Professional Documents
Culture Documents
www.elsevier.com/locate/ijts
Abstract
This paper presents a three-dimensional numerical and experimental geometric optimization study to maximize the total heat transfer rate
between a bundle of finned tubes in a given volume and external flow, for staggered arrangements of circular and elliptic tubes. Experimental
results were obtained for circular and elliptic configurations with twelve tubes, starting with an equilateral triangle configuration, which fitted
uniformly into the fixed volume with a resulting optimal dimensionless tube-to-tube spacing of S/2b = 1.5, where S is the actual spacing and
b is the smaller semi-axis of the ellipse. Several experimental configurations were built by reducing the tube-to-tube spacings, identifying
the optimal spacing for maximum heat transfer. Similarly, it was possible to investigate the existence of optima with respect to two other
geometric degrees of freedom, such as tube eccentricity and fin-to-fin spacing. The results are reported for air as the external fluid in the
laminar regime, for ReL = 852 and 1065, where L is the swept length of the fixed volume. Circular and elliptic arrangements with the same
flow obstruction cross-sectional area were compared on the basis of maximizing the total heat transfer. This criterion allows one to isolate
and quantify the heat transfer gain, by studying arrangements with equivalent total pressure drops independently of the tube cross-section
shape. This paper continues with three-dimensional numerical optimization results for finned circular and elliptic tubes arrangements, which
are validated by direct comparison with experimental measurements. Global optima with respect to tube-to-tube spacing, eccentricity and
fin-to-fin spacing (S/2b
= 0.06 for ReL = 852 and 1065) were found and reported in dimensionless terms. A relative heat
= 0.5, e
= 0.5, f
transfer gain of up to 19% is observed in the optimal elliptic arrangement, as compared to the optimal circular one. The heat transfer gain,
combined with the relative material mass reduction of up to 32% observed in the optimal elliptic arrangement in comparison to the circular,
show that the elliptical tubes arrangement has better overall performance and lower cost than the traditional circular tubes geometry.
2003 Elsevier SAS. All rights reserved.
Keywords: Constructal design; Finned tubes; Elliptic tubes; Spacings; Maximal heat transfer density
1. Introduction
The optimization of industrial processes for maximum
utilization of the available energy (exergy) has been a very
active line of scientific research in recent times. The increase
in energy demand in all sectors of the human society requires an increasingly more intelligent use of available energy. Many industrial applications require the use of heat
exchangers with tubes arrangements, either finned or nonfinned, functioning as heat exchangers in air conditioning
systems, refrigeration, heaters, radiators, etc. Such devices
have to be designed according to the availability of space
* Corresponding author.
478
Nomenclature
a
Ac
b
Ba
cp
Greek symbols
thermal diffusivity . . . . . . . . . . . . . . . . . . . m2 s1
penalty factor
mesh convergence criterion, Eq. (25)
fin-to-fin spacing . . . . . . . . . . . . . . . . . . . . . . . . . m
dimensionless temperature
dimensionless average fluid temperature
fluid kinematic viscosity . . . . . . . . . . . . . m2 s1
density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . kgm3
dimensionless fin density in direction z
Subscripts
m
maximum
opt
optimal
out
unit cell outlet
s
solid tube wall and fin material
w
tube surface
1, 2, 3 directions x, y and z
free stream
Bordalo and Saboya [3] reported pressure drop measurements comparing elliptic and circular tube and plate fin
heat exchanger configurations, with one, two and three-row
arrangements. Reductions of up to 30% of the loss coefficient (pressure drop coefficient per unit row due only to the
presence of the tubes) were observed, in favor of the elliptic configuration. The comparison was performed between
circular and elliptic arrangements with the same flow obstruction cross-sectional area, for 200 Re 2000 (1.8
u 18.2 ms1 , with = 1.65 mm), which cover the air
2. Theory
A typical four-row tube and plate fin heat exchanger
with a general staggered configuration is shown in Fig. 1.
Fowler and Bejan [12] showed that in the laminar regime,
the flow through a large bank of cylinders can be simulated
accurately by calculating the flow through a single channel,
such as that illustrated by the unit cell seen in Fig. 1. Because
of the geometric symmetries, there is no fluid exchange or
heat transfer between adjacent channels, or at the top and
side surfaces. At the bottom of each unit cell, no heat transfer
is expected across the plate fin midplane. In Fig. 1, L, H and
W are the length, height and width (tube length) of the array,
respectively. The fins are identical, where t is the thickness
and , is the fin-to-fin spacing.
479
The governing equations for mass, momentum and energy conservation are simplified in accordance with the assumptions of three-dimensional incompressible steady-state
laminar flow with constant properties for a Newtonian fluid,
e.g., [13]
U1 U2 U3
+
+
=0
(1)
X
Y
Z
U1
U1
U1
U1
+ U2
+ U3
X
Y
Z
1 2 U1 2 U1 2 U1
P
+
=
+
+
(2)
X ReL X2
Y 2
Z 2
U2
U2
U2
+ U2
+ U3
U1
X
Y
Z
1 2 U2 2 U2 2 U2
P
+
=
+
+
(3)
Y
ReL X2
Y 2
Z 2
U3
U3
U3
+ U2
+ U3
U1
X
Y
Z
1 2 U3 2 U3 2 U3
P
+
=
+
+
(4)
Z ReL X2
Y 2
Z 2
+ U2
+ U3
U1
X
Y
Z
1 2
2
2
=
+
+
(5)
PeL X2 Y 2 Z 2
The dimensionless variables have been defined based on
appropriate physical scales,
(X, Y, Z) =
(x, y, z)
,
L
P=
p
u2
(6)
480
(u1 , u2 , u3 )
T T
,
=
u
Tw T
(7)
U L
U L
,
PeL =
ReL =
=
(8)
+
+
ReL X2 Y 2 Z 2
(U1 , U2 , U3 ) =
U1 = 1,
=0
(9)
U2 U1
(B) and (C) U3 = 0,
=
=
=0
(10)
Z
Z
Z
U1 U3
(D) and (E) U2 = 0,
=
=
=0
(11)
Y
Y
Y
U1 U2 U3
(F)
=
=
=
=0
(12)
X
X
X
X
(G) U1 = U2 = U3 = 0,
=1
(13)
(H) U1 = U2 = U3 = 0,
=0
(14)
Z
Eq. (13) states that the tube walls are modeled as isothermal,
Tw = constant, this in order to account for the presence of
a well mixed fluid inside the tubes. In order to represent
the actual flow with boundary conditions (A) and (F), two
extensions need to be added to the computational domain,
upstream and downstream, as shown in Fig. 2. The actual
dimensions of these extralengths need to be determined by
an iterative numerical procedure, with convergence obtained
according to a specified tolerance.
The objective is to find the optimal geometry, such that
the volumetric heat transfer density is maximized, subject
to a volume constraint. The optimization study proceeds
Q/(Tw T )
kLHW/(2b)2
(15)
where the overall heat transfer rate between the finned tubes
and the free stream Q is divided by the constrained volume
LHW, k is the fluid thermal conductivity, and 2b = D is the
smaller axis of the ellipse or tube diameter. The first law for
one elemental channel requires
ec cp (Tout T )
Q = Nec Qec = Nec m
(16)
where Nec is the number of elemental channels. The elemental channel is defined as the sum of all unit cells in the
direction z. Therefore, m
ec = u [(S + 2b)/2](W nf tf )
is the mass flow rate entering one elemental channel, cp is
the fluid specific heat at constant pressure, and Tout is the average fluid temperature at the elemental channel outlet. The
number of fins in the arrangement is given by
nf =
W
tf +
(17)
481
W
m
=
nt ab(a tt )(b tt )
s L3 L3
+ f (LH nt ab)
(20)
3. Numerical method
The numerical solution to Eqs. (1)(14) was obtained by
using the finite element method [10], giving the velocities
and temperature fields in the unit cell of Fig. 2. The first
step in the numerical implementation was the elimination of
the pressure variable from Eqs. (2)(4), by using a penalty
model, approximating Eq. (1) as [15],
P
U1 U2 U3
+
+
=
X
Y
Z
(21)
C(U )
0
0
C(U )
0
0
U1
U 2
U 3
0
0
C(U )
U1
U 2
U 3
K12
K11 + 2K22 + K33
K32
K11 K 12 K 13
U1
K 21
K 31
U )T + L T = G
D(
K 22
K 32
K 23
K 33
U 2
U 3
K13
K23
K11 + K22 + 2K33
F1
F2
F3
(22)
(23)
482
where C(U ) is the capacity matrix that contains the advective terms of the momentum equations, which depends
nonlinearly on the solution U (tripartitioned vector: U 1 direction X, U 2 -direction Y and U 3 -direction Z, each one
with a number of components equal to the number of unknowns in the mesh); K11 , K12 , K13 , K21 , K22 , K23 , K31 ,
K32 and K33 are the stiffness matrices with constant coefficients (constant viscosity), that contain the viscous terms
of the momentum equations; K 11 , K 12 , K 13 , K 21 , K 22 , K 23 ,
K 31 , K 32 and K 33 are the penalty matrices, that contain the
terms due to the elimination of the pressure variable from
the momentum equations by using Eq. (21), which must be
computed with reduced integration (one point in each direc U ) is
tion with linear shape functions) to avoid locking; D(
the capacity matrix that contains the advective terms of the
energy equation; L is the diffusive matrix with constant coefficients (constant thermal conductivity), which contains the
diffusive terms from the energy equation; F1 , F2 and F3 are
the force vectors of the momentum equations that contain
the field forces and the velocity boundary conditions, and G
is the force vector of the energy equation that contains the
heat source terms of the energy equation and the temperature boundary conditions.
For the 3-D problem of Fig. 1, the computational domain
contains both the external fluid and the solid fin. Thus, the
solution of Eq. (8) is also required in order to obtain the complete temperature field. Instead of solving separately for the
two entities (fluid and solid) and imposing the same heat flux
at the interface solid-fluid, as a boundary condition, the solution is sought for the entire domain, simultaneously, with
the same set of conservation equations, imposing zero velocities in the solid fin and the appropriate constant coeffi in Eq. (23),
cients for the solid fin in the diffusive matrix, L,
which are obtained from Eq. (8). For the sake of brevity,
the mathematical details of the components of the above described matrices are not presented. However, the reader is
directed to the work of Reddy and Gartling [15], which was
the basis for the formulation implemented computationally
in the present study. A Fortran subroutine was written to implement the NavierStokes and energy equations in two and
three dimensions as isoparametric, four and eight-noded, linear elements, respectively, which were then aggregated to the
open code FEAP [10].
4. Experiments
An experimental apparatus was built in the laboratory to
produce the necessary experimental data to validate the 3-D
numerical optimization of finned arrangements, and to perform the experimental optimization of finned arrangements.
Fig. 3 shows the main features of the experimental apparatus utilized in this study. A small scale wind tunnel was built
from naval plywood to prevent deformation due to humidity.
A test section was designed in modular form as a drawer,
to allow the testing many configurations just by changing
between the side walls of the wind tunnel and at the midline
of the elemental channels. Three thermistors were placed
at the inlet (T1T3), five at the outlet (T8T12), and four
at the tube surfaces in one elemental channel (T4T7). An
additional thermistor (T13) was placed at the midpoint of
the extended region to measure the free stream temperature.
The thermistors at the inlet and outlet of the arrangement
permitted the determination of the vertical variation of
temperature in the arrangement. In all the tests performed,
the vertical temperatures remained within a 0.5 C margin
with respect to the average (vertical) temperatures calculated
at the inlet and outlet. The thermistors on the tube surfaces
showed that the temperature difference between tubes in one
elemental channel is negligible, namely, within 0.3 C of
the measured average of the four thermistors. Finally, the
additional thermistor placed at the extended region measured
free stream temperatures within 0.5 C of the measured
average arrangement inlet temperature.
Velocity measurements were made with a vane-type digital anemometer, model HHF 300A (OMEGA Engineering
Inc.), which was placed at the extended flow region, as
shown in Fig. 3. For the range of 0.1 to 35 ms1 , the velocity bias limit is 2.5% of the reading. The free stream
velocity was 0.1 and 0.13 ms1 in this study. To allow for
the continuous variation of the fan velocity, a variable power
source with 30 V and maximum current 2 A was used.
The pressure drop measurements were taken with a pressure transducer, model PX137-0.3DV (OMEGA Engineering Inc.), with a nominal range of (02068.5 Pa), which
was connected to a digital pressure meter, model DP25B-S
(OMEGA Engineering Inc.). The differential pressure maximum bias limit is 1% of the reading. The differential
pressure measurements had the purpose of determining the
pressure drop across each arrangement in all experiments,
Fig. 3.
The experimental work involved the acquisition of temperature data in real time. This task was performed through
the utilization of a computational data acquisition system
which consisted of a virtual data logger AX 5810 [17] and
four multiplexers AX758 [18] which allowed for the sequential data acquisition from 64 channels at time intervals of
1/256 s. All the data were processed by a suitable software
application to convert the sensors signals in readable temperatures.
The thermistors were calibrated in the laboratory to
determine the bias limits. The thermistors were immersed in
a constant temperature bath maintained by a bath circulator,
and a total of 64 temperature measurements were made
between 20 and 80 C, at 10 C increments. The largest
standard deviation of these measurements was 0.0005 C,
and, therefore, the bias limit was set at 0.001 C for
all thermistors. This bias limit is in agreement with the
0.0003 C of the same thermistors in a natural convection
experiment [19] and with the 0.0005 C bias limit listed in
an instrumentation handbook [20].
483
(24)
484
(25)
485
486
6. Conclusions
487
Fig. 12. Fin temperature distribution for four-row tubes and plate fin heat exchangers: (a) S/2b = 0.5, e = 1, (f = 0.006) and ReL = 852; (b) S/2b = 0.5,
e = 0.5, (f = 0.006) and ReL = 852.
Acknowledgements
The authors acknowledge with gratitude the support of
the Program of Human Resources for the Oil Sector and
Natural Gas, of the Brazilian Oil National AgencyPRHANP/MCT. A. Bejans work was supported by the grant
from the National Science Foundation.
References
[1] A. Bejan, Shape and Structure, from Engineering to Nature, Cambridge Univ. Press, Cambridge, UK, 2000.
[2] H. Brauer, Compact heat exchangers, Chem. Process Eng. August
(1964) 451460.
[3] S.N. Bordalo, F.E.M. Saboya, Pressure drop coefficients for elliptic
and circular sections in one, two and three-row arrangements of plate
fin and tube heat exchangers, J. Braz. Soc. Mech. Sci. XXI (4) (1999)
600610.
[4] S.M. Saboya, F.E.M. Saboya, Experiments on elliptic sections in one
and two-row arrangements of plate fin and tube heat exchangers, Exp.
Therm. Fluid Sci. 24 (2001) 6775.
[5] J.Y. Jang, J.Y. Yang, Experimental and 3-D numerical analysis
of the thermal-hydraulic characteristics of elliptic finned-tube heat
exchangers, Heat Transfer Eng. 19 (4) (1998) 5567.
[6] L.A.O. Rocha, F.E.M. Saboya, J.V.C. Vargas, A comparative study of
elliptical and circular sections in one and two-row tubes and plate fin
heat exchangers, Internat. J. Heat Fluid Flow 18 (1997) 247252.
[7] F.E.M. Saboya, E.M. Sparrow, Experiments on a three-row fin and
tube heat exchangers, J. Heat Transfer 98 (1976) 520522.