You are on page 1of 16

Applied Thermal Engineering 21 (2001) 771±786

www.elsevier.com/locate/apthermeng

Air-cooled heat exchanger inlet ¯ow losses


C.J. Meyer *, D.G. Kr
oger
Department of Mechanical Engineering, Private Bag X1, Matieland 7602, South Africa
Received 3 December 1999; accepted 15 July 2000

Abstract
The primary purpose of this experimental investigation is to determine the in¯uence that di€erences in
air-cooled heat exchanger geometry have on heat exchanger inlet air ¯ow losses. Heat exchanger geometry
would not only include details pertaining to the heat exchanger ®nned tubes, but also the orientation of the
heat exchanger ®nned surfaces with respect to the inlet air ¯ow as well as to the inlet air angle of incidence
on the heat exchanger. It is found that the inlet air ¯ow losses are independent of the average air velocity
through the heat exchanger. Inlet air ¯ow losses increase with a decrease in the inlet air angle of incidence.
The orientation of the heat exchanger ®nned surfaces has an e€ect on the inlet air ¯ow losses and is pri-
marily a function of the tube cross-sectional pro®le of the ®nned tubes. An equation based on the exper-
imental results is formulated to calculate the heat exchanger inlet air ¯ow losses. It is also shown that the
geometric details of the heat exchanger ®nned tubes or heat exchanger core used in the construction of the
heat exchanger could have a pronounced e€ect on the inlet air ¯ow losses. Ó 2001 Elsevier Science Ltd. All
rights reserved.

Keywords: Air-cooled heat exchanger; Inlet air ¯ow losses

1. Introduction
An air-cooled heat exchanger (ACHE) is a device that facilitates the transfer of heat energy
from a hot process ¯uid to ambient air. A typical forced draught ACHE consists of a heat ex-
changer built up of rows of ®nned tubes and an air pump, usually an axial ¯ow fan that forces
ambient air across the ®nned surfaces of the heat exchanger ®nned tubes. The hot process ¯uid is
cooled as it ¯ows through the heat exchanger ®nned tubes.
Forced draught ACHEs in the past have been employed as condenser units in power stations
[1,2] where the large size of these units and space limitations have popularised the idea of

*
Corresponding author. Fax: +27-21-808-4958.
E-mail address: cjm2@ing.sun.ac.za (C.J. Meyer).

1359-4311/01/$ - see front matter Ó 2001 Elsevier Science Ltd. All rights reserved.
PII: S 1 3 5 9 - 4 3 1 1 ( 0 0 ) 0 0 0 8 3 - 1
772 C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786

Nomenclature

K dimensionless pressure loss/rise coecient


p pressure (N/m2 )
Ry characteristic ¯ow parameter (m)
v velocity (m/s)
Greeks
D di€erential
q density (kg/m3 )
l dynamic viscosity (kg/ms)
h semi-apex angle (°)
Subscripts
c contraction
e exit
f friction
ih inlet conditions as a result of the semi-apex angle, h
o outlet

arranging the heat exchanger ®nned tube bundles in V-styled arrays as schematically shown in
Fig. 1. The use of this arrangement is, however, characterised by an increase in the heat exchanger
air ¯ow losses especially in instances where the semi-apex angle, h, of the V is small [3].
This increase in heat exchanger air ¯ow losses has prompted a number of theoretical and/or
experimental investigations. A number of investigations focused on air ¯ow losses associated with
the ¯ow-®eld downstream of V-styled heat exchangers [4±7].

Fig. 1. Schematic representation of a single unit steam condenser.


C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786 773

Although mainly concerned with changes in heat transfer characteristics, Fisher [8] reported a
progressive increase in heat exchanger ¯ow losses as a result of decrease in the semi-apex angle for
a range of heat exchangers. Preece and Hitchcock [9] reported a 7±28% di€erence in the static
pressure di€erences measured across heat exchangers inclined at a h-value of 11.5° with a change
in the orientation of the ®nned surfaces of the heat exchangers. Mohandes et al. [3] showed that
this high degree of variation encountered by Preece and Hitchcock [9] was markedly reduced when
total pressure, rather than static pressure changes, were considered.
In their investigation, Mohandes et al. [3] focused attention on the heat exchanger inlet air ¯ow
losses in an e€ort to relate the latter to the semi-apex angle of the V-styled heat exchanger. Apart
from proposing an approximate method to predict the increase in heat exchanger inlet losses,
Mohandes et al. [3] concluded that the di€erence in heat exchanger inlet air ¯ow losses due to a
change in the orientation of the ®nned surfaces of the heat exchanger ®nned tubes with respect to
the inlet ¯ow ®eld are small compared to the change in inlet losses due to a change of the semi-
apex angle of the heat exchanger.
Based on the experimental results, Meyer and Kr oger [10] reported the dramatic e€ect that
di€erences in heat exchanger inlet air ¯ow losses resulting from di€erent orientations of the ®nned
surfaces of the heat exchanger ®nned tubes have on the performance characteristics of a forced
draught ACHE. It is, however, mentioned that the heat exchanger used during the experimental
investigation is not necessarily representative of heat exchangers used in industrial applications.
The aim of the current investigation is to resolve some of the issues, raised in the preceding
paragraphs, by evaluating the isothermal performance characteristics of a wide range of V-styled
heat exchangers through experiment. The in¯uence of a change in orientation of the heat ex-
changer ®nned surfaces on the heat exchanger performance characteristics over a range of semi-
apex angles are investigated. The method proposed by Mohandes et al. [3] to determine the heat
exchanger inlet ¯ow losses is critically evaluated and an alternative method is proposed.

2. Analysis
A dimensionless pressure loss/rise coecient, K, is de®ned as
Dp
Kˆ ; …1†
1=2qv2
where Dp is the pressure rise/drop, q, the air density and v, the characteristic velocity.
According to Fig. 2, the total isothermal pressure loss coecient of a heat exchanger, Kh , based
on the average velocity through the heat exchanger can be expressed as
Kh ˆ Kih ‡ Kf ‡ Ke ‡ Ko ; …2†
where Kih , Kf , Ke and Ko are the inlet, frictional, exit and outlet loss coecients, respectively.
A characteristic ¯ow parameter, Ry, is de®ned as
Ry ˆ qv=l; …3†
where v is the characteristic velocity, q, the ¯uid density and l, the ¯uid dynamic viscosity.
774 C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786

Fig. 2. Flow losses through a V-styled heat exchanger.

3. Experimental equipment and procedure


3.1. Experimental equipment

A schematic representation of the experimental set-up and details pertaining to the test section
are shown in Figs. 3 and 4, respectively. A blower forces ambient air through the wind tunnel and
is immediately followed by a ¯ow straightener that serves to smooth the exiting air from the
blower. The mass ¯ow rate through the wind tunnel is controlled via a set of adjustable ¯aps
downstream of the ¯ow straighteners. After passing through the adjustable ¯aps, the air enters a
settling chamber furnished with a set of mesh screens. The exit of the settling chamber forms a
smooth contraction that together with the set of mesh screens within the settling chamber ensures
a smooth and uniform velocity pro®le at the settling chamber exit. The velocity magnitude at the
settling chamber exit is measured with a propeller-type anemometer. The heat exchanger test
section is connected to the settling chamber exit via a length of ducting that serves as a cross-over
piece between the rectangular cross-section of the settling chamber exit (0:33  0:66 m2 ) and the
square cross-section of the test section (0:6  0:6 m2 ).

Fig. 3. Schematic representation of the experimental set-up (elevation).


C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786 775

Fig. 4. Cross-sectional view of the test section (plan).

The test section is divided into two parts along the direction of air ¯ow by an adjustable di-
viding wall furnished with an inlet rounding. The exit of one division is totally blocked o€ and as a
result is ®lled with stagnant air during testing. The exit of the remaining division is constructed in
such a manner as to allow the heat exchanger to be positioned at a prescribed angle, h, with
respect to the inlet air as shown in Fig. 4. The air exits from the heat exchanger into the sur-
roundings.
During the course of the investigation, the performance characteristics of nine di€erent heat
exchangers with a frontal area of 600  600 mm2 were evaluated. A brief description of the dif-
ferent heat exchangers are given below. It should be noted that in the case of heat exchangers
constructed from more than one row of ®nned tubes, the term ``®rst row of ®nned tubes'' refers to
the ®rst row of ®nned tubes, which the air encounters as it ¯ow into the heat exchanger.
(1) 2RR, 4RR and 6RR: Heat exchangers 2RR, 4RR and 6RR are built up from the round
®nned tubes and are constructed using two, four and six rows of round ®nned tubes, respectively.
But for the number of rows of ®nned tubes used, the heat exchangers are geometrically identical,
Fig. 5(a) shows a cross-sectional view through heat exchanger 6RR.
(2) MDC: Heat exchanger MDC shown in cross-section in Fig. 5(b) is constructed using two
rows of plate ®nned heat exchanger ®nned tubes. Apart from using a coarser ®n pitch of 4 mm as
opposed to 2.5 mm, the ®nned tubes in the ®rst row of heat exchanger MDC are identical to those
used in the second row. It should be noted that the ®nned tubes in the second row are staggered
with respect to the ®nned tubes in the ®rst row.
(3) MSF: Heat exchanger MSF is built up of a single row of plate ®nned heat exchanger
®nned tubes of the type used in heat exchanger MDC. The ®nned tubes exhibit a ®n pitch of
2.5 mm.
(4) SM: Heat exchanger SM shown in cross-section in Fig. 5(c) is a double row heat exchanger
constructed from plate ®nned heat exchanger ®nned tubes that are geometrically similar, but
dimensionally smaller than the plate ®nned tubes used in heat exchangers MDC and MSF.
776 C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786

Fig. 5. Cross-sectional views through heat exchangers (a) 6RR, (b) MDC and (c) SM.

(5) LUM: Heat exchanger LUM is a single row heat exchanger constructed from the plate
®nned tubes as shown in Fig. 6. The plate ®ns exhibit a wave pattern in the direction of air ¯ow.
(6) HUNG: Heat exchanger HUNG is constructed from a heat exchanger core as opposed to
individual heat exchanger ®nned tubes. Fig. 7 shows the heat exchanger core in detail.
(7) RAD: Similar to the previous heat exchanger, heat exchanger RAD is constructed from a
heat exchanger core. This particular heat exchanger core as shown in Fig. 8 is often used in
automotive engine cooling systems.
C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786 777

Fig. 6. Finned tube used in heat exchangers LUM.

3.2. Experimental procedure

After mounting a particular heat exchanger in the test section of the wind tunnel as shown in
Fig. 4, the static pressure drop across the heat exchanger is measured over a range of volume ¯ow
rates. The di€erent volume ¯ow rates are chosen so as to yield an averaged velocity through the
heat exchanger ranging from 1 to 6 m/s. This velocity range is representative of the average heat
exchanger velocities encountered in most industrial applications.

Fig. 7. Core of heat exchanger HUNG.


778 C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786

Fig. 8. Core of heat exchanger RAD.

The procedure described in the previous paragraph is repeated for every individual heat ex-
changer over a range of h-values. Both orientations of the heat exchanger are considered for every
h-value tested. The latter refers to the position of the heat exchanger ®nned surfaces with respect
to the inlet air ¯ow ®eld. The ®nned surfaces can either be parallel or inclined to the incoming air
¯ow ®eld.

4. Discussion of results
From the experimental data, it was possible to determine the in¯uence of change in the average
air velocity through a heat exchanger on the Kh -value of the various heat exchangers tested at
di€erent inclinations to the uniform inlet ¯ow ®eld. Fig. 9 displays a plot of Kh vs. Ry for heat
exchanger 6RR at di€erent inclinations to the ¯ow ®eld denoted by the di€erent values of h. The
heat exchanger orientation is such that the heat exchanger ®nned surfaces are inclined with respect
to the inlet air ¯ow ®eld. It is observed that the characteristic lines are more or less parallel for
di€erent values of h which implies that the associated inlet air ¯ow losses are independent of the
magnitude of the average air velocity through the heat exchanger. Mohandes et al. [3] reported a
similar result. This trend was also observed for the other heat exchangers tested.
In order to relate the inlet loss coecient, Kih , in Eq. (2) to the inclination of the heat exchanger
with respect to the inlet ¯ow ®eld, Moore and Torrence [11] developed a simpli®ed model for the
¯ow between a set of parallel plates.
C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786 779

Fig. 9. Kh vs. Ry for heat exchanger 6RR.

The model, shown in Fig. 10, embodies the following assumptions:

· The ®nned surfaces have a negligible thickness resulting in a zero solidity ratio for the heat ex-
changer.

Fig. 10. Theoretical model of Moore and Torrence [77MO1].


780 C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786

· Upon entering the space between the ®nned surfaces ¯ow separation occurs. The pressure with-
in the separated region equals that of the inlet air and as a result the initial velocity between the
®nned surfaces equals the inlet velocity.
· Downstream of the separated region the air expands to ®ll the space between the ®nned sur-
faces.

The following expression for the inlet loss results:

Kih ˆ …1= sin h 1†2 : …4†

Figs. 11 and 12 contrast in the (Kh Khˆ90° ) values for the heat exchangers tested with Kih
calculated according to Eq. (4) over a range of h-values. Fig. 11 displays the data in the instance
where the ®nned surfaces of the ®nned tubes are inclined to the inlet velocity ®eld, and Fig. 12
shows the data in the case where the ®nned surfaces are parallel to the inlet velocity ®eld.
Although it can be seen from Eq. (2) that (Kh Khˆ90° ) is not equivalent to Kih , it should be
remembered that the expression for Kih as displayed in Eq. (2) embodies the assumption of a zero
solidity ratio due to the negligible thickness of the heat exchanger ®nned surfaces. Consequently,
no contraction losses are accounted for in Eq. (2). This is essentially what is expressed by the
(Kh Khˆ90° ) values of Figs. 11 and 12.
The experimental results displayed in Figs. 11 and 12 indicate a general increase in the dif-
ference between the inlet losses for the two orientations of the heat exchanger ®nned surfaces with
a decrease in the h-value. Table 1 lists the di€erences in the (Kh Khˆ90° ) values for the heat ex-
changer tested due to a change in the orientation of the ®nned surfaces of the heat exchanger
®nned tubes. The data of Table 1 indicate that the di€erence in (Kh Khˆ90° ) value is more pro-
nounced for heat exchangers utilising heat exchangers ®nned tubes where the cross-sectional
pro®le of the tubes are circular, i.e. heat exchangers 2RR, 4RR, 6RR and HUNG. It follows that

Fig. 11. (Kh Khˆ90° ) vs. h for the ®nned tube ®nned surfaces inclined to the inlet velocity ®eld.
C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786 781

Fig. 12. (Kh Khˆ90° ) vs. h for the ®nned tube ®nned surfaces parallel to the inlet velocity ®eld.

Table 1
The di€erence in the (Kh Khˆ90 ) values due to a change in the orientation of the heat exchanger ®nned tube ®ns with
respect to the inlet velocity ®eld
h (°) …Kh Khˆ90° †angle …Kh Khˆ90° †parallel
2RR 4RR 6RR MSF MDC SM LUM HUNG RAD
90 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
60 ± ± ± ± ± 0.237 ± ± ±
40 0.396 0.948 0.796 0.120 0.660 0.517 0.990 0.351 0.659
30 1.133 1.557 2.025 0.319 0.173 0.058 0.747 0.653 1.199
20 2.252 2.741 3.196 1.178 1.748 0.169 1.606 1.248 1.740
15 2.310 2.553 4.114 0.743 ± 1.972 0.452 3.712 0.912
10 5.534 5.282 6.782 1.356 2.317 3.350 3.105 6.990 2.062

the cross-sectional shape of the tubes rather than the shape of the ®nned surfaces of the ®nned
tubes in¯uence the di€erence in the ¯ow losses associated with the orientation of the ®nned
surfaces with respect to the inlet ¯ow ®eld. The results also indicate that the expression for Kih
developed by Moore and Torrence [11] is consistently lower than the experimentally determined
…Kh Khˆ90° † values.
After consideration of the theoretical model proposed by Moore and Torrence [11], Mohandes
et al. [3] observed that plots of …Kh Khˆ90° †=…1= sin h 1† vs. 1= sin h produced straight lines with
gradients close to unity for a range of perforated sheets and also for sets of parallel plates inclined
to the inlet air ¯ow ®eld. Mohandes et al. [3] subsequently proposed the following expression for
Kh :
Kh ˆ …2Kc0:5 ‡ 1= sinh 1†…1= sin h 1† ‡ Khˆ90° ; …5†
where Kc is the inlet contraction ratio.
782 C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786

Fig. 13. …Kh Khˆ90° †=…1= sin h 1† vs. 1= sin h for heat exchanger 2RR and 4RR.

Figs. 13±17 display plots of …Kh Khˆ90° †=…1= sin h 1† vs. 1= sinh for the di€erent heat ex-
changers tested during the course of this investigation. Note that the orientation of the ®nned
surfaces of the heat exchanger ®nned tubes are indicated in the legend of Figs. 13±17 by an ``A''
when the ®nned surfaces are angled and ``P'' when the ®nned surfaces are parallel to the inlet
velocity ®eld. The ®gures clearly indicate a linear relationship between …Kh Khˆ90° †=…1= sin h 1†
and 1= sin h although the gradients of these lines vary approximately from 0.54 to 1.0 as indicated

Fig. 14. …Kh Khˆ90° †=…1= sin h 1† vs. 1= sin h for heat exchanger 6RR and MDC.
C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786 783

Fig. 15. …Kh Khˆ90° †=…1= sin h 1† vs. 1= sin h for heat exchanger MSF and SM.

Fig. 16. …Kh Khˆ90° †=…1= sin h 1† vs. 1= sin h for heat exchanger LUM and HUNG.

by the lines ®tted through the data using a least squares approximation. Based on these ®ndings
the following expression for Kh is proposed:
Kh ˆ …b ‡ a= sinh†…1= sinh 1† ‡ Khˆ90° ; …6†

where a and b are the gradient and y-axis intercept of the ®tted lines of Figs. 13±17.
784 C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786

Fig. 17. …Kh Khˆ90° †=…1= sin h 1† vs. 1= sinh for heat exchanger RAD.

It is interesting to note from Figs. 16 and 17 that the data sets of heat exchangers whose geo-
metry most closely resemble sets of parallel plates (HUNG, RAD and LUM) as those tested by
Mohandes et al. [3] can be correlated with straight lines exhibiting gradients close to unity. In the
case of heat exchangers HUNG and RAD, the ®nned surfaces are at an angle to the inlet ¯ow ®eld
as opposed to heat exchanger LUM where the ®nned surfaces are parallel to the inlet ¯ow ®eld. In
case of the latter the ¯at pro®led tubes of the ®nned tubes act as sets of parallel plates. These
results would seem to indicate that Eq. (5) is a special case of Eq. (6) where the gradient a ˆ 1 and
the y-axis intercept b ˆ …2Kc0:5 1†.
An expression for the contraction coecient, Kc , for plate ®ns based on the free stream velocity
is given by Kays [12]

Kc ˆ …1=rc 1†2 =r221 ; …7†

where rc is the jet contraction ratio and r21 the solidity ratio of the plate ®ns.
A relationship between rc and r21 for parallel plates is given by Rouse [13] and parameterised
by Van Aarde and Kr oger [7] with the following expression:

rc ˆ 0:6144517 ‡ 0:04566493r21 0:336651r221 ‡ 0:4082743r321 ‡ 2:672041r421


5:963169r521 ‡ 3:558944r621 : …8†

By means of Eqs. (7) and (8), it is possible to calculate theoretical Kc -values for heat exchangers
HUNG and RAD and according to White [14] a theoretical Kc -value of approximately 0.04 is
applicable for heat exchanger LUM. These theoretical Kc -values are compared in Table 2 with the
Kc -values calculated from the lines ®tted through the relevant experimental data points of Figs. 16
and 17 using the assumption made by Mohandes et al. [84MO1] that a ˆ 1 and b ˆ …2Kc0:5 1† in
C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786 785

Table 2
Comparison of theoretically and experimentally determined Kc -values
Heat exchanger Kc
Theoretical Experimental
HUNG 0.14 0.54
RAD 0.05 0.55
LUM 0.04 0.09

Eq. (5). Table 2 indicates that the theoretically determined Kc -values are consistently lower than
those determined from the experimental data for the relevant heat exchangers.
Of all the ®tted lines shown in Figs. 13±17, the line that displays the smallest gradient correlates
the data points for heat exchanger LUM in Fig. 16 where the ®nned surfaces of the ®nned tubes
are at an angle to the inlet ¯ow ®eld. The small gradient could be attributed to the wave pattern of
the ®nned surfaces in the vicinity of the heat exchanger inlet. These curved surfaces could act as
inlet guides for the air ¯owing into the passages between the ®nned surfaces resulting in a di-
minished inlet loss.
The experimental data of heat exchanger RAD displayed in Fig. 17 are also of interest. Unlike
the data of the other heat exchangers, the measured inlet losses in the case wherein the ®nned
surfaces of the heat exchanger are parallel to the inlet ¯ow ®eld are higher than when the ®nned
surfaces are oriented at an angle to the ¯ow ®eld. This phenomenon can be attributed to the
unique staggered con®guration used for the position of the heat exchanger ®nned tubes that
would cause the angle of incidence between the inlet ¯ow ®eld and a line running through the
centre of a ®nned tube in the ®rst row and the corresponding ®nned tube in the last row to di€er
from the angle of incidence experienced for a di€erent con®guration.
Heat exchangers 2RR, 4RR and 6RR are nearly identical and di€er only in the number of rows
of ®nned tubes used which are 2, 4 and 6 respectively. It is expected that these heat exchangers will
exhibit identical inlet losses in the case where the ®nned surfaces of the ®nned tubes are at an angle
to the incoming ¯ow ®eld. Figs. 13 and 14, however, indicate that is not the case. The di€erences
in inlet losses can be attributed to distortion of the air ¯ow ®eld exiting from the heat exchanger
for h-values of 30° and lower. These outlet ¯ow ®eld distortions are most noticeable for heat
exchanger 2RR and to a lesser degree for heat exchanger 4RR. In the case of heat exchanger 6RR,
no outlet ¯ow ®eld distortions was observed for both orientations of the heat exchanger over the
range of h-values tested.

5. Conclusions
The experimental results show that the inlet air ¯ow losses are independent of the average air
velocity through the heat exchanger. The results further indicate an increase in the di€erence
between the inlet air ¯ow losses of a particular heat exchanger as a result of a change in the
orientation of the ®nned surfaces of the heat exchanger ®nned tubes as the h-value is decreased.
There are strong indications that the di€erence in inlet air ¯ow losses as a result of di€erent heat
exchanger orientations is related to the cross-sectional pro®le of the tubes of the heat exchanger
786 C.J. Meyer, D.G. Kroger / Applied Thermal Engineering 21 (2001) 771±786

®nned tubes and found to be more pronounced for ®nned tubes using round tubes as opposed to
®nned tubes utilising ¯at or elliptical tubes.
The dimensionless total pressure loss coecient, Kh , is best expressed by Eq. (6) where coef-
®cients a and b are determined experimentally. The expression proposed by Mohandes et al. [3] is
shown not only to be a special case of Eq. (6) applicable to heat exchangers characterised by
®nned surfaces closely resembling sets of parallel plates, but also to underestimate heat exchanger
inlet losses.
The empirical constants a and b displayed in Eq. (6) are shown to be sensitive to the geometrical
characteristics of the heat exchangers tested. The large degree of variation in the values of con-
stants a and b for the heat exchangers tested indicates that these values should best be determined
experimentally depending on the degree of accuracy required.

References

[1] G.J. Kosten, K.I. Morgan, J.M. Burns, P.L. Curlett, Operating and Performance Testing of the WorldÕs Largest
Air-cooled Condenser, American Power Conference, Chicago, 1981.
[2] H.B. Goldschagg, Lessons Learned from the WorldÕs Largest Forced Draft Direct Air-cooled Condenser, EPRI
meeting, Washington, 1993.
[3] M.A. Mohandes, T.V. Jones, C.M.B. Russell, Pressure loss mechanisms in resistances inclined to an air ¯ow with
application to ®n tubes, First National Heat Transfer Conference, Leeds, 1984.
[4] F.K. Moore, J.R. Ristorcelli, Turbulent ¯ow and pressure losses behind oblique high-drag heat exchangers, Int. J.
Heat Mass Transfer 22 (1979) 1175±1186.
[5] F.K. Moore, Flow Fields and Pressure Losses of V-bundles with Finite Resistance, ASME Paper 79-WA/HT-4,
1979.
[6] J.C.B. Kotze, M.O. Bellstedt, D.G. Kroger, Pressure drop and heat transfer characteristics of inclined ®nned tube
heat exchanger bundles, Proc. Eighth Int. Heat transfer Conf., San Francisco, 1986.
[7] D.J. Van Aarde, D.G. Kr oger, Flow losses through an array of A-frame heat exchangers, Heat Transfer Engng.
14 (1) (1993) 43±51.
[8] O. Fisher, Experimental investigations of forgo-type dry cooling elements, Proc. Fourth IAHR Cooling Tower
Workshop, Interlaken, Switzerland, 1984.
[9] R.J. Preece, J.A. Hitchcock, Comparative performance characteristics of some extended surfaces for air-cooled
applications, Chem. Engng. (1972) 238±244.
[10] C.J. Meyer, D.G. Kr oger, Plenum chamber ¯ow losses in forced draught air-cooled heat exchangers, Appl. Ther.
Engng. 18 (9,10) (1998) 875±893.
[11] F.K. Moore, K.E. Torrence, Air ¯ow in dry natural-draught cooling towers subject to wind, Cornell Energy
Report, Ithaca, New York, 1977.
[12] W.M. Kays, Loss coecients for abrupt changes in ¯ow cross-section with low Reynolds number ¯ow in single and
multiple-tube systems, Trans. ASME 72 (1950) 1067±1074.
[13] H. Rouse, Elementary Mechanics of Fluids, Wiley, New York, 1946.
[14] F.M. White, Fluid Mechanics, McGraw-Hill, New York, 1986.