HEAT TRANSFER

in TUBE BANKS
in CROSSFLOW

Nay AV Coltts ols

Pan OlTaky
<eks

Experimental and Applied Heat Transfer Guide Books

HEAT TRANSFER
in TUBE BANKS
in CROSSFLOW
This latest volume in the Experimental and
Applied Heat Transfer Guide Books series
provides results on local and average heat
transfer from smooth, rough and finned
tubes in crossflow of air and liquids at Rey-
nolds numbers from up to 10° and Prandtl
numbers to 10%.

The authoritative studies account for wide

ranges of relative transverse and longitudi-
nal pitches as well as surface roughness and
finning factors. The authors suggest a new
method of calculation, and introduce EuRe®,
an original dimensionless group.

All the material is correlated according to a

uniform approach, allowing the presentation
of a variety of governing relationships.
Moreover, the authors recommend the most
practical correlation graphs and formulas for
heat exchanger design.

The research data appearing throughout the

book are usually proprietary, and therefore a
welcome source of information for all stud-
ies involving heat transfer and crossflow.

The thoroughness of coverage and the

range of conditions studied make this vol-
ume a major sourcebook for researchers,
students, faculty, design engineers, equip-
ment operators, and all specialists con-
cerned with heat transfer, thermal ener-
getics, and the nuclear and chemical
industries.

(Continued on back flap)

\Mamata
 Digitized by the Internet Archive
In 2022 with funding from
Kahle/Austin Foundation

https://archive.org/details/heattransferintuO000zhuk
HEAT TRANSFER IN TUBE
BANKS IN CROSSFLOW
Experimental and Applied
Heat Transfer Guide Books

A. Zukauskas, Editor

A. Zukauskas and J. Ziugzda, Heat Transfer of a Cylinder in Crossflow

J. Vilemas, B. Cesna, and V. Survila, Heat Transfer in Gas-Cooled Annular
Channels
M. Tamonis, Radiation and Combined Heat Transfer in Channels
A. Zukauskas and A. Slantiauskas, Heat Transfer in Turbulent Fluid Flows
J. Stasiulevicius and A. Skrinska, Heat Transfer of Finned Tube Bundles in
n° rossflow
A. Zukauskas and R. Ulinskas, Heat Transfer in Tube Banks in Crossflow

IN PREPARATION
A. Zukauskas, V. Katinas, and R. Ulinskas, Fluid Dynamics and Flow-
Induced Vibrations of Tube Banks
HEAT TRANSFER IN TUBE
BANKS IN CROSSFLOW

A. Zukauskas
Academy of Sciences of the
Lithuanian SSR, Vilnius

R. Ulinskas
Institute of Physical and Technical
Problems of Energetics
Kaunas, Lithuanian SSR

© HEMISPHERE PUBLISHING CORPORATION

New York Washington Philadelphia London

Distribution outside North America

SPRINGER-VERLAG
Berlin Heidelberg New York ‘Tokyo
HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Copyright © 1988 by Hemisphere Publishing Corporation. All rights reserved. Printed in the
United States of America. Except as permitted under the United States Copyright Act of 1976,
no part of this publication may be reproduced or distributed in any form or by any means, or
stored in a data base or retrieval system, without the prior written permission of the publisher.
Originally published as Teplootdacha poperechno obtekaemykh puchkov trub by Mokslas, Vil-
nius, 1986, in the series Teplofizika, vol. 18.

Translated by Jamil E. Ghojel.

1234567890 BEBE 898

This book was set in Press Roman by Hemisphere Publishing Corporation. The editor was
Victoria Danahy; the production supervisor was Peggy M. Rote; and the typesetter was Cynthia
B. Mynhier.
BookCrafters, Inc. was printer and binder.

Library of Congress Cataloging in Publication Data

Zukauskas, A., 1923-

Heat transfer in tube banks in crossflow.

(Experimental and applied heat transfer guide books)

Translation of: Teplootdacha poperechno obtekaemykh puchkov trub.
Bibliography: p.
Includes index.
1. Heat exchangers. 2. Tubes—Fluid dynamics. 3. Heat—Transmission. I. Ulinskas, R.
(Romanas) II. Title. III. Series.
TJ263.Z46513 1988 621.402'5 87-19750
ISBN 0-89116-685-8 Hemisphere Publishing Corporation

DISTRIBUTION OUTSIDE NORTH AMERICA:

ISBN 3-540-18865-7 Springer-Verlag Berlin
eee

ee eee ee ee ee CONTENTS

Preface Vii
Nomenclature

INTRODUCTION
General
Thermal design principles of recuperators
Heat transfer from a tube in crossflow
Bank of tubes

LOCAL HEAT TRANSFER FROM BANKS

OF SMOOTH TUBES
Specific features of heat transfer
Local heat transfer in the low range of Reynolds number
Local heat transfer in the medium range of Re
Local heat transfer in the high range of Re
Local heat transfer in characteristic regions of tubes in in-line and
staggered banks
Temperature pulsations on tube surfaces

MEAN HEAT TRANSFER FROM BANKS OF SMOOTH

TUBES
SIE Heat transfer from a tube in a bank 56
SEZs Effect of the physical properties of the thermal carrier and the
temperature head on heat transfer 59
vi CONTENTS

3.3: Influence of flow regime and tube arrangement on heat transfer 63

3.4. Effect of number of tube rows on heat transfer 89
355) Effect of flow angle of attack 90
3.6. Calculation of mean heat transfer of tube banks 91

HEAT TRANSFER OF BANKS OF ROUGH TUBES 94

The specific features of heat transfer of rough surfaces 95

Heat transfer of a single rough tube 99
Local heat transfer of banks of rough tubes 101
Mean heat transfer of banks of rough tubes 113

HEAT TRANSFER OF BANKS OF FINNED TUBES 120

Heat conduction in finned tubes 121

Effectiveness of fins 125
Selection of the determining parameters 1133
Local heat transfer along the height and perimeter of a finned tube in a
bank 134
Mean heat transfer of banks of finned tubes 138

EFFECTIVENESS OF HEAT EXCHANGERS 157

Evaluation of heat-exchanger effectiveness 158

Determination of optimum characteristics of heat exchangers 170
Computer-aided design and calculation of heat exchangers 174
Possibility of calculating local and mean heat transfer 176

APPENDIXES 181

REFERENCES 187

INDEX 197
Sener
ee

ES
RE i PREFACE
isi ord ti

Heat transfer, resistance, and characteristics of a flow across banks of smooth,

rough, and finned tubes in a wide range of variation of typical parameters have
long been investigated at the Institute of Physical and Technical Problems of
Energetics of the Academy of Sciences of the Lithuanian SSR.
These questions were discussed in three monographs in the series “‘Ther-
mophysics’’: A. Zukauskas, V. Makarevicius, and A. Slan¢iauskas, ‘Heat
transfer in banks of tubes in crossflow of liquid,’’ J. Stasiulevicius and A.
Skrinska, “Heat transfer in banks of finned tubes in crossflow,’’ and A.
Zukauskas, R. Ulinskas, and V. Katinas, ‘“‘Hydrodynamics and vibrations of
banks of tubes in flow.”’
The authors of the present book accumulated additional material on the
study of local and average heat transfer for in-line and staggered banks of tubes
and the determination of their optimal pitch, roughness parameters, finning,
and effectiveness. A method is suggested for generalizing experimental data to
optimize the design of tube banks and to give the desired heat transfer and
resistance characteristics. The material is enhanced with new data and is pre-
sented to the reader in a convenient form—as graphs and generalized equations
normally required for practical computations.
In addition to the material in monographs and papers published during the
last few years, the authors of the present book made extensive use of data from
later investigations carried out by them with C; Sipavicius and other colleagues.
The authors express their sincere gratitude to J. Ziugzda, L. Burkoi, and to
all those who helped with the preparation of the manuscript.
Vil
ca
RAAT
: ie
277 Ee
4 - , omtee of 7 a
7» ia oe Gap ¢ Sid 7
wis of iy » & -
iG a. 2 7 a - °
ee 7
.
S
ty Ti 6 771A CWA SA TP VP Tle abe
. 7
" - » »lo Ye age aye : yy

; Js oe (SG See
.
v
= : 0

se “16 peng
a) -

= i
} < je ap 7] === (=e qs —
it
_

thames Rls ie ee ek aeeeeee ae dmnies Mali oy

wea taps. lush es en aiaics AG eaires Sbhw «few pepai
ha atioiaawshSecurloytelios | ae 4 stut er t Ten aly .
ter Wide pamettes © m oh
le? — wot ere arineos. ai Th © va
nie ee eau Hey a EGsige E
aA
eoutkeee fF
attneG a Ay toes iri aM VY wrduivcge
@
A
1 ee : >. ae i ace Pranpyy YY ein “h nary
A tie var ltigan ' Ca Vaart Vor alpedb @i wipes walt
YPSP Aipekie bee eeerhasat conte bite AetenitDo® y
~ wnat,1) ip a
=O Sorina CUM an Ligaen Gann, emmy zits mornrs 4
en ae GOR Lew aedateh OR teltineneys teal Sten
Jan ban ebnr rgvg
pov rants Vidgets oate teal aie
a
weitere oat a
Lene oA? UNA) omy wt (elias of | meters A LAT j east
Kee wales ( 1 Loree Somt 21 Tate Obie ortutd ye cagieerty-
=e) Sd nO : |
bee rh ~S (iw beanie e lorena wit Bs Pai on -
airtejeas Rvyeali 1 tony sihinry ttre}. PPOUEIVIF SS i be AY ul ;
ncaa lipaleaet
a0)yTh
W gooie Bentley requ, lu! cnr
hen) ey
i ad
~i) Ort oben ibid| of ed sal
ae
4 fe/ a Cl Ot 7 Sane wet dese a
schay bveee™ | otabutleny Sweat
Porson a Wenasa dod
et ne bean ow -
7 ais 7
 NOMENCLATURE

a= s5s,/ld relative transverse pitch

a thermal diffusivity, m’/sec
b = s,/d relative longitudinal pitch
b' = sj/d relative diagonal pitch of staggered bank
G = 2r,,/pu, coefficient of friction resistance
heat capacity, J/(kg - K)
tube diameter
heat transfer surface, m’
height of fin, m
height of roughness element, m; overall coefficient
of heat transfer, W/(m’ - K)
dimensionless roughness height
length, m
pressure, Pa
pressure drop, Pa
heat rate (heat flowrate), W
heat flux (density of heat flowrate), W/m*
fin pitch, m
transverse and longitudinal pitches between tubes of
a bank, m
diagonal pitch of staggered bank of tubes, m
temperature, °C, K
incoming velocity, m/sec
maximum velocity, m/sec
average velocity, m/sec
volume enclosed by heat transfer area, m’
velocity components, m/sec
fluctuation components of the velocity
dynamic velocity, m/sec
dimensionless velocity
Cartesian coordinates, m
local coefficient of heat transfer, W/(m’ - K)
fin parameter, dimensionless fin height
thickness of hydrodynamic boundary layer, m
displacement thickness, m
momentum thickness, m
x NOMENCLATURE

6; thickness of thermal boundary layer, m

0 temperature, read from the wall, °C
0 + q,/pC,Us dimensionless temperature
Il = F/V compactness of heat exchanger, 1/m
thermal conductivity, W/(m - K)
be coefficient of dynamic viscosity, Pa - sec
v coefficient of kinematic viscosity, m’/sec
p density, kg/m*
7 shear stress, Pa
gy angle, degree
Eu = Ap/pi’ Euler number
Nu = ad/d Nusselt number
Pr = via Prandtl number
St = a/pc,u, Stanton number

SUBSCRIPTS
f, 9 free stream (main flow)
w conditions on the wall
x local conditions
(<2) averaging
(PS fluctuation component

Other notations are given in the text.

 CHAPTER

ONE
INTRODUCTION

1.1. GENERAL

Consumption of fuel energy sources is rapidly increasing; therefore, the effi-

cient and rational use of these sources is considered an economical objective of
extreme importance. In power engineering and in industry, the greater part of
heat energy is transferred using various heat-exchange equipment. Generally,
heat exchangers operate on the principles, “‘gas-gas,”’ ‘‘vapor-liquid,’’ “‘liquid-
gas,’ or “‘liquid-liquid.’”’ Tubular heat exchangers, and other devices with
heating surfaces made of tubes, are widely used. Therefore, there is a need to
augment heat transfer processes and to increase the thermal efficiency of tubu-
lar heat exchangers. In the present book, the problems of heat transfer for tubes
with external flow are discussed.
Intensity of heat transfer depends mainly on the type of thermal carrier. For
example, for similar conditions and equal flow velocities, the heat transfer
coefficient in a stream of water is one to two orders of magnitude higher than in
an air stream, even though air is less aggressive chemically than water.
Flow velocity and stream conditions, which depend mainly on the type and
design of the heat exchanger, have great influence on the intensity of heat
transfer. In practice, the most widely used heat exchangers are shell-and-tube
exchangers with smooth, straight tubes, U-shaped tubes, or helical tubes (Fig.
1.1.). These are typically used at high pressures and temperatures. The ratio of
specific metal content to heat output of shell-and-tube heat exchangers is rela-
tively high.
The problem of intensification (augmentation) of heat transfer in gas-gas
2 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Figure 1.1. Types of shell-and-tube heat exchangers. a) with longitudinal and transverse flow
around tubes; b) with segmented baffles; c) with annular baffles; d) helical.

heat exchangers is of prime importance, since the heat transfer coefficient for
both carriers is low, and the size of the exchanger is large. Consequently, metal
consumption is considerable. The heat transfer coefficient for air is always
much lower than for liquids. Therefore, measures should be taken to increase
heat removal from the air side. In order to achieve this, finned surfaces, allow-
ing an increase of the heat transfer area by 20 times or more, are widely used.
In order to intensify heat transfer by finned surfaces, the fins are cut and
deformed. This increases the heat exchange at the roots where the velocity of
the thermal carrier is particularly low.
Plate and spiral heat exchangers are also used extensively in practice, since
they are easy to manufacture, compact, and have low metal content. The main
disadvantage of plate heat exchangers is their inability to operate at high pres-
sures. The majority of problems connected with heating, cooling, and medium
condensation are solved by using developed heat transfer surfaces. It is possible
to balance the heat transfer coefficient, decrease the weight, size, and cost of
heat exchangers.
In spiral heat exchangers, full counterflow is provided, there is no sharp
change in flow direction, and fouling and hydraulic resistance are lower than in
shell-and-tube heat exchangers. They are widely used for heating or cooling
highly viscous fluids by employing the liquid-liquid scheme.
Modern energy equipment in which thermal energy of burned fuel is con-
verted to electrical energy, and heat-exchange devices found in industrial equip-
ment do not always operate at optimum conditions. In this regard, the investiga-
 INTRODUCTION 3

tions directed towards determining the optimum conditions and parameters for
the operation of heat exchangers are of paramount importance.
In order to increase thermal efficiency, preserve scarce metals, and de-
crease the mass and size of heat-exchange equipment, it is necessary to
consider the possibility of the rational increase in compactness. The ques-
tion of heat transfer augmentation should be solved with the help of the most
rational means, depending on flow conditions and the type of thermal carrier.
The investigations aimed at improving heat-exchanger energy factors are of
great practical interest. These are usually conducted over a wide range of vary-
ing characteristics which determine the heat transfer and permit the derivation
of generalized relations for smooth, rough, and finned surfaces. The results of
these investigations make it possible to compare various heat transfer surfaces
and choose the optimum versions and conditions of operation of heat exchang-
ers.
In designing modern heat exchangers, it is imperative to know not only the
heat transfer in inner rows of the banks, but also the heat transfer in the first
rows, where it is usually lower. When the banks contain few rows, heat transfer
is determined from one row to another.
A significant portion of the heat transfer surface in a shell-and-tube heat
exchanger is subjected to flow at a certain angle. Therefore, when designing
such exchangers the variation of heat transfer as a function of the flow angle of
attack relative to the tube must be taken into account.
During usage, technically smooth tubes become rough due to technological
processes, fouling, and erosion of tube surfaces. Therefore, it is important to
know the dependence of heat transfer on the degree of surface roughness.
Consideration of these and a number of other factors is important to the
design of efficient heat exchangers.
Considerable attention has been paid recently to the problems of heat trans-
fer computations in heat exchangers. This is due to the development of nuclear
power generation and of the chemical industry, further improvement of various
industrial equipment, and the development of energy systems for the transporta-
tion sector.
At the Institute of Physical and Technical Problems of Energetics of the
Academy of Sciences of the Lithuanian SSR (IPTPE AS LitSSR), a study has
long been conducted on the mechanism of heat transfer and resistance in heat
exchangers and in their main components, namely, tube banks in crossflow,
which are widely used in numerous branches of the industry [1-6].
In this book, the results of experimental and theoretical investigations of
heat transfer in banks of smooth, rough, and finned tubes are presented. These
results have been obtained at IPTPE AS LitSSR and other scientific centers
both in the USSR and abroad for the last 25 years. Considerable attention has
been given to the characteristics of local heat transfer along the tube perimeter,
including separation and recirculation regions of liquid flow obtained by flow
4 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

visualization and photography. The problems of resistance in flows around tube

banks have been discussed in detail in later works by the authors [3, 5].
In order to increase the thermal efficiency of heat-exchange apparatus, the
mechanism of heat transfer is investigated and the possibility of heat transfer
augmentation is examined. Proper space is allocated to the analysis of the char-
acteristics of developed heat-exchange surfaces which are implemented in the
form of finned banks of tubes.
The material presented in the book is based on research carried out in flows
of air, water, transformer and aviation oils over the Reynolds number interval
from 1 to 2 x 10° and Prandtl numbers from 0.7 to 10°.

1.2. THERMAL DESIGN PRINCIPLES OF RECUPERATORS

Modern heat exchangers which operate within a wide range of velocities, physi-
cal properties of liquids, temperatures, and pressures must have the following
qualities:

© transmit the predicted amount of heat from one thermal carrier to another and
be stable in operation within the stipulated range of temperatures; the heat
flow must be carried out with the maximum possible heat transfer coeffi-
cients and at the given velocities of the thermal carriers;
¢ incur minimum hydraulic losses during the transfer of the given amount of
heat;
¢ be reliable and convenient in operation;
© be compact, light, strong, etc.

Satisfying all these requirements is extremely difficult. For example, aug-

mentation of heat transfer by increasing flow velocity leads to an increase in
hydraulic resistance. Therefore, the need arises to optimize a multitude of inter-
related factors.
In practice, two types of heat exchangers are widely used: non-
accumulators and accumulators of heat. In the first case, the heat flow is trans-
ferred through a wall from one thermal carrier to another. Such heat exchangers
are sometimes called recuperators. In heat exchangers that accumulate heat, the
same heat transfer surface comes into contact with the thermal carriers alter-
nately. The temperature of the wall and thermal carrier changes in time, causing
the process of heat transmission to be non-stationary. Such heat exchangers are
called regenerators. In the present book, heat transfer in tubular recuperators is
the subject of discussion.
The main problem accompanying thermal calculations is the determination
of heat exchange surfaces. If these are known, then the problem is to determine
the amount of heat transmitted, final temperatures of the thermal carriers, and
operating conditions of the heat exchanger.
 INTRODUCTION 5

Heat exchanger surface is determined by the simultaneous solution of the

heat balance equation

dQ = c,Gdt Cit)

and the equation for overall heat transfer

dQ = k AtdF (122)

where

dQ = CyiGdt, — CpGpadt, == k AtdF (1.3)

Subscript 1 denotes direction towards the hot thermal carrier, and subscript
2 denotes direction towards the cold thermal carrier; At = t, — t, is the
difference of mean temperatures of the hot and cold thermal carriers.
From the balance equation (1.3) in its integrated form, the heat transfer
surface of the element at a constant flow rate of the thermal carriers is deter-
mined by

pase, are (1.4)

ol
or

F=G, [ ets (1.5)

,
to

where C,1, Cp) is the specific heat capacity of the hot and cold thermal carriers;
t}, ti, tj, and t;, are the temperatures of the hot and cold thermal carriers at the
inlet and outlet of the heat exchanger. If the values of c,,, C,., dt,, dt,, k and At
change insignificantly along the length of heat transfer surface, then the heat
balance equation can be expressed as follows:

O =¢,,G, (t) —t{) = Cp, G2 (t2 -—3) =k (4

—ta) F (1.6)

The heat transfer surface is determined from equation (1.6):

P=se.GWti—t pik (y—2b) (1.7a)

or
6 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

F = cp,Gs(t3—t3)/k (4 — ts) (1.7b)

The character of temperature change is peculiar to recuperators with paral-
lel and counterflow directions of the thermal carriers (Fig. 1.2).
The water equivalent, sometimes called the water number, is of particular
importance:

Wi = Cy G,, W,= Cy2 G, (1.8)

Mass flow rate of a thermai carrier G is often replaced by volume flow rate V.
In this case, the water equivalent referred to 1 m° of the thermal carrier:

W = c,Vo (1.9)

Taking the water equivalent into account, equation (1.3) yields

dQ = —W, dt, = —W, dt, (1.10)

or

O =W, (t) — 11) =W2 (t2— 2) (1.11)

from which the flow rate of the thermal carrier or the amount of transmitted
heat can be found. From equation (1.11) an important relation follows

leet Mel (1.12)

Relation (1.12) is valid for the whole heat transfer area and for an individual
element dF:

‘lamina
t,
Wwe

Q a“
t G
t

0 is
a 6
Figure 1.2. Temperature change in parallel-flow (a) and counterflow (b) recuperators.
 INTRODUCTION 7

ae (1.13)

Relations (1.12), (1.13) allow to determine the optimum values of dt, and dt,
depending on the values of the water equivalent of the thermal carriers.
For significant temperature changes of thermal carriers along the length of
a heat transfer surface, the specific heat capacity and the coefficient of heat
transfer are functions of temperature:

CG) (1.14a)
Cy = f(t) (1.14b)

Then equation (1.7) transforms to

iH
F=G,
1
J ou.
Cpy (ti) dy
(1.15a)
aT

or

F=G, { eens
c- , Cpz (ta) dt
(1.15b)
i,
the solution of equation (1.15) is possible if the values c,, (¢;) C,» (t), and k(t, —
t,) are known. They are, however, very complex and are not convenient for
practical applications. Therefore, for separate parts of the tube bank, it is expe-
dient to use numerical integration of equations (1.15) with local values of the
integrated mean time intervals dt,, and dt, from which the thermophysical prop-
erties of the heat carriers are determined. It is assumed, in this case, that the
physical properties of the thermal carriers are constant within the bounds of the
single section of the tube bank.
The flow condition of the thermal carrier (parallel or counterflow) influ-
ences how dt, and dt, vary along the length of the heat transfer surface.
In practice, counterflow is far more common due to the larger heat flux it
gives as compared with parallel flow. Fig. 1.2 shows that the final temperature
of the cold thermal carrier t; for parallel flow is always lower than the final
temperature of the hot thermal carrier ¢{. For counterflow 4 can be higher than
Be
The mean temperature difference (head) is determined by various methods,
and it depends on the conditions of the heat transfer process. If the temperature
variation across the heat transfer surface is insignificant, then the mean temper-
ature head (difference) depends on the character of the thermal-carrier move-
8 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

ment (parallel or counter) and on the temperature at the inlet and outlet of the
tube banks. Consider the temperature change of the thermal carrier across the
heat-exchange surface for parallel flow and infinitesimal element of the ex-
changer surface area. From equation (1.10), we get

d (At) =d(4-t) =dy-dy= - (7+) dQ (1.16)

If

m= 1/W,+1/W,

then (107)

d(t; —t.) = —mdQ

Since, according to equation (1.2), d@ = kATdF for an element of heat transfer

surface area dF, then .

d (At) = — mk AtdF (1.18)

or

d (At) A
ee a —mkdk (1.19)

where

At=t,—te

If the values of m and k are constant, then equation (1.19) can be integrated
from At, to At and from 0 to F:

At iy F
J: t
SS =-mk i!dF (1.20)
At, 0

or

In (At/At,)= —mkF (2h)

from which

At=At, exp (—mkF) (1.22)

 INTRODUCTION 9

where At, = t/ — 1; is the temperature difference (head) at the inlet of the hot
thermal carrier; At is the local temperature head for an element of heat transfer
surface dF From equation (1.21), it is apparent that the temperature difference
across the heat transfer surface changes exponentially.
Similarly, the relation for calculating the local temperature difference is
obtained for counterflow. In this case, the temperature of both thermal carriers
along the heat transfer surface decreases. From Fig. 1.2, it follows that:

dQ = —W,dt,=—-W, dt, (1.23)

or

d (t, —t,.) = —mdQ (1.24)

where

m=1/W,-1/W,

The mean temperature difference on the heat transfer surface is determined

from equation

e.

At= — AtdF (1.25)

a (| 2

in which the value At is introduced from equation (1.21):

rs
Any
At= —+ | exp (—mkF)dF= — inkF
[exp (—mkF)—1] (1.26)
0

Substituting the values of mkF from (1.21) and (1.22) in equation (1.26), and
assuming that at the edge of the heat transfer surface At = At, (where t, = t; —
t5), we get

Ap oars
At= 7 (AG iAGD
1:27
ee)

that is, an expression of logarithmic mean temperature difference which is

applicable to both parallel and counter flows. For parallel flow At, = tj — &
and At, = t; — &, and for counterflow At, = t/ — ¢, and At = ff — by.
If the water equivalents of the thermal carriers are equal, then it follows
from equation (1.22) that At = At, i.e., the temperature difference of the heat
transfer surface is constant.
10 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

If the temperature of both thermal carriers along the length of the heat
transfer surface changes insignificantly, then the mean temperature difference
can be determined as the arithmetic mean value of extreme differences

At= > (At + Ate) (1.28)

The arithmetic mean value of the temperature difference is always higher than
the logarithmic mean value. However, at At,/At, > 0.5 the difference between
them is insignificant. Therefore, such an error is admissable for practical com-
putations.
The determination of the average temperature difference for crossflow of
thermal carriers is more complex. In various cases of cross and mixed flows,
auxiliary nomograms [7] are widely applied. From these nomograms a correc-
tion factor for equation (1.27) is found.
Fig. 1.3 shows the dependence of the correction factor € on the auxiliary
values

P=(th—1)(t{-1) (1.29)
and

W, , ” nv if
R=ap a(t — tlt2 — t2) (1.30)

for the crossflow of liquids in heat exchangers. From expression (1.27), the
mean temperature difference for counterflow is determined. The mean tempera-
ture difference for a complicated flow of thermal carrier in a heat exchanger
will be equal to

/ -——4 uw

= a

O Ot 02 U5 04 05 0h 02,038.092 y
Figure 1.3. Dependence of correction factor € on the auxiliary values P and R.
 INTRODUCTION 11

Ai, = Ai-e (1.31)

where Af is the logarithmic mean temperature difference found from equation

(£27);
The principal indicator for determining the surface area from equation
(1.15) is the overall heat transfer coefficient k. With increasing k, the heat
exchanger becomes more compact and loses less heat during heat transfer from
one thermal carrier to another. Therefore, if there are no other limitations,
higher values of k should be aimed at when designing a new heat exchanger.
In the case of a clean flat metal wall, the heat transfer coefficient is deter-
mined from

k=(L+54+4)" (1.32)

k=( Rae
: S inet en & ee) (1.33)

The values of a, and a, in equations (1.32) and (1.33) are the heat transfer
coefficients from the sides of both thermal carriers: 6 is the thickness of the flat
wall; and d,, d, are the internal and external tube diameters. In practice, for
d,/d, < 1.4, the error in calculating the heat transfer coefficient for a cylindri-
cal tube using equation (1.32) does not exceed 2-3%.
If the wall of the tube is multi-layered, then

In de + dee (1.34)
oa dy 2h; d; Ke d, Sha

from which it follows that the terms 1/a,d, and 1/a,d,,, are the thermal resis-
tance to heat transfer, and the term

ae eae
mn

a 2h; d;
i i

is the sum of thermal resistances due to heat conduction. The inverse of the heat
transfer coefficient, the value R = 1/k is called the overall linear thermal
resistance to heat transfer.
When the heat transfer surface is subjected to fouling in the form of sedi-
ments from the thermal carrier, scale from burned particles of the heated sub-
12 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

stance, and so forth, the thermal resistance following operation of tube banks
for a period of time acquires the form

Re=Re rae. =e) (35)

where Rema, is the maximum thermal resistance and a is a constant, determined

experimentally.
The overall heat transfer coefficient for a fouled surface k; = 1/R,;. The
local value of the thermal resistance during time 7 is determined from equation:

Re =(1/ae)—-(Cl/2) (1.36)

where a, and a are heat transfer coefficients for fouled and clean surfaces,
respectively. The overall coefficient of a fouled surface is determined from
equation:

kp=kid 4 Rh) (1.37)

In the absence of fouling R; = 0 and k; = k.
With increasing wall thickness, decreasing thermal conductivity, and
buildup of deposits the heat transfer coefficient decreases. If the heat transfer
coefficient changes insignificantly across the heat transfer surface, the arithme-
tic mean averaging of its values is performed. If the change is appreciable, then
the heat transfer surface is divided into separate sections for which the value of
k is determined. These values are then averaged for the whole surface:

kK, Fitk, F,+...+k, F,

a Pipi
te aera oo 2°)

where k,, k,, . . . , k, are the mean values of overall heat transfer coefficients
for sections F,, F,, . . . , F, of the heat transfer surface.
The determination of the overall heat transfer coefficient k entails the great-
est difficulty when conducting a thermal design of heat exchange apparatus.
The coefficient values of thermal conductivity of the wall material \ can be
obtained from reference tables, and heat distribution in the wall can be com-
puted from them. The determination of the heat transfer coefficients a, and a,
is a different matter altogether. The process of heat exchange between the wall
and the thermal carrier with which it is in contact is quite complicated. In many
cases, heat transmission is accomplished by convection, conduction, and radia-
tion simultaneously. Convective heat transfer is, in fact, a combined process of
heat transfer by convection and conduction; convection being the dominant
factor. Thus, for the general case

K=O, +O, (1.39)

 INTRODUCTION 13

where a, is the convective heat transfer coefficient which takes into account
thermal conduction and a, is the radiative heat transfer coefficient.
Heat transmission takes place from a hot body to a cold body, i.e., if there
is a temperature difference between the considered media. Knowing the temper-
ature gradient, the amount of heat transmitted by conduction and radiation can
be analytically determined.
The determination of the convection component «, is more difficult. For
forced convection between a liquid and a surface, the amount of heat trans-
ferred is governed by the temperature head and by the velocity of liquid flow.
In this case, not only heat is transmitted but also momentum. On the whole,
the process of convective transfer in heat exchangers is quite complex and
does not, in the majority of cases, lend itself to simple analytical calculations.
Convective heat transfer in tubes and ducts is analyzed in detail in numer-
ous works [3, 7-9]. In the present work, mainly the problems of convective
heat transfer for tube banks in flow are discussed.

1.3. HEAT TRANSFER FROM A TUBE IN CROSSFLOW

In the majority of cases, heat exchangers are made up of round tubes. Gener-
ally, flow and heat transfer for single tubes and banks of tubes are similar. Heat
transfer from a tube is closely related to the character of the flow. The principal
role in this interrelation is played by the laminar boundary layer which forms on
the frontal side of the tube. As we move away from the critical frontal point,
the boundary layer thickness 6 rises, leading to an increase in thermal resistance
and a decrease in heat transfer. At low values of Re, for a tube completely
immersed in a flow, the flow is uniform and the boundary layer separates from
the surface at the back side only. With rising Re, inertial forces start playing a
more noticeable role, and the laminar boundary layer separates from the
surface at the mid part of the round tube. Behind the tube, a complex vortex
flow is noticed. With further rise in Reynolds number (Re > 1.5 Xx 10°) a
critical flow regime is reached, whereby the boundary layer becomes gradually
turbulent, and the point of separation moves to the back side of the tube [5].
In the process of heat transfer from a nonisothermal surface of a tube in a
thermal carrier flow, a thermal boundary layer, similar to the hydrodynamic
boundary layer, is formed. The temperature in the layer varies from the wall
temperature ¢, to the temperature of flow ¢, For the theoretical determina-
tion of the local heat transfer, the thickness of the thermal layer 6, and the
temperature profile in the layer perpendicular to the surface are first deter-
mined.
The character of the temperature profile in the thermal boundary layer is
determined by Prandtl number.
In a laminar boundary layer
14 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

— (1.40)

When Pr = 1, the temperature and velocity profiles coincide. For Pr > 1 the
thickness of the thermal boundary layer is significantly lower than the thickness
of the hydrodynamic boundary layer. The main thermal resistance is concen-
trated near the surface of the tube. Heat transfer in this layer takes place only by
conduction. Since the thickness of the boundary layer is very small, the heat
transferred along the layer can be ignored, and the main heat transfer is deter-
mined in a direction perpendicular to the heat transfer surface. If this layer is
removed from the surface, heat will be transferred by convection. In the thin
layer of the thermal carrier, the flow from the wall can be determined by
Fourier equation

Gw= —A(Ot/dy)y=o (1.41)

The determination of the temperature gradient dt/dy directly at the wall of
the tube across the whole surface is a difficult problem. In practical computa-
tions of convective heat transfer from a tube, the Newton cooling law is em-
ployed

dQ=«(t,—t,) dF (1.42)

where dQ is the heat flow from the surface of tube element of area dF, and
temperature difference t, — t; = At is the local temperature head. The heat
transfer coefficient a is a function of thermal carrier velocity, tube size, physi-
cal properties of the thermal carrier, temperature head, and direction of heat
flow. The magnitude of a is a measure of heat transfer intensity. When deter-
mining heat flow from the wall to the thermal carrier (or vice versa), it is very
important to find the coefficient of heat transfer on the surface of the tube for
the heat transfer conditions of interest to us.
Surface are of the tube element dF can be found from equation

dQ = —i(dt/dy) dF,- (1.43)

By equating equations (1.41) and (1.42), we get

ts ay
If the temperature profile is known, the local heat transfer coefficient can
be easily determined from equation (1.44).
The study of heat transfer is generally reduced to the determination of the
heat transfer coefficient on the surface of a tube in various regimes of flow of
 INTRODUCTION 15

gas or viscous fluid. This problem is solved both experimentally and analyti-
cally. The temperature profile is a complicated function which depends on the
Prandtl number and flow parameters. To determine this profile experimentally,
the local temperature in the thermal boundary layer must be carefully mea-
sured.
In a laminar flow, the temperature profile at constant physical properties
can be computed from the energy equation for the thermal boundary layer of a
viscous incompressible fluid

ot te) o?
pep (4 ee + |= as (1.45)

with the boundary conditions

at y=0 u=v=0, t=t,

at Y=0O U=Hh, t=, (1.46)

To complete the solution, the equations of motion and continuity are also in-
cluded [5]:

ou Ou Op Ou 1.47
Use ee eas ieee (1.47)

Ou ov
cree (1.48)

In a turbulent boundary layer, an intermediate viscous sub-layer is formed

directly at the wall in which heat is transferred mainly by conduction. As we
move away from the wall, heat transfer is achieved mainly by turbulent mixing
which is taken into account in equation (1.45) by an additional turbulent ther-
mal conductivity coefficient X,-:

Cp (u Ox
ee oe (au j=
Oy
[a+%) <|+32
oy
[+20 a (1.49)

To solve equation (1.49), it is assumed that the turbulent heat transfer along the
y-axis is considerably higher than along the x-axis. In the limiting case y = 0
the turbulent thermal conductivity coefficient \, = 0. By moving away from the
wall, the role of turbulent heat transfer grows dominant. The turbulent thermal
conductivity coefficient is several times larger than the conductivity of the ther-
mal carrier.
To complete the equations of energy, motion, and continuity the relation-
ship between the coefficients of turbulent viscosity e, and thermal diffusivity €,
in the form of a turbulent Prandtl number is used
16 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Pr, = <_/£, (1.50)

which is determined from equation

excagee pr
gts
Ouloy
(1.51)

In the calculations, the value of Pr, is taken in the range 0.8-1.0 depending on
the value of Pr.
Computation for the turbulent boundary layer is more complicated than for
the laminar boundary layer. Turbulent coefficients of thermal conductivity and
viscosity depend on the longitudinal pressure gradient and on the turbulence of
flow. This is particularly evident for tubes in flow. Serious computational diffi-
culties arise near the region of turbulent boundary layer separation. In these
cases experimental investigations are vital, since they are necessary to deter-
mine local heat transfer and they help to refine the methods of numerical calcu-
lations.
Data from reference [4] shown in Fig. 1.4. indicate the variation of local
heat transfer around a cylinder with different Re. At low Re (curve 1) the heat
transfer in the frontal side of the cylinder is maximum, decreasing gradually

Figure 1.4. Effect of Re; on the local heat transfer for a tube in a flow of water at Tu = 1.5%; /)
Reg = 6 X 10°; 2) 10°; 3) 1.6 x 10°.
 INTRODUCTION 17

with the development of the laminar boundary layer, then increasing again in
the vortex region after separation (6 = 80°).
In the critical flow regime (curve 2) at ¢ = 80°, the turbulent boundary
layer develops leading to a sharp increase in heat transfer. At ¢ ~ 140° the
turbulent boundary layer separates.
In the supercritical flow regime (Re = 2 x 10°), the first heat transfer
minimum (Fig. 1.4., curve 3) corresponds to the direct transition from the
laminar boundary layer to the turbulent boundary layer (g = 30°), the second
corresponds to the separation of the turbulent boundary layer. It is pertinent to
mention, however, that the location of the laminar boundary layer on the tube
depends on the value of the Reynolds number and on the degree of turbulence in
the external flow.
The mean heat transfer of a tube in crossflow depends on the Reynolds
number, the physical properties of the fluid flowing around the tube, and the
hydrodynamic conditions of the flow.
In the thermal boundary layer, in a direction perpendicular to the heat
transfer surface, the physical properties of the thermal carrier change, influenc-
ing the temperature profile. The character of the temperature profile depends on
the direction of heat flow, and should be taken into account when computing
heat transfer.
Experimental data on heat transfer in flows of air, water, and transformer
oil agree well if, when generalizing, the temperature of the incoming stream ft,
and the velocity in the minimum cross section and tube diameter are taken as
the determining factors.
For the computation of mean heat transfer for a round tube, taking into
account the principal factors, the following generalized equation is used

Nu, =C Re?! . Pr} (Pr,/Pr,,)? (1.52)

Analysis of experimental data shows that the practical recommendations in [4]

for the calculation of heat transfer from a cylinder are better presented for the
flow zone as a function of Reynolds numbers (Fig. 1.5.). For the subcritical
region of flow, the exponent n for Pr is equal to 0.37, and for the supercritical
region n = 0.4. The exponent m for Re changes from 0.4 to 0.8. At low Re the
exponent varies from 0.4 to 0.5, and at higher Re it increases to 0.6. In the
critical flow regime (Re = 2 X 10°), it is rather difficult to determine a clear-
cut pattern of heat transfer change, and in the supercritical regime the exponent
m rises to 0.8.
It is experimentally established that the effect of the temperature head and
direction of heat flow in streams of viscous fluids can be; with an accuracy
sufficient for practical purposes, accounted for by the ratio Pr,/Pr,, proposed by
M. A. Mikheiev [7].
As experimental data [4] show, for a tube in crossflow the exponent p of
Pr,/Pr,, is equal, on the average, to 0.25 when the liquid is heated, and to 0.20
18 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Figure 1.5. Mean heat transfer for a tube at boundary conditions ¢,, = const and q,, = const in
flows of various fluids: /) air; 2) water.

when cooled. The maximum ranges of Pr,/Pr,, variations allow to assume p =

0.25 in calculations for both heating and cooling of the liquid. The ration Pr;,/
Pr,, = 1 for various gases, since the value of Pr on the wall and beyond the
bounds of the boundary layer is practically constant for moderate temperature
heads.
When a viscous liquid is heated Pr,/Pr,, > 1, and when cooled Pr,/Pr,, <
1. Pr,/Pr,, increases with increasing temperature head. Sometimes the effect of
the physical properties of a viscous liquid in the thermal boundary layer of a
tube is accounted for by the simplices p,/w,,, Ap/Ay» Cop/Cp- In flows of highly
viscous fluid, when the temperature changes across the boundary layer, mainly
viscosity changes and Pr,/Pr,, ~ p,/p,,.
The experimental data presented in Fig. 1.5 for the mean heat transfer from
a tube in crossflow of air, water, and transformer oil at the conditions q,(x) =
const and ¢,(x) = const at different heat flow directions agree well with each
other. In the range Re = (1.5 — 3.5) x 10°, the transition from one form of
flow to another takes place according to a complicated curve typical for the
critical flow regime. Thus, by increasing the Reynolds number, heat transfer
from a tube can be appreciably increased.
One major factor leading to an increase of heat transfer from a tube is the
incoming turbulent flow. By increasing turbulence from 1 to 15%, heat transfer
 INTRODUCTION 19

of a round tube increases on the average by 40%. At the same time, with
increasing turbulence at much lower Re a transition to critical and supercritical
flow regimes occurs.
Banks of tubes act as vortex generators, and, depending on the location of
the tubes, a corresponding turbulence level is established in each bank. There-
fore, heat transfer for tubes in inner rows is considerably higher than heat
transfer for tubes in front rows.

1.4. BANKS OF TUBES

In practice, heat exchangers made from banks of tubes are widely used. The
tubes in the banks are often arranged in an in-line or staggered order (Fig.
1.6.). Tube banks are characterized by the relative transverse (a = s,/d), longi-
tudinal (b = s,/d), and diagonal (b’ = s,/d) distances between the axes of the
tubes (pitches).
Heat transfer from banks of tubes depends on flow conditions. A tube in
the first row of the bank differs from a single tube as regards flow conditions.
This is because of the presence of other tubes in the same row, and also due to
the influence of the following (second) row. Therefore, heat transfer for the
first row differs somewhat from heat transfer for a single tube. In the inner
rows of the bank, the tubes are in contact with a flow of high degree of turbu-
lence which leads to an increase in heaf transfer.
Even earlier investigations indicated that, from the heat transfer standpoint,
the staggered arrangement is more effective.
More advanced investigations followed later. To start with, the influence of
bank type and the relative transverse and longitudinal pitches on the intensity of
heat transfer were investigated [10-12]. The majority of experimental investi-
gations showed that the heat transfer in a tube in a bank is greater than for a
single tube and depends on the longitudinal and transverse pitches.
The development of boiler technology required a more detailed study of
heat transfer for banks of tubes in gas flow. Numerous works were dedicated to
this question for a long period of time [13-17 and others].

Figure 1.6. Tube arrangements in a bank. The minimum flow section is 1-1.
20 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

With the development of the theory of similarity, the experimental results

were generalized according to this theory and were expressed by the equation of
similarity

No=c Re? (1.53)

where various values for c and m were proposed, depending on the pitch. This
allowed the transition to a wider generalization of data obtained by different
authors.
V. M. Antufiev and G. C. Beletsky [13] proposed single formulas for the
calculation of heat transfer and developed nomograms based on these formulas.
In work [14], based on the experimental data in [12] and investigations by other
authors, tables of c and m were derived. When the generalizations of different
authors were compared, discrepancies of several tens of percent were notices.
This was due in part to inadequacies of experiments and to the different meth-
ods of generalization. Some investigations were conducted on real specimens,
and often on fully heated models. Later, in the majority of experiments, the
method of local modelling was employed, i.e., one tube in a row was either
heated or cooled. The method of local modelling is based on the theory of
similarity and is subject to special investigations [15].
When generalizing the results in all the mentioned works, the diameter of
the tube in a bank is taken as the characteristic dimension, and the velocity in
the minimum flow area of the bank or of the transverse row of tubes as the
design velocity.
Analysis of the functions proposed by different authors showed that, in the
majority of cases, the choice of the determining temperature was the reason for
the discrepancy. This is the temperature at which the physical parameters in-
cluded in the similarity variables are evaluated. This discrepancy is particularly
noticeable with increasing temperature head between the gas and the wall. In
some works [10] the temperature of the wall was taken as the determining
temperature, and in others [14] the arithmetic mean temperature in the bound-
ary layer. In other cases some constants were determined from the wall temper-
ature, and other constants from the temperature of the flow. However, to take
the effect of the temperature head into account, an additional parameter of the
type T,/T,, was introduced in some cases [13].
M. A. Mikheiev [7], after studying heat transfer from a single tube in
crossflow and analyzing the results of works by different authors regarding
banks of tubes, came to the conclusion that the influence of temperature head
and direction of heat flow on heat transfer is best accommodated when the
physical quantities are evaluated at the free-stream temperature ty. N. V. Kuznet-
sov [16] established that if the physical gas constants are evaluated at the free-
stream temperature, then the results can be generalized with sufficient accuracy
without the need to introduce an additional parameter to account for the tempera-
ture factor. As was mentioned above, the choice of the determining temperature
 INTRODUCTION 21

was often the reason for the different ways the influence of tube location in a bank
on heat transfer was accounted for, and which affected the formulas.
Later, N. V. Kuznetsov [17], after adopting the free-stream temperature as
the determining one, systematized the obtained results in [11, 16] together with
the experimental data in works of other authors [10]. The effect of the pitch on
heat transfer for in-line was not detected by him, but this factor is mentioned
for staggered banks and it is suggested that it be taken into account by introduc-
ing an additional parameter (s, — d)//(s; — d) to the constant c. As regards the
index m for the majority of work, its value for staggered banks was found equal
to 0.6, and for in-line banks is about 0.65.
References [18, 19] are of interest as regards the influence of the pitches of
tube banks on heat transfer. In the former, heat transfer for a tube was deter-
mined as a function of the distance to the turbulence generating grid. It was
shown that heat transfer increases rapidly when this distance decreases. In the
second work, it was experimentally established that heat transfer for a tube in a
bank was not a function of the initial turbulence of the flow. Apparently, the
bank created a corresponding turbulence mechanism.
Numerous investigations showed that heat transfer from the first row of
tubes in a bank amounts to 60% of heat transfer from the third row, after which
it begins to stabilize.
In the aforementioned works, heat transfer was studied at moderate (1 x
10°) Reynolds numbers. Later, heat transfer of condensed banks was studied in
more detail [20] together with heat transfer at high Re. In investigations with air
[21], it was found that at Re higher than 1.2 x 10° a sharp intensification of
heat transfer takes place, which is indicated by the break in the heat transfer
curve.
Works dedicated to the investigation of local heat transfer along the circum-
ference of a tube have great significance for the understanding of the physical
mechanism of heat transfer in tube bank systems in crossflow.
G. N. Kurzhilin and V. A. Schwab [22] were the authors of the first work
on a technique for the direct determination of local heat transfer for a single
cylinder. G. A. Mikhailov studied in detail the variation of heat transfer across
the circumference of a tube in a bank in a crossflow of air [23]. Experiments
with banks of tubes in a staggered arrangement showed that the distribution of
heat transfer along the perimeter of a tube in an inner row was quite identical to
the distribution of heat transfer in a single tube with maximum heat transfer at
the frontal critical point. In in-line banks, a zone of weak circulation of the
liquid was observed between the inner tubes; consequently, heat transfer from
the front and rear parts of the tubes was insignificant. Therefore, unlike stag-
gered banks, maximum heat transfer for in-line banks occurred at an angle 50°
from the frontal point.
Much less research has been directed towards the determination of heat
transfer for banks of tubes in crossflow of liquid droplets. Notwithstanding its
immense practical significance, research in this field was started relatively re-
22 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

cently. Works [24, 25] dealing with the investigation of heat transfer for a single
tube in flow of liquid facilitated the development of research methodology for
heat transfer in banks of tubes, and helped, to a certain extent, to elucidate the
influence of the type of liquid.
O. P. Bergelin, et al. [26, 27] employed complete modelling to investigate
heat transfer for tube banks of various arrangements in a flow of viscous oil.
The physical properties in the similar variables were evaluated at the stream
temperature, and the influence of the temperature heat taken into account by the
parameter (Hp! by) Investigations were conducted at Reynolds numbers less
than 1 x 10°.
V. P. Isachenko [28] used local modelling to investigate heat transfer for
staggered and in-line banks of tubes in crossflow of air, water, and transformer
oil. This work facilitated the development of a technique for the determination
of the influence of physical properties of the liquid and temperature head on
heat transfer in tube banks.
The first publications of the works of IPTPE AS LitSSR on the investiga-
tion of heat transfer for in-line [29] and staggered [30, 31] banks of tubes in
crossflow of different liquids go back to roughly the same period of time.
The investigation of heat transfer for flows of liquid metals is of immense
interest as a new branch of technology. Heat transfer for tubes in liquid metal
thermal carriers is studied in detail by V. M. Borishansky, et al. [32]. The
distribution curve of the heat transfer coefficient along the tube perimeter for
staggered banks in liquid metals indicates that the maximum heat transfer oc-
curs in the front side of tube and is absent in the back side. At Pr < 1, due to
the high conductivity of the thermal carrier, turbulence caused by the separation
of the boundary layer does not noticeably affect heat transfer. Therefore, the
character of mean heat transfer corresponds to the character of heat transfer for
banks of tubes in laminar flow of liquid.
For the determination of heat transfer for banks of tubes in flows of liquids
with various physical properties, the following similarity equation is usually
used

Nu=c Re™. Pr’ (1.54)

where the type of liquid is accounted for by the term Pr’.

Due to the lack of experimental data on the influence of the physical prop-
erties of the liquid on heat transfer, the exponent n is taken by different authors
from 0.3 to 0.4. Consequently, the scatter of the heat transfer coefficient
reaches 50% in the case of viscous liquids.
In the 1960s, due to the intensive development of several emerging technol-
ogies, keen interest was shown in the question of heat transfer for liquid drop-
lets. The available data, however, was insufficient to derive complete computa-
tional functions. The study of the effect of the type of liquid, direction of heat
 INTRODUCTION 23

flow, temperature head, arrangement of tubes, and other factors on heat trans-
fer and resistance of banks in a flow of liquid remained incomplete.
With the purpose of elucidating these questions, a comprehensive pro-
gram of experimental investigations of tube banks using identical experimen-
tal aparatus and a unified methodology was undertaken at IPTPE AS LitSSR.
The problem of the influence of configuration of tube banks in crossflow on
heat transfer and resistance was studied using 12 staggered and 15 in-line
banks with various longitudinal and transverse pitches in flows of water,
transformer oil, and air. Investigated were heat transfer from tubes in banks
as a function of flow velocity, the type of liquid, the effect of temperature
head, and the direction of heat flow in the subcritical region for the range of
Pr number from 0.71 to 500.
The extensive experimental material helped to explain the mechanism of
heat transfer and resistance in banks, construct a physical picture of the pro-
cesses taking place, determine the influence of the physical properties of the
liquid and tubes arrangement on heat transfer and hydraulic resistance of banks,
find the most rational forms of generalization, and to compare the effectiveness
of banks of different configurations.
Based on the analysis of obtained data, it was concluded that the free-
stream temperature tf;should be adopted as the determining temperature, and to
account for the changing properties of the liquid with temperature in the bound-
ary layer, the parameter (Pr,/Pr,,) to the power 0.25 should be introduced into
equation (1.54). The results of these investigations were generalized in a mono-
graph by A. Zukauskas, V. Makarevicius, A. Slanciauskas [1], which was
published in 1968. After the publication of the monograph, investigations at the
Institute continued to cover the critical and supercritical flow regimes, and also
the region of low Reynolds numbers. New data obtained at other scientific
centers are also presented in the present book.
Factors such as surface roughness and finning also influence the intensity
of local and mean heat transfer of banks. These questions are discussed in detail
in the following chapters of the book.
The choice of a rational heating surface for a heat exchanger is feasible
only on the basis of joint solution of the problems of heat transfer and hydraulic
resistance. Therefore, in the majority of cases, in addition to heat transfer in
banks of tubes, the peculiarities of flow and the determination of hydraulic
resistance are studied. Usually, the experimental data on resistance are general-
ized by the relation

Eu=k Re-" (1.55)

To determine the effect of bank geometry on resistance, nomograms or tables of

the values k and r were constructed. Detailed analysis of flow, resistance, and
vibrations of tube banks was presented in the latest work by the authors [5].
24 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

To further improve heat exchangers, considerable attention was given to

increasing their effectiveness and determining optimum design parameters. For
the solution of this problem, investigations were conducted to determine local
and mean heat transfer and hydraulic resistance for both a single row and a
whole bank. During these studies, relative pitches were varied in a wide range
for in-line and staggered banks of tubes. Attention was given to the intensifica-
tion of heat transfer by the elements of surface roughness and to the determina-
tion of their effect on heat transfer and hydraulic resistance of a bank in a wide
range of roughness-element height change for normal tube arrangements in
banks and high compactness.
When studying the effectiveness of finned banks of tubes, major attention
was given to the optimization of fin height as a function of the Prandtl number.
When designing future heat exchangers, it is expedient to employ com-
bined data on the analysis of heat exchanger compactness and its thermal and
hydraulic characteristics.
 CHAPTER

TWO
LOCAL HEAT TRANSFER FROM BANKS
OF SMOOTH TUBES

2.1. SPECIFIC FEATURES OF HEAT TRANSFER

The distribution of the local heat transfer coefficient along the tube perimeter
reflects the flow character in the boundary layer on the surface of the tube. By
analyzing the mechanism of this distribution in a wide range of Reynolds num-
bers and relative pitches, one can obtain the most optimum flow parameters,
tube arrangements in banks, and the generalizing relationships for the computa-
tion of the local heat transfer coefficient.
In the present book, the local heat transfer coefficient is presented in a
dimensionless form a/@ = f(g, a X b, z) and its mean value is given for each
specific case. To compare the values of this coefficient in air flow and various
liquids, the distribution of the dimensionless complex Kg = Nu,,/Pr;' (Pr,/Pr,,)’
along tube perimeter is introduced.
The effect of the Prandtl number and temperature head can be accounted
for by introducing the local parameters Pr,’ and (Pr,/Pr,,)’ after determining the
power indices n and p beforehand. Since there are regions with laminar and
turbulent boundary layers of the tube surface, the power index n is a function of
flow regime in the boundary layer. For laminar flow around a plate n = 0.33,
and for turbulent flow n = 0.43. These same values are also used in the case of
flow around a bank of tubes. The power index p, which takes into account the
temperature head, is determined experimentally, and can be assumed to be
equal to 0.25 for both regimes in the boundary layer with sufficient accuracy
for practical purposes. In cases of heat transfer in streams of highly viscous

25
26 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

liquids and at high temperature heads, the parameters Pr,’ and (Pr,,)’ have large
values and their role rises significantly. In case of cooling of the turbulent
boundary layer on a plate for the non-gradient flow of a highly viscous thermal
carrier, when the ratio Pr,/Pr,, varies up to 25 times, the power index is equal to
0.167. Therefore, for large values of Pr,/Pr,, in a dominant turbulent boundary
layer in a cooled tube bank, it is recommended to assume that p = 0.17.
Local heat transfer along the perimeter of a tube in a bank can be computed
from equation

Nu,, =c Ref Pr? (Pr,/Pr,,)? (2.1)

but in order to do this the local value of the power index m, of Re, must be
determined. On the basis of the values a/@ = f(y, a X b) it is easy to deter-
mine m, = f(g, a X b). Data generalized in this fashion are convenient to
compare with the results of theoretical investigations. In addition, they could
serve as initial data for such investigations.
Local heat transfer for a tube in a bank behaves the same way as local heat
transfer for a single tube. The distribution of heat transfer across the surface is
determined from the specific features of flow around tubes in a bank, which
largely depends on bank configuration.
Fig. 2.1 shows the characteristic features of heat transfer variation for
tubes in inner rows in in-line and staggered banks compared with heat transfer

J 50 PO GH cp Rl 7 ee eae
Figure 2.1. Variation of heat transfer along the perimeter of a /) single tube, 2) tubes
in staggered,
3) and in-line banks.
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 27

for a single tube. Due to flow impingement on tube surfaces in staggered banks
and the influence of free-stream high turbulence, heat transfer in the frontal
critical point is higher than heat transfer for the same point on a single tube.
Heat transfer from the remaining parts of the tube is also higher due to the high
vorticity of flow. The maximum heat transfer coefficient for in-line banks is
reached at some distance away from the frontal critical point.

2.2 LOCAL HEAT TRANSFER IN THE LOW RANGE

OF REYNOLDS NUMBER

In studying the features of heat transfer and its intensification in complex flows,
primary attention was given, up to now, to convective heat transfer in the
medium range of Reynolds numbers. At the present time, the problem of heat
transfer in the case of natural convection and at low Reynolds numbers has
become urgent. This is caused by the fact that, for natural convection, either no
energy is expended to move the thermal carrier or very little energy is needed
compared with the expenditure on turbulent flow for medium and high range
Reynolds numbers.
For a long time, it was considered that heat transferred by natural convec-
tion is insignificant compared with forced convection. Investigation of heat
transfer in natural and mixed convection was initiated to conserve some of the
energy expended to move the thermal carrier using new technical innovations in
the design of heat-exchanger surfaces which are capable of operating effectively
at low Reynolds numbers. In addition, when determining heat transfer by natu-
ral and mixed convection at low Re, wider use of well-tested numerical compu-
tation methods can be made, thereby reducing the amount of experimental in-
vestigations.
Let us consider the features of heat transfer in the case of mixed convec-
tion.
If the tube wall temperature is higher than the ambient temperature, then
the flow of the thermal carrier begins to move up, and in the case of a cold wall
it moves down. In natural or free convection the process of heat transfer is
achieved by conduction and convection. To determine the heat transferred by
natural convection in the range Re = 1-10, special investigations were con-
ducted in flows of air and viscous liquids for high Grashof numbers. It was
found that for natural fluid movement, the forces of internal viscosity affect
heat transfer significantly. Their value depends on the velocity distribution char-
acter in the vicinity of the bank and the thermal condition of the liquid near the
heat transfer surface. The complicated distribution character of temperature
profiles around the tube in a bank causes the acceleration of the liquid and,
consequently, the appearance of inertial forces.
Calculation of heat transfer in banks of tubes in the case of mixed convec-
28 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

tion is conducted on the basis of the boundary layer equation taking into consid-
eration the influence of volume forces and ignoring viscosity dissipation.
Density of the liquid is not strongly dependent on pressure and changes
mainly as a function of temperature

i
See ee Sg (2.2)

where 83, is the coefficient of volume expansion of the liquid. The dependence
of density » on pressure and temperature is determined from the Clapeyron-
Mendeleev equation

C= pi? (2.3)

For moderate temperature drops, the density of a viscous incompressible

liquid changes little, and the lift force is determined from the relation

& (er —9) =Z Boe (t— Fy). (2.4)

For natural convection expression, Eq. (2.4) is called the Boussinesq ap-
proximation.
The local heat transfer for natural convection is determined from boundary
conditions of the third kind:

(bn — &) = — At). ~0 (2.5)

and the frictional stress on the wall from

Tx = u(du/dY) yo (2.6)

If heat transfer is caused by mixed convection, the largest contributor to the

overall heat transfer coefficient at Re = 1-10 is natural convection.
For the determination of heat transfer from banks of tubes at mixed convec-
tion, special investigations were conducted and the obtained results were gener-
alized according to the similarity equation

Nu,=/(Grp Pr; Pc,/Proed,.b, 8) (21)

where @ is the angle subtended between the directions of forced flow and lift
force which is still called volumetric force, generated by thermogravitational
convection. The volumetric force

Fy= —geBo(ty—
ty) (2.8)
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 29

where {, is the volumetric expansion coefficient, and t, is the temperature of

undisturbed flow. The Grashof criterion is found from the equation of motion
as a ratio of inertial and volumetric forces:

Ri 3 rs

= ==> Bo (hy hp) (2.9)

where L is the characteristic dimension (in tube banks, L is the tube diameter).
At high Prandtl numbers and moderate temperature head, thermogravitational
convection appears in the vicinity of the tube surface. Ignoring the inertial term
in the equation of motion, the following relation is obtained

£0o wt) dh, uy = Rel Fs, (2.10)

A combination of dimensionless criteria, aimed at excluding Re; leads to

Reyleigh number:

Ra, = gd
OMG ats) (2.11)
3

Vea

which is used in the form of a function Nu; = f(Gr; Pr,) mainly when studying
heat transfer in cases of natural convection.
To determine the coefficient of heat transfer for banks of tubes, investiga-
tions were conducted at Re, = 0 for natural convection, and at Re; from 1 to
200 for mixed convection in flows of air, transformer and aviation oils.
For natural convection and liquid heating, zero angle was taken at the
bottom side of the tube as indicated in the lower section of Fig. 2.2. For mixed
convection zero angle was taken at the front side of the tube increasing from left
to right, as indicated by the top section of Fig. 2.2. Starting with Re, ~ 1, a
smooth nonseparating flow was visually observed. Apparently, in the interval
for Re; from 0 to = 0.1 a certain flow pattern exists in which the subtended
angle between the directions of forced flow and lift force 0 < 8B < 90°. For
natural convection, (Fig. 2.2) the local heat transfer on the surface of the tube
varies slightly. The choice of Pr; > 7000 and moderate temperature heads
(At = 10-45 °C) does not generate thermogravitational flow of the liquid, since
viscous forces exceed lift forces. Therefore, heat transfer takes place by con-
duction. The decrease in heat transfer at 135 < ¢ < 180° can be explained by
the fact that in this zone thermal boundary layer is somewhat thicker than at
Orarg*<185
In order to determine the temperature head at which lift forces exceed
viscous forces, special investigations were staged for a flow of aviation and
transformer oils in a transparent duct. The temperature of aviation oil t, was
maintained in the range 20-30°C which corresponds to the change of Pr, from
15936 to 7558. Beginning from At = 70-80°C, noticeable movement of the
30 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Go 60 ey
Figure 2.2. Distribution of local heat transfer on the surface of a tube in an inner row of a
staggered bank 2.0 x 2.0 for mixed convection. Rey = 203 Ky =U Pre (rfPr,) oo

hot liquid mass from a bank tube is observed leading to the increase of heat
transfer coefficient. As a result of thermogravitational flow generation, the
local heat transfer coefficient in the region 135 < yg < 180° increases. This
force can be determined from equation (2.8). In the case of heat transfer at
constant temperature of the thermal carrier the values g, p, and By are con-
stants. For large temperature heads, the dependence of p, on temperature is
determined from equation (2.2). Then the volumetric force is a function only of
the temperature head. The determination of this force and the beginning of heat
transfer by natural convection is of great importance for practical purposes. The
need might arise to heat large volumes of a liquid with a very high Pr, at low
temperature ¢, for example, a reservoir of fuel oil in boilers in the winter,
transformer oil in transformers which are cooled by natural convection and so
forth. While studying heat transfer by natural convection in highly viscous
fluids (aviation and transformer oil), it was noticed that upon reaching a tem-
perature head of the order 50-60 °C the thermal carrier is brought into a state of
motion due mainly to thermogravitational forces. With increasing At this mo-
tion is enhanced, and at At = 80-90° the liquid is brought into motion starting
from ¢ = 0° with an increased intensity at the top side of the tube in the region
g = 135-180°. These data allow the determination of the optimum working
parameters of banks with different pitches a x b as a function of Pr, values.
However, in practical conditions, such a large head is difficult to attain,
and the local heat transfer coefficient increases only in the region 135 < g <
180°. Therefore, if there is a need to heat a liquid with very high viscosity,
either the heat transfer surface is increased or forced convection is employed.
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 31

With this in mind, investigations were conducted at low Re, (Fig. 2.2). As the
figure shows, at a very low velocity of viscous fluid motion heat transfer in-
creases appreciably. An incoming velocity of uw) = 0.04 m/sec corresponds to
Re, = 5. (In the minimum flow area of the bank, velocity increases up to Z =
0.08 m/sec). The mean coefficient of heat transfer & is found to be equal to 204,
whereas in the case of natural convection @ = 81 at the same temperature head
At. Consequently, the mean heat transfer coefficient increases 2.5 times. Along
the tube perimeter in an in-line bank, the local heat transfer coefficient varies
considerably (this is particularly so at ¢ ~ 70° where a, is a maximum and at
ge = 180° where aq, is a minimum) which is tied with the presence of two
recirculation flow regions for the liquid—in the front and back sides of the tube.
Since velocity of liquid motion at the back side of the tube is low, the local heat
transfer coefficient is far less than its mean value.
For the determination of the effect of tube banks pitch on heat transfer,
banks with large (2.0 x 2.0) and small (1.25 x 1.25) pitches were investigated
[33]. Local heat transfer in large pitch banks is more uniform, an exception
being the region 120 < g~ < 180° in which heat transfer decreases because of
the large thickness of thermal boundary layer creating thermal resistance. In
small pitch banks, local heat transfer varies along the whole perimeter of the
tube. This is linked with the influence of neighboring tubes which reduce heat
transmission in the region 90 < ~ < 180° in particular. In inner rows of a
staggered bank 2.0 x 2.0, the mean heat transfer coefficient @ = 81, and in the
same row of 1.25 x 1.25 is equal to 66.5, which is an indication of pitch
influence on heat transfer. By increasing the pitch from 1.56 to 4.0 heat transfer
rises 1.7 times, approximately.
In the first row of the bank (Fig. 2.3) heat transfer on the surface of the
tube varies less than in inner rows since there is no influence of neighboring

(a
Figure 2.3. Variation of local heat transfer for a tube in the first row of a compact in-line band: /)
Rey = 3.0, & = 234; 2) Res = 100, & = 460; 3) Ree = 1500, a = 1610.
32 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

tubes on some parts of the row. When Re reaches 200 and increases further, the
effect of natural convection becomes insignificant.
Let us examine the specific features of local heat transfer in the first and
inner rows of tubes in in-line and staggered banks at the most typical pitches
and Reynolds numbers. The character of local heat transfer in the first and
inner rows of tubes of an in-line bank having small pitches is variable. Despite
the fact that the incoming stream collides with the first row tubes at the frontal
critical point, the maximum heat transfer as a function of Re; is found in the
region g = 75-90° (Fig. 2.3). This is associated with high acceleration of the
flow in the frontal side of the tube. After collision with the tube, a vortex zone
is generated in the flow in the region of the frontal critical point. With increas-
ing Re, the vortex zone decreases and local heat transfer increases (Fig. 2.3,
curves 2, 3). In the rear side of the tube, the level and character of the local heat
transfer depends on the intensity of circulation. At low Re; (in the order of
unity) a laminar boundary layer develops around the tube without separation,
and a/a@ decreases up to y = 180° without a change in character (Fig. 2.3,
curve 1). With increasing Re, separation of the boundary layer commences.
The location of the separation depends on Re and on the relative pitches of tube
banks. Separation of the boundary layer causes an increase in a/& in the rear
side of the tube starting from g = 110° (Fig. 2.3, curve 3).
The character of local heat transfer in an inner row of tubes of a staggered
bank differs considerably from heat transfer in the first row [33]. The value of
the local heat transfer coefficient in an inner row of a bank depends on the
velocity on the external border of the boundary layer and on flow turbulence,
and the character of its distribution depends on the location of the tubes in the
bank which is defined by the pitches a x b. The magnitude of flow turbulence
in the boundary layer and in the intertubular spacing depends on the pitches
a X b. The main liquid stream in in-line banks of tubes flows between rows.
The maximum heat transfer occurs on the section where the flow and the tube
collide, which is located at ¢ ~ 70-80° at low Reynolds numbers. In recircula-
tion zones, heat transfer is considerably lower in comparison with the mean
heat transfer of the tube. For example, in the region of frontal critical point
(yg = 0-20°) the value a/& amount to 0.5; in the region of the rear critical point
it is somewhat higher, and at Re, of the order of 1500 it reaches the mean value
of the heat transfer coefficient. The location of collision of the flow and the
tube, i.e., the maximum value of the heat transfer coefficient, changes with
changing longitudinal pitch b. With increasing b, the location of flow-tube
collision moves towards the frontal critical point, intensity of liquid motion in
recirculation zones rises, and at large values of b the collision of flow and tube
commences at the frontal critical point. Thus, the variation of the local heat
transfer coefficient in in-line banks of tubes depends on the longitudinal pitch.
Let us consider the effect of pitches a x b on the local heat transfer in
inner rows of in-line tube banks. Fig. 2.4 shows the distribution of a/a = f(¢,
Re,) at characteristic values of Re, of significantly varying magnitudes. With
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 33

RIlR

10

05

0 30 60 oo 120 {50 Y
Figure 2.4. Variation of local heat transfer for an inner row tube of an in-line bank: /) Re, = 2.0,
a = 190; 2) Rep = 100, & = 450; 3) Rey = 1330, a = 1530.

increasing pitches, liquid circulation intensity and the local heat transfer at the
frontal critical point increase. At low Re,, the rear side of the tube is still in
contact with the boundary layer; therefore, the local heat transfer decreases up
to g = 180°. Increasing Re, leads to the separation of the boundary layer and to
the intensification of local heat transfer. If at Re; = 2.0, a/@ = 0.3 at the
frontal critical point, then increasing Re, up to 110 gives a/a& = 0.70, i.e., it
increases by almost 2.5 times. In the region of the frontal critical point of a tube
located in an inner row of an in-line bank, the ratio a/@ is almost independent of
Re, In the rear side of the tube, with increasing Re, a/a increases and at Re; =
1330 it reaches unity, i.e., the local heat transfer coefficient becomes equal to
the mean heat transfer of the tube. With increasing pitches from 1.25 x 1.25 to
1.50 x 1.50, the variation of tube thermal load in a bank remains the same,
i.e., a/& fluctuates from 0.5 to 2, on the average.
We now examine the specific features of the distribution of the heat transfer
coefficient in the first and inner rows of an in-line bank having large pitches at
the most typical values of Re; At low Re,, increasing the bank pitch leads to a
further rise of the local heat transfer in the region g¢ = 0°. When the pitch
changes from 1.56 to 4.0, the ratio a/&at ¢ = 0-30° increases 1.9 times on the
average. This indicates that increasing the distance between the tubes facilitates
more intensive recirculating motion in the region of the frontal critical point.
Increasing Re leads to a larger increase in o/a@ in the region g = 0°. In the rear
side of the tube at low Re,, the local heat transfer in the region ¢ = 180° is
34 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

considerably lower in comparison with the obtained value at g = 0°. The ratio
2927/1, 1g0° = 4.
With increasing Re,, heat transfer in the rear side of the tube intensifies,
and the minimum value for a/& shifts from g = 180° to g = 120°. At Re, =
10° the ratio a, _9-/a,- 129° drops to 2.5. This indicates that the nonuniformity of
the tube thermal loading decreases from 4 at Re, = 1.5 to 2.5 at Re; = 10°, but
still remains relatively high. In the ideal case, the condition a/a = 1 should be
satisfied, at which thermal loading is uniform for the whole tube.
The local heat transfer in an inner row of an in-line bank with large pitch
increases the same way as in the first row. However, the presence of two
recirculating zones in the longitudinal row of a bank causes highly nonuniform
thermal loading of the tube. In the medium range of Re, from 1 to 10° the ratio
of the maximum and minimum heat transfer coefficients o,,,,/Qmin amounts to 3-
3.4. Analysis of the variation of local heat transfer on the tube in a bank shows
that accelerating the flow in the frontal side influences the character of heat
transfer distribution significantly.
In in-line banks, beginning from the second row inwards, the tubes are
obstructed by the preceding row, i.e., the so-called “‘shadow effect’’ is intro-
duced; therefore the heat transfer character of inner row tubes differs apprecia-
bly from the heat transfer character of the first row tubes. The same argument
applies to staggered banks. The area of a tube surface located in the “‘shadow”’
region depends on the longitudinal pitch. The whole front side of the tube is
shadowed if the pitch is small. If the longitudinal pitch is too large the “‘shadow
effect’’ disappears.
One would think that the character of local heat transfer in the frontal side
of the first row tube of a bank at ¢ = 0-90° must influence only the incoming
flow characteristics. As was the case of flow around a single tube, the maxi-
mum heat transfer coefficient must be at the frontal critical point. However, for
a pressure gradient the maximum heat transfer in a tube bank does not always
occur at g = 0°. Fig. 2.5 shows the ratio of the maximum value of local heat
transfer coefficient a,,,, in the frontal side of the tube to its value in the frontal
critical point @,_9- as a function of Reynolds number. At Re ~ 1 the maximum
value Of Q,4,/,-9 ~ 1.4 corresponds to a bank with a = 1.25; this value
decreases with increasing Re,. For a bank with a = 1.5 at the same Re, the ratio

1 2
Lect GG OlO at t2\, nbn BinCSinemee lien tates
Ge Saat
Figure 2.5. Influence of transverse pitch on local heat transfer from the first row of in-line banks:
PV L255" 255 2VALSE x 105 83) 210 ee)
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 35

> tS am
iW As tae ae

0 OE, 15 6 e) Ge BO Din f20. 750

Figure 2.6. Variation of local heat transfer from a tube in a compact staggered bank: a—first row:
1) Rey = 4.0, @ = 260; 2) Rep = 100, & = 600; 3) Rep = 530, & = 920; b—inner row: /) Rey =
3.0, & = 270; 2) Rep = 100, & = 820; 3) Rey = 1600, & = 2120.

Qmax/Q@y-9 is less, and with increasing Reynolds number the effect of pressure
gradient dp/dx disappears at Re; ~ 50. In a bank with a = 2.0 the ratio a,,,,/
a,-o ~ 1 from Re, =~ 1. This figure shows that heat transfer in the frontal
critical point of the first row of a tube bank is influenced by the pressure
gradient and that this influence is dependent on the values of a and Re.
Consider the particular features of the distribution of local heat transfer
coefficient of tubes located in the first and inner rows in staggered banks. We
shall analyze the most typical arrangement of a staggered tube bank. At low Re;
(Fig. 2.6 a, curve /), heat transfer in the first row of a compact staggered bank
is a maximum not in the frontal critical point, but at ¢ = 80-90°. This is the
most characteristic feature of flow around the first row of all compact banks.
When Re; increases from 4 to 500, a/a increases in the frontal side of the tube.
In the rear side at low Re; separation does not commence and a/@ decreases up
to ¢ = 180°. Increasing Re; leads to the separation of the boundary layer and
a/& increases in the rear side of the tube. For example, at Re, => 500 and g =
180° the ratio a/@ reaches the same magnitude as at ¢ = 0°. In an inner row of
the bank, the local heat transfer at ¢ = 0° is maximum for all values of Re;.
This indicates that the effect of increasing flow turbulence in the front tube
exceeds the effect of pressure gradient. Some change in the character of the
ratio a/@in the region g = 40-90° is caused by the effect of flow acceleration
and vortex dissipation. The distribution character of the ratio a/q in the rear
side of the tube in inner and first rows are analogous.
Heat transfer in the frontal side of a tube in a bank of medium compactness
(1.5 x 1.5) changes appreciably at all Reynolds number, with the exception of
low values (Re, = 1-5), when the tube is in a flow without separation. At
36 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

higher Re, the maximum value of a/@ is at ¢ = 0°, then it decreases to the
point of separation. With boundary layer separation, o/a increases in the re-
gion ¢ = 110-180°. At low Re, heat transfer in an inner row of such a bank is
at maximum at g = 0°. An increase in spacing between the tubes leads to a
decrease of the influence of acceleration in the region ¢ = 40-90°; therefore,
the curves of heat transfer from bank 1.5 x 1.5 are less deformed than similar
curves for bank 1.25 x 1.25 in the same regions and at the same values of Re,
An increase of Re; leads to a shift of boundary layer separation towards lower
values of y. If separation in the case of Rey = 100 commences at ¢ = 13555
then in the case of Re, = 1400 it commences at y ~ 115°.
The study of the character of local heat transfer in the first and inner rows
of a widely-spaced staggered bank is of practical interest. If the main thermal
loading of the tube is concentrated at the frontal side and the maximum loading
is at g = 0° at high Re, then for Rey = 1-5 the maximum a/@ is located at
gy ~ 60°. The reason for this is the longitudinal velocity gradient du/dx, the
maximum variation of which takes place in the interval ¢ = 0-60°. The mean
velocity in the minimum flow cross section of a bank 2.0 x 2.0 increases
almost two times. At high Re, and up to 10°, the velocity gradient is not capable
of changing the distribution character of a/& peculiar to a single tube in flow,
for which the ratio a/@ is a maximum at the place of flow-tube collision and
decreases with increasing boundary layer thickness until separation occurs.
Thermal nonuniformity of a tube in the first row of a bank with large pitch
varies from a/@ = 1.75 aty = 0° toa/a = 0.25 at ¢ = 180°. Nonuniformity
of thermal loading of a tube in an inner row of this bank decreases slightly
since, following separation of the boundary layer, intensity of heat transfer
increases in the region g = 120-180°. Distribution characters of a/& = f(y) at
Re, ~ 1, 100, and 1000 differ substantially. It is important to note the relatively
high heat flux from the tube at low Re;,. If at Re, = 2 the heat transfer coeffi-
cient &@ = 200, then with increasing Re, to 1000 it increases more than 7 times.
Consequently, if there are no special requirements of heat transfer from tube
banks, it is expedient to operate in a low Re, regime.
In Figure 2.7 experimental data on heat transfer from the first and inner
rows of a bank are presented in polar coordinates. These data readily show the
maximum and minimum values of a/& along the tube perimeter at different
values of Re, The figure shows that all Re, the heat transfer coefficient in the
rear side of the tube is not high and that its contribution to the total heat transfer
of a tube in a bank is insignificant.
In the region of flow-tube collision the value of a/& is a maximum. From
this site, the stream bifurcates into the main and recirculating flows. Heat trans-
fer in this region is little studied theoretically; however, its determination is of
immense practical interest. With the development of the boundary layer from
the site of impingement of the flow on a tube in a bank onwards, the intensity of
heat transfer decreases substantially. In practice, there are cases when the tube
is in a strong turbulent flow. Here the computation of heat transfer becomes
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 37

Figure 2.7. Experimental data on local heat transfer from a tube in an inner row of in-line bank
AMD SK BAY.

complicated due to the effect of considerably varying pressure gradients from

the site of flow-tube collision and from the generated vortices in this region. To
calculate heat transfer in the region of the critical point for a perpendicular
impingement of flow on a plate, the following equation from reference [4] is
used

Nuc 0.57 Reo Pr Ce12)

For the determination of heat transfer in the frontal critical point of a tube,
the following dependence is obtained in work [34]

Nuy = 1.14 Res pro3 (2.13)

in which the incoming flow velocity is the generalizing velocity.
In work [4] a function is obtained for the calculation of heat transfer in
the critical point in a wider range of Prandtl numbers, namely, from 10 to
100:

Nip 0 URE sip (2.14)

According to work [35], in the region of the frontal critical point (py <
30°) at Re, = 200 heat transfer is not dependent on the limiting conditions.
However, beyond this region heat transfer in the case q,,(x) = const intensifies.
38 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Based on the above-mentioned functions, the local heat flow is determined from
the equation

Gy = sty eeu ce pO th) (2.15)

To obtain data on heat transfer for a tube in a turbulent-free stream further

investigations are needed. Accumulation of experimental data on local heat
transfer for these conditions in a wide range of variations of pressure gradient
and turbulence of incoming flow can help to refine the calculation methods of
such flows.
The large increase in velocity of flow around a tube in a bank, starting
from the frontal critical point to the minimum flow cross section, substantially
influences the distribution character of the local heat transfer on the tube sur-
face. Turbulence depends on tube diameter and bank pitches. At low Re,, vortex
motion of liquid in the space between tubes influences bank resistance more
than heat transfer. This is observed during investigations in flows of high vis-
cous fluids such as aviation and transformer oils.

2.3. LOCAL HEAT TRANSFER IN THE MEDIUM RANGE

OF RE

The medium range of Reynolds numbers are considered to be in the range from
2 x 10°to2 x 10°. High vorticity in the space between tubes in banks causes a
transition from a laminar flow regime in the boundary layer to a turbulent flow.
The transition point coincides with large values of ¢, which reduces the dimen-
sions of the recirculation flow zone behind the tube.
With the appearance of turbulence in the boundary layer, the distribution
character of the local heat transfer coefficient changes substantially. The transi-
tion process from a laminar to turbulent regime in tubes, ducts, and on plates is
widely studied. The tendency of a fluid to cross over from one flow regime to
another is an important factor which must be taken into account when designing
heat exchangers and optimizing their technical characteristics. In a turbulent
flow, intensive impulse transfer across the flow takes place, which leads to a
fuller velocity profile across the duct and to the intensification of heat transfer.
For a plate in flow, transition to a turbulent regime occurs at Re,, = (u,d*/
Up)er = 420. A tube, perpendicularly located to flow, is a badly streamlined
body; therefore, pressure gradient and turbulence of flow in the longitudinal
direction vary, influencing velocity distribution in the boundary layer, and Re,,
could decrease downstream. When studying the local heat transfer on the sur-
face of a tube in a bank, it is important to know when transition begins from
laminar flow regime in the boundary layer to turbulent flow. For this purpose
one must know the length of the laminar, transient, and turbulent boundary
layers and the latter’s point of separation. The determination of these character-
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 39

istics is the most important task when calculating local heat transfer on the tube
surface since, for each flow regime, the power index m in the expression a =
a” varies significantly. These data must be defined for tube banks since the
longitudinal pressure gradient and turbulence level strongly depend on the
pitches and Re,.
Apart from the indicated factors, surface roughness influences the point of
transition. Heat transfer from the wall of the tube to the liquid promotes turbu-
lence generation, and the opposite stabilizes the flow.
When designing an efficient heat exchanger, it is vital to know the tempera-
ture distribution on the surface of the tube, the local heat transfer coefficient,
and the velocity profile in every section. These characteristics vary with Re, and
the pitches of the banks. Distribution of the velocity profile, temperature, and
their pulsations in the boundary layer and in the intertubular spacing gives an
idea about the magnitude of turbulent heat transfer and indicates the way to
improve heat exchangers on the basis of optimum ratio of momentum and heat
transfer. In real conditions, heat transfer does not take place at q,,(x) = const or
t,(x) = const. If the heat-exchanger tubes are manufactured from materials
having a low thermal conductivity coefficient, heat transfer takes place nearer
to the condition q,,(x) = const. If the tubes are made from materials with a high
thermal conductivity coefficient, heat transfer takes place nearer to the condi-
tion ¢t,(x) = const. At g,,(x) = const the temperature of the tube wall changes
substantially. If the values of the mean heat transfer coefficients are known
from figures, or they can be determined from generalizations in the form Nu, =
c Re;'Pr; (Pr,/Pr,,)’ equations for the calculation of mean heat transfer, and also
from graphics of a/a@ = fig, a X b, Re), then the local heat transfer can be
easily determined. If the temperature of the thermal carrier is known and the
temperature head is given from the graphic a/& = f(¢), then it is easy to
determine the local heat flow on the surface of the tube. If such graphics are not
available, then calculations must be made to derive the local heat transfer char-
acteristics separately for laminar and turbulent sections of the boundary layer
and for the recirculation zone. For the transition zone, the mean value of the
heat transfer coefficient can be assume for laminar and turbulent boundary
layers at the point of separation.
In some cases, the incoming flow at the front row of bank tubes could be
either laminar or turbulent owing to the influence of the diffuser, bend, gate
valve, regulator, or instruments.
Here, the stabilized laminar flow upstream of the bank can become turbu-
lent. Also, flows of the following forms can appear:

e stabilized laminar;
¢ unstabilized laminar which takes place over some length along the duct with-
out disturbance, and then becomes stabilized;
e stabilized turbulent;
e unstabilized turbulent.
40 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

If turbulent flow appears as a result of disturbances at Re, ~ 2320 (d, =

4F/P; F is the cross section of duct; P is the perimeter of duct), then it does not
become laminar even after their elimination, although a laminar flow can be
established in very smooth ducts up to Rez, ~ 1.1 x 10°. Practically, at Re,, >
3000, the flow in the duct is considered turbulent. To determine the local heat
transfer from the tube of the first row in a bank, attention must be paid to the
velocity of incoming flow. For example, because of turns at elbows, the incom-
ing flow at the first row of tubes is extremely irregular, which reflects strongly
on heat transfer from the banks in large heat exchangers.
Heat transfer in the first rows of tube banks in a turbulent flow is far less
investigated than heat transfer in inner rows.
When fluid flow around the frontal side of a tube in the first and inner rows
of a staggered bank, the flow accelerates, and after crossing the minimum
cross-sectional area it decelerates. The location of the minimum cross-sectional
area depends on the pitches and could be at ¢ = 90° or at another angle. For
in-line banks in flow it is always located at ¢ = 90°. The acceleration of flow
effectively decreases the thickness of the boundary layer, increasing heat trans-
fer; however, transverse velocity fluctuations which enhance turbulent heat
transfer are reduced. Vortices generated in intertubular spacing and in the
boundary layer spread under the influence of negative pressure gradient. The
local heat transfer, as a result, decreases. The decrease in velocity in intertubu-
lar spacing leads to an increase in boundary layer thickness and to the separa-
tion of the boundary layer. In the region of positive pressure gradient velocity
and temperature fluctuations increase, particularly the transverse component of
velocity, which promotes local heat transfer.
Local heat transfer in tube rows of an in-line bank (Fig. 2.8) differs from
analogous heat transfer of a staggered bank as from the second row. The maxi-
mum value of a(¢) occurs at the point of flow-tube collision, from both sides of
which a boundary layer develops and a(¢) decreases. Local heat transfer inten-
sifies in the recirculation zone.
Observe how local heat transfer changes on a tube surface in a staggered
bank by rows at medium Re, range (Fig. 2.9). In the first row of the bank
(curve /) local heat transfer from the whole surface of the tube is less than in
the second and inner rows. The growth of flow turbulence from the first to
inner rows is the main factor for increasing heat transfer of the tube.
As Figs. 2.8 and 2.9 illustrate, for flow through the first rows of banks,
turbulence caused by the first and second rows plays a principal role, and
the difference in heat transfer of the third and fourth rows lies within the
bounds of measurement errors. In the inner rows of in-line and staggered banks
the character of heat transfer is different, particularly in the frontal side of
tubes.
It is important to note the specific features of heat transfer in inner rows of
in-line and staggered banks at different Reynolds numbers. In the region g ~
90-110° (Fig. 2.9, curve 3), heat transfer increases as a result of the flow
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 41

0 30 60 0 20 {0 »
Figure 2.8. Variation of local heat transfer in a tube in the first (1), second (2), and inner (3) rows
of an in-line bank 2.0 x 2.0 at Rep = 1.4 x 104 [23].

Figure 2.9. Variation of local heat transfer in a tube in the first (1), second (2), and inner (3) rows
of astaggered bank 2.0 x 2.0 at Rey = 1.4 x 10* [23].
42 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

transition from laminar to turbulent, and at ¢ ~ 150°, the increase is due to the
separation of the turbulent boundary layer.
Local heat transfer in an inner row of in-line banks (Fig. 2.8, curve 3)
varies slightly. The maximum value of a/a@ occurs at the point of collision of
flow and tube, after which o/@ decreases. In the region of boundary layer
transition from laminar to turbulent (¢ ~ 100-115°) and separation of the
turbulent boundary layer a/& increases. The relation between the maximum and
minimum values of a/a@ in in-line banks amounts to about 2.5, which indicates
stronger nonuniformity of thermal loading of the tube in the medium range of
Reynolds numbers.
As conducted experiments show, for a non-gradient flow of the laminar
boundary layer, the turbulence of the external flow does not give rise to an
increase of the local heat transfer coefficient. But for gradient flow, even slight
turbulence (1-2%) of the external flow leads to an increase of local heat trans-
fer by 40-60%.

2.4. LOCAL HEAT TRANSFER IN THE HIGH RANGE OF RE

The critical Reynolds number for a single tube is Re,, = ud/v, ~ 2 X 10°. At
this Re, value, flow regime transition from laminar to turbulent takes place. At
Re; ~ 2 X 10°,the laminar boundary layer separates from the tube at an angle
gy < 90°. When the turbulent boundary layer appears, separation shifts in the
direction of flow on the surface of the tube to g = 120°, leading to a sharp
decrease of the pressure-resistance coefficient of the tube in a bank c,, = 2P/
Fp’. At values higher than 2 x 10°, the region of high Re, begins. It is
pertinent to mention that the Value Re,, ~ 2 x 10° is not rigid, particularly for
tube banks in flow. It depends on flow turbulence, pressure gradient F = (v,/
po) (dp/dx) in the bank, the distance between flow-tube collision and bound-
ary layer separation points, and on the surface of the tube. In tube banks these
characteristics strongly depend on the pitches a X b. Therefore, Re,, can vary
within very wide bounds, i.e., from 10* to 5 x 10°. The value Re,, ~ 10° is
caused by high turbulence flow observed in medium-compact banks with small
diameter tubes in flow, and Re,, ~ 5 x 10° pertains to compact banks with
large diameter tubes, in the spacing of which transformation of turbulent flow
of the liquid into laminar flow occurs. The regime of local heat transfer at
Re, S$ 2 X 10° is called supercritical; however, there are other terms. For
example, in work [34] the regime of heat transfer up to Re; < 2 Xx 10° is called
subcritical in accordance with the pressure resistance coefficient c,, which is
constant for the tube bank under consideration in the Re, range from 10* to 2 x
10°. The range of Re, numbers, corresponding to the interval from where the
point c,, begins to decrease to its minimum value, is called critical, and to the
interval from the minimum c,, up to approximately Re; = 2 X 10° is called
supercritical, and when c,, becomes asymptotic to the constant value c, =
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 43

f(Re,) at Rep > 2 X 10° it is called transcritical. Some authors pay great
attention to individual characteristic phenomenon during the transient flow re-
gime. In so doing, the heat transfer zone in tube banks at Re, > 2 x 10° is left
without being investigated. The complexity of these investigations lie in the
difficulty of obtaining the range of Re; values from 2 x 10° and higher because
of large hydraulic resistance in tube banks. Furthermore, the influence of
pitches a x b on local and mean heat transfer is unknown in this range of Re,
Heat transfer in the rows and in the transverse row of tubes, pressure resistance
in the rows, and total hydraulic resistance of tube banks must be determined as
a function of pitches and Strouhal number. These data will stimulate the study
of heat transfer mechanism at extremely high flow velocities and have great
practical value for the development of transport power engineering.
Let us consider the particular features of local heat transfer at high Re; (up
to 2 x 10°) as a function of pitches a x b in the first and inner rows of in-line
and staggered tube banks in accordance with works [36-38]. In Fig. 2.10, data
are presented on local heat transfer of a tube in the first row of in-line banks as
a function of a x b variation from 1.25 to 2.0 and of Reynolds numbers. These
data allow the determination of the effect of transverse and longitudinal pitches

=e
pp Wisi 2200

OR OOO PT
Figure 2.10. Local heat transfer in the first row of in-line tube banks. a: /) Rey = 1.4 x 10° 5a =
8810; 2) Rey =4.4 x 10°, & = 17770; 3) Rey= 1.3 x 10°, & = 39400; b: 1) Rey= 1.9 x 10°,
& = 14100; 2) Rep = 3.7 X 10°, & = 22600; 3) Rey = 1.4 x 10°, & = 45700.
44 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

on local heat transfer at high Re, As can be seen, at Rey < 2 X 10°, the
laminar boundary layer separates from the front side of the tube irrespective of
a x b. Separation point lies at ¢ ~ 90° (Fig. 2.10, curve /). Heat transfer in
the rear side of the tube is considerably higher than in the frontal side of it. The
maximum heat transfer is at g = 180°.
With increasing Re, the character of local heat transfer changes. The value
of a/@ in the frontal side of the tube decreases, transition from laminar bound-
ary layer to turbulent takes place, and separation of the turbulent boundary
layer occurs at y = 130-140°. During the transition from laminar to turbulent
boundary layer (Fig. 2.10, curve 3), the heat transfer intensifies, reaching
maximum at y ~ 110°. At high Re, pressure gradient along the flow affects
the distribution of a/@. In a bank with = 1.25. (Fig. 2.10a) the longitudinal
pressure gradient is considerably larger than in a bank with a = 2.0.
A decrease of the longitudinal pitch b (Fig. 2.10b, bank 2.0 x 1.25) leads
to a more uniform distribution of local heat transfer in the rear side of the tube.
In the case of transition of flow regime in the boundary layer from laminar to
turbulent at Re; up to 2 x 10°, the extremum value of a/@ at ¢ ~ 110° is the
same as at g = 180°.
With a further decrease of longitudinal pitch, heat transfer in the frontal
side of the tube decreases noticeably, and in the rear side it increases. Nonuni-
formity of tube thermal loading in banks 1.20 x 1.09 and 1.4 x 1.12 in-
creases, compared with a bank 2.0 x 1.25. In the case of very high Re,, for
example 10° — 2 x 10°, the maximum heat transfer in the first row of in-line
banks is located at ¢ ~ 107°. Heat transfer at this point is 3 times higher in
comparison with the frontal critical point and 4 times higher than at the point of
flow transition from laminar to turbulent. For every flow regime in the bound-
ary layer there is a specific heat transfer intensity. The big difference in heat
transfer for laminar, transient, and turbulent boundary layers is caused by the
effect of flow turbulence in intertubular spacing and by the pressure gradient.
Consider the features of heat transfer in an inner row of the tubes of the
same banks as a function of Re, in subcritical and supercritical flow regimes. In
the subcritical regime (Fig. 2.11), at the point of flow-tube collision, a maxi-
mum heat transfer is observed which shifts to higher values of y with decreas-
ing pitch a and lies in the range g = 35-55°. With increasing Re, up to
(3-6) x 10° the flow regime around tube banks changes considerably. The
maximum value of local heat transfer of all banks, irrespective of pitches a x
b, occurs at maximum velocity. If, in the first row, the transition from the
laminar flow regime in the boundary layer to turbulent occurs at Re, > 2 X
10°, then in an inner row this takes place at Re, < 10° (Fig. 2.11a, b, curve J).
This can be explained by the effect of flow turbulence which is higher in a deep
row of a tube bank than in the first row. Since both transverse and longitudinal
bank pitches were varied in the investigations, it was possible to observe how
the maximum value of the heat transfer coefficient «,,,, changes relative to the
mean value @ as a function of these pitches. It was found that in a bank 2.0 x
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 45

a 7

; a2 68
e129 X20

| a V ‘
— iM IV row
10 os. ut. = <= 4 3 *~
- ° ‘ ny ~

de ene)
12208 , i.
i |

1.0 ia Se eae

06 ~ 200 x125
|
06 Mere
0 30 60
Figure 2.11. Local heat transfer in an inner row of in-line tube banks. a: /) Res = 9.6 x 10%
& = 13170; 2) Reg =4.6g10°, & = 39515;3)Rep = 1.25 x 10°, & = 75160; b: 1) Rey= 8.7 x
10*, & = 13210; 2) Re = 5.2 x 10°, & = 48020; 3) Rey = 1.1 x 10°, & = 71810.

1.25 Qmax/Qmin = 1.1, ina bank 1.25 x 2.0 it reaches 1.3, and with increasing
pitches a and b to 2.0, the ratio a,,,/Qmin Tises to 1.35. Consequently, in the
supercritical flow regime around bank tubes when spacing between tubes is
decreased, i.e., the pressure gradient increase, flow turbulence decays and the
pressure gradient becomes the dominant factor.
The behavior of the local heat transfer and the distribution character of a/&
on the surface of the tube in an inner row of compact in-line banks 1.20 x 1.09
and 1.4 x 1.12 at high Re; (exceeding 2 x 10°) is presented in Fig. 2.11. The
extremum of a/& lies between the values 0.9 and 1.3.
To analyze local heat transfer on the surface of a tube, one must know the
zones occupied by laminar, transient, and turbulent flow regimes in the bound-
ary layer. Also, the length of the recirculating flow zone on the tube must be
known. These data can then be compared with the results of theoretical calcula-
tions. Thus, it will be possible to reveal those spots on the tube in which heat
transfer must be intensified, which is vital when designing dependable and
highly effective heat exchangers.
The magnitude of these zones and their variation as a function of Re, be-
come distinctly apparent when they are compared with data on local heat trans-
fer at medium and high Re, ranges (Fig. 2.12). With increasing Re,, the lengths
of the zones with laminar (x,), transient (x;g), and turbulent (x;) flow charac-
ters change significantly. Flow collision with the tube in the first and inner deep
rows in staggered banks occurs at gy= 0°. At Re, < 2 X 10°, maximum heat
46 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

0 JO 60 90 120 150 Y

Figure 2.12. Variation of lengths of separate zones of the boundary layer as a function of Reynolds
number in an inner row of a staggered bank 2.0 x 2.0: 1) Reg = 1.05 10°, @ = 13530; 2) Re =
1.13 x 10°, a = 80256.

transfer is observed at the point of flow collision with the tube (Fig. 2.12, curve
1). With increasing Re, the maximum shifts to the region of minimum cross
section of the bank (curve 2), but decreases considerably in the frontal side of
the tube. Transition from the laminar flow regime in the boundary layer to the
turbulent shifts appreciably in the direction of the frontal critical point. The
magnitude of flow recirculation zone varies little with changing Re,
On the basis of the study of the distribution of the local heat transfer
coefficient in in-line and staggered tube banks of symmetrical and nonsymmet-
rical arrangements in a wide range of varying relative pitches a x b at medium
and high Re,, the dynamics of the characteristic points of flow regime change in
the boundary layer were determined. In all tube banks, with increasing Re,, the
transition from a laminar flow regime in the boundary layer to a turbulent flow
occurs at considerably smaller angles ¢ which is caused by increasing flow
turbulence. The transition in in-line banks commences at 7p ~ 90° at medium
range Re, and when Re;= 10° is reached it shifts to gz ~ 60°. The develop-
ment of the turbulent boundary layer begins at y; ~ 120°, and in the Re; range
from 2 x 10*to4 x 10° it changes only slightly. In the Re, range from 4 Xx 10°
to 10° the angle y; shifts to 83-113°. Separation point of the turbulent boundary
layer lies at ¢ ~ 150° on the average, and only after Re, reaches 4 x 10° does
it become dependent on the pitches of the banks. In banks of lower compactness
(a xX b = 2.0 X 2.0) with a large transverse relative pitch, separation com-
mences at g¢ ~ 130°. In compact banks or in banks with a low transverse pitch
(a X b = 1.25 X 2.0) separation is delayed to ¢ ~ 170°.
In staggered banks, the flow pattern changes substantially, which reflects
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 47

on the location of the characteristics points on the surface of the tube. With the
increase of Re, from 2 x 10*to2 x 1-° the transition point at ~7p 90° shifts to
Yrr ~ 20°. However, in staggered banks with small transverse and large longi-
tudinal pitches, for example ata x b = 1.25 x 2.0, the transition point varies
approximately up to Re, = (8-9) x 10°. This can be explained by the fact that
at small transverse relative pitches the velocity of the incoming flow increases
significantly and its fluctuations decay, and after leaving the region of minimum
cross-sectional area the velocity decreases in the enlarged area and the fluctua-
tions also decay until it reaches the following row. When Re; = 10° is reached,
the velocity increases substantially and the generated fluctuations lead, as in
other tube banks, to an early transition to turbulent flow regime in the boundary
layer. The point of boundary layer separation depends on the relative pitches
and lies within the bounds ¢ = 150-170°.
Based on the obtained experimental data, the lengths of laminar, transient,
and turbulent zones of the boundary layer on the surface of the tube were
determined; also obtained was the value of the recirculation zone in in-line and
staggered banks. From the standpoint of compactness, staggered tube banks are
of greater practical interest. With increasing Re, the laminar zone length in
staggered tube banks decreases considerably. The relative length x,/d ~ 0.8-
0.9 at Re, = 2 X 10*, and x,/d ~ 0.2 or less at Re, = 2 X 10°. As a result, the
length of the transient zone increases substantially. At Re = 2 x 10* the
relative length x,p/d ~ 0.2, and at 2x 10° it is equal to 0.5-0.6 or more. The
relative length of the turbulent zone of the boundary layer in the same Re, range
rises from 0.25-0.3 to 0.6. Banks with small transverse and large longitudinal
pitches are an exception. In such banks the relative length of the turbulent zone
in the boundary layer x,/d has a large value for the whole investigated Re,
numbers range.
The recirculation zone length x z in the range Rey = 2 X 10*-2 x 10° is
small. In staggered and in-line tube banks xp = (0.1-0.3)d. Knowing these
characteristics on the surface of the tube, the flow can be controlled. By chang-
ing the level of flow turbulence, the distance between tubes, and their arrange-
ments in banks, the required flow in separate regions on the tube surface can be
obtained. Also, heat transfer in intertubular spacing can be intensified and the
power expended to move the thermal carrier optimized.
The resistance of a tube in a row depends on the length of the recirculation
zone xz. As noted earlier, the magnitude of x, is determined by the location of
boundary layer separation from the tube. To design heat exchangers with opti-
mum energy parameters, the magnitude of the recirculation zone in various tube
banks within the maximum possible range of Reynolds numbers must be known.
Such data can be used to compare flow separation computing methods from
which the separation point and recirculation flow zones can be determined. Great
attention is presently being paid to developing and mastering computational tech-
niques for complex flows with a separation and recirculation zone. Therefore,
based on obtained experimental data on the distribution of the local heat transfer
48 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

coefficient in tube banks, and also based on flow visualization and photography,
the points of the boundary layer separation in staggered and in-line tube banks are
determined for pitches which are of practical interest, and for Re, variation from
1 to 2 x 10°. As Fig. 2.13 shows, the separation point on the tube surface
depends strongly on Re, The more compact the tube bank, the larger the Re,
numbers range in which nonseparating flow is preserved. In a staggered bank
2.0 x 2.0 the flow separates at Re, = 12, in a bank 1.5 xX 1.5 it separates at
Re, = 40, and in a bank 1.25 x 1.25 separation commences only at Re, = 90.
This agrees qualitatively with a single tube in flow, where, with increasing pack-
ing coefficient, the nonseparation flow zone increases. With the further increase
in Re,, separation shifts towards the frontal critical point, and at Re; = 107: g=
110-125°, depending on the bank pitches. The influence of tube bank compact-
ness on the separation point is preserved even at these values of Re, Separation
point in compact banks lies at higher values of the angle ¢. With a further
increase in Re,, transition of laminar flow in the boundary layer to turbulent takes
place, and separation shifts significantly towards the frontal critical point, thereby
decreasing the zone of flow recirculation.
A similar picture is observed for in-line banks in flow. Compared with
staggered banks, nonseparating flow in in-line banks is preserved at much
lower values of Re,;. From the lower section of Fig. 2.13 it follows that separa-
tionin compact in-line banks 1.25 xX 1.25 occurs at Re = 9, in banks 1.5 x
1.5 separation commences at Re = 7, and in banks 2.0 xX 2.0 nonseparating
flow continues up to Re, = 4. Beginning from Re, = 300 to Re, = 10* the
separation point in a wide range of varying relative pitches is dependent only on
+O +x4+0+ x yaa
< DLL
+- 150% {50 oO Eta ReonTp
x- 20* 20 (oe
re Sea

ie
ace pet tr

o°%o,90
gorthiciae
aisles

te te rit KT —

(0 0’ 0 10 10 0
6
10 Re
Figure 2.13. Dynamics of boundary layer separation points in staggered and in-line tube banks as
a function of Rey.
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 49

0 20 40 60 Y

Figure 2.14. Comparison of data on local heat transfer in the frontal side of the tube at low and
medium range Re: J) single tube, calculations according to [4]; 2-5) first row of in-line and
staggered banks 2.0 x 2.0 (Rep: 2) 6.1 x 10*, 3) 10°, 4) 1.5 x 10°, 5)3.8 x 10°); 6, 7)first row
of an in-line bank 1.25 x 2.0 (Reg: 6) 5.5 x 10*, 7) 1.5 x 10°).

the magnitude of Re, This indicates that in symmetrical in-line banks the main
flow takes place between the rows, irrespective of their geometric characteris-
tics. At Re, = 10° the angle ¢ reaches 110°. With in creasing the Re,, the
separation point becomes dependent on the relative pitches, and Re, has only an
insignificant effect.
The character of heat transfer of the tube changes, mainly when the veloc-
ity in the banks varies strongly. This is evident from Fig. 2.14 in which curve /
corresponds to the heat transfer of a single tube according to the analytical data
of work [39]. Curves 2-5, which correspond to experimental data on heat
transfer of a tube in the first row of in-line and staggered banks with a = 2.0,
indicate that the local heat transfer decreases gradually, starting from the frontal
critical point. With decreasing a to 1.25, the longitudinal velocity gradient and
heat transfer on the frontal side of the tube surface increase considerably (Fig.
2.14, curve 6, 7). (During the comparison of the data the incoming velocity up
entered into the Reynolds number.)
Data of local heat transfer studies in an inner row of staggered banks are in
good agreement with the results of the works of other authors [40-43] at low
and medium Re, ranges. The character of local heat transfer in the frontal side
of the tube little depends on Re; and the pitches a x b. Intensity of heat transfer
50 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

0 0 4 0G fh Sle 70 GU ace
Figure 2.15. Comparison of data on local heat transfer of staggered banks at medium and high Re,
ranges: 1) 1.3 X 1.25, Rep=4.3 x 104 [41];2)1.5 x 2.0, Rep= 2.5 x 10* [42];3)1.5 x 1.5,
Rey = 10° [37, 38].

in the rear side of the tube is very sensitive to changes in a X b and Re,
particularly in the transient flow zone. Heat transfer of symmetrical banks in-
tensifies during transition more than heat transfer of nonsymmetrical banks.
Data on local heat transfer of staggered banks are compared in Fig. 2.15.
These data were obtained by various authors at higher Re, and other values of
pitches a xX Db. It is apparent that the change in transverse and longitudinal
pitches of staggered banks in the range from 1.3 to 2.0 has little influence on
the character of heat transfer. The authors of works [36, 42] obtained the same
character of heat transfer. The extremum heat transfer lies at ¢ ~ 110-125°,
which corresponds to the end of transition and beginning of turbulent boundary
layer development. Separation point of the turbulent boundary layer is situated
at ¢ = 140-155°.
From the obtained data on the distribution of local heat transfer across the
tube surface in the first rows of banks, local heat flowrate on the heat transfer
surfaces can be determined in various flow regimes, and a heat exchanger with
optimum configuration can be designed.

2.5. LOCAL HEAT TRANSFER IN CHARACTERISTIC

REGIONS OF TUBES IN IN-LINE AND STAGGERED BANKS

For the improvement of heat exchangers, the specific features of flow and heat
transfer across the whole surface of the tube in a bank must be well known.
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 51

Therefore, generalization of heat transfer in characteristic regions of a tube in in-

line and staggered banks was conducted in a wide range of Re, and pitches a x b.
Such data are needed to control the heat transfer process and compare it with the
theoretical solutions in the case of flows with boundary layer separation.
The data presented in Table 2.1 were obtained during the generalization of
the results of local heat transfer investigation at the point of flow-tube collision
on the basis of the function Nu, = cRe}' Pr; (Pr,/Pr,,)’.
Consider the features of heat transfer in recirculation zones of in-line and
staggered tube banks. In in-line banks, two recirculation zones are present; the
first, from the collision point of flow with the tube to the frontal critical point,
and the second, from the separation point to the rear critical point. For a change
of pitches a, b from 1.25 to 2.0 and Re; from 20 to 2000 in in-line banks, heat
transfer in the first recirculation zone can be expressed by the function

Nu,, p= 1.07 (a— 1)** Re?-* Pr?° (Pr, /Pr,,)°?° (2.16)

In the second recirculation zone of in-line and staggered banks, heat

transfer has a more complicated character. The intensity of heat transfer is
determined by the velocity of recirculation flow, which is developed in the
vicinity of the tube surface. This is reflected on the exponent m for Re, In an
inner row of in-line banks this exponent varies slightly in the indicated range
of pitches. For banks of low compactness in the region of separation, the
exponent m is greater. Consequently, the circulation of the liquid is more
intensive. In staggered banks, there is one recirculation zone in the rear side
of the tube.
When the results of the investigations of heat transfer in the second recircu-
lation zone of in-line banks (Fig. 2.16) in the given region of pitch variation are
generalized, the following function is derived

Nuprg 30.587 (a — Wt Ree Prp* (Pr7/Pr,,)9 Ca?)

which is valid in the Re, range from 20 to 1000. From Re,= 1000 the flow in
this zone intensifies. The generalized experimental data at Re, > 10° are pre-
sented in Table 2.2.

Table 2.1. Data for the calculation of local heat

transfer at the point of flow-tube collision in banks
1.25 x 2.0
a ee

Bank c m Rey
DOP a a eee eee

In-line 0.777 0.52 20-2000

Staggered i iil 0.51 20-2000
ee
52 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

oI hd,
er )
Spe liga)|
ey

10° Re
f
Figure 2.16. Heat transfer in recirculation zone of in-line banks of tubes. K; = Kp/(a — ie eo
Kp = Nu; cllaeaaey (Pr//Pr,,)°?> . R is the length of the recirculation zone from the separation point
to the rear critical point.

The exponent m in staggered banks depends considerably on both the

pitches a x b and the distance from the point of collision of the flow with the
tube. From the site of boundary layer growth in the region of the frontal critical
point, for all banks of varying pitches from 1.25 to 2.0, the exponent m is
higher than in the case of in-line banks. Consequently, fluid circulation in the
vortex region is more intensive. In compact banks 1.25 xX 1.25, moving down-
stream from the rear critical point (g = 180°) leads to an increase of m, and in
the region of separation it somewhat decreases. As for in-line banks, the situa-
tion is reversed. This indicates that the main fluid flow in in-line banks is

Table 2.2. Data for the calculation of heat transfer

in the second recirculation zone of in-line
and staggered tube banks 1.25 x 2.0.

Bank c m Rey

In-line 0.114 0.7 10-4 x "10°

0.024 0.82 Axa es? O10"
Staggered 0.196 0.62 300-2000
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 53

concentrated in the channels between tube rows; therefore, the flow is not
capable of generating intensive fluid circulation in the vortex region. In stag-
gered banks, the flow in the intertubular spacing intensifies the fluid cir-
culation in the vortex region. With increasing pitches m decreases, which is
an indication of a close correlation between the intensity of circulation in a
vortex and bank compactness. Therefore, if there is a need to increase heat
transfer in a vortex region, it is expedient to employ compact staggered tube
banks.
The generalization of data on heat transfer of separate zones in in-line
banks shows that at low and high values of Re; transition from laminar flow in
the boundary layer to turbulent flow is delayed. In staggered banks the transi-
tion occurs at low Re,
If separation of the boundary layer appears followed by recirculation in the
back side of the tube, then the impingement of fluid on the tube takes place at
the rear critical point. The study of heat transfer intensity at this point is of
scientific interest, and the theoretical calculations will entail great difficulties.
Generalization of the results on local heat transfer at the rear critical point on a
tube in in-line banks yields the following relationship.

Nuy, ¢~ 130° = 0.646 (a — 1)°-38 Re? Pr236 (Pr,/Pr,,)°25 (2.18)

in the range of variation of pitches a, b from 1.25 to 2.0 and Re, from 20 to
2000.
For staggered banks in the same variation range for pitches c = 0.22, m =
0.65 at the rear critical point, and heat transfer is independent of the pitches of
the bank.
The aforementioned formulas for the calculation of heat transfer at charac-
teristic regions along the tube perimeter can, hopefully, aid further endeavors in
studying the specific features of heat transfer in complex flows and be useful in
designing more effective heat exchangers.

2.6. TEMPERATURE PULSATIONS ON TUBE SURFACES

Flow relative to the tubes in a heat exchanger can be across, along the axis, or
at an angle. Banks of tubes in crossflow present the most resistance to flow.
Heat transfer in this case is greater than in banks of tubes with axial flow or
flow at an angle.
The zones of flow collision with the tube, separation, and recirculation
leads to vortex-motion, time-dependent instability, causing temperature pulsa-
54 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

tions of the wall. At low temperature pulsations of the wall, thermal stress is
not detected in the metal. Higher pulsations of wall temperature cause local
thermal stresses which decrease the service life of heat transfer surfaces. If the
thermal conductivity of the tube material is high, temperature pulsations on the
surface are somewhat lower, since the temperature becomes more even along
the perimeter due to heat flow.
The amplitude of wall temperature pulsations at low and high Re,, as mea-
surements indicate, depends on flow velocity and local temperature head. Fig.
2.17 shows the curves of wall temperature pulsations at the frontal critical point
and in the region of flow separation at ¢ = 125° on a tube in a compact
staggered bank 1.25 xX 1.25 at low Re, At the frontal critical point, i.e., at the
site of flow collision with a tube in the first and inner rows of the bank, the
pulsation of the mean value of wall temperature amounts to 0.2-0.35°C. This
indicates that, at low Re, the incoming stream near the surface induces small
pulsations of wall temperature. They increase to 1 °C and more in the region of
flow separation. Measurements of wall temperature pulsations were mainly
conducted at a constant mean temperature head At = 10°C; however, measures
at At = 20° and 30° show that the pulsations of wall temperature increase with
increasing temperature head. At high Re, the pulsations of tube wall tempera-
ture in the separation zone in the first and inner rows of a bank reach 4°C.
However, they have a different character in inner rows due to the presence of
small vortices in the flow structure in the separation region.
To reduce nonuniformity of thermal loading, it is preferable for the heat
exchanger to operate at low Re, of the order 1-2000, and at lower temperature
heads. To decrease the resistance of the tubes and flow instability in the recircu-
lation zone, the tubes must be either well-streamlined or have baffles at 9 =
180°. The latter decrease the pulsations of flow temperature in the recirculation
zone, thus decreasing the pulsations of wall temperature and increasing heat
transfer surface area.
Should there be a need for the heat exchanger to operate in regions of

ig) 50 45
Figure 2.17. Pulsations of wall temperature in an inner row of a staggered bank 1.25 x 1.25 at
Rey = 1600 and Pr; = 1820.
 LOCAL HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 55

moderate and high Re, it is expedient to use a developed heat transfer surface in
the form of finned tubes. Based on measured temperature distribution at the tip
and root of the fin, local intensifiers are fabricated by cutting and twisting the
fins in the region where the heat transfer coefficient has minimum values. By
inducing turbulent flow at a certain site, the local heat transfer coefficient in-
creases. With this method, more uniform thermal loading can be achieved for a
small loss of hydraulic resistance of the bank. Heat transfer in such banks of
tubes are 40-60% higher than heat transfer of smooth tubes.
Some reduction of nonuniformity of thermal loading in cases of smooth
tube banks can be achieved by changing the transverse and longitudinal pitches,
and also on the basis of the local heat transfer coefficient determined on the tube
surface at various values of Re.
In high-temperature flows, the wall temperature pulsates at characteristic
sites of the tube where the flow is unsteady. Thermal stresses are then created
and erosion takes place which, eventually, decreases the durability and reliabil-
ity of heat exchangers.
 CHAPTER

THREE
MEAN HEAT TRANSFER FROM BANKS
OF SMOOTH TUBES

3.1. HEAT TRANSFER FROM A TUBE IN A BANK

Banks of tubes in crossflow are badly streamlined bodies. Their resistance at a

given flow velocity of the thermal carrier is higher compared with well-
streamlined bodies, such as a plate in parallel flow, flow in a tube, or a channel
having a cross section of any form. Usually, heat transfer in banks of tubes is
significantly greater than heat transfer in streamlined bodies. This is the main
reason behind the wide use of tube banks. The determinant factor in choosing
the form of the tube is the correlation between the energy expended on the
movement of the thermal carrier and the amount of heat transferred. Therefore,
the transferred heat and the power consumed are extremely important parame-
ters.
The velocity of the thermal carrier is but only a value characterizing the
regime of heat-exchanger operation, and one of the components for determining
the consumed energy. The amount of heat transferred is proportional to the heat
transfer coefficient, which depends on the velocity of the thermal carrier to a
power not usually exceeding unity. During transition from one flow regime to
another, the exponent of the velocity becomes somewhat larger than unity only
in a narrow region of Re, The energy expended is related to the velocity of the
thermal carrier raised to the third power. Heat transfer in banks of tubes in
crossflow is higher compared with parallel flow. Therefore, a comparison of
heat exchangers with streamlined and nonstreamlined elements shows that at
Re = idem, as far as the amount of transferred heat is concerned, they can be

56
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 57

equivalent only if the velocity of the thermal carrier in the heat exchanger with
streamlined elements is significantly larger. The advantages of using non-
streamlined elements, in particular, banks of tubes in crossflow, is demon-
strated by the fact that a more compact heat exchanger can be constructed from
them. In this respect, data about heat transfer in banks of tubes in a wide range
of thermal-carrier velocity variations are extremely vital.
Various thermal carriers are employed in heat transfer devices: air, fuel
combustion products, water, and other fluids with different viscosities which
change considerably during the heat transfer process. The heat transfer coeffi-
cient in a flow of air or gas is always lower than in a flow of water or other
viscous fluids. To achieve equality of heat transfer coefficients from the side of
air and liquid, the velocity of the gas has to be increased. Gas velocity in heat
exchangers considerably exceeds liquid velocity. The study of heat transfer of a
gas and various liquids in a wide range of velocity variation is of practical
interest.
When designing heat exchangers, one should be able to determine the in-
tensity of heat transfer in tube banks with the aim of exploring their optimum
variants.
As investigations show, heat transfer of tubes in the first row is lower than
in inner rows. But, at low Re, when the flow hitting the first row is laminar, it
becomes slightly turbulent, and heat transfer in the following banks somewhat
increases. For a specific tube arrangement in in-line banks, heat transfer of an
inner row at low Re; is even less than heat transfer in the first row in view of
shadowing. With increasing Re,, the flow, after passing the first and second
rows of tubes, becomes turbulent and heat transfer of the inner row in-
creases. Starting with the third row of a bank, heat transfer in the follow-
ing rows stabilizes and practically coincides with the heat transfer of the third
row. Comparison of heat transfer of tubes of the first and inner rows gives an
indication of the influence of turbulent flow on the intensity of heat transfer.
Heat transfer of tubes in a inner row, as a rule, increases with decreasing
longitudinal pitch of the bank. In-line banks of tubes with small longitudinal
pitch are an exception, where a decrease in pitch can lead to a decrease in heat
transfer. This is related to the decrease in intensity of recirculation between
neighboring tubes with decreasing longitudinal pitch at low and moderate Re,
in particular.
Depending on the longitudinal pitch and flow turbulence, heat transfer of
tubes in an inner row can increase by 30-100% compared with heat transfer of
the first row. In the majority of cases, heat transfer of tubes in the second row is
10-30% lower than in an inner row.
Heat transfer of a tube in a bank depends mainly on the velocity of the
thermal carrier, tube arrangement, properties of the incoming fluid, thermal
loading, and the direction of heat flow. In a dimensionless form this depen-
dence is as follows
58 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Nu=f (Re, Pree ee ene

key Ne
ee
Chics,
eee
d
“) (3.1)

Consequently, heat transfer of tubes in a bank must be studied proceeding

from the functional dependence (3.1).
For the generalization of experimental data, empirical relationships be-
tween the similarity variables are often used

Nu = c Re” Pr”

where c and m depend, in a specific range of Reynolds numbers, on the ar-

rangement of banks. The magnitude of the exponent n is stipulated by the effect
of the physical properties of the fluid on heat transfer. The constants c, m, and n
are determined on the basis of experimental data, generalization, and analysis
of heat transfer characteristics.
During the heat transfer process, the temperature of the fluid flowing
around various rows of tubes in a bank varies noticeably and, consequently, its
physical parameters also change. The question of taking into consideration the
effect of the physical properties of the fluid on heat transfer is closely related to
the consideration of the effect of the variation of these properties when the
temperature changes in the boundary layer. This is related to the choice of the
so-called determining temperature at which the values of the physical parame-
ters in the similarity variables are evaluated. The analysis of data on heat trans-
fer of tube banks in flows of various liquids and air showed that the temperature
of the incoming fluid stream t, must be taken as the determining temperature.
The diameter of the tube is taken as a characteristic dimension, and the velocity
of the fluid in the minimum cross-sectional area between tubes as the design
velocity. When comparing data obtained for a single tube in an infinite flow,
tubes in a duct, and tubes in the first row of a bank, a more specific parameter
should be taken as a determining velocity. It is expedient, therefore, to intro-
duce the concept of average velocity. This is the integral value of the velocity
along the whole length of flow path around the tube:

=> ihit(x) dx (3.2)

where / is the length of the arc along the tube perimeter.

Analysis of experimental data showed that the maximum velocity, which
differs slightly from the average, can be used to calculate heat transfer in tube
banks. Compact tube banks, in which the average velocity corresponds to the
incoming velocity, is an exception.
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 59

3.2. EFFECT OF THE PHYSICAL PROPERTIES

OF THE THERMAL CARRIER AND THE TEMPERATURE
HEAD ON HEAT TRANSFER

The majority of works dedicated to the investigation of heat transfer of tube

banks is conducted for air and at relatively low temperatures; therefore, the
effect of the physical properties of the fluid on heat transfer is difficult to
determine from the obtained data.
For the determination of the effect of Pr, on heat transfer, its dependence on
flow velocity in the form Nu, = f(Re,) was first defined. Experiments were
conducted at a constant temperature of the incoming flow and at a low tempera-
ture head, which was maintained constant while the flow velocity was varied.
The exponent m of Re, was determined for one of the banks in flows of air,
water, aviation, and transformer oils. In Fig. 3.1 data are presented on mean
heat transfer at low velocities of the thermal carrier (aviation oil), and at me-
dium and high velocities (water and transformer oil). It is difficult to obtain Re,
values more than 2 x 10° in flows of highly viscous fluids; therefore, water
was used (Pr;= 3-7) which allowed to obtain Re, = 2 x 10°. To start with, in
order to compare the data in a dimensionless form, investigations in the stream
of water were carried out in the same velocity range as for the streams of
viscous fluids. As is evident from the figure, the heat transfer coefficient is
proportional to the flow velocity raised to a power from 0.33 to 0.8. The

Ee RN el | AEG
Ee ea eee
cs
ee

46 810° 2 i bade Zi 4 6 810 vA u

Figure 3.1. Dependence of the mean heat transfer coefficient on flow velocity in an in-line bank
2.0 x 2.0: 1) aviation oil; 2) transformer oil; 3) water.
60 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

exponent of the velocity, depending on the pitches and configuration of the tube
banks, reaches unity in the region of high Re,
Following the determination of the exponent of Re, one can proceed to
determine the effect of Pr, on the heat transfer of tube banks. Special attention
was paid to this, since obtaining experimental data in a wide range of Re, and
Pr, numbers simultaneously is technically difficult. For this, a large-scale, high-
power experimental setup is needed. Also, the accurate determination of the
influence of Pr; on heat transfer is important, since the role of that influence
rises with increasing Pr,
If the tube is in a turbulent fluid flow, laminar, transient, and turbulent
boundary layers, and also recirculation, are formed on different parts along the
perimeter. It is expected that the magnitude of the exponent of Pr, will be
characteristic for these flow regimes in the boundary layer as it was in the case
of a plate. However, the effect of the pressure gradient and the specific features
of heat transfer in recirculating flow must be taken into account.
In the case of a plate in a laminar flow regime, the value of Pr, is accounted
for by an exponent equal to 0.33. In a turbulent boundary layer this exponent
increases up to 0.43. The effect of Pr, in the transient flow region is accounted
for by the exponent 0.43, although it is preferable to use a mean value, i.e.,
0.38 [3].
When calculating the mean heat transfer for a tube in crossflow, the power
exponent of Pr, in the subcritical flow regime is found to be equal to 0.37 [4],
and in the supercritical regime it is equal to 0.4 [44].
Heat transfer in tube banks is considerably more complex than in a plate in
parallel flow. In the frontal side of the tube and up to the separation point, a
laminar boundary layer is formed for values of Re, reaching 2 X 10°. After
separation, recirculation occurs. At higher than Re, ~ 2 X 10° values, all three
forms of flow on a tube in a bank are formed in the boundary layer—laminar,
transient, and turbulent. It is, therefore, very important to determine the degree
of influence of Pr; in tube banks.
On the basis of solving the equation of the boundary layer (see Chapter 1),
the effect of Pr; on heat transfer for tube banks was determined, and the ob-
tained results were compared with the experimental data. It is established that
the exponent n of Pr; varies in a complex fashion: in some regions of the
boundary layer it is dependent on the fluid flow regime and changes within
wide bounds. Calculations and experiments show that at high Re, values, due to
the elevated level of flow turbulence, the exponent n for the Pr, number in-
creases and assumes a value of 0.4, on the average.
Based on experimental data and on the function Nu,Re;” = f(Pr,), the
mean value of the exponent n of Pr, were determined in a wide range of Pr,
variation at various flow regimes.
In fluids, Pr, varies in a wide range from 1 to 10000. Fig. 3.2a presents
experimental data on heat transfer of tube banks at low Re, (from 1 to 10°) in
flows of aviation and transformer oils, and shows that the exponent n = 0.36.
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 61

oe
@
0.

Figure 3.2. Determination of the exponent n of Pry number. a—in in-line bank 2.0 x 2.0 at low
Reg. 1) aviation oil, 2) transformer oil, K; = Nu,/Re# b—in staggered banks at moderate Reg 1)
air, 2) water, 3) transformer oil, 4) 2.6 x 1.3; 5) 1.30 x 1.30; 6) 1.95 x 1.3.

Experimental data on heat transfer of staggered tube banks in flow of air,

and various fluids at medium range Re, values are generalized in Fig. 3.2b. It is
evident that in the region of predominantly laminar and mixed flows n = 0.36.
A similar exponent is also obtained for other tube banks.
On the basis of data on heat transfer of in-line and staggered tube banks,
with various geometrical characteristics in flows of air [38] and water [44] it
was found that n = 0.4 in the region of high Reynolds numbers (Re, > 2 X
10°) (Fig. 3.3). Therefore, the exponent n can be assumed, with sufficient
accuracy for practical computations, to be equal to 0.36 in a very wide range of
Reynolds numbers (Re, = 2 X 10°); and for Re, = 2 X 10° it can be assumed
equal to 0.4.
During the determination of the exponent n, data obtained at low thermal
loadings were used. The effect of the varying temperature was excluded by
dividing by a previously chosen parameter (Bis Pioace

6 810 y Pe AGP
Figure 3.3. Determination of the exponent n for Prin tube banks at high Reg 1) air; 2) water; 3)
staggered bank 1.5 xX 1.5; 4) in-line bank 1625) x-1.25;,Ky = Nu//Ref
62 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

For the determination of the influence of the temperature head on heat

transfer, water and electrical calorimetric measurements were employed to ob-
tain the regimes of fluid heating and cooling.
When studying the effect of the temperature head on the direction of heat
flow in tube banks, thermal loading was varied in a wide range.
As measurements of temperature profiles across the boundary layer of high
viscous fluids (aviation and transformer oils) show, the temperature profile
undergoes the main deformation due to Pr, variation, and the velocity profile
undergoes much lesser deformation.
Experimental data were analyzed in the form of a dependence of the com-
plex similarity variables K, = Nu,/Re;Pr;on the parameter Pr,/Pr,,. The experi-
mentally obtained values of m and n were introduced as a function of flow
regime. By so doing, the influence of the numbers Re, and Pr, was excluded. At
low Re, the exponent m changes from 0.33 to 0.46, and at moderate and high
Re,, m varies from 0.6 to 0.8. In the region of Re; from 1 to2 x 10° n = 0.36,
and at Re, > 2 X 10° n = 0.4.
Fig. 3.4 shows some data on the influence of the temperature head and the
direction of heat flow on the mean heat transfer of tube banks. As is evident
from Fig. 3.4, the exponent of Pr,/Pr,, is equal to 0.25 in in-line banks when the
thermal carrier is heated and cooled.
The results of experimental investigations [1] of heat transfer in staggered
banks during heating and cooling of water and transformer oil are presented in

02
Figure 3.4, Influence of the temperature head and the direction of heat flow on heat transfer at low
and moderate Rey: 1) water; 2) transformer oil; 3) aviation oil; 4) in-line bank 2.0 x 2.0; 5,6)
staggered banks 1.30 x 1.30 and 1.97 x 0.99.
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 63

Fig. 3.4 in the form of the function K, = f(Pr,/Pr,,). when the physical proper-
ties of the liquid are generalized by a curve with an exponent of Pr,/Pr,, equal to
0325:

3.3. INFLUENCE OF FLOW REGIME AND TUBE

ARRANGEMENT ON HEAT TRANSFER

The mean heat transfer of a tube in a bank is mainly dependent on the velocity
of flow. Depending on flow velocity in the minimum cross-sectional area of
tube banks, one can determine the dominant laminar, mixed, and turbulent flow
regime. For every regime there is a characteristic value of the exponent m of
the Re number: for the laminar flow m = 0.4, mixed flow m = 0.6-0.65, and
turbulent flow m = 0.8. During transition of the flow regime from laminar to
mixed, m acquires the value 0.5, and from mixed to turbulent m = 0.7-0.9.
These values of m are average values for a large number of tube banks. Re-
search showed that for every bank of tubes, depending on the pitches a x b,
there is a corresponding value of m which deviates somewhat from the above-
mentioned average values. Thus, from the value of m one can assess the flow
regime and the intensity of heat transfer.
In a steady-state flow heat transfer of a tube in a bank is significantly
greater than heat transfer of a single tube. This is related to the small thickness
of the boundary layer on the surface of a tube in a bank and to the elevated
turbulence of flow in intertubular spacing.
At low values of Re,, heat transfer of a tube in the first row of a bank
practically coincides with heat transfer of a single tube and tubes in an inner
row. With increasing Re,, flow turbulence in intertubular spacing of the banks
increases and heat transfer in inner rows becomes more intensive than in the
first row. This is due to the fact that the rows of tubes in a bank act as turbu-
lence generators in the flow. In the majority of cases, heat transfer towards the
core of tube banks stabilizes as from the third or fourth row. In an inner row,
heat transfer of a tube, as a rule, increases with decreasing longitudinal pitch.
In-line banks with a small longitudinal pitch are an exception to this rule. A
decrease in pitch in this case leads to flow laminarization, weakening of circula-
tion between neighboring tubes, and a probable decrease of heat transfer.
Let us analyze heat transfer in the region of dominant laminar flow which
encompasses the values of Re, up to 10°, approximately. In the case of low Rey,
natural convection, the role of which decreases with increasing Re,, influences
the mean heat transfer of the tube in a bank. In the case of natural convection in
tube banks, the main role is played by the magnitude of the temperature head
between the hot (or cold) wall and the fluid. Depending on the magnitude of the
temperature head, the flow of the fluid could be laminar or turbulent. By regu-
lating the temperature head, heat transfer can be intensified in tube banks. In
accordance with the functional dependence Nu, = f(Gr,, Pr,), heat transfer by
64 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

natural convection can be increased not only by increasing the temperature

head, but also by changing other values in the Rayleigh number (Ra, = Gr,
Pr,). Note that for liquid thermal carriers the Gr, number is quite often replaced
by the Ra, number, since heat transfer by natural convection is strongly depen-
dent on Pr, in flows of viscous fluids at Pr, > 10.
Ra, number increases significantly in high viscosity fluids. If Pr, in gases
reaches unity, then in liquids with high viscosity Pr, > 1. For example, at t; =
10°C, Pr, = 35241 in aviation oil. Lift force in the vicinity of the tube surface
appears because of density differences of hot and cold fluid particles. Shear
stress at the wall 7, = u(du/dy),_. depends on the viscosity of the thermal
carrier and the derivative (du/dy),_,. To decrease 7,,, 4 must be decreased at a
constant value of (du/dy),_9. When working with highly viscous thermal carri-
ers, ». can be decreased by raising the temperature of the fluid. For example, if
the temperature of the wall ¢,, is increased up to the average value of 80°C, the
ratio p4,/p,, Will amount to 64.1. Such a substantial change of viscosity near the
wall surface promotes heat transfer increase by natural convection.
The value (du/dy),_) depends on the character of velocity distribution in
the vicinity of tube surface. If there are laminar and turbulent regimes in the
boundary layer during natural convection, then the value (du/dy),_, in each of
them is different. In view of the greater intensity of motion of fluid particles in
a turbulent boundary layer, the velocity profile is fuller, i.e., the velocities near
the wall (du/dy),_) are much higher than in a laminar boundary layer. There-
fore, with increasing the temperature head At = t,, — t, viscosity decreases.
This means that 7,, also decreases but (du/dy),_, on the other hand, rises on
some section of the tube as a result of turbulent boundary layer initiation.
Therefore, the final value of 7,,(x) depends on the interrelation of the values of
pw and (du/dy),_9, and at high At on the tube it will be variable.
At low Rey, heat transfer is caused by free and forced convection. If only
free convection is involved, heat transfer in a bank is determined by the func-
tion:

Nu; =Cy Raye (Pr;/Pry)? (3.3)

For mixed convection, one must know what portion of the heat is transferred by
natural convection. At Re; = 0, the thermal carrier flows in the direction of the
lift force, i.e., the angle 6 subtended by the directions of the forced flow and
the lift force is equal to 90°. At some low values of Re, the angle 8 decreases to
zero, i.e., forced convection starts to dominate. Special investigations were
conducted with tube banks 1.25 x 1.25, 1.50 x 1.50, and 2.0 x 2.0 to
determine the heat transfer at Re; = 0. Curve 4 in Fig. 3.5 corresponds to heat
transfer of a single tube [45] when heated and cooled in air, water, aviation oil,
transformer oil, and compressor oil, and can be generalized by the function

Nu, = 0.5 Ra?:5 (Pr,/Pr,,)9-25 (3.4)

 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 65

i
Figure 3.5. Heat transfer of horizontal staggered tube banks with freely flowing air, aviation, and
transformer oils. Bank: 1) 1.25 x 1.25; 2) 1.5 x 1.5; 3) 2.0 x 2.0; Ky = Nu, Pr/Pr,)~°?5

which is reduced for air to

Nu,
= 0.46 Gr? (3.5)

As is evident from Fig. 3.5, the correlation between data on heat transfer in
tube banks and curve 4 is satisfactory. In the case of large pitches a x b the
data lie above the curve, and for small pitches they lie below the curve. Appar-
ently, in compact banks at high temperature heads, when a turbulent boundary
layer and a vortex flow region are formed on the tube, bank compactness
reduces the intensity of the heated fluid and changes its character. In the case of
natural convection, the intensity of heat transfer in tube banks depends mainly
on the flow regime in the boundary layer and on the magnitude of the vortex
region.
Small flow velocities corresponding to Re; = 1-5 change the character of
heat transfer of a tube in a bank significantly [33]. At Re, =5 the exponent ny of
Ra, decreases to 0.08. With increasing Re, to 100, m) decreases to 0.024, and at
Re, = 220 n) becomes equal to zero. Therefore, beginning from Re, ~ 200,
Ra, = 1 which indicates that heat transfer in tube banks is mainly by forced
convection.
In the case of mixed convection, heat transfer increases significantly even
for a small increase of Re; Therefore, when Re; is slightly higher than unity it
is of great practical significance. In a mixed convection regime, the energy
expended on the movement of the fluid is minimal. When heat is transferred by
mixed convection, heat transferred by forced convection is first determined;
then that portion which is transferred by natural convection is found. There are
practical cases when, due to the large dimensions of heat exchangers and high
temperature heads, the velocities induced by natural convection significantly
exceed the velocity of forced flow, and vice versa. This happens, for example,
when cooling electronic equipment, in instrument manufacturing, during the
natural cooling of hot billets, during the heat treatment of components, and so
66 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

forth. In all these cases cooling is to be facilitated by both natural and mixed
convection. Both forms of convection, therefore, play an important role. When
calculating the amount of heat removed, one must know how much heat is
removed by natural convection alone, and by how much it can be increased in
the case of mixed convection.
When developing new and highly effective equipment, an important role is
played by non-stationary flow which emerge during the starting and stopping of
heat exchangers, or during transition from one stationary regime to another. In
such flow regimes, heat transferred by natural and mixed convection varies.
Large variations of temperature of heat-exchanger surfaces in the transient re-
gime is accompanied by mixed convection; therefore, it is imperative in this
case to determine the heat flow transferred by both natural and mixed convic-
tion. Generally, the heat transferred by natural or forced convection can be
determined. If the ratio between the two forms of heat transfer is known, then a
specific heat flow can be determined for each of them. According to work [46]
and others, one can make use of Gr,/Re? in which Gr, and Re, are determined
from the local value x on the tube. The parameter Gr,/Rez expresses the ratio of
the lift force, generated by density variations in the fluid, to the inertial force;
therefore, it is advisable to use it when forced convection dominates. If natural
convection is dominant, the velocity of forcibly moving fluid is small, and heat
transfer is best determined relative to the parameter Re,/Grz. This parameter
most clearly defines the perturbation caused by the slow flow. It is shown in
work [46] that the magnitude of the lift force is strongly dependent on Pr, At
constant value of Gr,/Rez, increasing Pr; leads to a significant increase in heat
transfer. When Gr, < Re:, the effect of the lift force can be ignored. When Gr,
and Re? are of the same order, heat transfer by both natural and forced convec-
tion must be considered. More detailed information about heat transfer by natu-
ral and mixed convection is given in works [47, 48].
As is evident from the analysis of experimental data [33] about the effect of
mixed convection on heat transfer in tube banks, the extreme values of Re, = 0,
1, and 200 are of particular interest. The value Re; = 0 reflects heat transfer by
natural convection (if it is defined by the velocity of the forced flow). As far as
the physical nature of heat transfer is concerned, Re, = 1 is concerned as an
important value, since inertia forces in this case are equal to viscous forces.
At values of Re; higher than 200, the natural convection component in tube
banks can be ignored. Practically, it is important to know by how much heat
transfer in tube banks increases with transition from natural to forced convec-
tion.
Consider heat transfer in the first and inner rows of in-line and staggered
tube banks in a wide range of Re, Pr, and pitches a x b.
In practice, heat exchangers are operated in a wide range of thermal-carrier
velocity. Energy-wise, compact heat exchangers are the most effective. When
designing heat exchangers with various tube arrangements in banks (in-line or
staggered) and spacing between tubes, it is important to proceed from the ex-
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 67

changer’s intended use, conditions of operation, type of thermal carrier, spe-

cific features of fouling and cleaning of heat transfer surfaces, formation of
scale, and other factors. Heat exchangers with a X b < 1.25 xX 1.25 are
considered compact, and witha x b = 2.0 x 2.0 they are said to be widely
spaced. Thus, heat exchangers with 1.56 < a x b < 4.0 are considered of
medium compactness. Such a classification by compactness is related to heat
exchangers which are used in power engineering and industry, and are made
from tubes of 8-50 mm diameter. Heat exchangers used in transportation are
made from tubes of significantly less diameter; Therefore, their compactness is
considerably higher at the same pitches. Having this in mind, investigations
were carried out as IPTPE AS LitSSR to study heat transfer in a wide range of
variation of fluid velocity and pitches a x b. The results of the investigations
are generalized by similarity equations and can be used for small diameter
tubes, which are typical for the transportation sector. They can also be used in
the case of viscous liquid flow in the investigated range of Re, and Pr, values.
In this work, pitches were chosen first, since their study is more pertinent
for power engineering and industry. Based on the specific features of flow, tube
banks with small transverse and large longitudinal pitches or vice versa were
studied. Particular attention was paid to the determination of heat transfer of
extremely compact banks in which spacing between the tubes was equal to
fractions of a millimeter.
Consider heat transfer of the most typical tube banks at low Re; (1
Reps 1.35 Xx 10°) from the data of work [49] (Fig. 3.6). In the first row of
staggered bank 1.25 x 1.25, heat transfer in the range of Re, from 1 to 10° is
somewhat higher than in an inner row. In an in-line bank 1.25 x 1.25 they are
both equal. This is due to the fact that the front rows cause the flow to become
slightly turbulent in a compact bank. As a result, heat transfer in inner rows
intensifies. In in-line banks an inner row tube is lined exactly behind the pre-
ceding row, and heat transfer from it coincides with heat transfer from the first
row. This is due to lower flow velocity in the recirculation zones as a result of
the so-called “‘shadow”’ effect.
Heat transfer in the first row of a staggered bank becomes significantly
lower than in an inner row as from Re; = 10”; in in-line banks, on the contrary,
it is slightly higher. This can be explained by the fact that, with increasing Rey,
turbulence generated by the first row of staggered banks continues to positively
affect heat transfer in inner rows. The increase in Re; in in-line banks mainly
influences the frontal side of a tube in the first row; the frontal and rear sides of
a tube in an inner row are mainly in a low velocity recirculating flow.
Let us analyze the specific features of heat transfer in another extreme
case, namely in widely spaced banks 2.0 x 2.0 (Fig. 3.7). From Re, ~ 1 to
Re; = 10°, heat transfer of a tube in the first row of a staggered bank is lower
than in an inner row by approximately 7%. With further increase in Re,, the
difference increases up to 35%. At Re; < 10°, heat transfer of a tube in the first
row of an in-line bank 2.0 x 2.0 is higher by 25% than in an inner row.
68 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Zon
enna LA

Figure 3.6. Mean heat transfer of tubes in different rows of staggered and in-line banks 1.25 x
1.25: 1) first row; 2) fourth; 3) fifth. Kp = Nug Pre °°° (Pry/Pr,,)~°7°

2 a ell FF WF a bat

eeAi
oe
ie

Deed G1CalOaes tee al) ne

Figure 3.7. Mean heat transfer of tubes in different rows of staggered and in-line banks 2.0 x
2.0: 1, 3) inner rows; 2-4) first rows. The value of Ky as per Fig. 3.6.
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 69

Comparison of heat transfer of in-line banks 1.25 x 1.25 (Fig. 3.6) and 2.0 x
2.0 (Fig. 3.7) indicated that with increasing pitches the intensity of fluid flow in
recirculation zones falls, since heat transfer of a tube in an inner row decreases.
This is also evident from the difference in the flow around staggered and in-line
banks.
The exponent m of Re, in a zone with dominant laminar flow changes
considerably when Re, increases. At Re; < 10° m =0.33-0.4 and with further
increase in Re, m increases to 0.5 reaching, ultimately, 0.6 and 0.63 which are
typical values for staggered an in-line banks in a mixed flow region.
The first row of heat-exchanger tubes accounts for relatively large part of
the heat transfer surface area, since it is located on the largest circumference of
the tubular grid. When designing highly effective heat exchangers, one should
pay attention to heat transfer from the first row of tubes. Let us compare heat
transfer of the first rows of tube banks and generalize the obtained data using
the similarity equations. As is evident from Fig. 3.8, the difference in heat
transfer of staggered and in-line banks 1.25 x 1.25 and 1.5 x 1.5, up to
Re, = 10’, is insignificant. With increasing Re,, heat transfer of the first row of
a staggered bank 1.5 x 1.5 is higher by 10%, and of an in-line bank by 13%.
At Re, > 2 X 10° heat transfer of the first row of a staggered bank 1.5 x 1.5
is lower than heat transfer of a similar bank 1.25 x 1.25 by 12%, on the
average. Heat transfer of the first row of an in-line bank 2.0 x 2.0 is consider-
ably higher than of banks 1.25 xX 1.25 and 1.5 x 1.5 for the whole Re; range.

5f
ee to eras
a8 10° 2 OEY B80 Re,
Figure 3.8. Heat transfer of tubes of the first rows of staggered and in-line banks. Banks: /)
1.25 x 1.25; 2) 1.5 x 1.5; 3) 2.0 x 2.0. The value of Ky as per Fig. Br6:
70 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

This indicates that, at a certain spacing between tubes, flow intensity in recircu-
lation zones is at maximum.
When the obtained data on heat transfer of the first rows of staggered banks
are generalized, the following functions result:

Nu, = 1.0 Re?4 Prlee(Pee/ Pry yee atl < Re-< 10°

25 é 3 (3.6)
Nuy= 6.8 Rep0.45 Prop,.0.36
> (PrP r,)orate 02 2 Rex 2: id

and for the same rows of in-line banks

Nup 31,02 RepetPre’* (Pr

Prgyrate Reg alOy

Nu,=0.7 Re®5 Pr?-36 (Pr,/Pry)®? at Re, <1.5- 10° (3.7)

Formulas (3.6 and (3.7) are valid when a varies from 1.25 to 2.0, and b from
162510220:
Fig. 3.9 shows a comparison of data on heat transfer of inner rows of
staggered and in-line banks. In staggered banks, heat transfer as a function of
pitches a x b varies less than in in-line banks. This is so because the structure

teed Keron tealdeecco prceadiow el 2

“ail

0 2 — LG BIOS 2 6 RRO & Ze ee Re

Figure 3.9. Heat transfer in inner tube rows in staggered and in-line banks. Banks: IDY MIs), 3
1.25; 2) 1.5 x 1.5; 3) 2.0 x 2.0; 4) 2.0 1.25; The value of Ky as per Fig. 3.6.
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 71

2: ZOF2,0
SUL ALS

2 4680 2 4680 2 4 6 Re
Figure 3.10. Heat transfer of staggered and in-line tube banks at low Rey: 1-3, 5-7, 10) from data
by IPTPE AS LitSSR; 4, 9) [28]; 8) [26, 27]. The value of Ky as per Fig. 3.6.

of the turbulent flow in the spacing between tubes is distributed more uni-
formly, and the flow motion is sinusoidal. The turbulence intensity in a stag-
gered bank is caused by the bank pitches. Recirculation zones occupy the larg-
est part of the surface area in the frontal and rear sides of in-line tube banks.
The main flow, with the highest velocity in the bank between the rows, comes
in contact with a small part of the surface of the tubes at 90° and 270°. There-
fore, discrepancy between heat transfer curves of in-line tube banks with differ-
ent pitches is caused, mainly, by the intensity of fluid flow in the recirculation
zones. The size of the recirculation zones is mainly determined by the longitu-
dinal pitch b. According to work [2], at b > 1.7, the influence of a neighboring
tube in a longitudinal row decreases. The part of the tube surface which is in
contact with flow increases. This also enhances the intensification of heat trans-
fer in in-line banks.
When designing heat exchangers, it is imperative to know heat transfer
intensity in inner rows. Comparison of obtained data on heat transfer of a tube
in an inner row of banks with the results of other works (Fig. 3.10) shows good
correlation at Re, > 10° and slight discrepancy at Re; < 10°. At low Re, the
effect of turbulence level of the incoming fluid flow could become stronger.
Since the low values of Re, are characteristic of mainly high viscous fluid flow,
some discrepancies of the results at low Re, could be the results of varying
72 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

accuracy in determining the thermophysical properties of the thermal carrier in

our present and earlier works.
Heat transfer of a tube in an inner row of in-line banks, for the whole
region of predominantly laminar flow, is less than in staggered banks. As a
result of the generalization of available data, the following functions were ob-
tained for the calculation of the mean heat transfer in the region of dominant
laminar flow around tubes in an inner row of a bank:
for staggered banks in the range 1.0 < Re; = 40

Nu, = 1.04 Ref* Pr? (Pr,/Pr,,)°-?° (3.8)

in the range 40 < Re, = 1.5 X 10°

Nu, 0371 Re? Proce (Pry Pro (3.9)

and for in-line banks in the range 1.0 < Re, = 100

Nu; = 0.9 Re}4 Pro? (Pr7/Pr,,)o (3.10)

and in the range 100 < Re, < 10°

Nu, = 0.52 Ref? Pr?-*6 (Pr,/Pr,,)°2 (3.11)

From formulas (3.8)-(3.11), the total heat transfer by both natural and
forced convection is determined. If the designed heat exchanger is intended to
operate in the range of Re, from 1 to 200, the portion of heat transferred by
natural convection alone must be determined from equation (3.4). The obtained
result is then compared with the data computed from equations (3.8)-(3.11).
When Re; exceeds 10°, the flow in a tube bank changes significantly. The
flow in the intertubular spacing becomes vortex with increasing turbulence, and
the developed laminar boundary layer on the tube comes under the influence of
turbulence. All this is reflected on heat transfer of the tube. Such a flow regime
continues up to Rey = 2 X 10°. Transition from a dominant laminar flow
regime to a mixed flow begins and ends at different values of Re, and depends
on the arrangement of the tubes. This transition in in-line banks commences
slightly later than in staggered banks, due to the lower turbulence level of the
flow in the intertubular spacing.
Transition to mixed flow starts at Re; = 2 xX 10° in the majority of stag-
gered tube banks. Banks with very large transverse and small longitudinal
pitches (2.6 x 1.3 and 2.1 x 0.61) are an exception. In these banks, transition
lasts till Re, = 10°, since they are similar in a peculiar way to in-line banks due
to the formation of longitudinal slots along their length.
Heat exchangers operate mainly at moderate Reynolds numbers (10° <
Re, < 2 X 10°). Let us examine heat transfer of symmetrical and asymmetrical
tube banks of different configurations in this regime. Fig. 3.11 shows data on
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 73

=O)
ve

Pad bs ae) a ee 2 Re,

nN
Oo Ss N NS D & Ss
Figure 3.11. Mean heat transfer of symmetrical staggered banks of tubes of inner rows. Banks: /)
2S Xe 2552) ES e323) IES 54) 220i 220 she: valuevofi Ky as per Fig. 3.6.

heat transfer of the most widely used symmetrical banks of tubes of inner rows.
When designing new heat exchangers, it is useful to know by how much heat
transfer of a bank changes when pitches a x b change slightly. Changing the
pitches of the banks is necessary when seeking their optimum thermal and
hydraulic characteristics. This is more so since, before designing commences,
the permissible hydraulic resistance which is very sensitive to pitch variation is
assumed. The dependence of the hydraulic resistance of the banks on the mag-
nitude of the pitches is given in work [5]. Increasing the pitches from 1.25 x
1.25 to 1.30 x 1.30 (Fig. 3.11) has little effect on the level of heat transfer
K, = f(Re,); therefore, heat transfer of the banks is less sensitive to their
change than the hydraulic resistance. At Re; > 10*, the generalized heat trans-
fer curve of banks 1.5 x 1.5 and 2.0 x 2.0 lies somewhat lower than the heat
transfer curve of more compact banks.
Consider how heat transfer of nonsymmetrical staggered tube banks (Fig.
3.12) changes for significant variations of the transverse pitch and for a small
longitudinal pitch (bank 2.6 x 1.3 according to work [1] and bank 2.0 x 1.25
according to the data of IPTPE AS LitSSR). Also consider the opposite case for
small transverse and large longitudinal pitches (banks 1.3 x 2.9 and 1.25 x
2.0). Starting at Re, = 10°, heat transfer of the first tube rows becomes lower
than heat transfer of inner rows, since turbulence in the core of the bank is
higher. In medium range banks with a > b (bank 2.6 x 1.3; 2 x 1.25 and so
74 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

meas eaeae:
gue

Cette 2 BER 6 2
Figure 3.12. Mean heat transfer of asymmetrical tube banks: a—banks: J) 2.6 x 1.3; 2) 2.0 x
1.25 (first row); 3) 2.6 x 1.3; 4) 2.0 x 1.25 (inner row); b—banks: 7) 1.3 x 3.9 (first row); 2)
1.3 x 3.9 (inner row); 3) 1.25 x 2.0 (inner row). The value of Ky as per Fig. 3.6.

on), heat transfer of the first rows of tubes is lower by 45-60% than heat
transfer of inner rows up to the value Re, = 10*. In the interval of Re, from 10*
to 10° this difference decreases to 33%. This is explained by the fact that heat
transfer intensity of the rear side of the tube in the first row rises at high Re,
The flow becomes less stable and any disturbances caused by pitch variation
intensity slightly, leading to accelerated transition from one flow regime to
another (Rey ~ 2 x 10°). In tube banks with a > b flow turbulence in the
intertubular spacing, in the case Re, = idem, is higher than at a < b. This is so
since turbulence generated by tube rows at high Re, decays towards the core of
the bank (bank 1.3 x 3.9, 1.25 2.0, and so on). This is evident from com-
parison of heat transfer in tube banks with decreasing longitudinal pitch and
increasing transverse pitch. In staggered banks 1.97 x 0.99 and 2.1 x 0.61,
i.e., with b < 1, the direction of flow changes. For example, if the value K; =
64 in a bank 1.3 x 3.9 at Re, = 10°, then K, = 95 ina bank 2.1 x 0.61 and
equal to 100 in a bank 1.97 x 0.99. Heat transfer increases further at Re, =
10°. If K,;= 250 in a bank 1.3 x 3.9, then K, = 400 in banks 2.1 x 0.61 and
1.97 x 0.97. Consequently, by changing the pitches of the banks, particularly
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 75

the longitudinal pitch b, optimum variants of tube banks can be attained as

regards heat transfer.
Similar results were obtained in works [13, 26, 27, 50] for flow of air.
Heat transfer increases mainly with decreasing longitudinal pitch, and to a
lesser extent with increasing transverse pitch.
Fig. 3.13 shows data on heat transfer of an asymmetrical bank 2.1 x 0.61
for streams of water, air, and transformer oil in a mixed flow regime. It is
evident from Fig. 3.13 that heat transfer in an inner row is considerably higher
than in the first row of tubes for this bank and for bank 1.95 x 1.3.
Detailed analysis showed [1] that the variation of the constant c can be
accounted for by a parameter raised to the power 0.2 for a/b < 2. For a/b > 2,
the constant c = 0.4. The minimum flow cross section in these tube banks is
formed diagonally; therefore, the variation of the constant c is linked with some
variations of flow conditions.
Following a final generalization of data on heat transfer of a tube in inner
rows of various staggered banks, the following formulas are obtained:
for a/b < 2

Nu; =0,35,(a/6)°2 Re? Pr7** (Pr,/Pr,,)? (3.12)

and for a/b > 2

aN
1

Figure 3.13. Mean heat transfer of tubes in different rows of a staggered bank 2.1 x 0.61 asa
function of Reg Rows: /) first; 2) seventh; 3) sixteenth. The value of Ky as per Fig. 3.6, @)
transformer oil; +) air, O) water; / and 3) different liquids.
76 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Nu, =0.40 Re? Pr?-6 (Pr,,/Pr,,)%2° (3.13)

Heat transfer of staggered banks at 10° < Re, = 2 X 10° can be deter-
mined from equations (3.12) and (3.13) in a wide range of variation of pitches
a x b.
Consider heat transfer of in-line banks with different pitches a x b. To
obtain large heat transfer surfaces and low hydraulic resistances, in-line tube
banks must be used. Heat transfer in in-line banks is somewhat lower in com-
parison with staggered tube banks. However, the hydraulic resistance is signifi-
cantly less, particularly when the transverse pitch is larger than the longitudinal
pitch (e.g., in banks 2.6 X 1.1, 2.6 X 1.3, 2.0 X 1.25, and others). With
decreasing transverse and longitudinal pitches of in-line tube banks, their ther-
mal and hydraulic characteristics become similar to staggered tube banks. The
flow in intertubular spacing of dense tube banks (for example, 1.1 x 1.1,
1.25 x 1.25, 1.3 x 1.3) changes to a fast-moving current. When the pitches
are increased in symmetrical banks, the flow around the tubes becomes turbu-
lent. A decrease in the longitudinal pitch gives rise to the ‘“‘shadow”’ effect by
the preceding tube. As a result, heat transfer decreases due to the fact that a
considerable part of the tube in the frontal and rear sides is affected by low
velocity fluid recirculation.
With increasing transverse pitch, the character of the flow in a bank be-
comes analogous to the character of flow in a channel; therefore, heat transfer
decreases.
Data on heat transfer of tubes in the first and inner rows of symmetrical in-
line banks are presented in Fig. 3.14. The curve in Fig. 3.14a represents heat
transfer of the symmetrical banks 1.25 x 1.25, 1.3 x 1.3, and 1.5 x 1.5.
Beginning from Re; ~ 10° to Re, = 2 X 10°, the transition from a predomi-
nantly laminar flow regime to a mixed regime is vividly clear. Heat transfer
intensity in the transition region increases substantially and, following the tran-
sition at Re, > 2 X 10°, the exponent of Re, increases significantly. Such a flow
regime continues up to Re; ~ 2 X 10°. From Re, = 10° upwards, the heat
transfer of the first rows of tubes decreases further, compared with heat transfer
of the inner rows. This is an indication of how turbulence, generated by the first
rows of tubes in a bank, affects heat transfer of inner rows. At Re, = 10°, the
difference in heat transfer of the first and an inner row amounts to 24%. At
Re; = 10°, this difference rises to 70%.
Fig. 3.14b shows data on heat transfer of tubes of symmetrical banks with
large pitches (2.0 X 2.0, 2.5 X 2.5, 3.0 x 3.0), and also with pitches between
2.0 x 2.0 and 3.0 x 3.0. A comparison of the curves in the upper and lower
parts of Fig. 3.14 indicates that heat transfer of symmetrical banks with large
pitches is slightly higher than heat transfer of banks having pitches between
1.25 x 1.25 and 1.5 x 1.5. In moving towards less compact tube banks, the
symmetrical bank 1.5 x 1.5 is considered to be an intermediate bank, since
with further increase of pitches heat transfer begins to rise. The increase in heat
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 77

LeageG pee SE WEG Gi NG hp 2 Re,

Figure 3.14. Mean heat transfer of symmetrical in-line tube banks: a—banks: J) 1.25 x 1.25;
1.3 X 1.3; 1.5 x 1.5 (first row); 2) same as before (inner row); b—banks: /) 2.0 x 2.0; 2.5 x
2.5; 3.0 x 3.0 (first row); 2) same as before (inner row). The value of Ky as per Fig. 3.6.

transfer in symmetrical banks when their pitches are increased can be explained
by flow activation in the regions of recirculation. Turbulence generated by the
first rows penetrates the boundary layer developed on the tube more effectively
than in compact symmetrical in-line tube banks. In the latter, turbulence decays
as a result of the flow being compressed between longitudinal rows. Heat trans-
fer difference between the first and an inner row is lower than in compact
banks; and this can also be explained by the aforementioned reasoning. The
difference amounts to 35% in the range of Re; values from 10° to2 x 10° (Fig.
3.145).
Practice shows that the tubes get fouled when solid fuels, such as oil shale,
with low heating values are burned. Therefore, it is necessary to increase the
pitches of the tube banks from 3.0 x 3.0 to 10 x 10. Investigations show that
heat transfer and hydraulic resistance of tube banks with pitches 3.0 x 3.0 <
a X b < 10 x 10 differ from those of a single tube. Vortex shedding down-
stream of a tube row in banks with sparsely spaced tubes causes strong flow
turbulence which considerably influences heat transfer of the following row,
despite the fact that it is situated at a large distance, for example, s) = b X
d = 10. For the calculation of heat transfer and hydraulic resistance of in-line
78 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

and staggered tube banks with pitches a x b > 3.0 X 3.0, the relations
presented in works [51, 52] can be used for the medium range of Re, values. It
is pertinent to mention that the hydraulic resistance of tube banks with very
large pitches (a x b > 3.0 x 3.0) is substantially less than in the case of tube
banks of normal density. For example, at Re, = 10°-10°, the coefficient of
hydraulic resistance in an in-line bank 6.0 x 2.0 is less by an order of magni-
tude than in a bank 2.0 x 2.0, by two orders of magnitude less than in a bank
1.25 x 1.25, and approximately by three orders of magnitude less than in a
bank 1.1 x 1.1. Heat transfer of a bank 6.0 x 2.0 is 3.5 times less than heat
transfer of a bank 1.1 x 2.0, and only 50% less than heat transfer of a bank
2.0 x 2.0. Therefore, by using sparsely spaced banks, the energy expended to
move the thermal carrier can be substantially reduced. However, the dimen-
sions of the heat exchanger in this case increase significantly.
Let us consider the variation of heat transfer of in-line banks with an asym-
metrical tube arrangement. Fig. 3.15a gives data on heat transfer of the most
typical tube banks with large transverse and small longitudinal pitches. In such
tube banks, transition from a regime of a predominantly laminar flow to a
mixed flow occurs at Re, = 600, and continues to Re, = 10°. From Re, = 10°,

4
10
;

ite
Bae
TT] FH

Hitt meal

Le
Hill IP:

Dey A witli 6:68100 Wham el68 Oe a LOZ 4 Re

Figure 3.15. Mean heat transfer of asymmetrical in-line tube banks: a—banks: 7) 2.6 x 35
2.6 X 1.1; 2.0 x 1.25; 1.95 x 1.3 (first row); 2) 2.6 x 1.3; 2.0 x 1.25; 1.95 x 1.3 (inner
row); 3) 2.6 x 1.1 (inner row); b— banks: J) 1.3 x 2.6; 1.25 x 2.0; 1.05 x 1.85 (first row); 2)
1.3 X 2.6; 1.25 x 2.0 (inner row); 3) 1.05 x 1.85 (inner row).
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 79

the exponent of Re, increases, and a regime of mixed flow commences. In

practical calculations, heat transfer of banks with pitches of the type a x b can
be generalized, with sufficient accuracy, by a single relationship.
Starting from Re, = 10°, heat transfer of the first row in banks with a > D
becomes lower than heat transfer of an inner row by approximately 40%.
Heat transfer of the most characteristic tube banks with small transverse
and large longitudinal pitches is presented in Fig. 3.15b. In banks with pitches
of the type a < b, heat transfer is more sensitive to changes in a transverse
pitch and drops with its decrease.
Heat transfer in the first row is less than in an inner row at much lower Rey,
in comparison with banks in which a = b ora > b. The difference in heat
transfer of the first and inner rows of banks with a < b amounts to 40% at
Re, = 200, and with increasing Re, the difference reaches 65%.
In banks with a < b, due to a small transverse pitch, the incoming flow
accelerates significantly and then expands downstream of the transverse row as
a result of large longitudinal spacing between tubes. Therefore, large pressure
losses are incurred. Despite this fact, heat transfer of these banks is lower than
heat transfer of symmetrical in-line banks and banks with a < b. When design-
ing banks with a < D, it should be remembered that their choice at low Re, ina
regime of predominantly laminar flow is recommended. This is because the
difference in heat transfer of banks with different pitches decreases together
with the difference in heat transfer of the first and inner rows. In order to
generalize the research results obtained in work [1] and also later data, a com-
parison is conducted of the characteristics of various in-line tube banks (Fig.
3.16). Apparently, the constant c obtained for in-line banks as a function of
their configuration and the exponent m of Re, are different, whereas m = 0.6
for all staggered tube banks. For constant transverse pitch of in-line tube banks,
m increases with decreasing longitudinal pitch. According to the data of work
[1], the variation of m is influenced by the magnitude of the relationship of
longitudinal and transverse clearances 6, = s, — d and 6, = s, — d. Analysis
conducted in dimensionless variables 6,/d = (s, — d)/d = a — 1 indicates that
the value of m rises from 0.6 to 0.63-0.65 (depending on the pitch b) when a-1
varies from 0.2 to 2.0, i.e., in banks with a@ = 1.2-3. For lower values of 5,
the exponent m has a maximum value. At b-1 = 0.1, the value of m = 0.74,
and when b-1 = 1 is reached, m asymptotically approaches the constant value
0.63. This is noticed in banks with a pitch a ~ 2.5. In compact banks with
pitch a = 1.25-1.3 the value of m = 0.64 at b-1 = 0.2, and at b-1 = 2 it
approaches the value m ~ 0.6 asymptotically. For the medium range of Re,
values, i.e., in the regime of mixed flow of the majority of in-line banks, the
exponent of Re, can be assumed to be equal to 0.63. For the calculation of heat
transfer in an inner row of the most widely used in-line banks, the following
equation is recommended in the range of Re, from 10° to2 x 10°

INU 2= 0.27, Ree? Pry3? (Pr, /Pr,,)>*" (3.14)

80 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Figure 3.16. Heat transfer of tubes of various configurations in an inner row of in-line banks at
medium range Reg Banks? 7); 1305x1730) 1h6)52:072.0' 9220: 21421022) Oe esa)
1.95 x 1.30; 4) 2.6 x 1.1; 5) 1.30 x 2.0; 6) 1.3 x 2.6; The value of Ky as per Fig. 3.6.

Fig. 3.17 shows a comparison of heat transfer of the most characteristic stag-
gered tube banks with various pitches a x b. They are distinguished by the fact
that at medium Re, range the exponent m is equal to 0.6 on the average, and at
high Re, m = 0.80. The influence of widely varying pitches on heat transfer is
evident from this figure. When the longitudinal pitch is decreased, heat transfer
increases. With increasing transverse pitch, heat transfer varies to a lesser
extent.
After establishing that the longitudinal pitch of in-line tube banks greatly
affects heat transfer, further investigations (Fig. 3.18) on heat transfer of com-
pact banks of small diameter tubes were carried out. When the pitch b was
varied from 1.09 to 1.12, the longitudinal clearance between the tubes
amounted to 1.44-1.92 mm. Since the tubes were mounted on guides with
accurately drilled holes, the distance between the tubes could be determined
with an accuracy of + 0.01 mm. The longitudinal pitch was smaller than the
transverse pitch, which was varied slightly in order to determine the optimum
(for heat transfer) variants of tube banks (Fig. 3.18). The most effective bank
was 1.3 X 1.12. Generalization of data on such compact tube banks gives the
following dependence for the calculation of mean heat transfer in the Re, range
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 81

GA
wy
BAG
Za ich
SG
isl |sees
ee |

jp Tee on PP Ee I ee ee Oe 4 6 810° Re,

Figure 3.17. Heat transfer of tubes of various configurations in an inner row of staggered banks.
Banks31)2:6 10:92) ale 0.6173) 13329574) 21 X90161575) 1297 x10:99516) 95a x
1327) leSex<ale3 AO) Onxaless/9)meSexXwles- LO)2 exe 8531) ilese xa al2) le Sex Lal;
13) 2.0 x 1.25; 14) 1.25 x 2.0; 15) 1.25 x 1.25. The value of Kyas per Fig. 3.6.

Figure 3.18. Mean heat transfer of compact in-line tube banks. Banks: 7) 1.4 x 1.12; 2) 1.2 x
1.09; 3) 1.3 x 1.12; 4 1.2 x 1.12. The value of Ky as per Fig. 3.6.
82 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

from 10‘ to2 x 10°:

Nu, =0.308 Re} Pr??6 (Pry/Pry)°?* (GH15)

Heat transfer of very dense tube banks has specific features. Investigations were
conducted to study heat transfer of in-line banks in which the clearance between
the tubes amounted to fractions of a millimeter. In such a bank, the incoming
flow increases significantly near the minimum cross-sectional area. However,
for small clearances between tubes, the maximum velocity influences only a
small part of the tube surface. Evaluation of experimental data at maximum,
mean integral (from y = 0 to g = 90°) and incoming velocities indicates that
the use of the maximum velocity gives large discrepancies of heat transfer
curves of various tube banks. If the data on heat transfer of condensed banks
with different pitches are evaluated at the incoming velocity, correlation is good
and they can then be compared with other results. This is confirmed by the data
on heat transfer of the dense tube banks 1.008 x 1.008, 1.02 x 1.02, 1.026 x
1.026, 1.053 x 1.053, according to reference [1], which were evaluated at the
incoming velocity. When the pitches are decreased, heat transfer increases
slightly. Bank 1.008 x 1.008 has the lowest heat transfer. This indicates that
excessive compression of the flow dampens vortices generated by the tubes of
the bank leading to a decrease in heat transfer.
When the data are generalized (Fig. 3.18), the following dependence for
the calculation of dense tube banks is obtained in the Re; range from 4 x 10° to
4 x 10* (Reynolds number Re, is evaluated at the incoming velocity):

Nu, = 2.05 Ren Pr (PriPr,)— (3.16)

In our search for effective heat exchangers, we conducted investigations of heat

transfer of banks from coils with small turn radius in in-line arrangement. The
coil was constructed from a densely coiling tube 6 mm in diameter, such that
the turns were touching. The outer diameter of the coil D = 23 mm. The coils
were arranged in banks with pitches: s = 23 mm (when neighboring turns
touched), s = 25 mm (for a clearance of 2 mm between the turns), and s = 27
mm (for a clearance of 4 mm). The fluid flowed outside and inside the coil.
Such heat exchangers have low hydraulic resistance since three flow patterns
are combined in them, namely, longitudinal, at an angle, and transverse. With
increasing distance between the coils, heat transfer decreases. Any further in-
crease of the clearance 6/D, for example to 0.17, is pointless since heat transfer
remains unchanged, and the flow becomes analogous to the flow around a
single coil (6—clearance between coils, D—external diameter of the coil). For
the calculation of heat transfer of in-line banks in the form of coils, the follow-
ing dependencies are obtained:

Nu, =0,101 Re9- Pr?:36 (Pr,/Pr,,)?25 (3.17)

 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 83

when 6 = 0, i.e., the coils are in contact,

Nu, = 0.0904 Re? Pose (Prj/ Pr)?" (3.18)

when the parameter 6/D varies between 0.087 and 0.17 for Re, from 2 x 10° to
2 x 10°. Reynolds numbers were determined from the diameter of the tube
from which the coil was constructed and from the average mass velocity in the
smallest cross-sectional area of the bank. The average mass velocity Z,, is de-
fined by the equation

Um =G/ f+ (3.19)
where G is the mass flow rate of air f is the smallest cross-sectional area of the
bank and p is the density. The banks of coils, constructed from small diameter
tubes, are compact and have low coefficients of hydraulic resistance, compared
with tube banks in crossflow. Therefore, they can be employed in various
branches of emerging technologies where low energy is expended on the move-
ment of the thermal carrier and high compactness of heat exchangers is a pre-
requisite.
When the Re; number reaches 2 X 10°, the flow regime changes. Accord-
ing to data on the investigation of heat transfer in tubes, channels, and plates,
the exponent m of Re, rises to 0.8 and stays constant in the Re; range from 2 xX
10° to 10’. The heat transfer coefficient is significantly higher than in the low
and medium ranges of Re, values.
For tube banks in crossflow, the heat transfer process is different. Depend-
ing on the character of tube arrangement is the bank, m varies within the
bounds 0.8-1.0. The flow structure strongly influences the intensity of heat
transfer at high Re; which is the main reason for the increase of m. Thus, when
one is determining the optimum heat transfer parameters in tube banks, one
must find the arrangement for which the exponent of Re; is maximum. This
must be remembered when heat transfer is being intensified by increasing the
velocity of the thermal carrier. Experiments conducted at IPTPE AS LitSSR
showed that heat flows of the order 10° W/cm’ can be obtained at Rey = 2 X
10°. The maximum velocity in the compact bank, in this case, reaches 33 m/
sec. At such a high fluid velocity, the pressure in the frontal side of the tube
decreases significantly, causing cavitation. To avoid cavitation damage the pres-
sure in the system must be increased in proportion to the velocity of the thermal
carrier.
Despite the high heat flow obtainable at Re; = 2 X 10°, even higher
performance criteria are sought in some branches of modern technology. There-
fore, the study of heat transfer in tube banks in the Re, range from 2 x 10° to
10’ is of scientific interest. However, to achieve these values, extremely high-
power devices are needed. Also, the intensification of heat transfer in the
above-mentioned range is not practically worthwhile. Starting from Re, number
84 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

of the order 10°, boiling on the surface vastly improves the heat transfer charac-
teristics, leading to an increase in heat flow. Heat flow per unit area in this case
can amount to 5 x 10° W/cm’.
With the purpose of studying heat transfer at high Re, in tube banks, com-
prehensive investigations were staged at the IPTPE AS LitSSR. The works of
J. Stasiulevicius and P. Samoska [53] were among the first of a series of investi-
gations. Their work dealt with the determination of the mean heat transfer of
tube banks with various transverse and longitudinal pitches at high Reynolds
numbers in a flow of air. Later, investigations continued in flow of water [36,
37, 54] and air [55-58]. Great attention was paid to studying the peculiarities of
local heat transfer on the tube surface, and to deriving equations to calculate the
mean heat transfer at high Re, in a wide range of variation of pitches a x b. It
was observed that the break in the mean heat transfer curves during transition to
higher Re, numbers occurs in the Re, range from 2 Xx 10° to 5 x 10°. Follow-
ing the break, the exponent of Re; increases and heat transfer becomes signifi-
cantly more intensive. For example, at Re, = 4.7 x 10* the mean heat transfer
coefficient & of in-line bank 1.50 x 1.50 becomes equal to 8910, and when Re,
reaches 1.27 x 10° it becomes equal to 106000; i.e., it increases 11.9 times.
The exponent m for high Re, varies from 0.65 to 0.8 for the investigated tube
banks.
In Fig. 3.19 data are presented on the mean heat transfer of the first and
inner rows of tubes of in-line and staggered banks 2.0 x 2.0.
The mean heat transfer of tubes in the first row of banks is significantly
lower than in the inner rows. The values of mean heat transfer in in-line and

6 810 2 NAEP Re,

Figure 3.19. Mean heat transfer in the first and inner rows of banks: 1) 2.0 X 2.0 (first row); 2)
2.0 x 2.0 (inner row of staggered banks); 3) 2.0 x 2.0 (first row); 4) 2.0 x 2.0 (inner row of in-
line banks). The value of Ky = Nu, Pr~°# Pr¢/Pr,,) °°
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 85

3
Coo” 9 eh Te AIMS 2 Re
Figure 3.20. Mean heat transfer of in-line tube banks in flow of air at high Re, @) first rows, O)
inner rows. The value of Kyas per Fig. 3.19.

staggered banks with large pitches coincide. This indicates that the neighboring
tubes in the first and second rows do not influence heat transfer. In the investi-
gated range of Reynolds numbers, heat transfer of banks is generalized by the
equation

Nu, = 0.103 Re?-® Pr? (Pr,/Pr,,)°° (3.20)

At high Re,, heat transfer in an inner row of a bank depends substantially

on the character of tube arrangement. Fig. 3.19 shows that prior to the break in
the curves, m is equal to 0.61 for the staggered bank 2.0 x 2.0, and 0.64 for
the in-line bank. Following the break in the curves, m for the in-line bank
increases up to 0.78, and for the staggered bank it increases up to 1.0. The
large increase in m at high Re; can be attributed to the influence of two factors:
unequal longitudinal pressure gradient dp/dx, and varying turbulence near the
tube surface—in other words, the two values which are most sensitive to the
change in bank geometry, and which affect the character of boundary layer
development on the tube and the intensity of flow in the recirculation zones.
Fig. 3.20 shows data on the mean heat transfer of tubes in the first and
inner rows of an in-line bank 1.5 x 1.25, and in an inner row of a bank 2.0 x
1.25. At Re, < 2 X 10°, the exponent m of Re; in the equation of heat transfer
is equal to 0.6-0.65. At high Re, m increases to 0.76-0.8, indicating a change
in the mechanism of the heat transfer process. An increase in m is mainly
86 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

related to the shift of the transition point, of the laminar boundary layer to
turbulent boundary layer, towards the frontal critical point. This was shown
above when local heat transfer was analyzed. As a result of the aforementioned
argument, the length of the turbulent boundary layer on the tube surface in-
creases. As evidenced by the presented data, the character of transition to the
zone of more intensive heat transfer is different for the first and inner rows. In
the first row, a characteristic deviation of heat transfer from dependencies
raised to a power is observed in the region Re; = (2.5-7) X 10°, as was the
case with heat transfer of a single tube in low turbulence flow. This deviation of
heat transfer is related to the formation of a separation cavity and to the sharp
change of heat transfer in the frontal region of the tube. As was stated above,
such a deviation from a power dependence is not observed for a single cylinder
in a high turbulence flow. A similar picture is observed in an inner tube row of
in-line banks, and also in an inner tube row of staggered banks (curve J) 1.5 xX
1.25, curve 2) 2.0 x 1.25, Fig. 3.21). The exponent m for staggered tube
banks at Re, > 2 x 10° is somewhat higher and is equal to 0.8. Similar results
were obtained in a flow of water in a regime of supercritical flow: for the in-
line bank 2.0 x 2.0 m = 0.78, and for the staggered bank 2.0 x 1.25 m =
0.8.
Figs. 3.22 and 3.23 compare data on heat transfer of in-line and staggered
asymmetrical tube banks and compact banks 2.0 x 1.25, 1.25 x 2.0, 1.5 x
1.25, 1.5 x 0.865, and 1.25 x 0.865 at high Re, values. At Re; < 2 X 10:
the results agree well with previously analyzed data on heat transfer in tube
banks. The influence of the relative pitches of a bank on heat transfer of tubes at
high Re, numbers is similar to their influence in the medium range of Re,

6 810 et aedOE V8 Re,

Figure 3.21. Mean heat transfer of tubes in an inner row of staggered banks in flow of air at high
Reg The value of Ky as per Fig. 3.19.
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 87

6
6 8 ]0 Rey

Figure 3.22. Heat transfer of tubes in an inner row of in-line banks at high Re, The value of Ky as
per Fig. 3.19.

values. Data on heat transfer in flow of water and air, expressed as a depen-
dence K; = f(Re;), agree well with each other for each tube bank. As can be
seen from Fig. 3.22, m ~ 0.63 prior to the break in the curves, i.e., to Re;<
2 x 10°. Following the break in the curves, the exponent increases to 0.8. Heat
transfer in the in-line banks 2.0 X 1.25 is more intensive than in banks 1.5 x
1.25, 1.25 x 1.25, and 1.25 x 2.0. Banks with small transverse (a = 1.25)

mies
a YAK 125 «0,865
we Ls {25 «425

Awe eRe
Figure 3.23. Heat transfer of tubes in an inner row of staggered banks at high Reg The value of Ky
as per Fig. 3.19.
88 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

and large longitudinal (b = 2.0) pitches have the lowest heat transfer. For such
a case, the exponent m is equal to 0.74 at high Re, numbers.
As research shows, heat transfer in the first row of an in-line bank 2.0 x
1.25 is higher by 18% than in a bank 1.25 x 2.0 At high Re,, heat transfer of
the first row of an asymmetrical bank is generally lower than in an inner row.
Generalization of heat transfer of tubes in the first rows of asymmetrical in-line
banks gives the following dependence when the pitch a varies from 1.25 to 2.0,
and pitch b from 0.865 to 2.0

Nu, =0.113 Re? Pr? (Pr,/Pr,,)°?° (3.21)

in the range of Re, from 2 X 10° 2 x10":

To calculate the mean heat transfer of an inner tube row of symmetrical and
asymmetrical in-line banks, the following dependence is recommended in the
range of Re, from 2 Xx 10° to 2 x 10°

Nu,=0,033 Re?® Pr? (Pr, /Pr,,)?? (3.22)

In asymmetrical staggered tube banks at Re; > 2 Xx 10°, the exponent m in the
heat transfer equation deviates little from the value 0.8. For example m = 0.8
for bank 2.0 x 1.25, and m = 0.75 for bank 1.25 x 2.0. A decrease in the
longitudinal pitch b leads to a significant increase in heat transfer (Fig. 3.23);
however, m changes little. For bank 1.5 x 0.865 m = 0.77, and for bank
1.25 x 0.865 m= 0.76. Bank 1.25 x 2.0 has the lowest heat transfer, i.e.,
with increasing longitudinal pitch, heat transfer of the bank decreases.
As research shows, the mean heat transfer of the first rows in tube banks
with a > b (for example, a X b = 2.0 X 1.25 and others) is 1.8 times lower
compared with heat transfer in an inner row; and in tube banks with a < b (for
example, a X b = 1.25 X 2.0 and others) is 2.1 times lower. An increase in
longitudinal pitch and a decrease in transverse pitch leads to a decrease in heat
transfer of the first tube rows. The mean heat transfer of tubes in the first row
of asymmetrical staggered banks can be determined by the following depen-
dence in the range of a and b variation from 1.25 to 2.0 and 0.865 to 2.0,
respectively

Nu, =0.025 Re?-”” Pr?-4 (Pr,/Pr,,)°-25 (3223)

and in an inner row by the dependence

Nuy=0.31 (a/b)? Re? Pr (Pr,/Pr,,)925 (3.24)

 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 89

3.4. EFFECT OF NUMBER OF TUBE ROWS

ON HEAT TRANSFER

Investigations indicate that heat transfer of a tube in a bank is dependent on its

location in the bank. Heat transfer of the first tube row in banks either coincide
with or slightly deviate from heat transfer of an inner row at Re; S 10°. In-line
banks with small longitudinal pitch and sparsely spaced banks (e.g., a X b =
1.5 x 1.5, 2.0 x 2.0) are an exception to this rule. Heat transfer of the first
tube row of the latter is higher than heat transfer of deep rows, which is ex-
plained by the “‘shadow”’ effect in an inner row. With transition to the medium
range of Re, values (10° < Re, = 2 X 10°), heat transfer of inner tube rows
start to increase as a result of higher turbulence, which is generated by the first
tube rows.
In the region of high Re,, heat transfer of inner tube rows is higher still than
the first rows. In staggered banks the turbulence generating role of the first and
second rows is more pronounced than in in-line banks. The level of turbulence
in tube banks is dependent on the Reynolds number. For example, up to Re; <
10°, heat transfer of the second tube row of a staggered bank 1.25 x 1.25 is
slightly less than heat transfer of an inner row, and at Re, > 10* this difference
diminishes. At medium and high Re; ranges, heat transfer in staggered banks
becomes stable, starting from the third tube row. In in-line banks, heat transfer
stabilizes from the fourth row and it does not differ from heat transfer of an
inner row.
A comparison of heat transfer of the first and inner rows of different banks
indicates a strong influence of turbulence on heat transfer. According to the data
presented in Table 3.1 from work [1], heat transfer of inner tube rows increases
mainly with decreasing longitudinal pitch. The difference in heat transfer be-
tween the first row and an inner row in in-line banks is somewhat smaller than
in staggered banks.
When the number of rows in a bank along the direction of flow is small, the
lower heat transfer of the first rows must be taken into account by the coeffi-

Table 3.1. Degree of intensification of heat transfer of tube banks

Staggered Degree of In-line Degree of

tube banks intensification, % tube banks intensification, %

OS 115330) 175 1.30 x 1.30 160

2.60 x 1.30 170 2i0Exe 220 145
1.95 x 1.30 170 1.26 x 2.0 150
TEXOSS S580) 165 LO x BO 145
30) XaledS 140
1.30 x 3.90 125 First row 100
EEE
90 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

cient c., the dependence of which on the number of rows in a bank is shown in
Fig. 3.24. The dependence c, = /(Z, Re,) is plotted on the basis of experimental
data on the heat transfer of tubes in different rows of banks of various arrange-
ments and geometries with varying Reynolds numbers. According to Fig. 3.24
c. = 1 only for banks with the number of longitudinal rows exceeding 16. This
means that the average heat transfer of tubes of the whole 16-row bank is equal
to heat transfer of tubes of an inner row of a bank. The correction for the
number of rows depends on the values of Re; and z and is accounted for by the
dependence

Nug=c,NueSa¢ (225)

3.5. EFFECT OF FLOW ANGLE OF ATTACK

Heat transfer surfaces are composed of not only tubes in crossflow (6 = 90°),
but also of tubes arranged at an angle relative to the incoming flow (6 < 90°).
To avoid tube overheating it is very important to know the temperature and
heat-flow distribution along the tube perimeter at different angles of attack of
flow. Research work on heat transfer of tube banks at angels of attack B < 90°
is scarce. Thus, works [59, 60] are devoted to the investigation of the mean
heat transfer of tube banks in flow of air, and work [61] to the study of local and
mean heat transfer of staggered and in-line banks with transverse pitches
1.34 xX 1.16 and 1.34 x 1.34 respectively, in flows of viscous fluids at angles
of attack 25, 40, 60, 75, and 90°. Data on local heat transfer of a tube in an
inner row of a staggered bank 1.34 x 1.16 at angles of attack 8 = 25 and 75°
and for the same Re; show that nonuniformity of heat transfer along the tube
perimeter increases with decreasing 6. Generally speaking, the qualitative pat-
tern of heat transfer along the perimeter changes insignificantly. The determin-
ing velocity, for the above-mentioned study, was taken as the flow velocity in

are ee ea
cna)
Cahmeatal
172 [ee |

Oo Bano Oe Be JOP AR ae Oh ae
Figure 3.24. Correction for the number of rows when calculating average heat transfer of tube
banks (solid line—in-line bank, broken line—staggered bank).
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 91

Figure 3.25. Correction for the effect of angle of attack when calculating heat transfer of a tube in
a bank O @—staggered bank, A A—in-line bank (O, A—first row, @ A—inner row).

the minimum cross-sectional area of the tube bank corresponding to the angle
B = 90°.
Such a condition of comparison (at the same values of Re;) is very conven-
ient for practical computations of mean heat transfer of inclined tube banks. If
heat transfer of a tube bank in crossflow (G = 90°) is known, a correction
factor must be introduced, in accordance with the curve presented in Fig. 3.25,
for the calculations at any other values of the angle @. The coefficient c, is the
ratio of heat transfer of the tube at a specified value of angle 6 to heat transfer at
B = 90°:

Ce => Nug/Nug~90° (3.26)

The dependence (3.26) is obtained from experimental data for flow of water. It
is evident from Fig. 3.25 that the data on the heat transfer of staggered and in-
line tube banks are generalized by a single curve, and that c, decreases with
increasing angle of attack.

3.6. CALCULATION OF MEAN HEAT TRANSFER

OF TUBE BANKS

It was noted that the mean heat transfer of a tube in a bank depends on its
location relative to the core. In addition, the difference in heat transfer of tubes
of the first and inner rows is dependent on the Reynolds number, the geometric
characteristics, and the arrangement of the banks. Tube banks with a different
number of longitudinal banks are employed in practice. The difference in heat
transfer of tubes of the first rows in multi-row banks has less influence on the
92 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

average heat transfer of the whole bank than in banks with small number of
rows. This influence in the latter could be significant and must be accounted
for.
Generally, the average heat transfer of the whole bank can be computed by
the formula

Nu =Nu,- cz: (3.27)

where Nu, is the heat transfer of a tube in an inner row of banks of smooth tubes
in crossflow at 8 = 90°; c, is a coefficient which accounts for the dependence
of the average heat transfer on the number of rows of a bank (3.25); and cg is a
coefficient which accounts for the decrease in the average heat transfer due to a
decrease in angle of attack (3.26).
The generalization and analysis of data showed that calculation of heat
transfer of a tube in a bank is better carried out by separate equations for
characteristic regions as a function of Re;. To determine Nu,, the equations
presented in Figs. 3.26 and 3.27 must be used. The mean flow temperature

Rep
Figure 3.26. Data for the calculation of heat transfer of a tube in an inner row of in-line bank of
smooth tubes at 8 = 90°.
1: 1) Nuy=0.9 Re?* Pr9* (Pry/Pr,)*5, 2) Nu,=0.52 Re? Pr28¢ (Pr, /Pr,,)0.25
I: Nuy=0,27 Re9*** Pr36 (Pry /Pr.,)?-25 i iE gaa
Ill: Nuy=0.033 Re9® Pr? *(Pr,/Pr,,)°-25
 MEAN HEAT TRANSFER FROM BANKS OF SMOOTH TUBES 93

Figure 3.27. Data for the calculation of heat transfer of a tube in an inner row of staggered banks
of smooth tubes at 8 = 90°.
I: 1) Nuy=1,04 Re}* Pr}°* (Pr,/Pr,,)°*5, 2) Nuy=0.71 Re%® Pr3-3¢ (Pry/Pr,,)°-2.
II: Nuys =0,35 (a/b)? Rep* Pr9** (Pry/Pr,)*5 at a/b < 2, Nuy=0.4 Re?:®? Pr9*® (Pry/Pry)%*®
at. a/b>2.
III: Nuy=0,031 (a/b)°? Re}? Pro* (Pr,/Pr,,)°?5

should be taken as the determining temperature, and the average temperature of

the surface of all tubes of the bank as the surface temperature. The mean
velocity in the minimum cross-sectional area of the bank should be the deter-
mining velocity, and the diameter of the tube as the determining dimension.
The coefficients c, and cz are determined from Figs. 3.24 and 3.25.
 CHAPTER

FOUR
HEAT TRANSFER OF BANKS OF ROUGH TUBES

Special attention is currently being paid to the intensification of the convective

heat transfer process in heat-exchange devices. During the interaction of the
solid, heat-conducting, and impermeable surface with a surrounding gas or
liquid, the main resistance to heat transfer is caused by the boundary layer. The
thicker the thermal boundary layer and the lower the thermal conductivity of the
fluid, the lower the heat transfer will be. Therefore, the main methods of
intensifying convective heat transfer are directed towards the destruction or
turbulization of the boundary layer. The methods which employ additional ex-
ternal energy (ultrasonic, vibrations, electrical field, mechanical action, and so
forth) are called active. The methods which do not use external energy are
known as passive. The latter include the use of finned and other developed heat
transfer surfaces when the thermal carrier has a low heat transfer coefficient,
and the use of boundary layer spoilers.
One of the most effective passive methods of intensifying convective heat
transfer between the wall and the surrounding turbulent flow is the creation of
artificial wall surface roughness. Heat transfer in this case, unlike finned sur-
faces, does not increase as a result of increasing heat transfer surfaces, but
rather due to decreasing thermal resistance of the boundary layer when vortex
flow is generated at the wall by the roughness elements.
Intensity of heat transfer and the choice of the optimum surface roughness
depend, to a large extent, on the hydrodynamic regime in the boundary layer.
The most advantageous regime in this regard is turbulent or transient flow
regimes.

94
 HEAT TRANSFER OF BANKS OF ROUGH TUBES 95

4.1. THE SPECIFIC FEATURES OF HEAT TRANSFER

OF ROUGH SURFACES

The heat transfer coefficient on the surface of a smooth plate in a laminar

boundary layer is a ~ °°, and in a turbulent boundary layer a ~ 7°. With
the transition from laminar flow regime to turbulent flow regime heat transfer
changes proportional to the velocity and approaches z'*. However, the transient
flow regime occupies a very small part of the surface and the turbulent regime,
which is going to be discussed below, is dominant.
Investigations indicate that the dependence of heat transfer increases on
Reynolds number and the velocity for a plate and various fluids is identical:
Nu, = Re”.
Due to the fact that thermal conductivities of the viscous intermediate layer
for different fluids are not identical, and due to the effect of external turbulent
fluctuations, heat transfer in flows of water, air, and transformer oil takes place
in different fashions. There are two types of flows in tube banks: in the region
near the wall, and in the intertubular spacing of a turbulent flow.
In the wall region of rough tube banks in crossflow, layers in which veloc-
ity and temperature profiles obey different laws are developed. These are the
intermediate viscous sublayer, transient, and fully turbulent layers in the inter-
tubular spacing. The intermediate viscous layer exists when the elements of
surface roughness are too small compared with the thickness of the boundary
layer.
Despite the fact that the roughness elements generate small vortices in the
vicinity of the tube surface, the turbulent heat transfer g, = — pc,v'0’ in the
intermediate viscous sublayer is insignificant. Depending on the magnitude of
roughness elements and the pressure gradient dp/dx in the transient and turbu-
lent parts of the boundary layer, turbulent heat transfer reaches a maximum at a
dimensionless distance from the wall, 10 < y* < 50. In the intertubular
spacing, the velocity gradient perpendicular to the wall du/dy and the tempera-
ture gradient dt/dy decrease substantially. This leads to a decrease of turbu-
lent heat transfer, notwithstanding the high decree of flow turbulence in this
region.
The root-mean-square value of temperature fluctuations in a rough tube
bank is distributed across the thickness of the boundary layer almost the same
way as the velocity fluctuations. As our measurements show, when the
pressure-gradient parameter p* changes from a negative value (flow accelera-
tion) to a positive value (deceleration) the value

Vit? /9.
decreases substantially. The maximum value of

V t2/9,
96 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

lies in the range of y* from 10 to 50. In the region y* = 10° the value

1? 19.5

amounts to approximately 7% of the maximum value. For example, in a stag-

gered bank 1.25 x 1.25

Vr? [Ss Jmax = 1:3

at Rey2 x 10° and Pr, = 0.7.

The main difficulties in determining heat transfer arise when the interaction
of molecular and molar transfer is considered with rising Pry.
The sudden transition from a region of predominantly molecular transfer to
a region of predominantly molar transfer is characteristic of a boundary layer.
The effect of surface roughness is manifest mainly in the turbulent flow regime
of the thermal carrier. The main heat transfer in this case takes place in a
direction perpendicular to the rough surface.
The following expression, derived from the equation of one-dimensional
heat flowrate, holds for the turbulent boundary layer

1 1 Ex Chall at
Pr, v ) dy+ = (4.1)
(-e +

where 3* = 0/0., 0. = q,,/pCyUs, Pr, = €,/e, is the turbulent Prandtl num-

ber; €, and e, are the coefficients of turbulent viscosity and thermal conductiv-
ity,

Ux = |) apd te

Equation (4.1) shows that when Pr, increases, the relative effect of turbulent
transfer increases, since at Re, = idem the second term of equation (4.1) could
become larger than the first which represents molecular transfer, by several
orders of magnitude.
Thus, the heat flowrate in the laminar boundary layer is proportional to
Pe *, and in the turbulent boundary layer the exponent of Pr; must be higher as
a result of the additional molar transfer in the viscous intermediate layer.
As was stated earlier, heat transfer as a result of the thermal conductivity of
the viscous intermediate layer is different for different thermal carriers. When
Pr; increases, resistance of the viscous intermediate layer to heat transfer in-
creases progressively. According to the calculations in [3, 62] 25% of the total
temperature heat At = ¢,, — ft,in flow of air is due to the viscous intermediate
sublayer. This figure increases to 90% for flow of transformer oil (Pr, = 55).
The main part of the thermal resistance in flow of air is concentrated in the
transient region of the turbulent layer, and for liquids with Pr, > 1 it is concen-
 HEAT TRANSFER OF BANKS OF ROUGH TUBES 97

trated in the viscous intermediate sublayer. Hence, to intensify heat transfer,

one must exert influence on the flow layers near the wall.
Effective increase of heat transfer from the wall to the flow is feasible by
generating vortex flow in the near-wall zone by the roughness elements. In
addition, more effective methods are being explored to intensity heat transfer
with the help of artificial roughness. Due to the variation of thermal resistance
across the boundary layer, the roughness of the surface surrounded by a gas and
viscous liquids must be diverse. Furthermore, the process of molecular- and
molar-transfer interaction must also be studied. Therefore, heat transfer inside
the boundary layer will be discussed.
The equations of the thermal turbulent boundary layer for each of these
regions were solved [62], and the ratio of the total thermal conductivity of the
turbulent flow A to the molecular thermal conductivity of the liquid \ was
obtained.
Table 4.1 indicates that this ratio for gases in a turbulent core is equal to
18, and in liquids with Pr; = 100 the ratio reaches 742 in the buffer zone.
Thus, in order to achieve the necessary intensification of heat transfer in gases,
all the near-wall region must be made turbulent. In liquids with large values of
Pr, only the viscous intermediate layer must be disturbed. This means that the
higher the Pr,, the lower the roughness protuberances at which significant heat
transfer increase is achieved. It is important to note that the highest heat trans-
fer effectiveness in flows of liquids is provided by the regime of partial mani-
festation of roughness (5 < k* < 70).
The principal difference between flows on rough and smooth surfaces is
manifested in the creation of additional tangential stress by roughness elements,
and in the specific transfer features in the region between the roughness ele-
ments. The effect of the additional tangential stress can be observed from the
measured velocity profile.
Due to the significant increase in surface friction, velocity profiles are
strongly displaced. Heat transfer, in the meantime, is determined by vorticity
which is characterized by the same intensity as near-wall turbulence on a
smooth plate.
At k* < 5 the roughness elements are completely immersed in the viscous
intermediate layer and have no influence on the transfer mechanism. At k” >

Table 4.1. Value of A/) as a function of Pr, and y*

é
Pry y A/» yi A/® y Al

0.7 4.7 1.037 28.3 4.74 188 18

5) 4.8 1°53 28.3 Vers) 137 263
50 4.9 6.3 29.3 394 131 2570
100 4.9 12e5 28.2 742 146 5392
98 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

70 a regime of total roughness manifestation is established. In the range of k*

variation from 5 to 70, roughness is only partially manifested. At any value of
k* > 5 the well-known universality of velocity distribution near the wall is
violated, and the velocity profile shifts by a distance determined by the specific
geometry of the roughness. When the elements are densely arranged, a stagna-
tion zone develops between them, and the height k cannot serve as a determin-
ing geometrical characteristic.
Flow around the roughness element in a turbulent boundary layer is accom-
panied by flow separation and re-attachment to the wall and further develop-
ment to the conditions of equilibrium flow. Maximum heat transfer [63] is
observed in the region of attachment, and its value predetermines the mean heat
transfer between the roughness elements and, consequently, the effectiveness of
heat transfer of the rough surface. As experimental investigations show, the
length of the separation (recirculation) zone directly correlates with the value of
heat transfer. It also indicates the formation of a flow structure which possesses
various transport properties stipulated by the Reynolds number, geometrical
form, and height of the obstacle.
Special attention must be paid to the short separation region. Firstly, less
energy is spent on its formation; secondly, the vortices generated in this region
have little energy potential to make the viscous flow on the wall turbulent. The
highest heat transfer rate is observed during the formation of the short separa-
tion region [63]. When the height H of the obstacle is increased, heat transfer
increases slightly. In flows with Pr; > 1 such flow regime is distinguished by a
pronounced heat transfer maximum, provided the optimum distance between
the roughness elements amounts to (8-10)H.
Generally, the heat transfer coefficient of a rough surface is a function of
roughness-element height k, the arrangement form k/s (s is the distance between
the roughness elements), and the values of the Reynolds and Prandtl numbers.
The influence of roughness-element heights is manifested in the direct interrela-
tion of the values of Reynolds number and surface friction. The latter is also
defined by the form and arrangement of roughness elements.
Investigations of heat transfer of rough plates indicate that it intensifies
with increasing Pr, At the present time, generalizing formulas for applied
calculations are lacking. Only the dependence proposed in works [64, 65] are
known.
The heat transfer process in a turbulent flow of a fluid in tubes, down-
stream from the entrance, differs from the heat transfer process of a plate in an
infinite flow, since the boundary layers converge at some distance. Heat trans-
fer for turbulent fluid flow in rough tubes is investigated in detail in several
works [64-77].
Theoretical calculations are experimental investigations of heat transfer in
rough tubes in flows of water and transformer oil [66] show that the viscous
intermediate layer creates the main portion of the total thermal resistance of the
flow. The presence of artificial roughness on heat transfer surfaces causes the
 HEAT TRANSFER OF BANKS OF ROUGH TUBES 99

destruction and turbulization of the viscous sublayer, as a result of which heat

transfer in the tube intensifies substantially. Such an intensification by the flow
with Pr; > 1 increases proportionally with increasing Pr, of the thermal carrier.
Subsequent works [67-69] also showed that heat transfer in tubes in-
creases, under the influence of roughness, for both gas and liquid flows having
Preeti
Heat transfer process in rough tubes, particularly at the entrance, is similar
to the aforementioned heat transfer process on rough plates. The author of work
[67] conducted calculations of thermal resistance of the boundary layer in tubes
using a three-layer model. He concluded that at Pr; = 20, the main part of the
thermal resistance is concentrated in the viscous intermediate sublayer (99%),
and at Pr; = 0.7 it is concentrated in the transition zone layer. Thus, the height
of the roughness element in tubes, as was the case on a plate, must be different
for gases and liquids. The higher the value of Pr,, the lower the relative height
of the roughness elements at which maximum heat transfer is reached.
The dynamics of the obstacles in tube flow were investigated in works [66-
68]. As was the case in experiments with a plate, heat transfer on the wall in the
recirculaton zone at a distance s/H = 10-14 increases 2.6-2.7 times. The
optimum value of the ratio of the pitch to roughness-element height lies in the
bounds 6-11. Artificial roughness employed in modern heat exchangers to in-
tensify heat transfer are mainly rectangular, triangular, trapezoidal, or round
protuberances. Channels with rectangular protuberances have the largest hy-
draulic resistance. By giving the protuberances some streamlining, the friction
coefficient can be reduced; however, this leads to a decrease of the heat transfer
coefficient.
In works [67, 69] various formulas were proposed for the calculation of
heat transfer in rough tubes encompassing a wide range of Re; values.
In the present work, the profiles of velocity and temperature in the bound-
ary layer are generalized in universal coordinates. Heat transfer for forced fluid
flow is viewed as a functional dependence

St =f (Ress, Pry, ke ip) (4.2)

Mainly results of investigations conducted at IPTPE AS LitSSR are used.

4.2. HEAT TRANSFER OF A SINGLE ROUGH TUBE

The investigation of the specific features of heat transfer of a single rough tube
is vital from the theoretical and practical point of view. If the local and mean
characteristics of heat transfer on the surface of a single rough tube are known,
they can be compared with similar data for various surfaces.
Investigations of heat transfer characteristics in a flow of air [70] indicated
that the optimum roughness provides an increase of heat transfer by 60-80%,
100 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

and friction resistance by only 10%. Also investigated were the local and mean
heat transfer of a rough tube for flow of water at Re, range from 4 Xx 10° to
1.6 x 10° and turbulence of the external flow from 1 to 7% [71]. The height of
the roughness elements was varied from 0.15 to 1.2 mm. The limiting condition
corresponded to the case q,, (x) ~ const. It was found that when Re; and the
turbulence of the flow are increased, transition from the laminar boundary layer
to the turbulent boundary layer on the rough tube surface occurs more rapidly
than on a smooth tube.
Fig. 4.1 shows that the flow regimes are determined mainly by the Reynolds
number and turbulence of the external flow; however, surface roughness also
plays a major role in this case. Additional turbulence of the boundary layer
caused by the roughness elements influences heat transfer. The intensity of this
influence depends on the ratio of roughness-element height k to the thickness of
the boundary layer 6. When k is significantly smaller than 6 the velocity fluctua-
tions caused by the roughness show no appreciable influence on heat transfer. If
the values of k and 6 are comparable, the velocity fluctuations in the boundary
layer in conjunction with the external turbulence increase the intensity of heat
transfer.
At Tu = 4%, separation of flow from the rough tube commences at a
smaller ¢ (130°) than separation of flow from a smooth tube. This is caused by
the increase in boundary-layer thickness as a result of flow separation from the
roughness elements. Hence, heat transfer in the rear side of the rough tube

0 30 60 90 120 150 7)

Figure 4.1. Influence of roughness and flow turbulence on the local heat transfer of a tube: Ik =
0, Rey= 1.28 x 10°, Tu = 4%;2)Rep =3 x 10°, Tu = 4%; k = 0.15 mm; 3) Rep= 1.55 x
10°, Tu = 7%, k = 0.15 mm. Ky = NuyRej °5Pr7 937 (Prp/Pr,,) 7925,
 HEAT TRANSFER OF BANKS OF ROUGH TUBES 101

(Fig. 4.1) is less intensive than in the case of a smooth tube. This is caused by
the increase of boundary layer thickness and by the widening wake behind the
tube.
The mean heat transfer of tubes increases with increasing relative rough-
ness k/d; at lower Re, transition to a critical flow regime is also observed.
Analogous results were also obtained in work [70].
The mean heat transfer in the region of subcritical Re, values, for moderate
heights of roughness elements (k = 0.15 mm), increases insignificantly in
comparison with heat transfer of a smooth tube. This is the cause of the ele-
vated level of total velocity fluctuations in the laminar boundary layer, wherein
velocity fluctuations generated by the roughness elements prevail. When Re; is
increased, the parameter kK” = u. k/y in the laminar boundary layer reaches a
critical value causing a transition of the boundary layer from laminar to turbu-
lent. The Reynolds number for a rough tube is also a function of the roughness
parameter of the whole tube, i.e., it depends on the mean coefficient of the
friction resistance. With the increase of Re,, the dynamic velocity u. and,
hence, the roughness parameter k*, increase. However, a substantial increase
in us leads to the increase in k.
Therefore, the transition to the critical flow regime on a rough tube (k =
0.15 mm) commences at considerably lower values of Re; (Re; = 10°) than on a
smooth tube if the other conditions remain unchanged (Re; = 2 X 10°).
In the critical flow regime of water (Pr; ~ 6), with the increase of the
height of roughness elements to 0.15 mm, heat transfer rises by approximately
85%; and with an increase to 1.2 mm it rises by 100%. Therefore, in the
critical flow regime of water, surfaces with small relative roughness (kK = 0.15
mm) are the most effective for heat transfer.
The intensity of heat transfer is determined by the overall turbulence in the
boundary layer. The effect of this turbulence in the transient zone on the rough-
ness of the tube at Tu = 1% is similar to the effect of the overall turbulence on
a smooth tube at Tu = 4%.
In the region of critical Re, values, the roughness elements (k = 0.15 mm)
create some turbulence which is characteristic of this roughness. An additional
influence on the overall turbulence is exerted by the turbulence of the incoming
flow which determines the intensity of heat transfer.

4.3. LOCAL HEAT TRANSFER OF BANKS

OF ROUGH TUBES

Large scale investigations of heat transfer of tubes in crossflow are being con-
ducted at IPTPE AS LitSSR. The experimental data [72] showed that roughness
significantly affects both the location of boundary layer transition from laminar
to turbulent and its separation. As a result of using surface roughness, the
length of the turbulent boundary layer increases and heat transfer intensifies.
102 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Therefore, heat transfer of tubes in crossflow can be substantially increased

with the help of roughness elements.
The results of our measurements [72] indicate that when the heights of the
roughness elements are increased, the transition of flow regime in the boundary
layer from laminar to turbulent occurs appreciably earlier. For example, in
bank 2.0 x 2.0 in flow of transformer oil at Re; = 1.4 Xx 10’ and k = 0 (Fig.
4.2 curve 1), the laminar boundary layer is preserved to the angle ¢ = 100° at
which it separates from the wall. In the meantime, in the case of a single tube in
flow, the laminar boundary layer is separated at ¢ ~ 82°. The application of
fine roughness at the same Re; leads to the generation of turbulent boundary
layer. Curve 2 in Fig. 4.2 corresponds to the value k/d = 6.67 X 10° at
which transition of flow from laminar to turbulent commences at ¢ = 90° and
its separation at ¢ ~ 135°. Fine roughness elements substantially increase the
heat transfer coefficient along the tube perimeter, apart from a small section at
g = 80-100° where the heat transfer coefficient in the transition region is a
minimum. A further increase of the height of roughness elements to k/d =
40 x 10° (Fig. 4.2, curve 3) leads to a shift of the transition point from the
laminar boundary layer to the turbulent layer to g = 60°. The local heat
transfer in a significant part of the frontal tube side in the range ¢ ~ O0-70° is
less than the heat transfer of a smooth tube in a bank.
This can be explained by the small vortex stagnation zones in the fluid
which are formed between the large roughness elements. Comparison of curves
2 and 3 shows that k/d is best increased to a certain value, following which the
local heat transfer coefficient becomes less than for a smooth tube in a bank. An

0 50 60 90 120 ~— 50 Y
Figure 4.2. Local heat transfer in an inner row of a staggered bank 2.0 x 2.0 at Rey = 1.4 x 104
and Pry=er84: 1) k = 0; 2) k/d = 6.67 x 107%: 3) kid = 40 x 107. Kyf = Nu,Re;
etch ™ Pr;f °26
(Pr//Pry) :
 HEAT TRANSFER OF BANKS OF ROUGH TUBES 103

0 30 6 M9 120 0 ¥
Figure 4.3. Local heat transfer in an inner row of a staggered bank 1.25 x 1.25 at Rey = 2 X
10°, Prp= 0.7; 1) k= 0; 2) kid = 8 X107>. Ky—as
per Fig. 4.2.

increase in roughness in the rear side of the tube also leads to a decrease in the
local heat transfer in the region g¢ = 120-180°, despite the fact that separation
of the boundary layer commences at ¢ ~ 165°. Therefore, the large heights of
roughness elements, heat transfer intensifies in the turbulent part of the bound-
ary layer only; and for small heights, heat transfer intensifies along the whole
perimeter of the boundary layer. In order to obtain the same degree of intensifi-
cation in a flow of air, the height of the roughness elements must be twice as
much, on the average.
Let us consider the effect of k on the local heat transfer of tubes of compact
banks 1.25 X 1.25 and 1.25 x 0.935. Heat transfer in bank 1.25 x 1.25 (Fig.
4.3) intensifies along the whole tube perimeter. As can be seen from the figure
(curve 2), fine roughness elements can increase heat transfer significantly in the
region g = 0-120°. Starting from ¢ = 180° (the rear critical point), the local
heat transfer increases as the laminar boundary layer grows downstream of the
main flow. Due to the low velocity in the recirculation zone, heat transfer
increases slightly.
Stagnation zones are formed between the roughness elements and heat
transfer decreases in the frontal side of the tube bank 1.25 x 0.935 (Fig. 4.4).
This is the reason for the decrease in heat transfer in the region g ~ 30° at
kid = 40 x 107° (curve 3) in comparison with heat transfer at k/d = 6.67 x
107? (curve 2). When ¢ increases, thermal and hydrodynamic boundary layers
and heat transfer decrease. A comparison of local transfer at two values of
roughness clearly demonstrates its effect on the point of transition of flow
regime in the boundary layer from laminar to turbulent. When k/d= 6.67 x
10~*, the transition occurs at y ~ 70°, and when k/d is increased to 40 x Ome
it shifts to ¢ ~ 50°, making the flow in the boundary layer turbulent. The
104 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

OU IIR GR. AGG) I

Figure 4.4. Local heat transfer in an inner row of a staggered bank 1.25 x 0.935 at Rep = 4 x
10* and Pry = 84: 1) k = 0; 2) kid = 6.67 x10~*; kid = 40 x 107°. Ky—as per Fig. 4.2.

presence of small-size roughness elements (k/d = 6.67 xX 10~*) leads to an

increase of heat transfer on a large section of the tube surface (from ¢ = 15 to
y = 180°), whereas an increase of roughness to k/d = 40 X 107° leads to an
increase in heat transfer on a smaller section of the surface. The effective
influence of surface roughness is observed at g = 60-150°.
In order to determine the optimum height of surface roughness, the varia-
tion of the dependence K = f(k/d, ¢) was considered along the tube perimeter
for a change of Pr, from 80 to 300. It was found that the optimum roughness
height in the region of flow acceleration is k/d = 10 x 107’, and in the region
of deceleration k/d = 25 x 10°°. Analysis of experimental data showed that,
depending on the flow conditions, it might be expedient to vary the roughness
height along the tube perimeter. However, the manufacture of such a tube is
impractical; therefore, it is recommended to adopt a mean optimum roughness
height along the tube perimeter.
Figs. 4.2-4.4 show that the character of local heat transfer of a rough tube
is dependent not only on the heights of the roughness elements, but also on the
bank pitches a x b. From this viewpoint, banks with small pitches, for exam-
ple from 1.25 x 0.935 to 1.25 x 1.25, are the most optimal.
The main purpose of studying the effect of roughness elements on heat
transfer was to determine the optimum value in banks as a function of k/d, Pry,
anda x b.
With this in mind, the temperature profile was measured at different dis-
tances from the surfaces of smooth and rough tubes in flows of air and trans-
 HEAT TRANSFER OF BANKS OF ROUGH TUBES 105

former oil. Presented in Fig. 4.5 are data of such investigations in a thermally
efficient bank 1.25 x 1.25. The data is presented as dimensionless variables
© = f(np, ¢) where © = (¢ — t/t,— t,), n¢ = y/6, 5, — thickness of the
thermal boundary layer. To plot the temperature profile, (y; + k/2) was taken
as a reference point for the coordinate y where y, is the distance determined
from the top of the roughness element. In tube banks, the pressure gradient
downstream of the flow changes substantially; therefore, it is convenient to
adopt the dimensionless parameter p* = p(dp/dx)/7*” p'” as its characteristic
value.
The Prandtl number significantly influences the character of the tempera-
ture distribution in the boundary layer. As Fig. 4.5 shows (curves J-3 corres-
pond to flow air, and curves 4-6 to flow of transformer oil), the thermal bound-
ary layer in a flow of a viscous fluid is substantially thicker than in flow of air.
The negative values of p* correspond to the acceleration of flow, and the
positive value to its deceleration. For a decelerating flow of air (curve 3, low
values of Pr,) and viscous fluids (curve 6, very large values of Pr,), the thick-
ness of the thermal boundary layer is greater compared with an accelerating
flow. Consequently, to intensify heat transfer in flows of viscous fluids at Re, =
idem, rough surfaces of smaller height must be employed. It is useful also to
decrease the distance between the tubes to an optimum value, since the main
flow in the intertubular spacing does not partake in the heat transfer and the
energy needed for the movement of the thermal carrier is expended ineffici-
ently.
Of important practical interest is the study of the influence of the roughness
elements of different heights on heat transfer near the wall, particularly in the
viscous intermediate sublayer and in the transition region of the boundary layer.
The character of heat transfer in the boundary layer can be determined by the
temperature profiles expressed in the coordinates o* = fi") which acquire the
following form for a viscous intermediate layer

9+ = Pry (yr+ 5) wel (4.3)

and outside the limit of this layer, by the dependence

(>r+ 5) Uy
Ore ag yl el iy (Pik? sp*) (4.4)

where A, = Pr,/x, Pr, is the turbulent Prandtl number, x is a universal constant

in the distribution of the mixing-path length for both smooth and rough sur-
faces. Presented in Fig. 4.6 are experimentally obtained temperature profiles in
an inner row ofabank 1.25 x 1.25 at Rey = 2 x 10°, Pr;= 0.7 (curves 3-5),
and Pr, = 84 (curves 6-8) for various characteristic values of the dimensionless
106 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

= 2
[fas 7)

Nr
Figure 4.5. Temperature profile at Rep = 2 X 10°, k/d = 5.6 x 1073, Pry = 0.7 and the
following
p*: 1) —2.5 x 1073, 2) 0; 3) +265 x 107%, at Pr, = 84, and the following p™: 4)
—4 x 10-3, 5)0; 6) +265 x 107°.

O=-O70744

2 4680 2 4 680 2 4 Libs

ly k/2Mu,

Figure 4.6. Universal temperature profiles at Rep = 2 x 10°, k,/d = 5.6 X (eed Pry = 0.7. ke
and p*, respectively: 3) 124.4, —2.24 x 10-7; 4) 113.4, 0; 5) 72.2, +10.4 x 10 3; at efi
84, kt and p* respectively: 6) 80, —1.12 x 1077; 7) 70.5, 0; 8) 45, +36.2 x 1073.
 HEAT TRANSFER OF BANKS OF ROUGH TUBES 107

roughness k* and parameter of the pressure gradient p*. Curve J corresponds

to the temperature profile in the viscous intermediate sublayer on a smooth
surface for a non-gradient flow according to the equation

$+ =Pr,y* (4.5)
and curve 2 corresponds to the same conditions and flow according to the
equation

Ot =4.2In yt +3.8 (4.6)

The region with a logarithmic distribution of temperature profile is described by

Eq. (4.4). Temperature profiles in conditions of pressure gradient differ signifi-
cantly from the corresponding profiles for a non-gradient flow. At p” < 0,
curves 3 and 6 in Fig. 4.6 lie above the temperature profile for non-gradient
flow, and curves 5 and 8 at p* > 0 lie below.
According to the experimental data in Fig. 4.6, the slope of the curves
d* = fly") depends on the coefficient A,. If it is assumed that Pr, in the region
of y*, where the experimental data are obtained, varies only slightly, then the
value of x found from A, = Pr,/x is substantially larger than what is usually
assumed for calculations (x = 0.39-0.41). The value of the constant c, de-
pends mainly on the Prandtl number and, to a lesser extent, on the type of
roughness and pressure gradient.
As can be seen from the results presented in Figs. 4.2-4.6, surface rough-
ness substantially changes the character of local heat transfer on the tube sur-
face in a bank and affects the thickness of the thermal boundary layer. The
temperature fluctuations in the region of flow acceleration, generated by some
roughness elements, attain a maximum value at y* ~ 80. In the region of flow
deceleration, the maximum of the fluctuations is shifted to y* 10°. In the
investigated interval of non-gradient roughness values k* ~ 0-200, the maxi-
mum temperature fluctuations take place at the start of the logarithmic law of
temperature distribution in the boundary layer. It follows from this that the
intensification of heat transfer is brought about in the immediate proximity of
roughness elements.
Measurements results (Fig. 4.2-4.6) and the results of the investigations
presented in work [5] serve as a basis for the determination of effectiveness of
roughness in functional forms:

St/Stg=f (k*, Re;, Pr,) (4.7)

St/e,=f(kt, Rey, Pr) (4.8)

where St and St, are the Stanton numbers presented in Fig. 4.7 for rough and
smooth tubes, respectively. It is interesting to note the amount of increase of
108 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

St/St,

14

42 Re, = 600-4:10°
p20 =50

10 |
0 50 100 k.
Figure 4.7. Increase of heat transfer in rough tube banks in flows of air and viscous liquids.

heat transfer of banks of rough tubes in comparison with smooth tubes when the
dimensionless height of roughness elements k* varies within a certain range.
This range encompasses the values of k* from 1 to those in excess of the height
in a regime of complete manifestation of roughness in flows of gases and vis-
cous liquids. As is evident from Fig. 4.7, when the banks are densely arranged
and the roughness elements have a pyramidal shape, heat transfer in the regime
of complete roughness manifestation increases by 35%. With further increase of
roughness from k” ~ 100, heat transfer stays unchanged (with the exception of
a slight rise as a result of the heat transfer surface increase due to roughness). In
a flow of transformer oil, in a regime of partial manifestation of roughness up
to k* = 30, heat transfer rises by approximately 55%. From Fig. 4.7, it
follows that the intensity of heat transfer in flow of air is less, at the same Re,
than in viscous liquids, this intensity being reached at a considerably larger
height of roughness elements.
When the height of roughness elements is increased, the friction coefficient
c, substantially increases [5]. Although c; for banks in turbulent flow is small
compared with the pressure resistance coefficient c,,, it nevertheless, plays an
important role when the heat transfer mechanism near the surface is studied in
the form of effective heat transfer 2St/c; The analyzed experimental data in-
cluded both the height of roughness elements and the equivalent sand roughness
k, in order to compare the data with the investigations of other authors. The
methods of determining k, are given in [5] and others. Fig. 4.8 shows the
function 2St/c, = f(k”, Pr, at a constant Re, = 2 X 10°. The pressure gradient
parameter p* varies in a wide range (the negative values of p* correspond to
the acceleration of flow, and the positive values to the deceleration of flow). In
a flow of transformer oil in a regime of partial roughness manifestation at kK; =
 HEAT TRANSFER OF BANKS OF ROUGH TUBES 109

20, the ratio 2St/c; is maximum. With a further increase of k; , the heat transfer
effectiveness decreases, and at k;; ~ 80 it becomes less than the value typical
for smooth walls. The function 2St/c; for air is distributed differently: it de-
creases gradually when k; increases. This indicates that the intensification of
heat transfer in a flow of air using rough surfaces is not effective. As increase
in heat transfer can be achieved only for large values of k; . As shown by Fig.
4.8, this value is k; =~ 120-150. From the expression

k,=kty (Vs/Ux) (4.9)

at a given working air temperature, v, is determined, and from the working

interval of Re;,,u. is found from the function c; = f{Re, k,); finally, the optimum
height of roughness element k, is computed.
It was shown in Chapter 2 that the turbulent boundary layer on a tube in a
bank occupies a large part. The intensification of heat transfer using surface
roughness is observed in the transient and turbulent flow regimes in the bound-
ary layer. Therefore, of important practical interest are the equations for the
calculation of local heat transfer in the region of heat transfer intensification by
the elements of roughness. It is convenient to generalize the local heat transfer
as a dependence of the local heat transfer coefficient [5]

C, =at Ress (4.10)

on the parameter of pressure gradient

(4.11)

1’ 2 4680 2 4 6 BK
Figure 4.8. Dependence of effectiveness in an inner row of a staggered bank on the dimensionless
height of sand roughness in flows of air and transformer oil.
110 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

gt = f — .-dy (4.12)
S

v=t,-t0=t, - t,. The Stanton number St = a@/pc,u, is obtained from

experimental data, and Re,,. from the velocity in the external part of the bound-
ary layer and thickness of energy loss. Such an approximation of the experi-
mental data (Fig. 4.9) permits the derivation of equations which are convenient
for the computation of local heat transfer. Curve J corresponds to a smooth
surface. When the height of roughness elements increases, several curves,
parallel to curve J, are obtained. Curve 2 corresponds to the variation of k*
from 20 to 200, and is obtained for a wide range of variation of the parameter
I’, (from acceleration to deceleration of the flow).
When the experimental data on the local heat transfer in the zone from the
turbulent boundary layer up to the separation point, are generalized, the follow-
ing analytical expressions are obtained:
for a smooth tube bank in a flow of air

St = 0.0275 Rezo (4.13)

and for a rough tube bank in a flow of air

St = 0.0465c. (kt) Rese (4.14)

for a smooth tube bank in flows of viscous liquids

St = 0.022 Reso Prz%-§ (Pr,/Pr,,)%25 (4.15)

for a rough tube bank in flows of viscous liquids

St =0,0375c (kt) Rezo5 Pr7=-S (Pr,/Pr,,)°25 (4.16)

Figure 4.9. Dependence of the local heat transfer coefficient on the parameter of pressure gradient
in an inner row of staggered banks: /) smooth; 2) rough tubes.
 HEAT TRANSFER OF BANKS OF ROUGH TUBES 111

Here, c,(k;’) is a function of k* and has the form:

Cs (kt)=1.7 10-4k+ +0.0275 (4.17)

Equations (4.13)-(4.17) are valid in the range of k; from 20 to 150,
the pressure gradient parameter I’, from —1.0 to 0.5, and Re;.. from 400 to
2000.
There are no equations, at the present time, for the practical calculation of
the effectiveness of heat transfer of the turbulent boundary layer on a rough
surface in conditions of negative pressure gradient in transition to the positive
gradient. In work [64], the following equation is proposed for the case of a
plate in flow with complete roughness manifestation:

ca
_28t —+A=c(kt)™ Prt (4.18)
l Cyr/2

The values of the constants and the exponent in equation (4.18) depend on the
type of surface roughness. If, according to work [64], the friction coefficient c,
is replaced by the coefficient of resistance €, then the following data are ob-
tained in the case of sand roughness: A = 8.48, c =5.19, m = 0.2, = 0.44.
According to work [3], in the case of a plate with pyramidal roughness, the
constant c in equation (4.18) changes along the length of this plate from 2.7 to
5.4 at x =O and x = 1 m, respectively; and the values of A = 6.5, m = 0.2,
and n = 0.63.
In work [65], an equation which generalizes the experimental data on heat
transfer of various surfaces is given for densely arranged roughness elements:

| (4.19)
2.12 In (8*/k) +0.55 /k+
= (Pr,y?/?—0.2)
+9

which agrees well with experimental data obtained in the regime of complete
roughness manifestation at Pr; < 10. For the regime of partial roughness mani-
festation, it is recommended to employ linear interpolation between the values
of heat transfer coefficients of smooth and rough plates in the regime of full
manifestation.
Great attention has been paid, when generalizing experimental data of tube
banks with densely arranged pyramidal roughness, to the derivation of a for-
mula for the calculation of heat transfer effectiveness. For this purpose, hydro-
dynamic R(k*) and thermal G(k*, Pr,) functions of surface roughness have been
used in logarithmic velocity distribution [5] and temperature. Fig. 4.10 indi-
cates the distribution of these functions with varying k*. Curve / in Fig. 4.10a
corresponds to the equation
112 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

AT I TR Oy?
Figure 4.10. Dependence of the hydrodynamic (a) and thermal (b) functions on the dimensionless
roughness height in tube banks.

RE)
= 2 sink 5.5 (4.20)

for the regime of full roughness manifestation, and curve 2 for sand roughness.
The function R(k*) depends not only on k*, but also on the shape of the
roughness elements and the relative pitch s/k. The results obtained for a smooth
hydraulic zone (k* < 5) correlate well with the theoretical curve 7. With a
further increase of k*, the function R(k*) decreases and stabilizes in the region
of full roughness manifestation. Experimental data in the range of ¢ from 60 to
100°, i.e., to the site of boundary layer separation (curve 3) can be well ap-
proximated by the equation

20.3
RG =a +03 +6.2 (4.21)

The thermal roughness function G(k*, Pr,) is dependent on k* (Fig. 4.10b),

Pr,, and the geometrical parameters of roughness. The experimental data in the
range of p* from —2 x 10°’ to +25 x 10°? are well approximated by the
equation

G(k+, Pry) =4.9 (k+)0? Pro (4.22)

Equation (4.22) correlates with the data presented in works [64, 73]. On the
basis of equations (4.21) and (4.22), a formula is derived for the computation of
effectiveness of local heat transfer in rough tube banks, which takes into ac-
count the change of roughness parameters and pressure gradient:
 HEAT TRANSFER OF BANKS OF ROUGH TUBES 113

Cy
2St oy
{4.9
Va (k+)02 Pr;prod— Ota
(A*)*:?
20.3
ec.

Formula (4.23) is valid in the range of k* from 20 to 200, p* from —2 x 107?

to +25 x 107°, and Pr, from 0.7 to 100.

4.4. MEAN HEAT TRANSFER OF BANKS

OF ROUGH TUBES

Despite the fact that heat transfer of smooth tube banks in crossflow was exten-
sively investigated, the effect of surface roughness on heat transfer in them has
not been sufficiently studied.
As the analysis of the works of other authors indicates, the available data
on heat transfer of rough tube banks were obtained mainly for a flow of gas and
give no answers to the questions concerning the intensification of heat transfer
in viscous fluids.
Work [74] is considered one of the first to investigate the mean heat trans-
fer in banks of rough tubes at values of Re, up to (7-8) x 10*. Roughness was
applied to the tubes with the help of special knurling tools which formed trape-
zoidal protuberances on the surface. Heat transfer of banks 2.0 x 2.0 with
staggered and in-line tube arrangements were studied in the region of variation
of the roughness dimensionless height k/d from 0 to 6.4 x 107’. It was estab-
lished that, in staggered banks at Re; = 5 X 10° and k/d = 6.4 x 10°’, the
heat transfer coefficient increases 1.8 times. In accordance with the law of
resistance change, heat transfer also changes. However, this work was con-
ducted for narrow range of Re; in a flow of air.
In work [75], heat transfer in a bank 2.0 X 1.4 with staggered and in-line
tube arrangements in the region of Re; from 2 Xx 10* to4 x 10° and k/d from 0
to 9 x 10°’ was investigated. It was established that, under the influence of
surface roughness, the character of heat transfer and flow in the boundary layer
changes: Re,, at which transition to a regime of predominantly turbulent flow
occurs, decreases from 2 x 10° for a smooth tube to 2.5 x 10* for a tube with
kid = 17 x 107°. Heat transfer of a rough surface is higher by 50% than a
smooth one. When the value of Re, in flow is below the critical, surface rough-
ness does not intensity heat transfer.
The mean heat transfer of a staggered bank 2.06 x 1.37 was investigated
in work [76] in the region of Re; from 10* to 10°, for roughness height variation
in the region k/d from 0.6 x 10~° (a technically smooth surface) to 3 x 107°.
In the present work, equations for the calculation of mean heat transfer of
rough tube banks are derived on the basis of experimentally determined local
heat transfer coefficients. The mean heat transfer of rough tubes depends on
Re,, tube arrangement in the bank, physical parameters of the thermal carrier,
114 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

and the shape and dimension of the roughness elements. This dependence can
be expressed in a dimensionless form:

Nuy=/(Re;, Pree Pr Prasad Sod.) 2k5) (4.24)

The results of the investigations are generalized by the similarity equation

Nu, =c (a/b) Re! Pr? (Pr,/Pr.)? (k/d)! (4.25)

in which the physical parameters are evaluated at the temperature of the incom-
ing flow ¢, the mean velocity in the minimum cross-sectional area of the bank
i, and the external tube diameter d. The exponents of the similarity variables,
the relative roughness k/d, and the relative pitches a and b, and also the con-
stants in equations (4.24) and (4.25) were determined from experimental data.
Investigations were conducted with banks of rough tubes having a xX b =
2.0 x 2.0, 1.25 x 0.935 for the range of k/d values from 0 to 40 x 10a: [55
77]. The effect of the parameters k/d, a, and b on heat transfer of these banks
was also determined.
In Fig. 4.11, data are presented on the mean heat transfer of the first and
inner rows of a staggered bank 2.0 x 2.0, at low and medium range Re; values,

Figure 4.11. Mean heat transfer of a staggered tube bank 2.0 x 2.0: J-3) inner row, 4-6) first
row; k/d = 0; 6.67 x 1077 and 40 x 1073, respectively.
 HEAT TRANSFER OF BANKS OF ROUGH TUBES 115

as a function of k/d changes from 0 to 40 x 107’. Fig. 4.11 shows that rough-
ness begins to affect heat transfer at Re; ~ 2 x 10°. At this value of Re, a
transient flow regime commences corresponding to the break in curves of heat
transfer. The exponent m for Re; increases to 0.68 for the first tube rows of the
bank, and to 0.7 for inner rows. Up to the break in the curves, heat transfer of
rough tubes in banks is generalized by a single curve. Following the break in
the curves, heat transfer is stratified, depending on the height of surface rough-
ness. When k/d is increased, the difference in the mean heat transfer between
smooth and rough surfaces rises. In an inner row, at Re, = 2 X 10°, the value
of K, increased by 4% for k/d = 6.67 x 107° and by 11% for k/d = 40 x
10°; whereas in the first row K; increased by 9 and 20%, respectively.
The thickness of the thermal boundary layer in a flow of viscous fluid is
considerably smaller than in a flow of air at Re, = idem, and it decreases with
increasing Re,. Consequently, the effect of roughness, for a constant element
height, on heat transfer at low and medium range Re; values is only apparent in
flows of viscous fluids. In a flow of air, the effect of roughness is manifest at
higher Ref{Re, = 5 X 10° — 7 x 10*) (Fig. 4.12). The exponent m of Re, for
the first row of tube bank 1.25 x 1.25, following the transient flow regime at
kid = 8 x 107°, increases to 0.7, and for an inner row it increases to 0.82.
Heat transfer in an inner row of rough tubes, compared with heat transfer in a
bank of smooth tubes, increases by 14% at Re; = 10° and by 75% at Re; = 10°.

m ces SEES at

L ge lll | Wy
:
|

425* {25 | | |
| SAS

BE,
he }

9
\ a. |
S 3
Slwecd| | lar A
a
6 LA

eae
ae
otOe
SS
of
peat LE 6| ail| av?
ee. ye
aL : —}—-——-++-—_++ et —
aa

yA
74

6 D-O—+ E | eee a! + |
(@)

oe 6 bao Be 4 6 BW Re
4-6) first row;
Figure 4.12. Mean heat transfer of a staggered bank 1.25 x 1.25: J-3) inner row,
kid = 0,3 x 1073, and 8 x 107%, respectively.
116 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

The influence of roughness on the mean transfer of a bank 1.25 x 0.935 is

manifest to a greater extent (Fig. 4.13) and at a lower Re, than in a bank 2.0 x
2.0. The exponent of Re, following the transient flow regime increases to 0.74
for the first row of bank 1.25 x 0.935 with k/d = 40 x 1073, and to 0.77 for
an inner row. Heat transfer in this bank increases noticeably. For example, at
Re, = 10* and k/d = 40 x 10° heat transfer in an inner row of the bank is
larger by 53% than heat transfer at k = 0.
Fig. 4.14 shows a comparison of the obtained results on the mean heat
transfer with data from well-known references in a range of Re, from 10° to
2 x 10°, a from 1.25 to 2.0, and b from 0.935 to 2.0 (Fig. 4.14).
As is evident from Fig. 4.14 (curves J and 2), the agreement of the data of
the investigation of bank 2.0 x 2.0 with the results of work [74] is good in the
region of Re, up to 7 X 10*. The effect of roughness on heat transfer is
insignificant. In the region of Re; up to 4 x 10*, bank 1.25 x 0.935 (curves
3-5) has better heat transfer characteristics, and the influence of roughness is
more apparent than in bank 2.0 x 2.0. In the range of high Re, values, the
obtained results agree well with the data of work [76].

Lee aeRO 2 a Byay 2 4 Re

Paw 4.13. Mean heat transfer of a staggered bank 1.25 x 0.935: /-3) inner row, k/d = OF3rx<
10°, and 8 x 1Ome: 4-5) first row; k/d = 0 and 8 x 105%; respectively.
 HEAT TRANSFER OF BANKS OF ROUGH TUBES 117

10° Re.3
Figure 4.14. Comparison of mean heat transfer of staggered banks for different pitches and rough-
ness (from data of the present work): J, 2)a x b = 2.0 x 2.0, k/d = 0 and 40 x 10-7; 3-5)
a xX b = 1.25 x 0.935, k/d = 0, 15 x 1073, and 40 x 1077; 6, Na x b = 1.25 X 1.25, k/d
= Oand 8 x 1077; 8, 9)a x b = 2.0 x 1.37 k/d = 0 and 6.4 x 107? [76]; 10, I])a x b =
2.0 x 2.0, k/d = 0.04 x 107? and 1.2 x 1073 [74], respectively.

On the basis of generalization of experimental data, the following equation

is derived for the calculation of the mean heat transfer of banks of rough tubes
in the Re; range from 10° to 10°

Nu; =U (jp Re, Pree (Pro/Pro\eo(kid et (4.26)

and in the Re; range from 100.2 <M10°

Noy 0) 1{a/by Rey’ Pry (Pai Pr) * (he (4.27)

The mean heat transfer can be computed from these equations with an accuracy
of +15% for a in the range from 1.25 to 2.0, b from 0.935 to 2.0, and k/d
from 6.67 X 10°° to 40 x 10°.
118 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

The exponents of a/b, Re, and k/d in Eqs. (4.26) and (4.27) are determined
by generalizing data on mean heat transfer obtained in the present work and in
works [72, 74, 76, 77] for banks 2.0 x 2.0, 2.0 x 1.4, and 2.06 x 1.37.
For the determination of the effect of roughness height on heat transfer in
flows of air and highly viscous fluids, the functional dependence c, = f(k/d,
Re,, Pr,) has been compiled (Fig. 4.15) where the coefficient

c, = Nu,/Nu

expresses the degree of heat transfer intensification by a rough surface in com-

parison with a smooth one.
Curves /-3 in Fig. 4.15 are obtained as a result of investigating compact
banks of rough tubes in flows of viscous fluids at k/d = 6.7 x 107°, 15 x
10°°, and 40 x 107’, respectively; and curves 4, 5 in a flow of air at k/d =
10°* and 8 x 10°’. It follows from Fig. 4.15 that surface roughness in a flow
of viscous fluid begins to increase the mean heat transfer at lower values of Re,
(Re, = 10°, curves /-3) than in a flow of air (Re, ~ 10*, curves 4, 5) and c, has
higher values.
The increases of heat transfer in staggered tube banks, with the help of
roughness, causes hydraulic resistance coefficient to rise. This is because some
energy of the moving fluid is expended to generate local vortices. Expressing
the level of rising hydraulic resistance by the ratio Eu,/Eu, one can obtain a
parameter which is important for the design of heat exchangers:

_ _Nux/Nu
fer Eas (4.28)

Analysis of the function 7 = f(Re, Pr, k/d, a x b) indicates that the

magnitude of 7, in banks witha x b = 2.0 X 2.0 ina flow of viscous fluid at
Pr; = 80-300 and k/d = (0-40) x 10~*, decreases from 1 to 0.92. In compact
banks with a X b = 1.25 xX 1.25 and in highly compact banks with a x b=

Figure 4.15. Correction for the effect of roughness when calculating heat transfer of a tube in a
staggered bank: J-3) k/d = 6.67 x 107°, 15 x 1073, 40 x 1073, Pry = 84; 4, 5) k/d = 1 x
107°, 8 x 1073, Pry = 0.7, respectively.
 HEAT TRANSFER OF BANKS OF ROUGH TUBES 119

1.25 x 0.935, n decreases from | to 0.6 at Re; = 10° and from 1 to 0.75 at
Re, = 10°-10°. The values of 7 in a flow of air in the indicated ranges of
characteristic parameters is even lower. To intensify heat transfer in flows of
highly viscous fluids, it is expedient to employ a surface roughness of height
k = 0.4-0.8 mm for a dense arrangement and of height k = 0.8-2 mm for a
flow of air.
 CHAPTER

FIVE
HEAT TRANSFER OF BANKS OF FINNED TUBES

When designing heat exchangers, special attention must be paid to their eco-
nomical operation, thermal efficiency, and compactness.
For the construction of effective heat exchangers with developed surfaces,
finning is widely used. This is imperative in order to achieve equality of heat
fluxes from the air side and the liquid side. Usually, the heat transfer coefficient
of air is substantially less than the heat transfer coefficient of liquids. However,
the heat transfer coefficient in flows of high viscous liquids, such as trans-
former oil, aviation oil, and other oils is less than in a flow of water. The choice
of the thermal carrier depends on the specific requirements demanded by the
application of heat exchangers. Heat transfer intensification, flow turbulization,
and the use of rough surfaces do not achieve the equality of heat transfer
mentioned above.
Therefore, two options are left: change the velocity of the thermal carrier
or the surface of heat transfer. However, substantial increases in velocity are
not permissible sometimes, since the specific requirements of the heat transfer
process might be disturbed. Also, high velocities of the thermal carrier lead to
increased pumping losses. Therefore, the best way to enhance heat transfer is to
employ finned tubes in heat exchangers.
Finning assumes various forms (Fig. 5.1). Continuous radial fins are
widely used. The shape of the fins are chosen depending on the field of heat-
exchanger application. The big strides made in the manufacture of fins of vari-
ous shapes allow an increase in their effectiveness.
Let us analyze the specific features of heat transfer in banks of finned
tubes. For comparison purposes, the values of the mean heat transfer coefficient

120
 HEAT TRANSFER OF BANKS OF FINNED TUBES 121

Figure 5.1. Typical shapes of finned tubes: a) radial fins; b) straight fins; c) continuous corrugated
fins.

for smooth tube banks in flows of air, water, and oils in the cases of natural and
forced convection are presented in Table 5.1.

5.1. HEAT CONDUCTION IN FINNED TUBES

During the propagation of heat flowrate from the internal tube side to the
external side, the temperature drops towards the tip of the fin. When the height
of the fin is increased, the temperature of its tip reaches the temperature of the
external fluid flow, and any further increase of fin height is not worthwhile.
Hence, the limiting condition for heat transfer rise of the fin can be described
by equation

dQldr 0 (5.1)

Solving Eq. (5.1) for straight fins of rectangular profile, assuming that the heat
transfer and thermal conductivity coefficients along the height of the fin and the
temperature across its thickness are constant values, gives

Table 5.1. Values of the mean heat transfer coefficient

Character of heat transfer a Character of heat transfer a

Natural convection: Forced convection:

air 5-40 air 10-800
water 200-800 water 100-2 x 10+
oil 60-2.5 x 10°
i
122 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

V/a = 6/2 6:2)

The left-hand side of equality (5.2) represents the heat transfer resistance of the
fin. The right-hand side represents heat conduction resistance of the fin corre-
sponding to half its thickness. If both these resistances are equal, finning is not
worthwhile.
It is advantageous to employ finned surfaces when 2)/adé > 5. In practice,
however, finning is often determined by other factors such as hydraulic resis-
tance of the finned surface, weight and dimensions of the heat exchanger.
One of the most important optimum finning characteristics is the Biot num-
ber (Bi = al/d,, where 1 = V,/F; is a characteristic length, V, is the fin
volume, and F; is the fin surface area). The Biot number is the ratio of heat
conduction resistance to R, to the resistance of fin surface R.-

Bi = R,/R. (S<3)

where R, = 6/2d, R. = I/a.

Normally, R, is always substantially less than R.; therefore, Bi has a value
of the order 0.1. If it is assumed that Bi ~ 0.1, it follows that the mean heat
transfer coefficient and the characteristic length must be small compared with
fin heat conduction. According to Table 5.1, tube finning is advantageous for
natural and forced convection if coefficient a of the thermal carrier (gas or oil)
is small.
Let us consider heat conduction of tubes with round fins of constant thick-
ness 6 (Fig. 5.2). Due to the complexity of the investigated problem it is as-
sumed that heat conduction of the fin, the temperature of the cooling medium,
the heat transfer coefficient along the whole fin surface, and the temperature at
the root of the fin are all constants. It is pertinent to mention that only the last
two conditions merit the greatest attention.

LLL
NA
LA NMOL 7,

Figure 5.2 Schematic of a constant-

thickness fin on a round tube.
 HEAT TRANSFER OF BANKS OF FINNED TUBES 123

The temperature of the surroundings tis arbitrarily taken as equal to zero.

The temperature of the fin varies only along the radius t = f(r). The tempera-
tures at the root and tip of the fin are t, and t,, respectively. The heat transfer
coefficient on the fin surface is equal to a. For an elemental ring having the
radii r and r + dr in a stationary regime we can write

OF Oa dO (3.4)

O, = ~2nddr < (5.5)

where 3 = t — bt,

Or su =Q,+ S*10, dr= —2nd3 (<=

d%
r+-<z
d* >
rdr+-
d%
dr) (5.6)
a

O10 dO Sos (
da*?s
rdr+ <dt dr) (5.7)

But, on the other hand, heat transferred from the elemental ring to the sur-
roundings

dQ = 2«9 2nrdr (5.8)

By equating the right-hand sides of Eqs. (5.7), (5.8) we obtain

da?s 1 dd 2a
dr? Pdr kS Pen eed

Designating

V2a/48=8 -and Br=z

we obtain

qe etd es: «
RR G2)
Expression (5.10) is a Bessel equation which can be easily solved if the bound-
ary conditions are known. In this case, heat transfer on the end face can be
simplified, with a sufficient degree of accuracy, by arbitrarily increasing the
height of the rib by half its thickness.
To determine heat conduction of helical fins (Fig. 5.3), we use the same
simplifying conditions assumed for the derivation of the differential equation
for round fins.
124 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Figure 5.3 Continuous helical fin of

trapezoidal profile.

The main differential equation for helical fins of a trapezoidal section is

obtained by writing the heat balance for a helical element of height dr between
sections AA’D'D and CC’B'B.
Let us consider the special case when the fin has a rectangular cross sec-
tion, i.e., the thickness is constant along the height: dé/dr = 0 (Fig. 5.4). In
this case, the main differential equation is

Git or HBO =0 (5.11)

where

-L. Figure 5.4 A fin of rectangular profile.

 HEAT TRANSFER OF BANKS OF FINNED TUBES 125

r d9
srg Pp ar

is the curvature of the helix, s’ = s/2r.

The general solution of this equation, for the determination of the tempera-
ture along the fin, has the form
1 1

8=(s? +P)
'

* [96s? +79)* RBG

as
if
-
=

Oke D2
is2 ssh
Teor
ae
|
a
The amount of heat transferred by the helical fin, as in the case of a round fin,
is determined by assuming that all the heat flow passes through the area of the
fin root. In the present case, the area of the root is computed from the formula

= 25]/ s?4+(2z7r,) (5.13)

Finally, the amount of heat transferred is expressed by the equation

O = 4mN5 (s'? + 2) [61 cth [8 (R—r,)+]9, (e

e
s?4+p?

x esch [8 (R —r1)] \ (5.14)

When analyzing heat conduction in a finned tube it was assumed that the
temperature varies only along the radius of the fin. In reality, heat flow in the
fin is two-dimensional, since the temperature varies across the fin thickness. In
practice, however, when Bi ~ 1 or less the heat flow in the fin can be regarded
as one-dimensional [78] (the error does not exceed 5%).

5.2. EFFECTIVENESS OF FINS

If the heat transfer surface of one side of the wall is increased with the help of
metal fins, it is to be expected that the heat flow per unit surface area of the fin
carrying wall will increase in direct proportion to the heat transfer surface area.
However, because of the presence of a temperature gradient on the fins, the
effective temperature head is decreased. Therefore, the total increase in heat
flowrate will be lower than the expected flow.
In order to make the computation of heat transfer for finned surfaces more
convenient, the fin effectiveness coefficient, sometimes called fin efficiency, is
126 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

introduced. This coefficient indicates the intensification of heat transfer when a

given surface is finned.
The expression ‘‘intensification of heat transfer’’ only partially reflects the
true increase in heat transfer when the layer near the wall or the main flow is
made turbulent, since it rises mainly on account of the heat transfer surface
formed by the fins.
Heat transfer process in finned tubes is extremely complex and differs
substantially from heat transfer in smooth tubes. Heat flow from finned surfaces
depends on the dimensions of the fins, their shape, and the material they are
made from. The fin effectiveness coefficient represents the ratio of the actual
amount of heat transferred from the finned surface Q, to the amount of heat Q,
which could have been transferred had the whole fin surface had the tempera-
ture of the root, i.e.,

qe =Qr/Q, (at)

This condition could be satisfied if the thermal conductivity of the fin material
has an infinite value (A,, — ©). In practice, any material used to construct
finned tubes has a specific value of X,,. In addition, when a finned tube is placed
in the flow of a thermal carrier, thermal and hydrodynamic boundary layers are
created as a result of which a temperature gradient is formed towards the tip of
the fin. Therefore, the conditions of infinite value for thermal conductivity of
the fin and constancy of temperature along the fin height cannot be fulfilled, in
practice. However, fins made from materials with high thermal conductivity
(e.g., copper, aluminum) are highly effective.
In order to determine the variables on which fin effectiveness depends, the
following analysis is presented.
Let the amount of heat given by the surface of a fin to the fluid surrounding
it be expressed in the form

Fr
Or=[ «dF; (5.16)
0

where a is the fin heat transfer coefficient. Assume further that thermal conduc-
tivity of the fin is equal to infinity, and the heat transfer coefficient across the
whole fin surface and the temperature of the surroundings are constant. It is
then found that the temperature of fin surface approaches the temperature at the
fin root 3,. Then

Fr

Q,=%, [ adF' (8:17)

0
 HEAT TRANSFER OF BANKS OF FINNED TUBES 127

From equations (5.16) and (5.17) an expression for the fin effectiveness
coefficient is found

Fr Fr

ne= f add Fy / 9, ii ad F (5.18)

0 0

Assuming the value of a to be constant, a simple and convenient expression is

obtained, as a first approximation, for the determination of 7,.

Hr= 3/9, Gat9)

from which it is evident that the ratio of the average temperature difference of
the finned surface and the surroundings can be taken as the fin effectiveness
coefficient. From equations (5.16) and (5.19) the following expression is ob-
tained

QO=69, nr Fr (5.20)

which is convenient to use in practical calculations. To do this, however, the

heat transfer coefficient, the wall temperature at the fin root on the tube, and
the temperature of the thermal carrier must be known. The fin effectiveness
coefficient is determined from specification manuals and other sources [79-83]
as a function of the ratio D/d (D is the fin diameter and d is the diameter of the
tube carrying the fins). The shape and parameters of the fin are determined,
depending on the requirements imposed on the heat exchanger. The diagrams
for the determination of the fin effectiveness coefficient for different values of
D/d are usually plotted in the form of the function y, = f(6h), where the
parameter

B= \/2a]a,3 (5.21)
represents heat transfer in the fin. Here, a is the heat transfer coefficient of the
fin, \,, is the thermal conductivity coefficient of the fin metal, and 6 is the
thickness of the fin. If the fin has a trapezoidal shape, the ratio of thicknesses
6,/6, at the tip and root is taken as the characteristic thickness. It is not worth-
while to construct fins having n- < 0.6.
Let us consider the methods of calculating the coefficient n, for the most
common developed surfaces. The effectiveness of straight fins with a rectangu-
lar profile can be determined from equation

np = thx|x (5.22)

where
128 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

x=Bh, B=//2a/28, h=R-ry

The effectiveness of straight fins with other profiles (triangular, concave-

and convex-parabolic) are derived by solving analytical expressions presented
in the graphic form n; = f(8h). Equation (5.22) is often used to calculate the
effectiveness of round fins.
Fig. 5.5 presents the values of the effectiveness coefficients computed for
round fins of constant thickness which are used, depending on the parameter
Bh, when the effectiveness of helical fins are calculated. The coefficient of fin
effectiveness n, is determined as a function of

Bh’= V/2a/28 h’

and D/d. Here h’ is the arbitrary fin height used to evaluate the heat transfer of
the face of the fin, and is determined from the condition that the fin height
increases by half its thickness. Thus, for a trapezoidal fin

h’ =h+(8,+85)/2 (5.23)

04 = 08 12 16 20 24 ph
Figure 5.5. Dependence of the effectiveness coefficient of round transverse fins with a rectangular
profile on Bh.
 HEAT TRANSFER OF BANKS OF FINNED TUBES 129

Comparison of the effectiveness coefficients of round fins of rectangular

and narrowing profiles show that for the same fin height and identical cross
sectional areas fo, the effectiveness of the narrowing fin is higher. The differ-
ence in effectiveness of the two profiles rises with an increase in the value of
the parameter Bh.
The use of the fin effectiveness coefficient in calculations of heat transfer in
finned tubes helps avoid complex analytical expressions and makes these calcu-
lations simpler and more convenient.
The variation of heat transfer along the height of the fin is taken into
account by an additional correction factor y which, in the majority of practical
cases, is taken to be equal to 0.85-0.9. Otherwise, is determined according to
[6] and [92] by the following equations, respectively

4=0.97 —0.0568h
or

b=1—0.0588h, (5.24)
Therefore, fin effectiveness is dependent on its height, shape, material, and the
heat transfer coefficient on the surface. It is assumed that there is no contact
thermal resistance between the fin and the tube carrying that fin.
The difference in fin effectiveness of various cross sections amounts to
approximately 10%, and when Gh < 1, it is even less.
However, a fin is not characterized by just the fin effectiveness coefficient.
It is also important to take into consideration the volume and weight of the fin.
This is important, in particular, when rare metals are used to make fins.
By solving the differential equation (5.9), the profile of the fin with mini-
mum cross-sectional area can be found.
According to work [79], the optimum cross-sectional area of a rectangular
fin is

ee (# ) G25)
Na

where c is constant equal to 0.252 for straight fins of rectangular profile, 0.174
for fins of triangular profile, and 0.167 for fins of parabolic profile. For a
constant amount of heat transferred, a fin of rectangular profile will require
about 50% more material to construct, compared with an optimum fin of para-
bolic profile. Equation (5.25) indicates that the value of A decreases as the heat
transfer coefficients and thermal conductivity of the material increase. By vary-
ing these values, the optimum fin weight can be determined since it depends on
the cross-sectional area and specific weight of the metal. However, the volume
of the fin increases as a function of g, cubed where q, is the heat flow on the
130 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

tube carrying the fins, and decreases with increasing wall temperature ¢,.
Therefore, it is expedient to reduce the height of the fins and increase their
number. However, the number of fins is limited by the thickness of the bound-
ary layer which develops on their surfaces. When the distance between fins is
too small, the hydraulic resistance will be higher, since, according to Darsi

JE hfe!
Cid Ca?
i.e., the pressure drop is inversely proportional to the equivalent diameter of
the channel d, = 4F/P, where F is the cross-sectional area of the channel, P is
the perimeter of the channel, & is the friction coefficient, L is the length of the
channel, and & is the mean velocity of fluid flow. Hence, the developing bound-
ary layers on the fins must not come into contact with each other. The space
between the fins u = s — 6 must not be less than twice the thickness of the
boundary layer, i.e., u > 26,,,,, where s is the pitch of the fin, 6 is the thickness
of the fin, and 6,,,, is the maximum thickness of the boundary layer on the fin.
However, since the flow in the spacing between the fins is highly turbulent [5],
the optimum distance between fins can apparently be reduced to u = 26,,,,. If it
is assumed that the temperature of the fluid surrounding the finned tube is equal
to zero, the amount of heat emitted by the fin surface according to equation
(5.19) will be

p= Ol hp Fe (5.27)

Consequently, the amount of heat emitted by the smooth part of the finned tube
will be

O, =antF, (5.28)

Hence the total amount of heat

Q=Or+Q,=4t, Ge Fert %t,F, (5.29)

When the heat transfer coefficient is arbitrarily replaced by the averaged value
a, which takes into account heat transfer from the surface of the fin, the surface
of the non-finned part of the tube, and the effectiveness of the fin, the total heat
transferred by the finned surface is given by

Or aih (5.30)
where

0, = On Fy/F +a, F,/F (5.31)

 HEAT TRANSFER OF BANKS OF FINNED TUBES 131

jy es a oe (5.32)
Assuming a, ~ a we obtain

kp = % [= ar +( a “t)|=4 ! —(1 =e) = (5.33)

The reduced heat transfer coefficient a, takes into account the resistance to heat
conduction caused by the shape, section, and the fin material; the thermal
resistance to convective heat transfer when the finned surface is immersed in
the fluids; and the effectiveness of fin heat transfer. In experimental practice,
the reduced heat transfer coefficient is determined from the equation

a, = QO/FO, (5.34)
Fins with trapezoidal cross section, often seen in practice, can be ac-
counted for in effectiveness calculations by the correction coefficient € which
can be determined, depending on the parameters 6h and 6,/6,, graphically from
Fig. 5.6.
For the calculation of heat transfer through a finned tube (Fig. 5.4) be-
tween two thermal carriers, the following equations can be written

O=Gy, ee (t 51 — twa) (5,35)

~ 2e) Att)
aor (didy) Sig)

QO =a, F(t — ts) (5.37)

where d = 2r,. From Eqs. (5.35)-(5.37) we obtain

1,06

Figure 5.6. Dependence of the coeffi-

i which accounts for the variation
pe aie a) cient g,
40
in thickness of trapezoidal fins, on the pa-
0 06 Wa 18 fa rameters Bh and 55/6).
132 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

O=- Aner E93 Oph) Ges

wae, oa) ; : (5.38)
nm
Ber f wi eee
rel &

Dividing the heat flow by the total surface of the finned tube gives

O!/F=q=Kk (tp — tp) (5.39)

where

If the heat flow is divided by the surface of the tube carrying the fins F, we
obtain

=4y (tn — ty)

O/F,=9, (5.40)

where

aR eae 1 F,\72
es rae eeu = | Sau a =)

The wall thickness of the tube is usually small in comparison with its
diameter; therefore, simplified formulas are used in practice. By assuming
d/d) — 1, and expanding the term In (d/d) into a series, one can limit oneself
to the first term of the series

In d/dy = 28'/d, (5.41)

If we further assume that F) = F,, we obtain a simplified equation for the

calculation of the heat transfer coefficient of a finned tube in both cases:

{ F S Se at BA
. ai Mer Fue ee rt (5.42)

aay pall D 1 F,\-2

hale aie (5.43)
 HEAT TRANSFER OF BANKS OF FINNED TUBES 133

Figure 5.7. Schematic diagram for the deter-

mination of the equivalent diameter.

5.3. SELECTION OF THE DETERMINING PARAMETERS

When the similarity variables are computed, a linear determining dimension

must be assumed. There is no consensus, up to the present time, about the most
rational determining dimension for banks of finned tubes. Many authors adopt
the diameter of the tube d carrying the fins as a determining dimension in order
to simplify the computations. In some works the equivalent diameter is adopted
for the same reason

d, = 4f/P (5.44)

where f is the transverse cross-sectional area of the bank and P is the wetted
perimeter (Fig. 5.7). The latter is convenient to use if friction is the main factor
influencing the velocity distribution in a heat exchanger having a constant
cross-sectional area along the flow. In work [84], the following expression for
d, is proposed for finned banks:

Peele ao Fy,
de=[2 i. Sree \] F d (5.45)

and in work [85], the expression

t= 2 [y(s§=)=25h] (5.46)
2h+s

Sometimes the fin pitch, flow length, the diameter of an equivalent surface,
or some arbitrary value / are adopted as determining dimensions. The use of the
diameter of an equivalent surface is convenient due to the fact that both smooth
and finned tubes with equal external areas will have

dy = Fd/F, (5.47)
134 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

The sections of the carrying tube between fins and the plane fins, which are
similar to channels with characteristic dimensions, are taken into account by the
following arbitrary value [84]

Peay eo * 0.785 (D? — d?) (5.48)

All the foregoing expressions of the determining dimension are employed

when experimental data are evaluated.
Opinions differ regarding the selection of the particular geometrical param-
eters for the analysis of experimental data. The parameters used include h/d,
6/d, d/s, h/s, and others. In solving this problem, the relative depth of the space
between fins //s is of particular interest. It is established that, at some values of
h/s, stagnant regions are formed which exclude part of the surface from active
heat transfer.
There is a notion that the optimum distance between fins must not be less
than twice the thickness of the boundary layer. This value for laminar and
turbulent flows can be determined using the techniques presented in work [3].
The optimum distance between fins is not the same for laminar and turbulent
flows.
The ratio h/s is closely related to the finning coefficient «. This coefficient
is also used as an independent variable, and it clearly reflects the effect of the
geometrical finning parameters on heat transfer.
For the determination of the similarity variables for a body in crossflow, it
is important to select the determining velocity. The velocity of flow just ahead
of the tube Up, the velocity in the smallest bank section “7, or the mean value
between &@ and U, is used as a determining velocity.
For tubes in crossflow, the velocity of flow rises from U, to % and from
ge = 0° to g = 90° as a result of the narrowing cross-sectional area. In the
region 90° < » < 180°, the velocity change is reversed.

5.4. LOCAL HEAT TRANSFER ALONG THE HEIGHT

AND PERIMETER OF A FINNED TUBE IN A BANK

Calculation of the local heat transfer on the surface of finned tubes is an impor-
tant practical problem. For the determination of fin effectiveness, experiments
must be conducted with fins of various heights in flows of air and viscous
fluids, and the temperature distribution on the fin surface must be measured.
Local heat flows from the fin to the thermal carrier are determined from the
data of temperature measurements in the spacing between fins. Let us consider
some measurement results of local characteristics of banks of finned tubes.
The complex character of flow in the spaces between fins and tubes is
reflected in the temperature distribution in the boundary layer on fins. For the
 HEAT TRANSFER OF BANKS OF FINNED TUBES 135

measurement of local heat transfer characteristics, finned tubes with d = 120

mm, fin height h = 20 mm, and fin spacing s = 24 mm were used. A calori-
metric tube (Fig. 5.8) was specially constructed for this purpose. The interior
was lined with heated thin foil. To determine the surface temperature, an ele-
ment was prepared from thin foil and heated from a second source. Thermocou-
ples were welded to the inner surface of the element. By turning the calorimeter
around its axis, the wall-temperature distribution along the height of the fin and
circumference was obtained in the form of a function t, = f(y/h, 9). By
simultaneously moving the sensor between the fins, the temperature distribution
in the spacing between the fins was determined.
Fig. 5.9 shows the temperature distribution in the spacing between the fins.
In the frontal side of the tube (0° < g < 90°), the flow temperature differs
slightly from the temperature of the incoming flow ¢; (Fig. 5.9, curve J), de-
spite the proximity to its nonisothermal surface. This is related to the thorough
mixing of the thermal carrier in the wake of the preceding row. The velocity of
the thermal carrier in the rear side of the tube (90° < go < 18°) decreases,
causing recirculation flow, and the temperature increases substantially. The
temperature of the flow drops with increasing distance from the wall; however,
at the same distance from the wall it exceeds the temperature of the thermal
carrier in the frontal side of the tube (Fig. 5.9, curve 3). The thermal boundary
layer increases significantly in the region ¢ = 130-150°. Temperature increase
of the thermal carrier in the spacings between fins and tubes.in the rear side of

3) thin foil; 4)
Figure 5.8. Electrically heated tube: /) duralumin finned tube; 2) fabric housing;
ring; 9) movement
moving sensor; 5) fluoroplastic ring; 6) bronze cylinder; 7) contact; 8) fabric
/0) micrometer; //) spring; 12) welding points for thermocouples; 13) fabric; 14)
mechanism;
conducting wire.
136 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Figure 5.9. Temperature distribution in

the spacings between fins and tubes along
the tube perimeter for y, mm: J) 0.9; 2)
10; 3) 33.75; Rep = 1.6 x 10°, Pry =
O47:

the tube negatively affects the heat transfer intensity in this region. The dimen-
sionless derivative of the temperature dO/dn, along the tube perimeter varies
substantially (0 = (t — t)/(t, — t,), m = y/6,). The local temperature head
At = t,, — t,along the height and the fin perimeter varies within a wide range.
Local temperature-head distributions in the spacings between fins and tubes are
taken as initial data for the calculation of the fin effectiveness coefficient.
Fig. 5.10 shows the distribution of the local heat transfer coefficient on the
surface of a finned tube in an inner row of staggered banks at the most charac-
teristic Reynolds numbers. The local heat transfer is of a complex nature (Fig.
5.10, a, b, c, curve 1): the value of a; at all Re, values is maximum on the fin
surface, and minimum at the fin root. This difference acquires the largest val-
ues in the frontal side of the tube. At low Re,, it reaches a maximum in the
region of the frontal critical point, which is related to the specific features of the
flow at low Re; A large thermal boundary layer, which decreases with increas-
ing Re,, forms on the frontal critical point of the tube. The difference between
the values of a; at the tip and root of the fin, for the rear side of the tube in the
region ¢ ~ 130-160°, decreases substantially or vanishes altogether.
This indicates that, following boundary layer separation, a thermal bound-
ary layer forms in the recirculation flow from the rear critical point in a direc-
tion opposite to the main flow. The coefficient a; = f(y/h, yg) on the fin surface
has values lying between a; at the root and the tip of the fin. The difference
between the values of a; is closely related to Re, which strongly influences the
character of boundary layer developement on the surface of the finned tube.
The data of a; = f(y/h, ¢) is used to compare the results of theoretical investi-
gations, and also to calculate fin effectiveness.
The analysis of results shows that the local heat transfer coefficient on the
fin surface increases starting from the fin root to its tip. Such a tendency is
observed along the whole perimeter of the finned tube; however, the rate of
increase of da,/dh is not the same. The maximum rise of a; along the height is
observed at ¢ = 0°. For example, at Re; = 10°, a; increases up to 60%. In the
 HEAT TRANSFER OF BANKS OF FINNED TUBES 137

a. Lcd
IN oe a Re = 11-10"
é = 220 |

Figure 5.10. Distribution of local

heat transfer along the perimeter of a
finned tube and fin height: a—Rey =
1.1 x 10°, Pry = 220; b—Rey
1.2 x 10°, Pry = 220; c—Rey
1.6 x 10°, Pre = 0.7. I) y = hy; 2)
y = 0.3 h; 3) y = 0 (inner row).

region ¢ = 60°, the rise of a; is minimal. When Re, is increased from 10” to
10°, nonuniformity of heat transfer on the surface of the finned tube decreases.
The mean heat transfer in the rear side of the tube at Re, ~ 10° is higher than in
the frontal side; and starting from Re, = 10° it is the opposite.
The pitches of finned tubes influence the character of their local heat trans-
fer significantly. Investigations were conducted with compact banks 1.13 x
1.06 and 1.66 xX 1.53 while varying h from 1 to 20 mm. Investigations were
carried out for Reynolds numbers ranging from 10° to 3 x 10° in the first row
of banks in flows of air, transformer oil, and aviation oil. Fig. 5.11 shows the
variation of a,/a = f(y) for the staggered banks at two values of a x b and
h = 3 mm with varying Re, Here &, is the local heat transfer coefficient along
the tube circumference which is averaged along the fin height and determined
from equation

sya De E (5.49)
138 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

10
A © f
08 teeied ienoe axb NS ° Va |
a—* dO, a-{5x{06 3 wee os er
e ee @ _

0 e-(66x 155
j
3 |
06 :
0 50 60 50 120 150 y
Figure 5.11. Effect of a x b and Re; on the relative heat transfer coefficient along the perimeter
of finned tubes in banks: 1, 3) Rep = 4 x 10°, Pre = 220; 2, 4) Rey = 3 x 10*, Pry = 220 (first
row).

Here

4, =qil(ti— ty) (5.50)

where F,, g,;, and ¢; are the local characteristics on the surface of the finned
tube.
When bank compactness II increases from 41 (bank 1.66 xX 1.53) to 82
(bank 1.13 x 1.06) at Re, = 4 x 10°, heat transfer in the frontal side of the
tube decreases. However, it increases in the rear side compared with less com-
pact banks. Therefore, nonuniformity of heat transfer distribution in banks of
finned tubes can be regulated by changing bank pitches.
An increase in bank compactness leads to an increased influence of neigh-
boring banks and tubes in the row on the distribution character of &,. This is
particularly pronounced in the frontal side of the tube.

5.5. MEAN HEAT TRANSFER OF BANKS OF FINNED TUBES

Based on the investigation of local heat transfer on the surface of banks of

finned tubes, the mean heat transfer is determined

a= Q/F(t—-t,) (5.51)

where Q = AU - Jin the case of an electric calorimeter, and Q = c, G(t, — t)

in the case of a water calorimeter. The average temperature of the whole sur-
face of the finned tube is proportional to the measured temperatures
 HEAT TRANSFER OF BANKS OF FINNED TUBES 139

ie
be iP
ee GER = (5.52)
0

The mean heat transfer of a finned tube in a bank depends on the flow
regime, properties of the thermal carrier, temperature head, geometrical char-
acteristics of tube fins, tube arrangement in a bank, and the number of rows in
it. This dependence can be expressed in the following form

Nu,=f(Re,, Pr,, Pr,/Pr,, ®, s,/d. said, z) (5.53)

where ® are the parameters characterizing fin geometry.

Many investigations have been devoted to the heat transfer of tubes with
round fins of constant thickness, and tubes with helical fins in flows of gas.
In works [86, 87], a comparison is conducted of heat transfer of staggered
and in-line banks with identical pitches a and b. Heat transfer of four-row
staggered and in-line banks of finned steel tubes is investigated in work [86]
using the method of total thermal modelling in a flow of air for 2 x 10° <
Rep = 2 x 10°,18 sa 521, a= b,0.07 < hid < 0.54,02 = s/ld =
0.27. The determining parameters were chosen as follows: the diameter d of the
tube carrying the fins, the velocity % in the narrow section of the bank, the
arithmetic mean temperature t,,, of the flow and tube walls. The reduced heat
transfer coefficients were determined experimentally. It was established that
heat transfer of a staggered bank can be twice as high as the heat transfer of an
in-line bank. For a low finning coefficient (« = 1.82), the difference in heat
transfer is small and does not depend on Re,, whereas, for high € (« = 9.66) the
difference becomes dependent on Re; decreasing from 100 to 25% with an
increase in Re, from 6 x 10° to 8 x 10°.
In work [87] heat transfer of seven-row in-line and staggered banks of
finned steel tubes were investigated in flow of air. Experiments were conducted
using the method of local thermal modelling. The determining parameters were
the fin carrying tube diameter d, the velocity % in the smallest cross section of
the bank, and the arithmetic mean temperature ¢, of air at the inlet and outlet of
the bank. The reduced heat transfer coefficients were determined experimen-
tally fortl0"'=/Res='6 X 10°, 1.7’ S!a =93.0/1.2 sbi 3.0; hid = 0.28,
s/d = 0.19, and e = 5.1. It was noticed that the heat transfer of staggered tube
banks is higher by 10-30% than heat transfer in in-line banks; the difference in
heat transfer decreased with increasing Re,.
In work [88], heat transfer of eight-row banks of finned bimetallic tubes in
flow of air was investigated using the total modelling method. It was established
that heat transfer is influenced only slightly by the tube arrangement in the
bank. Experiments were conducted in the range of Re, from 4 x 10° to 6 x
10*, a from 1.91 to 3.51, b from 1.03 to 2.35, h/d from 0.35 to 0.44, and s/d
from 0.111 to 0.37.
140 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

The authors of work [89] reached the same conclusions after investigating
heat transfer of 14 banks of finned tubes in a flow of air at Re, from 10° to2 x
10°. The finning parameters in this instance were 0.088 = h/d =< 0.69,
0.078 < s/d < 0.31, 1.54 < a = b < 8.22. The characteristic parameters
were the fin carrying tube diameter and the velocity in the minimum bank
cross-sectional area.
Work [90] revealed a significant influence of a finned tube arrangement on
heat transfer in a staggered bank. Investigated were 10 six-row banks of tubes
having spiral fins (d x s xXh = 22 X 3.3 X 8) ina flow of air by the method
of local thermal modelling in the region of a from 1.82 to 2.41 and b from 1.36
to 1.89 at Rey = 2 X 10°. The determining parameters were the equivalent
diameter d,, the flow temperature 4, and the velocity in the minimum bank
cross-sectional area %. It was established that the coefficient a, rises with in-
creasing transverse bank pitch, and falls with increasing longitudinal pitch. This
can be evaluated by the parameters a°° and b~°*, respectively.
The positive effect of increasing a, and the negative effect of increasing b
on heat transfer of finned tube banks is also observed in works [6, 91].
The effect of bank pitch of finned tubes on heat transfer is evaluated in
works [92, 93] by the parameters (s, — d)/(s; — d) , is where s; is the
diagonal bank pitch.
As evidenced by works [88-93], a difference of opinion exists regarding
the evaluation of bank pitch influence on heat transfer. This influence is already
mentioned in much later works [6, 90-93].
There are two different methods for the evaluation of the effect of bank
row numbers on heat transfer. In the first method, the number of rows in a bank
is changed and the mean heat transfer coefficient is determined, i.e., LNu/z,
where Nu characterizes heat transfer of one transverse row and z is the number
of rows. In the second method, the number of bank rows is constant and heat
transfer is measured in each and every bank row. Works [94-96] can be as-
cribed to the first method.
The investigation of heat transfer of in-line and staggered banks of finned
tubes, using the method of total thermal modelling, is presented in work [94].
Comparison of the mean heat transfer of one-, two-, three-, and four-row banks
is carried out on the basis of a, which is determined from the surface of a
smooth tube. It is concluded that heat transfer in the range 6 x 10° < Re <
2 x 10° decreases in an in-line bank and increases in a staggered bank with an
increasing number of transverse rows. The determining parameters adopted
were the fin-carrying diameter, the velocity in the smallest bank section, and the
temperature of the boundary layer.
Similar results are obtained in works [95, 96]. It is noted in [92] that, with
an increasing number of transverse rows, the mean heat transfer of finned
staggered banks increases, and heat transfer of in-line banks decreases, reach-
ing the established value for four rows.
Works [97-99] can be ascribed to the second method. The investigations
 HEAT TRANSFER OF BANKS OF FINNED TUBES 141

[97, 98] were conducted by the method of local thermal modelling in a flow of
air at Re, ~ 10*. The coefficient a, was computed from the total external heat
transfer surface. It was found that beat transfer stabilizes in the second or third
row of a staggered bank and in the third row of an in-line bank. The heat
transfer in the first row amounts to 70-80% of heat transfer of the stabilized
rows.
In work [99] 14 staggered (1.68 < a < 3.98, 0.75 < b < 2.8) six-row
banks of finned copper tubes (d X h X s x 6, = 16.1 X 4 x 2.85 x 0.57)
were investigated in conditions of total diermal modelling in the range 6 x
10 n=< Re, = 6 X 10°. It was concluded that the stabilization of heat transfer
depends on the transverse pitch a: for a < 1.8 it commences in the second row,
and for a > 1.8 it commences in the third row. The coefficient a, was deter-
mined from the total heat transfer surface. The characteristic parameters were
the velocity in the smallest bank section 7%, the mean bank temperature t, and
fin pitch s.
Analysis of works [94-96, 97-99] indicates that, by increasing the number
of rows to four, heat transfer decreases in an in-line bank, and increases in a
staggered bank. The latter is the result of increased turbulence of the incoming
flow at the finned tube. The decrease of heat transfer in an in-line bank can be
explained by the “‘shadow”’ effect of the preceding tube. As a result, a consid-
erable part of the finned tube in the following transverse bank row is excluded
from active heat transfer.
The study of the effect of the number of transverse rows, using the second
method, revealed that the stabilization of heat transfer occurs in the second or
third rows of a staggered bank, and in the third row of an in-line bank (both
multi-row banks). This is important to know during an experimental investiga-
tion of heat transfer by the local modelling method, since the calorimeter must
be situated no nearer than the third transverse row of the bank.
The determination of the optimum characteristics of tube finning is of para-
mount importance for the design of highly effective compact heat exchangers.
The profile and the thickness of fins have little effect on the heat transfer of
finned tubes [100]; therefore, special attention was paid by researchers to the
study of the effect of fin height and pitch.
In work [101], the heat transfer of finned tubes was investigated as a
function of the parameters h/s, h/d, and s/é,. For 0.9 < h/s <= 3.0, 0.025 s
h/d s 0.48, 0.33 < s/é, = 0.5, the parameter h/s has more influence on heat
transfer compared with the other parameters. This is also confirmed in works
[86, 89, and 102].
Based on the present investigation and analysis of works [86, 103, and
others], one can conclude that the relative depth of the spacing between fins,
when h/s < 1.8, does not affect heat transfer. Heat transfer is proportional to
the surface of fins. When h/s > 1.8, the intensity of the heat transfer decreases
and this proportionality is violated.
The information available in the literature about optimum fin height is
142 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

contradictory. Thus, a fin height h = 10 mm (D/d = 1.87) is recommended in

work [103]. In work [104], since the dimensions of the tube bank with the given
finning is less and the amount of heat transferred is the same (other conditions
remaining the same), the recommended height h = 3 mm (D/d = 1.24).
A small distance between the fins is recommended. In work [105], s = 3
mm for s/d = 0.15, and in [2] not less than s/d = 0.06, which corresponds with
s = 1.4 mm. According to work [106], the permissible degree of fin proximity
depends on the velocity and turbulence of flow in the spacing between the fins,
and must not be less than twice the thickness of the boundary layer.
According to Blasius, the thickness of the boundary layer developing on the
fin surface in the direction of a laminar flow is equal to

d=5.0/ vx/U, (5.54)

and for a turbulent flow, according to Karman

O07 OGY) (5155)

Fin thickness, as noted in works [107, 139] must be as small as possible;

and the profile of the fin must tend to a shape formed by the arcs of a circle.
When the coefficient a, is determined experimentally and the data are
analyzed using this value, there is no need to compute heat transfer across the
material of the fin. However, in this case this coefficient, which is related to a
specific heat transfer surface, cannot be used to derive generalizing equations.
For the generalization of experimental data, convective coefficients of heat
transfer are used.
In work [84], on the basis of obtained data and the results of works of other
researchers on the value of a,, the coefficient @ was computed and the follow-
ing similarity equation proposed

Nu =c Rey ¢~ 0.375 Pip? (5.56)

where c = 0.45 and 0.3 for staggered and in-line banks, respectively. The
experimental results of 23 staggered and 35 in-line banks of finned tubes in the
Re, range from 10” to 10° and e from 5 to 12 were evaluated. Pitch characteris-
tics of banks are not taken into consideration in Eq. (5.56).
For the calculation of convective coefficients of heat transfer of staggered
and in-line banks, similarity equations are suggested in work [108], in which
the pitch characteristics of banks are taken into consideration. However, the
accuracy of the suggested equations is, unfortunately, low (+20%).
When the experimental data on heat transfer of staggered banks of finned
tubes were generalized in [109], it was established that the exponent of the
Reynolds number in the similarity equations depends on the coefficient of tube
finning. The method was further developed in work [92]. It was established in
 HEAT TRANSFER OF BANKS OF FINNED TUBES 143

[110] that this exponent is dependent on the parameters of the tube arrangement
in the bank s,/s,.
After studying heat transfer in staggered finned banks at high Re, values,
the authors of work [6] suggested the following generalizing equations:
in the range 2x10" = Re, = 2 x 10°

a == 0-0 91(a/ bo? (sid) (hid) "4 Reo§ pro (757)

in the range 2 x 10° < Refs 14 x10

Nu; = 0.008 (a/b) (s/d)°"* (hid)? Re? Prot (5.58)

for s/d from 0.12 to 0.28, h/d from 0.12 to 0.59, and a/b from 1.3 to 2.85.
The aforementioned techniques of generalizing experimental data [6, 84,
86, 92, 108, 109] are in need of improvement. Thus, the dependence of the heat
transfer coefficient on pitch characteristics of the bank a and b are not taken
into account in work [84]; in work [109], its dependence on a is ignored. Also
not taken into account is the dependence of heat transfer on the amount of
transverse rows in banks with sparsely spaced rows in [6, 84] and thermophysi-
cal properties of the thermal carrier in [109]. The latter are accounted for in
[84, 92] from experimental data of a smooth plate for a laminar boundary layer.
Since experimental data obtained in a flow of gas were evaluated in works
[6, 84, 92, 108, 109], the adopted value of the exponent of the Prandtl number
did not lead to a large error when a was computed. However, this would have
led to significant errors in the case of viscous fluids.
Thus, in works [7, 20, 86, 91, 92, 107, and others], the exponent of the
Prandtl number is taken to be equal to 0.33, and in [111] equal to 0.36. These
values were established from experimental data on heat transfer in a laminar
boundary layer on a plate, or from data of the mean heat transfer of smooth
tube banks.
At IPTPE AS LitSSR, investigations were carried out with finned tubes in
flows of air, water, aviation and transformer oils. This allowed the determina-
tion of the exponent of Pr, in a wide range (from 0.7 to 4.4 x 10°) (Fig 35a):
For example, the exponent is equal to 0.36 for banks with different pitches
[112]. It is noticed that, for Re, equal to or exceeding 2 Xx 10*, this exponent in
the range 0.7 < Pr, < 10 increases. However, investigations at high Pr, and Re,
have not yet been conducted in finned tube banks. Apparently, starting from
Re, > 2 X 10*, the value of the exponent can be taken to be equal to 0.4, as in
the case of banks of smooth tubes for which it was established [44] that the
exponent n = 0.4 from Re; = 2 xX 10°. Experimental data obtained for a flow
of air and generalized in work [6] for use for viscous fluids incorporated a value
of 0.4 for the exponent of Prandtl number in the region Re, > 2 X 10*. In
works [3, 113], it is analytically established that the exponent is a function of
Re, and Pr, numbers. For example, according to work [113], this exponent is
144 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

i)
Jbem
a

—-+-

rT:

Le
a
ea
|ti
|OH

Figure 5.12. Determination of the exponent of Pr, for banks of finned tubes with a X 5 andd X
s X A, respectively: J) 1.13 x 1.06, 45 X 9 X 3; 2) 1.33 X 1.23, 45 X 9 X 7.5 3) 1.66 X
1.53, 45 x 9 x 15; +—air; A—water; O—transformer oil; @—aviation oil. K, = Nu;Re” 7(Pry/
oe ie)

equal to 0.37 in the region of the frontal critical point of the tube for Pr; < 10.
When Pr, is increased (Pr > 10) the exponent decreases to 0.35.
The effect of the temperature head and the direction of heat flow on heat
transfer in the present work has been accounted for by the ratio Pr,/Pr,. This
ratio is sometimes replaced by x,/u,, in some investigations, since, when the
temperature changes across the boundary layer in the case of viscous fluids,
mainly viscosity changes. However, the ratio Pr,/Pr, reflects the change of
physical properties across the boundary layer more accurately.
The results obtained for different pitches of finned tube banks are presented
in Fig. 5.13. Curves /, 2 correspond to the heating of viscous fluids, and curve
3 to the heating and cooling in a flow of transformer oil during calorimetric
measurements [112]. As is evident from the Figure, the exponent of Pr,/Pr,, is
equal to 0.25. This value stays unchanged for both finned and smooth tube
banks when a viscous fluid is cooled or heated, i.e., it is independent of the
direction of heat flow. When a laminar boundary layer on a plate is cooled, the
exponent drops to 0.19 [113]; and when a turbulent boundary layer is cooled, it
drops to 0.17 [3].
Heat exchangers made from one row of finned tubes are often used in
practice. Fig. 5.14 shows data on the heat transfer of a single row of finned
tubes of a typical compact bank. Also shown is the heat transfer of an inner row
of finned tubes of multi-row staggered banks. Heat transfer of the single-row
bank, at low Re, coincides with the heat transfer of the first bank row within the
error bounds of the experiment. Heat transfer of an inner row, e.g., at Re; =
2 x 10, is higher by 12% than heat transfer of the single-row bank. It is
 HEAT TRANSFER OF BANKS OF FINNED TUBES 145

Figure 5.13. Determination of the exponent

of the parameter Pr,/Pr,,. /) 1.13 x 1.06; 2)
1.33 x 1.23; 3) 1.66 x 1.53. Ky = Nur
Rem aPta

Gt
ae

2 ROE Paes ary 2 htea eet isd es

Figure 5.14. Comparison of data on the mean heat transfer of a single-row bank with a = 1.66
and a multi-row bank 1.66 x 1.53 of finned tubes with d x s x h = 15 x 3 x 5; O—fifth row,
A—first row, +—one row. Ky = Nur Bra. (PryPn)
146 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

interesting to note that the exponent of the Reynolds number, in the case of the
single-row bank, remains unchanged at 0.47 up to Re, = 3 X 10°. In the first
row of the multi-row bank, the value of the exponent (0.48) remains unchanged
up to Rey = 9 X 10° only, which indicates a high turbulence level in the bank.
At a much higher Re, the difference in heat transfer of the single-row bank and
the inner row bank amounts to 30% in the mixed flow regime.
Heat transfer in the first and inner rows of banks with different pitches and
fin height (fins with a trapezoidal cross section) was investigated. Pitches were
varied in a wide range: a—from 1.13 to 1.66, b—from 1.06 to 1.53, h—from 1
to 15 mm. Table 5.2 shows the values of the constant c and the exponent m in
the function Nu; = c Re;'Pr; (Pr//Pr,,)’.
The data on heat transfer in the first row, as seen from Fig. 5.15, are
distributed within wide ranges, depending on the characteristic parameters of
tube banks. Heat transfer in bank 1.66 x 1.53, h = 5 (curve /) is the most
intensive. Heat transfer in bank 1.66 xX 1.53, h = 7.5 in the region of low Re,
and in bank 1.13 x 1.06, 4 = 3 in the medium Re, range (curve 8) is minimal.
This can be explained by the fact that turbulence in the incoming flow, up to
Re, < 10°, is less and, when the fin height is increased, it weakens even
further. At Re > 10’, flow turbulence is higher. However, the small distances

Table 5.2. Values of the constant c and the exponent m for the first row
of staggered banks of finned tubes

Number
of banks axb WU SS 1G Rey c m

] HASSE NESS IS xeeee5 2-10?—9,2-10? 0.64 0.48

3-10%—1.6-10 0.071 0.76
2 Ss 3eales [nese s) 4-10?—3.8-108 0.38 0.55
3,.8-10%— 104 0.093 0.72
8} 1,13 x 1.06 13x3x1 10°—4-108 0.766 0.46
4-10°—3-10 0.097 0.71
4 1.66x 1,53 45955 6-10? —3,5-10° 0.48 0.49
3.5-10°—4- 104 0.063 0.74
Si 1.66x 1.53 45 <9 XS 5,6-10? —4.8- 103 0.54 0.46
4.8-108—4,5-10' 0.06 On
6 1eGG> alos 45x9x3 5-10°—6- 108 0,623 0.46
6-10?—4,5-10! 0.1 0.657
7 iLsisisse Wee3} 4S 97 e5 2-10%—1.4-10' BS) 0.56
1.4-104— 8-10! 0.062 On
8 1.13 x 1.06 45x9x3 3.5-10*—3.7-104 0.425 G3
3.7-104—1,.2-108 0.058 0.7
 HEAT TRANSFER OF BANKS OF FINNED TUBES 147

2 le lel eeTi 2 Ce y, 4 6 & Re,

Figure 5.15. Experimental data on the mean heat transfer of tube banks with various finning
characteristics in the first row: 7) 1.66 x 1.53, A = 5; 2) 1.33 x 1.23, h = 2.5; Sila dSex<anle OG:
h-= 1; 4) 1.66 x 1.53, h = 15; 5) 1.66 x 1.53, h = 7.5; 6) 1.66 x 1.53, hk = 3; 7) 1.33 x
1.23, h = 7.5; 8) 1.13 X 1.06, h = 3; Ky— as per Fig. 5.14.

between the tubes and fins in highly condensed banks with small fin height
transforms the flow into laminar, and heat transfer decreases significantly
(curve 6). Depending on the finning parameters of the banks, the character of
transition from laminar to mixed flow changes in the range of Re, from 9 x 10°
to 3 x 10° (curve J in Fig. 5.15). This is an indication of the substantial
influence of the cross section of flow on the intensity of heat transfer of banks.
Such a transition was observed in only one of the investigated banks. Heat
transfer intensity rises at the start of transition and drops at the end. This
indicates that by changing some finning characteristics of the tube banks, the
intensity of heat transfer can be controlled.
Transition from a laminar flow regime to mixed flow changes, depending
on the pitches of tube banks and fin height, in a wide range of Re, In banks
with large intertubular spacing, it commences at low Re; (Fig. 5.15, curves J,
2, 5), and vice versa (curve 8). For example, at Rey = 2 X 10* the difference
in heat transfer of banks / and 8 is very substantial. Therefore, it is useful to
design heat exchangers to operate in this regime. As Table 5.2 shows, the
maximum value of the constant c, prior to the transition point, corresponds to
banks J and 3; and following the transition point it corresponds to banks J and
2.
The mean heat transfer of an inner row was also determined for the same
 148 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

“2 Fifth row

2 aig i ee Eg a ee ee ae
Figure 5.16. Experimental data on the mean heat transfer of an inner row (the fifth) of tube banks
with various finning characteristics. Ky and tube banks as per Ish, Sills). Fifth row

banks. The curves generalizing the experimental data are presented in Fig.
5.16; the constant c and the exponent m are presented in Table 5.3.
As Fig. 5.16 shows, following the transition to a mixed regime, the expo-
nent m varies slightly, and the constant c varies significantly. Let us consider the
effect of fin height on heat transfer in bank 1.66 x 1.53 having a characteristic

Table 5.3. Values of the constant c and exponent m for the inner row of
staggered banks of finned tubes

axb aExXaSoxan Rey Cc m

1,66 x 1,53 15x3x5 2-10?—8.6- 10? 1.07 0.40

8.6-10?—6,2-10° 0.15 0,69
feSishioe lS} LSpc3 x2.) 3.9-10?—2,3-104 0.23 0,66
1,13 x 1.06 Hay 6 S356 I 10°—3.-104 0.262 0.64
1,66 x 1,53 45x9x15 6-10?—6-104 0.134 0.68
1.66 x 1,53 4S5x9x7.5 5-10?—4,3-104 0.150 0.68
1.66 x 1.53 45x9x3 5-10? —4.3-104 9,184 0.66
esi Se 1 28! 45x9x 7.5 5-10?—8-105 0.211 0.64
3B) WOE 45x9x3 7-10?—3-108 0.216 0.62
 HEAT TRANSFER OF BANKS OF FINNED TUBES 149

. compactness II = 415 ath = 5 mm. (Investigated were banks witha x b =

1.66 X 1.53 = idem, s = idem and h = var, curves 4-6.) When the fin height
h is decreased, heat transfer increases. The exponent m stays practically un-
changed (0.66-0.68). For example, at Re = 10‘ and the other constant parame-
ters the same, heat transfer increases by 23% when fin height is decreased from
15 to 3 mm.
Investigations were carried out with tube banks having identical finning
characteristics and different pitches (Fig. 5.16, curves, 5, 7 and curves 6, 8).
Comparison of the data shows that with an increasing compactness of tube
banks, heat transfer rises up to a certain value and then it drops. For example,
at Rey = 5 X 10° heat transfer of bank 1.33 x 1.23 andh = 7.5 is higher by
approximately 10% than heat transfer of bank 1.66 x 1.53 with the same tube
finning. With increasing Re, this difference decreases, and at Rey = 6 X 10°
heat transfer of both tube banks becomes the same. with a further increase in
Re,, heat transfer of the more compact bank becomes less, and at Re; = 4 X
10° the difference amounts to 10%. Consequently, the flow in more compact
tube banks becomes laminar with increasing Re, For example, for the same
finning characteristic, heat transfer of the more compact bank 1.13 x 1.06
(Fig. 5.16, curve 8) is lower than heat transfer of bank 1.66 x 1.53 in the
entire investigated values of Re; At Rey = 7 X 10° the difference in heat
transfer amounts to 8%, and when Re; is increased to 4 x 10° the difference
becomes equal to 20.4%.
In order to determine the characteristic flow velocity in compact banks of
finned tubes, the experimental data were evaluated for a varying Reynolds
number which was determined from the incoming or maximum velocities. It
turned out that when the experimental data were presented as a function K=
J(Re,), the use of the maximum velocity resulted in a 12% scattering, and the
incoming velocity resulted in up to an 80% scattering or more. Therefore,
during the investigation of banks of finned tubes (including compact tubes) in
the ranges ofa from 1.13 to 1.66, b from 1.06 to 1.53, and Re, from 10° to3 x
10° the maximum velocity was considered to be the determining velocity.
Comparison of heat transfer of the first and inner rows of staggered tube
banks shows that it is higher by 41% in the first rows. This is caused by the
type of tube finning and the pitch values a x b and Re,
On the basis of experimental data, analysis of the effect of finning charac-
teristics on heat transfer was conducted for identical pitches a x b. The main
finning characteristics are fin height h, fin pitch s, and the coefficient of bank
finning « = F/F,. In the latter parameter, F is the total surface of a finned tube,
and F, is the surface of a tube without fins. The coefficient € is of significance
for the analysis and generalization of data on the heat transfer of banks of
finned tubes. In the present work, fin height and the finning coefficient were
varied, while the other parameters remained constant. It is established in work
[6] that the effect of fin pitch is taken into account by the parameter s/d raised
to the power 0.18. The same exponent has also been adopted in the present
150 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

FECWes aaah On 4 Re,

Figure 5.17. Determination of the fin height influence on heat transfer of a staggered bank 1.66 x
1.53, h: 1-3; 2) 7.5; 3) 15 mm. Ky—as per Fig. 5.14.

work. Fig. 5.17a presents data on heat transfer of a bank 1.66 x 1.53 as a
function of Re, for three values of fin height. Fig. 5.17b shows that the expo-
nent of the ratio h/d is equal to —0.14. For the calculation of the expression
K, = Ny Pre (Pe/Bes\e S the value of the exponent m was taken to be
equal to the experimental value.
Although the exponent of h/d is the same as in work [6], its Reynolds and
Prandtl ranges of application have been widened from 10° to 3 x 10° and from
0.7 to 3.8 x 10°, respectively.
The effect of the tube finning coefficient on heat transfer in a wide range of
€ variations is shown in Fig. 5.18. The exponent m of Re; in expression K, = K
Re; ” is determined on the basis of the experimental data of the investigated

Figure 5.18. Influence of the finning

coefficient on heat transfer: 7) work
[84]; 2) [112]; 3) [88]; 4) [20]; 5) (86);
6) [114]; 7) [115]. Kz = KRe #
 HEAT TRANSFER OF BANKS OF FINNED TUBES 151

banks. For example, the value of m in the expression for K, for all investigated
banks is taken to be constant and equal to 0.625 in work [84], whereas its
experimental values changed from 0.51 to 0.67. In the region e > 6 the expo-
nent of e was found to be equal to — 0.375 (Fig. 5.18, curve J). The adoption
of a mean value of m for the analysis of data could be the reason behind some
scattering of the calculated points. In the present investigation, considerable
attention has been given to the little-investigated range of « from 1 to 6. The
exponent, in this region of finning coefficient values is found to be equal to
—0.14. It is pertinent to mention that more accurate results are obtained when
the actual value of the exponent m of Re; is used instead of the mean value.
Fig. 5.19 presents a comparison of results of heat transfer investigations
obtained by various authors. The curves represent generalized data of groups of
tube banks. Curves J and 8 represent data on heat transfer of a single bank.
Curve J represents heat transfer of a bank of finned tubes witha x b = 2.0 Xx
1.73 andd X s X h = 22 X 4 X 10 mn, obtained using local modelling.
Curve 8 represents heat transfer of bank 1.68 x 1.43 with finning characteris-
ticsd X s X h = 16.1 X 2.85 xX 4. Heat transfer was determined using total
thermal modelling, i.e., with all tubes heated. According to Fig. 5.19 the data

3
10 YO

0 |.
Y

1 se 2
1
10 | | a,
3,4

0,8
i Font ] : ! 1

? 0° 0° 10

< : : i aut hors: 7) [116]; 2) [20]; 3)

Figure 5.19. Comparison of experimental data from works of various
[92]. K' =Nu, (h/d)%"™4
[114]; 4) [88]; 5) [117]; 6, 7) (112, 115]; 8) [99]; 9) [86]; 20) (6); 12)
Pry 0,36 (Pr;/Prw) 0,35 (a/b) —0;2 (s/d) —0,18
152 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

obtained in the present work agree well with the results of other investigations
conducted in a flow of air [6, 20, 86, 88, 92, 114, 116, 117]. The exponent
increases from 0.58 at low Re; to 0.65 at medium range Re, and up to 0.8-0.95
at high Re, numbers. On the basis of data obtained at IPTPE AS LitSSR in the
range of variation of Re, from 10° to2 x 10°, Pr, from 0.7 to 4.4 x 10°, and
also results of investigations of other authors, the following generalizing equa-
tions were derived for the calculation of heat transfer of inner rows of staggered
banks with finned tubes:
for Re, from 104.to.2 Sesi0*

Nu, = 0.192 (a/b)? (s/d)-18 (hid) ~°-14 Re®-% Pr?-36 (Pr,/Pr,,)°?> (5.59)

for Re from 2 x 10* to 2 x 10°

Nu, = 0.0507 (a/b)? (s/d)™8 (hid)~°14 Re?-8 Pre“ (Pr,/Pr,)°” (5.60)

when the relative pitches are: a from 1.1 to 4.0, b from 1.03 to 2.5; the finning
parameters are: h/d from 0.07 to 0.715, s/d from 0.06 to 0.36.
For Re, from 2 x 10° to 1.4 x 10°

Nuy= 0.0081 (a/b)? (s/d)8 (/rd)-9-14 Re®%5 Pro4 (Pr,/Pr,,)°29 (5.61)

when the relative pitches are: a from 2.2 to 4.2, b from 1.27 to 2.2; the finning
parameters are: s/d from 0.125 to 0.28, h/d from 0.125 to 0.6.
The determining parameters chosen were: the velccity in the minimum
cross-sectional area of the bank (HZ), the diameter of the fin-carrying tube (d),
and the mean temperature of flow in the tube bank (t,).
As is evident from Eqs. (5.59)-(5.61), the exponent n of the Pr; number for
a mixed flow regime is found to be equal to 0.36, and for a predominantly
turbulent flow is assumed to be equal to 0.4. The exponent p of Pr,/Pr,, is equal
to 0.25 in the whole range of Re; values for the cases of both heating and
cooling. The determining variables in Nusselt and Reynolds numbers are the
velocity in the minimum cross-sectional area of the bank and the diameter of the
fin-carrying tube.
The accumulated experimental data on heat transfer in the region of low
Re, is still insufficient and the exponent of Re, could be even less than 0.65. The
data of an investigated bank of finned tubes 1.66 x 1.53 withd x s x h =
15 x 3 xX Sat Re, < 8 X 10°, for which m was found to be equal to 0.4 [112],
is an indication of this.
Fig. 5.20 shows the values of the relative correction factors for the mean
heat transfer coefficient of staggered and in-line finned tube banks with sparse
tubes. The correction factors are determined according to the equation

c, = Nu,/Nu (5.62)
 HEAT TRANSFER OF BANKS OF FINNED TUBES 153

Figure 5.20. Relative correction factor

for heat transfer of 7) staggered and 2)
in-line finned tube banks with sparse
=) 4 3) Z = rows.

where Nu, and Nu are the average Nusselt numbers for a bank with any number
of transverse rows and a multi-row bank, respectively.
The difference in the curves c, = f(z) for in-line and staggered tube banks
can be explained by the fact that front tubes in staggered banks act as turbulence
inducers, as a result of which heat transfer of inner tube rows increases in
comparison with heat transfer of the first row. In in-line banks the first row
blocks the flow from the following rows, decreasing their heat transfer.
The character of curves c, = f(z) is somewhat influenced by Reynolds
numbers, pitches of tube banks, and finning characteristics. The correction
factor c, for staggered tube banks with wire finning is equal to 0.88 when z = 1
and to 1 when z = 4. In staggered banks of finned tubes, c, varies from 0.65 to
1.02 when z varies from 1 to 20.
Analysis of the aforementioned data showed that formulas for the calcula-
tion of heat transfer of banks of finned tubes are best ascribed to certain ranges
of Reynolds numbers. For the calculation of Nu numbers the formulas given in
Fig. 5.21 can be employed. In these formulas, the determining flow tempera-
ture is the average temperature of the thermal carrier, the determining surface
temperature is the mean temperature of the finned surfaces of all bank tubes,
the determining velocity is the mean velocity in the smallest cross-sectional area
of the bank, and the determining dimension is the diameter of the fin-carrying
tube. Some recommendations for the calculation of heat transfer of finned sur-
faces of other forms can be found in works [83, 118-123].
Membrane heat transfer surfaces, which are superior to other types of tube
banks, are being used lately. In work [124] are presented data on the local
transfer and flow characteristics of membrane tube banks for a wide range of
relative pitches. It is concluded that the local heat transfer of in-line membrane
tube banks have large nonuniformity, which is somewhat reduced in staggered
 154 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

| i

i } 9
6
| I Pe a I et oul
3

| e
| { | |
10
[ Tee r q
| : |
| =~ ' | |
| | oi ,| |
|
|
|
| jl \

T ‘ +
|
+ ie

| | : |

02 |
| |
H |

{
! | |
; ee l
|
—— |
|
|
‘|
| |
4 5
10° 10° 10 10 Re
Figure 5.21. Data for the calculation of heat transfer of banks of finned tubes:
I; Nuy=0.192 (a/b)°? (sid)e-** (hid)— °"* Reo** Prov’? (Pr. Pro
M1. Nuy=0,0507 (a/b)? (s/d)°18 (h/cl) ~ 914 Re9® Pr?* (Pr,/Pi,,)°?> for zones I and II
l.l<a<4.0; 1,03<b<2,5; 0,06 <s/d<0,36; 0.07 <h/d<0,715;
HIE. Nuys =0,0081 (a/b)°-? (s/d )?? (h/d) 9-14 Rep? Pro* (Pry/PryJ?>, 2.2<a< 4.2;
1.27<6<2,2; 0 125<s/d<0.28; 0.125<h/d<0.6

membrane oanks. Changing the relative pitches of membrane tube banks at

Re, > 10° has little influence on the distribution of the relative local heat
transfer a/& = f(y), where a, & are the local and mean heat transfer coeffi-
cients of membrane tube banks, respectively.
As noted in work [125], the most promising heat transfer surfaces are two-
layer surfaces comprised of fins and membranes.
Work [126] experimentally investigated the heat transfer, hydraulic drag,
and the efficiency of compact, staggered bundles with plano-oval, spirally
finned tubes (2.0 < a < 3.09, 1.26 < b < 1.33, 582 < II < 971 m’/m’, d,
x d,xhxsxX6=5.5 X 13.5 X 2.9 X 2.0 X 0.5) by local modelling at
10° < Re’ < 4 x 10°, where d, and d, are the small and large axes of the
plano-oval tubes and the supporting fins, respectively Re = Ud'/v, d’ = 4fL/
F, and fis the minimum opening of a tube bundle cross section, L is the depth of
all of the bundles, and F is the total heat transfer surface area of the bundle. It
 HEAT TRANSFER OF BANKS OF FINNED TUBES 155

was established that from an energy standpoint the most efficient one was the
bundle with an average compactness, i.e., witha x b = 2.54 x 1.33.
Experimental data are being gradually accumulated on local characteristics
of flow and heat transfer of finned tubes. In work [127], the local heat transfer
was determined for finned tubes in a staggered bank (a x b = 2 x 1.35, d x
s xh x 6 = 133 x 13 x 38 x 3.3) at Re, = 2.3 x 10‘ using the method of
spot heating of the model.
Investigations of the variations of local velocity, flow turbulence, tempera-
ture, and local heat transfer coefficients of a totally heated, finned tube model
dxXs xh xX 6 = 152 x 17 x 40 X 5 mm (in the second and sixth rows ofa
staggered bank witha x b = 2.68 x 1.37) at Re, = 1.25 x 10° are reported
in work [128].
A detailed analysis of heat transfer in pin and radial fins with different
shapes is given in work [129]. The advantages of the proposed new parameters
for the determination of fin effectiveness are discussed. Fin effectiveness, ac-
cording to the authors, is the value 7, i.e., the ratio Q/Q,. Q is the actual heat
flowrate from the fin; Q, is the heat flowrate which could have been achieved
had the fin temperature been constant and equal to the temperature of its base,
i.e., t,(h) = t,. Q, can be characterized as the heat flowrate given by a fin of
the same dimensions having an infinitely large thermal conductivity coefficient.
For the calculation of heat transfer of in-line tube banks in the ranges of Re,
from 5 x 10° to 10°, a from 1.72 to 3.0, b from 1.8 to 4.0, and € from 5 to 12
according to the equation given in work [84], we propose

Nu, = 0.303 Re?-85 ¢-0.375 Pr9-36 (Pr y/Pr,,)°26 (5.63)

According to work [139], the mean heat transfer of staggered tube banks
with cylindrical spikes in the ranges of Re, from 2 x 10° to 10’, € from 2.88 to
4.95, a from 1.65 to 4.35, and b from 1.27 to 1.79 is determined by the
formula

I 0 10Sce oe Re)? Pr? (Pr./ Pr (5.64)

where

AL a—| 7
v==
(sg!d)—1

and sj is the diagonal bank pitch. The mean heat transfer of in-line banks of the
same tubes in the range of b from 1.47 to 2.05, a = 2.2 is determined by the
formula

Nu, = 0.428 9-89 5044 Re®54 Pr2.36 (Pr//Pr,,)25 (5.65)

156 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

The mean heat transfer of staggered tube banks with wire finning can be
determined from equation [130]

Nup=4229:(d/s) 8 Ugh) ao botRets Pel*9 (Pi Pr,ees (5.66)

which is valid at Re, from 980 to 1.9 x 10*, J/h from 0.061 to 0.236 (Ip is the
loop pitch), d/s from 1.41 to 2.72, a from 2.1 to 4.0, and b from 1.84 to 4.0.
In work [131], data on heat transfer of an in-line bank 1.68 x 1.28 with
longitudinal fins (h/d = 0.015, h/5 = 0.52) on tubes in flow of gas are
reported for high Re, range from 10* to 10°. It is noted that a 30% increase of
heat transfer is achieved.
Staggered banks of spiked tubes are more effective than in-line banks of
identical tubes in the range of low Rey.
The mean heat transfer of staggered tube banks with trapezoidal and rec-
tangular fins is computed [132] from equation

Nu, =0,.16 ¢9-24 Reps Pi? (Brrr. (5.67)

which is used in the ranges of Re; from 8 x 10° to 8 x 10°, € from 0.59 to
2:27 ,<a from: 135'to 2:5; and dbitrom/1-1)to:2:5.
It is important to note that, in order to increase the heat transfer of finned
tube banks with high fins, considerable attention is being paid lately to their
angle, twisting, and deformation. By such means, heat transfer can be increased
at the fin root. Analysis of heat transfer in various banks of finned tubes shows
that the optimum angle of the fin is equal to 45°. Larger angles (of the order of
60°) increase the hydraulic resistance of the bank substantially without improv-
ing heat transfer.
As our investigations show, special attention must be given to banks of
tubes with wire finning. These tubes have higher heat transfer rates and lower
coefficients of hydraulic resistance, compared with traditional banks of finned
tubes.
aa
i ed es ee ee

CHAPTER

a ene aSihas ptorehee get ere ee wy ect SIX

eearte Se
EFFECTIVENESS OF HEAT EXCHANGERS

Comprehensive investigations of convective heat transfer of various tube banks

in crossflow, conducted both in the USSR and abroad during the last decade,
could be used as reliable initial data for various arrangements and designs.
A given technical condition can be met by several designs of heat transfer
surfaces. For the determination of the best design, comparison of the acceptable
alternatives are compared on the basis of one or more optimal criteria. The
design principles of effective surfaces, their shapes, and arrangements are pre-
sented in works [133-147 and others]. For the comparison of heat transfer
surfaces, a criterion called the energy coefficient is proposed

E = Q/N (6.1)

The intensity of heat transfer surfaces is determined by the heat flowrate Q,

and the effectiveness E is determined by dividing the heat transferred by the
power N expended to bring the thermal carrier into motion in the heat ex-
changer. From the functional relationship

Nu,=f(Re,, Pr,, a, b, k/d, hid, s/d, ¢, Tl) (6.2)

it follows that the heat transfer coefficient increases with the velocity of the
thermal carrier. However, the power needed to move the fluid increases
proportionally to the velocity cubed

= Aue (6.3)

157
158 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

where A is a parameter which is slightly dependent on or independent of the

velocity.
From the analysis of works [134-147 and others] it follows that the main
design criterion for new heat exchangers is high intensity of heat transfer for a
given value of the power expended to move the fluid. It is, therefore, impera-
tive to study and compare the available experimental results on heat transfer and
hydraulic resistance of tube banks of various shapes and arrangements in
crossflow. It is important to determine the heat transfer effectiveness as a func-
tion of the physical properties of thermal carriers in a wide range of Prandtl
numbers.
Factors which enhance intensification of convective heat transfer must be
brought to light. Intensification can be brought about by a change in tube shape,
use of fins and intensifiers, and tube arrangement in banks. On the basis of the
obtained data, guidelines can be set for the design of effective heat transfer
surfaces.

6.1. EVALUATION OF HEAT-EXCHANGER EFFECTIVENESS

Complex aF is one of the most important characteristics of heat exchangers for

one heat transfer surface (for the whole heat exchanger: kF). The higher the
value of this complex for given temperature head and power to bring the fluid
into motion, the higher the thermal energy transferred by the heat exchanger.
Analysis of data about the influence of different factors on aF allows the
determination of the qualitative characteristics of heat transfer surfaces.
The effectiveness of any heat transfer surface is

E= F

f
d? Ay At

Or VP
Nuys

Eu Re}
(6.4)

where

Nu, =c Re? Pr’.(Pr,/Pr,,)?

Eu=k Rey

F is the heat transfer surface and f is the minimum cross section. The character-
istics of heat transfer surfaces can be compared with each other if the conditions
correspond and At = idem.
However, the complexes dh, At(of - v7) and Nu,/Eu Rey, for given condi-
tions, are not identical even for two similar surfaces and fluids. This is so since
the first complex includes the diameter d and the second value Pr;. In order to
make these complexes more universal some transformations are necessary.
 EFFECTIVENESS OF HEAT EXCHANGERS 159

It is known that the Prandtl number is determined by the physical proper-

ties of the thermal carrier, and the exponent n of Pr, is dependent on the flow
regime and surface shape and normally varies between 0.33 and 0.4. When
n = 0.36 the value of Pr, characterizes the thermal carrier in the mixed flow
regime. Therefore, the value Pr,’ is best related to the complex d?Af At/p;,* v7.
aw into consideration that a = Nu, \j/d, % = Re, »,/d and Ap =
ps W Eu, we reduce the parameter F/f and Eu to a single transverse tube bank
and multiply and divide equation (6.4) by D*. Then

fe
fi \D
ee BEE
Or VF
EAL kc See (6.5)

where k is a constant and r is an exponent in the dependence of Eu on Re;. By

designating

=, Fa fd \F pe DAA PZ (Pry/PryyPAr Sy ah
| eg taro: ee
ae a (>) ed teas weSs
we obtain

in= EE E, (6.7)

To make the conditions of comparison of different surfaces and fluids identical,

it is assumed that At =1°C, (Pr,/Pr,)’ = 1, D = 1m.
It is evident from equation (6.4) that the effectiveness of heat exchangers
depends on the parameter F’/f which, in the case of a tube bank in crossflow,
can be determined by the formula

ft,a ize GEbizee a= (6.8)

F xdiz,z tdlz,z TZ

where z, is the number of tubes in a transverse row and z is the number of rows
in a bank. The parameter f/F depends on the transverse pitch a and on the
number of rows. If resistance is ascribed to one row, Eu and f/F are reduced by
z. As a result, surface compactness of one row will be characterized by the ratio
f,/F, in equation (6.6). With the help of the component of effectiveness £,, an
element of the bank (Fig. 6.1) can be characterized (F; is the tube surface in an
element 1 m long). The effect of dimensions of the heat transfer surface on its
effectiveness is taken into consideration by the value E,. The ratio F; to the area
S; X S, represents the compactness II of the heat transfer surface per unit
volume.
The parameters of the thermophysical properties of various thermal carri-
ers are variable. For a single surface E, = idem, and since Re; = idem, then
E, = idem. Within these conditions, the variable component E will be only the
160 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Figure 6.1 Schematic diagram of a

bank element for in-line (a) and stag-
gered (b) bank arrangements.

complex E,, which represents the effectiveness of the thermal carrier. The latter
is chosen depending on the conditions of the technological process. If a number
of thermal carriers satisfy the given system, then the optimum fluid is chosen by
comparing their characteristics. The complex E, allows the determination of the
influence of the thermophysical properties on effectiveness in dimensionless
form, thus making the design of heat exchangers easier.
When E, = idem, the effectiveness of a heat transfer surface is character-
ized by the product EF, x E;. Complex £; can be divided into two parts: Re, :
and (c/k) Re; * According to work [132], the second part is a shape crite-
rion of the heat transfer surface and characterizes both the heat transfer and
hydrodynamics of the surface.
In order to shed more light on the effective tube banks, let us analyze the
components of effectiveness E,, E,, and E separately. We start with component
E;.
When highly effective heat exchangers, composed of banks of smooth
tubes in crossflow of a fluid, are designed, the tube diameter d and the relative
transverse and longitudinal
bank pitches a and b must be chosen carefully.
It is noted in works [133, 134] that the technological qualities of the heat
transfer surface in dense banks of small-diameter tubes is better. Work [135]
indicates that decreasing d affects the reliability of the heat exchanger, compli-
cates its construction, but increases its durability. Therefore, a compromise
solution must be sought. In some cases, energy effectiveness and reliability are
the determining factors in choosing the design of a heat exchanger.
Following some manipulation, energy effectiveness can be expressed in the
following way [133]

Ee
ie iim Rue whi (6.9)

When analyzing the effect of d, a and b on E, a unit volume of 1 m’ and

constant temperature ¢, and velocity u of the thermal carrier must be assumed.
The effect of d on bank effectiveness was evaluated for the ranges of d from 10
to 100 mm, values of Re, from 10’ to 2 x 10°, and variation of the pitches of
symmetrical banks a X b from 1.1 xX 1.1 to 2.0 x 2.0.
 EFFECTIVENESS OF HEAT EXCHANGERS 161

Figure 6.2 Single channel bank.

The design parameter can be expressed as

F TAZ Ze
if Ged) Bi (6.10)

where B is the width of the bank. When the tube is smooth, / = B and equation
(6.10) is simplified to:

Fe nde (6.11)

Such representation of the parameter F/f corresponds to a single channel

bank the length of which is s,(z — 1) + d (Fig. 6.2). Further manipulation
yields the ratio of relative values of the circumference and the spacing between
the tubes (Fig. 6.3):

F 7

This shows that if the value of a is constant, then the heat transfer surface and
cross-sectional area of the row are independent of the tube diameter. Any de-
crease in the diameter of tubes leads to a corresponding increase in their num-
ber, and the total perimeter (the sum of circumferences of the tubes) remains
constant. This follows from the dependences

Deka ,;= keds, wD = thy ay = kod, (6.13)

5)

A An
ale 5 ‘ i 4 ee

Figure 6.3 Elements of single-row (a)

a & and multi-row (b) banks.
162 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

or

xD =k, xd, =kytdy=... =k, 7d). (6.14)

where D is the diameter of a standard bank.

In order to keep the heat transfer surface of a bank unchanged, the condi-
tion s, X z = const must be adhered to. In this case, the bank will be less dense
in the longitudinal direction. However, constancy of the true tube cross section
is a prerequisite. Then, when d is decreased and s, = const, the depth of the
bank increases, leading to an increase in its volume.
When a and b = const, V = const, and taking into account equation (6.14)
relative to the area of the true cross section, we note that with decreasing d the
heat transfer surface increases proportionally to k; = D/d, and the number of
tubes proportional to kj.
Bank compactness, on the other hand, is determined by a single element,
i.e., by the ratio of the circumference of the tube to the area of a rectangle with
sides equal to the transverse and longitudinal pitches:

eae enSy * So
(6.15)

or

l=
5 ri
(6.16)

Consequently, II and d are inversely proportional.

Now we can analyze the effect of the design parameters of tube banks on
effectiveness. From the analysis of the functional dependence k, = f(d), where
K,; = Nu,/Pr; (Pe/Pry)', it follows that when d increases, K;, increases if the
other parameters are kept constant. However, for the evaluation of effective-
ness, we must know the character of dependence d on heat transfer per unit area
of a unit volume. When d increases a decreases, since a = 1/d”~', where m is
the exponent of Re,. Let us analyze the character of this change. Consider the
function a = f(d) for symmetrical in-line banks 1.1 x 1.1; 1.25 x 1.25;
1.5 x 1.5; 2.0 x 2.0 in the Re, range from 10° to 2 x 10° according to the
equation Nu, = cRe; - Pr,'(Pr,/Pr,,)’. Within this range of Re, m = 0.65;
therefore, the value d°® enters into Re, and d' into Nu,. Finally, we conclude
that a is a function of d~°*’. For example, when d is decreased 10 times, a
increases 2.2 times. This increase can be explained by the changing local flow
conditions and heat transfer at constant values of u and Pr,.
A similar analysis of the effect of d on Eu in the function Eu = f(d)
indicates that when the tube diameter d is decreased 10 times, Eu increases 1.3
times. The increase in the resistance coefficient with decreasing d can be ex-
 EFFECTIVENESS OF HEAT EXCHANGERS 163

plained by the changing velocity gradient near the surface of the tubes. De-
creasing d leads to a shift in the flow regime towards laminar flow and to a
change of velocity gradient across the whole cross section of the stream be-
tween the tubes. In a turbulent flow, on the other hand, the value of the velocity
gradient is a maximum only in the boundary layer in the vicinity of the tube
surface.
The effectiveness of banks as a function of tube diameter can be expressed
in the following form

emer > rele

Teo: eae Ay Oe OM ake
Be Md)
By substituting the corresponding values of a, r, n, c, and k in Eq. (6.17) for a
fixed value of ¢ and for a particular thermal carrier, we obtain the following
dependence

E = A/d (6.18)

where A is a coefficient comprising parameters which are, according to the

theory of similarity, independent of d:

Te Ap pr” G iar 3 (6.19)

al or vee fk

Thus, when the diameter d of the tubes of the considered banks are decreased
by 10 times, the value of E increases 1.7 times. When d is decreased from 100
to 70 mm, E increases slightly, and from 70 to 10 mm, it increases substantially.
Any further increase in d above 100 mm decreases E only slightly [151]. The
rate of effectiveness increases with decreasing pitches. a xX b from 1.25 xX
1.25 to 1.1 xX 1.1 is less than in the range 1.5 x 1.5 to 1.25 X 1.25. This is an
indication of the presence of an extremum in the function E = f(a xX b).
Consequently, the effectiveness of heat transfer increases when we move
from banks with large pitches to more dense arrangements. It is important to
note that the effect of pitches a and b as a function of Re, is not consistent, but
the character of the change is approximately the same. This is explained by the
rising compactness, which means an increase in the heat transfer surface area
per unit volume of the bank. On the other hand, when a and 5 are decreased to
some values, the circulation intensity of the fluid in the vortex zones of the
tubes increases.
On the basis of analysis of the positive and negative factors influencing
bank compactness, the function (1/a — 1) x (c/k) = f(a — 1) was studied in
order to determine the most effective tube bank [152]. It is evident from Fig.
6.4 that the value of the function E = f(a x b) is a maximum for banks with
pitchesa x b = 1.1 X 1.1.
164 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

On 05 10 | 20 30
ay
Figure 6.4. Determination of the design parameter a — 1 of symmetrical in-line tube banks.

Therefore, the optimum design parameters of tube banks can be determined

for all flow regimes of interest, and for various values of a and b.
Let us consider the effectiveness of thermal carriers. The thermal carrier
for a heat exchanger is chosen according to the conditions of the technological
process. There are cases in power, chemical, food, radioelectronic and other
industries when there is a choice to employ several thermal carriers. The opti-
mum fluid is then determined by comparing several thermophysical properties
of the fluids, taking into account all the physical and technical requirements. A
number of techniques for comparing these properties are known.
In work [153], a comparison of the effectiveness of gaseous thermal carri-
ers is presented for banks in crossflow.
The effectiveness of the thermophysical properties of liquids used for cool-
ing electronic equipment is analyzed in work [154] on the basis of comparison
of the Muromtsev number (Mo = pncs/p'). It is noted that comparison of the
functions Mo = f(t) is valid for identical flow forms only. When the flow forms
are different, the ratio Mo,/Mo, becomes dimensional.
In the present work, the effectiveness of thermal carriers is evaluated on
the basis of the dimensionless complex £,. E, includes the terms of the
Muromtsev number and a linear dimension (difference in temperature), i.e., the
temperature head and the temperature factor. Complex E, allows the analysis of
the effect of thermophysical properties of thermal carriers in a simpler dimen-
sionless form, obtains the extremum in the function E = f(t,), and simplifies
the thermal design of heat exchangers.
Complex E, = D’ \,Pr; (Pr;/Pr,,)” At/(p, ° v) is the ratio of heat transfer
and power expended to move the thermal carrier.
In order to unify the conditions of comparison of various thermal carriers,
we assume At = IK and (Pr,/Pr,,)’ = 1. Then

E, = [Ag Pr3f(op vp] -¢ (6.20)

where
 EFFECTIVENESS OF HEAT EXCHANGERS 165

=D At(Pr7/Pry)?= 1

Analysis of work [150] shows that, when comparing various thermal carriers,
complex £, must be analyzed in the following form:

Ey =A,Pr? (Pr,/Pr,,)? D+ At/(o,¥}/D) (6.21)

The right-hand side of equation (6.21) is the ratio AQ/AN, the numerator of
which represents heat flow AQ, and the denominator represents the power
needed to overcome viscosity resistance (AN = p,v;/D). When designing heat
exchangers, it is imperative to know the dependence of the complex E, on the
parameters vy, Pr, and also on the value t, which significantly affects Pr, of
viscous thermal carriers.
The products ),Pr; and OY; are considered to be the main components of
the effectiveness of thermal carriers. The exponent of Pr; is assumed to be equal
to 0.36 and 0.4 for mixed and turbulent flow regimes, respectively. On the basis
of the function \,Pr; = f(p,, t), the working parameters of the pressure p; and
temperature ¢, at which heat transfer effectiveness for a given fluid is a maxi-
mum are determined.
From expression (6.21), an analysis of the functions E, = f(t,), E, = f(Pr,),
and E, = f(v,) was conducted for fluid heating and cooling. The effectiveness of
thermal carriers was calculated from equation (6.5), the other parameters remain-
ing unchanged. For various fluids, only the complex APLMo;) was varied as a
function of the temperature. The obtained results (Fig. 6.5) correlate well with
the function Mo = f(t,) in work [154]. The small discrepancy is attributed to the
different values of the initial constant ascribed to the surface configuration and
Reynolds number. Fig. 6.5 shows that water is the most effective thermal carrier,
and aviation oil is the least effective. Transformer oil is superior to air at tempera-
tures higher than 57°C. Since the value of »; in the complex APr/(097) varies
considerably in flows of viscous fluids, the effectiveness of highly viscous fluids
(e.g., aviation oil) increases substantially with rising temperatures. It is conven-
ient to determine the effectiveness of fluids in the form of the function E, =
f(Pr,). As work [152] indicates, this function tends to a maximum when Pr,
approaches unity. Such a behavior is observed up to the critical state of the fluid.
As the critical state is approached, a characteristic dependence of E, on Pr; which
is peculiar to each fluid is noticed.
Thus, complex E, is a criterion of the effect of the physical properties of
thermal carriers on the effectiveness of heat transfer. E, has no bearing on the
construction of the heat transfer surface.
Let us consider the effect of the form and arrangement of heat transfer
surfaces on the effectiveness of tube banks by analyzing the function E; =
fkRe,”* ). The form of a tube bank characterizes the shape of the tubes, and the
arrangement indicates the location of tubes in an element of the bank. The form
and arrangement of a bank affect the intensity of heat transfer via the resistance,
166 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

yA
asd
ae
mi
0 20 40 60 80 foo ¢t
Figure 6.5. Effectiveness of thermal carriers at Reg = 2 X 10°: 1) air; 2) water; 3) transformer
oil; 4) aviation oil MS-20.

the increase of which leads to an increase in heat transfer. However, this phe-
nomenon is not observed in some tube bank designs, and any further increase in
pumping losses is not worthwhile.
This indicates that the optimum parameters of form and arrangement of
tubes in banks must be known in order to determine the effectiveness of heat
transfer surfaces. The following complex can be used as a criterion

c Re}
3 (Eu Rete
exRejai
(C2) Ret ® eRetae
Wakes mec
ane Rey = (6.22)

which represents the ratio of the components of heat transfer and pumping
losses.
Other methods of arrangement optimization of smooth tube banks are also
known [155]; however, bank compactness is dependent in them on technological
limitations. With the development of new processes for the manufacture of heat
exchangers at the present time, the possibility of refining these methods has
increased considerably. Therefore, the limitations on the manufacture of small-
pitch banks are not taken into consideration in the present investigation. The
selection of the optimum forms and arrangements of heat transfer surfaces is
conducted on the basis of the thermal effectiveness of the heat exchanger.
 EFFECTIVENESS OF HEAT EXCHANGERS 167

For the determination of the most effective parameters of heat exchangers,

experimental data obtained in various installations using liquid and gaseous ther-
mal carriers for a wide range of Reynolds and Prandtl numbers were analyzed by
a unified method. Some of the obtained results are shown in Fig. 6.6. This figure
illustrates that the effectiveness of banks increases when their compactness in-
creases. Further analysis of data on compact in-line and staggered tube banks
(1.01 xP0t 1014 X 1014; 1.02 « 1.02; 1.027 x 1.027; 1.055 x 1.055)
indicates that the effectiveness of in-line banks begins to decrease at a—1 <
0.05. Compactness of staggered tube banks, on the other hand, can be increased
without an apparent decrease in effectiveness. This is explained by better flow in
staggered banks due to the absence of a recirculation region in the frontal side of
tubes. Based on the analysis of bank effectiveness variation, one can determine
the optimum pitches, surface roughness, finning parameters, and other design
parameters. The character of the curves FE, = ARE,” ) in Fig. 6.6 supports the
prediction arrived at in work [150] regarding the optimum design parameter a —
he:
The generalized experimental data obtained at IPTPE AS LitSSR in the
form of function F, = f(Eu Re;) for the Re, range from 1 to2 x 10°, Pr, range
from 0.7 to 3 x 10* ina wide range of relative tube bank pitches, are arranged.
in Fig. 6.6 as a band. The slope of the bank is equal to — 0.765 in the range of
Re, from 1 to 2 x 10°, and to —0.73 when Re, = 2 X 10°. (Generalization of
experimental data in the form of the function K, = f(Re,) is not very conven-
ient, since they are widely scattered, depending on pitches and Re,). Experi-
mental data generalized in the form of the function FE, = f(Eu Re;) show that
tube banks are classified depending on their effectiveness. Thus, the known
data about tube banks can be generalized and classified according to effective-
ness. Then

E=c, (k Rep**¥ (6.23)

where c, is a constant depending on the design characteristics of banks, k and r
are the constant and exponent in equation Eu = kRe; and e is an exponent.
Fig. 6.7 shows the dependence of the constant c, on the relative pitches of
tube banks and the parameter

a--1
o=
)/0.25a?+ 62-1 ( )

Dependence (6.24) is valid for many tube banks and a wide range of a X b.
The largest value of c, is observed in the range Re; = 10° 2a 10s for
staggered banks (inflection point at a—1 = 0.6), and the least value for the
same banks in the range Re, = 2 x 10° — 2 x 10° (inflection point ata—1 =
0.4). In-line banks are evaluated as a function of pitch b. From the variation of
168 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

psig NOR Sie, 7a Ahk C0 a ee 3

OS O10 IO 0 | HOO 0 0 0 Os Ne ao EuRe,
Figure 6.6. Comparison of the effectiveness of heat transfer surfaces: J) staggered and in-line
banks of normal density; 2) banks of dense arrangement (a ~ 1.05); 3) highly dense banks (a ~
1.01).

c, as a function of a and b at a constant value of e, the most effective tube banks

are deduced (Table 6.1).
Analysis of function F, = f(Eu Re; ) indicates that some tube banks have
higher hydraulic resistance which, is ineffective as a means for increasing the
heat transfer coefficient. Therefore, tube banks can be classified by their effec-
tiveness. Such a classification shows that the optimum selection of the design
parameters of tube banks allows the increase of their heat transfer by 20%.
Of tube banks with various design characteristics, the most effective will be
the one which has the lowest hydraulic resistance when Q = idem. The hydrau-
lic resistance is closely related to the flow velocity. by changing the velocity, the
hydraulic resistance and heat transfer can be controlled. Transition to a flow
regime with high velocities leads to a decrease in weight of the heat exchanger.
As was shown by the analysis of the components of effectiveness E,, E,, and
E;, the desired mass and dimensional characteristics of tube banks can be ob-
tained by changing each of the three components.
 EFFECTIVENESS OF HEAT EXCHANGERS 169

f |

aes i f
hoe!
mz

10" ak | ipa ;
8 A Bek Ried. | |
| | | | J
6-4 | a + a

j ee eaialeeed Ge
10 2 ES MY TYP 2 4 (a-N(o-1}
Figure 6.7. Dependence of constant c, on the design parameters a-1 and b-1 of in-line and stag-
gered tube banks: /) Rey = 16 = 2 «10°, c; = f(a-1) (staggered banks), ¢ = 1-2; 2) Re; =
10° — 2 x 10°, c, = f(b-1), a x b = (1.007-2.52) x (1.007-2.26) in-line banks; 3) Rey = 2 x
10 -2~x 10° c,; = f(a-1), ¢ = 1-2, staggered banks.

Special attention is being paid to compactness II, which can be determined

from Table 6.2. when the tube banks are designed.
By increasing compactness of heat exchangers, the required heat flow can
be obtained as a result of increasing the heat transfer surface. By simultane-
ously reducing flow velocities of the thermal carrier, energy can be conserved
as a result of decreasing pumping losses.

Table 6.1. Values of a x b, c,, and e for in-line and staggered tube
banks.

Banks aXxXb Cc) e Re,

In-line (1.68 — 2.52) x (1.26 — 2.26) 0.172 —0.76 10°—2-10°

(1,02 — 1.055) x (1,02 — 1.055) 0,128
Staggered C= aloe O—= 025 0.178
a—ae03 o=0,94
a= 1.1 ed Ne 0.203 —0,77 10% —2- 10°
az 1.5 =O.»
a> 1.50 G=iI=7 0.244

Staggered a= 1,03 o=1 0.0424

a=TAy = lI 0.0506 —0,73 2-10°—2-10°
a=1,26—-2,48 o=1—2 0.0575
nee EE
170 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

Table 6.2. Formulas for the determination of compactness of tube banks

No. Surface designation Calculation formula

1. Tube banks of any design Fi(s, + 52)

2. Banks of smooth tubes mTla:b-d
h
ay Banks of finned tubes ~z+4 FI
a-bed
: ENS )
4. Banks of tubes with longitudinal membranes ed (ope 1h— 4

Geog

6.2. DETERMINATION OF OPTIMUM CHARACTERISTICS

OF HEAT EXCHANGERS

In order to increase the effectiveness of heat exchangers, the tube diameter and
the parameter a— 1 must be reduced to their optimum values. When a, b and V
are identical

F,/Fy=d)|d; (6.25)

where F is some initial surface of heat exchanger and F; is an arbitrary variant.

For F = idem

V |Vo=dildo (6.26)

When determining the optimum parameters of banks, the analysis should in-
clude the compactness of the heat exchanger. Then

ITE, =f (xi) (6.27)

where II is the compactness and x; is an arbitrary design parameter (s,, 5, k, h,

s, and so forth). Function (6.27) can be written in a dimensionless form

Ky Ey=f (xi) (6.28)

where Ky = II/IIy. II is the compactness of the tube bank under consideration;
in the general case I], = 1 m’/m’ (or standard bank).
When designing heat exchangers, the main parameters and operation re-
gimes are usually assumed (Table 6.3).
 EFFECTIVENESS OF HEAT EXCHANGERS 171

Table 6.3. Initial data for the design of effective heat exchangers.

No. Parameter Symbol Units

i Temperature at heat exchanger

inlet ty K
2 Temperature at heat exchanger
outlet tp K
3. Physical properties of the fluid at
mean temperature:
density p kg/m?
thermal conductivity oN W/(m - k)
kinematic viscosity y m/sec
4. Mass fluid flowrate G kg/sec
5 Heat transferred Q WwW
6. Permissible limits of variation of
hydraulic resistance a, Pa
ke Fluid velocity Up, u m/sec
8. Design parameters:
type of heat transfer surface given or assumed
o Determining dimension d, m
10. Bank compactness II m?/m?
fil Dependencies for the calculation of Nu; S(Re,)
heat transfer from the literature
12. Dependencies for the calculation of Eu = f(Re,)
hydraulic resistance from the literature

The determination of the effective parameters of tube banks must be car-

ried out in the following order:

1. Evaluate p, v,; , and Pr, from the given temperature of the thermal car-
rier.
2. Determine the effective thermal carrier on the basis of analyzing the com-
plex £, as a function of temperature and Pr,.
3. If the velocity of the thermal carrier Up, its variation limits, and the per-
missible pressure drop across the bank are not given, the velocity and flow
regime must be assumed, and A, is determined from the dependence Eu =
f(Re,).
4. If the temperature head At is not given, the mean temperature head is
adopted and the quantity of heat Q (which must be transferred) is deter-
mined. The value of Q is dependent on complex kF, where k is the heat
transfer coefficient.
5. Assume the initial design parameters of the tube banks and determine the
power N needed to bring the fluid into motion.
172 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

10° io’ 0° a-f

Figure 6.8. Determination of optimum values of parameter a—1 for in-line banks of smooth tubes
at Rep = 10° and the values of b—1: 7) 1-0.1; 2) ~ 0.25; 3) ~ 0.5; 4) ~ 1.0.

6. Determine compactness II of tube bank.

7. Determine £,.
8. Compute the function II E, = f(x,); for the condition II E; — max, the
effective tube bank is determined. When function IT E, — max is analyzed,
the parameters a-1, h/d, k/d, s/h, and so forth, for optimum design are
determined.

The results of the computation of functions IT EF, = f(a—1) and II E; =

f(h/6) are presented in Figs. 6.8 and 6.9. The curves allow the determination of
the optimum values of the analyzed parameters. Fig. 6.8 shows that the opti-
mum pitch is a ~ 1.1 for a wide range of the parameters a—1 and b—1. The
function II E; = f(a—1) is a maximum at this same value of a in the arrange-
ment of parameter range ¢ = 1.7-3.
Function EF, = f(k Re;,**) or FE, = f(Eu Re,) together with tube bank
compactness II allow the reduction of the time needed for experiments and
computations. Analysis of the function indicates that tube banks in crossflow

2 iY fo 5 20 hfé
Figure 6.9. Determination of optimum finning parameters at Rew = orx 10*: 7) one parameter
optimized—fin height; 2) two parameters optimized—height and pitch of fin (d = 23 mm).
 EFFECTIVENESS OF HEAT EXCHANGERS 173

10° 10° 10” 10" 10" 10” EuRe,

Figure 6.10. Determination of effective banks of finned tubes for various values of a x b and h:
I) 1.33 X 1.23, h = 2.55 mm; 2) 1.13 x 1.06, A = 1 mm; 3) 1.66 x 1.53, h = 3 mm; 4)
1.66 x 1.53, h = 5mm; 5) 1.66 x 1.53, h = 7.5 mm; 6) 1.33 x 1.23, A = 7.5 mm; 7) 1.66 x
1.53, h = 15 mm; 8) 1.13 x 1.06, A = 3 mm.

are more effective at low and medium range Re, values, and tube banks in
parallel flow are more effective at high Re, values.
Fig. 6.10 shows an example for the determination of the effectiveness of
finned tube banks. Based on experimental data, function K, = f(Eu Re,) is
plotted for the analyzed tube banks. K; and IIK; are determined at Eu Re; =
idem. By analyzing IIK; in relation to bank pitches and tube finning characteris-
tics, the value of function ILK; > max is determined. This value corresponds to
the best indicators for heat transfer, hydraulic resistance, and compactness of
the group of typical finned tube banks under consideration.
Let us analyze the values of the power Eu Re; = 10’ and 10", presented in
Fig. 6.10, for typical banks of finned tubes. These tube banks are distinguished
by their compactness and large fin height for constant pitch between the fins
h = 3 mm and fin-carrying tube diameter d = 15 mm. Let us consider the
following banks: 1)a x b = 1.13 X 1.06, A = 1 mm, II = 246;2)a x b=
1.33 X 1.23, h = 2.5 mm, II = 335, and3)a xX b = 1.66 X 1.53, h = 5
mm, II = 415. For the first bank (curve 2) having h/d = 0.067, the values of
the function IIK; = 5.6 x 10° and 1.7 x 10* correspond to the values of Eu
Re; = 10’ and 10''. For the second bank (curve /) having h/d = 0.166, IIK; =
7.7 x 10° and 2.5 x 10‘, and the third bank (curve 4) having h/d = 0.33,
IlK; = 8.1 X 10° and 2.6 x 10* (in both cases corresponding to Eu Re; = 10°
and 10''). Comparison of the values of function ILK; ata X b = var, h/d =
var, II = var and at several values of the power number Eu Re; shows that the
most effective bank is the one with h/d = 0.33 and II = 415. Function ILK,, at
Eu Re; = 10’ and 10", is higher by 6 and 5% respectively, compared with a
bank having h/d = 0.166 and II = 335 and by 45 and 42% compared with a
bank having h/d = 0.067 and II = 246. Analysis of the heat transfer of banks
of finned tubes as a function of the power number Eu Re; and finning character-
istics shows that bank compactness is an important parameter. The use of the
power number Eu Re; and the function IIK — max substantially decreases the
time spent on experimentation and computations.
174 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

6.3. COMPUTER-AIDED DESIGN AND CALCULATION

OF HEAT EXCHANGERS

The design and calculation of heat exchangers is a time-consuming task per-

formed by highly qualified personnel. Nonetheless, there is no guarantee that
the optimum version will be arrived at, since errors in selecting the working
velocities and the determining temperature are possible. The latter determines
the characteristics of the thermophysical properties of the thermal carrier. Inac-
curacies in the determination of a change in the thermophysical properties of
fluids and the temperature factor lead to large errors. Usually, the characteris-
tics of the thermophysical properties are evaluated at the arithmetic mean tem-
perature at the inlet and outlet of the heat exchanger. The change of viscosity
and Prandtl number as a function of temperature is not linear. This is particu-
larly true for fluids with high viscosity which could, depending on the differ-
ence between the mean temperature of the heat transfer surface and fluid tem-
perature, change considerably.
Thermal conductivity and density of liquids change insignificantly with
temperature, and the density of gases vary significantly with temperature.
When the change of the physical properties of a thermal carrier with tempera-
ture is not accurately accounted for, erroneous Nusselt, Reynolds, and Prandtl
numbers are obtained, affecting subsequent calculations. The use of computers
allows the introduction of accurately determined values of y(t), A(t), o(t), and
c,(t) for a specific thermal carrier. Alternately, the following approximating
dependencies can be stored in the computer’s memory

pes = (t/t)
A/Ap = (t/t) (6.29)

ele, =(t/t,)"
CplCps = Git).

and accurately determine, from a given variation of fluid temperature, the effect
of the thermophysical properties towards the center of the tube bank.
With increasing temperature the coefficient of dynamic viscosity decreases
in liquids and increases in gases. The viscosity of gases as a function of temper-
ature can be determined from equation

(6.30)

where py is the coefficient of dynamic viscosity at t = 0°C, T is the absolute

temperature at which the coefficient of viscosity is determined, and c is Suther-
 EFFECTIVENESS OF HEAT EXCHANGERS 175

land’s constant, which is equal to 114 for air, 118 for nitrogen, 138 for oxygen,
274 for carbon dioxide, and 198 for methane.
Detailed analysis of the methods for determining viscosity [156] shows that
the most accurate method is the use of the dependence

p= eFl7 (6.31)

for p-Terphenyl A = 1.878 x 10°‘ and B = 1813 according to work [156]

and others. For practical purposes, it is convenient to use the relation

si ighe (Gc) (6.32)

where the exponent n is dependent on the temperature and has different values for
gases and liquids. In our calculations, the following values of n were adopted in
Eqs. (6.29) for the temperature range 20-60°: for water n, = 0.9, n. = 0.06,
for transformer oil n, = 1.4, m = —0.12, for glycerinn, = 3.1, n, = 0.085.
Data obtained for other temperature ranges are given in work [156 and others].
When experimental data were generalized in work [157], the following
formula for the determination of the thermal conductivity coefficient of air was
found to hold in the temperature range 200-1000 K at atmospheric pressure

»=0.0012 +0.0808- 10-2 7+ 0.0321 - 10-872 — 0.942: 10-9. T34.

40.0468 -10-22 74 (6.33)
where T is in K.
According to work [158] the following formula can be used in the tempera-
ture range from —50 to +50°C

d= do[1+ G55 (55 -1)] (6.34)

1/2

which can be used, albeit with less accuracy, for higher temperatures. The
coefficient E'”/23.5 is equal, for example, to 1.01 for glycerin, 0.79 for methyl
alcohol, 0.80 for ethyl alcohol, and 0.83 for toluene. \y is the thermal conduc-
tivity coefficient at 0°C. For the majority of normal liquids, an increase in
temperature by 1°C, in the range from 300 to 450K, leads to a decrease of the
thermal conductivity coefficient by 0.15%, on the average.
When designing a heat exchanger, the construction parameters must be
determined together with the heat transfer surface. It must also be ensured that
the heat exchanger would operate in the given ranges of velocities, tempera-
tures, and thermal loading.
In order to make the design of heat exchangers easier, equations which are
convenient to use in practical calculations are given in the present work. These
176 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

equations can be used to calculate the local and mean heat transfer of smooth
and developed heat transfer surfaces in flows of gases and viscous liquids.
When designing heat exchangers with the aid of a computer, one must feed into
it the following initial data:

¢ temperature of the thermal carrier at the inlet and outlet of the heat ex-
changer
maximum velocity of the thermal carrier
working temperature and pressure
required value of the heat transfer coefficient (maximum value)
characteristics of the physical properties of the thermal carrier
thermal loading

If the heat exchanger is designated for use in difficult conditions, for exam-
ple, with the possibility of surface fouling by the fluid, sedimentation and scal-
ing, corrosion and erosion of surfaces, and so forth, all these factors must be
taken into account in the calculations, together with the various limitations and
permissible errors.
The thermal design of heat exchangers includes the solution of the equa-
tions of heat balance and heat transfer. Presented in the present book are gener-
alizing equations for the calculation of heat transfer of banks of round tubes and
other shapes, and also banks with various types of finned tubes. When the heat
transfer coefficient is calculated, corrections are introduced into the effective-
ness coefficient of the fin under consideration. These corrections are
determined from empirical equations or on the basis of experimental data.
The calculation of heat transfer in the spacing between the tubes of heat
exchangers entail some difficulties which are caused by several reasons. Thus,
baffles in multipass shell-and-tube heat exchangers change the direction of flow,
and the clearances between bank tubes and the baffles and between the baffles and
the casing of the heat exchanger cause loss of energy. This influences the value of
the heat transfer coefficient since part of the thermal carrier does not follow the
crossflow pattern. Some heat exchangers have a bypass which significantly af-
fects heat transfer. Works [159, 160] present equations for the calculation of some
forms of flow losses. The effect of the bypass on heat transfer is taken into
account in work [159], and losses due to flow of the fluid in the clearances
between the casing of the device and the baffles are considered in reference
[160].

6.4. POSSIBILITY OF CALCULATING LOCAL AND MEAN

HEAT TRANSFER

Experimentation is the traditional method for obtaining information about heat

transfer of tube banks in crossflow of liquids. For this purpose, a tube bank
 EFFECTIVENESS OF HEAT EXCHANGERS 177

model or a full-scale heat exchanger is immersed in a system containing gas or

liquid so that the tube banks are in crossflow of the thermal carrier. The tubes
are heated or cooled, the temperature distribution on the surface changed, pres-
sure losses due to resistance of the tube system measured, and dimensionless
empirical dependences are obtained for the calculation of the local and mean
characteristics of heat transfer and resistance. Investigations are conducted at
the same Reynolds numbers which are of interest to researchers. If dependences
for the calculation of other alternatives of tube arrangements in banks are
needed, a new model is constructed and the investigations repeated. In order to
conduct investigations to obtain generalizing dependences in a wide range of
varying characteristic bank parameters and Reynolds numbers, much time and
means are required. Since heat exchangers of various types are widely used in
industry, and since it is difficult to cover all versions available experimentally,
researchers are turning their attention towards the analytical solution of the
problem. One of the most promising methods is the investigation of complex
flows taking reliable experimental data into account. On the basis of the pro-
posed method of calculating complex flows [167], computation of local and
mean heat transfer is conducted [168] for staggered tube banks in crossflow
with a X b = 1.67 X 2.44 at Re, = 75 and witha x b = 0.89 x 1.77 at Re,
from 1 to 10°. Good correlation between calculations and experimental data is
obtained. The problem of calculating the mean heat transfer at higher Re, val-
ues, which is of more practical significance, is not yet resolved.
The possibilities of theoretical calculation of heat transfer and resistance in
tube banks are discussed in work [169]. The results of the calculations are
compared with the experimental data for staggered tube banks with a x b =
Oa 2 ial lax 22 lol xX 2.2). optained at Re= «2 x 10°. As
velocity profiles indicate, the boundary layer in the frontal side of the tubes in a
bank at Re; < 10° is laminar despite the 25% turbulence of the external flow.
Agreement between the analytical and experimental data is satisfactory, and the
difference exceeds 15% for only a compact tube arrangement in the bank.
To design new and highly effective heat transfer equipment, local heat
transfer coefficients must be known, together with the resistance along the
perimeter of the tubes in the bank. One must also know the transfer of momen-
tum and heat in the boundary layer in the vicinity of the tube surface and in the
spacing between tubes. These data help solve the most important contemporary
problems when designing new and effective heat exchangers, namely,

° effectively make use of the pressure drop in the heat exchanger

e increase the efficiency of the whole device
¢ control flow in the intertubular spacing
e ensure optimum conditions for the process of heat and momentum transfer
e improve the quality of the heat exchanger
e decrease the weight and dimensions of the heat exchanger
178 HEAT TRANSFER IN TUBE BANKS IN CROSSFLOW

e determine any overheating in tube cross sections

e increase reliability and service life of heat exchangers

The numerical calculation of turbulent boundary layer in a thin shearing

layer is not considered difficult at the present time. However, as far as complex
flows are concerned, such as at the point of separation of the turbulent bound-
ary layer in tube banks in crossflow, elliptical equations must be used. To solve
these equations concrete turbulence models are resorted to. Analysis of availa-
ble turbulence models shows that they are not capable of expressing all flow
complexities and they need to be corrected. For any turbulence model to be
used in conjunction with tube banks in crossflow, the constants in the differen-
tial equations must be determined. Unfortunately, the accuracy of the available
data of the investigations of tube banks in crossflow does not satisfy the con-
temporary demands. The used of the known turbulence models cannot, there-
fore, give the results needed for practical purposes, which is indicated in works
[170-172]. The majority of the devised methods were checked for thin shearing
layers described by parabolic equations; therefore, they need further improve-
ments in order to be used for calculating complex flows.
Additionally, the accuracy of the data of experimental measurements using
thermoanemometers is adequate to check the models of turbulence in the
boundary layer, but not in the case of recirculating flow. Therefore, investiga-
tions in a number of leading scientific centers are being conducted with large
scale models which can obtain recirculating flows with different degrees of
turbulence of the incoming flow. A turbulence model for the calculation of
separating flows is also being developed and optimized on the basis of careful
investigations. The investigation of flow structure in separating flows is vital in
order to widen our knowledge of complex flows. The elliptical equations must
therefore be solved on the basis of the results of the experimental investigations
of separating flows with recirculating stream. As shown by the latest investiga-
tions, the turbulence model ‘‘k — e’’ must be modified on the basis of the
experimental data of the measurement of turbulence characteristics with the
purpose of obtaining accurate results for the separating region.
The integral method, described in works [174-179 and others], is success-
fully used for the practical calculations of local heat transfer and resistance
when designing thermal energy devices. This method is inferior to the other
numerical methods as far as accuracy is concerned, but much less labor inten-
sive. In the majority of cases, it is enough to know the mean characteristics of
heat transfer in the boundary layer on the surface surrounded by the fluid when
designing new devices. This is achieved by the integral calculation method. The
feasibility of its use for practical purposes was noted in the national and interna-
tional literature.
In conclusion, the following basic tasks can be noted when designing mod-
ern heat exchangers.
 EFFECTIVENESS OF HEAT EXCHANGERS 179

. Further fundamental investigations of heat transfer in a wide range of Rey-

nolds and Prandtl numbers
. Development of methods of heat-transfer intensification for modern heat
transfer equipment
. Expansion of investigations that are directed towards the development of
heat transfer processes and devices which will be energy efficient and
operable on secondary energy sources
. Development of new and improved methods of computer-aided design with
the purpose of speeding implementation of the obtained scientific results in
the production of effective devices. This could also lead to a substantial
increase in productivity and product quality.
— | ,Ab Vey Al ay

dos Chctham ele mae EBD ol

| etm A ean ban dike
ocwfi wi na eigsh iw «i nibs Ww fae GeTAVGL
-
| ATGAe, cole MS §
soit ahem <i) to aie hopitnen cal ty pakiggienn
5 ey ; ‘WALD Mo>. WEL anesof oe
POIs, CAs Vajap ays i On Y
| ar? poreberet
e
ib.
L okin Pe Oe, PTL
| “Svar Limes Cif. ad}, :
fg tiga ctattay aad et
(7) (rey)
tierra >
 APPENDIXES

I. CHARACTERISTICS OF THE THERMOPHYSICAL

PROPERTIES OF FLUIDS

Table 1. Air.

oes ey Ges , eee Boe <

20 1.205 1.005 2.59 15.06 0.703

30 1.165 1,005 2.67 16,00 0.701
40 1,128 1.005 2.76 16,96 0.699
50 1,093 1,005 2.83 17.95 0.698
60 1.060 1,005 2.90 18.97 0.696
70 1.029 1.009 2.97 20.02 0.694
80 1,000 1,009 3,05 21.09 0.692
90 0.972 1.009 3.06 22.10 0.690

Table 2. Water.
20 998.2 4.18 59.9 1,006 7.02
30 995.7 4.17 61.8 0.805 5.42
40 992.2 4.17 63.4 0.659 4.31
50 988.1 4,17 64.8 0.556 3,54
60 983.2 4.18 65.9 0.478 2.78
70 977.8 4.19 66.8 0.415 2.55
80 971.8 4.20 67.5 0,365 DpH)

90 965.3 4.21 68.0 0.326 IFO

181
 182 APPENDIXES

Table 3. Transformer Oil.

30 890 1.936 12.34 15,90 Dyy
40 882 1.964 12.20 10.80 153
50 872 1,995 12,05 7,80 112.6
60 862 2.02 11,93 5,87 85.6
70 852 2.05 11,79 4.60 68.2
80 843 2,07 11.66 Bae 55.2
90) 833 2.10 LS 3.08 46,8

Table 4. Aviation Oil.

aie} 2) 6
5 3 c, * 10 f INE Oe PietOns
ise ee J/kg - K W/m: K m/sec ae

20 895.3 1.985 0,140 981.0 12406

30 889,5 2.014 0.140 467.5 5995

40 833.6 2.043 0,138 244.2 3194

50 877.6 2.072 0,137 Sie 1820

60 871.7 2.098 A, 85.6 1144

70 865.8 2.118 0.135 55.6 754

80 860.0 2.139 0.134 37.5 515

90 854,0 2.160 26.0 359

100 848.0 2.181 0,133 18.1 252

II. TUBE BANKS INVESTIGATED AT IPTPE AS LitSSR

Table 5. Staggered banks of smooth tubes.

No. No.
cena axb Re Pr oer axb Re Pr

| 2,6x1.3 270—1,6-108 0.7-—270 13 1.47 x 1.04 10°—2,1-108 0.7

D 2.6x0.92 290—1.4-10° 0.95—220 14 NOSe3.9 160—2-105 0.7

— 382

3 2.48x1,.28 104—8-10° 0.7 15 1 eae rs) 170—3-105 2.8

— 290

A 2 xa 10!—1,4-108 0.7 16 jag sen ee 120—3-105 0.7

-- 315

5 2.11.85 260—1,6-108 2.8—220 17 shoes) 13} 160—1.3-10§ 0,7—550

6 2.1x0.61 170—1.6-10° 0.7—245 18 125ba 5 1.0—3-108 0.7

— 5000

ii 2.0x 1.25 170—2-105 0,7 —7.0 19 1.25 x 0.865 40 —2.5-108 0.7

 APPENDIXES 183

Table 5 (Continued)

of bank 407 * 8 Re Pr Brack axb Re Pr

8 1.97x0.99 95—2,3-10° 2.8—305 20 1,28 x 1.5 3-10°—1,4-108 0.7

salts 96—1.8-10° O7—21'5 21 1,26 x 0,96 4-10*—2.5-108 0.7
10 1,581.37 2-104—9,6-105 0.7 17) 1,19 x 0,94 3.2-104— 2-108 0.7
SSeS: 1.0—2-108 0.7—5000 23 1,026x0,89 105—2-10% 2.29
— 4,8
12 1,5x0.865 1,2-10°—2-108 0.7 24 2.0 x 2.0 1—2-10° 0,7
— 5000

Table 6. In-line banks of smooth tubes (mean heat transfer) [1, 36-38, 49].

No. axb Re Pr No. axb Re Pr

of bank of bank

1iee.63-1510 10?—1,6-105 0.7 —400 16 IPS <elins 700 — 108 0,7

— 400
2 2.60~x 1.30 10°—1,6-10° 0.7-—400 17 1.26 x 2.00 700—6-105 0.7
— 400
3 2,52x1.89 1.6-104—8,1-105 0.7 18 1.26x1,26 4,5-10°'—1,4-108 0.7
4 2.45 x 2.00 200 —2.5-10° 0.7—400 19 1.2531, 25 1—2-108 0.7
— 5000
5 2.00
x 2,00 1—2-10° 0.7—5000 20 1,055x1,055 5-10°—1,4-108 3-—5.0
Gue.00'<1.25 5-104—2-108 0.7 —7,0 21 1.055 x 0.89 104—2.3-10° 2.3-—5.0
7 1.95x 1,30 300 —2,5-105 0.7—400 22 1,053x1,053 3-104—1.3-10° 2.9—5.0
8 1.89x 1.89 1,6-104—9-10° Oe, 23 1.05 x 1,85 4-108—4.-10° 0.7—7,7
9 1.89x1.26 2.5-10!—9-10° 0.7 24 1,026x1.026 1.1-104—1.6-10® 3.5—5.0
10 1.68x1.70 2.0-104—9-10° 0.7 25 1.02 1.02 2.3-104—1.9-10® 3,.0—7.0
11682526 2-10 — 108 0.7 26 1.013x1.013 2-10°—1,6-108 3.0—5,0
12 OS>o Las: 9-10*—1.6- 108 0,7 27 1.01 x 1.01 3-104—2,2-10° 2.8—5.0
13) -1,65:< 2,0. 110—4-105 0.7—400 28 1,008
x 1.008 3-10®—3.-105 2,7-—5.5
14a) 105'—1,4-10° 0,7 29 1,007
x 1,007 7-104—3-108 2.8—5.5
1 Seo xX2.0 500 —4- 10° 0.7—400 30 1.25 x 2.0 5-104—2-10° 0.7—7.0
 184 APPENDIXES

Table 7. Staggered finned banks (local and mean heat transfer) [6, 112].

ie axb Re Pi h, mm s, mm € 8/3;

1 4.13 x 1,46 1-10*—5-108 0,7 10 6.5 5.23 2.0

2 411x214 1-10!—S5-10° 0.7 10 6,5 D525 2.0
3 4.04 x 1,98 1. 10*—5-108 0,7 10 6,5 a5 2.0
4 4.04 x 1.78 1. 10'—S5-10° 0.7 10 6.5 see 2.0
5 3,74 x 2,01 1-10'—6-10° 0,7 10 6.5 S220 2.0
6 3,13 x 1,78 1-104—5-10° 0.7 10 6,5 Y23 2.0
7 2,97 x 1,65 1.3-10*—8-10° 0.7 13S 6 8.2 2.0
8 274X211 [-10*— 6-108 0,7 10 6.5 5.23 2.0
9 2.74 x 1.64 1-10'—6-105 0.7 10 6.5 NWA, 2.0
10 2.67 x 1.46 1-10*—5-108 0.7 10 6.5 5,23 2.0
il 2.67 X147 1-10'—6.5-105 0.7 9 6 S15 2.0
12 2.67 x 1.46 1-10'—6,5-10° 0,7 6 6 3.47 2.0
13 2.67 x 1.46 1.4-10*—1-108 0.7 4 4 3.18 2.0
14 2.38x 1.46 5. 10'—1,1-108 0.7 13.5 6 foo 2.0
15 2.38x 1.46 2-104*—1,1-108 0.7 13.5 4 10.6 2.0
16 224 130 2,2-104—1,1-108 0.7 9 4 6.25 2.0
‘WE PLE
ARSHWOR 2.2-10'—1.2-108 OF 9 6 4.78 2,0
18 PUG i ioe 2.6-10*—8.5-10° 0.7 6 6 3,29 2,0
19 PNG ENS oe 8.5-104—1.3- 10° 0.7 6 4 4,42 2.0
2 2.213 8.5-10%— 1,3-108 0,7 4 4 3.09 2.0
ai PAT cee 2,2-10%-- 12-108 0.7 9 8 3.86 2.0
22 1.66 x 1,53 1.8-108—S-10* 84 — 220 18 9 Swis 6.95
23 1,66 x 1.53 8-10®—5-108 84—220 1) 9 2.66 4.95
24 1.66 x 1,53 5-10°—4.5-10' 84—220 3 9 1,43 4.1
25 1,66x 1.53 2-102—2.5-10! 84—220 5 5.13 6.95
26 1,53 x 1.56 10°—1.6-105 0.7 20 24 2,96 1,29
27 1.33'X1,23 2-108—6-10! 84 — 220 Tie 9 2.66 4.95
28 L338 x 1.23 3,8-10?—2.5-10! 84—220 238 3 2.66 4.95
29 1,33 x 1.06 4-10°—1.1-108 84-220 3 9 1.43 4.1
30 1.13x 1.06 1-10°—3-10 = 84—220 1,43 4.1
 APPENDIXES 185

Table 8. Rough staggered banks (local and mean heat transfer) [77].

Rite aX Ke Pr k, mm (k/d)- 108

1 P0220 100— 2-104 84 — 300 0 0

2 2.02.0 — 2-104
100 84 — 300 0.2 6.67
3 2.0 x 2.0 100 — 2-101 84 — 300 Wee 40
4 P25 525 10'—7-10° 0.7 — 220 0 0
5) eS 6 11 ss 10*—7-10° 0.7 — 220 0.45 3
6 258125 10'—7- 10° 0.7 — 220 Hae 8
Ti 1 PSS 12S) 2-10° 0.7 0.2 133
8 leo Sale ZO? 0.7 k= 051 k,/d=0,67
9 Dalen) 2-10° 0.7 [et 0)ss) Kejd=2.0
10 LEDS eZ 2-10° 0.7 k,=0.6 k,/d=4.0
11 1625°<0:935 400 —4-10% 84 — 300 0 0
WV? 1,25 x 0,935 —
400 4-104 84-3600 0.2 6.67
13 1.25 x 0,935 —
400 4-10? 84 — 300 0,45 1S
14 1.25 x 0.935 400 — 4-103 84 — 300 ie 40

Table 9. Inclined tube banks (mean heat transfer) [58].

No. axb Re je No. axb Re Pr 6

of bank of bank

Staggered banks In-line banks

1 1,34x1,16 10*—2-10§ 0.7—7.0 90 1 1,341.34 10#—2-10° 0.7—7,.0 90
D 1.34x«1,16 10'—2-10° 0O7—-7,0 75 22 1341.34 104—2-10° 0,.7—7.0 75
3 1,341.16 10*—2-10° 0.7—7.0 60 3 1.34 1,34 104-2-108 0.7—7.0 60
4 1.34x1.16 10!—2-10° 0,7—7,0 40 4 1,341.34 10!—-2-10° 9.7—7.0 40
5 1,341.16 104—2-10° 0O.7-7.0 25 5 {.34x1.34 10!—2-10° 0.7-7.0 25

Table 10. Staggered finned banks from round helical finned aluminum
tubes (mean heat transfer) [180].

No. of bank axb Re* ee h, mm s, mm € 6, mm

I Sols 2.103 2-10! 0.7 2,8 x 5,06 0.5

D; ef lexaleas 2-108—2-10* 0,7 2.8 z 5,06 0.5

3 1.90 x 1.43 2-108— 2-10! 0.7 se) : 5.06 0.5

 186 APPENDIXES

Table 11. Staggered finned banks from plane-oval finned aluminum tubes
(mean heat transfer) [126].
No. of bank axb Re* Pr h, mm s, mm € 6, mm

1 2.00 x 1.26 4-10°—4-10° 0.7 Ds 2, 5,06 0.5

2 Pea aless 4-10?—4-103 0.7 2.9 2, 5.06 0.5
3 3.09 x 1,33 4.10?—4-108 0.7 2 2 5.06 0,5

Table 12. Staggered banks from plane-oval finned aluminum tubes with
angled fins (mean heat transfer) [126].

No. of bank axb Res Pr h, mm s, mm € 6, mm

| 25451033 4-10?—4-108 0.7 2.9 2 5,28 0.5

Ret = Un ny, a= 4/6

f-—Minimum cross-sectional area of bank, m?

L—Depth of the whole bank, m
F—Total heat transfer surface of bank, m2
 REFERENCES

: Zukauskas, A., Makarevicius, V., and Slanciauskas, A. Heat transfer in banks of tubes in
cross-flow of liquids. Vilnius, Mintis. 1968. 192 pp.
; Zukauskas, A., Heat transfer from tubes in cross-flow. In: Adv. Heat Transfer. New York,
London, Academic Press, 1972, Vol. 8, pp. 93-160.
. Zukauskas, A. Convective transfer in heat exchangers. Washington, D.C., Hemisphere Pub-
lishing Corporation (In press).
Zukauskas, A., and Ziugzda, J. Heat transfer of a cylinder in cross-flow. Washington, D.C.,
Hemisphere Publishing Corporation, 1985, 208 pp.
é Zukauskas, A., Ulinskas, R., and Katinas, V. Fluid dynamics and flow-induced vibrations of
tube banks. Washington, D.C., Hemisphere Publishing Corporation, 1987.
. Stasiulevicius, J., and Skrinska, A. Heat transfer of finned tube bundles in crossflow. Wash-
ington, D.C., Hemisphere Publishing Corporation, 1987.
Mikheiev, M. Fundamentals of heat transfer. Moscow, Mir Publishers, 1968. 376 pp.
Isachenko, V. P., Osipova, V. A., and Sukomel, A. S. Heat transfer. Moscow, Energiya,
1975. 486 pp.
Petukhov, B. S. Heat transfer and resistance of tubes in laminar flow of liquids. Moscow,
Energiya, 1967. 411 pp.
. Antuf’iev, V. M., and Kozachenko, L. S. Heat transfer and aerodynamic resistance of heated
convective surfaces. Moscow, ONTI, 1938.
. Kuznetsov, N. V. Heat transfer and resistance of tube banks in cross-flow for different ar-
rangements. Izd. AN SSSR, 1937, No. 5, pp. 675-695.
. Pierson, O. L. Experimental investigation of the influence of tube arrangement on convection
heat transfer and flow resistance in crossflow of gases over tube banks. Trans. ASME, 1937,
Vol. 59, No. 7, pp. 563-572.
Antuf’iev, V. M., Beletsky, G. S. Heat transfer and aerodynamic resistance of tubular sur-
faces in cross-flow. Moscow, Mashgiz, 1948. 123 pp.
Grimison, E. D. Correlation and utilization of new data on flow resistance and heat transfer
for crossflow of gases over tube banks. Trans. ASME, 1937, Vol. 59, No. 7, pp. 583-594.
IBY. Kirpichev, M. V., and Mikheiev, M. A. Modelling of thermal devices. Moscow-Leningrad.
Izd. AN SSSR, 1936. 360 pp.
16. Kuznetsov, N. V. and Turilin, S. I. Effect of temperature conditions on heat transfer and
resistance of tubular surfaces in cross-flow. Izd. WTI, 1952, No. 11, pp. 23-27.

187
188 REFERENCES

. Kuznetsov, N. V., and Karasina, E. S. Formulas for heat transfer coefficient for smooth-tube
banks in cross-flow. Teploenergetika, 1954, No. 6, pp. 31-35.
. Kirpichev, M. V., and Eigenson, L. S. A thermal turbulence-measuring device. Izd.
Energeticheskovo in-ta AN SSSR, 1936, Vol. 4, No. 1, pp. 81-89.
. Bressler, R. Versuche uber den Druckabfall in quer angestromten Rohrbundeln. Forschung
auf dem Gebiete des Ingeneurwesens, 1958, Vol. 24, No. 3, pp. 365-368.
. Kays, W. M., and London, A. L. Compact heat exchangers. McGraw-Hill Book Company,
1964. 272 pp.
. Lyapin, M. F. Heat transfer and aerodynamic resistance of smooth-tube banks for high Re,
numbers of gas flow. Teploenergetika, 1956, No. 9, pp. 49-52.
22s Kruzhilin, G. N., and Schwab, V. A. Investigation of field on the surface of circular cylinder
in cross-flow of air at Reynolds numbers in the range 21 X 10°. ZhTR, 1935, Vol. 5, No. 4,
pp. 707-710.
23% Mikhailov, G. A. Convective heat transfer in banks of tubes. Sovietskoye kotloturbos-
troeniye, 1939, No. 12, pp. 434-437.
24. Gomelauri, V. I. Heat transfer in a stream of fluid droplets. Tr. Energeticheskovo in-ta AN
GSSR, 1948, Vol. 4, No. 4, pp. 2-10.
255 Zukauskas, A. A., Heat transfer in a cylinder in cross-flow of liquid. Teploenergetika, 1955,
No. 4, pp. 38-40.
26. Bergelin, O. P., Brown, G. A., and Doberstein, S. C. Heat transfer and fluid friction during
flow across banks of tubes. Trans. ASME, 1952, Vol. 74, No. 6, pp. 953-960.
Pfs Bergelin, O. P., Brown, G. A., Hull, H. L., and Sullivan, F W. Heat transfer and fluid
friction during viscous flow across banks of tubes. Ibid. 1950, Vol. 72, No. 6, pp. 881-888.
28. Isachenko, V. P. Heat transfer from banks of tubes in cross-flow of different liquids. Te-
ploenergetika, 1955, No. 8, pp. 19-22.
29! Zukauskas, A.A., Makarevicius, V. I., Indriunas, A. I., and Slanciauskas, A. A. Heat trans-
fer from in-line bank of tubes in cross-flow of liquid. Tr. AN LitSSR., Series B, 1957, Vol.
4(12), pp. 143-149.
30. Zukauskas, A.A., Makarevicius, V. I., Indriunas, A. I., and Slantiauskas, A. A. Heat trans-
fer from staggered tube banks with a = 2.1 and b = 1.9 in cross-flow of a liquid. Ibid.,
1958, Vol. 4(16), pp. 165-168.
Si: Stasiulevicius, J., and Samoska, P. S. Heat transfer and aerodynamic resistance of staggered
banks in cross-flow of air in the range Re; > 10°. IFZh, 1964, Vol. 7, No. 11, pp. 10-15.
32. Borishansky, V. M., Andreievsky, A. A., and Zhinkina, V. V. Heat transfer from banks of
tubes in cross-flow of water and metallic sodium. In book: General problems of heat transfer.
Heat and mass transfer. Minsk, 1962, Vol. 3, pp. 475-480.
33: Zukauskas, A., and Ulinskas, R. Some aspects of the heat transfer for banks of tubes in cross-
flow for the low range of Reynolds numbers. In: 6th Int. Heat Transfer Conf., Toronto,
Canada, Aug. 7-11, 1978, Vol. 4, pp. 243-248.
34. Schlichting, G. Boundary-layer theory. New York. McGraw-Hill, 1960.
35. Krall, K. M., and Eckert, E. R. G. Heat transfer to a transverse circular cylinder at low
Reynolds numbers including refraction effects. Heat Transfer, 1970, Ser. FC5, Vol. 3, Paris-
Versailles.
36. Zukauskas, A., and Ulinskas, R. Analysis of lateral streamline flow of water through tube
bundles during the process of heat transfer at critical Reynolds numbers. Jnt. Chemical Engi-
neering, 1977, Vol. 17, No. 4, pp. 673-676.
37. Zukauskas, A., Ulinskas, R., and Marcinauskas, K. Influence of the geometry of the tube
bundle on the local heat transfer rate in the critical region of streamiine flow. Int. Chemical
Eng., 1977, Vol. 17, No. 4, pp. 744-751.
38. Poskas, P., Survila, V., and Zukauskas, A. Local heat transfer rate at a tube in a corridor
bundle lateral to an air stream at large Reynolds numbers. Int. Chemical Eng., 1978, Vol. 18,
No. 2, pp. 337-343.
 REFERENCES 189

39) Frossling, N. Verdunstung, Warmetbergang und Geschwindigkeitsverteilung bei zweidimen-

sionaler und rotationsimmentrischer Grenzschichtstromung. Lands. Univ. Arsskr. Avd. 2,
1940. Vol. 36, No. 4, pp. 25-35.
40. Mayinger, T., and Schad, O, Ortliche Warmeubergangszahlen in querangestromten Stabbun-
deln. Warme-und Stofftibertragung, 1968, Vol. 1, pp. 43-51.
. De Bortoli, R. A., Grimble, R. E., and Zerbe, J. E. Average and local heat transfer for cross
flow through a tube bank. Nucl. Sci. Eng., 1956, No. 1, pp. 15-21.
. Winding, C. C., and Cheney, A. I. Mass and heat transfer in tube banks. Industr. Eng.
Chem., 1948, Vol. 40, No. 6, pp. 11-17.
. Neal, S. B. H. C., and Hitchcock, J. A. The development of improved heat transfer surface
for tubes in cross-flow, using a large scale model technique. In: 4th Int. Heat Transfer Conf.,
Paris-Versailles, 1970, F.C. 7. 8, pp. 1-11.
. Zukauskas, A., and Ulinskas, R. Heat transfer at high Reynolds numbers in flow of water
over tube banks. Thermal Engineering (Int. Quart. J.ISME), 1980, Vol. 1, No. 1, pp. 25-29.
. Mikheieva, I. M. Heat transfer in various free-moving liquids. In book: Heat transfer and
thermal modelling. Moscow, Izd. AN SSSR, 1959, pp. 226-238.
. Eckert, E. R. G., and Drake, J. R., Analysis of Heat and Mass Transfer. McGraw-Hill Book
Co., USA, 1972. 806 pp.
. Ialuria, Y. Natural convection. Heat and Mass Transfer, USA, Pergamon Press, 1980. 4C0
PP.
. Martynenko, O. G., and Sokovshin, Yu. A. Heat transfer by mixed convection. Minsk:
Nauka i teknika, 1975. 256 pp.
; Zukauskas, A., Ulinskas, R., and Sipavicius, C. Average heat transfer and pressure drop in
cross flow of viscous fluids over tube bundles at low Re. Heat Transfer-Soviet Research,
1978, Vol. 10, No. 6, pp. 90-101.
50. Lokshin, V. A., Antonov, A. Ya., Mochan, S. I., and Revzina, O. G. Generalization of data
on heat transfer for clean smooth tube banks in cross-flow. Teploenergetika, 1969, No. 5, pp.
21-25.
Se Boissier, A., Chatillon de M., Gautier, D., and Perez, R. Les pertes de charge et le transfert
thermique, cote gaz, dans les ¢ changeurs de chaleur a tubes lisses, a circulations ortho-
gonales. Bull. de la direct. des études et recherches. Ser. A, 1971, No. 2/3, p. 241.
52. Stasiulevicius, J. K., and Samoska, P. S. Heat transfer for smooth-tube staggered banks in
cross-flow of air at high Re numbers. Tr. AN LitSSR. Series B, 1963, Vol. 4(35), pp. 77-81.
53: Stasiulevicius, J. K., and Samoska, P. S. Heat transfer and aerodynamics of staggered banks
of tubes in cross-flow of air in the region Re > 10°. IFZh, 1964, Vol. 7, No. 11, pp. 11-15.
54. Zukauskas, A., and Ulinskas, R. Heat transfer efficiency of tube bundles in crossflow at
critical Reynolds numbers. Heat Transfer-Soviet Research, 1978, Vol. 10, No. 5, pp. 9-15.
Se, Hammeke, K., Heinecke, E., and Scholz, F. Warmeubergangs und Druckverlustmessungen
an querangestromten Glattrohrbundeln, insbesondere bei hohen Reynoldszahlen. Int. J. Heat
Mass Transfer, 1967, Vol. 10, pp. 427-446.
56. Zukauskas, A. Heat transfer of banks of tubes in crossflow at high Reynolds numbers. In:
Heat Exchangers: Design and Theory Sourcebook (edits. N. Afgan and E. U. Schlunder).
Washington, D.C., McGraw-Hill Book Co., USA, 1974, pp. 75-100.
aie Niggeschmidt, W. Druckverlust und Warmeubergang bei fluchtenden, versetzten und teilver-
setzten querangestromten Rohrbundeln. Diss., University of Darmstadt, 1975.
58. Zukauskas, A. Heat transfer of banks of tubes in crossflow at high Reynolds numbers. In:
Heat Exchangers: Design and Theory Sourcebook, (edits. Afgan, N. H., and Schlunder, E.
U.) McGraw-Hill Book Co., Washington, D.C., 1974, pp. 75-100.
5: Kazakevich, F. P. Investigation of heat transfer in tube banks at different angles of attack of
gaseous stream. Teploenergetika, 1954, No. 8, pp. 22-28.
60. Lokshin, V. I. Influence of angle of attack on heat transfer in tube banks. Teplosilovoye
khozyaestvo, 1940, No. 8, pp. 29-32.
190 REFERENCES

61. Zukauskas, A., Katinas, V., and Perednis, E. Average heat transfer coefficients of tube
bundles in flows of viscous fluids at different angles of attack. Heat Transfer-Soviet Research,
1983, Vol. 15, No. 2, pp. 22-30.
62. Zukauskas, A. Heat transfer in turbulent flows of various fluid. In: Turbulent forced convec-
tion in channels and bundles. Theory and Applications to Heat Exchangers and Nuclear
Reactors. Edit. Kakac, S., and Spalding, D. B., Hemisphere Publishing Corporation, Wash-
ington, D.C., 1979, Vol. 2, pp. 547-561.
63. Zukauskas, A., and Slanciauskas, A., and PediSius. A. Heat transfer in a turbulent boundary
layer behind a two-dimensional bluff body at different Pr numbers. Heat Transfer, Munchen,
1982, Vol. 3, pp. 217-227.
. Dipprey, D. F., and Sabersky, R. H. Heat and momentum transfer in smooth and rough tubes
at various Prandtl numbers. Int. J. Heat Mass Transfer, 1963, Vol. 6, No. 5, pp. 329-335.
. Yaglom, A. M., and Kader, B. A. Heat and mass transfer between rough wall and turbulent
fluid flow at high Reynolds and Peclet numbers. J. Fluid Mech., 1974, Vol. 62, pp. 601-623.
. Gomelauri, V. I. Influence of artificial roughness on convective heat transfer. Tr. In-ta phyziki
AN GSSR, 1963, Vol. 9, pp. 111-145.
. Migay, V. K. Heat transfer in rough tubes. Izv. AN SSSR. Energetika i transport, 1968, No.
3, pp. 97-107.
. Isachenko, V. P., Agababov, S. G., and Galin, N. M. Experimental investigation of heat
transfer and hydraulic resistance for tubes with artificial surface roughness in turbulent flow
of water. Tr. MEI, 1965, 63, pp. 27-37.
. Kalinin, E. K., Dreitser, G. A., and Yarkho, S. A. Intensification of heat exchange in ducts.
Moscow, Mashinostroyenie, 1972, 320 pp.
. Achenbach, E. The effect of surface roughness on the heat transfer from a circular cylinder to
the cross flow of air. Int. J. Heat Mass Transfer, 1977, Vol. 20, No. 4, pp. 359-369.
: Zukauskas, Ase Ziugzda, J., and Daujotas, P. Effects of turbulence on the heat transfer of a
rough-surface cylinder in cross-flow in the critical range of Re. In: 6th Int. Heat Transfer
Conf. Toronto, Canada, Aug. 7-11, 1978, Vol. 4, 231-236.
. Zukauskas, A., and Ulinskas, R. Surface roughness as means of heat transfer augmentation.
Heat Exchangers. New York, Hemisphere Publishing Corportion, 1982, pp. 311-321.
. Hudina, M. Rough surface, heat transfer and pressure drop. Eidg. Institut fur Reaktorfors-
chung Wurelingen, Schweiz, 1982, EIR Bericht No. 474, pp. 1-220.
. Puchkov, P. I. Influence of surface roughness on heat transfer in banks of tubes in cross-flow.
Kotloturbostroyenie, 1948, No. 4, pp. 5-6.
. Groehn, H. G., and Scholz, F. Heat transfer and hydraulic resistance in coaxial banks of tubes
with artificially applied surface roughness in cross-flow. Teplomassoobmen-5, 1976, Vol. 1,
Part 2, pp. 37-42.
76. Groehn, H. G., and Scholz, F. Anderung von Warmeubergang und Stromungswiderstand in
querangestromten Rohrbundeln unter dem Einfluss verschiedener Rauhigkeiten sowie An-
merkungen zur Wahl der Stoffwertbezugstemperaturen. In: Proc. 4th Int. Heat Transfer
Conf., Ser. FC 7.10 P., 1970, Vol. 3, pp. 1-11.
ile Zukauskas, A., Ulinskas, R., and Daunoras, P. The influence of surface roughness on the
heat transfer and drag of tube banks in crossflow. Heat and Technology, Italy, 1985, Vol. 11,
No. 5, pp. 1-46.
78. Thomas, L. C. Fundamentals of heat transfer. Englewood Cliffs, New Jersey, 07632, 1980,
702 pp.
IEE Schneider, P. J. Conduction Heat Transfer. Cambridge, Mass. Addison-Wesley Publishing
Co., 1955, 395 pp.
80. Kern, D. Q. Process Heat Transfer. New York. McGraw-Hill Book Co., 1950, 871 pp.
81. McAdams, W. H. Heat Transmission. New York. McGraw-Hill Book Co., 1954, 532 pp.
82. Kern, D., and Kraus, A. Extended surface heat transfer. McGraw-Hill Book Co., New York.
1972.
 REFERENCES 191

83. Roezen, L. I., and Dulkin, I. N., Thermal calculation of finned surfaces. Moscow, Energiya,
1977. 254 pp.
84. Schmidt. Th. E. Der Warmeubergang an Rippenrohren und die Berechnung von Rohrbiindel-
Warmeaustauschern. Kaltetechn., 1963, Vol. 15, No. 4, pp. 98-102; No. 12, pp. 370-378.
85. Yudin, V. F, and Tokhtarova, L. S. Resistance of banks of finned tubes in cross-flow.
Energomashinostroyenie, 1974, No. 1, pp. 30-32.
86. Brauer, H. Warme- und stromungstechnische Untersuchungen an quer angestromten Rip-
penrohrbundeln. Chemie-Ing. Technik, 1961, Vol. 33, No. 5, 6, pp. 327-335, 431-438.
87. Yudin, V. FE, and Tokhtarova, L. S. Heat transfer and resistance of staggered and in-line
finned banks. Energomashinostroyenie, 1964, No. 1, pp. 11-13.
88. Jameson, S. L. Tube spacing in finned-tube banks. Trans. ASME, 1945, Vol. 67, No. 8, pp.
633-642.
89. Brigs, E. D., and Young, E. H. Convection heat transfer and pressure drop of air flowing
across triangular pitch banks of finned tubes. CEP Symp. Series, 1963, Vol. 59, No. 41, pp.
1-10.
90. Zozulya, N. V., Khavin, A. A., and Kalinin, B. L. Investigation of arrangement effect on
heat transfer and hydraulic resistance in a bank of helically finned tubes. Information letter,
No. 31, In-t teploenergetiki AN USSR, 1962.
Mle Mirkovic, Z. Heat transfer and flow resistance correlation for helically finned and staggered
tube banks in crossflow. In: N. H. Afgan and E. U. Schlunder. Heat Exchangers: Design and
Theory Sourcebook, 1974. 893 pp.
O2F Yudin, V. F. Heat transfer of cross-finned tubes. Leningrad, Mashinostroyenie, 1982. 189 pp.
93) Zozulya, N. V., Kalinin, B. L., and Khavin, A. A. Influence of arrangement of banks from
aluminum finned tubes on heat transfer. Teploenergetika, 1970, No. 6, pp. 31-32.
94. Brauer, H. Warmeubergang und Stromungswiderstand bei fluchtend und versetzt angeordne-
ten Rippenrohren. Dechema Monographic., 1962, Vol. 40, pp. 41-76.
95: Yudin, V. F., and. Tokhtarova, L. S. Influence of number of transverse finned tubes of stag-
gered and in-line banks on heat transfer and resistance. Energomashinostroyenie, 1971, No.
4, pp. 41-42.
96. Kuntiesh, V. B., and Iokhvedov, F. M. Influence of number of rows and arrangement of
finned banks in cross-flow on local heat transfer of last rows of tubes. Izv. vuzov SSSR.
Energetika, 1979, No. 3, pp. 56-59.
OU Zozulya, N. V., Kalinin, B. L., and Khavin, A. A. Investigation of heat transfer along the
rows of staggered banks of aluminum tubes with helical fins. Khimicheskaya promieshlen-
nost’ Ukraini, 1968, No. 5, pp. 29-31.
98. Kuntiesh, V. B., and Iokhvedov, F M. Experimental investigation of local heat-transfer
coefficients for tubes with helical fins in finned banks in cross-flow. Izv. vuzov SSSR.
Energetika, 1977, No. 2, pp. 105-110.
oe Liogky, V. M., and Tupitsyn, Yu. K. Certain features of heat transfer in finned-tube bundles
in crossflow. Heat Transfer-Soviet Research, 1978, Vol. 10, No. 6, pp. 112-117.
100. Antuf’iev, V. M., and Beletsky, G. C. Heat transfer and aerodynamic resistance for tubular
surfaces in cross-flow. Moscow, Mashgiz, 1948. 119 pp.
101. Krischer, O., and Kast, W. Warmeibertragung und Warmeespannungen bei Rippenrohren.
VDI-Forschung, 1959, Vol. 25, No. 474, pp. 1-58.
102. Kuntiesh, V. B., and Iokhvedov, F. M. Influence of relative depth of fin spacing on ther-
mal effectiveness, convective heat exchange of banks of finned tubes, and intensification
of heat transfer in them. Izv. AN SSSR. Energetika i transport, 1970, No. 4, pp. 127-
136.
103. Yudin, V. F., Tokhtarova, L. S., and Andreiev, P. A. Heat transfer and resistance of stag-
gered banks with fins of various height and pitch. Tr.TsKTI, 1966, 73, pp. 98-106.
104. Antuf’iev, V. M. Investigation of effectiveness of various forms of finned surfaces in cross-
flow. Teploenergetika, 1965, No. 1, pp. 81-86.
192 REFERENCES

105. Antuf’iev, V. M. Effectiveness of various forms of heated convective surfaces. Moscow-

Leningrad, Energiya, 1966, 184 pp.
106. Grass, G., and Coenen, F. P. Systematische Untersuchungen uber den Warmeubergang und
Stromungwiderstand von Rippenrohren. Atomkernenergie, 1959, No. 2, pp. 41-48.
107. Eckert, E. R., and Drake, R. M. Analysis of heat and mass transfer. Hemisphere Publishing
Corporation, 1987.
108. Vampola, I. Generalization of realtionships of heat transfer and pressure drop for banks of
finned tubes in cross-flow of gas. In book: Heat and mass transfer. Minsk, Vol. 1, 1985, pp.
260-269.
109. Kuznetsov, E. F. Heat transfer and resistance for heat exchange surfaces of air and gas
coolers for compressors. In book: Turbine and compressor engineering. Leningrad,
Mashinostroyenie, 1970, pp. 78-100.
110. Liogky, V. M., and Pis’menny, E. N. About one mechanism of heat exchange in staggered
banks of tubes with circular fins in cross-flow. Izv. vuzov SSSR.Energetika, 1982, No. 11,
pp. 107-111.
111. Groehn, H. G. Warme- und stromungstechnische Untersuchungen an einem querdurchstrom-
ten Rohrbundel: Warmeaustauscher mit niedrig berippten Rohren bei grossen Reynolds-
Zahlen. Ber. Kernforschungsanlage, Julich, 1977, N. 1462, p. 72.
112. Zukauskas, A. A., Ulinskas, R. V., and Zinevicius, F V. Heat transfer and resistance of
staggered banks of finned tubes in cross-flow. Izv. AN SSSR, Energetika i transport, 1983,
No. 4, pp. 117-124.
113; Makarevicius, V. I. Heat transfer during physical and chemical changes. Vilnius, Mokslas,
1978, 228 pp.
114. Schmidt, Th.E. Heat transmission and pressure drop in banks of finned tubes and in laminated
coolers. Gen. Discussion Heat Transf., Inst. Mech. Eng., 1951, pp. 186-188.
115. Zukauskas, A. A., Ulinskas, R. V., and Zinevicius, F. V. Local characteristics of heat trans-
fer and flow for staggered banks of finned tubes. Tr. AN LitSSR. Series B, 1984, Vol. 2(141),
pp. 46-55.
116. Koshmarov, Yu. A., Svirshevsky, S. B.; and Inozemtseva, E. N. Heat transfer and resistance
of finned tubes at low Reynolds numbers. Temat. sb. nauch. tr. MAI, 1978, 463, pp. 33-37.
117. Yudin, V. F., Tokhtarova, L. S., Lokshin, V. A., and Tulin, S. N. Gereralization of experi-
mental data on convective heat transfer in cross-flow for banks of tubes with transverse band
and circular fins. Tr. TsKTI, 1963, 82, pp. 108-134.
118. Andreiev, A. P., Gremilov, D. I., and Fedorovich, E. D. Heat-exchange devices for nuclear
energy plants. Leningrad: Sudostroyenie, 1969. 353 pp.
Pi9s Berman, Ya. A., Man’kovsky, O. N., Marr, Yu. N., and Rafalovich, A. P. Cooling systems
for compressor installations. Leningrad, Mashinostroyenie, 1984. 228 pp.
120. Fraas, A., Ozisik, M. Heat exchanger design. John Wiley & Sons, Inc., New York-London-
Sydney.
Pail Grigor’iev, V. A., Krokhin, Yu. I. Heat and mass transfer devices for cryogenic engineering.
Moscow, Energoizdat, 1982. 312 pp.
122. Burkov, V. V., and Ideikin, A. I. Auto-tractor radiators. Leningrad, Mashinostroyenie, 1978.
216 pp.
123: Dul’nev, G. N. Heat and mass transfer in radio-electronic devices. Moscow, Vieshaya
shkola, 1984. 247 pp.
124. Migay, V. K., Bystrov, P. G., and Morgunoy, A. V. Investigation of local heat transfer of
heated convective membrane surfaces of boilers. Teploenergetika, 1982, No. 10, pp. 43-46.
125s Lipets, A. U., Nekrasov, M. I., Pikus, V. Yu., and Dvoinishnikov, V. A. Two-layer finned
tubes for boiler economizers. Energomashinostroyenie, 1983, No. 8, pp. 41-42.
126. Ilgarubis, V. S., Ulinskas, R. V., Butkus, A. V., and Zukauskas, A. A. Hydraulic resistance
and average heat transfer in compact banks from flat-oval finned tubes. Tr. AN LitSSR. Series
B, 1987, Vol. 1(158), pp. 49-56.
 REFERENCES 193

127. Lymer, A., and Ridal, B. F. Finned tubes in cross-flow of gas. J. Brit. Nucl. Energ. Conf.,
1961, Vol. 6, No. 4, pp. 307-313.
128. Neal, S. B. H. C., and Hitchcock, J. A. A study of the heat transfer process in banks of
finned tubes in cross flow, using a large scale model technique. In: Proc. 3rd. Int. Heat
Transfer Conf., Chicago, 1966, pp. 290-298.
129: Kraus, A. D., and Snider, A. D. New parameters for heat exchange problems in fins. J. Heat
Transfer, 1980, Vol. 102, No. 3.
130. Tulin, S. N. Refinement of computation formulas for heat transfer and resistance in banks of
tubes with wire fins. Energomashinostroyenie, 1971, No. 3, pp. 31-33.
S16 Groehn, Kh. G., Scholz, F. Heat transfer and hydraulic resistance for coaxial banks of tubes
with artifically applied surface roughness in cross-flow. In book: Heat and mass transfer-5,
Minsk, 1976, Vol. 1, Part 2, pp. 37-42.
132% Yudin, V. F., Tokhtarova, L. S., Lokshin, V. A., and Tulin, S. I. Generalization of experi-
mental data on convective heat exchange for staggered banks of tubes with fins in cross-flow.
Tr.TsKTI, 1968, 86, pp. 132-142.
. Zukauskas, A., and Ulinskas, R. Efficiency parameters for heat transfer in tube banks. Heat
Transfer Engineering, 1984, Vol. 6, No. 1, pp. 19-25.
134. Gukhman, A. A. Some aspects of the theory of high intensity convective heat-transfer pro-
cesses. ZhTF, 1953, Vol. 23, 6 pp. 1064-1114.
. Gukhman, A. A. Intensification of convective heat transfer and the problem of comparative
evaluation of heat exchange surfaces. Teploenergetika, 1977, No. 4, pp. 5-8.
. Kirpichev, M. V. The most optimum form for heated surfaces. Tr. ENINN AN SSSR, 1944,
Vol. XII, pp. 5-8.
. Kirpichev, V. A., and Leifman, V. I. Graphical method for comparative evaluation of effec-
tiveness of heated convective surfaces. Teploenergetika, 1975, No. 3, pp. 34-36.
. Andreiev, M. M., Berman, S. S., Buglaev, V. T., and Kostrov, Kh. N. Heat-exchange
devices for energy installations. Moscow, Mashgiz, 1963, 240 pp.
. Lyshevsky, A. S., Sokolov, V. G., Sychev, V. M., and Kutukov, A. A. Investigation of heat
transfer and resistance of banks of tubes with cylindrical spikes in cross-flow of gas stream at
low Reynolds numbers. In book: Heat supply and heat/mass transfer. Rostov n/D, 1977, pp.
63-71.
. Mitskevich, A. I. Method for evaluation of effectiveness of convective heat transfer.
Tr.TsKTI, 1967, 78, pp. 3-25.
. Mitskevich, A. I. Effectiveness of heat transfer surfaces. Teplo- i massoobmen, Vol. 1,
Minsk, 1965, pp. 270-276.
. Yudin, V. F. Method for comparative evaluation of heated convective surfaces. Energo-
mashinostroyenie, No. 5, 1969, pp. 31-34.
. Antuf’iev, V. M. Effectiveness of various forms of heated convective surfaces. Moscow-
Leningrad, Energiya, 1966. 184 pp.
. Cox, B., Jallouk, P. A. Methods for evaluating the performances of compact heat transfer
surfaces. Trans. ASME, 1973, No. 4, pp. 35-42.
. Antuf’iev, V. M. Comparative investigations of heat transfer and resistance of finned sur-
faces. Energomashinostroyenie, 1961, No. 2, pp. 12-16.
. Soland, J. G., Mack, W., and Rohsenow, W. M. Evaluation of surface characteristics of heat
exchangers with plate fins. Heat transfer, 1978, Vol. 100, No. 3, pp. 150-156.
. Kern, J. Zur Bewertung von Kompakt-Warmeaustauschern. Warme-und Stoffubertragung,
1980, Vol. 13, pp. 205-215.
148. Kudrevaty, D. M. Method for comparing heat transfer surfaces. Energomashinostroyenie,
1972, No. 1, pp. 21-23.
149. Kovalenko, L. M. Comparison of thermo-energy effectiveness of convective surfaces of heat
transfer for various forms used in devices for the chemical industry. Khimicheskoye i neftye-
noye mashinostroyenie, 1979, No. 3, pp. 14-17.
194 REFERENCES

150. Baclic, B. S. A simplified formula for cross-flow heat exchanger effectiveness. Trans.
ASME, 1978, Vol. 100, No. 4, pp. 746-747.
ISS Ulinskas, R., Zakrevskis, V., and Zukauskas, A. The effect of one design factor on the
efficiency of in-line tube bundles in crossflow. Heat Transfer-Soviet Research, 1978, Vol. 10,
No. 6, pp. 102-111.
1523 Zakauskas, A., and Ulinskas, R. Developments in the compact heat exchanger field. Compact
Heat Exchangers, American Soc. Mech. Eng., HTD, Vol. 10, Chicago, Illinois, Nov. 16-21,
1980, pp. 29-36.
SY Kalafati, D. D., and Popalov, V. V. Comparative scale for effectiveness of heat transfer of
gaseous thermal carriers. Teplotekhnika, 1975, No. 9, pp. 67-69.
154. Kraus, A. D. Cooling of electronic equipment. Hemisphere Publishing Corporation, 1983.
1S) Kalafati, D. D., and Popalov, V. V. Optimization of tubular grid of in-line bank in cross-flow
on the basis of heat transfer effectiveness. Izv. vuzov SSSR. Energetika, 1977, No. 8, pp.
89-93.
156. Chechetkin, A. V., High-temperature thermal carriers. Moscow, Energiya, 1971. 496 pp.
157. Vargaftik, N. B., Filipov, L. P., Tarzimanov, A. A., and Yurchak, R. P. Thermal conductiv-
ity of gases and liquids. Moscow, Izd. Komiteta standartov, 1970, 156 pp.
158. Missenard, A. Conductivité thermique des solides, liquides, gaz et de leurs mélanges. Paris,
Eyrolles, 1965.
159. Andreiev, V. A. Heat-exchange devices for viscous liquids, Moscow-Leningrad. Gosener-
goizdat, 1961. 174 pp.
160. Andreiev, M. M., Berman, S. S., Buglaev, V. T., and Kostrov, V. T. Heat exchange devices
for energy installations. Moscow, Mashgiz, 1963. 240 pp.
161. Gavrilov, A. F, and Malkin, B. M. Fouling and cleaning of boiler heated surface. Moscow.
Energiya, 1980. 328 pp.
162. Kucherenko, D. I., Kalinin, A. M. Cleaning methods of heat exchange devices from deposits
and growth. Tr. in-ta VODGEO, 1977, pp. 44-46.
163. Taborek, I., Aoki, T., Ritter, R. B., Palen, I. W., and Knudsen, J. G. Predictive methods for
fouling behavior. Chem. Eng. Progr. 1972, Vol. 68, No. 7, pp. 69-78.
164. Chepurnoy, M. N., Schneider, V. E., Berkuta, A. D., and Klimenko, M. I. Calculation of
heat-exchange devices taking into account the kinetics of scale formation on the heated sur-
face. Teplophyzika i teplotekhnika, 1977, 33, pp. 83-87.
165. Burkov, V. V., Indeikin, A. I. Auto-tractor radiators. Leningrad, Mashinostroyenie, 1978.
216 pp.
166. Silvestrini, R. Heat exchanger fouling and corrosion. CEP, 1979, pp. 29-35.
167. Gosman, A. D., Pun, W. M., Runchal, A. K., Spalding, D. B., and Wolfshtein, M. Heat and
mass transfer in recirculating flows. Academic Press, London and New York, 1969.
168. Launder, B. E., and Massey, T. H. Numerical calculation of viscous flow characteristics and
heat transfer in banks of tubes. Teploperedacha. 1978, No. 4, pp. 1-8.
169. Launder, B. E. Heat and mass transfer. In book: Turbulence. Moscow, Mashinostroyenie,
1980, pp. 235-290.
170. Launder, B. E., Reece, G. I., and Rodi, W. Progress in the development of a Reynolds-stress
turbulence closure. J. Fluid Mech., 1975, Vol. 68, pp. 537-566.
7A Hanjalic, K., and Launder, B. E. A Reynolds-stress model of turbulence and its application to
thin shear layers. Ibid. 1972, Vol. 52, pp. 609-638.
V2: Launder, B. E., and Spalding, D. B. Mathematical models of turbulence. New York, Aca-
demic Press, 1972. 169 pp.
Wis: Durst, F., and Rastogi, A. K. Theoretical and experimental investigations of turbulent flows
with separation. Turbulent shear flows, 1977, University Park, Pennsylvania, USA, Sym-
poisum, 18-20 April, 1977, Vol. 1, pp. 18.1-18.9
174. Kutateladze, S. S., and Leontiev, A. I. Heat and mass transfer and friction in turbulent
boundary layer. Moscow, Energiya, 1972, 344 pp.
 REFERENCES 195

175. Kutateladze, S. S. Fundamentals of theory of heat transfer. Novosibirsk, Nauka, 1970, 660
Pp.
176. Zukauskas, A., Ulinskas. R. Investigation of parameters of a turbulent boundary layer in an
adverse pressure gradient. Heat Transfer-Soviet Research. 1981, Vol. 13, No. 5, pp. 25-35.
Tits Ziesina-Molozhen, L. M., Ziesin, L. V., and Polyak, M. P. Heat transfer in turbo-machinery.
Leningrad, Mashinostroyenie, 1974. 336 pp.
178. Reynolds, A. J. Turbulent flow in engineering applications. Moscow, Energiya, 1979, 408
PP.
Wf). Kline, S. I., Morkvin, M. V., Sovran, G., and Cockrell, D. I. Computation of turbulent
boundary layers. 1968. In: AFOSR-IFP—Stanford Conf., Stanford Univ., California, Aug.
18-25, 1968, Vol. 1, p. 590.
180. Patankar, S. Numerical heat transfer and fluid flow. Hemisphere Publishing Corporation,
New York, 1980.
i toere ob dior, add he

borat

, i AP? had go> Mh

# > Lemeewll (* optical Aas
| all
 INDEX

Air, 181 Fins, effectiveness of, 125

Aluminum, 126 Flow laminarization, 63
Auxiliary nomograms, 10 Flow velocity, 1
Aviation oil, 182 Forced convection, 72
Fouling, 12
Fourier equation, 14
Banks of tubes, 19
Bessel equation, 123
Biot number, 122
Boundary layer, 60, 134 Gas velocity, 57
as a function of Reynolds number, 46 Grashof:
separation, 48 criterion, 29
spoilers, 94 numbers, 27
thickness, 13, 101 Heat exchange, 1
Boussinesq approximation, 28 apparatus, 12
Heat exchangers, 67, 176
design and calculation, 173
Clapeyron-Medeleev equation, 28 effectiveness of, 157
Copper, 126 gas-gas, |
Crossflow, 13, 17, 83, 133 liquid-gas, 1
liquid-liquid, 1
plate, 1
Fin geometry, 142 Snellemiens)
Fin thickness, 142 spiral, 1
Finned: tube, 1, 3
aluminum tubes, 185-186 vapor-liquid, 1
biometallictubes, 139 Heat transfer, 9, 13-14, 17, 21, 39-40, 59,
copper tubes, 141 79
surfaces, 2, 94 calculation of, 176
tubes, 3, 120, 130, 137, 154 coefficient, 2
banks of, 134, 138 curves, 71
heat conduction of, 121 from a tube in crossflow, 13
typical shapes of, 121 from banks of smooth tubes, 25, 56

197
198 INDEX

Heat transfer (continued) parallel flow, 6

influence of flow regime, 63 thermal design principles of, 4
influence of tube arrangement on, 63 Reynolds number, 4, 13, 17, 21-23, 25, 27,
in high range of Re, 42 32, 34-35, 38, 42-43, 47, 58, 61,
in low range of Reynolds number, 27 72, 82-83, 84, 89-90, 95, 98, 100-
in medium range of Re, 38 101, 136-137, 142, 146, 149, 153,
of banks of rough tubes, 94 1652 1675 1745 177579
of rough surfaces, 95 Rough tubes, 3, 94, 99, 101, 113
of staggered tube banks, 61 Round fins, 123, 128
of tube banks, 89, 91
process, 7, 99
specific features of, 25-26, 56 “Shadow effect,’’ 34, 141
through a finned tube, 131 Shell-and-tube exchangers, 1, 3
Helical fins, 123, 128
types of, 2
Helical tubes, 1
Single channel bank, diagram of, 151
Hot thermal carrier, 9
Smooth tube(s), 3, 25
Hydraulic resistance, 73
banks of, 113
Hydrodynamic boundary layer, 13-14
Staggered banks, 50, 68-71, 80-81, 113
heat transfer of, 76
of finned tubes, 140
Inertial force, 66 of smooth tubes, 182
In-line banks, 50, 68-71, 80-81 Stagnation zones, 103
heat transfer of, 78 Stanton number(s), 107, 110
Intertubular spacing, 74 Strouhal number, 43
Surface friction, 98

Laminar boundary layer, 13, 32, 42, 72, 95,

101, 144
Temperature:
Laminar flow regime, 60, 95
fluctuations in a rough tube, 95
Lift force, 66
gradient, 14
Liquid flow, velocity of, 13
profile, 17
Liquid heating, 29
pulsations on tube surfaces, 53
Thermal:
boundary layer, 14, 17, 105, 136
Moderate temperature drops, 28
carrier(s), 94, 164
Muromtsev number, 164
efficiency, 3
resistance, 11, 13-14
roughness function, 112
Natural convections, 29, 64, 66, 72
Thermophysical properties of fluid, 181
Newton cooling law, 14
Transformer oil, 181
Nonsymmetrical staggered tube banks, 73 Trapezoidal:
Nusselt number(s), 152-153, 174 fins, 128
profile, 124
protuberances, 113
Plate and spiral heat exchangers, 2 Tube(s):
Prandtl numbers, 4, 51, 25, 29, 37, 96, 98, banks, 3, 12, 19, 29, 84, 164, 182
105, 107, 143, 159, 167, 174, 179 diameter, 29
finned, 3
helical, 1
Rayleigh number, 29, 64 rough, 3, 94
Recuperators: rows, 89
counterflow, 6 smooth, 3
 INDEX 199

Tube(s): (continued) Velocity of liquid flow, 13

U-shaped, 1 Vortex:
wall temperature, 27 generators, 19
Tubular heat exchangers, 1 zone, 32
Turbulence, 77, 89, 100-101
Turbulent boundary layer, 15, 17, 42, 95,
111
Water, 181

U-shaped tubes, 1
=p,
oA, & |) a oinlas
. ,
(Continued from front flap)

The Experimental and Applied Heat Transfer

Guide Books series, under the editorship of
Professor A. Zukauskas, offers a selection
of heat transfer titles—here available in
English—from Thermophysics, a series of
books published by the Institute of Physical
and Technical Problems of Energetics of the
Lithuanian Academy of Sciences.

A. ZUKAUSKAS is Vice President of the

Academy of Sciences of the Lithuanian SSR,
a position he has held since 1966. He gradu-
ated from Kaunas University, and became a
full Professor in 1967.

R. ULINSKAS is a Sector Chief at the Insti-

tute of Physical and Technical Problems of
Energetics at the Academy of Sciences of
the Lithuanian SSR. He is a graduate of the
Kaunas Polytechnic Institute.
 Of Related Interest

FLOW-INDUCED VIBRATION OF CIRCULAR CYLINDRICAL STRUCTURES

Shoei-Sheng Chen, Argonne National Laboratory, Argonne, Illinois
This volume presents a comprehensive treatment of the vibration of circular cylin-
ders. Dr. Chen examines this industrially important subject from the perspectives of
quiescent fluid, axial flow, and crossflow—then offers key applications of the ana-
lytical methods and experimental data useful for design evaluation. Over 200 illus-
trations, tables, and graphs complement the textual material.
1987, 464 pages, ISBN 0-89116-602-5

HEAT TRANSFER OF FINNED TUBE BUNDLES IN CROSSFLOW

J. Stasiulevicius and A. Skrinska, both of the /nstitute of Physical and
Technical Problems of Energetics, Kaunas, Lithuanian SSR
Edited by A. Zukauskas, Academy of Sciences of the Lithuanian SSR, Vil-
nius
In this work, findings on heat transfer and hydraulic drag in bundles of finned tubes
in crossflow are correlated for Reynolds numbers from 10* to 10°. These studies
illustrate fin, local and mean heat transfer coefficients, effects of geometric param-
eters of the fins, effect of tube location within the bundle on heat transfer and
hydraulic drag, and resistance of finned tube bundles.
1988, 224 pages, ISBN 0-89116-360-3
(Experimental and Applied Heat Transfer Guide Books)

HANDBOOK OF HYDRAULIC RESISTANCE

Second Edition, Revised and Augmented
I. E. Idelchik, State Scientific Research Institute for Industrial and Sanitary
Gas Purification, Moscow, USSR
English edition editor, Erwin Freid
“There is no other comparable book in English. This new edition contains a wealth
of data on pressure drops in piping networks and hydraulic machinery in the form
of tables and graphs. . . . most of the data are also given by correlating equations
for easy use in computer programs.’’— APPLIED MECHANICS REVIEWS
1986, 640 pages, ISBN 0-89116-284-4

For additional information and ordering, write to:

© HEMISPHERE PUBLISHING CORPORATION
79 Madison Avenue, New York, N.Y. 10016

AUDI
AVA
Hemisphere ISBN 0-89116-685- oo dl <7
P7-
Springer-Verlag ISBN 3-540-18865-7°