Anda di halaman 1dari 151

About the Author

Rajiv Mukherjee is a consultant in unfired heat transfer based in New Delhi, India. He
has 36 years of experience in the thermal design, revamping, and troubleshooting of aircooled and shell-and-tube heat exchangers, and considerable experience in the design of
heat exchanger networks. He has written several articles in reputed journals and presented many papers at technical symposia. Rajiv has also served as faculty for numerous
courses on heat exchanger design and operation, energy conservation, and heat exchanger
networks, and presently teaches an intensive in-house refresher course on the design and
operation of heat exchangers that can be offered at any plant site or office location around
the world. He is an honors graduate in chemical engineering from Jadavpur University,
Kolkata, India.
In his spare time, Rajiv enjoys reading (Swami Vivekananda and Kahlil Gibran are big
favorites), writing, and listening to music. He lives in New Delhi with his wife, Kalpana.
Their daughter, Shilpi, and her husband, Bappa, presently live with their sons, Sohum and
Shivum, in Tokyo, Japan.
Rajiv may be contacted by e-mail at rajiv.mukherjee@vsnl.com or by telephone at
0091-11-2551 8281 or 0091-98711-20126.

R. Mukherjee

PREFACE

My desire to write this book was precipitated by the absence of such a practical book.
Recent heat exchanger design literature has been predominantly occupied by proceedings
of conferences. There is no book in the market that explains the logic of heat exchanger
thermal design and gives practical suggestions and recommendations for actually designing industrial heat exchangers. So, having written my earlier book, Practical Thermal
Design of Shell-and-Tube Heat Exchangers, which received a fairly good response, I decided to write a sequelone on air-cooled heat exchangers.
The theoretical aspects of single-phase heat transfer and condensation have been very
well presented in several books. So, what was really required was a practical how-todesign book with numerous worked-out examples or case studies to embellish or illustrate a
particular technique, facet, or style of design. The thousands of air-cooled heat exchanger
designs that I have been associated with over the last three decades have provided numerous
examples. They say that one picture is more eloquent than a thousand words. If you extend
this logic, one appropriate illustration by a case study is eminently more didactic than a long
dissertation on a particular subject.
This book has been written in the same style, language, and format as the one on shelland-tube heat exchangers. For the sake of convenience, both English and metric units have
been used throughout the book. There are 26 case studies, all aimed at embellishing,
illustrating, reinforcing, or demonstrating a feature, rationale, or methodology of design
elaborated or advocated in the text. Not only are the case studies based on the HTRI
software, the entire book is founded on the platform of HTRI know-how, which has become
a way of life for me for almost three decades.
Being a practical book, theory is limited to a bare minimum, and the accent is on
fundamentals, on design logic, on the interplay of parameters, on cause and effect, on
understanding why things happen the way they do. For example, why does a light
hydrocarbon condenser tend to have only four rows of tubes, whereas a heavy hydrocarbon
liquid cooler tends to have more rows of tubes? Or why do we choose 1/2 in. (12.7 mm)
high fins in certain situations but 5/8 in. (15.875 mm) high fins in others? Or why is the
process fluid break point between an air-cooled heat exchanger and its downstream trim
cooler related to the design ambient temperature? And many, many others.
This book has been written primarily for the heat exchanger thermal designer. However,
I think it will also be useful to process engineers, a significant part of whose routine job is to
specify heat exchangers. This book has not been written in an esoteric style for this very
reason. Since operating aspects are also often discussed, I trust it will be of interest to plant
operation specialists as well.
It is my fond hope that even B.S. and M.S. chemical and mechanical engineering
xi

students will find the book interesting, informative, and useful. I still remember when I was
an undergraduate studentI used to long for more practical, real-life information about
industrial practice. If one considers that many engineering graduates end up working in the
chemical process industries, there may be a lot of merit in adding such a flavor to heat
transfer in the university curriculum, as indeed it is to all other fields of human learning. The
juxtaposition of industrial equipment design practice with basic theory will go a long way in
making the subject more interesting and meaningful.
The thermal design of air-cooled heat exchangers is a fascinating activitysometimes
even more so than that of shell-and-tube heat exchangersfor the simple reason that there
are more variables: even the coolant (air) flow rate is a variable! This book will have served
its purpose if it can inspire the reader to consider the thermal design of air-cooled heat
exchangers as a joyous activity rather than a mundane chore.
I will be grateful for any feedback regarding any aspect of this book, and the same may
be sent to rajiv.mukherjee@vsnl.com or rajivmuk2003@yahoo.com.
R. Mukherjee
Heat Transfer Consultant
New Delhi, India

xii

Dedication

This book is dedicated to my parents, who would have been proud to see
this work. And that is an understatement. To my dear wife, Kalpana, who
has been supporting and inspiring me for over three decades now; to our
daughter, Shilpi, our son (-in-law), Bappa, and their sons Sohum and
Shivum; but most importantly, it is dedicated to the reader, whose approbation and appreciation would make all the toil worthwhile.

iii

Acknowledgments

I am indebted to Almighty God for having given me the education, intelligence, opportunity,
strength, and fervor to write this book.
I am also indebted to all those from whom I learned the design of air-cooled heat
exchangers over the years, especially to Wim Bos and Peter van der Broek of Lummus
Nederland B.V., who in 1971, initiated me on the path of air cooler design.
I will always be grateful to Cindy Mascone, ex-technical editor at Chemical Engineering
Progress, and her one-time compatriot, Gail Nalven, who led me to believe that I could write
a book. I will never forget Gails words, You do not know how prolific you are! As for
Cindy, she is the finest editor I have ever worked with.
This book might not have been possible without the wonderful exposition of air-cooled
heat exchanger technology by Heat Transfer Research, Inc. (HTRI). My long experience in
the field of air-cooled heat exchangers has been very largely honed on the platform of HTRI,
whose software I have been using since 1974.
I am grateful to HTRI, Begell House, Hudson Products Corp., Moore Fans Ltd., TEMA
(Tubular Heat Exchanger Manufacturers Association), and Cal Gavin Ltd. for permission to
use some of their diagrams and photographs in this book. These have been duly
acknowledged where they appear.
I am indebted to Bill Begell who decided to publish this book, and to all the people at
Begell House who were responsible for its production. Special thanks go to Donna
Thompson who did a splendid job of copyediting this book, as she did with the previous one
on shell-and-tube heat exchangers. Donna, I have enjoyed working with you again.
How can I forget my good friend Graham Polley in the UK? It was he who led me to
Bill in the first place.
I am thankful to Geoff Hewitt, who is the editor of the present series of books, for
having readily accepted this book into his fold.
Thanks are also due to my wonderful friend Sam Chapple of Edmonton, Canada, who
guided me on some important issues in the text.
I must also express my gratitude to another good friend, Lalit Shingal, who helped me
with the reproduction of many diagrams that appear in the book.
What we are able to accomplish in our lives, whether professionally or otherwise, is the
result of the Lords grace and the encouragement and support we receive from myriad
sources. This book is therefore truly a collaborative effort, and the credit belongs to the
human fraternity at large, rather than to any individual.

iv

CHAPTER 1

Introduction
Although air is much more freely available than water and costs nothing, process cooling
has historically been accomplished by cooling water. This is attributable to the much
lower cost of cooling by water, thanks to its substantially higher thermal conductivity and
lower temperature. However, with increasing shortages of cooling water and a consequent increase in its cost, air cooling has become more and more popular. Today, aircooled heat exchangers (ACHEs) are a common sight in the chemical process industries
(CPIs).
The first cost of an ACHE is much greater than that of a water-cooled heat exchanger
for the same heat duty, but its operating cost is usually much less. The operating cost with
water cooling comprises the cost of the initial raw water itself, makeup water, treatment
chemicals, apportioned cost of the cooling tower, and of course the pumping cost. For aircooled heat exchangers, the operating cost is only the cost of the power required to make the
air flow across the tube bundles. Thus, on an overall cost basis, ACHEs often compare quite
favorably with water-cooled heat exchangers.
The design of ACHEs comprises two distinct activities, namely, thermal design and
mechanical design. In thermal design the basic sizing of the heat exchanger is accomplished,
whereas in mechanical design the thicknesses and precise dimensions of the various
components are determined and a bill of materials is produced. Detailed engineering
drawings are then prepared based on which actual fabrication drawings are made. In this
book, as the title suggests, we shall talk principally about thermal design.
With the availability of sophisticated software, there has been an undue dependence on
them as black boxes, without the designer being truly in control of the design process and
understanding the nuances of design. A proper and sound understanding of the fundamental
principles and interplay of parameters is essential in order to produce an optimum design.
The principal purpose of writing this book is to help the heat exchanger thermal designer
attain such an understanding.
Presently, there is no book available on practical ACHE thermal design. This book is
based on the authors experience of over 36 years in the thermal design of ACHEs for the
chemical process industry, and reflects many real-life situations that were far from
straightforward. This book has been written in a structured, logical, and didactic manner, and
special effort has been made at bringing out the interplay of parameters for a thorough
understanding of basic issues.
As Example is better than precept, several case studies are presented in this book in
order to vividly bring out a particular methodology, principle, or practice that has been
advocated. The reader is invited to run these examples with further variations in the
parameters being examined, in order to develop a comprehensive understanding.
1

It is well known that the thermal design of ACHEs is still largely an enigma, with far
fewer engineering and fabricating companies practicing the trade than the thermal design
shell-and-tube heat exchangers. This is really quite surprising, considering that thermal
design of ACHEs is simpler and more straightforward than that of shell-and-tube heat
exchangers! This book will have served its purpose if it encourages more companies to
overcome this diffidence and take up the thermal design of ACHEs.
Now, coming to the individual chapters themselves, Chapter 2 dwells on the advantages
and disadvantages of air cooling, while Chapter 3 discusses the optimization of air and water
cooling. In some instances, only cooling by air need be employed, whereas in others only
cooling by water is adequate. However, in the vast majority of cases that fall between these
two extremes, cooling by both air and water is favorable.
Chapter 4 gives a detailed rundown of the various components and constructional
features of ACHEs, since a good understanding of the same is vital to the thermal design of
this equipment. This chapter will also be of considerable interest to mechanical designers of
ACHEs, since it explains the implications of several constructional features on thermal
design.
Chapter 5 discusses various basic concepts that form much of the foundation of
knowledge for ACHE design. The simultaneous optimization of airside and tubeside
calculations is certainly not an easy task. However, with the help of logical explanations,
arguments, and case studies, the design methodology is made easy to understand and apply.
Chapter 6 is on the thermal design of condensing ACHEs. After a brief classification of
condensers and a brief account of the mechanisms of condensation, practical guidelines for
thermal design are discussed. These include isothermal, narrow-range and wide-range
condensation, the effect of pressure, the handling of desuperheating and subcooling, nozzle
sizing, and the handling of condensing profiles and physical property profiles.
In Chapter 7, with the help of numerous case studies, optimization of ACHEs is
demonstrated vis--vis tube OD, fin height, fin spacing, number of tube rows, fan power
consumption, tube pitch, and the number of tube passes.
In Chapter 8, physical properties and heat release profiles are discusses at length. The
reader is offered guidance on how to feed heat release profiles, a matter that is not as simple
as it may appear.
Chapter 9 explains why overdesign is provided, and elaborates on the modalities of
overdesign for single-phase and condensing services.
After reviewing the various categories of fouling and the parameters that affect it,
suggestions are offered in Chapter 10 on how to specify fouling resistance. Comprehensive
guidelines are then suggested and analyzed in order to minimize fouling.
Chapter 11 is on the control of ACHEs, where various methods of control are discussed
in detail. Unlike water-cooled shell-and-tube heat exchangers, ACHEs offer very good
control on the process.
Chapter 12 deals with operating problems in air-cooled heat exchangers. Various
potential problems and ways to avoid them are discussed for both the tubeside and the
airside cases.
In Chapter 13, many special applications are elaborated on, including combined
services, recirculation ACHEs, humidified ACHEs, tube inserts, variable finning density,
natural convection, and vacuum steam condensers.

CHAPTER 2

Advantages and Disadvantages


of Air Cooling
Let us take a look at the advantages and disadvantages of air cooling as compared to water
cooling

2.1 Advantages of Air Cooling


Air cooling offers many advantages over water cooling. We have already discussed the
cost advantage of air cooling over water cooling. Besides this advantage, the use of air as
a cooling medium eliminates certain inherent disadvantages associated with water cooling:
a) The location of the cooler and thereby a plant is independent of a source of water
supply such as a river or a lake, or even a sea; hence, the plant can be located in
any geographic area. To use water as a cooling medium, however, the plant has
to be located at a site close to a large natural body of water such as a lake, river,
or sea. This could very easily entail a penalty in terms of transportation of raw
materials or finished products.
b) Air coolers are far more environment friendly since thermal and chemical pollution of the source of water are eliminated. In once-through cooling water, such as
with sea water, warmer water is returned to the body from which the water is
drawn, thereby leading to a rise in temperature of that body of water. This has a
direct adverse effect on the life and longevity of the aquatic plant and animal
species inhabiting the body of water. In recirculating cooling water systems
(which are the norm), the outlet warm water is cooled by a cooling tower so as to
eliminate this increase in temperature of the discharge water with its associated
adverse effect on aquatic life.
c) Maintenance costs are lowered considerably since frequent cleaning of the water
side of coolers (necessitated by fouling such as scaling, biofouling, sedimentation, etc.) is eliminated.
d) The installation is simpler since water piping and water pumps are eliminated.
Another advantage with air coolers is that they continue to operate (although at a reduced
capacity) by natural convection even when there is a power failure. In some cases, this
can be as much as 6070% of the design duty. In the case of water cooling, however, a
power outage usually means a plant shutdown, which results in direct loss in production.
Yet another advantage with air cooling is that air-cooled heat exchangers offer very
effective control of the process fluid outlet temperature (and thereby the heat duty) through
3

the following various means:


a) switching fans on/off
b) use of two-speed motors
c) use of autovariable fans
d) use of louvers
e) use of variable-speed drives
These will be discussed in detail in Chapter 11.
On the other hand, water cooling does not render an effective means of control of the
process fluid outlet temperature (and thereby the heat duty). This is because of two reasons:
(a) the MTD is predominantly controlled by the cold end temperature difference which does
not change with a reduction in the cooling water flow rate and (b) the cooling water film
resistance is a very small percentage of the overall resistance to heat transfer. Consequently,
a reduction in the cooling water flow rate has a negligible effect on the performance of a
water-cooled cooler.
In an air-cooled heat exchanger, however, a reduction in the air flow rate has a much
more pronounced effect on the performance of the cooler because both the MTD and the
overall heat transfer coefficient change significantly. This is because the airside heat transfer
coefficient controls the overall heat transfer resistance quite strongly, and the MTD also
varies significantly with a change in the air flow rate, and thereby the outlet air temperature.
This is illustrated in the following case study.

CASE STUDY 2.1: EFFECT OF REDUCTION OF AIR FLOW RATE


A quantity of 1,700,000 lb/h (771,115 kg/h) of hot water is to be cooled from 174F
(78.9C) to 140F (60C), representing a heat duty of 57.665 M Btu/h (14.53 M kcal/h).
The allowable pressure drop of hot water is 10 psi (0.7 kg/cm2) and its fouling resistance
is 0.001 h ft2F/Btu. For cooling by air, the design air temperature is 107F (41.7C). For
cooling by water, the cooling water inlet temperature is 93F (33.9C), its allowable pressure drop is 10 psi (0.7 kg/cm2), and its fouling resistance is 0.002 h ft2F/Btu (0.0004 h
m2C/kcal).
The air-cooled heat exchanger design was prepared first and its principal construction
parameters are indicated in Table 2.1a. The total air flow rate was 7,700,000 lb/h (3,493,700
kg/h). To demonstrate the effect of a reduction in the total air flow rate, the same was
changed to 6,900,000 lb/h (3,130,000 kg/h) and then to 6,100,000 lb/h (2,767,000 kg/h). The
principal performance parameters for all three total air flow rates are shown in Table 2.1b. It
will be seen that there is a significant reduction in the overdesign with a lowering in the air
flow rate, from 6.2% in the first case to 2.9% in the second case, and finally to 13.7% in
the third case. This is due to an appreciable reduction in both the MTD [from 33.2F
(18.4C) to 28.6F (15.9C)] and the airside heat transfer coefficient [from 178.8 Btu/h ft2F
(873 kcal/h m2C) to 163.2 Btu/h ft2F (797 kcal/h m2C)].
Next, the water-cooled hot water cooler design was then prepared, the principal
construction parameters of which are shown in Table 2.1c. The cooling water flow rate is
3,400,000 lb/h (1,542,200 kg/h). In order to demonstrate the effect of a reduction in the
coolant flow rate, the same was reduced by the same amount as the air flow rate in the case
of the air-cooled heat exchanger design. The principal performance parameters of all three
designs are shown in Table 2.1d. It will be seen that the reduction in overdesign is far less
than that of the air-cooled heat exchanger, from 7.3% in the first case to 3.5% in the second
4

Table 2.1a: Principal construction parameters of air-cooled hot water cooler


No. of bays in parallel
No. of bundles per bay
No. of tubes per row
No. of tube rows no. of tube passes
Tube/fin material
Tube OD thickness
Tube length, ft (m)
Fin height fin thickness
Fin density
Transverse pitch, in (mm)
Bundle width, ft (m)
Total bare tube area, ft2 (m2)
Total extended area, ft2 (m2)
No. of fans per section fan dia., ft (m)
Motor power, HP (kW)

4
2
46
52
CS/Aluminum
1 in (25.4 mm) 12 BWG (2.77 mm)
34 (10.36)
5/8 in (15.875 mm) 0.016 in (0.4 mm)
11 per in (433 per meter)
2.625 (67)
10.2 (3.11)
16,047 (1491)
381,080 (35,416)
2 14 (4.27)
30 (22.4)

Table 2.1b: Effect of variation in air flow rate on performance of air-cooled hot water cooler
7,700,000
(3,492,700)
138.1 (58.9)
0.43 (10.9)
2.1 (0.15)
838 (4092)
178.8 (873)
114.9 (561)
33.2 (18.4)
6.2
26.5 (19.7)

Total air flow rate, lb/h (kg/h)


Air outlet temperature, F (C)
Static pressure, in. WC (mm WC)
Tubeside pressure drop, psi (kg/cm2)
Heat transfer coefficient,
Tubeside
Btu/h ft2F (kcal/h m2C)
Airside (bare tube)
Overall
MTD, F (C)
Overdesign, %
Absorbed power, HP (kW)

6,900,000
(3,130,000)
141.7 (60.9)
0.35 (8.9)
2.1 (0.15)
838 (4092)
171.1 (835)
111.7 (545)
31.2 (17.3)
2.9
19.7 (14.7)

6,100,000
(2,767,000)
146.3 (63.5)
0.3 (7.6)
2.1 (0.15)
838 (4092)
163.2 (797)
108.3 (529)
28.6 (15.9)
13.7
14.2 (10.6)

Table 2.1c: Principal construction parameters of water-cooled hot water cooler


TEMA Type
No. of shells
Shell ID, in. (mm)
No. of tubes no. of tube passes
Tube OD thickness, in. (mm)
Tube length, ft (m)
Tube pitch, in. (mm)
Type of baffles baffle cut orientation
Baffle spacing, in (mm) no. of tube rows overlap
Connections: shellside/tubeside, nominal, in (mm)
Heat transfer area, ft2 (m2)

AEL (Fixed tubesheet)


1
45 (1143)
1578 1
0.75 (19.05) 14 BWG (2.108 mm)
20 (6.1)
1.0 (25.4) triangular
Double segmental horizontal
19 (483) 6
16 (400)/20 (500)
6083 (565)

case, and finally to 0.1% in the third case. This is because although there is a similar
reduction in the overall heat transfer coefficient as in the case of the air-cooled heat
exchanger, the drop in the MTD is far less.

2.2 Disadvantages of Air Cooling


Let us now consider the limitations of air-cooled heat exchangers as compared to watercooled heat exchangers. This is a comprehensive list and only some of them will be presented for a particular situation or application.
5

Table 2.1d: Effect of variation in cooling water flow rate on performance of water-cooled
hot water cooler
Cooling water flow rate, lb/h (kg/h)
Cooling water outlet temperature, F (C)
Pressure drop, psi (kg/cm2)
Shellside
Tubeside
Heat transfer coefficient,
Shellside
Btu/h ft2F (kcal/h m2C)
Tubeside
Overall
MTD, F (C)
Overdesign, %

3,400,000
(1,542,000)
110 43.3)
3.8 (0.27)
3.0 (0.21)
1539 (7514)
1065 (5200)
185.6 (906)
54.8 (30.4)
7.3

3,000,000
(1,360,800)
1123 (44.6)
3.8 (0.27)
2.3 (0.16)
1539 (7514)
970 (4736)
182.5 (891)
53.8 (29.9)
3.5

2,700,000
(1,224,700)
114.4 (45.8)
3.8 (0.27)
1.9 (0.134)
1539 (7514)
897 (4380)
179.8 (878)
52.8 (29.3)
0.1

High initial cost


Since air has a much lower thermal conductivity and specific heat than water, an aircooled heat exchanger has a much lower overall heat transfer coefficient and consequently a much larger bare tube heat transfer area than a water-cooled heat exchanger for
the same heat duty. Thus, the initial cost of an air-cooled heat exchanger (cooler hardware only) is considerably more than that of a water-cooled heat exchanger. However,
should the process fluid require ordinary carbon steel and the cooling water require admiralty brass or copper-nickel or duplex special stainless steel, the cost disadvantage of the
air-cooled heat exchanger reduces appreciably. The more superior the required cooling
water metallurgy, the lower the cost disadvantage of the air-cooled heat exchanger.
Costly winterization
In cold climates, extensive winterization arrangements have to be incorporated to negotiate subzero temperatures, thereby increasing the first cost even further. Winterizing an
air-cooled heat exchanger means rendering it operable even under winter conditions; that
is, the process stream will not freeze or congeal. These arrangements include heating by
steam coils, incorporating louvers to reduce air flow rate, reversing air flow, and even an
elaborate recirculation system that will be discussed in detail in Chapter 13.
Lower economical approach temperature
Because of the very low overall heat transfer coefficient, an economical approach temperature between the outlet process fluid and the ambient air is generally in the range of
1820F (1012C) whereas in water-cooled heat exchangers, this approach temperature
can be as low as 57F (34C). The logic here is that a very low MTD, coupled with the
typically low overall heat transfer coefficient of an air-cooled heat exchanger, will lead to
an enormous heat transfer area and thereby first cost. The relatively large MTD for air
cooling as compared to that for water cooling will neutralize to a large extent the inherently higher first cost of an air-cooled heat exchanger. This disadvantage is overcome by
having air cooling followed by trim cooling with water; that is, by off-loading the last
part of the cooling duty to water cooling. However, should cooling water be unavailable
at a particular site, this could represent a major limitation.
Larger plot area
Because of the larger heat transfer area, an air-cooled heat exchanger requires a considerably larger plot area than a water-cooled heat exchanger. However, this disadvantage is
6

overcome by locating an air-cooled heat exchanger on a pipe rack, so that no valuable


plot area is wasted. Process equipment such as shell-and-tube exchangers, accumulators,
and pumps are usually located at grade level under the pipe rack.
Fan noise
Due to the low specific heat and density of air, air-cooled heat exchangers have to force
large quantities of air across the tube bundles, resulting in a high noise level. However,
with the improvement in fan technology, including the development of low-noise fans,
the noise level can usually be restricted to permissible levels.
Limitations in plant layout
Air-cooled heat exchangers cannot be located near large obstructions, such as buildings,
since air recirculation can set in and affect the cooling performance adversely. However, this
need not represent a major limitation since a judicious plant layout can usually circumvent
this problem.
Availability of fewer vendors
Since air-cooled heat exchangers employ relatively sophisticated design technology, the
number of vendors who can offer air-cooled heat exchangers is far smaller than the number
of vendors who can offer water-cooled heat exchangers, which are shell-and-tube heat exchangers.
Problems associated with laminar flow
It is well known that viscous liquids yield laminar flow and a rather poor heat transfer coefficient when flowing inside tubes due to the boundary layer separation. Such liquids yield
considerably higher heat transfer coefficients when flowing on the outside of heat exchanger
tubes, due to the much higher turbulence with a staggered tube arrangement. Therefore, for
cooling viscous liquids, air-cooled heat exchangers become even more expensive due to the
extremely low tubeside heat transfer coefficient and thereby the higher heat transfer area.
Besides, such liquids often have a high pour point, which necessitates the use of an air recirculation arrangement, thereby increasing the first cost even more. However, this situation
can be remedied to a large extent by the use of wire-fin tube inserts. This is addressed later
in Chapter 13.

CHAPTER 3

Optimization of Air and Water Cooling


In applications where the process outlet temperature is relatively low, cooling by only air
may not be feasible. For example, cooling a light hydrocarbon liquid to 113F (45C) by
air may not be feasible at a site where the design ambient temperature is 107.6F (42C).
In such cases, a combination of cooling by air followed by trim cooling (cooling by water) has to be adopted.
Occasionally there are services where cooling by air may not be economically viable.
Thus, at a site where the design ambient temperature is 107.6F (42C), cooling by air may
not be viable for a naphtha stabilizer condenser service wherein the inlet/outlet temperatures
are 122/113F (50/45C), due to the very low temperature difference between the process
stream and the air. Here, only cooling by water should be employed.
In other services, where both the inlet and the outlet temperatures of the process stream
are relatively high, e.g., 248F (120C) and 140F (60C), only air cooling can be employed.
However, such services as described above are quite rare and invariably the process inlet
temperature will be fairly high, such as 212F (100C), whereas the process outlet
temperature will be rather low, such as 113F (45C). In such services, the most optimum
arrangement is to have an air-cooled heat exchanger, followed by a trim cooler employing
cooling water in a shell-and-tube heat exchanger.
Thus, for some services, only air cooling should be employed; for some others, a
combination of air and water cooling should be employed; while for the balance, only water
cooling should be employed. The vast majority of services in the chemical services fall in
the middle category where both air and water cooling should be used.
The optimum temperature break point between air and water cooling (that is, the
temperature at which a process fluid should leave an air-cooled heat exchanger and enter a
water-cooled heat exchanger) should be established by overall economics for every project,
because it will depend on equipment cost (air-cooled and water-cooled heat exchangers),
cost of water (total cost, as discussed in Chapter 1), and the cost of power. Generally
speaking, this optimum temperature is about 2732F (1518C) greater than the design
ambient temperature.
It is important to realize here that for a combination of air and water cooling, the aircooled heat exchanger will handle the major heat duty (75% or more of the total heat duty),
thereby resulting in a considerable reduction in the cooling water flow rate.
When using a combination of air and water cooling, it is usually best to design the aircooled heat exchanger for a somewhat lower ambient temperature (than what would be used
if there were no trim cooling), and then to design the trim cooler for the process fluid
temperature, which would be discharged by the air-cooled heat exchanger at the higher
ambient temperature. This is because with the higher ambient temperature, the decrease in
8

Table 3.1a: Heat exchanger service for Case Study 3.1


Fluid
Flow rate, lb/h (kg/h)
Temperature in/out, F (C)
Operating pressure, psia (kg/cm2a)
Total allowable pressure drop, psi (kg/cm2)
Fouling resistance, h ft2 F/Btu
(h m2 C /kcal)
Heat duty, MM Btu/h (MM kcal/h)
Vapor
Density in/out, lb/ft3 (kg/m3)
properties
Viscosity in/out, cp

Liquid
properties

Stabilizer overhead
127,870 (58,000)
153.3 (67.4)/116.6 (47)
172 (12.1)
8.5 (0.6)
0.00195 (0.0004)
18.72 (4.717)
1.643 (26.31)/1.382 (22.14)
0.01/0.0095

Specific heat in/out, Btu/lb F (kcal/kg C)


Thermal conductivity in/out, Btu/h ft F (kcal/h m C)
Density in/out, lb/ft3 (kg/m3)
Viscosity in/out, cp
Specific heat in/out, Btu/lb F (kcal/kg C)
Thermal conductivity in/out, Btu/h ft F (kcal/h m C)

0.45/0.43
0.0134(0.02)/0.0128 (0.019)
31.28 (501)/31.84 (510)
0.1/0.131
0.71/0.7
0.0605 (0.9)/0.0659 (0.098)

the cost of the trim cooler will be less than the increase in the cost of the air-cooled heat
exchanger, thereby resulting in a higher overall cost. This is illustrated by the following case
study.

CASE STUDY 3.1: SELECTION OF DESIGN AMBIENT TEMPERATURE


FOR AIR AND WATER COOLING
Consider the stabilizer condensing duty specified in Table 3.1a. The allowable pressure
drop of 8.5 psi (0.6 kg/cm2) is for both the air-cooled heat exchanger and the trim cooler.
The maximum and minimum ambient temperatures at the site were 100.4F (38C) and
64.4F (18C), respectively. The inlet temperature of the cooling water was 89.6F
(32C), its fouling resistance was 0.002 h ft2F/ Btu (0.0004 h m2oC/kcal), and its permitted pressure drop was 10.7 psi (0.75 kg/cm2). The optimum break temperature between
air and water was established as 131F (55C). Thus, an air-cooled heat exchanger was to
be designed for condensing the stabilizer overhead from 153.3F (67.4C) to 131F
(55C) and a trim cooler (using water) was to be designed to condense the balance vapor
from 131F (55C) to 116.6F (47C). Since there had to be a trim cooler, it was decided
to consider the design ambient temperature as 95F (35C).
An air-cooled heat exchanger was designed, its principal construction and performance
parameters being as per Table 3.1b.
By carrying a performance run, it was established that if the ambient temperature were
100.4F (38C), the stabilizer overhead outlet temperature would be 133.3F (56.3C).
Therefore, the trim cooler was designed for this heat duty; that is, for condensing the
uncondensed vapor from 133.3F (56.3C) to 116.6F (47C). A single TEMA (Tubular
Heat Exchanger Manufacturers Association) type AES shell having a heat transfer area of
2233 ft2 (207.5 m2) was found to be adequate. The principal construction and performance
parameters of this design are indicated in Table 3.1c.
Now, let us consider what would happen if the air-cooled heat exchanger were to be
designed for an ambient temperature of 100.4F (38C) and the trim cooler for condensing
the stabilizer overhead from 131F (55C) to 116.6F (47C). The bare tube area of the air9

Table 3.1b: Principal construction and performance parameters of air-cooled heat


exchanger for Case Study 3.1
Air inlet temperature, F (C)
1. No. of bays
2. No. of bundles per bay
3. No. of tubes per bundle per row no. of tube rows
4. No. of tube passes
5. Tube OD thickness, in. (mm)
6. Tube length, ft (m)
7. Fin OD thk. tube pitch, in. (mm)
8. Fin density, no./in. (mm)
9. Total bare tube area, ft2 (m2)
10. Heat transfer
Tubeside
coefficient, Btu/h ft2
Airside

F (kcal/h m2 C)
Overall
11. Pressure drop
Tubeside, psi (kg/cm2)
Airside, in. (mm) WC
12. Airflow rate, lb/h (kg/h)
13. MTD, F (C)
14. Power per fan, HP (kW)
15. Total fan power, HP (kW)
16. Fan diameter, ft (m)
17. Overdesign, %

95 (35)
100.4 (38)
1
1
2
2
38 6
44 4
2
2
0.984 (25) 0.098 (2.5)
41.0 (12.5)
2.24 (57) 0.016 (0.4) 2.638 (67)
11 (433)
4714 (438.1)
5459 (507.3)
364.7 (1781)
336.7 (1644)
188.9 (922.3)
184.8 (902.4)
98.3 (479.9)
95.1 (464.1) ()
5.3 (0.37)
4.17 (0.29)
0.5 (12.64)
0.45 (11.5)
1,984,100
2,160,500
(900,000)
(980,000)
33.1 (18.4)
29.0 (16.1)
28.7 (21.4)
28.7 (21.4)
57.4 (42.8)
57.4 (42.8)
14 (4.267)
15 (4.571)
11.0
8.6

cooled heat exchanger increases from 4714 ft2 (438.1 m2) to 5459 ft2 (507.3 m2). The
principal construction and performance parameters of this design are shown in Table 3.1b.
Notice that the fan diameter also goes up from 14 ft (4.267 m) to 15 ft (4.57 m) in order to
provide the minimum 40% bundle coverage. However, since its heat duty is lower, the trim
cooler heat transfer area decreases from 2233 ft2 (207.5 m2) to 1891 ft2 (175.7 m2).
The above results are summarized in Table 3.1d. It will be seen that when the air-cooled
heat exchanger is designed for 100.4F (38C), (a) the total heat transfer area of the aircooled heat exchanger and the trim cooler is significantly higher and (b) the heat transfer
area of the air-cooled heat exchanger too is higher. Therefore, since air-cooled heat
exchangers cost significantly more than shell-and-tube heat exchangers for the same bare
tube heat transfer area, the installed cost of this option is considerably higher. Consequently,
it will be more economical to design the air-cooled heat exchanger for an ambient
temperature of 95F (35C) and the trim cooler for a stabilizer overhead temperature that
would be the outlet from the air-cooled heat exchanger when the ambient temperature is the
maximum (100.4F or 38C).
From the above example, we see that is more economical to design an air-cooled heat
exchanger for an ambient temperature that is somewhat lower than the maximum ambient
temperature, and to pass on the shortfall in the heat duty of the air-cooled heat exchanger to
the trim cooler when the ambient temperature is the maximum expected.
Evidently, the above approach is possible only for air-cooled heat exchangers that are
followed by a trim cooler. If an air-cooled heat exchanger is not followed by a trim cooler, it
will obviously have to be designed for the maximum expected ambient temperature.
10

Table 3.1c: Principal construction and performance parameters of water-cooled heat


exchanger for Case Study 3.1
1. Shellside inlet temperature, F (C)
2. Heat duty, Btu/h (kcal/h)
3. Type of exchanger
4. Shell ID, in. (mm)
5. Number of tubes
6. Total heat transfer area, ft2 (m2)
7. Tube pitch, in. (mm) layout angle
8. Number of tube passes
9. Baffling
Baffle spacing, in (mm)
Baffle cut (dia), %
10. Shellside
Cross-flow
velocity, ft/s (m/s)
Window flow
11. Shellside pressure drop, psi (kg/cm2)
12. Tubeside velocity, ft/s (m/s)
13. Tubeside pressure drop, psi (kg/cm2)
14. Heat transfer
Shellside
coefficient, Btu/h ft2
Tubeside
F (kcal/h m2 C)
Overall
15. Overdesign, %

133.3 (56.3)
131 (55)
5.85 (1.475)
4.89 (1.232)
Floating head (TEMA AES)
30.9 (785)
28.7 (730)
560
474
2233 (207.5)
1891 (175.7)
1.024 (26) square
4
4
10.8 (275)
10.8 (275)
25
25
5.61 (1.71)
5.45 (1.66)
5.09 (1.55)
5.35 (1.63)
2.13 (0.15)
2.42 (0.17)
4.82 (1.47)
4.76 (1.45)
1.9 (0.58)
1.84 (0.56)
241 (1177)
259 (1263)
996 (4865)
996 (4865)
109 (533)
112.4 (549)
8.3
7.9

Table 3.1d: Overall comparison of the two cases for Case Study 3.1

1. Bare tube area of


air-cooled heat
exchanger, ft2 (m2)
2. Bare tube area of
trim cooler, ft2 (m2)
3. Total bare tube
area, ft2 (m2)

Air-cooled heat exchanger designed for


95F (35C), trim cooler designed for duty
corresponding to air-cooled heat
exchanger performance at 100.4F (38C)

Air-cooled heat exchanger


designed for 100.4F
(38C), trim cooler designed
for corresponding duty

4714 (438.1)

5459 (507.3)

2233 (207.5)

1891 (175.7)

6947 (645.6)

7350 (683)

Further Reading
1.

Maze, R.W., 1975, Air Cooler or Water Tower: Which for Heat Disposal, Chem. Eng.,
Jan. 6.

11

12

CHAPTER 4

Construction Features of
Air-Cooled Heat Exchangers
4.1 Introduction
Before we start discussing the thermal design of air-cooled heat exchangers, it will be
necessary to have a detailed look at the constructional features.
The principal components of an air-cooled heat exchanger are:
Tube bundle
Fans and drive
Plenum chamber
Structure
Before discussing these principal components, let us first consider some important terms
in air-cooled heat exchanger parlance.
Tube bundle: A tube bundle is an assembly of headers, tubes, tube supports, and
frames (Fig. 4.1).
Bay or section: A bay or a section is composed of one or more tube bundles served by
two or more fans, complete with structure, plenum, and other attendant equipment
(Fig. 4.2). Thus, a bay is the smallest independent part of an air-cooled heat exchanger that is repeated for multibay or multisection units.
Unit: A unit is composed of one or more tube bundles in one or more sections for an individual service (Fig. 4.3).
Bank: A bank or battery of air-cooled heat exchangers comprises one or more sections
or units arranged on a continuous structure (Fig. 4.4).

Fig. 4.2 A bay or a section

Fig. 4.1 Tube bundle


13

Bundle

Bay

Plan view

Bundle
Fan

Unit A

Unit B
Bank

Front elevation

Fig. 4.3 A unit

4.2

Tube Bundle

As defined above, a tube bundle is an assembly of tubes, headers, tube supports, and side
frames. We shall now take these up one by one.

Fig. 4.4 A bank or battery of air-cooled heat exchangers


14

4.2.1 Finned tube


Because of the extremely
low heat transfer coefficient of air, which is a
direct result of its low
thermal conductivity, it
becomes imperative to
employ extended surface
on the airside. The commonest and cheapest form
of extended surface is the
finned tube (Fig. 4.5).
The base tube may be
of any commercially available material suitable for
the process, based on

considerations of corrosion, pressure, and temperature, but it is usually of carbon steel or


stainless steel. The fins are invariably of circular cross section, although fins of elliptical
cross section have also been used. Although occasionally of steel for high-temperature
services, fins are normally of aluminum because it has the most favorable thermal
conductivity-to-cost ratio, besides having good cold-working properties. Since these fins can
be as thin as 0.016 in. (0.4 mm), it is common to pack in 11 fins per inch (433 fins per
meter). However, several vendors prefer to limit the fin density to 10 per inch (394 per
meter) on account of airside fouling, while others are known to employ even 12 per inch
(472 per meter). Limiting the fin density to 11 per inch (433 per meter) appears to be a very
sensible practice. The manufacturers of finned tubes find it economically practical to limit
the fin density to between 7 and 11 per inch (276 to 433 per meter).
Since the coefficient of linear expansion of aluminum is about twice that of carbon steel,
a gap resistance between the tube and the fin material develops and this increases as the
operating temperature is increased, due to the increased difference of the coefficients of
linear expansion. Thus, depending on the type of bond between the tube and the fin,
maximum operating temperatures have been established. These are specified for the various
types of finned tubes in Sections 4.2.1.1, 4.2.1.2, 4.2.1.3, and 4.2.1.4.
The standard tube OD is 1 in., although 1-1/4 in., 1-1/2 in., and even 2 in. are employed
where the tubeside pressure drop is controlling. The corresponding metric values are 25 mm,
32 mm, 38 mm, and 50 mm. The standard fin heights are 3/8 in. (10 mm), 1/2 in. (12 mm),
and 5/8 in. (16 mm), with the latter two being far more popular than the former. Such a
standardization becomes necessary, considering an economical production program of
finning machines.
There are various types of finned tubes, as described below.
4.2.1.1 Single L-footed finned tube
This is a circular fin wrapped helically around the tube under tension (Fig. 4.5a). Full
coverage of the base tube by the L-foot offers good protection against atmospheric corrosion. However, these fins tend to become loose with time, thereby causing an appreciable
deterioration in the airside performance due to the air gap between the tube and the fins
(see Ref. [1]). Consequently, their use is generally restricted to applications where the
process inlet temperature is less than 248F (120C).
However, even for such applications, their use is generally not very popular because the
airside performance of these finned tubes is more likely to deteriorate with time due to the
loosening of the fins. An exception is in corrosive marine atmospheres, such as on offshore
platforms, where the good protection provided by these tubes against atmospheric corrosion
of the base tubes makes them superior to conventional grooved fins. However, in such
applications, it would be advisable to derate the effectiveness of these tubes to cater to the
probability of loosening of the fins from the base tube.
4.2.1.2 Double L-footed finned tube
These tubes offer an even better coverage of the base tube (Fig. 4.5b) but since they are
much more costly, may be preferred only in extremely corrosive atmospheres. The upper
limit of process fluid inlet temperature for these finned tubes is 338F (170C). However,
claims of bond improvement and improved corrosion protection have been questioned by
some users. The use of these finned tubes is quite rare in the chemical process industries.

15

4.2.1.3 Grooved or embedded or G type finned tubes


In these tubes, the fin is embedded in the tube by first plowing a groove in the tube wall,
and then stretching the fin material into the groove under sufficient tension to achieve
specified bond strength (Fig. 4.5c). Evidently, G-finned tubes require a thicker tube
wall than L-footed tubes. As per clause 4.1.11.3 of API (American Petroleum Institute)
661, the minimum tube wall thickness is 0.083 in. (2.1 mm) for carbon steel and lowalloy steel, and 0.065 in (1.65 mm) for stainless steel. For embedded fin tubes, however,
this thickness is reckoned from the bottom of the groove. Hence, embedded fin tubes
have to be thicker than L-footed fin tubes by the groove depth (which is usually the
equivalent of one gauge).
Grooved finned tubes can tolerate process fluid temperatures of up to 752F (400C)
due to their strong fin bond (usually no fin-bond resistance penalty is applied because it is

(a)

(b)

(c)

(d)

(e)

Fig. 4.5 Various types of finned tubes: (a) single L-footed; (b) double L-footed; (c) grooved (Gtype); (d) bimetallic; (e) extruded
16

considered negligible) and they are a very commonly used type of finned tubes. They can
also withstand cyclic operation without any loss of fin-tube contact. The disadvantage of
these tubes is that their base tube material is exposed to the atmosphere and, therefore, their
use in aggressive atmospheres (such as marine applications) is not recommended.
4.2.1.4 Bimetallic finned tubes
Bimetallic finned tubes have G-fins embedded in an outer tube of aluminum, which is
stretched over the base tube (Fig. 4.5d). In applications where the process fluid is at high
pressure and/or is corrosive, thereby requiring the use of an expensive alloy, it may be
cheaper to use bimetallic finned tubes with a thin-wall inner tube, than to use a heavy
base tube of high alloy with G-fins. The upper temperature limit for these finned tubes is
554F (290C).
4.2.1.5 Extruded finned tubes
These are basically a double tube construction. An outer tube with a large wall thickness,
usually aluminum, is swaged over an inner base tube and extruded into high fins, all in one
operation. An uninterrupted bond over the entire tube length is formed, thereby protecting
the base tube from the outside (Fig. 4.5e). The fin bond is thus considered superior to that of
any other interference fit fin type.
Due to the high fin material usage and the more expensive manufacturing process, these
finned tubes have the highest cost. Their applications are similar to those of bimetallic finned
tubes.
For a detailed discussion of fin bond resistance and the maximum permissible operating
temperature for the above types of high-finned tubes, the reader is referred to [1]. For a
detailed discussion on specifying the right fin type for different applications, please see [2].
4.2.2 Headers
Headers serve to introduce the hot fluid into the tubes and collect the cooled fluid at the
end of the flow passage. They carry the inlet and the outlet nozzles and other connections
(such as vents and drains), as well as the pass partition plates required for multipass exchangers.
Headers are so arranged that movement within the side frame is possible to contain
thermal expansion. As per API 661, clause 6.1.6.1.2, if the temperature difference between
the inlet to one pass and the outlet from the adjacent pass is greater than 200F (111C), split
headers (Fig. 4.6.) or other means of restraint relief are to be employed. This condition also
applies when the maximum operating temperature is greater than 350F (177C).
There are various types of header
construction, each having specific
advantages and disadvantages. The
most common header types are as
follows.
4.2.2.1 Plug type header
This is the most commonly used type
of construction (Fig. 4.7a) and comprises a rectangular, welded box with
inlet nozzles in the top plate and outlet

Fig. 4.6 Split header


17

nozzles in the bottom plate. The tubes are either welded to the tubesheet or expanded into
tube holes in the tubesheet. The plug hole opposite each tube in the plug sheet allows mechanical cleaning of each tube and plugging in case of leakage.
This type of header is relatively cheap and can be used for pressures up to 3000 psig
(211 bar g). The disadvantage of plug headers is that for frequent cleaning of tubes for dirty
services (fouling resistance greater than 0.00195 h ft2F/Btu or 0.0004 h m2C/kcal),
removal of the large number of plugs becomes time consuming and therefore costly.
Thus, plug headers are preferred for clean and moderate-to-high pressure applications.
4.2.2.2 Cover-plate type header
For fouling services, cover-plate type headers (Fig. 4.7b) are preferred since it is much
easier to remove cover plates than the numerous plugs of plug type headers. However, at
higher pressures, this header type becomes expensive since the cover plate becomes very
thick. Hence, cover-plate type headers are usually not used for pressures in excess of 569
psig (40 bar g). For easy removal, cover-plate type headers are equipped with jackscrews
and lifting lugs.
4.2.2.3 Manifold headers
Round manifold headers (Fig. 4.7c) are used in very-high-pressure applications. The
tubes are welded to the manifold by means of stubs and will usually have return bends,
i.e., the tubes will be U-tubes.
Due to manufacturing limitations, the number of tube rows per manifold is restricted to

(a)

(b)

(c)

(d)

Fig. 4.7 Various types of headers: (a) plug; (b) cover plate; (c) manifold; (d) bonnet
18

one or two. Thus, the choice in the number of tube passes becomes restricted. For example,
if a manifold has a single row of tubes, the number of tube passes has to be equal to the
number of rows. If a manifold has two rows of tubes, the number of tube passes has
necessarily to be half the number of tube rows.
Since cleaning of the inside of tubes can only be carried out either chemically or by
cutting of the U-bends, this type of construction is not recommended for dirty services.
However, for pressures above 3000 psig (211 bar g), it usually becomes advantageous to use
manifold type headers rather than plug type headers.
4.2.2.4 Bonnet type headers
In a bonnet type construction (Fig. 4.7d), a semicircular bonnet is welded or bolted to the
tubesheet. This is a relatively inexpensive construction, but there is an inherent disadvantage: the piping must be removed for cleaning, or even plugging a leaking tube. Hence,
its use is not common.
4.2.3 Tube Supports
Since air-cooled heat exchanger tubes are usually very long, they require periodic support
so that they do not sag. Therefore, the finned tubes are supported either by special aluminum support boxes or by zinc collars cast on the tubes themselves.
4.2.4 Side frames
Side frames (Fig. 4.8) perform two roles:
1) They support the headers and the finned tubes, thereby making the tube bundle a
rigid, self-contained assembly that can be transported and erected conveniently.
2) They serve as seals and prevent bypassing of air.
4.2.5 Tube-to-tubesheet joint
As in shell-and-tube heat exchangers, tubes are expanded into double grooves in the
tubesheet (Fig. 4.9a) at low-to-medium pressures. At high pressures, however, tubes are
strength welded to the tubesheet (Fig. 4.9b).

4.3 Fans and Drives


Side frame

4.3.1 Fans
Axial fans (Fig. 4.10)
employed in air-cooled
heat exchangers have
the common characteristic of displacing very
large volumes of air
against a low static
pressure (typically, 0.6
0.8 WC or 15.220.3
mm WC). These fans
have characteristic performance curves that
are proprietary to each

Side frame

Fin OD

Fig. 4.8 Side frame


19

Sealing
strip

Before expansion

After expansion

(a) Expanded

(b) Welded

Fig. 4.9 Tube-to-tubesheet joint: (a) expanded; (b) welded

fan manufacturer. Hence, the designer should have access to fan curves giving information regarding volume (or mass) of air, static pressure, absorbed power, and noise. Some
fan manufacturers even furnish computer software to aid the designer in proper fan selection. Also, some contemporary software packages incorporate fan performance curves of
the leading fan manufacturers in their software.
The fan diameter for air-cooled heat exchangers usually varies between 6 ft (1.83 m)
and 18 ft (5.49 m), although fans having smaller and larger diameters are employed in
special circumstances. A fan consists of two basic components: the hub and the blades.
4.3.1.1 The hub
The hub (see Fig. 4.10) is the component that is mounted on the fan shaft and the blades
are mounted on the hub. Hub material may be cast iron, cast aluminum or fabricated
steel. Manufacturers usually conduct static and dynamic balancing of the hub in the shop.
The hub is usually of two types:
a) Manually adjustable, where the blade angle can be altered only when the fan is
stationary.
b) Autovariable, where the hub carries a device (usually a pneumatic controller) that
can alter the blade angle even while the fan is in motion in order to control air
flow. Control is usually effected by means of a signal from a TIC (temperature
indicator controller) responding to the
outlet temperature of the process fluid.
4.3.1.2 Blades
Blades (Fig. 4.10) can either be of metal
(usually aluminum) or FRP (fibreglass
reinforced plastic). Plastic blades are
suitable only for temperatures of up to
158F (70C); thereby representing a
limitation for induced-draft air-cooled
heat exchangers.
Fan performance (air flow rate and
static
pressure) is determined by the
Fig. 4.10 Axial fan (Courtesy Moore Fans Limited)
20

number of blades, tip speed, blade angle, and blade width. The effect of a change in the tip
speed of a fan on its performance is dramatic. The volume of air flow varies directly with the
tip speed; the pressure varies as the square of the tip speed and the horsepower varies as the
cube of the tip speed. The tip speed is normally limited to 200 ft/s (61 m/s) since noise
becomes excessive at higher values.
An increase in the number of blades of a fan increases its ability to work under pressure.
Thus, the tip speed of a six-blade fan can be reduced to deliver the same volume of air, as
compared to that of a four-blade fan. However, this can be carried too far in that as the
number of blades increases beyond six, multiblade interference may actually reduce the
efficiency of the fan, since each blade works in the disturbed wake of the preceding blade.
Therefore, the number of blades has to be selected carefully by the fan vendor.
All the blades of a fan should be set at the same angle for smooth operation. Usually, the
blade angle is set between 12 and 27. This is because performance deteriorates at low
angles and becomes unstable at high angles. The volumetric flow rate varies as the blade
angle tangent ratio (BART) to the 1/3 power and pressure to the 2/3 power. Therefore, HP
varies directly with BART. The blade angle should also be carefully selected by the fan
vendor.
A fan with a wider blade width can be operated at a lower tip speed to achieve the same
performance. Consequently, fans with wider blades operate less noisily. This feature is
exploited by fan vendors who offer special low-noise fans.
4.3.1.3 Fan laws
Fans of the same basic design and dimensions operate theoretically in accordance with
certain fan laws. In practice, these laws do not apply exactly because of design considerations and manufacturing tolerances, but they are useful in estimating the approximate
outputs of similar fans of different diameters and speeds, as applied to normal air delivery. These laws can be expressed as follows:
a) The volume of air flow varies as rpm and as (fan diameter)
b) The pressure developed varies as (rpm) and as (fan diameter)
c) The power absorbed by the fan varies as (rpm) and as (fan diameter)5
It is important to note, however, that these laws apply to the same operating point on a
fan characteristic. They cannot be used to predict other points on the fans curve.
These laws are most often used to calculate changes in the flow rate, pressure, and
power of a fan when the size or rotational speed is altered. They assume no change in fan
efficiency for any given point on the fan curve, when there is a change in speed.
4.3.1.4 API specifications
Section 4.2.3 of API 661 stipulates the following for fans and fan hubs:
1) There should be at least two fans along the tube length; however, a single-fan design can be agreed to in exceptional circumstances (such as very small units) between purchaser and vendor. This two-fan requirement is apparently based on
considerations of reliabilityshould one fan stop functioning due to a belt
breakage or other reason, the other fan will be running so that the unit will continue to run, albeit at a somewhat lower heat duty. Furthermore, at lower loads
and at cooler ambient temperatures, one fan may be stopped for better control of
the process outlet temperature as well as for power savings. Another reason is
that even when autovariable fans are used, both fans need not be autovariable;
21

one autovariable fan and one manually adjustable fan can achieve the necessary
control.
2) Fans should be of the axial-flow type and each fan should occupy at least 40% of
the tube bundle face area served by it. This is to ensure a reasonably good distribution of air across the face of the tube bundles.
It may be added here that when the width of a bay is about one-half of the
tube length, it represents an ideal situation because the fans then see a square
cross section of the bay since there are two fans in a bay. For example, if the
width of a bay is 20 ft (6.095 m) and the tube length 40 ft (12.191 m), each of the
two fans will deliver air across a 20 ft (6.095 m) 20 ft (6.095 m) cross section.
A 14 ft (4.27 m) diameter fan with a cross-sectional area of 153.9 sq. ft (14.3 sq.
m) will just fall short of the 40% requirement specified above, so that the use of a
15 ft (4.57 m) diameter fan will be necessary. The reader should note that a tube
length-to-bay width ratio of 2:1 is not a must, but it is desirable that it does not
exceed this ratio substantially. Furthermore, when this ratio tends toward 3:1, a
design with three fans per section will perform much better than one with two
fans.
3) The fan dispersion angle (see Fig. 4.11) should not exceed 45 at the centerline of
the tube bundle. This requirement is also based on proper air distribution.
4) The radial clearance between the fan ring and the fan tip should not exceed 0.5%
of the fan diameter or 0.75 in. (19 mm), whichever is less; fan stalling may occur
at larger clearances. Compliance with this requirement is strongly recommended
since it yields significantly better fan performance but is not expensive to implement. Table 4.1 gives a clearer picture of the effect of tip clearance on the performance of a 5 ft diameter fan. Notice that until a 0.5% tip clearance, the fans

4
ma5
x.

45
Max.
dispersion
angle

Fig. 4.11 Fan dispersion angle


22

Table 4.1: Effect of fan tip clearance on its performance


Tip clearance
in. (mm)
0.074 (1.9)
0.15 (3.8)
0.2 (5.1)
0.3 (7.6)
0.4 (10.2)
0.464 (11.8)

Tip clearance as % Fan performance


of fan diameter
efficiency (%)
0.12
99.5
0.25
98.5
0.33
97.4
0.5
95.0
0.67
91.0
0.77
88.0

perform quite well; but thereafter, there is a rapid deterioration in efficiency.


5) Fan tip speed should not exceed 200 ft/s (61 m/s) unless approved by the purchaser, and in no case exceed 266 ft/s (81 m/s). As stated earlier, the noise level
of a fan increases sharply beyond this tip speed.
Some additional recommended design guidelines are as follows:
1) The minimum distance between the plane of the fan and the tube bundle (that is,
the plenum height) should be one-half the fan diameter for forced draft units and
one-third the fan diameter for induced draft units. These requirements are for
maintaining favorable aerodynamics and thereby superior performance of fans.
2) For both forced draft and induced draft units, the height of the fan ring should be
at least one-sixth of the fan diameter.
3) Air seals should be provided between tube bundles and between tube bundles and
the plenum chamber in order to minimize air bypassing. Any gap wider than 0.4
in. (10 mm) should be considered excessive and therefore sealed.
4.3.2 Fan drives
The power required by the fans to move air across the tube bundles is provided by an
electric motor, steam turbine, gas or gasoline engine, or hydraulic motor, with the electric
motor being the overwhelming choice. Polyphase, squirrel-cage totally enclosed fancooled (TEFC) induction type motors are usually used. Steam turbine drivers are required
to be as per API Standard 611.
The power is transmitted from the motor (or the turbine) to the fans through direct drive,
V-belt drive, HTD (high-torque drive), or gear drive. A direct drive (Fig. 4.12a) has the fan
shaft directly connected to the driver and is usually used with fan diameters of 5 ft (1.53 m)
or less, as well as drives of 5 HP (3.73 kW) or less.
A V-belt drive (Fig. 4.12b) is used when the rpm of the fan is less than the rpm of the
driver. V-belt drives may be used with motor drives rated 30 HP (22 kW) or less, as per API
661 clause 4.2.8.2.10.
HTD may be used with motor drives rated 50 HP (37 kW) or less, as per API 661 clause
4.2.8.2.11. Unlike flat and V-belts, HTD belts do not rely on friction for its pulling power.
HTD belts utilize a revolutionary new tooth design that substantially improves stress
distribution and higher overall loading.
HTD belts do not stretch due to wear, are corrosion resistant, and operate at reduced
noise levels. The belts are capable of transmitting higher torque at lower speed, thus
improving the horsepower capacity of toothed belts. The belts do not depend on thickness to
develop great tensile strength. There is no loss of speed caused by belt creep or slippage as
with flat and V-belts.
23

HTD systems have unusually high mechanical efficiency. Further, transmission


efficiency is not lost with use. The belt construction ensures very little heat buildup since
friction is not required to pull the load. Since belt tension is reduced, significant power
savings are thus achievable, especially on larger HP installations.
Maintenance is simple. No adjustments are required due to stretch or wear. HTD belts
are ideal where proper maintenance is difficult or where downtime could prove to be
extremely expensive.
For electric motors rated above 50 HP (37 kW), right-angle gear drives (Fig. 4.12c)
must be used (API 661 clause 4.2.8.3.1). All steam turbine drivers must employ right-angle
gear drives (API 661 clause 4.2.8.3.3)
4.3.3 Plenum chamber
The air delivered by a fan is
distributed to (forced draft) or
collected from (induced draft)
the tube bundle by a plenum
chamber that consists of ductwork in the form of a rectangular box (Fig. 4.13) or a
cone/rectangle transition piece
(Fig. 4.14). For forced draft
units, the plenum chamber can
be square (or rectangular) or
conical whereas for induced
draft units they are invariably
conical. A partition is provided
between fans, and the gap between the tube bundle and the
partition plate should not exceed
20 mm.
For forced draft units, the
plenum chamber has a conical
inlet at the bottom to reduce inlet
losses. When low-noise fans are
employed, the conical inlet is
replaced by a bell mouth.

(a) Direct Motor Drive

(b) Belt Drive

4.4 Configuration
of ACHEs
Horizontal configuration
Air-cooled heat exchangers
(ACHEs) are usually configured
(c) Right-angle gear drive with fan support
in the horizontal disposition
since maintenance becomes easFig. 4.12 Fan drives: (a) direct drive; (b) V-belt drive; (c)
ier (see Fig. 4.15).
right-angle gear drive
24

A-frame configuration
This design is almost exclusively employed in power plants for condensing turbine exhaust steam (see Fig. 4.16). The tube bundles are mounted on a triangular frame with the
fans located below. The inclination from the horizontal is
Bundle
usually between 45 and 60.
The A-frame configurations permit a 3040% reduction in the plot area as
compared to a horizontal conPlenum
figuration. Additionally, and no
less importantly, the A-frame is
ideally adapted for condensing
since it facilitates condensate
Fan ring
drainage. The common header
at the top of the unit allows
uniform steam distribution with
minimum pressure loss, which Fig. 4.13 Box type plenum (redrawn with permission from
HTRI)
is important for the efficient
operation of vacuum steam
Bundle
condensers. The A-frame configuration is in fact the basis of
several patented freeze-proof
designs.
Vertical configuration
Plenum
Vertical configurations are
generally employed for packFan ring
aged units such as compressors with their intercoolers
(see Fig. 4.17). Evidently,
Fig. 4.14 Transition plenum (redrawn with permission from
they are used where floor HTRI)
space is at a premium and
could thus be used advantageously in offshore platforms
as well. They are much more
prone to deterioration in performance due to crosswinds.
Furthermore, multipass designs are not feasible for condensing services.

4.5 Natural Draft versus


Mechanical Draft
In natural draft there are no
fans, so that the flow of air is Fig. 4.15 Horizontal configuration of an air-cooled heat exby natural convection due to changer (redrawn with permission from HTRI)
25

the stack effect across the tube bundle.


An external stack is sometimes incorporated to increase the draft and thereby
the cooling. The principal application is
in dry cooling towers in power plants
where a large chimney (dry cooling
tower) establishes an appreciable draft.
Some process licensors specify external
stacks for process condensers so as to
achieve 65 or 70% plant throughput
even under power failure conditions.
Most air-cooled heat exchangers are
of the mechanical draft type. Vast
amounts of air are moved across finnedtube bundles by axial fans driven by
electric motors. There are two principal
Fig. 4.16 A-frame configuration of an air-cooled categories of mechanical draftforced
heat exchanger (redrawn with permission from draft and induced draft. In forced draft
HTRI)
(see Fig. 4.18), the fans are mounted
below the tube bundles and blow air
across the finned tubes. In induced draft
(see Fig. 4.19), the fans are located above
the tube bundles and suck air across the
finned tubes. Each type has its advantages
and disadvantagesand therefore preferred application, as elaborated below.

4.6 Forced Draft versus


Induced Draft
Forced draft advantages
The principal advantages are as follows:
Fig. 4.17 Vertical configuration of an air-cooled 1) Since both fans and motors/drive
heat exchanger (redrawn with permission from transmissions are located below the tube
HTRI)
bundles, accessibility of the same for
maintenance is far better.
2) Since the fans are located below the
tube bundles and handle the colder incoming air, the air pressure drop and
therefore the fan power consumption are
somewhat lower.
3) Fan blade life is longer since exposure is to cold inlet air.
4) It is possible to have a recirculation
air-cooled heat exchanger system to
avoid freezing and other solidification
Fig. 4.18 Forced draft (redrawn with permission problems only with forced draft fans.
from HTRI)

26

Forced draft disadvantages


The principal disadvantages are as follows:
1) Poorer distribution of air across the tube bundles, since the air leaves the tube
bundles at a much lower velocity.
2) Greater possibility of hot air recirculation as a result of the lower discharge velocity and the absence of a stack. Hot air recirculation results in an increase of the
air inlet temperature and consequently a decrease in the MTD. In low-MTD applications, the deterioration in performance can be significant. Consequently, the
forced draft type is not preferred where the cold end temperature approach (the
difference in temperature between the process outlet and the inlet air) is less than
914F (58C).
3) Exposure to the elements (sunlight, rain, hail, and snow), unless louvers or roofs
are provided at the top of the tube bundles. This results in poorer stability and
process control, as well as possible damage to the finned tubes.
4) Due to a very small stack effect, natural draft capability in the event of fan failure
is rather low.
Induced draft advantages
The principal advantages are as follows:
1) Better air distribution across the tube bundles with better cooling.
2) The probability of hot air recirculation is considerably lower. The air velocity at
the discharge is usually over twice that at the entrance.
3) Due to the much higher stack effect, natural draft capability under fan failure
conditions is much higher.
4) Better process control and stability from effects of rain, snow, hail, or sunlight, as
well as protection from the damaging/negative effects from the same.
5) No possibility of damage to fans and/or drive due to leaking products, where corrosive.
Induced draft disadvantages
The main disadvantages are as follows:
1) The fans and drives are less accessible for maintenance, being located above the
plenum chamber. Further, maintenance work may even have to be carried out in
the hot air caused by natural convection.

Fig. 4.19 Induced draft (redrawn with permission from HTRI)


27

2) Higher air pressure drop and thereby motor power because of handling hotter air,
which is lighter.
3) In order to prevent damage to fan blades, V-belts, bearings, and other mechanical
components, the exit air temperature has to be limited to about 194F (90C). Fiber-reinforced plastic blades (which give superior performance) cannot be used at
an air temperature higher than 158F (70C).

References
[1] Taborek, J., 1987, Bond Resistance and Design Temperatures for High-Finned Tubes - A
Reappraisal, Heat Transfer Eng., 8(7), pp. 2634.
[2] McHugh, S., and Chappell, S.E. 1999, Specify the right fin type for air-cooled heat exchangers, Hydrocarbon Process., Sept., pp. 6772.
[3] API, 1992, Air-cooled Heat Exchangers for General Refinery Services, API Standard 661, 3rd
Ed., April, American Petroleum Institute, Washington, DC.

Further Reading
1. Monroe, R.C., 1979, Improving Cooling Tower Fan System Efficiencies, Combustion, May,
pp. 2026.
2. Gardner, K.A., 1945, Fin Efficiency of Several Types of Straight Fins, Trans. ASME, 67, pp.
621631.

28

CHAPTER 5

Thermal Design of Single-Phase


Air-cooled Heat Exchangers
5.1 Introduction
The design of air-cooled heat exchangers comprises two distinct activities: thermal design and mechanical design. In thermal design, the heat exchanger is sized, which means
that all the principal construction parameters such as number of bays, number of bundles/bay, number of tube rows, number of tubes per row, tube OD and thickness, fin OD
and density, tube length, tube pitch, number of tube passes, and nozzle sizes are determined. In mechanical design, detailed calculations are carried out to determine the dimensions of various components such as tubesheets or plugsheets, header boxes, flanges,
etc. and a complete bill of materials and set of engineering drawings are generated. In this
book, we shall talk predominantly about thermal design.
The basic equations for tubeside and airside heat transfer and pressure drop are well
known and are presented in several books (see references). This chapter will dwell on the
application of these and other correlations for the optimum thermal design of air-cooled heat
exchangers. Before we proceed any further, let us see what the broad objectives of a thermal
designer are when he or she sets out to produce a thermal design.
An air-cooled heat exchanger, like any other heat exchanger, must satisfy the following
basic equation:

A=

Q
U (MTD)

(5.1)

where
A = heat transfer area
U = overall heat transfer coefficient
MTD = mean temperature difference
The overall heat transfer coefficient is determined as follows:
I
l
1
=
+ + r f (tubeside) + r f (air) + rw
U hair ht
where

hair = airside heat transfer coefficient


29

(5.2)

ht = tubeside heat transfer coefficient


rf = fouling resistance
rw = tube wall resistance
Since the airside heat transfer coefficient is generally much lower than the tubeside heat
transfer coefficient, it becomes necessary to use extended the surface (finned tubes) so as
to make the airside heat transfer coefficient compatible with the tubeside heat transfer
coefficient. The fin height and fin density (fins per in. or m) can be varied so as to incorporate the optimum extent of the extended surface. The more the airside heat transfer coefficient is controlling, the greater will be the extended surface required. Thus, for aircooled steam condensers or hot water coolers, which have a very high tubeside heat
transfer coefficient, the airside will be highly controlling and the designer should therefore use the highest fin density (11 fins per in. or 433 fins per m) and 5/8 in. (15.875 mm)
fin height. For viscous liquid coolers, however, where the tubeside heat transfer coefficient is much lower so that the airside heat transfer coefficient is not controlling, it will
be prudent to use a lower fin density such as 57 fins per in. (197276 fins per m) and a
fin height of 0.5 in. (12.7 mm). In an extreme situation with a very low airside heat transfer coefficient, even bare tubes may be considered.

5.2 Broad Objectives of Thermal Design


The basic aims of a thermal designer are as follows:
a) Produce a thermal design that has a low overall cost; the lower, the better. The
overall cost of a heat exchanger is the sum of the initial cost and the operating
cost. The initial cost is evidently the fixed cost or the first cost of the heat exchanger. The operating cost is the sum of the pumping cost, the maintenance
cost, and the downtime cost. The maintenance cost is the sum of the cost of periodically cleaning the exchanger, the cost of any antifoulant treatment, and the
cost of any repair or replacement.
Thus, it is not enough to produce a design having a very low fixed cost if its
operating cost is high due to, say, frequent fouling and thereby the requirement
for cleaning. Designers often lose sight of the operating cost of a heat exchanger
and should always attempt to minimize fouling and also minimize pressure drop.
Now, this represents a direct conflict because as we shall see later on in the book,
the best way to minimize fouling is to maximize velocity (within limits of erosion, of course) that will directly maximize pressure drop and thereby power consumption. Obviously, then, the designer has to optimize the design so that while
the velocity is not low enough to exacerbate fouling, the pressure drop is not excessively high.
In this context, the selection of the materials of construction is very important. The materials should be good enough to permit the heat exchanger to function for the lifetime of the plant (typically 2025 years) without major repairs and
without replacement of components (such as tubes). However, the materials
should not be inordinately expensive because then the first cost of the heat exchanger will become unnecessarily high.
b) Utilize allowable pressure drops as fully as possible.
It will be easily appreciated that the higher the velocity of a given stream, the
higher will be its heat transfer coefficient. However, accompanying the high heat
30

transfer coefficient will be a high pressure drop. So, while the former (high heat
transfer coefficient) will tend to reduce the first cost of the heat exchanger, the
latter (high pressure drop) will tend to increase the operating cost of the heat exchanger. Thus, a very important goal for a good thermal design is the best utilization of the allowable pressure drop. This is discussed in more detail later in this
chapter.
It sometimes so happens that the permitted pressure drop is unnecessarily
high to produce a good design and if, in such cases, the pump specifications have
not been frozen, they can be revised to take advantage of the lower (than anticipated) pressure drop. However, if the pump specifications have already been frozen, the possible saving in pumping power cannot be realized and the differential
pressure drop will just have be let down (typically) across a control valve.
c) Maintain adequate tubeside velocity to minimize fouling.
This has just been discussed above, and is treated in much more detail in Chapter 10.

5.3 Data to be Furnished for Thermal Design


Before coming to the actual thermal design of an air-cooled heat exchanger, let us take a
look at the data required for the same.
The following information must be furnished by the process licensor for the hot process
stream (wherever applicable) before thermal design can be taken up:
Flow rate
The complete requirements of vapor, liquid, steam, water, and noncondensable flow rates
must be furnished, as applicable, at both the inlet and outlet of the heat exchanger.
Inlet and outlet temperatures
Evidently, the inlet and outlet temperatures have to be specified, as they will go toward
the determination of the heat duty, and also toward the calculation of the mean temperature difference.
Heat duty
It is a good idea for the thermal engineer to corroborate the heat duty since licensors occasionally slip in this aspect. For sensible cooling services, the heat duty is simply the
product of the mass flow rate, the average specific heat, and the difference between the
inlet and the outlet temperatures. For condensing services, the total heat duty is the sum
of the sensible vapor cooling duty, the sensible liquid cooling duty, and the condensing
duty, which is the product of the amount of vapor condensed per hour and the latent heat
of condensation.
Heat release profiles
By heat release profiles are meant plots of the following variables with temperature,
wherever applicable: heat duty, weight fraction vapor, and vapor molecular weight. Evidently, for single-phase services, the last two are not applicable. Besides, the plot of heat
duty versus temperature is essentially linear so that no heat release profile is really required.
However, for any service involving phase change, heat release profiles as defined above
are a must. If the temperature difference between the inlet and the outlet is rather small, such
31

as 9 oF (5 oC) or 18 oF (10 oC), a straight-line heat duty versus temperature may be specified
since the curvature will be minimal. Heat release and other profiles will be discussed in
detail in Chapter 8.
Operating pressure
This is not really required for liquids since their properties do not vary with pressure to
any significant extent. However, it is required for gases and vapors since their properties,
particularly gas density, vary with pressure. However, if the physical properties are furnished, the operating pressure is no longer required for single-phase gas and condensing
vapor streams.
Allowable tubeside pressure drop
This is a very important parameter for air-cooled heat exchanger design and the process
licensor should be aware of the significance of the same for thermal design. The higher
the pressure drop, the higher will be the heat transfer coefficient and thereby the lower
the heat transfer area and fixed cost. However, the operating cost will be higher. Consequently, the allowable pressure drop represents the optimum balance between fixed cost
and operating cost of a heat exchanger such that the total cost is minimal. Generally, for
liquids, a value of 710 psi (0.50.7 kg/cm2) is permitted per shell. A higher value is usually warranted for viscous liquids, especially if routed through the tubeside. For gases, the
usually allowed value is 0.72.8 psi (0.050.2 kg/cm2), a very typical value being 1.4 psi
(0.1 kg/cm2).
It must be stated here that whereas typical values are generally applicable, specific
instances must be investigated more thoroughly. For example, if it is found that the
allowable pressure drop for a particular stream represents a severe constraint in producing a
satisfactory thermal design, the effect of a higher allowable pressure must be examined to
arrive at the optimum design based on minimum total cost.
It may be stated here that this aspect is very important for good thermal design of heat
exchangers and indeed for any good design: the designer must not follow the beaten path but
always question the various parameters specified and examine alternatives. It should be
remembered that some of the parameters specified are not really sacrosanct but are only
based on hereditary engineering practice. A special situation may call for special measures.
The author has found that it always helps to keep asking oneself: Is there not a better way
of doing this? Why don't I see what happens if I change this parameter?
Fouling resistance
This is another extremely important parameter and one that is unfortunately based more
on experience than fundamental understanding, thanks to the complexity of the phenomenon. If the fouling resistance of a particular stream is not furnished, the heat exchanger designer should adopt the same from TEMA standards or from past operating
experience. This subject is discussed in far greater detail in Chapter 10.
Physical properties
Principally viscosity, thermal conductivity, density, and specific heat, preferably at both
inlet and outlet temperatures. Viscosity data must be supplied at inlet and outlet temperatures, especially for liquids, since the variation with temperature is considerable and irregular (neither linear nor semilog nor log-log). Additional properties required are latent
32

heat and surface tension for condensing services. Physical properties are discussed at
length in Chapter 8.
Line sizes
It is desirable to match nozzle sizes with line sizes since no expander or reducer will then
be required. However, criteria for nozzle sizing (velocity and v2) are usually more stringent for heat exchanger nozzles than for lines. Nozzle sizing is based on pressure drop,
which in turn is based on expansion and contraction losses, whereas line sizing is based
on line pressure drop, which is dependent on velocity and the length of the line. Consequently, nozzle sizes are sometimes required to be one size (or even more in exceptional
circumstances) larger than the corresponding line sizes. This is especially true for small
line sizes where the change in flow area from one pipe size to the next is quite considerable.
Tube size
By tube size is meant tube OD, thickness, and length. As per the API 661 Standards, 1 in.
(25.4 mm) is the smallest OD of tubes to be used in air-cooled heat exchangers. This is
somewhat surprising, considering that shell-and-tube heat exchangers in the same chemical process industries can be designed and built with 0.75 in (19.05 mm) tubes. This is
one of the mysteries that the author has never been able to unravel!
Since air-cooled heat exchangers are rather large and occupy large plot areas, they are
invariably located over pipe racks. In such situations, the tube length is usually 1.64 ft (0.5
m) greater than the pipe-rack width for reasons of mechanical convenience. Thus, the tube
length of an air-cooled heat exchanger gets fixed by the pipe-rack width of the unit in which
it is going to be located.
Occasionally, an air-cooled heat exchanger is located on a technological platform and
in such situations its tube length can be optimized so as to yield the most cost-effective
design.
The selection of the tube length of an air-cooled heat exchanger is discussed in more
detail in Section 7.1.
Materials of construction (MOC)
The materials of construction of the tubes, tubesheets, and headers should be specified by
the process licensor. Since a process or other stream entering an air-cooled heat exchanger is usually not at an elevated temperature, the most common tube material is carbon steel. In a shell-and-tube heat exchanger, the materials of construction of the components that face both the shellside and tubeside fluids (for example, tubes and floatinghead covers) have to be so selected so that they can withstand both the shellside and tubeside fluids. Often, a material that is suitable for the tubeside fluid is not suitable for the
shellside fluid. In such cases, material selection can become difficult and finally end with
a very expensive material. In an air-cooled heat exchanger, however, the situation is
much simpler since air is not corrosive and material selection is solely on the basis of the
tubeside fluid. The fin material universally used for air-cooled heat exchangers is aluminum since it exhibits the most favorable thermal conductivity-to-cost ratio. Copper has a
thermal conductivity far greater than that of aluminum but its cost is disproportionately
higher.

33

Corrosion allowance
Corrosion allowance for the various pressure parts have to be specified by the process
licensor. No discrete corrosion allowance is applied on tubes since the standard tube
thicknesses recommended by TEMA already incorporate a corrosion allowance.
Special considerations
All pertinent requirements, such as cycling, upset conditions, alternate cases of operation,
and whether operation is continuous or intermittent, should all be specified so that all
demands made on a heat exchanger during its expected lifetime can be taken into account
for design.
Multiple operating cases
Sometimes air-cooled heat exchangers have to be designed for two or more operating
cases corresponding to various feedstocks or different modes of plant operation. The alternate cases of operation have to be assessed carefully since a single case need not represent the controlling case on all counts. Heat transfer area, tubeside pressure drop, and fan
power consumption are the broad controlling parameters. More often than not, a single
case is controlling from all points of consideration. Occasionally, it may happen that one
case is controlling for the required heat transfer area and another case for the tubeside
pressure drop.
In such situations, it is advisable to run all cases to ensure satisfactory operation in each
condition. In fact, it is advised that unless the designer is very experienced, all cases are run
to ensure that no error of judgment is committed in identifying the controlling case(s). The
most common mistake occurs when the highest heat duty case is assumed to be the
controlling case. The highest heat duty need not be the controlling case if the mean
temperature difference for that case is disproportionately higher than for another apparently
noncontrolling case. To illustrate this vividly, consider an air-cooled heat exchanger that has
to be designed for the following two cases:
Case 1: Heat duty = 5.5 M kcal/h
Case 2: Heat duty = 4.9 M kcal/h
However, while carrying out the design, it is found that the mean temperature difference
(MTD) and overall heat transfer coefficient (U) are as follows:
Case 1: U = 66 Btu/h ft2 F (322 kcal/h m2 C/kcal), MTD = 117F (65C)
Case 2: U = 62.9 Btu/h ft2 F (307 kcal/h m2 C/kcal), MTD = 99F (55C)
From the above values of heat duty, overall heat transfer coefficient, and mean
temperature difference, the required heat transfer area works out to be 262.8 m2 and 290.2 m2
for Cases 1 and 2, respectively. Thus, although its heat duty is lower, Case 2 is the
controlling case for heat transfer area because its MTD is even lower.
Here, it may be noted that when there are multiple cases of operation, the ratio of heat
duty to MTD is a good indicator of which case is controlling since the overall heat transfer
coefficient usually will not vary considerably unless the flow rates and/or physical properties
are rather different. It is therefore advised that unless the controlling case is clearly visible, it
is best to run all the cases to make sure that the heat exchanger will work satisfactorily for all
the cases.
Another possible error in identifying the controlling case is in the wrong determination
of the MTD by basing it on only the inlet and outlet temperatures of a condensing process
stream. The heat release profile must be considered for an authentic calculation of the MTD.
34

Of course, if the condensation takes place across a small temperature range so that the heat
release profile is essentially linear, the error will be negligible.

5.4 Tubeside Calculations


5.4.1 Effects of tubeside velocity
Both the heat transfer coefficient and pressure drop vary with tubeside velocity, but the
latter varies much more strongly than the former. A good design will make the best use of
the allowable pressure drop since this will yield the highest heat transfer coefficient and
thereby the lowest heat transfer area and cheapest design.
Let us now see how tubeside velocity is increased. If the entire tubeside fluid were to
flow through all the tubes (which is called a single tube pass), it would result in a certain
velocity. Usually, this velocity is unacceptably low since it yields a very low heat transfer
coefficient and therefore has to be increased. By incorporating pass partition plates (and
corresponding gaskets) in the headers, the tubeside fluid is made to flow several times
through a fraction of the total number of tubes at a time. Thus, if an air-cooled heat
exchanger tube bundle has 200 tubes and 2 passes, the fluid flows through 100 tubes at a
time, and twice along the exchanger: first from left to right and then from right to left (or
vice versa). The velocity will evidently be twice that of the velocity if there were only one
pass.
The number of tube passes is usually 1, 2, 4, 6, 8, and so on. Evidently, an air-cooled
heat exchanger with U-tubes can only have an even number of tube passes. An odd number
of passes is usually avoided since the outlet piping has to be taken back all along the length
of the air-cooled heat exchanger. However, in pressure-drop limiting cases, it often becomes
unavoidable to have a single tube pass. Therefore, the number of tube passes is either one or
an even number.
A very large number of tube passes leads to difficulties in fabrication, especially when
the number of tube rows is small. However, in the vast majority of cases, the number of tube
passes is not required to be greater than six. It will be evident that the number of tube passes
will be high when the number of tubes is rather large and/or when the tubeside mass flow
rate is rather low. This situation is exacerbated when the tubeside viscosity is high, because
then a higher mass velocity is required to obtain a satisfactory Reynolds number.
5.4.2 Tubeside heat transfer coefficient
The tubeside heat transfer coefficient is a function of the Reynolds number, the Prandtl
number, and the tube diameter. The Reynolds number is given as

DG
,

where

D = tube ID
G = mass velocity
= viscosity
The Prandtl number is given as

35

c
,
k

where
c = specific heat
= viscosity
k = thermal conductivity

If these are broken down into the fundamental parameters, they are the physical properties (namely, viscosity, thermal conductivity, and specific heat), tube diameter, and, very
importantly, mass velocity.
The variation in liquid viscosity being quite considerable, this physical property has the
most dramatic effect on the heat transfer coefficient.
Let us take a look at the fundamental equation for turbulent heat transfer inside tubes:
Nu = 0.027 (Re)0.8 (Pr)0.33

(5.3)

or
0.8

DG c
hD
= 0.027

k
k

0.33

(5.4)

or
0.8

DG c
hD
= 0.027

k
k

0.33

k

D

(5.5)

Thus, the heat transfer coefficient varies to the 0.8 power of mass velocity. It also varies
to the -0.2 power of tube ID, mass velocity remaining constant. That is to say, for the
same total tubeside flow area per pass (which will yield the same mass velocity), a
smaller tube size will yield a higher heat transfer coefficient.
As regards the variation of heat transfer coefficient with fluid viscosity, it will be seen
that there are two opposing tendencies; one in which viscosity is a parameter of the
Reynolds number, and the other in which it is a parameter of Prandtl number.
Thus,

h = ()0.330.8
() 0.47
which is to say the heat transfer coefficient is inversely proportional to the 0.47 power of
viscosity.
An interesting aspect to observe here is that even for the same Reynolds number, two
liquids of fairly different viscosities will yield fairly different heat transfer coefficients due to
their different Prandtl numbersthe liquid having the higher viscosity will have the higher
Prandtl number and thereby the higher heat transfer coefficient. The Reynolds numbers can
be the same if the higher viscosity stream has a correspondingly higher mass flow rate.
It will also be seen that the heat transfer coefficient is directly proportional to the 0.67
exponent of the thermal conductivity.
Curiously, it appears that the heat transfer coefficient is proportional to the 0.33
36

exponent of the specific heat. While this is true, it should also be realized that the heat duty
will increase directly with specific heat so that all other things remaining the same, a higher
specific heat will result in a considerably higher heat duty than heat transfer coefficient!
The above leads to some very interesting generalities of heat transfer. A high thermal
conductivity promotes a high heat transfer coefficient. Thus, water has an extremely high
heat transfer coefficient [thermal conductivity around 0.37 Btu/h ft oF (0.55 kcal/h m oC)],
followed by hydrocarbon liquids [thermal conductivity between 0.12 and 0.18 Btu/h ft oF
(0.08 and 0.12 kcal/h m oC)], and then followed by hydrocarbon gases [thermal conductivity
between 0.03 and 0.045 Btu/h ft oF (0.02 and 0.03 kcal/h m oC)]. Liquid ammonia has a
thermal conductivity in between that of water and hydrocarbon liquids.
Hydrogen is an unusual gas because it has an exceptionally high thermal conductivity
(same as or even greater than that of hydrocarbon liquids), as well as an exceptionally high
specific heat; hence, its heat transfer coefficient is in the range of hydrocarbon liquids.
Typical heat transfer coefficients of various fluids are as follows:
Cooling water: 1200 Btu/h ft2 oF (6000 kcal/h m2 oC)
Hydrocarbon liquids: 50260 Btu/h ft2 oF (2501300 kcal/h m2 oC)
Hydrocarbon gases: 1050 Btu/h ft2 oF (50500 kcal/h m2 oC)
The rather large variation in the heat transfer coefficient of hydrocarbon liquids is due to
the rather large variation in their viscosity, from less than 0.1 cp for ethylene and propylene
to greater than 500 cp and even more for heavy hydrocarbon liquids such as vacuum residue
and bitumen.
The large variation in the heat transfer coefficient of hydrocarbon gases is attributable to
the large variation in operating pressure. With increase in operating pressure, gas density
increases. Since pressure drop is (a) directly proportional to the square of the mass velocity
and (b) inversely proportional to the density, a higher mass velocity can be maintained when
the density is greater, for the same pressure drop. This higher mass velocity translates into a
higher heat transfer coefficient. Let us demonstrate this with the help of two case studies,
one for a gas at a high pressure and the other for a gas at a low pressure.
A more elaborate list of typical overall heat transfer coefficients is sourced from Ref. [1]
and presented toward the end of this chapter in Tables 5.6 ac.

CASE STUDY 5.1: HIGH-PRESSURE GAS COOLER


Consider the high-pressure gas cooler service elaborated in Table 5.1a. The gas pressure
is 886 psia or 62.3 kg/cm2 abs. At this pressure, the gas inlet density is 3.06 lb/ft3 (49.1
kg/m3) and the outlet density is 3.38 lb/ft3 (54.2 kg/m3), which are quite high.
The design was to be carried out using 0.984 in. (25 mm) OD 0.098 in. (2.5 mm) thick
34.5 ft (10.5 m) long carbon steel tubes, having fins of aluminum.
Based on the above process data, a thermal design was prepared, the principal
construction and performance parameters of which are detailed in Table 5.1b. The following
important features may be noted:
1) By virtue of the high gas density, the tubeside gas mass is very high: 92.7 lbf/s ft2
(451.9 kg/s m2).
2) As a result, the midpoint tubeside Reynolds number is also very high, at 724,953.
Consequently, the tubeside heat transfer coefficient is also very high, at 265.4
Btu/hft2F (1296 kcal/h m2C/kcal). This is unusually high for gases and is possible only because of the high pressure and density, which makes it possible to sus37

Table 5.1a: Principal process parameters of air-cooled heat exchanger for Case Study 5.1
1. Process stream
2. Flow rate, lb/h (kg/h)
3. Temperature in/out, F (C)
4.Operating pressure, psia (kg/cm2 a)
5. Allowable pressure drop, psi (kg/cm2)
6. Heat duty, MM Btu/h (MM kcal/h)
7. Density in/out, lb/ft3 (kg/m3)
8. Viscosity in/out, cp
9.Thermal conductivity in/out, Btu/h ft F (kcal/h m C)
10. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
11. Design air temperature, F (C)

Hydrocarbon gas mixture


440,900 (200,000)
199.4 (93)/149 (65)
886 (62.3)
3.6 (0.25)
13.1 (3.3)
3.06 (49.1)/3.38 (54.2)
0.0132/0.012
0.0238 (0.0354)/0.0214 (0.0318)
0.00098(0.0002)
104 (40)

Table 5.1b: Principal construction and performance parameters of air-cooled heat exchanger for
Case Study 5.1
1. No. of bays
2. No. of bundles per bay
3. Tube OD thk length, in. (mm)
4. Fin height, in. (mm) fin density, fins/in. (fins/m)
5. No. of tubes per row no. of rows tube pitch, in. (mm)
6. No. of tube passes
7. Total bare tube area, ft2 (m2)
8. Fans per section fan diameter, ft (m)
9. Total air flow rate, lb/h (kg/h)
10. Air outlet temperature, F (C)
11. Airside midpoint velocity, ft/s (m/s)
12. Airside pressure drop, in. WC (mm WC)
13. Individual fan power, HP (kW)
14. Total fan power, HP (kW)
15. Connections per bundle, in. (mm)
16. Tubeside mass velocity, lbf/s ft2 (kg/s m2)
17. Tubeside Reynolds no., inlet/midpoint/outlet
18. Heat transfer coefficient, Btu/h ft2 F/ Tubeside
(kcal/h m2 C)
Airside
Overall
19. Thermal resistance, %
Airside
Tubeside
Fouling
19. Tubeside pressure drop, psi (kg/cm2)
20. Overdesign, %

1
2
0.984 (25) 0.098 (2.5) 413 (10500)
0.63 (16) 11 (433)
46 8 2.36 (60)
1
3363 (312.5)
2 13 (3.963)
1,543,000 (700,000)
139.3 (59.6)
19.5 (5.93)
0.37 (9.37)
17.3 (12.9)
34.6 (25.8)
In: 2 10 (250)
Out: 1 10 (250)
92.7 (451.9)
686,756 / 724,953 / 755,431
265.4 (1296)
181.9 (888)
92.6 (452.3)
50.97
34.91
11.31
3.3 (0.23)
15.3

tain a high tubeside mass velocity. We have just seen that a typical tubeside heat
transfer coefficient for hydrocarbon gases is in the range of 1050 Btu/h ft2 oF
(50500 kcal/h m2 oC).
3) There is another reason why the tubeside heat transfer coefficient is so high, and
that is the high tubeside flow rate, as a result of which only one tube pass is required to obtain a very high tubeside velocity. With a single tube pass, it is evidently possible to obtain a much higher tubeside velocity than with two tube
38

Table 5.2a: Principal process parameters of air-cooled heat exchanger for Case Study
5.2
1. Process stream
2. Flow rate, lb/h (kg/h)
3. Temperature in/out, F (C)
4.Operating pressure, psia (kg/cm2 a)
5. Allowable pressure drop, psi (kg/cm2)
6. Heat duty, MM Btu/h (MM kcal/h)
7. Density in/out, lb/ft3 (kg/m3)
8. Viscosity in/out, cp
9. Thermal conductivity in/out, Btu/h ft F (kcal/h m C)
10. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
11. Design air temperature, F (C)

Nitrogen gas
518,100 (235,000)
248 (120)/131 (55)
142 (10)
1.4 (0.1)
15.32 (3.86)
0.525 (8.42)/0.624 (10.0)
0.0217/0.0191
0.018(0.027)/0.0157(0.0234)
0.00098 (0.0002)
104 (40)

passes. This is because with two tube passes, the length of travel is double that
with a single tube passing. Since the tubeside pressure drop varies directly with
the length of travel and to the 0.8 power of mass velocity, a lower number of tube
passes will translate into a higher mass velocity for the same pressure drop. The
higher mass velocity will in turn translate into a higher tubeside heat transfer coefficient.
4) The thermal resistance on the tubeside is 34.91% while that on the airside is
about 51%. As we will see later in Chapter 7, a higher fin height is usually favorable for cases where the airside thermal resistance is controlling.
5) The fan power consumption is only 17.3 HP (12.9 kW). This is because in the
present design, the tubeside pressure drop is very close to the permissible value
and, hence, any reduction in the number of tubes by virtue of a higher airside
heat transfer coefficient is not possible since we have to employ a fixed tube
length of 34.5 ft (10.5 m). Therefore, it is prudent to minimize fan power consumption so as to minimize the operating cost.

CASE STUDY 5.2: LOW-PRESSURE GAS COOLER


Let us now take a look at another gas cooler. This time, a nitrogen compressor aftercooler
operating at a much lower pressure (142 psia or 10 atma). The principal process parameters of this service are indicated in Table 5.2a. Carbon steel tubes with aluminum fins
were to be used, the preferred tube size being 1 in. (25.4 mm) OD 12 BWG (2.77 mm)
thick 29.5 ft (9.0 m) long.
While carrying out the thermal design, it was found that for the given nitrogen flow rate
and with the given length of tubes, it was not possible to satisfy the permissible tubeside
pressure drop while using 1in. (25.4 mm) tubes, even with a single tube pass. Hence, the
tube size was increased to 1.25 in. (31.75 mm) 12 BWG (2.77 mm) thick, and since the
tubeside thermal resistance was expected to be controlling, a fin height of 0.5 in. (12.7 mm)
was used. The principal construction and performance parameters of the thermal design that
finally emerged are indicated in Table 5.2b.
It will be seen that the tubeside heat transfer coefficient is far lower than what it was in
the previous case study, just 45.8 Btu/hft2F (223.7 kcal/h m2C/kcal) as against 265.4
Btu/hft2F (1296 kcal/h m2C/kcal) in Case Study 5.1. This is directly attributable to the
much lower gas densities in this case, 0.525 kg/m3 (8.42 lb/ft3) at the inlet and 0.624 kg/m3
39

Table 5.2b: Principal construction and performance parameters of air-cooled heat exchanger for
Case Study 5.2
1. No. of bays
2. No. of bundles per bay
3. Tube OD thk length, in. (mm)
4. Fin height, in. (mm) fin density, fins/in. (fins/m)
5. No. of tubes per row no. of rows tube pitch, in. (mm)
6. No. of tube passes
7. Total bare tube area, ft2 (m2)
8. Fans per section fan diameter, ft (m)
9. Total air flow rate, lb/h (kg/h)
10. Air outlet temperature, F (C)
11. Airside midpoint velocity, ft/s (m/s)
12. Airside pressure drop, in. WC (mm WC)
13. Individual fan power, HP (kW)
14. Total fan power, HP (kW)
15. Connections per bundle, in. (mm)
16. Tubeside mass velocity, lbf/s ft2 (kg/s m2)
17. Tubeside Reynolds no., inlet/midpoint/outlet
18. Heat transfer coefficient,
Tubeside
Btu/h ft2 F/ (kcal/h m2 C)
Airside
Overall
19. Tubeside pr. drop, allow/calc., psi (kg/cm2)
20. Overdesign, %

2
2
1.25 (31.75) 0.109 (2.769) 354 (9000)
0.5 (12.7) 11 (433)
35 7 2.56 (65)
1
9243 (859)
2 11 (3.353)
2,142,700 (971,900)
133.7 (56.5)
21.9 (6.67)
0.7 (17.7)
20.0 (14.9)
80.0 (59.6)
In: 1 10 (250)
Out 1 10 (250)
24.9 (121.3)
162,286
45.8 (223.7)
142.8 (697)
32.9 (160.8)
1.4 (0.1)/1.37 (0.096)
10.5

(10.0 lb/ft3) at the outlet, as against 3.06 kg/m3 (49.1 lb/ft3) at the inlet and 3.38 kg/m3 (54.2
lb/ft3) in Case Study 5.1.
As a result of the much lower gas density, the tubeside mass velocity that could be
sustained is just 24.9 lb/s ft2 (121.3 kg/s m2), as compared to 92.7 lb/s ft2 (451.9 lb/s ft2) in
the earlier case study. Hence, the tubeside Reynolds number and thereby the tubeside heat
transfer coefficients were much lower in the present example.
5.4.3 Tubeside pressure drop
Thus far, we have discussed the variation of the heat transfer coefficient with the variation of tube ID, mass velocity, and physical properties. For a given fluid, the mass velocity exerts a very strong influence on the heat transfer coefficient. Whereas the tubeside
heat transfer coefficient varies to the 0.8 exponent of the tubeside mass velocity in turbulent flow, the tubeside pressure drop varies with the square of the mass velocity.
Further, the tubeside film resistance represents only a fraction of the total resistance to
heat transfer (the others being tubeside fouling resistance, airside film resistance, airside
fouling resistance, and tube wall metal resistance), so that for an increase in the tubeside
mass velocity, the increase in the overall heat transfer coefficient will be even less.
Consequently, there will be an optimum mass velocity above which it will be wasteful to
increase mass velocity any further. In other words, there is an optimum velocity for the
conversion of pressure drop to heat transfer.
Very high velocities also lead to erosion. However, the pressure drop limitation usually
becomes controlling much before erosive velocities are attained. The minimum recommended liquid velocity inside tubes is 3.3 ft/s (1.0 m/s) while the maximum is 8.29.8 ft/s
40

(2.53.0 m/s). In the case of highly viscous liquids, however, it is often not possible to
achieve a velocity of even 3.3 ft/s (1.0 m/s) unless a very high pressure drop is permitted.
Furthermore, the pressure drop is proportional to the square of the velocity and the total
length of travel. Thus, when the number of passes is increased from two to four to increase
the tubeside velocity, both the tubeside velocity and the length of travel are doubled, so that
the pressure drop increases by (2)2 2, or 8 times. Similarly, when the number of passes is
increased from four to six, the pressure drop increases by (1.5)2 1.5 or 3.375 times. Thus,
when the number of tube passes is increased for a given number of tubes and a given
tubeside flow rate, the pressure drop increases to the third power of this increase. In actual
practice, this increase is somewhat less since the friction factor reduces at the higher
Reynolds number, so that the exponent may be considered to be approximately 2.8 instead
of 3.
It is easily appreciated that an increase in tubeside pressure drop with increase in the
number of tube passes is a rather steep change. Consequently, it often happens that with two
passes and a given number of tubes, the pressure drop is much lower than the allowable
value; but with four passes, it exceeds the allowable value. In such circumstances, the tube
diameter and/or the tube length may be varied so that the allowable pressure drop is fully
utilized or at least, utilized to a very large extent. This will evidently yield the highest
tubeside heat transfer coefficient in a given situation.
The following tube diameters are usually used for air-cooled heat exchangers in the
chemical process industries: 1 in. (25 mm), 11/4 in. (31 mm), 11/2 in. (38 mm), and 2 in. (50
mm). The smallest size recommended by API 661 is 1 in. (25 mm) and this is by far the
most common. It is curious that although in. (19 mm) OD tubes are acceptable to TEMA
(Tubular Exchanger Manufacturers Association) for shell-and-tube heat exchangers, API
(American Petroleum Institute) do not permit the use of tubes smaller than 1 in. (25 mm)
OD for air-cooled heat exchangers for use in the same chemical process industries.
Apparently, the limitation on the smallest tube size, whether for shell-and-tube heat
exchangers or for air-cooled heat exchangers, is on considerations of fouling: the adverse
effect of fouling is greater in smaller diameter tubes than in larger diameter ones.
Occasionally, a rather high overdesign may have to accept for an air-cooled heat
exchanger only to satisfy the allowable tubeside pressure drop. That is to say, the number of
tubes is increased not to incorporate additional heat transfer area but just to reduce the
velocity and thereby the tubeside pressure drop. Evidently, this will lead to an unnecessary
overdesign. Such a condition is referred to as a pressure drop limiting design.
However, if such a stream goes to another heat exchanger subsequently, such as a trim
cooler, the extra heat duty that will be transferred due to the oversizing of the air-cooled heat
exchanger should be taken advantage of, and only the correspondingly lower heat duty
considered for the trim cooler.
Consider a hot stream that is to be cooled from 248oF (120oC) to 176 oF (80oC) by air in
an air-cooled heat exchanger, representing a heat duty of 11.91 106 Btu/h (3.0 106
kcal/h), and from 176oF (80oC) to 104oC (40oC) by cooling water in a trim cooler (shell-andtube heat exchanger), representing a heat duty of 11.19 106 Btu/h (2.82 106 kcal/h). The
total allowable pressure drop for the hot stream is 9.8 psi (0.7 kg/cm2), out of which 5.6 psi
(0.4 kg/cm2) is allocated to the air-cooled heat exchanger and 4.2 psi (0.3 kg/cm2) to the trim
cooler. While carrying out the design of the air-cooled heat exchanger, it is found that in
order to restrict the hot stream pressure drop to 5.6 psi (0.4 kg/cm2), there is an incidental
overdesign of 20%, which translates to a heat duty of 12.86 106 Btu/h (3.24 106 kcal/h),
corresponding to an outlet temperature of 170.2oF (76.8oC). Since the balance heat duty is
41

Table 5.3a: Principal process parameters of air-cooled heat exchanger for Case
Study 5.3
1. Process stream
2. Flow rate, lb/h (kg/h)
3. Temperature in/out, F (C)
4. Allowable pressure drop, psi (kg/cm2)
5. Heat duty, MM Btu/h (MM kcal/h)
6. Viscosity in/out, cp
7. Density in/out, lb/ft3 (kg/m3)
8. Thermal conductivity in/out, Btu/h ft F (kcal/h m C)
9. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
10. Design air temperature, F (C)

Heavy hydrocarbon liquid


198,414 (90,000)
356 (180)/158 (70)
7.1 (0.5)
22.0 (5.545)
4.2/14.0
51.59 (820)/55.56 (890)
0.0605 (0.09)/0.0672 (0.1)
0.002929 (0.0006)
107.6 (42)

(11.91 + 11.19 12.86) or 10.24 106 Btu/h [(3.0 + 2.82 3.24) or 2.58 106 kcal/h], the
trim cooler needs to be designed only for this heat duty and not for 11.19 106 Btu/h (2.82
106 kcal/h). Of course, the desired oversurfacing can be incorporated in both the air-cooled
heat exchanger and trim cooler.
In this context, it is important to realize that the total pressure drop for a given stream
has to be met. The distribution of pressure drop in the various heat exchangers for a given
stream in a particular circuit may be varied as found best to obtain good heat transfer in all
the heat exchangers.
For example, consider the total allowable pressure drop for a distillation column
overhead stream to be 4.2 psi (0.3 kg/cm2). This overhead stream has to flow through an aircooled condenser and subsequently through a water-cooled condenser. If a proper air-cooler
design can be made with a tubeside pressure drop of 1.7 psi (0.12 kg/cm2), the balance
pressure drop of (4.21.7) or 2.5 psi [(0.30.12) or 0.18 kg/cm2] is available for the trim
condenser.
5.4.4 Increased tubeside pressure drop
Occasionally, a very low allowable tubeside pressure drop is specified by a process licensor. This can penalize the air-cooled heat exchanger design considerably by requiring an
inordinately high heat transfer surface area to be provided. For gases and condensers, the
allowable pressure drop is generally between 0.7 and 2.84 psi (0.05 and 0.2 kg/cm2), depending on the operating pressurethe lower the operating pressure, the lower the allowable pressure drop. For liquids, the allowable pressure drop is generally 7 to 10 psi
(0.5 to 0.7 kg/cm2). However, if the liquid viscosity is high, a higher pressure drop is
warranted for an optimum design. The higher the viscosity is, the higher the pressure
drop that should be permitted. This is borne out by the following two case studies.

CASE STUDY 5.3: EFFECT OF ALLOWABLE TUBESIDE


PRESSURE DROPFIRST STUDY
An air-cooled heat exchanger was to be designed to cool a heavy hydrocarbon liquid. The
principal process parameters are elaborated in Table 5.3a.
Carbon steel tubes of 0.984 in. (25 mm) OD and 0.098 in. (2.5 mm) thickness were to be
used. Since the ACHE was to be located on a pipe rack of 39.4 ft (12 m) width, the tube
length had to be 41.0 ft (12.5 m). The design ambient temperature was 107.6F (42C).
Since the liquid viscosity was rather high, the allowable pressured drop of 7.1 psi (0.5
42

Table 5.3b: Principal construction and performance parameters of air-cooled heat exchanger for Case
Study 5.3

1. No. of bays
2. No. of bundles per bay
3. Tube OD thick length, in. (mm)
4. Fin height, in. (mm) fin density, fins/in. (fins/m)
5. No. of tubes per row no. of rows tube pitch, in. (mm)
6. No. of tube passes
7. Total bare tube area, ft2 (m2)
8. Fans per section fan diameter, ft (m)
9. Total air flow rate, lb/h (kg/h)
10. Air outlet temperature, F (C)
11. Airside velocity, ft/s (m/s)
12. Airside pressure drop, in. WC (mm WC)
13. Individual fan power, HP (kW)
14. Total fan power, HP (kW)
15. Connections per bundle, in. (mm)
16. Tubeside velocity, ft/s (m/s)
17. Tubeside Reynolds no., inlet/mmidpoint/outlet
18. Heat transfer coefficient,
Tubeside
Btu/h ft2 F/ (kcal/h m2 C)
Airside
Overall
19. Tubeside pressure drop, psi (kg/cm2)
20. Overdesign, %

Allowable tubeside pressure drop


7.1 psi (0.5 g/cm2)
20.8 (1.46 g/cm2)
3
2
2
2
0.984 (25) 0.098 (2.5) 492 (12500)
0.49 (12.5) 9 (354)
46 8 2.36 (60)
50 8 2.36 (60)
8
12
22,8000 (2119)
16,527 (1536)
2 14 (4.27)
2 14 (4.27)
5,081,300 (2,304,850)
2,822,300 (1,280,200)
125.6 (52)
140 (60)
17.32 (5.28)
13.5 (4.11)
0.45 (11.3)
0.3 (7.6)
21.6 (16.1)
12.2 (9.1)
129.6 (96.6)
48.8 (36.4)
In: 1 2 (50)
In: 1 3 (75)
Out 1 2 (50)
Out 1 3 (75)
1.08 (0.33)
2.3 (0.7)
1372/596/412
2559/1497/861
11.0 (53.6)
14.1 (69)
124.5 (608)
112.2 (548)
9.7 (47.3)
12.0 (58.4)
6.54 (0.46)
20.8 (1.46)
20.2
22.7

kg/cm2) was inordinately low. However, this condition was accepted and a thermal design
was prepared, as shown in Table 5.3b.
Due to the low allowable tubeside pressure drop, the tubeside heat transfer coefficient is
only 11 Btu/h ft2F (53.6 kcal/h m2C), which represents 88.4% of the total heat transfer
resistance. The tubeside velocity is only 1.08 ft/s (0.33 m/s), which will result in heavy
fouling. Also, it may be noted that since the airside heat transfer coefficient is not
controlling, a fin density of 9 fins per in. (354 fins per m) has been used, and not the
common density of 11 fins per in. (433 fins per m).
In an attempt to produce a more economical design with a higher tubeside velocity and
thereby lower fouling, an allowable tubeside pressure drop of 42.7 psi (3.0 kg/cm2) was
sought from the process licensor. However, only 21.3 psi (1.5 kg/cm2) was granted. The
revised design with this increased tubeside pressure drop is also elaborated in Table 5.3b.
It will be noted that in the revised design, the tubeside velocity increased from 1.08 fps
(0.33 m/s) to 2.3 fps (0.7 m/s). This results in less fouling inside the tubes. As a result of the
increase in tubeside velocity, the tubeside heat transfer coefficient increased from 11 Btu/h
ft2F (53.6 kcal/h m2C) to 14.1 Btu/h ft2F (69.0 kcal/h m2C), and the overall heat transfer
coefficient from 9.7 Btu/h ft2F (47.4 kcal/h m2C) to 12.0 Btu/h ft2F (58.4 kcal/h m2C).
Consequently, the number of sections could be reduced from three to two and the overall
bare tube heat transfer area from 22,800 ft2 (2119 m2) to 16,527 ft2 (1536 m2), thus resulting
in a considerable reduction in the initial cost and the plot area of the unit.
Also, the power consumption of each fan decreased from 21.6 HP (16.1 kW) to 12.2 HP
43

Table 5.4a: Principal process parameters of air-cooled heat exchanger for


Case Study 5.4
1. Process stream
2. Flow rate, lb/h (kg/h)
3. Temperature in/out, F (C)
4. Allowable tubeside pressure drop, psi (kg/cm2)
5. Heat duty, MM Btu/h (MM kcal/h)
6. Viscosity in/out, cp
7. Density in/out, lb/ft3 (kg/m3)
8. Thermal conductivity in/out, Btu/h ft F (kcal/h m C)
9. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
10. Design air temperature, F (C)

Heavy hydrocarbon liquid


652,300 (295,900)
255.7 (124.3)/194 (90)
10.7 (0.75)
21.5 (5.422)
11.3/15.0
51.8 (830)/52.4 (840)
0.05(0.075)/0.053(0.079)
0.0039 (0.0008)
107.6 (42)

(9.1 kW) and since there were fewer fans, the total power consumption decreased
substantially from 129.6 HP (96.6 kW) to 48.8 HP (36.4 kW). Therefore, while the pumping
power required for the process stream increased, the total fan power came down
significantly.
It will be noticed that with the use of a higher tubeside velocity, the inlet, midpoint, and
outlet Reynolds number went up from 1372, 596, and 412 to 2559, 1497, and 861,
respectively. Since a Reynolds number of 2100 is the upper limit of laminar flow, we have
not been able to get out of laminar flow except for in only a small region at the inlet of the
air-cooled heat exchanger. Consequently, the tubeside heat transfer coefficient has not really
increased as much as we would have desired. This is because the higher tubeside pressure
drop sought was not granted.
Comment
If a sufficiently high tubeside pressure drop can be permitted, it is possible to move totally out of laminar flow and derive a substantial improvement in the tubeside heat transfer coefficient as well as a substantial reduction in tubeside fouling. The following case
study will demonstrate this.

CASE STUDY 5.4: EFFECT OF ALLOWABLE TUBESIDE


PRESSURE DROPSECOND STUDY
An air-cooled heat exchanger was to be designed to cool a heavy hydrocarbon liquid. The
principal process parameters are elaborated in Table 5.4a.
Carbon steel tubes of 0.984 in. (25 mm) OD and 0.098 in. (2.5 mm) thickness were to be
used. Since the air-cooled heat exchanger was to be located on a pipe rack of 32.8 ft (10 m)
width, the tube length had to be 34.5 ft (10.5 m). The design ambient temperature was
107.6F (42C).
Since the liquid viscosity was rather high, the allowable pressured drop of 10.7 psi (0.75
kg/cm2) was inordinately low. However, this condition was accepted and a thermal design
was prepared, as shown in Table 5.4b.
Due to the low allowable tubeside pressure drop, the tubeside heat transfer coefficient is
only 9.7 Btu/h ft2F (47.1 kcal/h m2C), which represents 85.2% of the total heat transfer
resistance. The tubeside velocity is only 2.1 ft/s (0.64 m/s), which will result in heavy
fouling.
In an attempt to produce a more economical design with a higher tubeside velocity and
44

Table 5.4b: Principal construction and performance parameters of air-cooled heat exchanger for
Case Study 5.4

1. No. of bays
2. No. of bundles per bay
3. Tube OD thick length, in. (mm)
4. Fin height, in. (mm) fin density, fins/in. (fins/m)
5. No. of tubes per row no. of rows tube pitch,
in. (mm)
6. No. of tube passes
7. Total bare tube area, ft2 (m2)
8. Fans per bay fan diameter, ft (m)
9. Total air flow rate, lb/h (kg/h)
10. Air outlet temperature, F (C)
11. Airside velocity, ft/s (m/s)
12. Airside pressure drop, in. WC (mm WC)
13. Individual fan power, HP (kW)
14. Total fan power, HP (kW)
15. Tubeside velocity, ft/s (m/s)
16. Tubeside Reynolds no., inlet/midpoint/outlet
17. Heat transfer coefficient,
Tubeside
Btu/h ft2 F/ (kcal/h m2 C)
Airside
Overall
18. Tubeside pressure drop, psi (kg/cm2)
19. Overdesign, %

Allowable pressure drop


10.7psi (0.75g/cm2)
65.4 (4.6 kg/cm2)
4
2
2
2
0.984 (25) 0.098 (2.5) 413 (10500)
0.49 (12.5) 5 (197)
46 8 2.36 (60)
45 7 2.36(60)
6
25,555 (2375)
2 13 (3.96)
5,952,400 (2,700,000)
122.7 (50.4)
17.45 (5.32)
0.36 (9.11)
16.1 (12.0)
128.8 (96.0)
2.1 (0.64)
918/795/715
9.7 (47.1)
78.2 (381.9)
8.2 (40.2)
10.4 (0.73)
6.5

8
10,943 (1017)
2 13 (3.96)
3,615,500 (1,640,000)
132.3 (55.7)
21.8 (6.64)
0.47 (12.0)
26.4 (19.7)
105.6 (78.8)
6.56 (2.0)
2755/2480/2235
27.7 (135.4)
85.6 (418)
18.9 (92.3)
65.4 (4.6)
0.01

thereby lower fouling, an allowable tubeside pressure drop of 66.8 psi (4.7 kg/cm2) was
sought from the process licensor and it was granted. The revised design with this increased
tubeside pressure drop is also elaborated in Table 5.4b.
It will be noted that in the revised design, the tubeside velocity increased from 2.1 fps
(0.64 m/s) to 6.56 fps (2.0 m/s). This results in considerably less fouling inside the tubes. As
a result of the increase in tubeside velocity, the tubeside heat transfer coefficient increased
from 9.7 Btu/h ft2F (47.1 kcal/h m2C) to 27.7 Btu/h ft2F (135.4 kcal/h m2C), and the
overall heat transfer coefficient from 8.2 Btu/h ft2F (40.2 kcal/h m2C) to 18.9 Btu/h ft2F
(92.3 kcal/h m2C). Consequently, the number of sections could be reduced from four to two
and the overall bare tube heat transfer area from 25,555 ft2 (2375 m2) to 10,943 ft2 (1017
m2), thus resulting in a considerable reduction in the initial cost and the plot area of the unit.
Although the power consumption of each fan increased from 16.1 HP (12.0 kW) to 26.4 HP
(19.7 kW), the total power consumption reduced from 128.8 HP (96.0 kW) to 105.6 HP
(78.8 kW) due to the reduction of the number of fans. Thus, while the pumping power
required for the process stream increased substantially, the total fan power came down,
albeit to a lesser extent. Thus, the total operating cost due to pumping will be higher, but the
operating cost due to fouling will be significantly lower. The fixed cost of the second design
is less than half that of the former design since there are two bays instead of four, and seven
rows of tubes instead of eight. Thus, the total (fixed plus operating) cost of the latter design
will be less.
It will be noticed that with the use of a higher tubeside velocity, the inlet, midpoint, and
outlet Reynolds number went up from 918, 795, and 715 to 2755, 2480, and 2235,
45

respectively. Since a Reynolds number of 2100 is the upper limit of laminar flow, we have
been able to get totally out of laminar flow. Consequently, the tubeside heat transfer
coefficient has increased considerably.
Comment
It must be stated here that the tubeside is fundamentally not well suited to handling viscous fluids because of the deleterious effect of the boundary layer separation. Consequently, air-cooled heat exchangers are not recommended for cooling viscous liquids, and
cooling by closed-circuit tempered water on the shellside of shell-and-tube heat exchangers is far more efficient and economical. For the same pressure drop, the heat transfer coefficient for a viscous liquid on the shellside will be higher by an order of magnitude. However, if an air-cooled heat exchanger has to be used for cooling a viscous
liquid, the use of wire-fin tube inserts should be considered. By promoting radial mixing
from the wall of the tube to the center, such inserts improve the thermal performance
considerably and yield a much higher heat transfer coefficient for the same pressure drop.
This is discussed in detail in Section 13.4.

5.5 Airside Calculations


The most widely accepted correlations for the airside heat transfer coefficient and airside
pressure drop are those resulting from the experimental work of Edwin H. Young, Dale
E. Briggs, and Ken E. Robinson at the University of Michigan, Ann Arbor [2,3].
The general correlation for heat transfer across a bank of finned tubes is as follows:
0.2

Nu = 0.134 Re

0.681

Pr

0.33

S S

h b

0.1134

(5.6)

where
Nu = Nusselt number, dimensionless
Re = Reynolds number, dimensionless
Pr = Prandtl number, dimensionless
S = fin spacing, in.
h = fin height, in.
b = fin thickness, in.
The general correlation for pressure drop across a bank of finned tubes is
R
F = 18.93Re 0.316
d

0.927

Pt

Pl

0.515

where
f = friction factor, ft2/in2
R = MTD correction factor, (T2 T1)/(t2 t1)
T2 = tubeside outlet temperature, oF
T1 = tubeside outlet temperature, oF
t2 = airside outlet temperature, oF
46

(5.7)

t1 = airside outlet temperature, oF


d = tube OD, in.
Pt = transverse tube pitch, in.
Pl = longitudinal tube pitch, in.

5.6 Mean Temperature Difference (MTD)


The MTD calculations for air-cooled heat exchangers are somewhat different from the
MTD calculations for shell-and-tube heat exchangers, since pure cross-flow is employed
in air-cooled heat exchangers. Thus, the MTD curves furnished in the TEMA standards
are not applicable to air-cooled heat exchangers. One source of cross-flow MTD determination is [4]. There are different charts for one-pass cross-flow, two-pass cross-flow, and
three-pass cross-flow. The chart for one-pass cross-flow is depicted in Fig. 5.1 and the
chart for two-pass cross flow is depicted in Fig. 5.2.
Interestingly, as the number of tube passes increases in an air-cooled heat exchanger, the
MTD also increases, and for four or more tube passes, it becomes equal to the true
countercurrent MTD determined from the four terminal temperatures.
In water-cooled heat exchangers, the cooling water outlet temperature is usually limited
to 109.4F (43C) or 113F (45C), based on considerations of scaling by reverse solubility
salts. In the case of air-cooled heat exchangers, however, there is no such limitation.
Consequently, the air flow rate and thereby its outlet temperature can vary to a large extent.

Fig. 5.1 MTD correction factors for 1-pass cross-flow (Courtesy Hudson Products Corporation,
USA)
47

When the air mass flow rate is lowered, its outlet temperature increases, thereby reducing the
MTD (mean temperature difference). Furthermore, the airside heat transfer coefficient
reduces due to the lower mass velocity, thereby reducing the overall heat transfer coefficient.
Both these effects increase the required heat transfer area. The saving is in the power
consumption, since both the lower air flow rate and consequently the airside pressure drop
are lower. The optimum thermal design is the one that will best balance these opposing
tendencies. This balance will depend on the extent to which the airside heat transfer
coefficient is controlling. Optimization of thermal design is discussed in detail in Chapter 7.

5.7 Design Ambient Temperature


The selection of an appropriate design ambient temperature is of the utmost importance.
Without realizing the consequences, many customers tend to specify the highest temperature as the design air temperature. This is an extremely conservative practice and will
result in a considerable increase in the cost of an air-cooled heat exchanger. For example,
if a plant site has a summer peak of 113F (45C) and a winter low of 35.6F (2C), the
design air temperature should not be adopted as 113F (45C) but somewhat lower, say
107.6F (42C). This practice is based on the logic that it is not prudent to penalize the
air-cooled heat exchanger design by designing it for a peak temperature that may be
prevalent for only a few hours in the whole year. The cost difference between the two
cases (113F or 45C and 107.6F or 42C) may be appreciable, especially if the MTD is
low. A common practice is to select a temperature that is not exceeded during 23% of

Fig. 5.2 MTD correction factors for two-pass cross-flow (Courtesy Hudson Products Corporation,
USA)
48

Table 5.5a: Principal process parameters of air-cooled heat exchanger


for Case Study 5.5
1. Overhead flow rate, lb/h (kg/h)
2. Temperature in/out, F (C)
3. Operating pressure, psia (kg/cm2 a)
4. Allowable pressure drop, psi (kg/cm2)
5. Weight fraction vapor, in/out
6. Heat duty, MM Btu/h (MM kcal/h)
7. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
8. Design air temperature, F (C)

451,943 (205,000)
192 (88.9)/149 (65)
25.6 (1.8)
3.0 (0.21)
1.0/0.23
100.94 (25.44)
0.001 (0.000205)
104 (40)

Table 5.5b: Principal construction and performance parameters of air-cooled heat exchanger
for Case Study 5.5
1. No. of bays
2. No. of bundles per bay
3. Tube OD thk length, in. (mm)
4. Fin height, in. (mm) fin density, fins/in. (fins/m)
5. No. of tubes per row no. of rows
6. No. of tube passes
7. Total bare tube area, ft2 (m2)
8. Fans per bay fan diameter, ft (m)
9. Total air flow rate, lb/h (kg/h)
10. Air outlet temperature, F (C)
11. Airside velocity, ft/s (m/s)
12. Airside pressure drop, in. WC (mm WC)
13. Fan power, HP (kW)
14. Connections per bundle, in. (mm)
15. Heat transfer coefficient, Btu/h ft2 F/ Tubeside
(kcal/h m2 C)
Airside
Overall
16. Tubeside pressure drop, psi (kg/cm2)

4
2
1.26 (32) 0.098 (2.5) 492 (12500)
0.63 (16) 11 (433)
40 8
2
33,786 (3140)
2 15 (4.57)
6,500,000 (2,948,350)
168.5 (75.8)
17.45 (5.32)
0.55 (14)
23.4 (17.4)
In: 2 8 (200)
Out 2 6 (150)
282.6 (1380)
149.5 (730)
85.4 (417)
2.84 (0.2)

the total yearly hours of operation. Thus, a temperature variation chart at sufficiently
short intervals throughout the year is needed for a proper estimate of the design ambient
temperature. Such data are available from meteorological offices.
The sensitivity of the performance of an air-cooled heat exchanger with the design
ambient temperature is borne out by the following case study.

CASE STUDY 5.5: EFFECT OF DESIGN AMBIENT TEMPERATURE


An air-cooled heat exchanger was to be designed for condensing a distillation column
overhead. The principal process parameters are indicated in Table 5.5a.
Since the pipe-rack width on which the air-cooled heat exchanger was to be mounted
was 39.4 ft (12.0 m), a tube length of 41 ft (12.5 m) was to be used. The tubes were to be of
carbon steel and the fins of aluminum.
A thermal design was prepared, the principal construction and performance parameters
of which are indicated in Table 5.5b. A tube size of 1.26 in. (32 mm) OD and 0.098 in. (2.5
mm) thickness was selected as the optimum in this instance since it permitted the best
49

utilization of the allowable pressure


drop and thereby yielded the highest
tubeside heat transfer coefficient. With
Design air temperature,
MTD, F
Overdesign,
the more common 0.984 in. (25 mm
F (C)
(C)
%
OD) tubes, the allowable tubeside
104.0 (40)
36.9 (20.5)
5.4
pressure drop was exceeded by far
105.8 (41)
34.9 (19.4)
0.2
with two tube passes, and with one
107.6 (42)
32.9 (18.3)
5.2
tube pass, the tubeside heat transfer
109.4 (43)
30.8 (17.1)
10.7
coefficient was far less than the one
111.2 (44)
28.8 (16.0)
16.2
presently obtained with 1.26 in. (32
mm) OD tubes.
In order to demonstrate the effect of the design air temperature on the performance of
the air-cooled heat exchanger, the design air temperature was varied from 104F (40C) to
111.2F (44C) in steps of 1.8F (1C). The results are shown in Table 5.5c. It will be seen
that there is a sharp reduction in the MTD so that the air-cooled exchanger first becomes less
oversurfaced and then undersurfaced.
Interestingly, there are a few features in the present design that might catch the attention
of the reader. First, air-cooled hydrocarbon condensers usually have four to five tube rows;
but here, it is eight. This is because the plot area was limited so that the number of bays had
to be restricted to four.
Second, the airside velocity and thereby the airside pressure drop and fan power are all
much on the lower side (see Table 5.5b). This was not unintentional, but a deliberate effort
to minimize the operating cost of the air-cooled heat exchanger. Notice the permissible
pressure drop has been fully utilized. If we were to increase the airside heat transfer
coefficient at the expense of fan power, the overall heat transfer coefficient will increase,
thereby reducing the heat transfer required. Now, the heat transfer area can be reduced by
reducing either the number of tubes or the tube length, or both. In the present instance, the
tube length is fixed and cannot be reduced. If the number of tubes is reduced, the tubeside
pressure drop will exceed the permissible limit. If one were to suggest a reduction in the
number of tube passes from two to one in order to contain the pressure drop, the tubeside
heat transfer coefficient will reduce drastically, thereby making the unit undersurfaced.
Therefore, the heat transfer area of the present design cannot be reduced. Since this is the
case, it is best to minimize the fan power and thereby the operating cost, and that is precisely
what has been done.
Table 5.5c: Effect of design air temperature on
performance of air-cooled heat exchanger for Case
Study 5.5

5.8 Noise
With the present concern over the environment, noise pollution often represents a constraint in the design of air-cooled heat exchangers. Fans and drives have to be designed
and/or selected to comply with local laws for noise limitation.
Sound power level is an objective quantity and is the total sound emitted by a system to
the environment. Sound pressure level is a subjective quantity and is the measured sound
pressure related to a fixed value, largely depending on the distance from the source of the
noise.
The sound power level emitted by air-cooled heat exchangers to the environment is
produced almost entirely by the fans and drives, with the fan drive often accounting for as
much as 50% of the total. Typical sound power levels are 6080 dBA for electric motors, 60
65 dB for V-belts, and 70100 dB for gear boxes.
The major part of the noise produced from axial fans is by vortex shedding at the trailing
B

50

edge of the blades. The noise power produced by the fans will vary approximately to the
third power of the fan tip speed. Fan manufacturers generally tend to limit the tip speed to
about 200 ft/s (61 m/s). Special low-noise fans are also designed for quieter operation at a
given HPthese employ a larger number of blades and/or wider chord blades.
The overall sound power level (PWL) of an ACHE installation can be expressed as
PWL = K + 30 logTS + 10 logHP

(5.8)

where
PWL = sound power level in dB (reference level 10-12 watts)
K = a constant, established by performance tests or furnished by the fan manufacturer
TS = tip speed in m/s
HP = absorbed shaft horse power.
For a forced draft installation, the maximum sound pressure level at a distance of 3.3 ft (1
m) from the fan is given by
SPL = 46 + 30 logTS + 10 logHP 20 log D

(5.9)

where
SPL = sound pressure level in dB (reference level 0.0002 microbars)
D = fan diameter, m
For an induced draft installation, the SPL may be reduced by 3 dB, measured 3.3 ft (1 m)
below the tube bundle.
From the above, it can be seen that if the tip speed of a fan is reduced from 197 ft/s (60
m/s) to 131 ft/s (40 m/s), the SPL will reduce by 5.3 dB.
When two sounds of equal intensity are added together, the resulting sound level
increases by 10 log2 or 3 dB. Therefore, when the number of fans increases from one to two,
only 3 dB is added to the earlier sound level. The same effect results when the number of
fans is increased from two to four, four to eight, and so on.
The distance between the observer and the fan also has an important bearing. The sound
pressure level varies as 20 log(d2/d1), where (d2/d1) is the distance ratio. This translates into a
reduction of 6 dB for a doubling of the distance from the fan.

51

Table 5.6a: Typical overall heat transfer coefficients for air-cooled liquid coolers [1]
Service
Oils, 20 API
200F (93.3C)
300F (148.9C)
400F (204.4C)
Oils, 30 API
150F (65.6C)
200F (93.3C)
300F (148.9C)
400F (204.4C)
Oils, 40 API
150F (65.6C)
200F (93.3C)
200F (148.9C)
200F (204.4C)

Overall heat transfer coefficient


Btu/h ft2 F
kcal/h m2 C
1016
1322
3040

5080
65110
150200

1223
2535
4555
5060

60115
125175
225275
250300

2535
5060
5565
6070

125175
250300
275325
300350

Service
Heavy oils 814 API
300F (148.9C)
400F (204.4C)
Diesel oil
Kerosene
Heavy naphtha
Light naphtha
Gasoline
Light hydrocarbons
Alcohols and most
organic solvents
Ammonia
Brine, 75% water
Water
50% Ethylene glycol
and water

Overall heat transfer coefficient


Btu/h ft2 F
kcal/h m2 C
610
1016
4555
5560
6065
6570
7075
7580

3050
5080
225275
275300
300325
325350
350375
375400

7075

350375

100120
90110
120140
100120

500600
450550
600700
500600

Table 5.6b: Typical overall heat transfer coefficients for air-cooled


condensers [1]
Service
Steam
Steam, 10% noncondensables
Steam, 20% noncondensables
Steam, 40% noncondensables
Pure light hydrocarbons
Mixed light hydrocarbons
Gasoline
Gasoline-steam mixtures
Medium hydrocarbons with steam
Pure organic solvents
Ammonia

Overall heat transfer coefficient


Btu/h ft2 F
kcal/h m2 C
700750
140150
500550
100110
475500
95100
350375
7075
400425
8085
325375
6575
300375
6075
350375
7075
275300
5560
375400
7580
500550
100110

52

Table 5.6c: Typical overall heat transfer coefficients for air-cooled vapor coolers (Ref. [1])

Service

Overall heat transfer coefficient, Btu/h ft2 F (kcal/h m2 C)


10 psig
50 psig
100 psig
300 psig
500 psig
(0.7 kg/cm2 g)
(3.5 kg/cm2 g)
(7.0 kg/cm2 g) (21.1 kg/cm2 g) (35.2 kg/cm2 g)

1520 (75100)
Light hydrocarbons
Medium hydrocarbons
1520 (75100)
and organic solvents
Light inorganic vapors 1015 (5075)
810 (4050)
Air
1015 (5075)
Ammonia
1015 (5075)
Steam
2030 (100150)
Hydrogen, 100%
1728 (85140)
75% vol.
1525 (75125)
50% vol.
1223 (60115)
25% vol.

3035 (150175)

4550 (225250)

6570 (325350)

7075 (350375)

3540 (175200)
1520 (75100)
1520 (75100)
1520 (75100)
1520 (75100)
4550 (225250)
4045 (200225)
3540 (175200)
3035 (150175)

4550 (225250)
3035 (150175)
2530 (125150)
3035 (150175)
2530 (125150)
6570 (325350)
6065 (300325)
5560 (275300)
4550 (225250)

6570 (325350)
4550 (225250)
4045 (200225)
4550 (225250)
4550 (225250)
8595 (425475)
8085 (400425)
7580 (375400)
6570 (325350)

7075 (350375)
5055 (250275)
4045 (200225)
5055 (250275)
5560 (275300)
95100 (475500)
8590 (425450)
8590 (425450)
8085 (400425)

References
[1] Brown, R., 1978, A Procedure for Preliminary Estimate, Chem. Eng., 85(8), pp. 108111.
[2] Briggs, D.E., and Young, E.H., 1965, "Convection Heat Transfer and Pressure Drop of Air
Flowing across Triangular Banks of Finned Tubes", Chemical Engineering Progress Symp.,
Series 41, No. 59.
[3] Robinson, K.K., and Briggs, D.E., 1965, Pressure Drop of Air Flowing across Triangular
Pitch Banks of Finned Tubes, 8th Natl. Heat Transfer Conf., AIChE, Preprint 20.
[4] McDermott Inc., 1994, The Basics of Air-cooled Heat Exchangers, Hudson Products Brochure McDermott Inc., Houston.

Further Reading
1.

Zukauskas, A., 1981, Air-cooled Heat Exchangers, Heat Exchangers: Thermal Hydraulic
Fundamentals and Design, Kakac, S., Bergles, A.E., and Mayinger, F., eds., Hemisphere,
Washington, DC.
2. Kern, D.Q., and Kraus, A.D., 1972, Extended Surface Heat Transfer, McGraw-Hill.
3. Mcketta, J.J., ed., 1992, Heat Transfer Design Methods, Marcel Dekker, New York.
4. Farrant, P.E., 1983, Noise and its Influence on Air Cooled Heat Exchanger Design, Heat
ExchangersTheory and Practice, Hemisphere, Washington, DC.
5. Paikert, P., and Ruff, K., 1983, State of Art for Design of Air Cooled Heat Exchangers with
Noise Level Limitation, Heat ExchangersTheory and Practice, Hemisphere, Washington,
DC.
6. API, 1981, Measurement of Noise from Air-cooled Heat Exchangers, API Recommended Practice 631M, 1st Ed., June 1981 (Reaffirmed Oct. 1985), American Petroleum Institute, Washington, DC.

53

54

CHAPTER 6

Thermal Design of Condensing


Air-Cooled Heat Exchangers
6.1 Introduction
Thus far, we have been looking at single-phase air-cooled heat exchangers, namely, gas
and liquid coolers. In this chapter, we will look at air-cooled condensers.
The thermal design of condensers is a fascinating subject since there is considerable
variation in service, operating pressure, and condensing range. Due to large-scale research
carried out in recent years, the phenomenon of condensing is now quite well understood, and
this sophisticated knowledge is embodied in several proprietary software packages.
As with any complex subject, it is important to grasp the fundamentals and the interplay
of parameters in order to not only enjoy the activity, but to produce efficient and optimized
designs as well. The accent in the following sections will not be on a plethora of equations
of which there is a preponderance in the published literaturebut rather on their application
for optimum design of air-cooled heat exchangers.

6.2 Classification of Air-Cooled Condensers


Air-cooled condensers can be classified as per the following modalities.
6.2.1 According to service
Condensation in an air-cooled heat exchanger can either be total or partial. This will depend on the service and the operating conditions of pressure and temperature. A pure
component will condense isothermally and will invariably be totally condensed. A mixture of components, or a mixture of a condensable and a noncondensable, can be condensed either totally or partially, depending on the process requirements.
6.2.2 According to condensing range
If a condensing vapor is a pure component, condensation will evidently be isothermal.
However, if the condensing stream is a mixture of various components, the condensation
will be accomplished over a temperature range. The wider the mixture (in terms of volatility), the greater will be the condensing range, which is the temperature range through
which the process of condensation will take place.
When a pure component has a noncondensable associated with it, the rate of condensation of the condensing component is initially quite rapid, but slows down progressively
with temperature as the percentage of noncondensable increases. There could even be an
initial desuperheating zone if the stream is not saturated at the entry to the condenser.
55

Quite a different situation is encountered in an air compressor intercooler, where a very


small percentage of atmospheric water vapor enters the condenser along with bulk air which
has to be cooledas the air is cooled, some associated water vapor condenses. Vapor shear
is very high in such cases due to the extremely high vapor weight fraction.
Evidently, the determination of the heat transfer coefficient is much easier for purecomponent isothermal condensation than for condensation of a mixture through a
temperature range. In the latter case, sensible vapor cooling and diffusion enter the picture
and complicate matters considerably. This is discussed at length in Section 6.3.3. Purecomponent condensation yields a much higher heat transfer coefficient than condensation of
a mixture of different components over a temperature range, since (a) no sensible vapor
cooling is involved and (b) there is no mass transfer resistance or diffusion.
6.2.3 According to operating pressure
The operating pressure of a condenser can vary from high vacuum to hundreds of atmospheres. Since vapor shear plays a very important role in nonisothermal condensation and
since vapor density is directly proportional to the operating pressure of a condenser, the
latter exerts considerable influence in condenser design. The higher the operating pressure, the higher the condensing heat transfer coefficient.
The other reason is linked to pressure drop. In a condenser, we are essentially handling
vapor. In isothermal condensers, the heat duty is only the latent heat of condensation and
there is no vapor cooling. However, in nonisothermal condensers, there is vapor cooling
duty, besides the latent heat of condensation. The higher the vapor mass velocity, the higher
is the heat transfer coefficient of the vapor cooling duty and the higher the overall (weighted)
condensing heat transfer coefficient. This can be explained as follows.
Unlike liquids whose densities do not vary with pressure since they are incompressible,
vapor density is directly proportional to pressure. Thus, when the operating pressure of a
given condenser is higher, so is the vapor density. We all know that pressure drop is
proportional to the square of mass velocity and inversely proportional to density. For a given
pressure drop, the higher vapor density permits a higher vapor mass velocity inside the
tubes, which then translates into a higher condensing heat transfer coefficient.
Evidently, it is much more difficult to handle low-pressure services, primarily because
of the low vapor densities and the low allowable pressure drop. This is really a double
penalty since a lower vapor density produces a higher pressure drop, all other things
remaining constant. Thus, handling a low-pressure condensing service necessitates the use of
one or more of the following: a lower number of tube passes (often only one), a larger tube
diameter (1.25 in. or 32 mm, or even 1.5 in. or 38 mm), and/or a smaller tube length.
Common to all the above features are a larger flow area and/or a shorter flow length.

6.3 Mechanisms of Condensing


Before proceeding with the design of air-cooled condensers, it will be a good idea to take
a brief look at the mechanisms of condensing. Condensation occurs when a vapor comes
in contact with a surface that is at a temperature below its dew point. The normal mechanism for heat transfer in commercial condensers is filmwise condensation. Although
much academic investigation has been devoted to dropwise condensation, there has been
very little application of this for commercial purposes, since special surfaces are required
to maintain this mode of condensation. Furthermore, beneficial results are demonstrated
only at low liquid loadings. Finally, the airside thermal resistance is invariably control56

ling in air-cooled condensers, so that the increase in the overall heat transfer coefficient
due to a large increase in the tubeside condensing heat transfer coefficient is not considerable. Consequently, we shall only look at filmwise condensation in this book.
When condensation occurs, a film of liquid covers the heat transfer surface. The
thickness of the film depends on the rate of condensation and the rate of removal of the
condensate. The latter depends on the actions of shear and gravity. Therefore, it is very
important for condenser calculations to determine the flow regime, shear controlled or
gravity controlled, since different correlations have to be employed in the two regimes.
When the shear force is much greater than the force of gravity, the regime is shear
controlled. However, when the force of gravity is predominant, the regime is gravity
controlled.
When the condensing stream is a mixture of various components, with or without
noncondensables, the condensation process becomes far more complex than for a pure vapor
since it then involves mass transfer effects, which create additional thermal resistances,
thereby reducing the heat transfer coefficient considerably.
Let us look at some of the possible modes of condensation to have a better
understanding of the phenomenon.

57

Turbulent

Condensate

Coolant

Gravity-controlled condensation
Here, the force of gravity is considerably greater
than that of shear. This is the classical Nusselt mode
of condensation, wherein the film is laminar and
heat transfer is considered to take place totally by
conduction through the condensate film.
In practical applications, however, the liquid film
does not remain laminar but, after passing through a
transition region, becomes turbulent after a critical
Reynolds number is reached. When this happens, the
heat transfer coefficient increases.
The typical variation of the gravity-controlled
condensate heat transfer coefficient with liquid
Reynolds number is depicted graphically in Fig. 6.2.
This curve is general in that it portrays typical

Laminar Wavy

6.3.1 Vertical in-tube condensation


Let us consider condensation inside vertical tubes,
since this is the simplest situation where the mechanisms can be easily visualized. (Granted that aircooled heat exchangers are invariably horizontal,
but there are applications when the tubes are inclined or even vertical.) Figure 6.1 represents the
situation, wherein it will be seen that the condensate
film flows under the influence of both the shear
force and the force of gravity, and is restrained by
the shear force (friction) at the wall. Based on the
balance of these forces and the liquid and vapor
flow rates and their physical properties, the film
thickness can be determined.

Fig. 6.1 Condensation on a vertical


surface in the absence of vapor shear
(redrawn from the Heat Exchanger
Design Handbook 2002, with permission of Begell House, Inc.)

Condensate film heat


Transfer coefficient

Increasing
Prandtl No.

Laminar film

Transition rigime

Turbulent film

Liquid Reynold's No.

Fig. 6.2 Typical variation of gravity-controlled condensate film heat transfer coefficient with liquid Reynolds number (redrawn with permission from HTRI)

behavior, whether the condensation is inside vertical tubes or horizontal tubes and, in fact,
even when it is outside tubes.
Shear-controlled condensation
In this regime, the vapor velocity is very high and, therefore, the gravity component becomes negligible. The flow pattern is annular and is shown in Fig. 6.3a. The variation of
the condensate film heat transfer coefficient with vapor Reynolds number is essentially
linear and is represented graphically in Fig. 6.3b. Evidently, for the same Reynolds number, the heat transfer coefficient will be higher when the Prandtl number of the condensate is higher.
The main flow pattern of vertical in-tube down flow is the existence of annular flow for
a wide range of flow
conditions. In fact, for
a low flow rate case,
Fs
annular flow persists
throughout the entire
tube.

Vapour Core

Dv

Di
Fig. 6.3a Annular flow in shear-controlled condensation inside vertical
tubes (redrawn with permission from HTRI)
58

6.3.2 Horizontal intube condensation


Horizontal tubeside
condensation is commonly employed in
air-cooled heat exchangers and represents a more difficult
situation because of
the more complex
nature of the flow

Condensate film heat transfer coefficient

Prandtl no.

Vapor Reynolds no.

Fig. 6.3b Typical variation of ahear-controlled condensate film heat transfer coefficient with
vapor Reynolds number (redrawn with permission from HTRI)

patterns, especially in the gravity-controlled regime.


For both high and low total mass velocities, condensation begins in the annular regime.
Subsequently, in the case of high total mass velocity, the condensation progresses to slug
flow whereas in the case of low total mass velocity, it progresses to wavy or stratified flow.
This is represented in Fig. 6.4a. Cross-sectional views of stratified and annular flow are
shown in Fig. 6.4b.
In stratified flow, the heat transfer coefficient is a strong function of the height of the
stratified layer, which is difficult to predict. In the shear-controlled regime, however, the
conditions are much less complicated and, in fact, the heat transfer coefficient is virtually the
same as in vertical in-tube condensation.

High Flow

Annular mist

Annular

Annular with
high stratification

Slug elongated buble

Low Flow

Annular
mist

Annular Annular with


high
stratification

Wave

Stratified

Fig. 6.4a Stratified and annular flow in horizontal in-tube condensation (redrawn with permission
from HTRI)
59

(a)

(b)

Fig. 6.4b Stratified and annular flow in horizontal in-tube condensation (redrawn from the Heat Exchanger Design Handbook 2002, with permission of
Begell House, Inc.)

6.3.3 Condensation of mixed vapors and mixtures of vapors and noncondensables


Thus far, we have discussed condensation of pure vapors. However, a far more practical
application is the condensation of a mixture of vapors, with or without noncondensables.
Here, we may have total condensation of a multicomponent mixture or partial condensation of only some of the components.
This situation is rather complicated and the following features will have to be considered
for design:
a) Since the heavier components will condense first and the lighter components
later, condensation will take place over a temperature range.
b) Because the condensation is not linear, there will be sensible heat duty besides
condensing duty. Any vapor that has not been condensed at any given location
will have to be cooled to its dew point before it can condense. Similarly, any vapor that has been condensed will have to be cooled as liquid to the final outlet
temperature. Since the vapor cooling heat transfer coefficient is considerably
lower than that of condensing, it reduces the overall heat transfer coefficient appreciably. It follows, therefore, that the wider the range of components and,
thereby, the wider the condensing temperature range, the lower will be the heat
transfer coefficient, all other things remaining constant.
c) Since the composition of the vapor and the liquid vary continuously along the
path of condensation, so will their physical properties. Particularly significant is
the liquid viscosity.
d) The molecules of the heavy components must diffuse through the barrier of the
molecules of the lighter components to reach the condensing surface. Therefore,
the condensation rate will be controlled by the rate of diffusion as well as the rate
of heat transfer.
The heat transfer rate across the condensate film is calculated in the same way as for pure
components. The vapor-phase processes are (a) the diffusion of heavy molecules from the
bulk to the interface and (b) the cooling of the vapor phase to the saturation temperature
of the vapor mixture prevailing at the interface.
60

Pressure drop
The determination of pressure drop in a condenser is a very complex task since the velocity and flow pattern change constantly along the flow path. The various components of
pressure drop in an air-cooled condenser are as follows:
a)
b)
c)
d)

inlet and exit losses (contraction and expansion) in nozzles and headers
two-phase friction loss
static head
momentum change

The static head is usually insignificant in condensers.


The momentum change results in a pressure gain since there is deceleration of the vapor
as its flow rate decreases with the progress of condensation. However, this is insignificant
unless the condenser operates under vacuum, in which case the pressure gain could be
substantial.
The two-phase friction is usually the largest component of the overall pressure drop and
is determined stepwise along the length of the condenser, using Martinellis or other
correlations. The Martinelli correlation is particularly accurate for condensing inside
horizontal tubes.

6.4 Some Case Studies


Let us now take a look at some case studies to illustrate some of the points that have been
made in the foregoing sections. We will first look at an isothermal condenser, which is a
pure component condenser, and thereafter at condensers handling mixtures. Finally, we
will look at desuperheating and subcooling.

CASE STUDY 6.1: ISOTHERMAL CONDENSER


The principal process parameters of an isothermal condenser are detailed in Table 6.1a.
Note the small difference between the pentane inlet and outlet temperatures. This is due
to the pressure drop across the condenser and the relatively low operating pressure of the
system, whereby even a small pressure drop results in a noticeable drop in the saturation
temperature.
Tubes were to be of carbon steel, 0.984 in. (25 mm) OD, 0.0984 (2.5 mm) thick, and
29.5 ft (9.0 m) long with the usual aluminum fins.
Since the tubeside heat transfer coefficient was expected to be high, this being a pure
component condenser, and thereby the airside thermal resistance controlling, a higher fin
height (0.63 in. or 16 mm) was employed. A thermal design was made and the salient
construction and performance parameters are shown in Table 6.1b.
The first thing to notice in Table 6.1b is the high tubeside heat transfer coefficient of
413.7 Btu/h ft2 F (2020 kcal/h m2 C/kcal). This is because the condenser is condensing a
pure component isothermally, which implies that there is only phase change and no sensible
cooling. Thus, the tubeside thermal resistance is only 24.02% of the total resistance and, as
expected at the outset, the airside thermal resistance is controlling, being 54.78% of the total
resistance to heat transfer.
Despite the rather high overall heat transfer coefficient, this is a large condenser with
three bays and two bundles per bay, with a total bare tube heat transfer area of 11,524 ft2
(1071 m2). This is because of the rather large heat duty and the rather low MTD.
61

Table 6.1a: Principal process parameters for air-cooled condenser for Case Study 6.1
1. Fluid
2. Flow rate, lb/h (kg/h)
3. Temperature in/out, F (C)
4. Operating pressure, psia (kg/cm2a)
5. Weight fraction vapor, in/out
6. Total allowable pressure drop, psi (kg/cm2)
7. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
8. Heat duty, MM Btu/h (MM kcal/h)
9. Vapor properties
Density in/out, lb/ft3 (kg/m3)
Viscosity in/out, cp
Specific heat in/out,
Btu/lb F (kcal/kg C)
Thermal conductivity in/out,
Btu/h ft F (kcal/h m C)
10. Liquid properties
Density in/out, lb/ft3 (kg/m3)
Viscosity in/out, cp
Specific heat in/out,
Btu/lb F (kcal/kg C)
Thermal conductivity in/out,
Btu/h ft F (kcal/h m C)

Pentane
264,600 (120,000)
169.2 (76.2)/165.2 (74)
49.8 (3.5)
1.0/0.01
2.84 (0.2)
0.00146 (0.0003)
37.5 (9.44)
0.6 (9.609)/0.57 (9.1)
0.0085/0.0085
0.484/0.481
0.0114 (0.017)/0.0113 (0.0168)
35.2 (564)/35.3 (566)
0.142/0.144
0.636/0.632
0.055 (0.0818)/
0.0554 (0.0824)

Table 6.1b: Principal construction and performance parameters of air-cooled condenser


for Case Study 6.1
1. No. of bays
2. No. of bundles per bay
3. Tube OD thickness length, in. (mm)
4. Fin density, fins/in. (fins/m)
5. No. of tubes per row no. of rows tube pitch, in. (mm)
6. No. of tube passes
7. Total bare tube area, ft2 (m2)
8. Total extended area, ft2 (m2)
9. Fans per bay fan diameter, ft (m)
10. Total air flow rate, MM lb/h (MM kg/h)
11. Air outlet temperature, F (C)
12. Airside face velocity, ft/s (m/s)
13. Airside pressure drop, in. WC (mm WC)
14. Individual fan power, HP (kW)
15. Total fan power, HP (kW)
16. Heat transfer coefficient,
Tubeside
Btu/h ft2 F/ (kcal/h m2 C/kcal)
Airside
Overall

3
2
0.984 (25) 0.098 (2.5) 354(9000)
11 (433)
43 6 2.64 (67)
2
11,524 (1071)
275,316 (25,587)
2 12 (3.657)
4.41 (2.0)
144.7 (62.6)
10.3 (3.15)
0.43 (10.9)
19.6 (14.6)
117.6 (87.6)
413.7 (2020)
181.5 (886)
99.3 (485)

17. Thermal resistance, %

Airside
Tubeside
Fouling
Tube wall
18. Tubeside pressure drop, allow/calc., psi (kg/cm2)
19. MTD, F (C)
20. Overdesign, %

62

54.78
24.02
18.19
3.01
2.84 (0.2)/2.84 (0.2)
37.1 (20.6)
13.6

Note also in Table 6.1b that the permitted tubeside pressure drop of 2.84 psi (0.2 kg/cm2)
has been fully utilized, which means that the tubeside heat transfer coefficient is the highest
that can be achieved.
Just to give the reader an idea of the extent of change in performance with increased
tubeside mass velocity, let us assume that the allowable tubeside pressure drop is 10 psi, and
increase the number of tube passes from the present two to three. While the tubeside heat
transfer coefficient increases from 413.7 Btu/h ft2 F (2020 kcal/h m2 C/kcal) to 459.4 Btu/h
ft2 F (2243 kcal/h m2 C/kcal), an increase of only 11%, the tubeside pressure drop increases
sharply from 2.84 psi (0.2 kg/cm2) to 9.23 psi (0.65 kg/cm2), an increase of 225%. This is
typical of isothermal condensers. Even with a relatively low tubeside mass velocity, the
tubeside heat transfer coefficient is quite high, and an attempt to increase the mass velocity
does not yield an appreciable increase in the heat transfer coefficient, while the pressure drop
increases to the square of the mass velocity, just as in the flow of single-phase fluids.
It was stated in Chapter 5 that for sensible cooling of gases and liquids in the turbulent
regime, the heat transfer coefficient varies to the 0.8 power of mass velocity while the
pressure drop varies to the power 2. In the preceding analysis, we see that while the pressure
drop does increase to the power 2 of the mass velocity, the heat transfer coefficient increases
only to the power 0.26.
In the present condenser, the regime is shear controlled from the inlet (full vapor) until a
vapor weight fraction of about 0.4, and the tubeside heat transfer coefficient varies from 756
Btu/h ft2 F (3690 kcal/h m2 C/kcal) to 444 Btu/h ft2 F (2168 kcal/h m2 C/kcal). Thereafter,
the condenser is in the transition region until a vapor weight fraction of about 0.08, and the
tubeside heat transfer coefficient varies from 385 Btu/h ft2 F (1879 kcal/h m2 C/kcal) to
278 Btu/h ft2 F (1357 kcal/h m2 C/kcal). Finally, the condenser is in the gravity-controlled
region only in the last increment of the condenser with a tubeside heat transfer coefficient of
217 Btu/h ft2 F (1061 kcal/h m2 C/kcal). This large variation clearly demonstrates the
influence of vapor shear on the tubeside heat transfer coefficient.

CASE STUDY 6.2: NARROW-RANGE CONDENSER


Let us next examine a narrow-range condenser. The principal process parameters are indicated in Table 6.2a. The dew point is 159.8F (71C) and the bubble point is 141.8F
(61C); the condensing range thereby is 18F (10C). Tubes are to be of carbon steel,
0.984 in. (25 mm) OD, 0.0984 (2.5 mm) thick, and 29.5 ft (9.0 m) long with the usual
aluminum fins.
Since the tubeside heat transfer coefficient is still expected to be fairly high and,
therefore, the airside thermal resistance controlling, once again a larger fin height of 0.63 in.
(16 mm) was employed. The principal construction and performance parameters of the
thermal design that emerged are indicated in Table 6.2b.
Once again, this is a large condenser with four bays, two bundles per bay, and a total
bare tube heat transfer area of 15,720 ft2 (1461 m2). This is because of the rather large heat
duty and the rather low MTD. Compared to the previous case study, the heat duty is
somewhat higher and the MTD significantly lower; hence, this is an even larger condenser.
The most crucial feature to notice here is that the tubeside heat transfer coefficient has
come down appreciably from the previous case, from 413.7 Btu/h ft2 F (2020 kcal/h m2
C/kcal) to 257.9 Btu/h ft2 F (1259 kcal/h m2 C/kcal). This is only to be expected, since we
are now looking at condensation through a temperature range, albeit narrow, wherein
besides phase change, we also have sensible vapor cooling.
63

Table 6.2a: Principal process parameters for air-cooled condenser for Case Study 6.2
1. Fluid
2. Flow rate, lb/h (kg/h)
3. Temperature in/out, F (C)
4. Operating pressure, psia (kg/cm2a)
5. Weight fraction vapor, in/out
6. Total allowable pressure drop, psi (kg/cm2)
7. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
8. Heat duty, MM Btu/h (MM kcal/h)
9. Vapor properties
Density, lb/ft3 (kg/m3)
(average)
Viscosity, cp
Specific heat, Btu/lb F (kcal/kg C)
Thermal conductivity, Btu/h ft F (kcal/h m C)
10. Liquid properties
Density, lb/ft3 (kg/m3)
(average)
Viscosity, cp
Specific heat, Btu/lb F (kcal/kg C)
Thermal conductivity, Btu/h ft F (kcal/h m C)

Butane + Pentane
262,350 (119,000)
159.8 (71)/141.8 (61)
49.77 (3.5)
1/0
2.84 (0.2)
0.00146 (0.0003)
39.0 (9.83)
36.2 (580)
0.17
0.6
0.053 (0.079)
0.58 (9.3)
0.01
0.462
0.0116 (0.0173)

Table 6.2b: Principal construction and performance parameters of air-cooled condenser for
Case Study 6.2
1. No. of bays
2. No. of bundles per bay
3. Tube OD thickness length, in. (mm)
4. Fin density, fins/in. (fins/m) fin height, in. (mm)
5. No. of tubes per row no. of rows tube pitch, in. (mm)
6. No. of tube passes
7. Total bare tube area, ft2 (m2)
8. Total extended area, ft2 (m2)
9. Fans per bay fan diameter, ft (m)
10. Total air flow rate, MM lb/h (MM kg/h)
11. Air outlet temperature, F (C)
12. Airside face velocity, ft/s (m/s)
13. Airside pressure drop, in. WC (mm WC)
14. Individual fan power, HP (kW)
15. Total fan power, HP (kW)
16. Heat transfer coefficient,
Tubeside
Btu/h ft2 F/ (kcal/h m2 C/kcal)
Airside
Overall

4
2
0.984 (25) 0.098 (2.5) 492 (12500)
11 (433) 0.63 (16)
44 6 2.64 (67)
2
15,720 (1461)
76,963 (34,910)
2 12 (3.658)
6.504 (2.95)
134.4 (56.9)
11.2 (3.41)
0.48 (12.1)
24.8 (18.5)
198.4 (148)
257.9 (1259)
187.3 (914.5)
88.1 (430)

17. Tubeside pressure drop, allowable/calc., psi (kg/cm2)


18. Thermal resistance, %
Airside
Tubeside
Fouling
Tube wall
19. MTD, F (C)
20. Overdesign, %

64

2.84 (0.2)/1.6 (0.11)


47.04
34.18
16.13
2.65
28.8 (16.0)
2.15

The airside thermal resistance is still controlling, at 47.04%. Thus, the use of a higher fin
height of 0.63 in. (16 mm) is judicious.
Note that the tubeside pressure drop has been utilized only a little more than 50%.
However, an increase in the number of tube passes from two to three is not possible since
the pressure drop will increase by approximately (1.5)2.8 or 3 times, which will push the
pressure drop to well beyond the permissible limit.
One will also discern that the fan power consumption (24.8 HP or 18.5 kW) is
somewhat on the low side. Thus, since the airside is controlling, one could be tempted to
consider reducing the number of tubes per row and increasing the airside heat transfer
coefficient at the expense of a higher power consumption, to make up for the reduction in
heat transfer area. Unfortunately, as has been propounded in Chapter 5, the increase in
airside power consumption with increased velocity is far, far greater than the increase in heat
transfer coefficient. Thus, it is unlikely that a better design can be produced at the expense of
higher fan power consumption.
Coming to the breakup of the tubeside heat transfer coefficient, the condenser is in the
shear-controlled regime from the inlet down to a vapor weight fraction of about 0.55, with a
tubeside heat transfer coefficient of 386 Btu/h ft2 F (1884 kcal/h m2 C/kcal) to 282 Btu/h ft2
F (1376 kcal/h m2 C/kcal). Thereafter, it is in the transition region down to a vapor weight
fraction of about 16%, with a tubeside heat transfer coefficient of 277 Btu/h ft2 F (1352
kcal/h m2 C/kcal) to 223 Btu/h ft2 F (1088 kcal/h m2 C/kcal). Finally, the condenser is in
the gravity-controlled regime thereafter, with a tubeside heat transfer coefficient of 209
Btu/h ft2 F (1021 kcal/h m2 C/kcal) to 110 Btu/h ft2 F (537 kcal/h m2 C/kcal). Notice the
sharp reduction in the tubeside heat transfer coefficient, from 386 Btu/h ft2 F (1884 kcal/h
m2 C/kcal) at the beginning of the condenser to a mere110 Btu/h ft2 F (537 kcal/h m2
C/kcal) at the end. Once again, this vividly demonstrates the effect of shear on the
condensing heat transfer coefficient.
It may also be seen that as compared to isothermal condensation (in Case Study 6.1),
condensation of even a narrow-range condensing application yields a significantly lower
heat transfer coefficient.

CASE STUDY 6.3: WIDE-RANGE CONDENSER


We will now take a look at a wide-range condenser as specified in Table 6.3a. Saturated
vapor enters the condenser at 235.4F (113C). At the exit of the condenser, 52% vapor is
left uncondensed, so this is a partial condenser with a very wide condensing range. Thus,
we would expect a much lower tubeside heat transfer coefficient than what we saw in the
previous examples. To make matters worse, the operating pressure of the process stream
is very low. The consequent low vapor density will penalize the tubeside heat transfer
coefficient even further.
Tubes were to be of carbon steel, 0.984 in. (25 mm) OD, 0.118 in. (3.0 mm) thick, and
34.5 ft (10.5 m) long and the fins were to of aluminum.
Since the tubeside heat transfer coefficient is expected to be rather low and therefore
controlling, a smaller fin height of 0.49 in. (12.5 mm) was employed. The principal
construction and performance parameters of the thermal design that emerged are indicated in
Table 6.3b.
This is yet another large condenser having four bays, two bundles per bay, and a total
bare tube heat transfer area of 24,447 ft2 (2272 m2).
65

Table 6.3a: Principal process parameters for air-cooled condenser for Case Study 6.3
1. Fluid
2. Flow rate, lb/h (kg/h)
3. Temperature in/out, F (C)
4. Operating pressure, psia (kg/cm2a)
5. Weight fraction vapor, in/out
6. Total allowable pressure drop, psi (kg/cm2)
7. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
8. Heat duty, MM Btu/h (MM kcal/h)
9. Vapor properties
Density in/out, lb/ft3 (kg/m3)
Viscosity in/out, cp
Specific heat in/out, Btu/lb F (kcal/kg C)
Thermal conductivity in/out, Btu/h ft F
(kcal/h m C)
10. Liquid properties
Density in/out, lb/ft3 (kg/m3)
Viscosity in/out, cp
Specific heat in/out, Btu/lb F (kcal/kg C)
Thermal conductivity in/out, Btu/h ft F
(kcal/h m C)

Hydrocarbon mixture
348,300 (158,000)
235.4 (113)/147.2 (64)
28.4 (2.0)
1/0.52
1.4 (0.1)
0.00195 (0.0004)
60.2 (15.17)
0.2 (3.12)/0.17 (2.7)
0.012/0.01
0.46/0.44
0.016 (0.0238)/0.014 (0.021)
44.9 (720)/46.55 (746))
0.26/0.42
0.61/0.582
0.063 (0.094)/0.071 (0.106)

Table 6.3b: Principal construction and performance parameters of air-cooled condenser for Case
Study 6.3
1. No. of bays
2. No. of bundles per bay
3. Tube OD thickness length, in. (mm)
4. Fin density, fins/in. (fins/m) fin height, in. (mm)
5. No. of tubes per row no. of rows tube pitch, in. (mm)
6. No. of tube passes
7. Total bare tube area, ft2 (m2)
8. Total extended area, ft2 (m2)
9. Fans per bay fan diameter, ft (m)
10. Total air flow rate, MM lb/h (MM kg/h)
11. Air outlet temperature, F (C)
12. Airside face velocity, ft/s (m/s)
13. Airside pressure drop, in. WC (mm WC)
14. Individual fan power, HP (kW)
15. Total fan power, HP (kW)
16. Heat transfer coefficient, Btu/h Tubeside
ft2 F/ (kcal/h m2 C/kcal)
Airside
Overall

4
2
0.984 (25) 0.118 (3.0) 413 (10500)
11 (433) 0.49 (12.5)
44 8 2.36 (60)
One
24,447 (2272)
424,245 (39,428)
2 13 (3.962)
6.06 (2.75)
147 (63.9)
10.0 (3.04)
0.62 (15.77)
25.5 (19.05)
204 (152.4)
123.8 (604.4)
152.8 (746)
56.9 (278)

17. Tubeside pressure drop, allowable/calc., psi (kg/cm2)


18. Thermal resistance, %
Airside
Tubeside
Fouling
Tube wall
19. MTD, F (C)
20. Overdesign, %

66

1.4 (0.1)/1.4 (0.1)


37.2
46.02
14.61
2.16
48.2 (26.8)
11.5

Table 6.4a: Principal process parameters for air-cooled condenser for Case Study 6.4
1. Fluid
2. Flow rate, lb/h (kg/h)
3. Temperature in/out, F (C)
4. Operating pressure, psia (kg/cm2a)
5. Weight fraction vapor, in/out
6. Total allowable pressure drop, psi (kg/cm2)
7. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
8. Heat duty, MM Btu/h (MM kcal/h)
9. Vapor properties
Density in/out, lb/ft3 (kg/m3)
Viscosity in/out, cp
Specific heat in/out, Btu/lb F (kcal/kg C)
Thermal conductivity in/out,
Btu/h ft F (kcal/h m C)
10. Liquid properties
Density in/out, lb/ft3 (kg/m3)
Viscosity in/out, cp
Specific heat in/out, Btu/lb F (kcal/kg C)
Thermal conductivity in/out,
Btu/h ft F (kcal/h m C)

Hydrocarbon mixture
348,300 (158,000)
235.4 (113)/147.2 (64)
199 (14.0)
1/0.52
2.5 (0.18)
0.00195 (0.0004)
60.2 (15.17)
1.4 (21.84)/1.19 (18.9)
0.012/0.01
0.46/0.44
0.016 (0.0238)/0.014 (0.021)
44.9 (720)/46.55 (746))
0.26/0.42
0.61/0.582
0.063 (0.094)/0.071 (0.106)

The most crucial feature to notice here is that the tubeside heat transfer coefficient has
come down considerably from the previous casefrom 257.9 Btu/h ft2 F (1259 kcal/h m2
C/kcal) to 123.8 Btu/h ft2 F (604.4 kcal/h m2 C/kcal). This is only to be expected since we
are now looking at condensation through a much wider temperature range, involving a much
larger vapor cooling load and much higher diffusion resistance. The tubeside thermal
resistance is 46.02%, while the airside thermal resistance is 37.2%. Thus, the tubeside
thermal resistance is controlling.
From this and the earlier two case studies, we see clearly that the wider the condensing
range of a condenser, the lower the tubeside heat transfer coefficient we can expect.

CASE STUDY 6.4: EFFECT OF PRESSURE ON AIR-COOLED CONDENSERS


In Section 6.2.3, there was a discussion of the effect of operating pressure on the heat
transfer coefficient in air-cooled condensers, and it was explained how a higher operating
pressure leads to a higher tubeside heat transfer coefficient. This case study is now presented to vividly demonstrate this.
Consider the example presented in Case Study 6.3. Let us arbitrarily increase the
operating pressure sevenfold, from 28.4 psia (2.0 atma) to 199 psia (14 atma) and the
permissible tubeside pressure drop from 1.4 psi (0.1 kg/cm2) to 2.5 psi (0.18 kg/cm2), just to
see what happens to the condenser design. With a sevenfold increase in the operating
pressure, the vapor density will increase sevenfold as well. The revised process parameters
are shown in Table 6.4a. The changed parameters are shown in bold face, for the sake of
easy reckoning.
The revised thermal design is detailed in the second column of Table 6.4, with the first
column showing the original design (Case Study 6.3). Thanks to the much higher vapor
density and the somewhat higher allowable tubeside drop, it was possible to reduce the
number of bays from four to three, and the total bare tube heat transfer from 24,447 ft2 (2272
m2) to 20,820 ft2 (1935 m2).
67

Note the significantly higher tubeside mass velocity in the second design (17.03 lb/s ft2
or 83.14 kg/s m2) than in the first (8.48 lb/s ft2 or 41.4 kg/s m2). This results in a much higher
tubeside heat transfer coefficient, 171.7 Btu/h ft2 F (838.3kcal/h m2 C), and the consequent
lower heat transfer area.
Since the number of sections was reduced from four to three, the total face area came
down by 25%. Thus, for a meaningful comparison, the air flow rate had to be reduced so that
the fan power consumption remained the same. Notice that in the second design, the air
pressure drop was lower (0.57 in. or 14.4 mm WC versus 0.62 in. or 15.77 mm WC) but
since the volumetric flow rate per fan was higher (192,140 cfm or 5440 m3/min versus
180,960 cfm or 5124 m3/min), the fan power consumption was the same. Also, due to the
lower air flow rate in the second design, there was a small reduction in the MTD, from
48.2F (26.8C) to 46.1F (25.6C).
It will therefore be seen that even for a wide-range condenser, the tubeside heat transfer
coefficient need not be very low, provided the operating pressure is sufficiently high. Put in
other words, the tubeside heat transfer coefficient in a condenser can be expected to be
higher with an increase in the operating pressure of the system.
In terms of both the operating pressure and the condensing range of a condenser, we can
state that the higher the operating pressure and the lower the condensing range, the higher
will be the tubeside heat transfer coefficient.

6.5 Condensation with Desuperheating and/or Subcooling


The vapor that enters a condenser is usually saturated, so that condensation begins right
away. This is always the case when the vapor comes from the top of a distillation column.
However, it sometimes happens that the vapor is not a distillation column overhead, and
is superheated. In such a case, there will be desuperheating until the vapor becomes saturated at the operating pressure, followed by condensation.
The condensate leaving a condenser is usually a saturated liquid, but is sometimes
required to be subcooled for process reasons (e.g., the condensate is volatile and has to be
cooled to prevent flashing, it has valuable heat that can be recovered, or a distillation column
requires subcooled condensate as reflux).
Thus, we may have any of the following combinations in a condensing service:
1)
2)
3)
4)

only condensation
desuperheating and condensation
desuperheating, condensation, and subcooling
condensation and subcooling

The principal difference between condensation and the other two phenomena is that while
the condensation heat transfer coefficient is usually high, that of desuperheating and subcooling are rather low. The desuperheating heat transfer coefficient is low since it involves the sensible cooling of a vapor, while the subcooling heat transfer coefficient is
low since its mass velocity is low. Sometimes it is possible to increase the subcooling
liquid mass velocity by incorporating more tube passes for the subcooling zone.
Let us consider desuperheating and subcooling separately.
6.5.1 Desuperheating
The tube wall temperature will always be lower than the bulk vapor temperature, since
coolant flows on the other side. True desuperheating will exist only as long as the tube
68

wall temperature is greater than the vapor saturation temperature. If the tube wall temperature is less than the dew point of the vapor, it is said to be a wet wall condition.
However, if the tube wall temperature is greater than the dew point, it is said to be dry
wall condition.
The tube wall temperature will depend on both the heat transfer coefficient of the
desuperheating vapor and that of the air. Since the former will invariably be lower than the
latter, the tube wall temperature will tend to be closer to the air bulk temperature.
Consequently, only a part of the desuperheating heat duty will be transferred as gas cooling,
while the rest will be transferred as condensing. Evidently, how much of the heat duty will
be transferred as gas cooling (dry wall) and how much as condensing (wet wall) will depend
on the extent of superheating and the desuperheating and the air heat transfer coefficients.
This will be observed in the case study presented later in this section.
In actual practice, the phenomenon is somewhat complicated since, even with wet-wall
desuperheating, the bulk vapor is still superheated. Thus, although condensate forms at the
tube wall, the uncondensed superheated vapor reflashes some of the condensate. This
process continues until the bulk vapor cools down to the dew point when true condensation
begins. For all practical purposes, however, the heat transfer coefficient in the wet-wall
condition is virtually the same as in the true condensing mode; in reality, it is somewhat
lower.
All sophisticated heat exchanger thermal design software can handle desuperheating in
both the dry-wall and wet-wall modes described above.
The penalty (in the form of additional heat transfer area) associated with desuperheating
is usually rather small since the loss in the overall heat transfer coefficient is largely
compensated by the increase in the MTD. The greater the degree of superheating, the greater
is the decrease in the heat transfer coefficient; but then, the greater is the increase in the
MTD. Consequently, for the same total heat duty, the difference in the heat transfer area
between only condensing and condensing preceded by desuperheating is usually not
appreciable. However, the pressure drop may increase appreciably, especially if the
operating pressure is low.

CASE STUDY 6.5: CONDENSATION WITH DESUPERHEATING


Let us consider the propane condenser presented in case Study 6.1. In that example, the
propane entering the condenser was saturated. In order to demonstrate the effect of superheat on the performance of a condenser, we shall consider an arbitrary case of propane
superheating: 248F (120C). The saturation temperature is 165.2F (74C).
The principal process parameters for this new condition are shown in the second column
of Table 6.5a. For the sake of comparison, the corresponding parameters for Case Study 6.1
are shown in the first column of the same table. Note that the heat duty has gone up from
37.5 M Btu/h (9.44 M kcal/h) to 46.7 M Btu/h (11.77 M kcal/h), an increase of 24.7%.
A new design was generated and the principal construction and performance parameters
of the same are indicated in the second column of Table 6.5b. Once again, for the sake of
comparison, the corresponding values for Case Study 6.1 are shown in the first column of
the same table. The new design has a few more tubes per row (48 versus 43) and an
increased tube length (32.8 ft or 10 m versus 29.5 ft or 9.0 m). Since both the number of
tubes per row and the tube length were higher, the fan diameter was increased from 12 ft
(3.657 m) to 14 ft (4.267 m) to incorporate the minimum 40% area coverage.
69

Table 6.5a: Principal process parameters for air-cooled condenser for Case Study 6.5

1. Fluid
2. Flow rate, lb/h (kg/h)
3. Temperature in/out, F (C)
4. Operating pressure, psia (kg/cm2 a)
5. Weight fraction vapor, in/out
6. Total allowable pressure drop, psi (kg/cm2)
7. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
8. Heat duty, MM Btu/h (MM kcal/h)
9. Vapor properties
Density in/out, lb/ft3 (kg/m3)
Viscosity in/out, cp
Specific heat in/out,
Btu/lb F (kcal/kg C)
Thermal conductivity in/out,
Btu/h ft F (kcal/h m C)
10. Liquid properties
Density out, lb/ft3 (kg/m3)
Viscosity out, cp
Specific heat out,
Btu/lb F (kcal/kg C)
Thermal conductivity out,
Btu/h ft F (kcal/h m C)

Original design for


New design for
Case Study 6.1
Case Study 6.5
Pentane
264,600 (120,000)
169.2 (76.2)/165.2 (74)
239 (115)/165.2 (74)
49.8 (3.5)
1.0/0.01
2.84 (0.2)
0.00146 (0.0003)
37.5 (9.44)
46.7 (11.77)
0.6 (9.609)/0.57 (9.1)
0.53 (8.43)/0.57 (9.1)
0.0085/0.0085
0.0094/0.0085
0.484/0.481

0.518/0.481

0.0114 (0.017)/
0.014 (0.021)/
0.0113 (0.0168)
0.0113 (0.0168)
35.3 (566)
0.144
0.632
0.0554 (0.0824)

For the sake of a meaningful comparison, the overdesign in the new design was kept the
same as the earlier one, and the fan power consumption was only marginally higher.
Note how the tubeside heat transfer coefficient has come down from 413.7 Btu/h ft2 F
(2020 kcal/h m2 C) to 300.3 Btu/h ft2 F (1466 kcal/h m2 C), a reduction of 27.4%. This is
because in the desuperheating zone, especially in the dry-wall desuperheating zone, the
tubeside heat transfer coefficient is much lower than that in the condensing zone. The
variation in the tubeside heat transfer coefficient in Btu/h ft2 F (kcal/h m2 C) in the various
zones is as follows:
Dry-wall desuperheating zone: 73.4 (358) to 97.5 (476)
Wet-wall desuperheating: 122.2 (597) to 411.4 (2009)
Condensing: 555.3 (2711) to 720.9 (3520)
However, the overall heat transfer coefficient has changed far less, from 99.3 Btu/h ft2 F
(485 kcal/h m2 C) to 90.1 Btu/h ft2 F (440 kcal/h m2 C), which is a reduction of only
9.2%. This is because the airside thermal resistance is controlling far more than the tubeside thermal resistance (which is only to be expected in a condenser) and this value has
gone down by only about 2%.
Looking at the MTD, we notice that it has increased from 37.1F (20.6C) to 41.2F
(22.9C), an increase of 11.2%.
Thus, the net result is that for case Study 6.5, which is a heat duty 24.7% higher, the heat
transfer area is 24% higher. This corroborates what had been stated earlier in this section,
that the penalty (in the form of additional heat transfer area) associated with desuperheating
is usually rather small since the loss in the overall heat transfer coefficient is largely
compensated by the increase in the MTD. In fact, in the present instance, the increase in the
heat transfer area is slightly less than the increase in the heat duty! Note that for a
70

Table 6.5b: Principal construction and performance parameters of air-cooled condenser for
Case Study 6.5
Original design for
New design for
Case Study 6.1
Case Study 6.5
32
0.984 (25) 0.098 (2.5)
29.5 (9.0)
32.8 (10.0)
11 (433)
43 6
48 6
2 2.64 (67)
11,524 (1071)
14,289 (1328)
275,316 (25,587)
341,329 (31,722)
2 3.758
2 14 (4.267)
4.41 (2.0)
5.236 (2.375)
144.7 (62.6)
146.5 (63.6)
10.3 (3.15)
9.91 (3.02)
0.43 (10.9)
0.4 (10.2)
19.6 (14.6)
20.8 (15.5)
117.6 (87.6)
124.8 ((93))
413.7 (2020)
300.3 (1466)

1. No. of bays no. of bundles per bay


2. Tube OD thickness, in. (mm)
3. Tube length, ft (m)
4. Fin density, fins/in. (fins/m)
5. No. of tubes per row no. of rows
6. No. of tube passes tube pitch, in. (mm)
7. Total bare tube area, ft2 (m2)
8. Total extended area, ft2 (m2)
9. Fans per bay fan diameter, ft (m)
10. Total air flow rate, MM lb/h (MM kg/h)
11. Air outlet temperature, F (C)
12. Airside face velocity, ft/s (m/s)
13. Airside pressure drop, in. WC (mm WC)
14. Individual fan power, HP (kW)
15. Total fan power, HP (kW)
16. Heat transfer coefficient,
Tubeside
Btu/h ft2 F/ (kcal/h m2 C/kcal)
Airside
Overall
17. Thermal resistance, %

Airside
Tubeside
Fouling
Tube wall
18. Tubeside pressure drop, allow/calc., psi (kg/cm2)
19. MTD, F (C)
20. Overdesign, %

181.5 (886)

178 (869)

99.3 (485)

90.1 (440)

54.78
24.02
18.19
3.01
2.84 (0.2)/ 2.84 (0.2)
37.1 (20.6)
13.6

50.67
30.03
16.51
2.78
2.84 (0.2)/2.7 (0.19)
41.2 (22.9)
13.6

meaningful comparison, both the overdesign and the fan power consumption have been kept
virtually identical.
6.5.2 Subcooling
Subcooling is a rare phenomenon in air-cooled heat exchangers. More often than not, aircooled condensers are followed by shell-and-tube water-cooled trim condensers, so that
even if subcooling were required, it would be in the trim condenser and not in the aircooled condenser.
Subcooling in horizontal tubeside condensers (such as in air-cooled condensers) will be
efficient if the total mass velocity is high, when the condensation will progress from annular
to slug and finally to the full tube of liquid. However, if the total mass velocity is low, so that
the condensation progresses from annular to wavy and then to stratified, subcooling is
ineffective and difficult to predict accurately.
The usual problem of low velocity for subcooling that is found in horizontal shell-andtube condensers, however, can be addressed very effectively in air-cooled condensers by
dividing the subcooling zone into several passes, so as to yield a high tubeside liquid
velocity. Thus, if there are six tube rows in an air-cooled condenser, the first pass can have
three tube rows, the second pass two tube rows, and the final tube row can be divided into
71

two, three, or even four passes such as to achieve a decent liquid velocity and thereby a
satisfactory heat transfer coefficient. This is eminently feasible since the pressure drop in the
subcooling zone is far, far less than that in the condensing zone.

6.6 Nozzle Sizing


Nozzles are sized on the basis of v2, which determines the pressure drop and the tendency to erode. The latter assumes added significance for condensers, especially for saturated vapors. This is because there may be a few droplets of condensate in the vapor
stream entering a condenser, which will then travel at the vapor velocity, produce an extremely high v2, and thereby tend to erode the tubes.
It is a usual practice to limit the total pressure drop in the inlet and outlet nozzles of a
heat exchanger to 1520% of the total allowed, and the same holds true for condensers. This
is to leave sufficient pressure drop for the tubes proper, so that the heat transfer coefficient
can be maximized. Since there is expansion at the channel inlet and contraction at the
channel outlet, the pressure drop in the inlet nozzle is always greater than that in the outlet
nozzle, for the same v2. A v2 of 13442016 lb/ft sec2 (20003000 kg/m sec2) is usually
considered for the sizing of nozzles for gas or condensing services, since normally 1.42.8
psi (0.10.2 kg/cm2) is permitted for such services. For liquid coolers, where the allowable
pressure drop is generally 710 psi (0.50.7 kg/cm2), a v2 of 3360 lb/ft sec2 (5000 kg/m
sec2) is quite normal and 4030 lb/ft sec2 (6000 kg/m sec2) an upper limit. It must be evident
that the higher the value of v2 permitted, the smaller will be the size of a nozzle in a given
situation.
It must be stated here that it is very important to size condensate nozzles for proper
drainage, so as to provide weir type flow, rather than a flooded drain pipe. In the latter
situation, flooding may result, with a deleterious effect on the performance of the condenser.
In order to achieve weir type flow, it is a common practice to size steam condensate nozzles
for a velocity of 2.0 ft/s (0.6 m/s), and hydrocarbon condensate nozzles for a velocity of 3.3
4.0 ft/s (11.2 m/s).

In single-phase applications, the variation of heat


duty versus temperature, or
the heat release profile, is
essentially linear. Essentially, because vapor or
liquid specific heat does
increase with temperature,
but this variation is not
large enough to impart an
appreciable curvature to the
heat release profile.
However, in the case of
condensers, there is usually
an appreciable curvature in
the heat release profile. This

Temperature, oC

6.7 Condensing Profiles


and MTD

Length of condenser

Fig. 6.5a Condensing profile of a multicomponent mixture


72

Temperature, OC

70
60
50
40

Length of Condenser
Fig. 6.5b Condensing profile of a pure component

AB: Desuperheating, BC:


Condensing, CD: Subcooling

Temperature, OC

Length of Condenser
Fig. 6.5c Desuperheating, condensing, and subcooling profile

is particularly true when the condensing stream is a multicomponent mixture. Due to the
effect of partial pressures, the less volatile components will condense first, and the more
volatile components later. Thus, since more heat duty will be transferred per unit of
temperature difference at the hotter end of the condenser than at the colder end, the slope of
the curve will be steeper at the hotter end than at the colder end, as shown in Fig. 6.5a.
However, a pure component condenser will have an isothermal heat release profile, as
shown in Fig. 6.5b.
The situation is further complicated when desuperheating and/or subcooling zones are
also present. A typical desuperheating, condensing, and subcooling situation is shown in
Fig.6.5c. Evidently, if the condensing stream is a pure component, the profile in the
condensing zone will be linear.
It is extremely important to feed the condenser heat release profile to the thermal design
73

software. The entire profile will have to be divided into a sufficiently large number of
virtually straight-line segments, and the calculations performed segmentwise. This is
important because not only the MTD, but even the tubeside condensing heat transfer
coefficient, will vary significantly from inlet to outlet. The higher condensing duty at the
inlet end will produce a correspondingly higher heat transfer coefficient.
Besides heat duty, the vapor weight fraction will also have to be fed against temperature.
This can be easily understood, since the relative amounts of vapor and liquid in the
condensing stream will determine the heat transfer coefficient and pressure drop in various
segments (increments of tube length).

74

CHAPTER 7

Optimization of Thermal Design


of Air-Cooled Heat Exchangers
Optimization of thermal design is a much more challenging and rewarding task for aircooled heat exchangers than for shell-and-tube heat exchangers. In shell-and-tube heat
exchangers, the cooling water flow rate is virtually fixed by the cooling water inlet temperature and the maximum cooling water outlet temperature, usually a temperature difference of 1821.6F (1012C). However, in the case of air-cooled heat exchangers, the
air flow rate is much more flexible and has to be optimized carefully.
Besides the air flow rate itself, the other variables are tube length, tube OD, fin height,
fin spacing, number of tube rows, fan power consumption, tube pitch, and number of tube
passes that have to be optimized in terms of the lowest total cost (fixed cost plus operating
cost). Another complication is that the fan size is linked to the tube length and section width,
thereby making the total optimization rather difficult. Finally, the number of tube bundles
has to be an even number so that we have a whole number of bays. The use of a
sophisticated computer program thus becomes essential for this task. Let us consider the
individual variables in more detail.

7.1 Tube Length


The number of sections (or bays) should evidently be minimized. This is similar to the
number of shells in series/parallel in shell-and-tube heat exchangers. Usually, there are
two tube bundles in a section, although there can be more for very large units. Bundle
width is usually limited to 10.511.5 ft. (3.23.5 m) for reasons of convenience of transportation. Thus, a section having the usual two tube bundles is 2123 ft (6.47 m) wide.
A fan diameter of 12 ft (3.657 m), 13 ft (3.962 m), or 14 ft (4.267 m) fits in quite comfortably in this width. Since there must be at least two fans along the tube length (as per
the code API 661 for air-cooled heat exchangers), a tube length of 26.232.8 ft (810 m)
will satisfy the minimum 40% fan coverage criterion (also as per API 661).
Sometimes, an air-cooled heat exchanger section may become highly rectangular (that
is, the length-to-width ratio is 2.5, or even more). This may arise in the case of a large plant
having a large pipe-rack width (with a consequent large tube length) and low-duty air-cooled
heat exchangers. In such cases, three fans may be used along the tube length. Since the cost
increases in such designs, a three-fan-per-section design should be chosen as a last resort.
Since air-cooled heat exchangers are usually located on top of a pipe rack and across it,
the tube length is determined by the pipe rack width. The tube length is usually 1.64 ft (0.5
m) greater than the pipe rack width for convenience in affixing the plenum chambers. Thus,
if the pipe rack width is 26.2 ft. (8 m), the tube length will be 27.9 ft (8.5 m). If the plant is
75

Table 7.1a: Principal process parameters for air-cooled condenser for Case Study 7.1
1. Fluid
2. Flow rate, lb/h (kg/h)
3. Temperature in/out, F (C)
4. Operating pressure, psia (kg/cm2a)
5. Total allowable pressure drop, psi (kg/cm2)
6. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
7. Heat duty, MM Btu/h (MM kcal/h)
8. Vapor properties
Density in/out, lb/ft3 (kg/m3)
Viscosity in/out, cp
Specific heat in/out, Btu/lb F (kcal/kg C)
Thermal conductivity in/out,
Btu/h ft F (kcal/h m C)
9. Liquid properties
Density in/out, lb/ft3 (kg/m3)
Viscosity in/out, cp
Specific heat in/out, Btu/lb F (kcal/kg C)
Thermal conductivity in/out, Btu/h ft F
(kcal/h m C)

Distillation column overhead


1,025,100 (465,000)
255 (124)/131 (55)
55.5 (3.9)
2.0 (0.14)
0.001 (0.0002)
251.2 (63.3)
0.52 (8.25)/0.41 (6.53)
0.009/0.0088
0.46/0.43
0.0134(0.02)/0.0119 (0.0178)
40.65 (651)/ 42.3 (678)
0.36/0.58
0.54/0.51
0.053 (0.079)/0.07 (0.105)

much smaller and the pipe rack width is only 19.7 ft (6 m), a tube length of 21.3 ft (6.5 m)
will be adopted. Evidently, smaller fans of, say, 8.2 ft (2.5 m) will have to be employed in
this case.
Sometimes, air-cooled heat exchangers are grade mounted. In such cases, the above
limitation will not apply and a better optimization of the tube length can be realized. It is
evident that the number of tube bundles will have to be an even number since two of them
are combined to form one section.

7.2 Tube OD
Just as in shell-and-tube heat exchangers, an air-cooled heat exchanger will also be
cheaper the smaller the tube diameter. However, the smaller the tube diameter, the more
difficult is the cleaning of the tubes. A 1 in. (25 mm) OD is the smallest tube diameter
recommended by API 661 (Section 4.1.11.1) and most air-cooled heat exchangers are
constructed with tubes of this size. However, in pressure drop limiting cases, a larger tube
diameter may yield a better design. For example, with 0.984 in. (25 mm) OD tubes, a particular design may give a pressure drop exceeding the permitted value with two tube
passes. If the number of tube passes is reduced to one, the pressure drop will reduce to
approximately one-seventh the earlier value and therefore become okay. However, the
tubeside heat transfer coefficient will reduce appreciably, thus requiring a higher heat
transfer area. In this situation, 1.26 in. (32 mm) OD tubes with two tube passes may give
a much more economical design.

CASE STUDY 7.1 OPTIMIZATION OF TUBE OD


A distillation column overhead condenser was to be designed for a refinery. The principal
process parameters are elaborated in Table 7.1a. The tubes were to be carbon steel. Since
the pipe-rack width was to be 12 m, the tubes were to 41 ft (12.5 m) long.
A first design was attempted with 0.984 in. (25 mm) OD and 0.098 in. (2.5 mm) thick
tubes since the smallest tube diameter gives the most economical design. A configuration
76

Table 7.1b: Principal construction and performance parameters of air-cooled heat exchanger for Case
Study 7.1
First run
Second run
Third run
1. No. of bays
8
8
8
2. No. of bundles per bay
2
2
2
3. Tube OD, in. (mm)
0.984 (25)
4. Tube thk length, in. (mm)
0.098 (2.5) 492 (12500)
5. Fin height, in. (mm) fin density,
0.63 (16) 11 (433)
fins/in. (fins/m)
6. No. of tubes per row no. of rows
48 6
48 8
7. No. of tube passes tube pitch, in. (mm) 2 2.64 (67) 1 2.64 (67)
1 2.64 (67)
8. Total bare tube area, ft2 (m2)
45,483 (4427)
63,452 (5897)
9. Fans per bay fan diameter, ft (m)
2 15 (4.57)
10. Total air flow rate, MM lb/h (MM kg/h)
17.42 (7.9)
17.75 (8.05)
11. Air outlet temperature, F (C)
148.1 (68.3)
153.9 (67.7)
12. Airside velocity, ft/s (m/s)
18.2 (5.54)
18.64 (5.68)
19.06 (5.81)
13. Airside pressure drop, in. WC (mm 0.41 (10.5)
0.42 (10.6)
0.57 (14.4)
WC)
14. Individual fan power, HP (kW)
26.7 (19.9)
26.9 (20.1)
35.3 (26.3)
15. Total fan power, HP (kW)
427.2 (318.4) 430.4 (321.6)
564.8 (420.8)
16. Tubeside pressure drop, psi (kg/cm2)
4.69 (0.33)
0.85 (0.06)
0.57 (0.04)
17. Heat transfer coefficient,
Tubeside
191 (933)
124 (606)
105.8 (517)
Btu/h ft2 F/ (kcal/h m2
Airside
177.2 (865)
178.7 (872)
180.4 (881)
C/kcal)
Overall
80.6 (393)
65.8 (322)
60.5 (296)

18. Overdesign, %

8.8

18.6

0.2

Final run
8
2
1.22 (31)

45 6
2 2.87 (73)
55,349 (5144)
17.0 (7.7)
156.6 (69.2)
19.0 (5.78)
0.47 (12.0)
28.7 (21.4)
459.2 (342.4)
1.7 (0.12)
145.3 (710)
158.1 (772)
68.1 (333)
5.6

with eight bays having two tube bundles each was made. The other construction and
performance details are given in Table 7.1b. With two tube passes, the tubeside pressure
drop of 4.7 psi (0.33 kg/cm2) far exceeded the allowable limit of 2.0 psi (0.14 kg/cm2), and
was therefore not acceptable.
In an attempt to reduce the tubeside pressure drop, the number of tube passes was
decreased from two to one. As expected, there was a sharp fall in the tubeside pressure drop
from 4.7 psi (0.33 kg/cm2) to 0.85 psi (0.06 kg/cm2), which was well within the permitted
value. However, since the tubeside heat transfer coefficient also fell sharply from 191 Btu/h
ft2 F (933 kcal/h m2 C/kcal) to 124 Btu/h ft2 F (606 kcal/h m2 C/kcal), the condenser
became 18% undersurfaced.
The number of tube rows had to be increased from six to eight to make the condenser
design adequately surfaced. Construction details of this and the earlier design are also shown
in Table 7.1b.
Evidently, this design with 0.984 in. (25 mm) OD tubes and a single tube pass is
uneconomical since the permitted tubeside pressure drop is very poorly utilized, thereby
resulting in a rather low tubeside heat transfer coefficient and consequently a large heat
exchanger. Therefore, a design was attempted with 1.22 in. (31 mm) OD and 0.098 in. (2.5
mm) thick tubes, using two tube passes. The design was found to be far more economical
since the heat transfer area reduced appreciably, from 63,452 ft2 (5897 m2) to 55,349 ft2
(5144 m2). This is because the permitted tubeside pressure drop was almost fully utilized so
that the tubeside and thereby the overall heat transfer coefficient were much higher.
Construction details of the revised design are also given in Table 7.1b. Needless to say, this
design was selected as the final one.
77

Table 7.2a: Principal process parameters for air-cooled tempered


water cooler for Case Study 7.2
1. Fluid
2. Flow rate, lb/h (kg/h)
3. Temperature in/out, F (C)
4. Operating pressure, psia (kg/cm2a)
5. Total allowable pressure drop, psi (kg/cm2)
6. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
7. Heat duty, MM Btu/h (MM kcal/h)

Tempered (hot) water


485,000 (220,000)
249.8 (121)/140 (60)
78.2 (5.5)
7.1 (0.5)
0.00098 (0.0002)
53.34 (13.44)

7.3 Fin Height


The usual fin heights are 3/8 in. (9.5 mm), 1/2 in. (12.7 mm), and 5/8 in. (15.875 mm),
with the last two being by far the most popular. The selection really depends on the relative values of the airside and the tubeside heat transfer coefficients. Where the airside
heat transfer coefficient is controlling (that is, it is the major resistance to heat transfer), a
larger fin height of 5/8 in. (15.875 mm) will usually result in a better design. If, however,
the tubeside heat transfer coefficient is controlling, it will be prudent to use a smaller fin
height of 1/2 in. (12.7 mm).
A higher fin height leads to a greater efficiency of conversion of pressure drop to heat
transfer on the airside. However, the higher tube pitch means that a lower number of tubes
can be accommodated per row for the same bundle width. In cases where the airside heat
transfer coefficient is controlling, a higher heat transfer coefficient achieved by use of a
higher fin height results in a significant increase in the overall heat transfer coefficient,
thereby reducing the heat transfer area and thus the number of tubes per row. Thus, it usually
becomes more economical to use a higher fin height for services where the airside heat
transfer coefficient is controlling, such as in steam condensers and water coolers. For gas
coolers and viscous liquid hydrocarbon liquid coolers, however, the tubeside heat transfer
coefficient is controlling and hence the use of a smaller fin height is economically
advantageous.
The above logic is now illustrated by two case studies, one in which the airside heat
transfer coefficient is controlling and the other in which the tubeside heat transfer coefficient
is controlling.

CASE STUDY 7.2: OPTIMIZATION OF FIN HEIGHT:


AIRSIDE HEAT TRANSFER COEFFICIENT CONTROLLING
A closed-circuit tempered water cooler was to be designed in a refinery application. The
principal process parameters are detailed in Table 7.2a. Carbon steel tubes of 0.984 in.
(25 mm) OD and 41 ft (12.5 m) length were to be used. Fins were to be of aluminum of
the G or grooved type. The design ambient temperature was 107.6F (42C).
Two designs were prepared, one using 1/2 in. (12.7 mm) high fins and the other using
5/8 in. (15.875 mm) high fins. The overdesign and power consumption were kept the same
for a meaningful comparison. The salient details of the two designs are given in Table 7.2b.
As expected, since this is an airside heat transfer controlling case, the 5/8 in. (15.875
mm) high fin design proved to be more economical. For the same overdesign and power
consumption, the 5/8 in. (15.875 mm) fin height design had a significantly lower bare tube
area (11.5%). It is true that the finned area was greater (18%), but this represents a smaller
78

Table 7.2b: Principal construction and performance parameters of air-cooled heat


exchanger for Case Study 7.2

1. No. of bays
2. No. of bundles per bay
3. Tube OD thickness length, in. (mm)
4. Fin density, fins/in. (fins/m)
5. No. of tubes per row no. of rows tube pitch,
in. (mm)
6. No. of tube passes
7. Total bare tube area, ft2 (m2)
8. Total extended area, ft2 (m2)
9. Fans per bay fan diameter, ft (m)
10. Total air flow rate, MM lb/h (MM kg/h)
11. Air outlet temperature, F (C)
12. Airside face velocity, ft/s (m/s)
13. Airside pressure drop, in. WC (mm WC)
14. Individual fan power, HP (kW)
15. Total fan power, HP (kW)
16. Tubeside velocity, ft/s (m/s)
17. Heat transfer coefficient,
Tubeside
Btu/h ft2 F/ (kcal/h m2 C/kcal)
Airside
Overall
2

18. Tubeside pressure drop, psi (kg/cm )


19. MTD, F (C)
19. Overdesign, %

Fin height, in. (mm)


0.5 (12.7)
0.625 (15.875)
2
2
2
2
0.984 (25) 0.098 (2.5) 492 (12500)
11 (433)
Alternately 42 and
48 4 2.36 (60)
43 4 2.64 (67)
4
4
8283 (739)
7037 (654)
140,890 (13,094)
166,380 (15,463)
2 15 (4.572)
2 15 (4.57)
5.093 (2.31)
5.24 (2.376)
151 (66.1)
149.7 (65.4)
12.86 (3.92)
13.4 (4.08)
0.5 (12.65)
0.47 (11.86)
40 (29.8)
39.8 (29.7)
160 (119.2)
159.2 (118.8)
3.4 (1.04)
3.9 (1.19)
939.9 (4589)
1038 (5070)
174.4 (851.4)
199.9 (976)

120 (586)

133.3 (651)

5.75 (0.404)
58.9 (32.7)
5.4

0.5
60.1 (33.4)
6.2

increase in cost than the saving due to the lower tube cost. (The fabrication cost, which
represents a major component of the cost of a finned tube, depends on the length of tubing to
be finned and does not vary significantly with fin height.) Since in the 5/8 in. (15.875 mm)
fin height design there will be 11.5% less tubes to fins, the overall cost of this design will be
less since both the tube cost and the cost of finning will be lower.

CASE STUDY 7.3: OPTIMIZATION OF FIN HEIGHT:


TUBESIDE HEAT TRANSFER COEFFICIENT CONTROLLING
A bottom stream in a refinery had to be cooled to its run-down temperature. The principal
process parameters are detailed in Table 7.3a.
The specified tube size was 1 in. (25 mm) OD, .1 in. (2.5 mm) thick, and 41 ft (12.5 m)
long. The tube material was carbon steel.
Since the tubeside viscosity is rather high, the tubeside heat transfer coefficient will be
controlling. Therefore, a design was made with 1/2 in. or 12.7 mm high fins and 7 fins/in. or
276 fins per meter (see the following section for the logic for not using the maximum fin
density of 11 per in. or 433 per meter), as shown in Table 7.3b. The airside heat transfer
resistance was only 12.2% of the total resistance, whereas the tubeside heat transfer
resistance was 82.48%.
Next, an attempt was made to prepare a design with 5/8 in. (16 mm) high fins. The
number of tubes per row was reduced from 50 to 45 so that the bundle width would be
79

Table 7.3a: Principal process parameters for Case Study 7.3


1. Fluid
2. Flow rate, lb/h (kg/h)
3. Temperature in/out, F (C)
4. Total allowable pressure drop, psi (kg/cm2)
5. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
6. Heat duty, MM Btu/h (MM kcal/h)
7. Liquid properties
Density in/out, lb/ft3 (kg/m3)
Viscosity in/out, cp
Specific heat in/out, Btu/lb F (kcal/kg C)
8. Design air temperature, F (C)

Hydrocarbon liquid
205,700 (93,300)
356 (180)/158 (70)
22.8 (1.6)
0.00293 (0.0006)
22.86 (5.76)
820/890
4.2/15
0.58/0.54
107.6 (42)

Table 7.3b: Principal construction and performance parameters for Case Study 7.3

1. No. of bays
2. No. of bundles per bay
3. Tube OD thk length, in. (mm)
4. Fin density, fins/in. (fins/m)
5. No. of tubes per row no. of rows tube
pitch, in. (mm)
6. No. of tube passes

7. Total bare tube area, ft2 (m2)


8. Total extended surface area, ft2 (m2)
9. Fans per bay fan diameter, ft (m)
10. Total air flow rate, lb/h (kg/h)
11. Air outlet temperature, F (C)
12. Fan power consumption, HP (kW)
13. Tubeside pressure drop, psi (kg/cm2)
14. Overdesign, %

Fin height, in. (mm)


0.5 (12.7)
0.625 (15.875)
2
2
2
2
0.984 (25) 0.098 (2.5) 492 (12500)
7 (276)

50 8 2.36 (60)

45 9 2.64 (67)

10 (2 each in the first 2


10 (2 each in the first
rows, one each in the other
2 rows, one each in
6 rows)
the other 7 rows)
16,527 (1536)
16,720 (1554)
193,056 (17,942)
53,000 (24,040)
2 15 (4.572)
2 15 (4.572)
3,530,000 (1,600,00)
134.4 (56.9)
4 19.3 = 77.2
21.9 (1.54)
19.8 (1.39)
5.7
5.7

retained the same as in the base design, since the tube pitch was now 2.64 in. (67 mm) as
against the earlier 2.36 in. (60 mm). The airside flow rate was adjusted to 1660,000 kg/h so
as to yield the same power consumption. However, the tubeside pressure drop became
excessive at 1.85 kg/cm2. Reducing the number of tube passes to nine resulted in the design
becoming undersurfaced by 3.84%. Unfortunately, the number of tubes per row could not be
increased since the bundle width was already the permitted maximum of 3.2 m. As can only
be expected since the airside is not controlling, increasing the air flow rate made no
significant improvement in the performance. Therefore, the only alternative was to increase
the number of tube rows from eight to nine. The number of tube passes was retained as nine
(two in the first row and one each in the subsequent seven rows) and the air flow rate
reduced from 1,660,000 to 1,600,000 kg/h to have the same airside power consumption.
This design is elaborated in the second column of Table 7.3b. It is acceptable since it is
5.7% oversurfaced and the tubeside pressure drop 1.39 kg/cm2. However, while the bare
tube area was only marginally higher than that of the original design (1554 m2 versus 1532
m2), the finned area was considerably higher (24,037 m2 versus 17504 m2). Consequently,
the first design is lower in first cost and was therefore accepted.
80

Table 7.4: Variation in airside heat transfer coefficient, pressure drop and fan
power with fin density for Case Study 7.4
1. No. of fins/in. (fins/m)

11 (433)

10 (394)

9 (354)

8 (315)

7 (276)

2. Air pr. drop, in. (mm) WC

0.47
(11.86)
199.9
(976)
39.8
(29.7)
133.3
(651)
82.29

0.44
(11.29)
4.8
183.9
(898)
8.0
38.5
(28.7)
3.4
126.2
(616)
79.54

0.42
(10.73)
5.0
167.5
(818)
8.9
37.1
(27.64)
3.69
118.2
(577)
76.23

0.4
(10.19)
5.0
151.6
(740)
9.5
35.7
(26.7)
3.4
110
(537)
72.59

0.38
(9.66)
5.2
135.4
(661)
10.7
34.5
(25.7)
3.75
101.2
(494)
68.45

54.92

54.53

53.77

52.68

51.14

3. Stepwise reduction
4. Air heat transfer coefficient,
Btu/h ft2 F/ (kcal/h m2 C/kcal)
5. Stepwise reduction
6. Fan power, HP (kW)
7. Stepwise reduction
8. Overall heat transfer coefficient,
(kcal/h m2 C/kcal)
9. Air heat transfer coefficient/Air
pr. drop (metric)
10. Overall heat transfer
coefficient/Air pr. drop (metric)

An unusual situation
An unusual situation arises sometimes where, although the tubeside heat transfer coefficient is controlling, design with a higher fin height proves to be more economical. This
happens when for a given design with 1/2 in. (12.7 mm) high fins, the tubeside pressure
drop is not fully utilized and therefore the tubeside heat transfer coefficient is not maximized. Any increase in the number of tube passes results in the tubeside pressure drop
exceeding the allowable limit. Here, by switching over to 5/8 in. (15.875 mm) high fins,
the number of tubes can be reduced by about 10%, thereby leading to an approximately
8% increase in the tubeside heat transfer coefficient. Thus, the use of tubes with 5/8 in.
(15.875 mm) high fins results in a more economical design.

7.4 Fin spacing


The logic described above for fin height is also applicable herea higher fin density is
economically favorable for services where the airside heat transfer coefficient is controlling. Furthermore, fin spacing can be optimized while fine tuning a design. It usually varies between 7 and 11/in. (276 and 433/m). A higher fin density (number of fins per unit
length) will result in a higher airside heat transfer coefficient as well as a higher airside
pressure drop. This same is demonstrated in the following case study.

CASE STUDY 7.4: VARIATION OF AIRSIDE HEAT TRANSFER COEFFICIENT


AND PRESSURE DROP WITH FIN DENSITY
For the second design of the tempered water cooler [with 5/8 in. (15.875 mm) high fins]
presented above (Case study 7.2), the variation in the airside heat transfer coefficient and
pressure drop with fin density are shown in Table 7.4.
It is seen that for a typical air mass velocity, the reduction in the air heat transfer
coefficient is significantly greater than that in the air pressure drop and power consumption.
81

The ratio of air heat transfer coefficient to air pressure drop, which is a measure of the
efficiency of heat transfer, drops significantly as the fin density is reduced from 82.29 for 11
fins/in. (433 fins/m) to 68.45 for 7 fins/in. (276 fins/m).The ratio of the overall heat transfer
coefficient to the airside pressure drop also reduces, although at a much slower rate, as the
fin density is reduced from 11 fins/in. to 7 fins/in.
A higher fin density is always more efficient at converting pressure drop to heat transfer.
Therefore, whenever the airside heat transfer coefficient is controlling, a higher fin density
should be employed. However, where the airside heat transfer coefficient is not controlling,
it may be sensible to use a lower fin density to reduce power consumption. This is because a
higher airside heat transfer coefficient will not result in a significant increase in the overall
heat transfer coefficient, so that the increased power consumption will not be justified.

7.5 Number of Tube Rows


The minimum number of tube rows recommended for establishing a proper air flow pattern is four, although three tube rows can also be used in exceptional circumstances while
maintaining a higher overdesign. Most air-cooled heat exchangers have four to six tube
rows, although eight or even ten rows may be used occasionally.
The advantage of using a higher number of tube rows is that more heat transfer area can
be accommodated in the same bundle width. This leads to a reduction in the number of tube

10
9
8
7
6
5
4
3

an
t Tr
Hea

r
sfe

t
ien
ffic
e
o
C

1
1

8 9 10

Air Mass Velocity

Fig. 7.1 Variation of airside heat transfer coefficient, airside pressure drop, and fan power consumption with airside velocity
82

bundles and sections, thereby reducing the cost and the plot area of the air-cooled heat exchanger. The disadvantage is that besides increasing the fan horsepower (for the same air
velocity), it reduces the airside flow area for a given heat transfer area and thereby the air
flow rate itself, thus lowering the MTD. In fact, this is one of the fascinating aspects of aircooled heat exchanger designeven the coolant flow rate is a variable and has to be
optimized. For water-cooled heat exchangers, the cooling water flow rate is virtually fixed,
depending on its inlet temperature and the permitted maximum outlet temperature.
The airside pressure drop increases directly with the number of tube rows and so does
the heat transfer area. However, as the number of tube rows increases, the airside velocity
will have to be reduced for the same fan horsepowerthis will translate into a lower airside
heat transfer coefficient. This is why the number of tube rows has to be optimized carefully.
For the sake of convenience, this may be summarized by stating that the larger the number
of tube rows, the lower will be the airside heat transfer coefficient.
It follows from the above explanation that when the tubeside heat transfer is controlling,
a larger number of tube rows may be used since the lower airside heat transfer coefficient
will not hurt the design, since it is not controlling in the first place. However, when the
airside heat transfer coefficient is controlling, the number of tube rows will have to be
limited to a lower value for an optimum design. Thus, for services such as condensers, water
coolers and light hydrocarbon liquid coolers having a high tubeside heat transfer coefficient,
the optimum number of tube passes is likely to be four or five. However, for gas coolers and
viscous liquid coolers, the same is likely to be six, seven, or even higher. The maximum
number of tube rows is generally eight to ten, since with a larger number of tube rows, the
airside heat transfer coefficient will be inordinately low (due to the low velocity sustainable
for a normal power consumption) or the airside pressure drop will be inordinately high (if a
normal airside velocity is used).

7.6 Fan Power Consumption


Fan power varies directly with the volumetric air flow rate and the pressure drop. The
volumetric air flow rate varies directly with the air mass velocity, while the pressure drop
varies to the 1.75 power of the air mass velocity. Thus, the air pressure drop varies to the
1.75 power and fan power to the 2.75 power of the air mass velocity. In sharp contrast,
the air heat transfer coefficient varies to the 0.5 power of the air mass velocity. This is
represented graphically in Fig. 7.1.
Thus, when air mass velocity is increased, the air pressure drop and the fan power
increase rather sharply (especially the latter) whereas the air heat transfer coefficient
increases at a much slower rate. It therefore follows that there will be an optimum air mass
velocity beyond which any further increase will be wasteful. Evidently, this optimum air
mass velocity will depend on the extent to which the airside heat transfer coefficient is
controlling. In a situation where the airside heat transfer coefficient is highly controlling, this
optimum air mass velocity will be higher than in a situation where the airside heat transfer
coefficient is much less controlling.

CASE STUDY 7.5: OPTIMIZATION OF FAN POWER CONSUMPTION


Let us illustrate this with the help of an actual example. For the sake of convenience, we
will ignore the tube metal resistance and the airside fouling resistance (which, in any
case, are negligible) and the tubeside fouling resistance.
83

Table 7.5a: Effect of increased air flow rate in an airside controlling design

Heat transfer coefficient,


Btu/h ft2 F (kcal/h m2
C/kcal)

Airside

Base case
163.9 (800)

With 25% increase


in air flow rate
183.2 (894.4)

Tubeside

819.3 (4000)

819.3 (4000)

Overall

136.6 (667)

149.7 (731)

9.6

0.47 (12)
24.1 (18)

0.7 (17.74)
44.6 (33.25)

47.8
84.7

Airside pressure drop, in. (mm) WC


Fan power consumption, HP (kW)

% increase
11.8

Table 7.5b: Effect of increased air flow rate in a tubeside controlling design

Heat transfer coefficient,


Btu/h ft2 F (kcal/h m2 C/kcal)

Airside
Tubeside
Overall
Airside pressure drop, in. (mm) WC
Fan power consumption, HP (kW)

Base case
163.9 (800)
819.3 (4000)
39.0 (190.5)
0.47 (12)
24.1 (18)

With 25% increase


in air flow rate
183.2 (894.4)
819.3 (4000)
40.0 (195.4)
0.7 (17.74)
44.6 (33.25)

% increase
11.8

2.6
47.8
84.7

Case 1 (Airside controlling)


Let us consider a typical air-cooled heat exchanger where the airside thermal resistance is
controlling, such as a water cooler. The relevant performance parameters are portrayed in
the first column of Table 7.5a.
Now, let us consider a 25% increase in the air mass flow rate. The second column of
Table 7.5b indicates how the selected performance parameters will increase. (With increase
in air flow rate, the airside heat transfer coefficient increases to the power 0.5, the airside
pressure drop to the 1.75 power, and the fan power consumption to the 2.75 power.) Note
that the overall heat transfer coefficient has increased from 136.6 to 149.7 Btu/h ft2 F (667
to 731 kcal/h m2 C), which is an increase of 9.6%. This has been achieved at the expense of
a 48% increase in air pressure drop and an 85% increase in fan power consumption.
Case 2 (Tubeside controlling)
Now, let us consider a typical air-cooled heat exchanger where the tubeside thermal resistance is controlling, such as a low-pressure gas cooler or a wide-range condenser. For the
sake of comparison, let us consider the same airside performance parameters as in Case 1
but, of course, a much lower tubeside heat transfer coefficient and thereby an overall heat
transfer coefficient. These parameters are portrayed in the first column of Table 7.5b.
Now, let us consider a 25% increase in the air mass flow rate as in the previous example.
The second column of Table 7.5b indicates how the selected performance parameters will
vary. While the airside parameters vary precisely as in the previous example, the overall heat
transfer coefficient increases from 39.0 to 40.0 Btu/h ft2 F (190.5 to 195.4 kcal/h m2 C), an
increase of only 2.6%. This is because in this case the tubeside thermal resistance is
controlling.
Thus, for the same increase in air pressure drop and fan power consumption, the
84

increase in the overall heat transfer coefficient is 9.6% in Case 1 but only 2.6% in Case 2.
Therefore, while this increase in air flow may be justifiable in Case 1 (depending on the
actual situation of heat transfer area and air pressure drop/fan power consumption), it is
certainly unlikely to be justifiable in Case 2.

7.7 Tube Pitch


Although tubes can be laid out on a staggered or in-line pattern, the former is almost invariably employed (as shown in Fig. 4.8) since it produces a much better performance
(conversion of pressure drop to heat transfer). Tube pitch has a very profound effect on
airside performance. It will be evident that the transverse pitch is more crucial and is
what is implied by the term tube pitch. The longitudinal pitch has a much less profound
influence and is usually 8090% of the transverse pitch, in an effort to minimize the
height of the tube bundles and thereby the cost. Designers tend to use the following standard tube OD/fin OD/tube pitch combinations since they tend to be optimum:
1 in./2 in./2.375 in. (25 mm/50 mm/60 mm)
1 in./2.25 in./2.625 in. (25 mm/57 mm/67 mm)
However, it should be understood that in many situations, these may not be the optimum
and the tube pitch should be varied and the optimum established and adopted.
The normal range of tube pitch for 1 in. tube OD/2 in. fin OD combination is 2.1252.5
in. Similarly, for a 1 in. tube OD/2.25 in. fin OD combination, the normal range of tube pitch
is 2.3752.75 in. It should be realized that at a relatively lower tube pitch, the air pressure
drop and therefore the power consumption tend to be high for the same airside heat transfer
coefficient. That is to say, since the tube pitch is decreased, the airside pressure drop and
power consumption increase more rapidly than does the airside heat transfer coefficient.

CASE STUDY 7.6: EFFECT OF TUBE PITCH AND FIN DENSITY


ON AIRSIDE HEAT TRANSFER COEFFICIENT AND PRESSURE DROP
A study was carried out for a typical air-cooled heat exchanger having 1 in. (25.4 mm)
OD tubes and 1/2 in. (12.7 mm) high fins for the following values of tube pitch:
2.125 in. (54 mm)
2.25 in. (57.2 mm)
2.375 in. (60.3 mm)
2.5 in. (63.5 mm)
The fin density was varied as follows: 7 fins/in., 9 fins/in., and 11 fins/in.
The results of the study are shown in Table 7.6. Since it is a very commonly used
configuration, the combination employing 2.375 in. (60.3 mm) pitch and 11 fins/in. (433
fins/m) was considered to be the base case, and 100 units were assigned for the values of
both airside heat transfer coefficient and airside pressure drop. The performance of the other
combinations (in terms of heat transfer coefficient and pressure drop) is expressed in
comparison to these datum values. Finally, a ratio of the airside heat transfer coefficient to
the airside pressure drop, which is a measure of the efficiency of heat transfer, has also been
indicated.
As expected, it was found that the reduction in pressure drop was much sharper than that
in the heat transfer coefficient. Typically, for a step change in tube pitch, the pressure drop
85

Table 7.6: Effect of tube pitch and fin density on the heat transfer coefficient and pressure drop
for Case Study 7.6
Tube
pitch, in.
(mm)
2.125
(54)
2.25
(57.2)
2.375
(60.3)
2.5 (63.5)

7 fins/in. (276 fins/m)


Ratio
HTC/PD
HTC
PD

9 fins/in. (354 fins/m)


Ratio
HTC/PD
HTC
PD

11 fins/in. (433 fins/m)


Ratio
HTC/PD
HTC
PD

76.4

91.7

0.833

90.6

109.8

0.825

104.7

129.5

0.808

74.5

81.1

0.919

88.7

96.2

0.922

101.9

112.9

0.903

73.1

72.7

1.006

86.8

86.4

1.005

100.0

100.0

1.0

72.2

65.9

1.096

85.4

77.3

1.105

98.1

89.4

1.097

HTC = Airside heat transfer coefficient


PD = Airside pressure drop

reduced by about 10% whereas the heat transfer coefficient reduced by only about 2%. This
variation was largely the same for the various fin densities. The designer may therefore feel
that the higher the tube pitch, the better the performance of an air-cooled heat exchanger.
While this is essentially true, there is another extremely important factor that should not be
overlooked. Since the tube pitch is increased for a given number of tubes, the tube bundle
width will increase (even the fan diameter will increase at a certain stage), thus pushing up
the cost of the air-cooled heat exchanger. Consequently, the tube pitch should be optimized
in the overall context.
Although the optimum tube pitch may vary from situation to situation, designers
generally prefer to use 2.375 in. (60 mm) tube pitch for a 1 in. tube OD/ 2 in. fin OD (25 mm
tube OD/50 mm fin OD) combination since it generally represents an optimum between the
two opposing tendencies of performance efficiency and cost. For a 1 in. tube OD/ 2.25 in.
fin OD (25 mm tube OD/57 mm fin OD) combination, the following tube pitch values may
be used: 2.375 in. (60 mm), 2.5 in.(64 mm), 2.625 in. (67 mm), and 2.75 in. (70 mm). Here,
the tube pitch most commonly used is 2.625 in. (67 mm).
It is important to realize that tube pitch is a powerful variable in the design of air-cooled
heat exchangers and can be varied in steps of even .04 in. (1 mm) and not necessarily 1/8 in.
(3.18 mm). It can be fine tuned, depending on the situation at hand. For example, it can be
reduced in order to accommodate the required number of tubes per row within the maximum
permitted tube bundle width (which is determined by the convenience of transportation) at
the expense of a somewhat higher power consumption. Similarly, it can be reduced or
increased in order to employ a certain standard fan diameter so as to meet the API 661
stipulation of a minimum fan coverage of 40%.

7.8 Number of Tube Passes


The distribution of tubes in the various passes need not be identical. Thus, in a six tube
row air-cooled heat exchanger, there can be four tube passes, with two rows each in the
upper two passes and one row each in the lower two tube passes. This feature is especially useful in condensers where the flow area in each pass can be gradually reduced
since the liquid fraction and therefore the mixture density increases progressively,
thereby obtaining a more uniform pressure drop in the various passes than would be possible in an even distribution of 1.5 tube rows in all passes. This results in a higher overall
86

Table 7.7a: Principal process parameters for air-cooled condenser for Case Study 7.7
1. Fluid
2. Flow rate, lb/h (kg/h)
3. Temperature in/out, F (C)
4. Operating pressure, psia (kg/cm2a)
5. Weight fraction vapor, in/out
6. Total allowable pressure drop, psi (kg/cm2)
7. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
8. Heat duty, MM Btu/h (MM kcal/h)

Atmospheric column overhead


1,1106,700 (502,000)
293 (145)/149 (65)
54.0 (3.8)
1.0/0.01
7.1 (0.5)
0.00098 (0.0002)
270.0 (68.0)

Table 7.7b: Variation of performance of air-cooled condenser with tube pass distribution for
Case Study 7.7

1. Tubeside heat transfer coefficient,


Btu/h ft2 F/(kcal/h m2 C/kcal)
2. Overall heat transfer coefficient,
Btu/h ft2 F/(kcal/h m2 C/kcal)
3. Mean temperature difference, F (C)
4. Overdesign, %
5. Tubeside pressure drop, psi (kg/cm2)
6. Ratio of tubeside heat transfer coefficient to tubeside pressure drop

1st pass,
4 rows; 2nd
pass, 1 row

1st pass,
2.5 rows;
1st pass,
1st pass,
3 rows; 2nd 2nd pass, 2.5 2 rows; 2nd
rows
pass, 3 rows
pass, 2 rows

473.1 (2310)

527.6 (2576)

547.5 (2673) 559.6 (2732)

98.47 (480.8) 101.05 (493.4) 101.9 (497.6) 102.4 (499.9)


21.46 (38.63)
0.82
6.87 (0.483)

22.17 (39.9)
6.88
7.05 (0.496)

4782

5194

22.18 (39.92) 21.99 (39.59)


7.84
7.45
9.02 (0.634) 13.05 (0.918)
4216

2973

pressure drop-to-heat transfer conversion, thereby resulting in a lower heat transfer area.
This can be better understood as follows: If the number of tubes per pass were to be identical in all the passes, the initial passes (where there is more vapor and consequently the
density is lower) will give a much higher pressure drop while not giving a correspondingly higher heat transfer coefficient. The subsequent tube passes will give a much lower
pressure drop. The design will therefore not be optimum as regards best utilization of
available pressure drop.

CASE STUDY 7.7: OPTIMIZATION OF PASS DISTRIBUTION


An atmospheric column overhead condenser was to be designed for the crude distillation
unit of a refinery, using air as the cooling medium. The salient process parameters of the
same are detailed in Table 7.7a.
A design was made having eight bays and two tube bundles per bay. Each tube bundle
had 48 tubes per row and 5 tube rows. Tubes were of carbon steel, 0.984 in. (25 mm) OD,
0.098 in. (2.5 mm) thick, and 41 ft (12.5 m) long, having aluminum fins 0.49 in. (12.5 mm)
high and a fin density of 11 per in. (433 per meter). The number of tube passes was two.
However, it is not necessary to distribute the tubes 50:50 in the two passes. Hence, four
different pass distributions were tried in order to see which would be optimum and the
findings are shown in Table 7.7b.
87

Since the allowable tubeside pressure drop is exceeded in the last two runs, those
designs are evidently not acceptable and we have to select either of the first two designs. If
we look at the ratio of the tubeside heat transfer coefficient to the tubeside pressure drop, we
notice that it is higher for the second run than for the first. Thus, run 2 is the best design of
the four, affording an overdesign of 6.88% while run 1 has an overdesign of just 0.88%.
It should be noted that between the various designs, the variation in the tubeside
pressure drop is far greater than the variation in the tubeside heat transfer coefficient. This is
consistent with what we have been saying throughout this book; namely, with variation in
mass velocity, the pressure drop varies far more profoundly than the heat transfer coefficient
It should also be noted that for a condenser, having a full row in one pass usually gives a
better tubeside performance than having a fractional rows/pass arrangement. This means a
higher tubeside heat transfer coefficient for the same pressure drop or a lower pressure drop
for the same heat transfer coefficient. Thus, for a six row/four pass design, it is better to have
two passes of two rows each and two passes of one row each than to have four passes of 1.5
rows each.

Further Reading
1.
2.
3.

Baker, W.J., 1980, Selecting and Specifying Air-cooled Heat Exchangers, Hydrocarbon
Process., (May).
Monroe, R.C., 1985, Minimizing Fan Energy Costs: Part 1, Chem. Eng., May 27, pp. 141
142
Monroe, R.C., 1985, Minimizing Fan Energy Costs: Part 2, Chem. Eng., June 24, pp. 141
142

88

CHAPTER 8

Physical Properties
and Heat Release Profiles
The importance of feeding authentic physical properties cannot be overemphasized. The
author has seen numerous instances wherein the feeding of incorrect properties has produced incorrect results. Besides physical properties themselves, heat release profiles are
also very important, and must be fed accurately and meaningfully in order to produce
realistic and consistent results.
It must be stated here that when the feeding of an incorrect physical property produces
an absurd or exaggerated heat transfer coefficient or pressure drop, the experienced designer
may often be able to sense it, because these values will be well beyond the expected range.
But not so with the inexperienced or less experienced designer. Consequently, it is strongly
recommended that the input values be checked carefully in order to eliminate any probability
of incorrect results.

8.1 Physical Properties


Please refer to Section 5.4.2 where a brief discussion of physical properties was presented. Viscosity, thermal conductivity, specific heat, and density are the fundamental
properties in single-phase applications. Additionally, for condensing services, surface
tension also assumes significance.
It must be stated at the outset that physical properties should be furnished at both the
inlet and outlet temperatures, especially if the difference between the two temperatures is
highin which case the values of these properties are likely to vary appreciably.
The physical property that varies the most with temperature is liquid viscosity,
especially in the case of heavy liquids. To illustrate this, if the inlet temperature of a stream
to a heat exchanger is 140F (60C) and the outlet is 104F (40C), the variation is not large,
and even average physical properties may be fed without entailing any serious error.
However, if the inlet temperature is 284F (140C) and the outlet 140F (60C), it would be
inadvisable not to feed physical properties at both these temperatures.
In many designs, the heat transfer coefficient varies considerably from inlet to outlet,
primarily due to the variation of liquid viscosity. Coupled with this is the fact that the MTD
may also vary significantly from inlet to outlet, since the temperature approach (temperature
difference between the process stream and air) is invariably larger at the exchanger inlet than
at the exchanger outlet. If one considers the example of a viscous liquid cooler, both the heat
transfer coefficient and the MTD will reduce considerably from the exchanger inlet to the
exchanger outlet, so that the heat transfer area required per unit heat duty will increase from
the exchanger inlet to the exchanger outlet. Consequently, if average physical properties are
89

fed, the results will be unrealistic because a constant heat transfer coefficient will be applied.
It is the responsibility of the process licensor to furnish authentic physical properties.
This usually does not represent a problem since the output of any standard process simulator
includes all relevant physical properties. Even in cases where comprehensive physical
property data is not furnished, it is not difficult to obtain specific heat, density, and thermal
conductivity data of hydrocarbons, since these are very well documented [13].
The specific heat, viscosity, and thermal conductivity data of hydrocarbon vapors are a
function of molecular weight and temperature. The values of all these properties increase
linearly with temperature. Hydrogen has a considerably higher specific and thermal
conductivity than hydrocarbon vapors.
The density of a hydrocarbon vapors is expressed as (pM)/(zRT), where
P = operating pressure
M = molecular weight
Z = compressibility factor
R = universal gas constant
T = absolute temperature
At moderate temperatures and pressures, the compressibility factor may be considered to
be 1.0 without entailing any serious error.
The specific heat, density, and thermal conductivity of a hydrocarbon liquid vary with
temperature and API gravity. While density and thermal conductivity decrease with
temperature, specific heat increases with temperature. The viscosity of hydrocarbon liquids
varies irregularly with temperature and the same cannot be represented on any conventional
scale (linear, semilog, or log-log). This variation of various hydrocarbon liquid viscosities is
represented in special plots by ASME (Fig. 8.1) which are available in the TEMA standards.
It will be seen that the variation becomes extremely large for heavy liquids at low
temperatures.
Although all other physical properties may be obtained from various sources if they are
not furnished in the process data sheet, it will be prudent to insist on the liquid viscosity
values from the process licensor. Of course, this excludes standard pure components for
which data is available in [13].
It should also be stated here that, except for liquid viscosity, all other physical properties
vary linearly with temperature for all practical purposes and, hence, they need be furnished
to the heat exchanger software only at the inlet and outlet temperatures. However, as already
mentioned, the variation of viscosity with temperature is highly nonlinear and, consequently,
if the variation in liquid viscosity between the inlet temperature and the outlet temperature is
high, it is advisable to feed the viscosity values at intermediate temperatures as well.
While the HTRI (Heat Transfer Research, Inc.) software has the capability of evaluating
the intermediate values of viscosity accurately, not all software packages do. Evidently, the
number of intermediate viscosity values that should be fed will depend on the variation in
the viscosity. The important thing to do is to furnish a sufficiently high number of values so
that if a straight-line interpolation is implemented between any two points fed, the
representation will be quite authentic.
Thus, if the inlet viscosity is 2.0 centipoise at 248F (120C) and the outlet viscosity is
5.0 cp at 140F (60C), furnishing an intermediate viscosity value of 3.5 cp, say, at the mean
temperature may not be unreasonable. However, if the inlet and outlet viscosities at these
same temperatures are 10.0 cp and 54.0 cp, more points will have to be fed for a proper
90

representation. Evidently, the greater the number of points fed, the more accurate will be the
results. Most simulators permit the entry of values at a maximum of ten temperature points.
It is suggested that the intermediate points be so fed that the ratio between any two viscosity
values is more or less the same. Many designers feel tempted to feed the intermediate points
along roughly equal temperature increments. Thus, in the above case, the temperature
variation between inlet and outlet is (248 140) or 108F [(120 60) or 60C]. Thus, the
designer may feed intermediate viscosity values at two 36F (20C) intervals, i.e., at 212F
(100C) and 176F (80C). However, since the variation of liquid viscosity is not linear but
exponential, this will not result in a proper representation. The variation is much steeper at
the lower temperature range, so there should be more points at the lower temperature end.
In the case of a condenser where there is no liquid at the inlet, and condensation begins
at a slightly lower temperature (that is, the inlet vapor is somewhat superheated), a licensors
data sheet often indicates liquid physical properties only at the outlet temperature. This is
because, in the data sheet, there is provision for specifying physical properties only at the
inlet and outlet temperatures. For example, if the inlet temperature is 212F (100C) and the
outlet temperature is 104F (40C), liquid physical properties are furnished only at 104F
(40C). A common mistake in such cases is to feed the liquid physical properties only at the
outlet temperature. What most heat exchange thermal design software packages do in such
situations is to assume that the liquid physical properties at the inlet temperature are identical
to those at the outlet temperature. As already explained, this may result in error. What should
be done in such situations is to obtain the liquid physical properties at the inlet temperature
from the process licensor, standard physical property charts, or by sensible extrapolation,
and feed these data to the thermal design software. Often, detailed physical properties are

Fig. 8.1 ASTM plot of variation of hydrocarbon liquid viscosity with temperature (Reprinted with
permission from Standards of TEMA, 8th edition, 1999)
91

furnished by the licensor in the later data sheets (and not in the top data sheets), in which
case there is no problem at all.

8.2 Physical Property Profiles


Besides heat duty and weight fraction vapor, any other parameter that does not vary linearly with temperature should also be specified for an accurate thermal design. Physical
properties such as specific heat, thermal conductivity, and density of both vapor and liquid essentially vary linearly with temperature, as does vapor viscosity. However, liquid
viscosity does not vary linearly with temperature and, if the difference between the inlet
and outlet temperatures is not low, there could be a significant variation in this parameter
from the inlet to the outlet of the exchanger. In such a situation, the variation in liquid
viscosity with temperature should also be fed to the design software.
An interesting case is observed in the condensation of streams containing both hydrogen
and hydrocarbons. When such a mixture is cooled, only the hydrocarbon will condense, so
that the concentration of hydrogen will increase. Since the value of specific heat is the same
in English and metric units, only the values will be mentioned hereafter in this chapter. Now,
hydrogen has a much higher specific heat (typically 3.5, as compared to 0.5 for hydrocarbon
vapors), so that as the hydrogen concentration increases and the specific heat of the mixture
increases. Now, there is another effect on the specific heat of the mixture; namely, that of
temperature. The specific heat of both hydrogen and hydrocarbons decreases with a
reduction in temperature. However, the effect of the increase in the concentration of
hydrogen usually far outweighs that of the reduction in temperature, so that the specific heat
of the mixture increases as the hydrogen concentration increases with the reduction in
temperature.
In many such condensers, there is an initial desuperheating zone, after which the
condensation of hydrocarbon starts. Thus, from the inlet temperature to the dew point, there
will be a decrease in the total vapor specific heat, since the hydrogen concentration remains
the same and the temperature reduces. However, once condensation of a hydrocarbon
begins, the total vapor specific heat will begin to increase, and usually ends being

Fig. 8.2 Variation of vapor-specific heat with temperature in a hydrogen plant condenser
92

significantly higher than that at the inlet. In such cases, therefore, the variation of vapor
specific heat with temperature must be furnished to the thermal design software as well.
It should be understood that, besides accurately determining the heat transfer coefficient
zonewise, the vapor specific heat profile is also required in the above case for reconciling the
heat duty in each zone. An actual case study for a hydrogen plant condenser is shown in Fig.
8.2. A mixture of naphtha and hydrogen at 78.2 psia (5.5 kg/cm2 abs.) is condensed from
680F (360C) to 104F (40C). The variation in the total vapor specific heat is represented
by the curve ABCD. It will be seen that the specific heat decreases from 0.698 at 680F
(360C) to 0.51 at the dew point of 320F (160C), after which it increases sharply to 1.29 at
104F (40C). If this curve were not fed, a linear variation (shown by the straight line AD)
would be considered by the software. As a consequence, it would fail hopelessly to reconcile
the heat duty of each zone specified in the heat duty versus temperature profile, because it
would consider much higher values of mixture specific heat!
The above phenomenon is true of the thermal conductivity of hydrogen-hydrocarbon
mixtures as well. Thus, at 212F (100C), the thermal conductivity of hydrogen is 0.121
Btu/h ft F (0.18 kcal/h m C), whereas a typical hydrocarbon thermal conductivity at the
same temperature is 0.0148 Btu/h ft F (0.022 kcal/h m C). If the intermediate values of the
thermal conductivity of the vapor mixture are not fed, a linear interpolation would be
employed between the values fed at the inlet and outlet temperatures. Consequently, much
higher values of thermal conductivity would be considered, thereby leading to an
unrealistically optimistic design.

8.3 Heat Release Profiles


A heat release profile is a plot of the heat duty and weight fraction vapor versus temperature, and is an essential part of the process data sheet.
In single-phase services, the heat release profile is essentially linear. However, in
condensing services, the heat release profile is usually not linearthe slope varies from inlet
to outlet. If the temperature range is low to moderate (say, < 3654F or 2030C), the
variation in slope is usually small and a linear profile may be considered.
If a heat release profile is linear, the same is stated in the process data sheet, and a plot
need not be furnished to the computer program or considered for hand calculations.
If a heat release profile is not furnished and it is also not stated that the same is linear, a
straight-line profile may be assumed, provided the temperature range is low to moderate (<
54F or 30C). However, if the temperature range is higher, the licensor may be requested to
either furnish the heat release profile, or confirm that a linear profile may be considered.
When a heat release profile is not linear, there will be an error in the determination of
the MTD by considering it to be linear. Depending on the nature of the profile, it could be
higher or lowerusually it is lower. The greater the curvature of the profile, the greater will
be the difference in the MTD determined from the actual profile and an assumed linear
profile.
As can be easily understood, the heat duty versus temperature plot is essential for the
determination of MTD in the various zones.
The weight fraction vapor condensed versus temperature data is essential for two
reasons:
a) To reconcile the heat duty of each regionthis includes the phase change duty
and the sensible (vapor and liquid cooling) duty. Thermal design software packages usually evaluate the sensible vapor and liquid cooling duties from the respective flow rates, inlet/outlet temperatures, and specific heats. They then sub93

tract the total sensible cooling duty from the total heat duty, to obtain the condensing duty. The latent heat is then determined from the condensing duty and
the amount of vapor condensed. A negative latent heat, or an unusually high or
low latent heat, indicates an error in the data, which should then be examined and
rectified. This is a very important step in the design of condensers.
b) The vapor and liquid flow rates in the various locations (zones) are required to
compute the heat transfer coefficient and pressure drop.

8.4 How to Feed Heat Release Profiles


The number of points should be so chosen that each segment of the curve is virtually a
straight line. Referring to Fig. 8.3, the following points must be specified as a minimum:
A (80C), E (48C), F (44C), and G (42C).
Points B, C, and D need not be fed because they lie on the straight line AE. Additionally,
it is recommended that one intermediate point may be fed in sector EF and another in
sector FG for extra accuracy, since these sectors have some curvature. However, points E
and F are absolutely essential because, if not fed, the profile will be altered considerably.
No extra accuracy is achieved by feeding several points on the straight portion of a
curve. On the other hand, by feeding unnecessary points, the probability of a mismatch in
heat duty and consequent negative latent heat increases, especially if the temperature
increments and/or the amount condensed is small. It is therefore recommended that only the
minimum number of points required to represent the data authentically be fed.
When the divisions on the axes are quite large, it is sometimes difficult to read some
intermediate points from plots. In such cases, it is helpful to construct intermediate lines or
plot the same data on graph paper having more intermediate divisions.
The temperature points to be fed to the thermal design software should be so chosen that
the heat release curve is represented authentically. The values of weight fraction vapor
should be fed at these temperatures.

Fig. 8.3 How to feed heat release profiles


94

Even for services where only one stream has a nonlinear heat release profile, it is
beneficial to plot both the profiles together, since useful insight may be obtained from such a
composite plot, e.g., the existence of a pinch or a near pinch. Impossible situations with
the cold stream getting hotter than the hot stream at an intermediate temperature point can
also be detected and avoided without wasting any time in thermal design.
Sometimes, a heat release plot is furnished in a tabular fashion. Since it is not possible to
assess the curvature of a plot from tabulated heat duty versus temperature data, it is strongly
recommended that a curve be plotted first, either manually or by using a computer program.
Once this curve is plotted, it is relatively simple to decide how many points should be fed for
a proper representation. The methodology described above should then be followed.

References
[1] Gallant, R.W., and Railey, J.M., 1984, Physical Properties of Hydrocarbons, Vols. 1 and 2,
2nd Ed., McGraw-Hill, New York.
[2] Dean, J.A., ed., 1999, Lange's Handbook of Chemistry, 15th Ed., McGraw-Hill, New York.
[3] Perry, R.H., and Green. D.W., eds., 1997, Perry's Chemical Engineers Handbook, 7th Ed.,
McGraw-Hill, New York.

95

96

CHAPTER 9

Overdesign
9.1 Introduction
An overdesign factor (or margin) is often specified for the design of heat exchangers for
one or more of the following reasons:
a)
b)
c)
d)
e)

future increase in capacity


plant control flexibility
upset conditions
handling alternate feedstocks
uncertainty in results of process simulation

9.2 Mechanics of Overdesign


As the name implies, overdesign is the extra margin incorporated in a design. Unless otherwise specified, it implies the margin on the heat transfer surface. Thus, if an air-cooled
heat exchanger requires 1291 ft2 (120 m2) bare tube heat transfer area, and if 1453 ft2
(135 m2) is provided, we say that the overdesign is 12.5%. Designers prefer to incorporate an overdesign as insurance toward uncertainties that could exist in the flow rate,
temperatures, physical properties, and even the air-cooled heat exchanger design software
itself.
Evidently, if a heat exchanger is relatively small, a 10% or even a 20% overdesign may
not be objectionable because it is the extra absolute cost and not the percentage that is
significant. For larger heat exchangers, a smaller overdesign, such as a 5% margin, may be
more prudent. The higher the accuracy of the process data, and the higher the reliability level
of the design method (software), the lower the margin of overdesign that needs to be
incorporated.
For very large air-cooled heat exchangers, a smaller margin may be retained (down to
5%) for reasons of economy. Thus, for air-cooled heat exchangers having up to 2152 ft2 (200
m2) bare tube heat transfer area, a 10% margin may be retained, which may be reduced
progressively to 5% for 43044842 ft2 (400450 m2) and beyond. It will be evident that
these values are of an indicative nature only.

9.3 Overdesign in Single-Phase Heat Exchangers


It will be easily appreciated that an overdesigned heat exchanger will deliver more than
the design heat duty. In the case of sensible heat transfer, the air will get heated more and

97

Table 9.1a: Principal process parameters for Case Study 9.1, high
overdesign case
1. Kerosene flow rate, lb/h (kg/h)
2. Temperature in/out, F (C)
3. Heat duty, MM Btu/h (MM kcal/h)
4. Allowable pressure drop, psi (kg/cm2)
5. Inlet/outlet viscosity, cp
6. Inlet/outlet density, lb/ft3 (kg/m3)
7. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
8. Design air temperature, F (C)

396,830 (180,000)
205.2 (96.2)/140 (60)
12.82 (3.23)
7.1 (0.5)
0.48/0.72
46.8 (750)/48.6 (778)
0.00195 (0.0004)
95 (35)

the hot stream will get cooled more. For the final conditions of terminal temperatures and
heat duty, the overdesign will obviously have to be nil.
The increased heat duty that an overdesigned heat exchanger can perform, expressed as
a percentage of the design heat duty, is called overdesign on performance. Thus, if a stream
has to be cooled through 90F (50C) but actually gets cooled through 99F (55C), it has a
10% overdesign on performance.
The value of overdesign on performance is always less than the value of overdesign on
surface. This is because an increase in the heat duty of an overdesigned heat exchanger is
always accompanied by a decrease in the MTD. Therefore, the larger the reduction in the
MTD, the smaller will be the difference between overdesign on surface area and overdesign
on performance.
While overdesign on surface simply indicates the excess area available to cater to
uncertainties, or more pragmatically, the percentage of tubes that can be plugged in case of
leakage (provided, of course, there is a corresponding margin in the tubeside pressure drop),
overdesign on performance indicates how much better an exchanger may be expected to
perform in terms of heat duty. Heat exchanger services are usually analyzed and interpreted
in the fouled condition. However, when a heat exchanger is new and thereby cleanor after
it has been cleaned when used for some length of timeits overall heat transfer coefficient
is much higher than that in the fouled condition, so that it is oversurfaced, over and above
the overdesign the designer has incorporated in the original design. Evidently, the extent of
overperformance will depend on the extent to which the total fouling resistance is
controlling the heat transfer process. Consequently, a new or cleaned heat exchanger will
deliver a heat duty higher than the design value.
We will now take a look at a case study to illustrate some of the points that have just
been discussed.

CASE STUDY 9.1: EFFECT OF OVERDESIGNHIGH OVERDESIGN CASE


Consider the case of a kerosene cooler in an oil refinery heat exchanger having the principal process parameters specified in Table 9.1a. A thermal design was made for this service, and the principal construction and performance parameters are detailed in Table
9.1b. The overdesign on surface is 31.1%; that is, the bare tube heat transfer area provided [3810 ft2 (354.1 m2)] is 31.1% more than that required to handle the specified heat
duty.
Since the exchanger is oversurfaced, the kerosene will cool somewhat more and the air
will get heated somewhat more, until the required and provided heat transfer areas match. A
performance run or a simulation run was taken for this exchanger to ascertain its actual
98

Table 9.1b: Principal construction and performance parameters for Case Study 9.1, high
overdesign case
1. No. of bays
2. No. of bundles per bay
3. Tube OD thk length, in. (mm)
4. Fin height, in. (mm) fin density, fins/in. (fins/m)
5. No. of tubes per row no. of rows
6. No. of tube passes
7. Total bare tube area, ft2 (m2)
8. Fans per bay fan diameter, ft (m)
9. Total air flow rate, lb/h (kg/h)
10. Air outlet temperature, F (C)
11. Airside velocity, ft/s (m/s)
12. Airside pressure drop, in. WC (mm WC)
13. Fan power, HP (kW)
14. Connections per bundle, in. (mm)
15. Heat transfer coefficient,
Btu/h ft2 F/ (kcal/h m2 C)

Tubeside
Airside
Overall

16. Overdesign, %
17. Tubeside pressure drop, psi (kg/cm2)

1
2
0.984 (25) 0.098 (2.5) 492 (12500)
0.49 (12.5) 11 (433)
46 4
2
3810 (354.1)
2 14 (4.268)
2,337,000 (1,060,000)
117.9 (47.7)
12.04 (3.67)
0.43 (10.83)
32.7 (24.4)
In: 2 4 (100)
Out 2 4 (100)
178.8 (873)
165.3 (807)
69.5 (339.4)
20.2
3.27 (0.23)

Table 9.1c: Comparative statement of design and expected performance duties for Case Study 9.1,
high overdesign case

1. Kerosene stream Inlet temperature, F (C)


Outlet temperature, F (C)
2. Air
Inlet temperature, F (C)
Outlet temperature, F (C)
3. Heat duty, MM Btu/h (MM kcal/h)
4. MTD, F (C)
5. Overall heat transfer coefficient, Btu/h ft2 F
(kcal/h m2 C)
6. Overdesign, %

For design duty


205.2 (96.2)
140 (60)
95 (35)

Expected
performance
205.2 (96.2)
130.3 (54.6)
95 (35)

117.9 (47.7)
12.82 (3.23)
63.4 (35.2)
69.5 (339.4)

121.1 (49.5)
14.64 (3.69)
55.4 (30.8)
69.2 (337.6)

1.202

Nil

Percent change

14.2
() 12.5
() 0.53

expected operating performance. Table 9.1c gives a comparative statement of the design and
the expected duties.
It is seen that the exchanger will actually deliver a heat duty of 14.64 M Btu/h (3.69 M
kcal/h) instead of the design value of 12.82 M Btu/h (3.23 M kcal/h). Therefore, the
overdesign on performance is 3.69/3.23 or 1.142, or 14.2%. What has happened is that while
the heat duty has increased, the MTD has decreased from 63.4F (35.2C) to 55.4F
(30.8C), so that the heat transfer area required is equal to that provided. The overall heat
transfer coefficient has virtually remained the same, but this may change as well, depending
principally on the change in temperature and thereby the physical properties, principally
viscosity.
99

Table 9.2a: Principal process parameters for Case Study 9.2, low overdesign case
1. Distillation column overhead flow rate, lb/h (kg/h)
2. Temperature in/out, F (C)
3. Heat duty, MM Btu/h (MM kcal/h)
4. Allowable pressure drop, psi (kg/cm2)
5. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
6. Design air temperature, F (C)
7. Vapor properties

Density in/out, lb/ft3 (kg/m3)


Viscosity in/out, cp
Specific heat in/out, Btu/lb F (kcal/kg C)
Thermal conductivity in/out, Btu/h ft F
(kcal/h m C)

8. Liquid properties

Density in/out, lb/ft3 (kg/m3)


Viscosity in/out, cp
Specific heat in/out, Btu/lb F (kcal/kg C)
Thermal conductivity in/out, Btu/h ft F
(kcal/h m C)

474,000 (215,000)
269.6(132)/131 (55)
59.65 (15.03)
3.56 (0.25)
0.00195 (0.0004)
105.8 (41)
0.45 (7.2)/
0.4 (6.45)
0.013/0.011
1.28/1.64
0.094 (0.14)/
0.086 (0.128)
47.4 (760)/
51.17 (820)
0.75/2.13
0.54/0.45
0.065 (0.097)/
0.0706 (0.105)

In this example, it is seen that while the overdesign on surface is quite high (31.1%), the
overdesign on performance is much lower (14.2%).

9.4 Overdesign in condensers


In the case of total condensers, overdesign on surface will translate into subcooling. Since
subcooling is rather inefficient as compared to condensing in an integral shell, the increase in heat duty will be very small. However, in the case of partial condensers, there
could be an appreciable increase in heat duty due to overdesign. Let us now take a look at
another case study, this time of a partial condenser.

CASE STUDY 9.2: EFFECT OF OVERDESIGNLOW OVERDESIGN CASE


Table 9.2a elaborates the principal parameters of an air-cooled condenser in an oil refinery. A thermal design was prepared for this service, and the principal construction and
performance parameters are presented in Table 9.2b. The overdesign on surface is 6.8%.
A performance or simulation run was taken next, and the expected performance of the
heat exchanger is reported in Table 9.2c. It will be seen that while there is a very minor
change in the overall heat transfer coefficient, the MTD is lower by 4.9%. The heat duty is
60.7 M Btu/h (15.3 M kcal/h), as against the design heat duty of 59.65 M Btu/h (15.03 M
kcal/h). The overdesign on performance is therefore 60.7/59.7 or 1.7%. Once again, the
overdesign on performance is less than the overdesign on surface, but not to the extent we
saw in the previous case study.

9.5 The Overdesign Factor


The overdesign factor may be on the heat transfer surface only (excess area) when the
thermal design will have to be performed on the specified flow rates and the required
margin incorporated. Usually, however, the overdesign factor is specified for both the
flow rates and the heat duty. Since a heat exchanger can be designed only for a consistent
100

Table 9.2b: Principal construction and performance parameters for Case Study 9.2, low
overdesign case
1. No. of bays
2. No. of bundles per bay
3. Tube OD thickness length, in. (mm)
4. Fin height, in. (mm) fin density, fins/in. (fins/m)
5. No. of tubes per row no. of rows
6. No. of tube passes
7. Total bare tube area, ft2 (m2)
8. Fans per bay fan diameter, ft (m)
9. Total air flow rate, lb/h (kg/h)
10. Air outlet temperature, F (C)
11. Airside velocity, ft/s (m/s)
12. Airside pressure drop, in. WC (mm WC)
13. Fan power, HP (kW)
14. Connections per bundle, in. (mm)
15. Heat transfer coefficient, Btu/h ft2 F/
(kcal/h m2 C)

Tubeside
Airside
Overall

16. Tubeside pressure drop, psi (kg/cm2)

4
2
0.984 (25) 0.098 (2.5) 413 (10500)
0.49 (12.5) 11 (433)
40 6
2
16,570 (1540)
2 13 (3.962)
6,173,000 (2,8000,000)
145.9 (63.3)
11.1 (3.38)
0.56 (14.2)
24.2 (18.0)
In: 2 6 (150)
Out 2 4 (100)
124.6 (608)
159.8 (780)
58.0 (283.3)
2.7 (0.19)

Table 9.2c: Comparative statement of design and expected performance duties for Case Study 9.2,
low overdesign case
1. Overhead stream
2. Air

Inlet temperature, F (C)


Outlet temperature, F (C)
Inlet temperature, F (C)

Outlet temperature, F (C)


3. Heat duty, MM Btu/h (MM kcal/h)
4. MTD, F (C)
5. Overall heat transfer coefficient, Btu/h ft2 F (kcal/h m2 C)
6. Overdesign, %

For design
duty
269.6 (132)
131 (55)
105.8 (41)

Expected
performance
269.6 (132)
127.9 (53.3)
105.8 (41)

145.9 (63.3)
59.65 (15.03)
66.24 (36.8)
58.0 (283.3)
6.8

146.7 (63.7)
60.7 (15.3)
63.0 (35.0)
57.96 (283)
Nil

Percent
change

1.8
4.9
Negligible
6.8

set of parameters, this overdesign factor should be the same for the shellside flow rate,
the tubeside flow rate, and the heat duty.
Sometimes, a process data sheet specifies different overdesign factors for flow rate and
heat duty. For example, the shellside flow factor may be 120%, the tubeside flow factor
110%, and the heat duty 115%. In such cases, the matter should be reconciled with the
process licensor so that the same factor is specified for all three parameters. If necessary, the
licensor may specify an additional multiplier for flow rates for the purpose of determination
of pressure drop only, if the same is expected to increase under certain conditions.
In some instances, the process licensor may specify different factors for the shellside and
tubeside flow rates for an alternate case of operation, and request that for the thermal design
(geometry) finalized, the heat exchanger designer indicate the outlet temperatures of both
streams for certain specified inlet temperatures. This situation is acceptable since it is
realistic and there is no inconsistency. An alternate simulation run need only be taken to
indicate the desired data.
101

If the process licensor specifies overdesign on flow and duty, and also specifies excess
area, both should be complied with. The excess area required can be applied toward
uncertainties in simulation, whereas the margin on flow rates can be applied toward future
increase in plant capacity.
In order to limit the tubeside pressure drop to within the allowable limit, the overdesign
of a heat exchanger may sometimes be inordinately high, since any reduction in the number
of tubes (to reduce overdesign) may increase the tubeside pressure drop beyond the
allowable limit. This is called a pressure drop limiting design and is an acceptable
situation. Evidently, this will occur when the tube length is standardized and cannot be
reduced.
However, in such situations, it will make eminent sense to take advantage of the
inordinate overdesign while designing the corresponding trim cooler, if one is present. To
clarify, consider an air-cooled heat exchanger having a heat duty of 10.5 M Btu/h (2.65 M
kcal/h) and a subsequent trim cooler having a heat duty of 2.5 M Btu/h (0.63 M kcal/h). If,
by virtue of the higher overdesign in the air-cooled heat exchanger, as explained above, it is
able to deliver a heat duty of 11.3 M Btu/h (2.85 M kcal/h) after catering to a normal
overdesign margin, the trim cooler need be designed only for a heat duty of 2.5 (11.3
10.5) or 1.7 M Btu/h [0.63 (2.85 2.65) or 0.43 M kcal/h].

9.6 Tube Plugging


Although not usually specified for the same, and often not realized directly, an overdesign margin is often useful since it permits a certain extent of tube plugging before a
heat exchanger has to be taken out for repair, provided there is sufficient margin in the
tubeside pressure drop. Thus, even if only 5% of the tubes in a heat exchanger are
plugged, the tubeside pressure drop will increase by about 10%. We all know that, despite the best practices in material selection and exchanger design/fabrication/testing,
tubes do fail occasionally in real life. Therefore, in general, it is very useful if the permitted tubeside pressure drop is not consumed completely while designing a heat exchanger,
but a small margin left unutilized.

102

CHAPTER 10

Fouling: Its Causes, Consequences,


and Mitigation
Fouling may be defined as the deposition of undesirable matter on a heat transfer surface
and is an inevitable consequence of the process of heat transfer between two streams
across a metal wall. Taborek called fouling the major unresolved problem in heat transfer way back in 1972 [1], and the situation does not appear to have changed significantly. Since fouling has a strong impact on the energy efficiency of a heat exchanger, on
both heat transfer and pressure drop, it is essential for the designer to be fully aware of
the phenomenon, its consequences, and its mitigation.
Evidently, fouling can take place both inside and outside tubes. The deposition of the
foulant results in the following:
1) a reduction in the overall heat transfer coefficient due to the extra resistance to
heat transfer, thereby resulting in a larger heat transfer area
2) a reduction in the area of the flow passages, resulting in increased pressure drop
of the flowing streams
The extent of fouling varies markedly with the nature of the fluids being handled.
Consequently, exchangers that handle clean fluids may remain largely free of fouling,
whereas exchangers that handle dirty streams may be constantly afflicted by it.
A fact that is usually not recognized is that the increase in pressure drop is usually a
much more serious consequence of fouling than the increase in the thermal resistance and,
thereby, the reduction in the heat transfer coefficient. When a heat exchanger is taken out of
service for cleaning, it is invariably due to the reduced throughput as a result of partial
blockage of flow areas, rather than reduced thermal performance.
The adverse effects of fouling are as follows:
1) Increased capital cost due to the reduced overall heat transfer coefficient
2) Additional energy requirements to make up for the loss in performance. For example, when a given air-cooled heat exchanger underperforms, the cooling water
flow rate in the subsequent trim cooler has to be increased to make up for this
loss in performance, thereby increasing the cooling water pumping cost.
3) Maintenance costs for antifoulant, chemical treatment, and cleaning of fouled
surfaces
4) Downtime costs associated with the outage of the air-cooled heat exchanger for
cleaning
5) Reduction in throughput
103

In an air-cooled heat exchanger, there are five resistances to heat transfer, namely, airside
film, airside fouling, tubeside film, tubeside fouling, and tube wall. By virtue of being of
extended surface, airside fouling resistance is usually not very significant. A typical airside fouling resistance of 0.002 h ft2 F/Btu (0.00041 h m2C/kcal) on the extended surface translates into 0.000114 h ft2 F/Btu (0.000023 h m2 C/kcal) on the bare tube surface, considering a typical surface ratio (ratio of total extended surface to outside bare
tube surface) of 17.6. This value is so low as to be virtually insignificant. However, the
reduction of the airside flow passage due to external deposition often causes a significant
reduction in the air flow rate, which reduces both the airside heat transfer coefficient as
well as the MTD, thereby impairing the thermal performance significantly. In this chapter, we will first take a look at tubeside fouling; and later, at airside fouling.

10.1 Tubeside Fouling


10.1.1 Categories of fouling
There are six principal categories of fouling as follows:

1) Precipitation fouling (scaling) is the precipitation of dissolved substances on the


heat transfer surface.
2) Particulate fouling is the accumulation of particles suspended in a fluid on the
heat transfer surface. It is caused by gravity in some applications, where it is referred to as sedimentation fouling.
3) Corrosion fouling occurs when the heat transfer surface itself reacts to produce
corrosion products, which then foul the heat transfer surface. Since the heat transfer surface becomes rougher due to corrosion fouling, it produces nucleation sites
for precipitation and particulate fouling.
4) Chemical reaction fouling is the formation of deposits by chemical reaction
among the different constituents of the flowing stream. The surface material itself
does not enter into reaction.
5) Solidification fouling is the solidification of pure liquid or particular constituents
of a liquid solution on a subcooled heat transfer surface. One example is ice formation in air-cooled steam condensers in colder climates, such as the northern
parts of North America.
In a given air-cooled heat exchanger application, depending on the situation, one or
more of the above modes of fouling may occur. This is what makes fouling a very complex
and unpredictable phenomenon.
10.1.2 Progress of fouling
The buildup of fouling on a heat transfer surface is ideally represented by an asymptotic
curve as shown in Fig. 10.1. There are three distinct regions in this curve:

a) An initial period, 01, where there is very little evidence of any fouling. This
may be considered to be an initiation period, which may range from a few hours
to several weeks. This will depend on several parameters that are discussed in
Section 10.1.3.
b) Period 12, wherein a steady increase in fouling deposition is observed. This period could vary from a few hours to several months.
104

c) Period 23, where the rate of increase of fouling resistance decreases from that in
period 12, and the curve attains a plateau or the asymptotic fouling resistance
value.
In real life, however, a fouling curve may vary considerably from this idealized curve, so
that in some cases there may be no initiation period, and in others, the increase in fouling
resistance may be virtually linear. Whatever the nature of a fouling curve, there will come
a time, with most fluids, when the thermal and hydraulic performance of a given heat exchanger will deteriorate to such an extent that the heat transfer surface will have to be
cleaned to restore the original (or close to original) performance.
Evidently, the higher the fouling resistance considered for a particular stream, the longer
will a heat exchanger be able to operate before being subjected to a shutdown. However, the
higher fouling resistance will also mean a costlier heat exchanger. Therefore, the selection of
a design fouling resistance will have to be made on the basis of optimization of the total cost
(fixed cost plus operating cost).
Another factor that should be considered in this context is the normal turnaround of a
plant. All plants have to be shut down periodically for inspection and overhauling of
equipment, piping, instrumentation, etc. The run length usually varies from 12 to 36 months.
Thus, it is a desirable practice to consider this run length for the selection of design fouling
resistances of all flowing streams, so that the heat exchangers can be cleaned in the same
period.
10.1.3 Parameters that affect fouling
There are several parameters that affect the degree of the various types of fouling, as follows:

Fouling resistance

1) The nature of the flowing fluidwhether clean or dirty. This is the starting point
for all discussions on fouling, and it is only for dirty streams that all the subsequent considerations assume importance. For heat exchangers handling clean
streams, such as steam and very light hydrocarbons, fouling is not a problem at
all. On the other hand, heavy hydrocarbon streams, such as long residue or vac-

Time
Fig. 10.1 An idealized fouling curve
105

Fouling resistance

uum gas oils, foul readily and cause considerable deterioration in performance in
heat exchangers.
2) Flow velocity and temperature are perhaps the most crucial variables that control
the fouling process. A high velocity minimizes virtually all types of fouling.
Fouling is a dynamic process wherein deposition and removal of foulant occur
simultaneously. The net fouling at any given instance represents the equilibrium
balance between these two opposing phenomena, a deposition phenomenon and a
removal phenomenon.
With increase in velocity, the viscous sublayer close to the tube wall becomes thinner, thereby resulting in a reduction in the resistance to diffusion from
the bulk to the wall. This permits a comparatively higher rate of deposition for a
diffusion-controlled fouling process. However, at the same time, the higher velocity increases the shearing forces that tend to remove the fouled deposit. The
net rate of fouling will therefore depend on these two opposing effects of velocity. Usually, the rate of decrease of fouling due to the increase in the shear force
is greater than the rate of increase of deposition due to the reduction of the viscous sublayer. Consequently, higher velocities invariably result in less fouling.
The general nature of the degree of fouling versus flow velocity is represented in
Fig. 10.2.
3) The temperature of the fluid-foulant interface strongly influences the extent of various modes of fouling. Bulk fluid temperatures and their heat transfer coefficients, as
well as the fouling and tube wall resistances, will determine this interface temperature. When normal-solubility salt solutions are cooled, they exhibit fouling since the
solubility decreases at lower temperatures.
The rates of chemical reaction are a strong function of temperature. Thus, if a
fouling process involves a chemical reaction, the extent of fouling will depend on
temperature. The rate of increase or decrease of fouling with time will be related to
the rate constant of the chemical reaction itself.
Corrosion is basically a chemical reaction. Consequently, the fouling of surfaces by the products of corrosion will be dependent on temperature.
4) Material of construction and surface finish including the roughness, size, and den-

Flow velocity

Fig. 10.2 Variation of fouling resistance with flow velocity


106

sity of cavities affect crystalline nucleation, sedimentation, and the sticking tendency of deposits. It is generally believed that very smooth surfaces are less likely to
receive and retain dirt than are rough surfaces. However, it may be argued that this
will be true only for the initial fouling because thereafter the surface will no longer
be smooth. However, the practical experience is that polished or smooth surfaces
foul less than rough ones.
In such a complicated scenario where there are various modes of fouling, as well as
several factors that determine the degree of the various modes of fouling, it is evident that
the prediction of the extent of fouling is extremely difficult. Consequently, it becomes
unavoidable to adopt a qualitative approach and rely on past experience for the selection of
fouling resistances for various services. TEMA (Tubular Exchanger Manufacturers
Association) specifies point and range values of various streams encountered in the chemical
processing industries [2]. Although the TEMA guidelines specifically apply to shell-andtube heat exchangers, the fouling resistances specified therein apply to air-cooled heat
exchangers as well, since these are also tubular heat exchangers. (Fouling resistances are
much lower than these TEMA values for nontubular heat exchangers such as plate heat
exchangers, which generate far more turbulence.)
The TEMA values are for guidance only, and should be modified based on actual
operating feedback wherever available. If no actual data is available, the TEMA values
should be adopted. The selection of appropriate fouling resistances contributes significantly
in ensuring satisfactory operation of air-cooled heat exchangers.
10.1.4 How to provide a fouling allowance
Many heat exchangers operate satisfactorily for several years without cleaning. Other
heat exchangers are constantly afflicted by fouling. However, most heat exchangers experience some fouling so that it becomes necessary to provide sufficient heat transfer area
to enable them to operate for a reasonably long period of time (usually two or three years)
before requiring shutdown and cleaning.
The extra heat transfer area that must be provided to account for fouling is usually
determined by assigning a fouling resistance to the tubeside stream, and another fouling
resistance to the airside stream. The term fouling resistance is often incorrectly referred to
as fouling factor. This is an absolute misnomer because a factor is something one
multiplies with, whereas a resistance is something one adds.
An alternative approach could be to add a certain percentage of additional heat transfer
surface to that determined from the clean overall heat transfer coefficient, based on actual
field experience for a specific service. However, such quantified experience is rarely
available, so the use of the fouling resistance is the universal method.

Apply fouling allowance for fouling only.


It must be mentioned here that the application of fouling resistances should be aimed only
at making an allowance for anticipated fouling. It should not cater to uncertainties in design methodology (whether in process simulation or in the thermal design software), the
prediction of physical properties, or future increase in plant capacity. If such possibilities
exist, they should be catered to by specific allowances for each. For example, a 10%
margin may be retained for uncertainties in physical properties, and/or a 10% multiplier
may be applied to flow rates and heat duty in view of future capacity expansion. The advantage of this methodology is that there will be a much lower probability of overspecify107

ing the overall margin that is applied on the heat transfer surface than there would be
by applying a single overdesign factor based on an overall feel of the situation, to
cater to all the uncertainties and/or requirements.
Do not overspecify fouling resistance.
Designers often consider it prudent to apply conservative fouling resistances in order to
obtain a longer run length (period between successive cleanings). However, unnecessarily
high fouling resistances may actually cause more harm than good because of the following:
1) The application of unduly large fouling resistances will result in large heat exchangers, whereby it may not be possible to maintain a sufficiently high velocity
within the allowable pressure drop. For example, to maintain the same tubeside
velocity, a larger number of tubes will require a greater number of tube passes,
which will translate into a higher pressure drop. If the resultant pressure drop exceeds the allowed limit, the number of passes will have to be reduced, thereby
condemning the exchanger to a lower velocity and consequently heavier fouling!
It is for this reason that fouling is often referred to as a self-fulfilling prophecy.
2) The specification of an unduly high fouling resistance also sends a wrong signal
to the designer that the stream is dirty, thereby prompting efforts to unnecessarily
maintain a sufficiently high velocity to minimize fouling. Since this will result in
a much higher pressure drop, the whole practice will be wasteful, since it will
lead to unnecessarily higher energy consumption.
3) The application of high fouling resistances will also result in a large difference
between the clean and the dirty overall heat transfer coefficients. Thus, such heat
exchangers will be highly oversurfaced when clean (at start-up), in which condition they may be difficult to control. Additionally, in air-cooled heat exchangers,
the air temperature varies drastically between day and night and through the seasons. Fortunately, variation of air flow rate by means of autovariable fans or
variable speed drives affords very good control of air-cooled heat exchangers.
This is discussed in detail in Chapter 11.
Therefore, it is strongly recommended that only realistic fouling resistances be considered for heat exchanger design.
10.1.5 Selection of Fouling Resistance
As has been mentioned earlier, even after so many years of research on fouling by so
many organizations, the heat exchanger designer still has to invariably fall back on the
values of fouling resistances furnished in the TEMA standards. This is not surprising,
considering the variety of fluids being handled in the chemical process industries, along
with the variation in fluid velocities, wall temperatures, and materials of construction.
Thus, it becomes important for plant operators to monitor fouling for the various fluids
handled. While this is a tedious task, it is well worth the effort in terms of the realistic
fouling resistance data that will emerge.
Also, while the earlier TEMA editions furnished only point values of fouling resistance
of various fluids handled in the chemical process industries, the latest (1999) edition
furnishes range values for certain fluids. TEMA values are evidently indicative, and should
be used with discretion in the absence of any other authentic in-house data. Where range
108

values are given, selection must be made on the basis of specifics.


For example, the fouling resistance of kerosene has been specified in the TEMA
standards as 0.0020.003 h ft2 F/Btu (0.000410.000614 h m2 C/kcal). This can be
interpreted in the following manner. For kerosene produced in the crude distillation unit of
an oil refinery processing a light crude, the fouling resistance may be considered to be 0.002
h ft2 F/Btu (0.00041 h m2 C/kcal), whereas for kerosene produced in the delayed coking
unit of a refinery processing a heavy crude, the fouling resistance may be considered to be
0.003 h ft2 F/Btu (0.000614 h m2 C/kcal). The final selection should also depend on the
velocity of the kerosene in the given heat exchanger, as well as the temperature level of the
kerosene.
It is interesting to note that in the 1999 edition of the TEMA standards, the values of
fouling resistance of certain fluids have been increased, while those of certain other fluids
have been decreased from the values in the previous edition. Thus, the fouling resistance of
compressed air has been reduced from 0.002 to 0.001 h ft2 F/Btu (from 0.00041 to
0.000205 h m2 C/kcal) while that of reduced crude oil (long residue) has been increased
from 0.005 to 0.007 h ft2 F/Btu (from 0.00102 to 0.00143 h m2 C/kcal). Evidently, these
revisions have been incorporated on the basis of feedback received from plant operators, as
well as from a better understanding of the phenomenon of fouling.
To summarize, the selection of fouling resistance has to be done carefully, and should be
based on past experience. Values specified in the TEMA standards are for guidance only,
and should be tempered with operating feedback and engineering judgment. A proper
selection of fouling resistance will go a long way toward ensuring the satisfactory operation
of heat exchangers.
10.1.6 Design guidelines to minimize tubeside fouling
Although it is difficult to determine fouling resistances accurately, there are several qualitative as well as quantitative rules of thumb that should be followed in order to minimize
the extent of fouling:

10.1.6.1 Use large diameter tubes


For a given heat exchanger service, a smaller tube diameter results in a smaller and
cheaper heat exchanger. However, smaller diameter tubes are more difficult to clean internally, especially for relatively long lengths. Consequently, 1 in. (25 mm) OD tubes are
the smallest recommended by API 661 for air-cooled heat exchangers. Smaller diameter
tubes [3/4 in. (19 mm) OD] may be used for clean services after agreement between client and consultant. However, for dirty services [fouling resistance greater than 0.002 h ft2
F/Btu (0.00041 h m2 C/kcal)], it is generally recommended that the tube diameter be
not less than 1 in. (25 mm). For very dirty services, even 1-1/4 in. (31 mm) or 1-1/2 in.
(38 mm) OD tubes may be considered. This is because the increase in tubeside pressure
drop due to the same degree of fouling [say, a 0.02 in. (0.5 mm) thick deposit] will be
higher with smaller diameter tubes. Furthermore, smaller diameter tubes tend to get
plugged much more readily than larger diameter ones, and are more difficult to clean.
10.1.6.2 Maintain high velocity
A high velocity suppresses all types of fouling. For fouling liquids, a minimum velocity
of 3.3 ft/s (1.0 m/s) is generally recommended, although 5 ft/s (1.52 m/s) is preferable.
The dirtier the liquid, the higher should be the minimum velocity, in order to restrict fouling to an acceptable level.
109

Table 10.1a: Principal process parameters for Case Study 10.1


1. Liquid flow rate, lb/h (kg/h)
2. Temperature in/out, F (C)
3. Heat duty, MM Btu/h (MM kcal/h)
4. Allowable pressure drop, psi (kg/cm2)
5. Inlet/outlet viscosity, cp
6. Inlet/outlet density, lb/ft3 (kg/m3)
7. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
8. Fouling layer thickness, in. (mm)
9. Design air temperature, F (C)

639,000 (289,850)
255.2 (124)/194 (90)
21.5 (5.42)
15 (1.055)
11.3/15
51.4 (823)/52.5 (842)
0.0039 (0.0008)
0.016 (0.4)
107.6 (42)

Occasionally, a proper velocity cannot be maintained within the limitation of the


allowable pressure drop because a large number of tube passes is required to achieve this
velocity. In such cases, either a higher pressure drop should be allowed or the tube velocity
best maximized by varying the tube diameter and length. It is important to realize that while
a lower pressure drop will ensure a lower operating cost due to pumping, the resultant lower
tubeside velocity will result in a higher operating cost due to heavier fouling. Thus, a
reasonable balance should be struck between the cost of pumping power and that of fouling.
A problem usually arises with highly viscous and fouling liquids (which are not very
uncommon in oil refineries), where a very high-pressure drop is required to sustain a
satisfactory velocity. This could go up to 2856 psi (24 kg/cm2) or even more, which at
first may appear to be unacceptable, but which should be examined from an overall cost
optimization point of view. The standard values of pressure drop permitted for various heat
exchanger services [typically 1.42.8 psi (0.10.2 kg/cm2) for gases and condensing
services, and 10 psi (0.7 kg/cm2) for liquid services)] are only based on a general
optimization. However, 10 psi (0.7 kg/cm2) is not the proper value of allowable pressure
drop for viscous liquids, and depending on the viscosity, 2856 psi (2.04.0 kg/cm2), or even
greater, is more authentic.
It should be added here that a viscous fluid flowing on the tubeside requires a much
higher pressure drop for the same heat transfer coefficient, than when it is flowing on the
shellside of a shell-and-tube heat exchanger. This is why air-cooled heat exchangers are
fundamentally not suitable for cooling viscous liquids, and cooling by closed-circuit hot
(tempered) water in shell-and-tube heat exchangers is far better.
It will be easily realized that the higher the velocity in the final design, the higher will be
the heat transfer coefficient and the lower will be the incidence of fouling. It is therefore
suggested that in no case should the velocity be allowed to fall below 1.6 ft/s (0.5 m/s). The
magnitude of this problem can be fully appreciated when it is considered that plants often
run in turndown conditions due to various constraints, when the flow velocities are even
lower than the design values.
Let us now take a look at a case study that will demonstrate that given sufficient
tubeside pressure drop, it is possible to obtain a satisfactory tubeside velocity so that fouling
can be minimized.

CASE STUDY 10.1: MAINTAINING HIGH TUBESIDE VELOCITY


The principal process parameters of a viscous hydrocarbon liquid cooler are specified in
Table 10.1a. The inlet/outlet viscosity values are 11.3 and 15 cp, which are rather high
110

Table 10.1b: Principal construction and performance parameters for Case Study 10.1
1. No. of bays
2. No. of bundles per bay
3. Tube OD thk length, in. (mm)
4. Fin height, in. (mm) fin density, fins/in. (fins/m)
5. No. of tubes per row no. of rows
6. No. of tube passes
7. Total bare tube area, ft2 (m2)
8. Fans per bay fan diameter, ft (m)
9. Total air flow rate, lb/h (kg/h)
10. Air outlet temperature, F (C)
11. MTD, F (C)
12. Airside pressure drop, in. WC (mm WC)
13. Fan power, HP (kW)
14. Connections per bundle, in. (mm)
15. Tubeside velocity, ft/s (m/s)
16. Tubeside pressure drop, psi (kg/cm2)
17. Tubeside Reynolds no., inlet/midpoint/outlet
18. Heat transfer coefficient, Btu/h ft2 Tubeside
F/ (kcal/h m2 C)
Airside
Overall
19. Overdesign, %

Design 1
Design 2
4
2
2
2
1.0 (25.4) 0.109 (2.769) 408 (10360)
0.5 (12.7) 9 (354)
43 7
43 8

6
8
20,960 (1948)
11,965 (1112)
2 13 (3.962)
4,515,900 (2,048,400)
2,759,000 (1,251,500)
127.4 (53)
140 (60)
105.8 (58.8)
100.3 (55.7)
0.33 (8.33)
0.53 (13.42)
10.3 (7.68)
20.0 (14.9)
In: 2 2 (50)
In: 2 3 (75)
Out: 2 2 (50)
Out: 2 3 (75)
2.76 (0.84)
6.43 (1.96)
15 (1.053)
68.3 (4.8)
1166/992/896
2766/2299/2084
12.3 (60.1)
26.8 (130.7)
122.3 (597)
10.5 (51.4)
8.55

133.3 (651)
19.8 (96.7)
10.58

Table 10.2: Suggested values of fouling layer thickness for various values of
fouling resistance
Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
0.0049 (0.001)
0.0059 (0.0012)
0.0068 (0.0014)
0.0078 (0.0016)
0.0088 (0.0018)
0.0098 (0.002)

Fouling layer thickness, in. (mm)


0.02 (0.5)
0.024 (0.6)
0.028 (0.7)
0.0315 (0.8)
0.0354 (0.9)
0.0394 (1.0)

for a stream flowing inside tubes. The permitted pressure drop is 15 psi (1.055 kg/cm2),
which appears to be quite low, considering the high viscosity values.
Corresponding to the specified fouling resistance of 0.0039 h ft2F/Btu (0.0008 h
2
m C/kcal), a fouling layer thickness of 0.016 in. (0.4 mm) was applied, as per Table 10.2.
This will be discussed in the next section and a case study will be presented.
A thermal design was prepared with the specified pressure drop, and the principal
construction and performance parameters are elaborated in the first column of Table 10.1b.
This is a very large air-cooled heat exchanger with four bays, and a total bare tube heat
transfer area of 20,960 ft2 (1948 m2). The principal reason for this is the very low tubeside
heat transfer coefficient and, thereby, the very low overall heat transfer coefficient. Thanks
to the low tubeside allowable pressure drop, the tubeside velocity is only 2.76 ft/s (0.84 m/s),
which for such a viscous liquid translates into a very low tubeside heat transfer coefficient of
111

12.3 Btu/h ft2 F/ (60.1 kcal/h m2 C). Since the tubeside is highly controlling (85.5%), there
is no merit in having a high airside heat transfer coefficient, and wasting fan power. Hence,
the air flow rate was kept quite low, so that the power consumption is only 10.3 HP (7.68
kW) per fan. This would at least minimize the operating cost on the airside!
Now, let us see what would happen if a higher tubeside pressure drop is permitted.
Consider a tubeside pressure drop of 68.3 psi (4.8 kg/cm2). The new thermal design is
detailed in the second column of Table 10.1b. Note the huge reduction in the bare tube heat
transfer area, from 20,960 ft2 (1948 m2) to 11,965 ft2 (1112 m2). The number of bays has
been reduced from four to two. The key to this stark change is the much higher tubeside
velocity, 6.43 ft/s (1.96 m/s) compared to 2.76 ft/s (0.84 m/s) in the earlier design. As a
consequence, the Reynolds number at the inlet/midpoint/outlet of the exchanger has
increased from 1166/992/896 to 2766/2299/2084, thereby pushing up the tubeside heat
transfer coefficient from 12.3 Btu/h ft2 F/ (60.1 kcal/h m2 C) to 26.8 Btu/h ft2 F/ (130.7
kcal/h m2 C). Despite such a significant increase in the tubeside heat transfer coefficient, the
latter is still largely controlling the overall heat transfer process (74%). This is because the
flow is still in the laminar region on the tubeside, where it is simply not possible to achieve
the high heat transfer coefficients that are prevalent in the turbulent regime.
With the increase in the tubeside velocity, as one would expect, the tubeside pressure
drop has jumped up from 15 psi (1.053 kg/cm2) to 68.3 psi (4.8 kg/cm2). As for fan power,
while it has increased per fan from 10.3 HP (7.68 kW) to 20 HP (14.9 kW), the total power
consumption has actually reduced from 82.4 HP (61.4 kW) to 80 HP (59.6 kW). This is
because the number of bays has been reduced from four to two and thereby the number of
fans has reduced from eight to four!
Thus, while the operating cost (due to tubeside pumping power) of the second design is
significantly higher than that of the first design, the first cost and the operating cost (due to
fouling) of the second design are far lower. Therefore, an overall cost assessment of the two
designs will have to be carried out to determine which design has the lower total cost.
10.1.6.3 Allow sufficient margin in pressure drop
When heavy fouling is anticipated, it would be wise to leave sufficient margin (say, 30
40%, and perhaps even more) between the allowable and calculated values of pressure
drop. (A more scientific approach is to consider an appropriate fouling layer thickness, as
discussed in the next section.) The idea is to permit an exchanger to operate at the design
throughput, even with the increased pressure drop due to fouling. If the margin is not
provided, the flow rate will be reduced in the fouled condition, such that the available
pressure drop is not exceeded, thereby limiting plant capacity.
10.1.6.4 Use fouling layer thickness
Although heat exchanger designers invariably employ a fouling resistance to account for
fouling in terms of extra thermal resistance to heat transfer, they do not generally translate the expected fouling into a fouling layer thickness to account for increased pressure
drop. Evidently, as a fouling deposit builds up, it will reduce the flow area and consequently result in an increased pressure drop.
It is quite straightforward to translate the fouling resistance into a fouling layer thickness
if the thermal conductivity of the deposit is known. Unfortunately, the thermal conductivity
of fouling deposits from hydrocarbon liquids is not readily available. However, the thermal
conductivity of asphalt is 0.43 Btu/h ft F (0.64 kcal/h m C). Since fouling layer thickness
will be applicable only for very fouling fluids such as vacuum gas oils, fuel oils, reduced
112

Table 10.3a: Principal process parameters for Case Study 10.2


1. Liquid flow rate, lb/h (kg/h)
2. Temperature in/out, F (C)
3. Heat duty, MM Btu/h (MM kcal/h)
4. Allowable pressure drop, psi (kg/cm2)
5. Inlet/outlet viscosity, cp
6. Inlet/outlet density, lb/ft3 (kg/m3)
7. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
8. Design air temperature, F (C)

185,200 (84,000)
473 (245)/176 (80)
31.0 (7.8)
35.6 (2.5)
0.7/15.0
49.9 (800)/56.4 (903)
0.0039 (0.0008)
113 (45)

crude (long residue), short residue, and asphalt, let us consider a somewhat lower value of
0.34 Btu/h ft F (0.5 kcal/h m C) for the deposits from heavy and dirty hydrocarbon liquids
other than asphalt.
Let us also bear in mind that since fouling layer thickness is directly proportional to
thermal conductivity, the higher the thermal conductivity considered, the higher will be the
fouling layer thickness.
Considering a uniform thermal conductivity of 0.34 Btu/h ft F (0.5 kcal/h m C) yields
the results shown in Table 10.2. These values do not look unreasonable since physical
inspection of fouled tube bundles handling such services will tend to corroborate them.
Most sophisticated thermal design software packages have a provision for incorporating
a fouling layer thickness. The following case study is now presented to demonstrate the
effect of applying a tubeside fouling layer thickness on the performance of an air-cooled heat
exchanger.

CASE STUDY 10.2: USING A FOULING LAYER THICKNESS ON THE TUBESIDE


Consider the case of a liquid cooler in a refinery unit, having the principal process parameters specified in Table 10.3a.
A thermal design was made for this service and the principal construction and
performance parameters are detailed in Table 10.3b. The following may be noted:
a) There are 8 tube rows and 16 tube passes. The large number of tube passes was
required since the tubeside flow rate is relatively low compared to the heat duty
and the liquid is quite viscous.
b) The tubeside velocity is a satisfactory 3.1 ft/s (0.95 m/s) and the tubeside pressure drop is within the permissible limit of 35.6 psi (2.5 kg/cm2).
c) The airside pressure drop is quite low (0.44 in. or 11.1 mm WC) and so is the
power consumption per fan (13 HP or 9.7 kW). This is because the tubeside is
highly controlling for heat transfer (73.7%) whereas the airside thermal resistance is only 15.34%. Thus, since there is no incentive for a higher airside heat
transfer coefficient, it is prudent to minimize the operating cost (fan power).
Since our process stream is dirty, let us now take a look at what the HTRI software predicts when we apply a fouling resistance of .02 in. (0.4 mm), as suggested in Table 10.2
for a stream having a fouling resistance of 0.0039 h ft2 F/Btu (0.0008 h m2 C/kcal). This
will represent the performance of the exchanger in the fouled condition. These results are
also shown in Table 10.3b.
As expected, the tubeside velocity has increased from 3.1 ft/s (0.95 m/s) to 3.38 ft/s
(1.03 m/s) and the pressure drop from 32.4 psi (2.28 kg/cm2) to 38.7 psi (2.72 kg/cm2),
113

Table 10.3b: Principal construction and performance parameters for Case Study 10.2

1. No. of bays
2. No. of bundles per bay
3. Tube OD thickness length, in. (mm)
4. Fin height, in. (mm) fin density, fins/in. (fins/m)
5. No. of tubes per row no. of rows
6. No. of tube passes
7. Total bare tube area, ft2 (m2)
8. Fans per bay fan diameter, ft (m)
9. Total air flow rate, lb/h (kg/h)
10. Air outlet temperature, F (C)
11. MTD, F (C)
12. Airside pressure drop, in. WC (mm WC)
13. Fan power, HP (kW)
14. Connections per bundle, in. (mm)
15. Tubeside velocity, ft/s (m/s)
16. Tubeside pressure drop, psi (kg/cm2)
17. Heat transfer coefficient,
Tubeside
Btu/h ft2 F/ (kcal/h m2 C)
Airside
Overall
18. Overdesign, %

On tube ID
On tube OD

Fouling layer thickness


Nil
0.4 mm
2
2
0.984 (25) 0.098 (2.5) 354 (9000)
0.49 (12.5) 11 (433)
45 8
16
10,663 (991)
2 12 (3.657)
2,094,400 (950,000)
175.1 (79.5)
157.3 (87.4)
0.44 (11.1)
13.0 (9.7)
In: 2 3 (75)
Out: 2 3 (75)
3.1 (0.95)
3.38 (1.03)
32.4 (2.28)
38.7 (2.72)
35.6 (173.9)
38.1 (186.2)
28.5 (139.1)
29.3 (143)
137 (669)
137 (669)
21.0 (102.6)
21.3 (104.1)
12.4
14.0

which is beyond the permissible limit of 35.6 psi (2.5 kg/cm2). Thus, if the specified value of
fouling layer thickness is considered realistic, the present design is not acceptable and the
number of tubes per row will have to be increased from 45 to 48 in order to contain the
tubeside pressure drop to within 35.6 psi (2.5 kg/cm2).
The tubeside heat transfer coefficient based on the tube ID has increased from 35.6
Btu/hft2F (173.9 kcal/h m2C) to 38.1 Btu/hft2F (186.2 kcal/h m2C), which is an increase
of 7%. However, the tubeside heat transfer coefficient based on the tube OD has increased
from 28.5 Btu/hft2F (139.1 kcal/h m2C) to 29.3 Btu/hft2F (143 kcal/h m2C), which is an
increase of only 2.8%. This is because, with the addition of the fouling layer thickness of
0.0157 in. (0.4 mm), the tube ID has reduced from 0.787 in. (20 mm) to 0.756 in. (19.2 mm),
so that the ratio of tube OD/tube ID has increased from 1.25 to 1.302. Therefore, when the
tubeside heat transfer coefficient based on the tube ID is converted to the heat transfer
coefficient based on the tube OD, the extent of the change is less. As for the overall heat
transfer coefficient, it has increased only by 1.46%. Thus, while the tubeside pressure drop
has gone up significantly with the application of the fouling layer thickness, the overdesign
margin has increased marginally.
10.1.6.5 Use wire-fin tube inserts
Wire-fin tube inserts (see Chapter 13, Fig. 13.3) were developed to increase the heat
transfer coefficient for laminar flow inside tubes. However, it was discovered later that
the use of these inserts also resulted in a profound reduction in fouling [3,4]. This is only
to be expected, considering that (a) the principal action of these inserts is to prevent
boundary layer separation, and promote radial mixing from the tube wall to the center of
114

Table 10.4a: Principal process parameters for Case Study 10.3


1. Fluid

Distillation column
overhead

2. Flow rate, lb/h (kg/h)

467,400 (212,000)
262.4 (128)/131 (55)

3. Temperature in/out, F (C)


4. Weight fraction vapor, in/out
5. Operating pressure, psia (kg/cm2 abs.)

0.12/0.085
455.0 (32.0)

6. Total allowable pressure drop, psi (kg/cm2)

3.56 (0.25)

7. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)

0.00195 (0.0004)

8. Heat duty, MM Btu/h (MM kcal/h)


9. Vapor
Density in/out, lb/ft3 (kg/m3)
properties
Viscosity in/out, cp
Specific heat in/out, Btu/lb F (kcal/kg C)
Thermal conductivity in/out, Btu/h ft F (kcal/h m C)
10. Liquid
Density in/out, lb/ft3 (kg/m3)
properties
Viscosity in/out, cp
Specific heat in/out, Btu/lb F (kcal/kg C)
Thermal conductivity in/out, Btu/h ft F (kcal/h m C)
11. Design air temperature, F (C)

47.73 (12.026)
0.387 (6.2)/0.356 (5.7)
0.0135/0.011
1.38/1.685
0.094 (0.14)/0.0894 (0.133)
48.05 (770)/ 51.2 (820)
0.9/2.2
0.537/0.485
0.065 (0.097)/0.084 (0.125)
107 (41.7)

the tube, and (b) the lack of turbulence is the root cause for both inefficient heat transfer
and heavy fouling. The churning action not only increases the heat transfer coefficient by
increased convection, but also minimizes the deposition of foulants due to the increased
turbulence. Wire-fin inserts have been demonstrated to reduce tubeside fouling to a dramatic level for many dirty services in the chemical process industries, and this appears to
be an excellent application. However, for reasons unknown to this author, their use has
not really lived up to their potential. For a detailed discussion on wire-fin tube inserts including a case study, see Section 13.4.

10.2 Airside Fouling


Airside fouling is usually not reported to be a major problem; but in plants located in
dusty areas, it may become a constraint after several years of operation. Deposits on the
fins and tubes are usually restricted to the first two tube rows. Since tubes are located on
a triangular (staggered) pitch and finned tubes usually have 1011 fins/in. (394433
fins/m), cleaning of deposits can be difficult. Using compressed air generally does not
produce a satisfactory result. High-pressure jets of water may harm the extremely thin
fins of aluminum. The best results are obtained by wetting the finned tubes with a detergent solution, followed by hydrojetting at moderate pressure.
If an air-cooled heat exchanger is to be located in a dusty environment, a few design
features may be extremely useful in reducing the extent of fouling, as follows:
a) Use of induced draft
The velocity of air entering a tube bundle is appreciably less for induced draft
than for forced draft. Therefore, the amount of dust and other foulants carried
into a tube bundle will be less with induced draft.
b) Restricting the fin density
115

Table 10.4b: Principal construction and performance parameters for Case Study 10.3
1. No. of bays no. of bundles per bay
2. Fin height tube pitch, in. (mm)
3. Fin density, fins/in. (fins/m)
4. Number of tubes per row
5. Number of rows no. of tube passes
6. Total bare heat transfer area, ft2 (m2)
7. Approximate bundle width, ft (m)
8. Air flow rate, lb/h (kg/h)
9. Airside
outlet temperature, F (C)
pressure drop, in. (mm) WC
heat transfer coefficient, Btu/h
ft2 F (kcal/h m2 C)
10. Tubeside
pressure drop, psi (kg/cm2)

11. Thermal resistance, %

heat transfer coefficient, Btu/h


ft2 F (kcal/h m2 C)
Airside
Tubeside

12. MTD, F (C)


13. Overall heat transfer coefficient, Btu/h ft2 F/ (kcal/h m2 C)
14. Overdesign, %

42
0.49 (12.4) 2.36 (60)
11 (433)
35
62
14,320 (1331)
7.06 (2.15)
5,511,500 (2,500,000)
143.1 (61.7)
0.59 (15.0)
162.4 (793)
2.9 (0.204)
127.2 (621)
36.5
46.66
61.7 (34.3)
59.3 (289.7)
10.0

It may be a very good idea to limit the fin density to 7 or 8 fins/in. (276 or 315
fins/m) instead of the usual 11 fins/in. (433 fins/m).
c) Restricting the number of tube rows
By restricting the number of tube rows to four, it will be possible to obtain a
higher air velocity inside the tube bundle, which will minimize the extent of fouling.
d) Using a fouling layer thickness
In order to ensure that reduced flow of air (due to airside fouling) does not become a serious constraint in the performance of an air-cooled heat exchanger, the
effect of fouling can be simulated by incorporating a fouling layer thickness on
the airside. Some software packages have the facility of accepting a fouling layer
thickness on the airside. If a given software package does not have this feature,
the fin thickness may be fed correspondingly higher, e.g., 0.02 in. (0.5 mm) instead of 0.016 in. (0.4 mm), to simulate the effect.
Thus, satisfactory operation of an air-cooled heat exchanger can be ensured by anticipating and catering to the adverse effect of airside fouling.

CASE STUDY 10.3: USING A FOULING LAYER THICKNESS ON THE AIRSIDE


Consider a partial air-cooled condenser whose principal parameters are specified in Table
10.4a. There are quite a few interesting features to observe in the process data as follows:
a) The unusually high values of vapor phase thermal conductivity and specific heat
clearly reveal the presence of a high percentage of hydrogen in the overhead
stream.
116

Table 10.4c: Analysis of fouling layer thickness for Case Study 10.3

Air mmidpoint velocity,


ft/s (m/s)
Air pressure drop, in. WC
(mm WC)
Power consumption per
fan, HP (kW)
Airside heat transfer
coefficient, Btu/h ft2 F
(kcal/h m2 C)
Overall heat transfer
coefficient, Btu/h ft2 F
(kcal/h m2 C)
MTD, F (C)
Overdesign, %

FLT = 0

FLT = 0.004 in.


(0.1 mm)

FLT = 0.008 in.


(0.2 mm)

FLT = 0.008 in. (0.2


mm) and air flow rate
= 5.07 MM lb/h (2.3
MM kg/h)

22.8 (6.95)

24.6 (7.5)

26.7 (8.15)

24.6 (7.51)

0.59 (15.0)

0.67 (17.0)

0.77 (19.45)

0.67 (17.0)

22.9 (17.1)

25.5 (19.0)

28.7 (21.4)

22.9 (17.1)

162.4 (793)

168.5 (822.5)

175.5 (856.7)

168.6 (823.1)

59.3 (289.7)

60.1 (293.4)

60.9 (297.4)

60.2 (293.7)

61.7 (34.3)
10

61.7 (34.3)
11.4

61.7 (34.3)
12.9

60.5 (33.6)
9.3

FLT -> Fouling layer thickness

b) Most of the vapor has already condensed prior to entering this condenser. The
vapor weight fraction is 0.12 at the inlet (262.4F or 128C) and 0.085 at the outlet (131F or 55C), which makes it a very wide condensing-range mixture.
c) Although the operating pressure is fairly high, the vapor density is quite low,
thanks to the presence of a significant proportion of hydrogen.
d) Due to (b) and (c) above, the tubeside heat transfer coefficient may be expected
to be quite low, as discussed earlier in Chapter 6.
Coming to the thermal design, tubes were to be of carbon steel, 0.984 in. (25 mm) OD
0.098 in. (2.5 mm) thickness 34 ft (10.36 m) long. Fins were to be of aluminum. The
design ambient temperature was 107F (41.7C).
A thermal design was prepared and the principal construction and performance
parameters are detailed in Table 10.4b. Now, in order to demonstrate the effect of airside
fouling, this design was checked with fouling layer thicknesses of 0.004 in. (0.1 mm) and
0.008 in. (0.2 mm). The results are shown in Table 10.4c. It will be noticed that with the
application of a fouling layer thickness, there is a marked increase in the airside midpoint
velocity, pressure drop, and fan power consumption. However, the airside heat transfer
coefficient increases marginally.
Does this mean that the fan power consumption will be considerably higher if we are to
cater to a fouling layer thickness of, say, .01 in. (0.2 mm); that is, 28.7 HP (21.4 kW) instead
of 22.9 HP (17.1 kW)? Not necessarily. We can simulate the performance of this air-cooled
condenser by lowering the air flow rate by simple trial and error until we obtain the same fan
power consumption (22.9 HP or 17.1 kW). In the present instance, by lowering the air flow
rate from 5,511,500 lb/h (2,500,000 kg/h) to 5,070,058 lb/h (2,300,000 kg/h), we obtain a
fan power consumption of 22.9 HP (17.1 kW). Due to the lower air flow rate, there is a
small reduction in both the MTD and the airside heat transfer coefficient, so that the
overdesign reduces marginally from 10% to 9.3%. Thus, an airside fouling layer thickness of
0.004 in. (0.1 mm) or 0.008 in. (0.2 mm) does not have any serious repercussions on the
performance of an air-cooled heat exchanger.
117

However, if the fouling layer thickness is likely to be higher [for example, 0.016 in. (0.4
mm)], then the performance could deteriorate significantly. The important thing to realize is
that we should address all realistic operating conditions that an air-cooled heat exchanger
can be subjected to, so that it operates satisfactorily even under these conditions.

References
[1] Taborek, J., Aoku, T., Ritter, R.B., Palen, J.W., and Knudsen, J.G., 1972, Fouling The Major Unresolved Problem in Heat Transfer, Parts I and II, Chem. Eng. Prog., 68(2), pp. 5967
and 68(7), pp. 6978.
[2] Tubular Exchanger Manufacturers Association, 1999, Standards of the Tubular Exchanger
Manufacturers Association, 8th Edition, TEMA, New York.
[3] Gough, M.J., and Rogers, J.V., 1987, Reduced Fouling by Enhanced Heat Transfer Using
Wire-Matrix Radial Mixing Elements, AIChE Symp. Series, 83(257), pp. 1621.
[4] Gough, M.J., and Rogers, J.V., 1991, Getting More Performance from Heat Exchangers,
Processing, July, pp. 1516

Further Reading
1.
2.
3.
4.
5.
6.
7.
8.

Epstein, N., 1978, Fouling in Heat Exchangers, Proc. Sixth International Heat Transfer
Conf., Toronto, Hemisphere, New York, Vol. 6, pp. 235253.
Garret-Price, B.A., Smith, S.A., Watts, R.L., Knudsen, J.G., Marner, W.J., and Suitor, J.W.,
1985, Fouling of Heat Exchangers, Characteristics, Costs, Prevention, Control and Removal,
Noyes, Park Ridge, NJ.
Hewitt, G.F., ed., 2002, Heat Exchanger Design Handbook 2002 (HEDH2002), Begell
House, Inc., New York, Redding, CT.
Knudsen, J.G., 1984, Fouling of Heat Exchangers: Are We Solving the Problem? Chem.
Eng. Prog., Feb., pp. 6369.
Melo, L.F., Bott, T.R., and Bernardo, C.A., eds. 1988, Fouling Science and Technology, Kluwer, Dordrecht.
Somerscales, E.F.C., and Knudsen, J.G., eds., 1981, Fouling of Heat Transfer Equipment,
Hemisphere, New York.
Kakac, S., Bergles, A., and Mayinger, F., 1981, Heat Exchangers: Thermal-Hydraulic Fundamentals and Design, Hemisphere, New York.
Mukherjee, R., 1996, Conquer Heat Exchanger Fouling, Hydrocarbon Process., 75(1), pp.
121127.

118

CHAPTER 11

Control of
Air-Cooled Heat Exchangers
11.1 Introduction
The control of air-cooled heat exchangers is required to accomplish various goals as follows:
1) Control of distillation column operating pressure
2) Prevention of excessive cooling of liquid products (winterization) in order to
Prevent congealing (pour point)
Prevent solidification fouling
Incorporate a safe margin of subcooling for volatile streams
3) Control of reflux temperature (for total or partial condensers)
4) Energy conservation
5) Protection from adverse atmospheric effects such as hot air recirculation, excessive solar radiation, and heavy rainfall.

11.2 Methods of Control


11.2.1 Bypassing of process fluid
Partial bypassing of the process fluid around an air-cooled heat exchanger (Fig. 11.1) will
prevent any undesirable overcooling when the ambient temperature is lower than for
which the system was designed. Since this will result in a reduction in the tubeside flow
velocity, its use is not recommended for dirty streams, since the low velocity will aggra-

Fig. 11.1 Partial bypassing of process fluid around an air-cooled heat exchanger
119

vate fouling. Hot-vapor bypassing is a very popular method of control of the top pressure
of a distillation column.
11.2.2 Switching fans on/off
Switching fans on and off is a simple method of control, and is often used if there are a
large number of fans in a single service. Air flow control is in incremental steps. Power
savings are at the mercy of the operator, unless control is automatic. This mode of control
can cause water hammer in air-cooled condensers, or tube-to-tubesheet joint leakage due
to differential expansions of bundles.
11.2.3 Use of two-speed motors
This is a rather inexpensive method of controlling the air flow rate. The switching over
from normal speed to half speed may be manual or automatic. Since there are usually two
fans per bay, the variation of air flow rate will be in large step values. However, if the
number of sections and, therefore, the total number of fans is large, a more gradual control can be achieved. This method of control is not very popular for obvious reasons.
11.2.4 Use of louvers
A louver is a device that comprises a large number of blades mounted in a frame. It covers the entire face area of a tube bundle, and by manipulating the blades, the amount of
air that passes through the bundle can be controlled. Louvers can be manually controlled
from the platform or grade, or remotely operated by hand or by a TIC. The material of
construction is usually aluminum.
There are two basic designs of automatic louversparallel action and opposed action.
In the parallel-action louver, there is a series of blades in parallel having a common tie bar
and a common linkage. This type of louver begins to lose control when the blades are 3540
deg open.
The opposed-action louver has each blade opposed to the previous blade, with the
alternate blades having a common tie bar. An opposed-action louver in the closed position
and one in the half-open position are shown in Figs. 11.2a and 11.2b, respectively.
Manufacturers claim that the opposed- action louver offers effective control across the entire
90 deg of blade rotation. Consequently, where louvers are used for control, opposed blade
action is usually preferred.

(a)

(b)
Fig. 11.2 Opposed-action louver in the (a) closed position and (b) half-open position
120

Louvers placed at the top of the tube bundles for forced draft applications can be used to
control the air flow rate as desired. The control may be manual or automatic. An added
advantage of the use of louvers is that they afford protection to the tube bundles (finned
tubes) against hail. Yet another advantage is that for a combined section (a section having
two or more tube bundles handling different services), louvers can provide control of air
flow rate across individual tube bundles.
However, if the air flow rate through one tube bundle is reduced by partially closing its
louvers, the air flow rate through the other tube bundle(s) will increase to a certain extent.
Therefore, if fine control is required for a combined section, the fans will have to be
autovariable as well. It should be noted that only the use of autovariable fans can vary the air
flow rate uniformly through all the tube bundles served by themindividual control is not
possible. Due to the complex nature of control of air flow through individual tube bundles of
combined services, many licensors prefer not to combine services requiring precise control,
such as condensers.
Louvers are often used in conjunction with steam coils for cold start-up and protection
against freezing. A distinct disadvantage of the use of louvers is that it does not result in
power savings.
11.2.5 Use of autovariable fans
Autovariable, variable pitch, or controllable pitch are various terms employed to describe
axial fans that alter the pitch or blade angle while in operation, so as to deliver the precise
amount of air flow required to meet the heat duty of a given air-cooled heat exchanger, at
all times (Fig. 11.3).
An autovariable fan hub includes a device (usually a pneumatic controller) that can alter
the blade angle, even while the fan is in motion. Control is usually effected by means of a
signal from a temperature indicator controller (TIC) responding to the outlet temperature of
the process fluid. Thus, the air flow rate is reduced during cold ambient conditions, and
increased as the ambient temperature increases.
Evidently, the use of autovariable fans results in substantial energy savings, with
paybacks being as short as one year, or even less, in dry climates (having a large variation in
ambient temperature between summer afternoons and winter nights). Generally, 50%
autovariable and 50% manual fans are employed, since this is adequate to achieve the
required control for the severest variation in the range of climatic conditions encountered.
Since a bay usually has two fans, one fan is normally autovariable while the other is manual.

Fig. 11.3 Autovariable fan (Courtesy Moore Fans Ltd.)


121

The use of autovariable fans is a very common method of air-cooled heat exchanger control.
11.2.6 Use of variable frequency drives
Variable frequency drives (VFDs) provide the best way to control air flow, in terms of
energy efficiency and operating flexibility. They modulate the supply frequency to the
induction motor by changing the supply frequency and the speed of the motor, and hence
that of the fan is changed to deliver the required air flow rate at all times, meaning all
ambient temperatures. At lower air flow rates, the reduction in fan efficiency is much less
in variable frequency drives than in autovariable fans. Therefore, VFDs save more power
than do autovariable fans. Since VFDs also cost more than autovariable drives, the payback period of the two are quite similar.

Further reading
1.

Monroe, R.C., 1980, Consider Variable Pitch Fans, Hydrocarbon Process., Dec., pp. 122128.

122

CHAPTER 12

Operating Problems
in Air-Cooled Heat Exchangers
12.1 Introduction
In this chapter, we will take a look at some of the more common operating problems that
occur in air-cooled heat exchangers. First, we will discuss the problems that occur on the
tubeside and the problems that occur on the airside.

12.2 Problems on the Tubeside


12.2.1 Flow maldistribution
Since the process fluid has to be distributed from the header to the various tubes in each
row of tubes, some maldistribution is inevitable. The degree of maldistribution and the
consequent reduction in performance varies depending on whether the service is singlephase or condensing, with the latter being far more critical.

12.2.1.1 Single-phase services


Because of pressure changes along the inlet and the outlet headers and manifolds, varying
flow rates of the process fluid may be flowing through multiple tube bundles connected
in parallel. Consequently, the outlet temperature of the process fluid will be different
from the different bundles, resulting in a loss in performance of the unit. The extent of
this loss in performance will depend on the extent of maldistribution as well as the temperature approach between the process fluid outlet and the air inlet. The lower this temperature approach, the greater will be the reduction in performance for a given degree of
maldistribution.
Usually, the pressure drop of the process fluid in the manifolds and headers is
considerably less than that in the tube bundle itself (that is, in the tubes themselves).
Consequently, flow maldistribution is generally not a serious problem in single-phase aircooled heat exchangers, so that the piping manifold shown in Fig. 12.1a is normally
employed. Even when it does take place, there are usually no serious adverse effects. The
reduction in performance is usually well within the overdesign margin that is normally
incorporated.
However, when the process fluid happens to have a high pour point, overcooling in
some tubes due to maldistribution may lead to congealing. This potential blockage will lead
to further reduction in the flow rate, and a spiraling effect will result that may finally block
these tubes. Air cooling is generally not recommended for cooling such liquids, and if air
123

cooling is employed, steam coils are usually incorporated, and sometimes a sophisticated
recirculation system as well.
12.2.1.2 Two-phase services
Manifold
Flow maldistribution is practically inevitable with two-phase flow entering an air-cooled
heat exchanger. Although it can be minimized by employing a carefully designed manifold system as shown in Fig. 12.1b, it can never be totally eliminated. When flow maldistribution occurs between the various tube bundles of an air-cooled heat exchanger but is
limited to only flow rate and not the vapor quality, the effect will be precisely the same as
described above for single-phase streams. However, if the manifold is not designed carefully, separation of vapor and liquid may occur and thereby lead to a severe deterioration
in performance. Since specific heat is much lower than latent heat, the tube bundles that
receive predominantly liquid have a much lower heat duty to perform and therefore overperform; the tube bundles that receive predominantly vapor will have a considerably
higher heat duty to remove, and will naturally underperform.
Headers and tubes
In order to minimize lateral maldistribution of flow, and separated vapor and liquid flows
to the various tubes of a given tube bundle, it is recommended to
a) Use whole numbers of tube rows in a tube pass.
b) Design headers carefullyespecially to use an adequate number of nozzles for
each tube bundle. It is a good idea to use two nozzles at both inlet and outlet

(a) Single-phase service

(b) Two-phase service

Fig. 12.1 Piping manifold systems


124

wherever the bundle width is greater than 3.3 ft (1.0 m). The lower the operating
pressure, the greater is the difference in liquid and vapor densities. Thus, for
rather low operating pressures such as 28.4442.66 psig (23 kg/cm2 g), the use
of even three nozzles along the bundle width may be considered for wide tube
bundles [bundle width more than 8.2 ft (2.5 m)].
A more serious maldistribution occurs in condensers when there are two or more tube
rows in a given tube pass. Due to the separation of liquid and vapor by gravity in the
header, more liquid will tend to flow through the tubes of the bottom row(s) and more
vapors will tend to flow through the tubes of the upper row(s). This will result in unnecessary overcooling of the liquid and insufficient condensing of the vapor (Fig. 12.2).
12.2.2 Inadequate process cooling
Inadequate process cooling in air-cooled heat exchangers may be the result of one or
more of the following:

a) The actual heat duty is higher than the design heat duty due to a higher flow rate,
a higher inlet temperature, or even a change in the composition of the process
fluid.
b) The air flow rate may be lower than that for which the system was designed.
c) Excessive fouling inside tubes; that is, fouling greater than that anticipated at the
design stage. This is discussed in detail in Section 12.2.3.
d) Inadequacy of heat transfer surface area.
e) Process fluid maldistribution. This was just discussed in Section 12.2.1 above.
f) Partial hot air recirculation from the air-cooled heat exchanger outlet to the inlet,
resulting in an increase in the air inlet temperature to the unit, and thus a reduc-

Liquid(L) + Vapour/Gas(V)

Fig. 12.2 Maldistribution due to phase separation in header


125

tion in the MTD and thereby the performance (discussed later).


g) Air maldistribution caused by an inadequate face area of the tube bundle, insufficient height of the plenum chamber, blockages in the air flow path, or excessive
air leakage between tube bundles, and even within tube bundles (i.e., between
side frame and outermost tubes).
12.2.3 Excessive fouling
Fouling is an inevitable consequence of the heat transfer process itself. There are various
modes of fouling such as sedimentation, freezing, corrosion, chemical reaction, etc.
These may occur either singly or in any combination. The extent of fouling depends on
various factors such as the nature of the fluid, velocity, wall temperature, surface finish,
and the material of construction. Fouling can never be eliminated, but it can be minimized by good design practice. One of the principal features of good design is a sufficiently high tubeside velocity. The subject of fouling is discussed in detail in Chapter 10.
Section 10.1.6 specifically advocates certain guidelines to minimize tubeside fouling in
air-cooled heat exchangers.
One manifestation of excessive fouling inside air-cooled heat exchanger tubes is
congealing. Heavy hydrocarbon streams such as vacuum residue have a rather high pour
point, and hence the use of air-cooled heat exchangers for them is not warranted unless a
special recirculation arrangement is employed. In this setup, the exit warm air is partly
recycled and mixed with fresh ambient air so that the combined air stream is at the desired
temperature. This value should be such that the tube wall temperature remains above the
pour point. Since the ambient temperature varies considerably between day and night and
through the seasons, the extent of recirculation will have to be varied accordingly. This is
achieved by the use of autovariable fans and louvers, as shown in Fig. 13.1. Since this
arrangement becomes rather expensive, many plant operators prefer to use a more robust
closed-circuit tempered (dematerialized) water cooling system, with an inlet temperature
higher than the pour point of the congealing liquid. For a more detailed description of
recirculation air-cooled heat exchangers, please see Chapter 13.

12.3 Operating Problems on the Airside


12.3.1 High air inlet temperature
A high air inlet temperature will evidently lower the MTD, which is the driving force for
heat transfer, and thereby lower performance. It is therefore very important to select the
air design temperature carefully. This is discussed in detail later in the Section 12.3.4.
12.3.2 Hot air recirculation
Hot air recirculation has been known to be particularly deleterious to the thermal performance of air-cooled heat exchangers. It is much more pronounced under the influence
of crosswinds. As a result of partial hot air recirculation, a part of the air leaving an aircooled heat exchanger mixes with fresh air entering the unit, thereby resulting in an increase in the temperature of air entering the bundles. This will evidently result in a decrease in the mean temperature difference (MTD). The lower the MTD of an air-cooled
heat exchanger for the design conditions, the greater will be the reduction in the MTD
due to the same degree of hot air recirculation.
Since induced draft (ID) fans throw the exit air to a higher altitude than do forced draft
126

(FD) fans, the extent of hot air recirculation is less for ID fans than for FD fans. Thus, for
services where the process fluid is cooled and/or condensed to a relatively low temperature
(difference between the process fluid outlet temperature and the inlet air temperature) that is
less than 914F (58 C), ID fans are preferred.
Specific causes of hot air recirculation are as follows:
a) Adjacent units located at different elevations, so that the exit air from the unit located at a lower elevation can mix with the air entering another the unit located at
a higher elevation (Fig. 12.3a). A good practice is to locate all adjacent units
(having the same type of draft) at the same elevation.
b) Induced draft units located adjacent to forced draft unitsthe exit air from the
forced draft unit is more likely to partially mix with the air entering the adjacent
induced draft unit since it has no plenum at the bottom (Fig. 12.3b). Thus, adjacent units should have the same type of draft.
c) Units located in front of a downward obstruction such as a building (e.g., compressor house) (Fig. 12.3c). Units should be located clear of obstructions that are close
enough to result in hot air recirculation.
d) Adjacent units located with small-to-medium gaps between one another. Evidently,
the exit from one unit has a much greater probability of recirculating to the other,
depending on the wind direction (Fig. 12.3d). Therefore, adjacent units should be
located contiguously.
(a)

(b)

(c)
Prevailing Hot Wind
Compressor Building

(d)

Fig. 12.3 Hot air recirculation: (a) adjacent units at different elevations, (b) induced draft units
located adjacent to forced draft units, and (c) units located in front of an obstruction (d) adjacent
units located at a small gap
127

By applying good engineering judgment in plant layout, it is possible to restrict hot air
recirculation to a very minimum.
12.3.3 Inadequate air flow
After maldistribution of the process fluid in air-cooled condensers, inadequate air flow is
perhaps the most common reason for unsatisfactory performance of air-cooled heat exchangers.
The air flow rate may be lower than design, caused by one or more of the following:
incorrect fan selection
wrong installation of the fans (e.g., a large clearance between the fan ring and the
fan tip)
slippage of belts
incorrect blade angle
excessive air leakage between and/or around tube bundles due to imperfect sealing
The last point is often overlooked by operating personnel. As per API 661 [1], any clearance over 1 in. (25 mm) should be considered excessive and should be blocked to prevent
air bypassing.
12.3.4 Incorrect selection of design ambient temperature
The selection of the design air temperature is of paramount importance for the efficient
operation of air-cooled heat exchangers. In the case of water cooling, the variation of the
supply water temperature to the coolers between day and night during the summer
months (the controlling period) is not considerable. However, the variation in the ambient
temperature through the year is quite large, especially in dry climates. Therefore, the selection of the design air temperature for air-cooled heat exchangers is not very straightforward. The higher this temperature, the lower will be the mean temperature difference
(MTD) and, therefore, the higher the heat transfer area. Consequently, if a very high
value is adopted, the first cost will be high. On the other hand, if an unrealistically low
value is considered, the coolers will not perform satisfactorily whenever the ambient
temperature is higher than the design value.
The Gas Processors Suppliers Association (GPSA) recommendation is that the design
air temperature should be that value of ambient temperature that is not exceeded during 2%
of the time during the year. Evidently, this will have to be determined from meteorological
data.
A very important factor in this matter is whether or not the air-cooled heat exchanger is
followed by a trim (water) cooler. When the process fluid outlet temperature is rather low,
cooling of the last few degrees by air becomes very expensive, due to the extremely low
mean temperature difference (MTD). For example, if a process fluid is to be cooled from
212F (100C) to 122F (50C) at a plant site where the design water temperature is 91.4F
(33C) and the design air temperature is 107.6F (42C), it will be prudent to cool from
212F (100C) to ,say, 140F (60C) by air, and the balance by water.
If an air-cooled heat exchanger is followed by a trim cooler, the design air temperature
for the air-cooled heat exchanger may be somewhat on the lower side. When the air
temperature is higher than this value, the process fluid outlet temperature from the air-cooled
heat exchanger will be a little higher than the design value. However, the trim cooler can be
designed for the process fluid outlet temperature from the air-cooled heat exchanger,
128

corresponding to the highest expected ambient temperature. This is based on the logic that
since an air-cooled heat exchanger is much costlier than its trim cooler, the overall cost of
the air-cooled heat exchanger and the trim cooler will be lower if designed this way. To
illustrate, for a plant site where the maximum air temperature is 113F (45C), an air-cooled
heat exchanger may be designed for an ambient temperature of 107.6F (42C) and the trim
cooler designed for the process fluid outlet temperature from the air-cooled heat exchanger
corresponding to an ambient temperature of 113F (45C).
If an air-cooled heat exchanger does not have a trim cooler, the design ambient
temperature will have to be more conservative. However, the very fact that there is no trim
cooler indicates that the process fluid temperature is relatively high, in which case the
increase in heat transfer area due to the higher design ambient temperature will not be
substantial. This is because the mean temperature difference (MTD) is fairly high in the first
place, so that the reduction in the same due to a somewhat higher design ambient
temperature is not considerable. This concept was discussed earlier in Chapter 3, and Case
Study 3.1 exemplified the concept.
12.3.5 High noise level
A high noise level may be particularly unpleasant to plant personnel, and is generally attributable to one or more of the following:

a) Wrong selection of the fan and/or drive and/or motor. This is usually not a problem, as long as the matter is in the hands of fan suppliers, who have specialized
and sophisticated know-how.
b) Increased air flow (compared to design) due to wrong selection of blade angle, or
readjustment of blade angle for delivering higher air flow rate to meet increased
heat duty.
c) Worn-out fan and/or motor bearings.
d) Worn-out, misaligned, or slipping drive belts.
Generally speaking, high noise level is not a common problem in air-cooled heat exchangers.
12.3.6 Airside fouling
This is discussed at length in Section 10.2.

12.4 Performance Evaluation of Air-Cooled Heat Exchangers


In the event of unsatisfactory performance of an air-cooled heat exchanger, it will be necessary to establish the cause or causes responsible for the same before any remedial action can be taken. This diagnosis can be arrived at only after comprehensive testing is
carried out.
The air flow pattern in air-cooled heat exchangers is extremely complex as well as
difficult to measure. The air velocity is typically about 19.7 ft/s (6 m/s) through the fan,
about 9.84 ft/s (3 m/s) leaving the fan, and about 6.6 ft/s (2 m/s) at the inlet to the fan. The
best location to measure air velocity is at the fan inlet using a small time-averaging
anemometer.
The AIChE Equipment Testing Procedure for air-cooled heat exchangers [2] describes
in detail procedures for the measurement of flow rates, pressures, pressure drops, and
temperatures of both the process fluid and air, as well as motor horsepower. API
129

Recommended Practice 631M [3] describes in detail procedures for conducting noise tests in
order to establish that noise levels are within specified levels.

References
[1] API, 1992, Air-cooled Heat Exchangers for General Refinery Services, API Standard 661, 3rd
Ed., April, American Petroleum Institute, Washington, DC.
[2] AIChE, 1978, AIChE Equipment Testing ProcedureAir Cooled Heat Exchangers: A Guide
to Performance Evaluation, AIChE, New York.
[3] API, 1981, Measurement of Noise from Air-cooled Heat Exchangers, API Recommended
Practice 631M, 1st Ed., June (Reaffirmed Oct. 1985), American Petroleum Institute, Washington, DC.

Further reading
1.
2.
3.
4.
5.
6.
7.

Hewitt, G.F., ed.., 2002, Air-cooled Heat Exchangers, Vol. 3, Section 3.8, Heat Exchanger
Design Handbook 2002 (HEDH 2002), Begell House, Inc., New York, Redding, CT..
Larinoff, M.W., Moles W.E., and Reichhelm, R., 1978, Design and Specification of Aircooled Steam Condensers, Chem. Eng., May 22.
Berryman, R., and Russell, C., 1986, Assessing Airside Performance of Air-cooled Heat
Exchangers, Process Eng., April, pp. 5964.
AIChE, 1978, AIChE Equipment Testing ProcedureAir Cooled Heat Exchangers: A Guide
to Performance Evaluation
Shastri, S.S. et al., 2001, Enhance Air-cooled Heat Exchanger Performance, Hydrocarbon
Process., Dec., pp. 4955.
Berryman, R. and Russell, C., 1985, Troubleshooting air-cooled heat exchangers, Process
Eng., Apr., pp. 2528.
Mukherjee, R., 1996, Conquer Heat Exchanger Fouling, Hydrocarbon Process., 74(1)., pp.
121127.

130

CHAPTER 13

Special Applications
In this chapter, we will take a look at some special applications in air-cooled heat exchangers for addressing either special requirements or special services.

13.1 Combined Services


A difficulty arises when several small coolers have to be designed to be located on a
broad pipe rack of, say, 26.2 ft (8 m) width. The heat transfer areas being rather small,
single tube bundles of small width [say, 3.3 ft (1 m) to 9.9 ft (3 m)] may suffice. However, it is impractical to have long, slender bays with three, four, or even more fans,
drives, plenums, etc.
In such cases, it is a common practice to combine several such coolers in a single bay, so
that just the usual two fans per bay become feasible. For example, three tube bundles of
width 3.3 ft (1 m), 6.6 ft (2 m), and 9.9 ft (3 m), catering to different services, may be
combined to form a bay having a width a little over 19.7 ft (6 m).
The disadvantage in combining different services in one section is the loss of individual
control. Consequently, this practice is not recommended for condensers, and is limited to
product coolers. In other words, a bay of combined services will usually have no condenser,
but only run-down product coolers. Let us now demonstrate an actual case of combined
services.

CASE STUDY 13.1: COMBINED SERVICES


In a fluidized catalytic cracking (FCC) unit of an oil refinery, three air-cooled heat exchangers were to be designed for a design ambient temperature of 107.6F (42C) and the
process parameters detailed in Table 13.1a. Carbon steel tubes of 0.984 in. (25 mm OD)
0.098 in. (2.5 mm) thick 34.5 ft (10.5 m) long were to be used, since the air-cooled
heat exchangers were to be mounted on a pipe rack of 32.8 ft (10 m) width.
Since the individual heat duties were rather small, a combined design was prepared
wherein three tube bundles (one for each service) were combined in a single bay using Gfinned tubes having 11 fins/in. (433 fins/m) and 0.49 in. (12.5 mm) high fins of aluminum at
a transverse pitch of 2.36 in. (60 mm). The principal construction and performance
parameters of the three designs are elaborated in Table 13.1b.
Two fans of 12 ft (3.657 m) diameter each were selected to force the total air flow rate
of 1,253,600 lb/h (568,600 kg/h) against a static pressure of 0.48 in. (12.2 mm) WC,
consuming 18.55 HP (13.84 kW) each. Two motors of 22.5 HP (16.785 kW) each were
selected to drive the fans. Split headers were used for all three bundles, since all of them had
design temperatures in excess of 392F (200C), and the first two had a high temperature
131

Table 13.1a: Principal process parameters for Case Study 13.1


1. Flow rate, lb/h (kg/h)
2. Inlet temperature, F (C)
3. Outlet temperature, F (C)
4. Heat duty, MM Btu/h (MM kcal/h)
5. Operating pressure, psig (kg/cm2 abs)
6. Allowable pr. drop, psi (kg/cm2)
7. Fouling resistance, h ft2 F/Btu (h m2
C/kcal)
8. Inlet/outlet viscosity, cp
9. Inlet/outlet density, lb/ft3 (kg/m3)

ACHE # 1
108,000 (49,000)
327.2 410 (164)
150.8 (66)
10.4 (2.62)
145 (10.2)
10 (0.7)
0.00195 (0.0004)

ACHE # 2
42,400 (19,230)
329 (165)
150.8 (66)
1.02
136.5 (9.6)
7.1 (0.5)
0.00146 (0.0003)

ACHE # 3
163,600 (74,200)
350.6 (177)
280.4 (138)
1.64
130.8 (9.2)
7.1 (0.7)
0.00146 (0.0003)

0.8/3.2
800/865

0.3/0.7
603/698

0.28/0.36
670/705

difference (between the inlet and the outlet temperatures of the process fluid) as well.
Looking at the overdesign values, we see that they are 36.8%, 19.6%, and 24.6% for
ACHE #1, ACHE #2, and ACHE #3, respectively. The latter two values are somewhat on
the higher side, but considering the small sizes of these bundles, they are acceptable.
However, the overdesign value for ACHE #1 is rather high, at 36.8%. This is really a
pressure drop limiting case, since with a lower number of tubes and the same number of tube
passes, the tubeside pressure drop will exceed the permitted value of 10 psi (0.7 kg/cm2). If
the number of tube passes is reduced from six to four (not five, since an odd number of
passes is not preferred for reasons of piping inconvenience), the overdesign value falls to a
mere 6.2%, due to a steep reduction in the tubeside heat transfer coefficient from 123.7
Btu/hft2F (604 kcal/h m2C/kcal) to 77.8 Btu/hft2F (379.9 kcal/h m2C/kcal). Thus, since
the number of tubes cannot be reduced, it is better to retain six tube passes and maintain a
decent tubeside velocity.
Table 13.1b: Principal construction and performance parameters for Case Study 13.1

1. Number of tubes per row


2. No. of rows no. of tube passes
3. Fin height, in. (mm) fin density, fins/in.
(fins/m)
4. Tube pitch, in. (mm)
5. Approximate bundle width, ft (m)
6. Air flow rate, lb/h (kg/h)
7. Air inlet temperature, F (C)
8. Air outlet temperature, F (C)
9. Airside pr. drop, in. (mm) WC
10. Tubeside heat transfer coefficient, Btu/h
ft2 F/ (kcal/h m2 C/kcal)
11. Tubeside velocity, ft/s (m/s)
12. Tubeside pr. drop, psi (kg/cm2)
13. Overall heat transfer coefficient, Btu/h
ft2 F/ (kcal/h m2 C/kcal)
14. MTD, F (C)
15. Heat transfer area (bare), ft2 (m2)
16. Overdesign, %
17. No. of fans fan dia., ft (m)
18. Total power consumption, HP (kW)

ACHE # 1

ACHE # 2

ACHE # 3

48
66

15
64
0.49 (12.5) 11 (433)

13
62

2.36 (60)
2.36 (60)
9.7 (2.95)
3.1 (0.95)
806,900 (366,000) 248,500 (112,700)
107.6 (42)
107.6 (42)
161 (71.7)
175.3 (79.6)
0.48 (12.2)
0.53 (12.2)
123.7 (604)
187 (913)
3.5 (1.06)
9.8 (0.69)
58.0 (283)
92.7 (51.5)
2503 (232.6)
36.8

3.74 (1.14)
4.6 (0.32)
71.7 (350.3)

2.36 (60)
2.8 (0.85)
198,200 (89,900)
107.6 (42)
243.7 (117.6)
0.53 (13.4)
403 (1965)
7.94 (2.42)
9.4 (0.66)
88.5 (432)

86.2 (47.9)
134.1 (74.5)
782 (72.7)
678 (63)
19.6
23.7
2 12 (3.657)
2 18.55 (13.84) = 37.1 (27.68)

132

It may be noted that while the tubeside velocities for ACHE #1 and for ACHE #2 are
acceptable, that for ACHE #3 [7.94 ft/s (2.42 m/s)] is so high that it will reduce tubeside
fouling to a minimum. It is not often that a designer is able to achieve such a high velocity.
It may also be noted that the air outlet temperature from ACHE #3 is rather high, at
243.7F (117.6 C). This was a direct consequence of the high MTD that required a
relatively small heat transfer area and therefore a rather small face area, through which only
a relatively low flow rate of air can be passed.

13.2 Recirculation Air-Cooled Heat Exchangers


There are occasions when, for an air-cooled heat exchanger, a minimum tube wall temperature has to be maintained. An example is a sour water stripper overhead condenser
where, if the tube wall temperature falls below 158F (70C), solidification fouling takes
place due to the deposition of ammonium salts. Since the ambient temperature varies
from day to night and through the seasons, a special arrangement has to be employed in
order to maintain a constant air temperature (the design ambient temperature) at all times
to the fans delivering air to the condenser tube bundles (forced draft). A recirculation aircooled heat exchanger (Fig. 13.1) is employed for such a situation, wherein automatic
louvers at the top and side(s) of the housing (containing the entire paraphernalia of tube
bundles, steam coils, plenums, fans, motors, etc.) control the extent of recirculation. As
shown in Fig. 13.1, a part of the air leaving from the top of the tube bundles is recirculated and mixes with fresh air entering from the side(s), so that the combined temperature
is precisely the design ambient temperature. Evidently, the lower the ambient temperature, the greater will be the extent of recirculation. Automatically controlled louvers
(sensing the air temperature just below the fans) at both the top and the side(s) ensure the
desired air temperature at all times. This arrangement is also employed for cooling heavy
stocks having a high pour point.
One difficulty will be start-up during winter or even summer evenings, when the
ambient temperature will be below the design ambient temperature. In order to overcome
this problem, steam coils are located below the tube bundles, so that the cold ambient air can

Fig. 13.1 Recirculation air-cooled heat exchanger


133

be warmed up to the design ambient temperature by passing an LP steam through the steam
coils, and having total air recirculation until the air temperature builds up to the design
ambient temperature. The steam supply can then be stopped, since the air-cooled heat
exchanger is then able to take care of itself.

13.3 Humidified Air-Cooled Heat Exchangers


Normal air-cooled heat exchangers can only cool a process fluid to a temperature level
that is higher than the dry-bulb temperature of the ambient air. In hot and arid areas, the
utility of air cooling is therefore severely limited. However, by humidifying the hot and
dry air with water, the temperature can be brought down considerably, thereby enhancing
the capability of air cooling considerably.
Humidified air-cooled heat exchangers (Fig. 13.2) are necessarily of the induced draft
type. Water is sprayed into the air stream before it comes in contact with the heat exchanger
surface, thereby lowering the air temperature due to evaporation. Mist eliminators are
employed to prevent droplets of water from entering the tube bundle. Since only a small part
of the water that is sprayed into the air will evaporate, the balance will have to be collected
in a basin beneath the cooler and recirculated.
The installed cost of humidified air-cooled heat exchangers is evidently rather high.
Another limitation of humidified air cooling is that soft water will have to be used or else
there will be scaling of the finned tubes, with consequent deterioration in cooler
performance. The cost of makeup water will represent an additional expense. However,
since they enhance the cooling capability considerably, a proper economic study will have to
be carried out to establish their viability. It should be kept in mind that in arid areas water is
extremely scarce, so that this mode of cooling may necessarily have to be carried out for
services requiring cooling of streams to temperatures below summer day temperatures.
One important factor to be considered here is that humidification of the coolant air will
be required only during summer daytime hours, which may represent only 45% of the total
operating time of the equipment. Evidently, the selection of the design ambient temperature
will have to be done very carefully.

Fig. 13.2 Humidified air-cooled heat exchanger (Courtesy Hudson Products Corporation, USA)
134

Hudson Products Corp. in Houston, Texas, USA, design, manufacture, and supply
patented humidified air-cooled heat exchangers by the trade name of Combin-Aire aircooled heat exchangers.

13.4 Use of Tube Inserts


Tube inserts considerably enhance the tubeside heat transfer coefficient under laminar
flow conditions. The first-generation tube inserts were twisted tapes that imparted a
swirling motion to the tubeside fluid, thereby augmenting heat transfer. The secondgeneration tube inserts are wire-fin tube inserts (Fig. 13.3). By virtue of the increased
tubeside and thereby the overall heat transfer coefficient, wire-fin tube inserts are very
useful in air-cooled heat exchangers cooling viscous liquids, especially in offshore platforms where floor space is at a premium.
Wire-fin tube inserts are manufactured from formed wire loops, spaced radially and
axially within the tube, and supported from a central core. The inserts continually remove
low-velocity or stagnant fluid from the tube wall, and replenish it with fluid from the center
of the tube. By doing this, wire-fin tube inserts minimize the effect of frictional drag, thereby
preventing the formation of a stable boundary layer, and resulting in dramatically higher
rates of heat transfer inside tubes for a given pressure drop. This increase can vary from 2 to
even 25 times, depending on the Reynolds number in the bare tube situationthe lower the
Reynolds number, the greater the enhancement. In addition, shorter residence time, and
reduced temperature difference (between the tube metal and the fluid film) due to higher
tubeside heat transfer coefficient, can eliminate thermally dependent causes of fouling.
Now, for the same length of travel and velocity, these inserts will evidently result in an
increased tubeside pressure drop, which may be unacceptable. However, the number of tube
passes can be decreased appropriately, such that the heat transfer coefficient is appreciably
higher than that obtained with bare tubes, while the pressure drop is still within the allowable
limit. Thus, under laminar flow conditions, the efficiency of conversion of pressure drop to
heat transfer is higher with these inserts.
When wire-fin tube inserts were first applied, the principal benefit was expected to be
improved heat transfer performance. However, it was found that the extent of fouling was
reduced dramatically. This is not surprising, considering that the boundary layer separation,
and the absence of effective mixing (between the fluid at the tube wall and that at the center
of the tube), are the principal culprits responsible for both inefficient heat transfer and
aggravated fouling. The improved mixing results in (1) a shorter residence time of the fluid
within the tube and (2) a lower temperature difference between the tube metal and fluid film,

Fig. 13.3 Wire-matrix type insert (hiTRAN System) ( Cal Gavin Ltd., reprinted with permission)
135

Table 13.2a: Principal process parameters for Case Study 13.2


ACHE # 1
1. Flow rate, lb/h (kg/h)
4564 (2070)
2. Inlet temperature, F (C)
165.2 (74)
3. Outlet temperature, F (C)
129.2 (54)
4. Heat duty, MM Btu/h (MM kcal/h)
0.075 (0.019)
5. Operating pressure, psig (kg/cm2 abs)
88 (6.2)
6. Allowable pr. drop, psi (kg/cm2)
10 (0.7)
7. Fouling resistance, h ft2 F/Btu (h m2 0.00195 (0.0004)
C/kcal)
8. Inlet/outlet viscosity, cp
8.3/14.0
9. Inlet/outlet density, lb/ft3 (kg/m3)
52.3 (838)/
53.0 (850)

ACHE # 2
16,400 (7440)
150.8 (66)
129.2 (54)
0.179 (0.045)
654 (46.0)
14.2 (1.0)
0.00195 (0.0004)

ACHE # 3
78,300 (35,500)
167 (75)
129.2 (54)
1.484 (0.374)
99.5 (7.0)
14.2 (1.0)
0.00195 (0.0004)

10/21
52.4 (840)/
53.0 (850)

9.2/21
52.2 (836)/
53.0 (850)

thereby reducing the fouling propensity appreciably, especially for thermally dependent
applications.
Advantages of wire-fin tube inserts
Summarizing the above, wire-fin tube inserts offer several advantages, as follows:
1) They can be used to reduce the fixed cost of an air-cooled heat exchanger, or to
save pumping power on the tubeside, or both.
2) A very potent benefit is in revamps, where instead of supplementing or replacing
existing air-cooled heat exchangers, they are modified to have a lower number of
tube passes, and fitted with appropriate inserts, thereby eliminating the cost of
structural and piping modifications and associated downtime, which can be very
significant.
3) Another very significant advantage of using tube inserts is the reduction, or even
elimination, of thermally dependent fouling by virtue of radial mixing, shorter
residence time, and reduced temperature difference across the tube. In fact, where
excessive fouling plagues the performance of an air-cooled heat exchanger, mitigation of fouling may be the prime advantage sought. However, since dirty liquid
streams are generally also viscous, the benefits of enhanced heat transfer and reduced fouling in such situations usually go hand in hand.
A notable description of the mitigation of tubeside fouling by the use of wire-fin tube
inserts is [1]. Other distinguished papers on enhancement of heat transfer are [24].

CASE STUDY 13.2: USE OF WIRE-FIN TUBE INSERTS


For an offshore platform, three oil coolers were to be designed and incorporated in a plot
area of 7.2 ft. (2.2 m) 23.0 ft (7.0 m). The salient process parameters are detailed in
Table 13.2a.
The design ambient temperature was 104F (40C). Tubes of carbon steel 1 in. (25.4
mm) OD 12 BWG (2.77 mm) (minimum under groove) with aluminum fins were to be
used. Due to the rather high viscosity, laminar flow in all three coolers resulted in extremely
low tubeside heat transfer coefficients. Consequently, the required plot area was
considerably higher than that permitted. Furthermore, the tubeside pressure drops could not
be contained within the allowable limits.
136

Table 13.2b: Principal construction and performance parameters for Case Study 13.2
ACHE # 1
ACHE # 2
ACHE # 3
1. Number of tubes per row
2. Number of rows
3. Number of tube passes
4. Approximate bundle width, ft (m)
5. Air flow rate, lb/h (kg/h)
6. Air inlet temperature, F (C)
7. Air outlet temperature, F (C)
8. Tubeside pr. drop, psi (kg/cm2)
9. MTD, F (C)
10. Overall heat transfer coefficient,
Btu/h ft2 F/ (kcal/h m2C/kcal)
11. Heat transfer area (bare), ft2 (m2)
12. No. of fans fan dia., ft (m)

2
7
2
0.66 (0.2)
28,700 (13,000)
104 (40)
115 (46.1)
10 (0.7)
35.8 (19.9)
27.2 (133)

4
7
1
1.1 (0.34)
57,300 (26,000)
104 (40)
116.8 (47.1)
13.0 (0.91)
27.0 (15.0)
42.4 (207)

23
7
1
4.8 (1.47)
329,600 (149,500)
104 (40)
122.7 (50.4)
14.2 (1.0)
31.3 (17.4)
53.5 (261)

78 (7.2)

154 (14.3)
3 5 (1.524)

886 (82.3)

Therefore, revised designs were attempted incorporating wire-fin tube inserts


(HITRAN) of Cal Gavin, UK, with a tube length of 21.3 ft. (6.5 m) and 11 fins/in. (433
fins/m) of 1/2 in. (12.7 mm) height. It was found that a combined section, using one tube
bundle for each service, could be easily accommodated within the specified plot area. The
salient features of the thermal design of the three coolers are detailed in Table 13.2b.
The maximum overall plot width was 6.6 ft (2.01 m) and the maximum tube length was
21.3 ft (6.5 m). Three fans of 5 ft (1.524 m) diameter were employed to drive the total air
flow of 415,600 lb/h (188,500 kg/h) across the three tube bundles.
It should be highlighted here that the overall heat transfer coefficients of 27.2, 42.4, and
53.5 Btu/h ft2 F (133, 207, and 261 kcal/h m2 C, respectively) obtained with the wire-fin
tube inserts are higher by an order of magnitude, as compared to what could have been
achieved with bare tubes.

13.5 Use of Variable Finning Density


For cooling liquids when there is considerable variation in the viscosity from inlet to outlet, the use of variable finning density is a useful tool. With the increase of viscosity as
the fluid flows from top to bottom, the tubeside heat transfer coefficient decreases. It was
mentioned in Section 7.4 that a higher finning density is favorable when the airside heat
transfer coefficient is controlling, and a lower finning density is favorable when the tubeside heat transfer coefficient is controlling. Thus, as the tubeside heat transfer coefficient
decreases, the tubeside resistance becomes more and more controlling and a lower finning density becomes favorable. Consequently, it often becomes advantageous, while
cooling a viscous liquid, to have a higher finning density in the upper rows and a lower
Table 13.3: Results of Case Study 13.3
Original design
Alternate run 1
7 fins/in.
7 fins/in. (276 fins/m) in
(276 fins/m) in top 6 rows, 5 fins/in. (197
all 8 rows
fins/m) in lower 2 rows
Power consumption
per fan, HP (kW)
Overdesign, %

Alternate run 2
7 fins/in. (276 fins/m) in
top 6 rows, 5 fins/in. (197
fins/m) in lower 2 rows

25.87 (19.3)

24.66 (18.4)

24.0 (17.88)

5.7

5.1

4.45

137

Table 13.4a: Principal process parameters for Case Study 13.4


1. Stream
2. Flow rate, lb/h (kg/h)
3. Inlet temperature, F (C)
4. Outlet temperature, F (C)
5. Heat duty, MM Btu/h (MM kcal/h)
6. Operating pressure, psig (kg/cm2 abs)
7. Allowable pr. drop, psi (kg/cm2)
8. Fouling resistance, h ft2 F/Btu (h m2 C/kcal)
9. Fouling layer thickness, in. (mm)
10. Inlet/outlet viscosity, cp
11. Inlet/outlet density, lb/ft3 (kg/m3)
12. Design ambient temperature, F (C)

Vacuum residue
37,500 (17,000)
514.4 (268)
428 (220)
1.8 (0.454)
113.8 (8.0)
24 (1.7)
0.0098 (0.002)
0.04 (1.0)
16/20.9
872/881
107.6 (42)

finning density in the lower rows. The overdesign may reduce insignificantly, whereas
there is an appreciable saving in power consumption. Evidently, the greater the variation
in viscosity of the tubeside liquid, the greater is the scope for employing alternative finning.

CASE STUDY 13.3: USE OF VARIABLE FINNING DENSITY


Let us refer to Case Study 7.3 in Chapter 7, where there were two sections, each having
two tube bundles. Each bundle had 50 tubes per row, 8 rows and 10 passes. The number
of fins per meter was 276.
In order to study the effect of alternative finning, alternate runs 1 and 2 were taken with
varying finning densities. In alternate run 1, the number of fins per meter was retained as
seven per inch (276 per meter) in the top six rows, but was changed to five per inch (197 per
meter) in the lower two tube rows. In alternate run 2, the fin density was retained as seven
per inch (276 per meter) in the top four rows, but was changed to five per inch (197 per
meter) in the lower four tube rows. The results are shown in Table 13.3.
It will be seen that in alternate run 1, the power consumption per fan reduced from 19.3
kW to 18.4 kW whereas the overdesign reduced from 5.7 % to 5.1 %. In alternate run 2,
these values came down to 17.88 kW and 4.45% respectively. Thus, it will be seen that for a
negligible reduction in overdesign, the power consumption, which is a recurring expense,
reduced significantly.
By the judicious application of alternative finning in services where the tubeside heat
transfer coefficient varies significantly from the first tube row to the last, it is possible to
save energy by varying the fin density across an air-cooled heat exchanger.

13.6 Use of Natural Draft


In case a very small heat duty is to be performed and the tubeside heat transfer coefficient
is rather low, no fans need be used, so that power consumption is eliminated. This is illustrated by the following case study.

CASE STUDY 13.4: USE OF NATURAL DRAFT


Vacuum residue in an oil refinery was to be cooled in an air-cooled heat exchanger. The
principal process parameters are detailed in Table13.4a. Since this is a very dirty service,
a tubeside fouling layer thickness of 0.04 in. (1.0 mm) has been considered.
It will be noticed that the inlet and outlet viscosity values are very high, thereby making
138

Table 13.4b: Principal construction and performance parameters for Case Study 13.4
Run 1
Run 2
Run 3
1. Chimney height, ft (m)
2. No. of bays no. of bundles per bay
3. Fin height tube pitch, in. (mm)
4. Fin density, fins/in. (fins/m)
5. Number of tubes per row
6. Number of rows no. of tube passes
7. Approximate bundle width, ft (m)
8. Air flow rate, lb/h (kg/h)
9. Airside
outlet temperature, F (C)
pressure drop, in. (mm) WC
heat transfer coefficient,
Btu/hft2F/ (kcal/h
m2C/kcal)
10. Tubeside
inlet temperature, F (C)
outlet temperature, F (C)
velocity, ft/s (m/s)
pressure drop, psi (kg/cm2)
heat transfer coefficient,
Btu/h ft2 F/ (kcal/h m2
C/kcal)
11. Thermal
Airside
resistance, %
Tubeside
12. Heat duty, MMBtu/h (MM kcal/h)
13. MTD, F (C)
14. Overall heat transfer coefficient, Btu/hft2F/
(kcal/h m2C/kcal)
15. Heat transfer area (bare), ft2 (m2)
16. Overdesign, %

9.9 (3.0)

8.9 (2.7)
70,100 (31,800)
214.2 (101.2)
0.026 (0.67)
11.1 (54.3)

6.6 (2.0)
11
0.49 (12.4) 2.36 (60)
11 (433)
44
48
8.9 (2.7)
49,800 (22,570)
204.1 (95.6)
0.018 (0.45)
5.0 (24.3)

3.3 (1.0)

8.9 (2.7)
31,750 (14,400)
196.7 (91.5)
0.011 (0.28)
2.44 (11.9)

514.4 (268)
428 (220)
3.15 (0.96)
23.2 (1.63)
15.8 (77.2)

514.4 (268)
459.9 (237.7)
3.18 (0.97)
22.9 (1.61)
15.8 (77.2)

514.4 (268)
482.5 (250.3)
3.18 (0.97)
22.5 (1.58)
15.7 (76.7)

53.73
37.86
1.8 (0.454)
345 (173.9)
6.0 (29.2)

72.18
22.76
1.16 (0.292)
364.5 (184.7)
3.6 (17.6)

84.07
130.5
0.68 (0.171)
346.1 (192.3)
2.05 (10.0)

944 (87.7)
Nil

this service appear to be, at first glance, unsuitable for cooling by air. However, for reasons
of convenience, an air-cooled heat exchanger was specified and 0.984 in. (25 mm) OD
0.0984 in. (2.5 mm) thick 19.68 ft (6.0 m) long carbon steel tubes were to be used.
A natural draft air-cooled heat exchanger was designed, the principal construction and
performance parameters of which are detailed in the first column of Table 13.4b. It will be
seen that there is only a single bay having a single tube bundle with 40 tubes per row, four
tube rows, and eight tube passes. In order to provide a suitable natural draft, a 9.84 ft (3 m)
high stack was incorporated.
Note the extremely low airside pressure drop [0.026 in. (0.67 mm) WC] and heat
transfer coefficient [11.1 Btu/h ft2 F (54.3 kcal/h m2 C/kcal)], typical of air-cooled heat
exchangers operating under natural draft. Despite a tubeside velocity of 3.15 ft/s (0.96 m/s),
the tubeside heat transfer coefficient is also very low, since the liquid viscosity is very high.
The airside thermal resistance is 53.73% and the tubeside thermal resistance 37.86% of the
total.
As would be expected for natural draft, the air flow rate is very low and thereby the
airside outlet temperature is very high at 214.2F (101.2C). Nevertheless, because of the
high process temperatures, the MTD is still unusually high.
In order to demonstrate the significance of the chimney height, two additional runs were
taken for the same air-cooled heat exchanger construction, one with a chimney height of 6.6
ft (2 m) and the other with a chimney height of 3.3 ft (1 m). The results are shown in the
second and third columns of Table 13.4b. It will be seen that with a reduction in the chimney
height, there is a sharp fall in the air flow rate and thereby its heat transfer coefficient.
139

Consequently, the heat duty that can be handled reduces profoundly, from 1.8 M Btu/h
(0.454 M kcal/h) to 1.16 M Btu/h (0.292 M kcal/h), and finally to just 0.68 M Btu/h (0.171
M kcal/h).
Louvers were provided to cut off even the small air flow when the unit was not to be in
use, so as to prevent congealing of the vacuum residue under cold weather conditions.
It should be noted that since the tubeside heat transfer coefficient is rather low, the use
of fans would have resulted in a negligible increase in the overall heat transfer coefficient.
Therefore, the first cost would be significantly higher due to the fans, drives, plenums, etc.,
and the operating cost would also be higher due to the fan power. Thus, the use of natural
draft is optimal under such circumstances.

13.7 Air-Cooled Vacuum Steam Condensers


Steam turbines are very widely used in the chemical process industries for driving not
only electricity generators but also various types of pumps, fans, and compressors, as
well as other equipment. Steam condensers are required to condense the exhaust of these
turbines and return the same to the boiler. Such condensers can either be water cooled or
air cooled. The advantages and disadvantages of the two have been elaborated on in Sections 2.1 and 2.2. A typical A-frame air-cooled vacuum steam condenser is shown diagrammatically in Fig. 13.4.

Fig. 13.4 A-frame air-cooled vacuum steam condenser exchanger (redrawn with permission from
HTRI)
140

The main problem with the air-cooled vacuum steam condenser is not the condensation
of the steam but the evacuation of the noncondensables. Failure to eliminate the
noncondensables can cause the following:
1) freezing of condensate in winter
2) loss of performance due to blanketing of the heat transfer surface
3) absorption of noncondensables by the condensate and subsequent corrosion of
the tube metal
Thus, a successful air-cooled vacuum steam condenser must continuously and totally collect and eliminate all noncondensables from the system. The noncondensables are the
gases that enter the vacuum section of the power cycle from the atmosphere, as well as
from the chemicals used for the treatment of boiler feed water.
The trapping of noncondensables inside the condenser tubes is a direct consequence of
the variation of coolant air temperature across the tube bundle. Consider a single-pass
condenser having two or more tube rows. The tubes of the lowermost row are exposed to the
coldest air, while the tubes of the upper rows are exposed to progressively hotter air.
Therefore, the tubes in the lowermost row condense more steam (due to the higher MTD)
while those in the upper rows condense less and less steam. Consequently, the pressure drop
will be the highest in the tubes of the lowermost row, and progressively lower in the tubes of
the upper rows. This will cause a backflow of noncondensables from the tubes of the upper
rows to the tubes of the lowermost row. Figure 13.5 shows the simplest situation where there
are only two tube rows.
This backflow of noncondensables can eventually lead to gas blanketing of a substantial
fraction of the heat transfer surface. This problem will be less acute but not absent in a fourrow two-pass construction. In order to address this situation, a correction factor has been
proposed by Rozenman et al. [5], wherein extra heat transfer area has to be incorporated as
per the penalty factor evaluated.
Special patented designs have been developed by some air-cooled heat exchanger
vendors to address this situation. For a detailed presentation of the problem caused by
incomplete evacuation of noncondensables, the reader is referred to [5,6].

Fig. 13.5 Back-flow of noncondensables from the tubes of the upper row to the tubes of the lower row
in a 1-pass 2-row construction
141

References
[1] Gough, M.J., and Rogers, J.V., 1987, Reducing Fouling by Enhanced Heat Transfer Using
Wire-Matrix Radial Mixing Elements, 24th National Heat Transfer Conf., Pittsburgh, August, AIChE Symp. Series No. 83, pp. 16-21
[2] Mascone, C.F., 1986, CPI Strives to Improve Heat Transfer in Tubes, editorial survey,
Chem. Eng., Feb 3, pp. 2225.
[3] Bergles, A.E., 1978, Enhancement of Heat Transfer, 6th Int. Heat Transfer Conf., Aug. 7
11, Toronto, Paper No. KS-9.
[4] Marner, W.J., and Bergles, A.E., 1978, Augmentation of Tubeside Laminar Flow Heat
Transfer by Means of Twisted-Tape Inserts, Static Mixer Inserts and Internally-Finned
Tubes, 6th Int. Heat Transfer Conf., Aug. 711, Toronto, Paper No. FC(a)-17.
[5] Rozenman, J., Pundyk, J., and Fenoglio, F., 1973, The Effect of Unequal Heat Loads on the
Performance of Air-cooled Condensers, 14th National. Heat Transfer Conference, Atlanta,
Aug. 58.
[6] Larinoff, M.W., Moles, W.E., and Reichhelm, R., 1978, Design and Specification of Aircooled Steam Condensers, Chem. Eng., May 22.

Further reading
1.

Rubin, F.L., 1980, Winterizing Air-cooled Heat Exchangers, Hydrocarbon Process., Oct., pp.
147149.

142

INDEX

condensing profiles, 2
condensing range, vii, 55, 63, 65, 67, 68
condensing service
low pressure, 56
configuration
A-frame, 25
horizontal, 25
congealing, 119, 123, 126, 140
control
switching fans on/off, 4
use of autovariable fans, 4, 121, 126
use of louvers, 4, 121
use of two-speed motors, 4
use of variable-speed drives, 4
conversion of pressure drop to heat transfer,
40, 78, 85, 135
corrosion allowance, 34
cost
first, 1, 6, 7, 30, 31, 80, 112, 128, 140
fixed, 30, 32, 45, 75, 105, 136
initial, 6, 30, 43, 45
maintenance, 30
operating, 1, 3032, 39, 45, 50, 75, 105,
110, 112, 113, 140
overall, 1, 9, 30, 79, 110, 112, 129
total operating, 45
density, 2, 7, 15, 29, 30, 32, 37, 40, 43, 56,
65, 67, 81, 82, 86, 89, 90, 92, 117, 137
design air temperature, 4, 48, 50, 128
desuperheating, 2, 55, 61, 68, 69, 70, 73, 92
dry wall, 69, 70
penalty associated with, 69, 70
wet wall, 69
distribution of tubes in various passes, 86
draft
forced, 23, 24, 26, 27, 51, 115, 121, 126,
127, 133
induced, 23, 24, 26, 51, 115, 126, 127,
134
natural, 25, 27, 139, 140

air compressor intercooler, 56


air distribution, 22, 27
air flow
inadequate, 128
air mass velocity
optimum, 83
air-cooled steam condensers, 30, 104
airside heat transfer coefficient controlling,
30, 48, 78, 8183, 137
alternative finning, 138
ambient temperature, 810, 21, 42, 44, 48,
49, 78, 117, 119, 121, 122, 126, 128, 129,
131, 133, 134, 136
API 661, 17, 21, 23, 24, 33, 41, 75, 76, 86,
109, 128
approach temperature, 6
bank, 13, 46
bay, 13, 22, 29, 61, 63, 65, 87, 120, 121,
131, 139
boundary layer separation, 7, 46, 114, 135
channel, 72
chimney height, 139
combination of air and water cooling, 8
combined services, 2, 121, 131
condensate drainage, 25
condensate film, 57, 58, 60
condensation
dropwise, 56
filmwise, 56
inside vertical tubes, 57
pure-component isothermal, 56
transition region, 57, 63, 65
condensation inside vertical tubes, 57
condensation of a mixture of vapors, 60
condenser
isothermal, 56, 61, 63
narrow range, 63
wide range, 65, 68, 84
condensing heat transfer coefficient, 56, 57,
65, 74
143

economical design, 43, 44, 76, 81


fan
autovariable, 21, 108, 121, 122
blade, 28
blade angle, 20, 21, 121, 128, 129
blade width, 21
dispersion angle, 22
hub, 21
low noise, 7, 21, 24, 51
manually adjustable, 22
power consumption, 26, 34, 39, 65, 68,
70, 71, 75, 84, 117
ring, 22, 23, 128
shaft, 20, 23
tip, 22, 51, 128
tip clearance, 22
tip speed, 51
vendor, 21
fan blades
FRP, 20
fan drive
direct, 23
gear, 23, 24
V-belt, 23
fin density, 15, 30, 43, 79, 81, 82, 85, 87,
115, 116, 138
fin height, 2, 15, 30, 39, 46, 61, 63, 65, 75,
78, 81
fin spacing, 2, 46, 75, 81
fin thickness, 46, 116
finned tube
bimetallic, 17
flooding, 72
flow
annular, 58, 59
cross, 47
laminar, 7, 44, 46, 114, 135, 136
stratified, 59
turbulent, 40
flow maldistribution, 123, 124
flow regimes
annular, 58, 59
slug, 59
flow velocity, 106, 119
fouling
adverse effects of, 103
airside, 15, 40, 83, 104, 116, 117
categories of, 2, 104
corrosion, 104
excessive, 126, 136
particulate, 104
sedimentation, 104
solidification, 119, 133
tubeside, viii, 40, 44, 83, 104, 109, 113,

115, 126, 133, 136, 138


fouling layer thickness, viii, 111113, 114,
116118
fouling resistance, viii, 2, 4, 9, 18, 30, 32,
98, 104, 105, 107109, 111113
unduly large, 108
headers
cover-plate type, 18
manifold type, 18
plug type, 18
heat release profile, 2, 31, 34, 72, 73, 89, 93,
95
heat release profiles
tabular, 95
heat transfer coefficient
airside, 4, 29, 30, 39, 43, 46, 48, 50, 65,
78, 8185, 104, 112, 113, 117, 137
overall, 4, 5, 6, 29, 34, 40, 43, 45, 48, 50,
57, 60, 61, 69, 70, 77, 78, 82, 84, 85,
98100, 103, 107, 108, 111, 114, 135,
137, 140
tubeside, 7, 30, 35, 3741, 43, 44, 45, 46,
50, 61, 63, 65, 67, 68, 70, 7679, 81,
83, 84, 88, 111, 112, 114, 117, 132,
135140
heat transfer coefficients
typical overall, 37
high pour point, 7, 123, 126, 133
high velocity, 106, 108, 109, 133
hot air recirculation, 27, 119, 125128
humidified air-cooled heat exchanger, 134,
135
hydrogen, 92, 93, 116, 117
hydrogen-hydrocarbon mixtures, 93
interface temperature, 106
laminar flow, 7, 44, 46, 114, 135, 136
line size, 33
materials of construction, 30, 33, 108
mean temperature difference, 29, 31, 34, 48,
126, 128, 129
mechanical design, 1, 2, 29
MTD, 46, 27, 29, 34, 4648, 50, 61, 63,
6870, 72, 74, 83, 89, 93, 98100, 104,
117, 126, 128, 129, 133, 139, 141
multiple operating cases, 34
natural convection, 2, 3, 25, 27
negative latent heat, 94
noise, 7, 20, 21, 23, 50, 129, 130
noise level, 7, 23, 129, 130
nozzle sizing, 2, 33
nozzles, 17, 33, 61, 72, 124
condensate, 72
Nusselt number, 46
operating pressure, 32, 37, 42, 55, 56, 61,
144

65, 6769, 90, 117, 119, 125


operating problems, 2, 123
overdesign, 2, 4, 41, 70, 71, 78, 82, 88, 97
102, 108, 114, 117, 123, 132, 138
on performance, 98100
on surface, 98, 100
reasons for providing, 97
overdesign factor, 97, 100, 101
pass partition plates, 17, 35
physical properties, 2, 32, 34, 36, 40, 57, 60,
8991, 97, 99, 107
physical property profiles, 2
pipe rack, 7, 33, 42, 44, 75, 131
pipe-rack width, 33, 49, 75, 76
plenum chamber, 23, 24, 27, 75, 126
Prandtl number, 35, 36, 46, 58
pressure drop
allowable, 4, 9, 3032, 35, 41, 42, 50, 56,
72, 108, 110, 111
allowable tubeside, 4145, 50, 63, 88
in nozzles, 72
total, 42, 72
tubeside, 15, 32, 34, 3945, 50, 63, 65,
67, 77, 80, 81, 88, 98, 102, 109, 110,
112114, 132, 135, 136
utilization of, 77
pressure drop limiting design, 41, 102
process licensor, 26, 3134, 42, 43, 45, 90,
91, 101, 102
radial mixing, 46, 114, 136
recirculation air-cooled heat exchanger, 26,
126, 133
residence time, 135, 136
run length, 105, 108
sound power level, 50, 51
sound pressure level, 51
specific heat, 6, 7, 31, 32, 36, 37, 72, 89, 90,
92, 93, 116, 124
subcooling, 2, 61, 68, 71, 73, 100, 119
technological platform, 33
TEMA standards, 32, 47, 90, 108, 109
thermal conductivity, 1, 6, 14, 15, 32, 33,
36, 37, 89, 90, 92, 93, 112, 113, 116
thermal design, 1, 2, 13, 2932, 37, 39, 43,

44, 48, 49, 55, 61, 63, 65, 67, 69, 73, 75,
9195, 98, 100, 101, 107, 111113, 117,
137
optimum, 29, 48
trim cooler, 810, 41, 102, 103, 128, 129
tube bundle, 1, 7, 13, 14, 19, 2227, 35, 51,
7577, 83, 8587, 113, 115, 116, 120,
121, 123, 124, 126, 128, 131, 133, 134,
137139, 141
tube diameter, 35, 36, 41, 56, 76, 109, 110
tube inserts, 2, 114, 135, 136
wire fin, 7, 46, 114, 115, 135137
tube length, 17, 21, 22, 29, 33, 39, 41, 42,
44, 49, 50, 56, 69, 7476, 102, 137
tube passes
number of, 2, 19, 29, 35, 39, 41, 47, 50,
56, 63, 65, 7577, 80, 81, 83, 87, 108,
110, 113, 132, 135, 136
tube pitch, 2, 29, 75, 78, 80, 85, 86
longitudinal, 47
transverse, 47
tube plugging, 102
tube rows
number of, 2, 18, 29, 35, 75, 77, 80, 82,
83, 116
tube size, 33, 36, 39, 41, 49, 79
tube supports, 13, 14
tubes
bare, 30, 135, 137
finned, 1517, 19, 26, 27, 30, 46, 115,
121, 131, 134
tubesheet, 18, 19
tubeside heat transfer coefficient controlling,
78, 81, 137
tubeside velocity, 31, 35, 38, 41, 4345,
108, 110113, 126, 132, 139
tube-to-tubesheet joint, 120
U-tubes, 18, 35
variation of viscosity with temperature, 90
viscosity, 32, 3537, 42, 44, 60, 79, 89, 90,
92, 99, 110, 136139
winterization, 6, 119

145