JOURNAL OF THERMOPHYSICS AND HEAT TRANSFER

Vol. 20, No. 4, October–December 2006

Analytical Model for Convection Heat Transfer From Tube Banks

W. A. Khan,∗ J. R. Culham,† and M. M. Yovanovich†

University of Waterloo, Waterloo, Ontario N2L 3G1, Canada
DOI: 10.2514/1.15453
The main objective of this study is to investigate heat transfer from tube banks in crossflow under isothermal
boundary conditions. Because of the complex nature of fluid flow and heat transfer in a tube bank, the heat transfer
from a tube in the first row of an in-line or staggered bank is determined first. For this purpose, a control volume is
selected from the leading row of a tube bank and an integral method of boundary layer analysis is employed to
determine the average heat transfer from the front stagnation point to the separation point, whereas the heat transfer
from the separation point to the rear stagnation point is determined by an empirical correlation. To include the effect
of the remaining rows, an empirical correlation is employed. The models for in-line and staggered arrangements are
applicable for use over a wide range of Reynolds and Prandtl numbers as well as longitudinal and transverse pitch
ratios.

Nomenclature  = distance normal to and measured from surface of tube,

a = dimensionless longitudinal pitch
SL =D m
b = dimensionless transverse pitch
ST =D  = pressure gradient parameter
CV = control volume  = angle measured from stagnation point, radians
c = dimensionless diagonal pitch
SD =D  = ratio of thermal to hydrodynamic boundary layer
cp = specific heat of fluid, J=kg  K thickness
D = tube diameter, m
Fa = arrangement factor Subscripts
h = average heat transfer coefficient, W=m2  K
a = ambient
k = thermal conductivity, W=m  K
f = fluid
L = tube length, m
o = outlet
N = total number of tubes in bank
NT NL
p = pressure
NL = number of tubes in longitudinal direction
T = thermal
NT = number of tubes in transverse direction
w = wall
NuD = Nusselt number based on tube diameter
Dh=kf
Pr = Prandtl number

=
Q = total heat transfer rate, W I. Introduction
ReD = Reynolds number based on tube diameter
DUmax =
EAT transfer in flow across a bank of tubes is of particular
SD
SL
=
=
diagonal pitch, m
longitudinal distance between two consecutive tubes,
H importance in the design of heat exchangers. Heat exchangers
are found in numerous industrial applications, such as steam gen-
m eration in a boiler or air cooling in the coil of an air conditioner. Tube
ST = transverse distance between two consecutive tubes, m banks, used in heat exchangers, are usually arranged in an in-line or
s = distance along curved surface of tube measured from staggered manner and are characterized by the dimensionless trans-
forward stagnation point, m verse, longitudinal, and diagonal pitches, shown in Figs. 1 and 2.
T = temperature,  C This study is one of the first attempts to develop analytical models
Uapp = approach velocity, m=s for the heat transfer from tube banks (in-line and staggered). These
Umax = maximum velocity in minimum flow area, m=s models are developed in terms of longitudinal and transverse pitch
Us = velocity in inviscid region just outside boundary layer ratios and Reynolds and Prandtl numbers. Depending upon the ap-
m=s plication, they are classified as compact or widely spaced tube banks.
u = s component of velocity in boundary layer, m=s Typically, one fluid moves over the tubes, while the other fluid, at a
v =  component of velocity in boundary layer, m=s different temperature and pressure, passes through the tubes. In this
 = thermal diffusivity, m2 =s study, the authors are specifically interested in the convection heat

Tlm = log mean temperature difference,  C transfer associated with crossflow over the tubes.
T = thermal boundary layer thickness, m Based on the pertinent data available up to 1933, Colburn [1]
 = hydrodynamic boundary layer thickness, m proposed a simple correlation for heat transfer for flow across banks
of staggered tubes as follows:
Presented as Paper 958 at the 43rd AIAA Aerospace Meeting and Exhibit,
Reno, Nevada, 5–8 January 2005; received 7 January 2005; revision received Nu D  0:33Re0:6
D Pr
1=3
(1)
23 March 2006; accepted for publication 26 March 2006. Copyright © 2006
by the American Institute of Aeronautics and Astronautics, Inc. All rights This correlation works well for 10 or more rows of tubes in a
reserved. Copies of this paper may be made for personal or internal use, on staggered arrangement and for 10 < ReD < 40; 000. Then Huge [2],
condition that the copier pay the $10.00 per-copy fee to the Copyright Pierson [3], Omohundro et al. [4], Bergelin et al. [5–7], Jones and
Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923; include Monroe [8], Gram et al. [9], Žukauskas [10], Aiba et al. [11,12], and
the code $10.00 in correspondence with the CCC.
∗ Žukauskas and Ulinskas [13] reported extensive experimental data
Microelectronics Heat Transfer Laboratory, Department of Mechanical
Engineering; currently Associate Professor, Department of Mathematics,
for heat transfer and fluid friction during viscous flow across in-line
COMSATS Information Technology Center, University Road, Abbottabad and staggered banks of tubes under both isothermal and isoflux
22060, NWFP Pakistan. boundary conditions. Grimison [14] correlated the experimental data
† of Huge [2] and Pierson [3] for both arrangements and gave a
Microelectronics Heat Transfer Laboratory, Department of Mechanical
Engineering. correlation of the form
720
 KHAN ET AL. 721

D SL Nu D  0:35Fa Re0:57
D Pr
0:31
(7)

Uapp, T a can be employed with

Fa  1  0:1a  0:34=b (8)

Žukauskas [10] gave the following experimental correlation for the

average Nusselt number for a tube bank consisting of 16 or more
rows:
Nu D  FCRenD Prm (9)
where the coefficients C, m, n and the parameter F are given in
tabular form (Kreith and Bohn [17]). Launder and Massey [19], Fujii
and Fujii [20], Dhaubhadel et al. [21], Wung and Chen [22], and
ST Murray [23] presented numerical solutions of local heat transfer for
the tube bank problem for a wide range of longitudinal and transverse
pitches: Reynolds and Prandtl numbers.
Fig. 1 Schematic of an in-line arrangement. Beale [24] conducted a detailed numerical study of fluid flow and
heat transfer in tube banks. Using complex function theory, he
obtained a potential flow solution in the form of a power series. He
SL presented his results in the form of skin friction, pressure drop, and
D SD heat transfer for different thermal boundary conditions. Later Beale
and Spalding [25,26] extended the previous work for laminar fully
Uapp, T a developed crossflow and heat transfer in tube-bank heat exchangers.
They obtained a wide range of results for in-line square, rotated
square, and equilateral triangle configurations.
Wilson and Bassiouny [27] developed a mathematical model to
simulate the laminar and turbulent flow fields inside tube banks.
They solved the conservation equations of mass, momentum, and
energy using an implicit finite volume procedure. They found that
pressure drop and friction factor increased with the longitudinal
pitch. They recommended the use of a longitudinal pitch ratio
S L 3 to obtain the best performance and to achieve a high degree
ST of compactness in an in-line arrangement, whereas S L 1:5 was
needed to reduce friction and enhance NuD in the staggered
arrangement.
Mandhani et al. [28] solved the fluid flow and energy equations
numerically to obtain detailed temperature fields and the distribution
Fig. 2 Schematic of a staggered arrangement.
of Nusselt number on the surface of a typical cylinder in a cylinder
bundle for the steady incompressible flow of Newtonian fluids. They
Nu D  C RenD (2) found that the surface averaged value of Nusselt number increases
with decreasing values of porosity and increasing values of Prandtl
His empirical relation is valid for air only and pertains to tube banks and Reynolds numbers. Their results were found in satisfactory
having ten or more rows of tubes in the flow direction. For fewer agreement with previous numerical and experimental data for a
rows, Kays and London [15] gave a correction factor C2 such that single cylinder and for the tube banks.
The preceding literature review shows that almost all studies are
Nu D jNL <10  C2 NuD jNL 10 (3) experimental/numerical and no comprehensive analytical model
exists for any in-line or staggered arrangement that can be used for a
The values of C, C2 , and n are given, in tabular form, in most heat wide range of parameters. The empirical models were developed for
transfer text books (e.g., Holman [16] and Kreith and Bohn [17]) for specific fluids, values of longitudinal and transverse pitch ratios, and
both in-line and staggered arrangements. Grimison [14] also for a specific range of Reynolds numbers. Because of the complex
correlated the test measurements of Pierson [3] and Huge [2] using a nature of heat transfer in tube banks, the user cannot interpolate/
second method and derived the following expression: extrapolate these correlations for other values of longitudinal and
transverse pitch ratios, or for other fluids or for Reynolds numbers
Nu D  0:32Fa ReD
0:61
Pr0:31 (4) other than the specified range, often found in existing tube banks.
New comprehensive models, developed in this study, can be used for
where Fa was presented graphically by Grimison [14] for the vari- the following:
ous values of ReD , subject to the dimensionless transverse and
longitudinal pitches. 1:25 a; b 3:0; 103 ReD 2
105 ;
Hausen [18] modified slightly the above correlation and presented
1 Pr 1000
an empirical formula for the tubes Fa instead of the graphical
representation by Grimison [14]. For an in-line arrangement

Nu D  0:34Fa ReD
0:61
Pr0:31 (5) II. Analysis
Consider a uniform flow of an incompressible Newtonian fluid
can be used with
past a tube bank as shown in Figs. 1 and 2. The ambient temperature
Fa  1  a  7:17=a 6:52 is assumed to be Ta . The surface temperature of the tube wall is
p





















Tw >Ta . The flow is assumed to be laminar, steady, and two dimen-

0:266=b 0:82 0:12 1000=ReD (6) sional. In determining heat transfer from tube banks, the ambient
temperature of the incoming fluid is taken as the determining tem-
For a staggered arrangement perature whereas the diameter of the tube is taken as the characteristic
722 KHAN ET AL.

length in the definition of Reynolds and Nusselt numbers. These from the inner tube rows starts to increase as a result of higher
numbers are defined as ReD  DUmax =
and NuD  Dh=kf , where turbulence, which is generated by the first tube rows. When the
Umax is used as a reference velocity in the calculations of fluid flow number of rows in a bank along the streamwise direction is large, the
and heat transfer for both types of arrangements, and is given by higher heat transfer rate of the remaining rows must be taken into
account. So the average heat transfer from the whole bank can be
Umax  maxb=b 1Uapp ; b=c 1Uapp  (10) written as

p
























Nu D  C1 NuDf (15)

where c  a2  b=22 is the dimensionless diagonal pitch in the
case of the staggered arrangement. where NuDf is the dimensionless average heat transfer from a tube in
the first row of a bank of smooth tubes in crossflow and C1 is the
A. Heat Transfer from a Tube Bank coefficient that accounts for the dependence of the average heat
transfer on the number of rows of a tube bank. The coefficient C1 is
The total heat transfer rate from the tube bank depends upon the derived from the experimental data of Žukauskas and Ulinskas [13]
average heat transfer coefficient, the total surface area for heat for ReD > 103 for both arrangements and is given by
transfer, and the inlet and outlet fluid temperatures, and is given by

1:23  1:47NL1:25 =1:72  NL1:25  in-line
Q  hNDL
Tlm (11) C1  (16)
1:21  1:64NL1:44 =1:87  NL1:44  staggered

where the log mean temperature difference is given by For NL  16, the value of C1 is 1.43 for the in-line arrangement and
1.61 for the staggered arrangement.
Tw Ta  Tw To 

Tlm  (12)
ln Tw Ta =Tw To  B. Heat Transfer from a Single Tube in a Tube Bank
The average heat transfer coefficient of a single tube taken from
where To is the outlet temperature of the fluid and can be obtained the first row of an in-line or staggered tube bank can be determined by
from using energy balance: an integral method of boundary layer analysis. In this study, the von
  Kármán–Pohlhausen integral method is used to solve the momentum
DNh and energy equations for an isothermal boundary condition. A
To  Tw Tw Ta   exp (13)
Uapp NT ST cp fourth-order velocity profile in the hydrodynamic boundary layer
and a third-order temperature profile in the thermal boundary layer
The only unknown quantity in Eq. (11) is the average convection heat are used. For this purpose, a control volume is considered from the
transfer coefficient for the tube bank that depends on the geometry, first row of an in-line or staggered arrangement as shown in Figs. 1
the maximum velocity (Umax ) in the minimum free cross section and 2. The width of the control volume is taken as unity for
between two tubes, and the arrangement of tubes, as well as the convenience, and the length and height, in dimensionless form, are
physical properties of the fluid ( ;
; cp ; and kf ). The functional taken as a and b=2 (
ST =2D), respectively. Because the flow is
relationship for the average dimensionless heat transfer can be symmetrical about the horizontal centerline, the solution has been
written as obtained for half of the flow domain (i.e., for ABCEFG in Fig. 3). The
top and bottom surfaces of the control volume can be regarded as
Nu D  fReD ; Pr; b; a (14) impermeable, adiabatic, and shear free (no mass transfer and shear
work transfer across the boundary). The heat transfer between the
Extensive experimental investigations (Žukauskas and Ulinskas tube and stream is Q and the wall temperature is Tw . The governing
[13]) indicate that the average heat transfer from a tube in a tube bank equations, velocity and temperature distributions for the CV, are the
is also dependent on its location in the bank. The difference in heat same as described by Khan et al. [29] for a single isolated cylinder.
transfer from tubes of the first and inner rows depends on the The potential flow velocity outside the boundary layer was obtained
Reynolds number, the number of tubes, the longitudinal and by using complex variable theory and following Suh et al. [30] it can
transverse pitch ratios, and the arrangement of the tubes in a bank. be written as (see Appendix)
The increase in heat transfer is also observed due to flow blockage by
the upstream tubes. For ReD > 103 , they observed that heat transfer U  Umax f (17)

where
    
 cosh=a sin  sin   sinh=a sin  sin   cos  sin=a cos 
f  sin  2sin2  sinh sin  (18)
2a cosh=a sin  cos=a cos  a cosh
=a sin  cos
=acos 2

for an in-line arrangement and

   
 cosh sin =2a sin   sin  sinh sin =2a sin   sin cos =2a cos 
f  sin  2sin2 sinh
4a cosh sin =2a cos cos =2a 2a cosh
 sin =2a cos
 cos =2a2
cosh
sin  2b=2a sin 

cosh
sin  2b=2a cos
cos  2a=2a
  
sin  2b sinh
sin  2b=2a sin   sin
cos  2a=2a cos 
sinh  (19)
2a cosh
sin  2b=2a cos
cos  2a=2a2

for the staggered arrangement.

The following boundary conditions are specified for the control volume of Fig. 3.
1) On the curved surfaces of the tube, u  0, v  0, and T  Tw .
2) Along the top and bottom of the control volume and on the side-wall regions between tubes, v  0, w  0, and Q  0.
 KHAN ET AL. 723

τw = 0, Q = 0 The first term on the right-hand side gives the dimensionless average
G F heat transfer coefficient of the tube from the front stagnation point to
Umax Uapp , T a the separation point, and can be obtained, using Eqs. (23–25), for
η different pitch ratios and then correlated to obtain a single expression
s τ w sin θ P cos θ
τw θ
in terms of the ReD and Pr numbers for both in-line and staggered
S T /2
P
arrangements. This expression can be written as
Hydrodynamic
Tw D/2 boundary layer Nu Df1  C2 ReD
1=2
Pr1=3 (29)
θ
A
τw = 0, Q = 0
B 0 C
τw = 0, Q = 0
E where C2 is a constant that depends upon the longitudinal and
SL transverse pitches, arrangement of the tubes, and thermal boundary
conditions. For the isothermal boundary condition, it is given by:
Fig. 3 Control volume for prediction of heat transfer from a tube bank.
8
<  :016  :6a2 =0:4  a2  in-line
C2 
3) At large distances upstream of the CV, u  Uapp and T  Ta . : 0:588  0:004b0:858  0:04b 0:008b2 1=a staggered
Following Khan et al. [29] and assuming the presence of a thin
thermal boundary layer T along the tube surface in the CV, the (30)
energy integral equation for the isothermal boundary condition can
be written as Equation (30) is valid for 1:25 a 3 and 1:25 b 3 for both
Z  arrangements.
d T @T 
T Ta ud    (20) The second term on the right-hand side of Eq. (28) gives the
ds 0 @ 0 dimensionless average heat transfer coefficient of the tube from the
separation point to the rear stagnation point. The integral analysis is
Using a fourth-degree polynomial velocity profile and a third-degree unable to predict these heat transfer coefficients. The experiments
polynomial temperature profile that satisfy all the boundary (Žukauskas and Žiugžda [31], Fand and Keswani [32], and
conditions (Khan et al. [29]) and assuming   T = < 1 for Pr  1, Nakamura and Igarashi [33], among others) show that the heat
Eq. (20) can be integrated to give transfer from the rear portion of the cylinder increases with the
Reynolds numbers. From a collection of all known data, Van der
T d=dsUsT   12  90 (21) Hegge Zijnen [34] demonstrated that the heat transferred from the
rear portion of the cylinder to the air can be determined from
where Us is given by Eq. () and  is obtained from the momentum
integral equation and the definition of momentum boundary layer Nu Df2  0:001ReD (31)
thickness. The values of  are obtained cor-
responding to each position along the tube surface and are fitted by Thus, the total heat transfer coefficient from a single tube in the first
the least squares method and given by row can be written as
  7:36 3:74  27:952 96:643  157:834
Nu Df  C2 ReD
1=2
Pr1=3  0:001ReD (32)
135:87  58:65 10:10
5 6 7
(22)

Integrating Eq. (21) with respect to s, one can obtain local thermal
boundary layer thicknesses III. Results and Discussion
v

















































s











According to Žukauskas and Ulinskas [13], tube banks with
  u
T  u  b
a 1:25
1:25 are considered compact, and with b
a 
 ReD Pr 
1=2 1=3 t
3 90I1
(23) 2
2 they are said to be widely spaced. For both compact and wide
D   122 f2 2g tube banks, Incropera and DeWitt [35] solved a problem of a
staggered tube bank that is used for space heating. In this study, that
where g is the derivative of the function f with respect to  and problem is chosen for comparing the results of the present analysis.
I1 is given by Incropera and DeWitt [35] assumed steady state conditions,
Z negligible radiation effects, and negligible effect of change in fluid
I1  f  12d (24) temperature on fluid properties. They used the data given in Table 1
0 to calculate an air-side convection coefficient and heat transferred by
the tube bank.
The local heat transfer coefficient can be written as
h  3kf =2T (25) Table 1 Data used by Incropera and Dewitt [35] for a staggered tube
bank
Thus the dimensionless local heat transfer coefficient can be written
as Quantity Dimension
s
















































r












Tube diameter, mm 16.4
NuDf jisothermal 3 3   122 f2 2g Longitudinal pitch, mm 20.5, 34.3
 (26) Transverse pitch, mm 20.5, 31.3
1=2 1=3
ReD Pr 2 90I1  Number of tubes, staggered 8
7
Tube surface temperature,  C 70
The average heat transfer coefficient is defined as Air properties:
Z Z Z Approach velocity, m=s 6
1  1 s 1  Thermal conductivity, W=m  K 0.0253
h hd  hd  hd (27)
 0  0  s Density, kg=m3 1.217
Specific heat, J=kg  K 1007
In dimensionless form, the heat transfer coefficient can be written as Kinematic viscosity, m2 =s 14:82
10 6
Prandtl number 0.701
Nu Df  hD=kf  NuDf1  NuDf2 (28) Ambient temperature,  C 15
724 KHAN ET AL.

Table 2 Comparison of results for a compact tube bank (1:25  1:25) DeWitt [35]. The reason for higher heat transfer rates in the present
case might be due to the fact that Incropera and DeWitt [35] used the
NuD h, W=m2  K To ,  C Q, kW
same constants in their correlation for both cases, whereas the present
Incropera and DeWitt [35] 152.0 234.0 38.5 28.4 models are sensitive to pitch ratios. The comparison of Tables 2 and 3
Present analysis 186.8 288.3 39.2 25.5 shows that heat transfer of a bank decreases with increasing pitch
ratio.
Average heat transfer from a single tube in the first row of
Table 3 Comparison of results for a wide tube bank (1:9  2:1) symmetrical in-line tube banks with 1:25
1:25 and 2:0
2:0 pitch
ratios is shown in Figs. 4 and 5. In both cases, the average heat
NuD h, W=m2  K To ,  C Q, kW transfer increases with the Reynolds numbers and the behavior
Incropera and DeWitt [35] 87.9 135.6 25.5 19.4 approximates a linear dependence on the logarithmic scale. The
Present analysis 113.15 175.15 27.5 20.1 comparison of both figures show that the heat transfer increases in
symmetrical in-line tube banks with their transverse and longitudinal
pitch ratios. Turbulence generated by the first rows penetrates the
Incropera and DeWitt [35] solved this problem by using boundary layer developed on the tube more effectively than in
Žukauskas and Ulinskas [13] correlations, whereas the present compact symmetrical in-line tube banks. In compact banks, tur-
analysis uses an analytical model. The results are shown in Table 2 bulence decays as a result of the flow being compressed between
for a compact bank and in Table 3 for a widely spaced bank. Table 2 longitudinal rows [13]. The present results are compared with the
shows that the present analysis gives higher heat transfer rate (around
18%) than Incropera and DeWitt [35], whereas Table 3 shows that
the present heat transfer rate is 22% higher than Incropera and
103
103 Analytical (present model)
Analytical (present model) ˇ
Experimental (Zukauskas and Ulinskas [13])
ˇ
Experimental (Zukauskas and Ulinskas [13])
Staggered
2.0 x 1.25
In-line First row
Nuf

2
1.25 x 1.25 10
Nuf

2
10 First row

101 3
10 104 105
101 3 ReD
10 104 105
ReD Fig. 6 Average heat transfer from the first row of a 2:0  1:25
staggered tube bank.
Fig. 4 Average heat transfer from the first row of a 1:25  1:25 in-line
tube bank.

103 103
Analytical (present model) Analytical (present model)
ˇ
Experimental (Zukauskas and Ulinskas [13]) ˇ
Experimental (Zukauskas and Ulinskas [13])

In-line 2.6 x 1.3

2 x2 Staggered
First row First Row
Nuf

Nuf

102 102

101 3 101 3
10 104 105 10 104 105
ReD ReD
Fig. 5 Average heat transfer from the first row of a 2:0  2:0 in-line Fig. 7 Average heat transfer from the first row of a 2:6  1:3 staggered
tube bank. tube bank.
 KHAN ET AL. 725

103 103
Analytical (present model)
Empirical (Grimison [14]) In-line
ˇ
Empirical (Zukauskas and Ulinskas [13]) Staggered

In-line 1.25 x 1.25

1.25 x 1.25

NuD
NuD

102 102

101 3 101 3
10 104 105 10 104 105
ReD ReD
Fig. 10 Comparison of average heat transfer from in-line and
Fig. 8 Average heat yransfer from an in-line tube bank.
staggered tube banks.

103 for the same pitch ratios. Good agreement is found between the
Analytical (present model)
analytical and experimental results.
ˇ
Experimental (Zukauskas and Ulinskas [13]) The comparison of the average heat transfer from both in-line and
staggered tube banks for the same transverse and longitudinal pitch
ratios is shown in Fig. 10. For the same pitch ratio, the heat transfer is
found to be higher in a staggered bank than in an in-line bank. This is
Pr = 0.71 due to the fact that in a staggered bank the path of the main flow is
more tortuous and a greater portion of the surface area of downstream
ST / D = 2 tubes remains in this path.
NuD

2
10 SL / D = 2
IV. Conclusion
Heat transfer from tube banks in crossflow is investigated
analytically, and simplified models of heat transfer for both arrange-
ments (in-line and staggered) are presented. The coefficient C1 in
Eq. (16) is derived from the experimental data of Žukauskas and
Ulinskas [13] and C2 in Eq. (29) is determined by fitting the analyti-
cal results obtained for various pitch ratios in both arrangements. The
101 3 results obtained from this investigation are as follows:
10 104 105 1) Both models can be applied over a wide range of parameters and
ReD are suitable for use in the design of tube banks.
Fig. 9 Average heat transfer from a staggered tube bank. 2) The average heat transfer coefficients for tube banks in
crossflow depend on the number of longitudinal rows, longitudinal
and transverse pitch ratios, and Reynolds and Prandtl numbers.
3) Compact banks (in-line or staggered) indicate higher heat
experimental data of Žukauskas and Ulinskas [13]. The agreement is transfer rates than widely spaced ones.
found to be good in both cases. 4) The staggered arrangement gives higher heat transfer rates than
Figures 6 and 7 show the average heat transfer from a single tube in the in-line arrangement.
the first row of asymmetrical staggered tube banks with 2:0
1:25
and 2:6
1:3 pitch ratios. They show significant variation in the heat
transfer at large Reynolds numbers for a large change in the Appendix
transverse pitch ratio and a small change in the longitudinal pitch I. Cylindrical Tubes in an In-Line Arrangement
ratio. These results are also found to be in good agreement with the Following Suh et al. [30], the complex potential for in-line arrays,
experimental data of Žukauskas and Ulinskas [13]. Similar results subjected to uniform flow, can be written as
were obtained by Bergelin et al. [6,7] for the flow of air.
Average heat transfer values for the entire bank can be determined X
1

wz  Uapp z 
from Eq. (15) depending upon the number and the type of j;k 1
2z kSL  ijST 
arrangement of tubes. For a compact in-line bank 1:25
1:25 with
NL  16, the average heat transfer values are plotted versus ReD in
X 1
1
 Uapp z  (33)
Fig. 8. On the logarithmic scale, heat transfer values increase linearly 2 j;k 1 z kSL  ijST 
with the Reynolds numbers. The present values are compared with
the empirical correlations of Grimison [14] and Žukauskas and where j and k are the number of rows and columns. Using
Ulinskas [13]. Both correlations are found to be in good agreement
with the analytical results. Figure 9 shows the heat transfer from a X1  
1  z
widely spaced staggered tube bank (2:0
2:0). The results are  coth (34)
compared with the experimental data of Žukauskas and Ulinskas [13] j 1
z ijS T ST ST
726 KHAN ET AL.

the complex potential for in-line bank can be written as components of velocity at the surface of the tube can be written as

X 1

ur  1=r@ =@jrR and u  @ =@rjrR (43)
wz  Uapp z  coth z kSL   Uapp z  Tz
2ST k 1 ST 2ST
(35) which gives

where ur  0 and u  Uapp f (44)

X
1

Tz  coth z kSL  (36) where

k 1
2ST
 
 cosh=a sin  sin 
Using complex variable theory, it can be shown that f  sin  2sin2
2a cosh=a sin  cos=a cos 
Tz  ST =2SL  cotz=SL  (37)   
 sinh=a sin  sin   cos  sin=a cos 
 sinh sin 
Therefore, the complex potential will be a cosh
=a sin  cos
=a cos 2
(45)
Wz  Uapp z  
=4SL  cotz=SL  (38)

which gives the complex velocity W 0 z as follows: The resultant potential flow velocity will be
 

1  U  Uapp f (46)
W 0 z  Uapp (39)
4SL sin2 =SL  SL

At the surface of the tube, W 0 R  0, )

=4SL 
Uapp SL =sin2 R=SL  Therefore the required potential flow II. Cylindrical Tubes in Staggered Arrangement
function for in-line tubes will be Following Suh et al. [30], the complex potential for in-line arrays,
Wz    i  Uapp fz  C cotz=SL g (40) subjected to uniform flow, can be written as
   
where  and are the potential and stream functions and C is a z
Wz    i  Uapp z  C cot
constant, given by 2SL
 
C  SL =sin2 R=SL  (41) z SL  iST 
 cot (47)
2SL
It is interesting to note here that the potential flow field has no
dependence on the transverse spacing ST for the infinite number of where  and are the potential and stream functions and C is a
rows. The stream function in polar coordinates r;  can be constant, given by
obtained from Eq. (40) as
 
sinhC1 r sin  C  2SL =sin2 R=2SL  (48)
 Uapp r sin  C (42)
coshC1 r sin  cosC1 r cos 
The stream function in polar coordinates r;  can be obtained
where C1  2=SL is a constant. The radial and transverse from Eq. (47) as

   
2S R sinh
r sin =SL 
 Uapp r sin  L sin2
 2SL cosh
r sin =SL  cos
r sin =SL 
   
2SL 2 R sinh
r sin  ST =SL 
sin (49)
 2SL cosh
r sin  ST =SL  cos
r cos  SL =SL 

The radial and transverse components of velocity at the surface of the tubes can be obtained by using Eq. (43) and can be written like Eq. (44),
where

   
 cosh sin =2a sin   sin  sinh sin =2a sin   sin cos 
 cos 
f  sin  2sin2 sinh 2a
4a cosh sin =2a cos cos =2a 2a cosh sin =2a cos cos =2a2
cosh
sin  2b=2a sin 

cosh
sin  2b=2a cos
cos  2a=2a
  
sin  2b sinh
sin  2b=2a sin   sin
cos  2a=2a cos 
sinh  (50)
2a cosh
sin  2b=2a cos
cos  2a=2a2

The resultant potential flow velocity will be:

 KHAN ET AL. 727

U  Uapp f (51) [15] Kays, W. M., and London, A. L., “Compact Heat Exchangers,”
McGraw–Hill, New York, 1964.
[16] Holman, J. P., Heat Transfer, 7th ed., McGraw–Hill, New York, 1992,
pp. 307–310.
[17] Kreith, F., and Bohn, M. S., “Principles of Heat Transfer,” 5th ed., West
Acknowledgments Publishing, New York, 1993, pp. 469–485.
The authors gratefully acknowledge the financial support of the [18] Hausen, H., “Heat Transfer in Counterflow, Parallel Flow and Cross
Natural Sciences and Engineering Research Council of Canada and Flow,” McGraw–Hill, New York, 1983.
the Centre for Microelectronics Assembly and Packaging. [19] Launder, B. E., and Massey, T. H., “The Numerical Prediction of
Viscous Flow and Heat Transfer in Tube Banks,” Journal of Heat
Transfer, Vol. 100, No. 4, 1978, pp. 565–571.
[20] Fujii, M., and Fujii, T., “A Numerical Analysis of Laminar Flow and
References Heat Transfer of Air in an In-Line Tube Bank,” Numerical Heat
[1] Colburn, A. P., “A Method of Correlating Forced Convection Heat Transfer, Vol. 7, 1984, pp. 89–102.
Transfer Data and a Comparison With Fluid Friction,” Transactions of [21] Dhaubhadel, M. N., Reddy, J. N., and Telionis, D. P., “Finite Element
the American Institute of Chemical Engineers, Vol. 29, American Analysis of Fluid Flow and Heat Transfer for Staggered Banks of
Institute of Electrical Engineers, New York, 1933, pp. 174–210. Cylinders in Cross Flow,” International Journal for Numerical
[2] Huge, E. C., “Experimental Investigation of Effects of Equipment Size Methods in Fluids, Vol. 7, 1987, pp. 1325–1342.
on Convection Heat Transfer and Flow Resistance in Cross Flow of [22] Wung, T. S., and Chen, C. J., “Finite Analytic Solution of Convective
Gases Over Tube Banks,” Journal of Dynamic Systems, Measurement Heat Transfer for Tube Arrays in Crossflow: Part 2–Heat Transfer
and Control, Series G, Vol. 59, 1937, pp. 573–581. Analysis,” Journal of Heat Transfer, Vol. 111, Aug. 1989, pp. 641–
[3] Pierson, O. L., “Experimental Investigation of the Influence of Tube 648.
Arrangement on Convection Heat Transfer and Flow Resistance in [23] Murray, D., “Comparison of Heat Transfer in Staggered and In-Line
Cross Flow of Gases Over Tube Banks,” Journal of Dynamic Systems, Tube Banks with Gas Particle Flow,” Experimental Thermal and Fluid
Measurement and Control, Series G, Vol. 59, 1937, pp. 563–572. Science, Vol. 6, No. 2, 1993, pp. 177–185.
[4] Omohundro, G. A., Bergelin, and Colburn, A. P., “Heat Transfer and [24] Beale, S. B., “Fluid Flow and Heat Transfer in Tube Banks,” Ph.D.
Fluid Friction During Flow Across Banks of Tubes,” Journal of Heat Dissertation, Imperial College, London, 1992.
Transfer, Vol. 71, 1949, pp. 27–34. [25] Beale, S. B., and Spalding, D. B., “Numerical Study of Fluid Flow and
[5] Bergelin, O. P., Davis, E. S., and Hull, H. L., “A Study of Three Tube Heat Transfer in Tube Banks with Stream-Wise Periodic Boundary
Arrangements in Unbaffled Tubular Heat Exchangers,” Journal of Heat Conditions,” Transactions of the CSME, Vol. 22, No. 4A, 1998,
Transfer, Vol. 71, 1949, pp. 369–374. pp. 397–416.
[6] Bergelin, O. P., Brown, G. A., Hull, H. L., and Sullivan, F. W., “Heat [26] Beale, S. B., and Spalding, D. B., “A Numerical Study of Unsteady
Transfer and Fluid Friction During Flow Across Banks of Tubes–3: A Fluid Flow in In-Line and Staggered Tube Banks,” Journal of Fluids
Study of Tube Spacing and Tube Size,” Journal of Heat Transfer, and Structures, Vol. 13, 1999, pp. 723–754.
Vol. 72, 1950, pp. 881–888. [27] Wilson, A. S., and Bassiouny, M. K., “Modeling of Heat Transfer for
[7] Bergelin, O. P., Brown, G. A., and Doberstein, S. C., “Heat Transfer and Flow Across Tube Banks,” Chemical Engineering and Processing,
Fluid Friction During Flow Across Banks of Tubes–4: A Study of the Vol. 39, 2000, pp. 1–14.
Transition Zone Between Viscous and Turbulent Flow,” Journal of [28] Mandhani, V. K., Chhaabra, R. P., and Eswaran, V., “Forced
Heat Transfer, Vol. 74, 1952, pp. 953–959. Convection Heat Transfer in Tube Banks in Cross Flow,” Chemical
[8] Jones, C. E., and Monroe, E. S., “Convection Heat Transfer and Engineering Science, Vol. 57, 2002, pp. 379–391.
Pressure Drop of Air Flowing Across In-Line Tube Banks: Part 1– [29] Khan, W. A., Culham, J. R., and Yovanovich, M. M., “Fluid Flow
Apparatus, Procedures, and Special Effects,” Journal of Heat Transfer, Around and Heat Transfer from an Infinite Circular Cylinder,” Journal
Vol. 80, 1958, pp. 18–24. of Heat Transfer, Vol. 127, July 2005, pp. 785–790.
[9] Gram, A. J., Mackey, C. O., and Monroe, E. S., “Convection Heat [30] Suh, Y. B., Somasundaram, S., and, Anand, N. K., “Remarks on the
Transfer and Pressure Drop of Air Flowing Across In-Line Tube Banks: Potential Cross Flow Over Tube Banks,” Journal of Applied
Part 2–Correlation of Data for Ten Row-Deep Tube Banks,” Journal of Mechanics, Vol. 56, 1989, pp. 476–479.
Dynamic Systems, Measurement and Control, Series G, Vol. 80, 1958, [31] Žukauskas, A., and Žiugžda, J., Heat Transfer of a Cylinder in
pp. 25–35. Crossflow, Hemisphere, New York, 1985.
[10] Žukauskas, A., “Heat Transfer from Tubes in Crossflow,” Advances in [32] Fand, R. M., and Keswani, K. K., “A Continuous Correlation Equation
Heat Transfer, Vol. 8, 1972, pp. 93–160. for Heat Transfer from Cylinders to Air in Crossflow for Reynolds
[11] Aiba, S., Tsuchida, H., and Ota, T., “Heat Transfer Around Tubes in In- Numbers From 10 2 to 2
105 ,” International Journal of Heat and
Line Tube Banks,” Bulletin of the JSME, Vol. 25, No. 204, 1982, Mass Transfer, Vol. 15, 1972, pp. 559–562.
pp. 919–926. [33] Nakamura, H., and Igarashi, T., “Variation of Nusselt Number with
[12] Aiba, S., Tsuchida, H., and Ota, T., “Heat Transfer Around Tubes in Flow Regimes Behind a Circular Cylinder for Reynolds Numbers from
Staggered Tube Banks,” Bulletin of the JSME, Vol. 25, No. 204, 1982, 70 30000,” International Journal of Heat and Mass Transfer, Vol. 47,
pp. 927–933. 2004, pp. 5169–5173.
[13] Žukauskas, A., and Ulinskas, R., “Heat Transfer in Tube Banks in [34] Van der Hegge Zijnen, B. G., “Modified Correlation Formulae for Heat
Crossflow,” Hemisphere, Washington, D.C., 1988. Transfer by Natural and Forced Convection from Horizontal
[14] Grimison, E. D., “Correlation and Utilization of New Data on Flow Cylinders,” Applied Scientific Research, Series A. Mechanics, Heat,
Resistance and Heat Transfer for Cross Flow of Gases Over Tube Vol. 6, No. 2–3, 1956, pp. 129–140.
Banks,” Journal of Dynamic Systems, Measurement and Control, [35] Incropera, F. P., and DeWitt, D. P., “Introduction to Heat Transfer,”
Series G, Vol. 59, 1937, pp. 583–594. Wiley, New York, 2002.