Heat Transfer Characterization of Flat Plain

Fins and Round Tube Heat Exchangers

N. Kayansayan • The effects of outersurface geometry on the performance of flat plain
Dokuz Eyliil University, fins and round tube crossflow heat exchangers are considered. With the
Department of Mechanical Engineering, finning parameter varying from 11 to 23, a total of 10 geometrically distinct
Izmir, Turkey configurations were tested over a Reynolds number range of 100-30,000.
The tube outside diameter and collar thickness define the characteristic
dimension. The convective heat transfer coefficients are presented as plots
of the Colburn j factor versus Reynolds number and compare well with
previous studies. The dispersion in the majority of the data is + 10%. The j
factor, Reynolds number, and finning parameter are correlated.

Keywords: compact heat exchangers, fin-and-tube exchanger, forced

convection

INTRODUCTION tion factor to the Reynolds number given by (Ao/Ato)".

The Reynolds number in the analysis ranged between 100
Heat exchangers with fiat fins and round tubes are quite and 4000.
common in applications related to the air-conditioning, The work now presented documents the average heat
heating, and refrigeration industries. Owing to the com- transfer coefficients for 10 distinct fin-tube-bank configu-
plex pattern of fluid flow over the fin-and-tube surface, rations obtained from controlled experiments in a wind
the theoretical prediction of heat transfer coefficients is tunnel. In the experiments, the number of tube rows along
often precluded. The combined process of heat and mo- the flow direction was four, and the Reynolds number
mentum transfer serves to complicate the analysis. There- spanned the range from 102 to 3 x 104. The characteristic
fore, it is necessary to resort to experimentation in order dimension is the tube outside diameter including the
to construct useful models. collar wall thickness. This choice makes it possible to
A variety of flow configurations have been studied and correlate the heat transfer data in a compact form. Com-
documented in the literature. Reviews of the literature parison of the present results with previous studies is also
have been given by Webb [1] and McQuiston [2]. The provided.
results reported here, however, are unique in that the
present study not only extends the range of the geometri-
cal parameters of previous studies but also considers a EXPERIMENTAL SETUP AND
larger Reynolds number range. The following review is INSTRUMENTATION
not intended to be exhaustive, but rather to provide a Wind Tunnel
background for the present study.
Rich [3, 4] examined the effects of fin spacing and A wind tunnel facility similar to the one used in previous
number of tube rows on the heat transport of several heat compact exchanger analysis [7] was modified to accept
exchangers. Varying the number of tube rows from one to exchanger prototypes with approximately 0.25 m 2 frontal
six, Rich concluded that, depending upon the Reynolds area and to provide two-dimensional flow as free of vibra-
number, the average heat transfer coefficient for a deep tion and turbulence as reasonably possible for exchanger
coil may be higher or lower than that for a shallow coil. performance studies. A schematic diagram of the wind
In the Colburn j-factor correlation presented by tunnel is shown in Fig. 1. The system is designed to suck
Elmahdy and Biggs [5], the Reynolds number exponent, room air over the finned side of the exchanger while
m, was assumed to be a strong function of the physical circulating hot water through the tubes. The tunnel, made
parameters of the finned tube exchanger over the Reynolds of 0.5 mm thick galvanized sheet metal, was a square duct
number range 200-2000. Experiments were performed, 50 cm x 50 cm in cross section and 1100 cm in overall
and the m values for every individual exchanger with a length. To avoid the flow of dust particles into the system,
specified geometry were determined by a regression analy- the entrance section contains two 100 cm x 100 cm
sis method. screens of 10 meshes per cm, and 0.2 mm diameter steel
McQuiston [6] developed a very simple correlation for wire cloth.
four-row staggered banks with plain fins. It was found that Through a 50 cm long Zanker-type flow straightener [8],
the j-factors were best correlated by applying a multiplica- air flows approximately 500 cm in a straight horizontal

Experimental Thermaland Fluid Science 1993; 6:263-272

© 1993 by Elsevier Science Publishing Co., Inc., 655 Avenue of the Americas, New York, NY 10010 0894-1777/93/$6.00
263
264 N. Kayansayan

777 ~.~. I A /
///
/

Figure 1. A schematic diagram of the experimental apparatus and the instrumentation. 1, Blower; 2, wind tunnel; 3, insulation;
4, test exchanger; 5, thermocouple; 6, pump; 7, air-side measurement point; 8, straightener; 9, screen; 10, rotameter; 11, main
valve; 12, electronic variator; 13, flow adjustment valve.

duct before reaching the test section. As depicted in Fig. ture was allowed to vary within + 3°C of the preset value.
1, the duct wall surfaces 100 cm downstream and up- Owing to the large capacity of the boiler tank, stable
stream of the test section are furnished with a total of 12 temperatures at the exit were achieved.
holes 10 mm in diameter. Axisymmetric with these holes,
cylindrical Teflon elements, 10 mm I.D., are attached to Test Heat Exchanger
the tunnel to provide access holes for the velocity probe.
Air leaving the metering section flows through a sheet Figure 2 shows the fin layout and the tube circuit arrange-
metal transition section and enters the fan. At the fan ment of the exchanger that was studied in this experiment.
exit, the air is discharged to the outdoors. To minimize Table 1 presents the geometrical parameters of all the
heat loss to the surroundings, the tunnel outer surface is tested coils. Each core had fiat, continuous 0.2 mm thick
insulated with a 2 cm thick glass-wool layer. Additionally, aluminum fins with collars. The copper tubes of 0.5 mm
being supported by stands of perforated steel plates, the wall thickness, a product of Wieland Corporation, were
duct system is elevated 50 cm above the floor level of the manufactured with ___0.06 mm tolerance on the outside
laboratory room. diameter. After assembly, the tubes were mechanically
Power for the wind tunnel was provided by a Sontec expanded into the fins and tube sheets. The mechanical
Model 6938 fan driven by a 3 kW ac motor. The motor bond between the fins and tubes was checked and judged
was in turn powered by an electronic variac; a three-phase to be quite tight, and a negligible fin-tube thermal con-
motor control unit and the fan speed could be varied in a tact resistance was secured. The return bends were manu-
continuous manner from 0 to 1350 rpm. Thus it was ally soldered to the tube extensions. Thus, the tubes of
possible to alter the tunnel air velocity in the range of each row were interconnected, and four identical multi-
0-15 m / s . A digital display panel indicated the fan rota- pass crossflow circuits connected in parallel were ob-
tional speed. tained.
Avoiding any possible clogging, each circuit was tested
Hot Water System with pressurized air. Then the 25 mm steel tubing headers
for the supply and collection of hot water through the
The hot water system consists of a boiler of 115 kW circuits were attached. The tube sheets, which form a
heating capacity, a circulating pump, a flow-metering unit, casing for the core and possess mounting holes on their
and the test exchanger. All components of the system periphery, were fabricated of galvanized steel sheet 0.5
were interconnected with insulated steel piping 25 mm in mm thick.
diameter. Thus, a closed circuit between the boiler and
the test exchanger was established. The boiler contained Instrumentation
1500 liters of water and was fired by a burner. A Honey-
well thermostat, located at the exit, kept the water tem- The hot water supply to the test section was metered by
perature at a preset value of 80°C. The burner was con- an A S A glass tube variable-area rotameter. The meter
trolled by the thermostat so that the exit water tempera- had a sensitivity of 1 liter/min per cm of bob displace-
 Flat Plain Fins and Round Tube Exchangers 265

5212
.,__$2t2 t
]-T . . . . . .

. - - k.

oV I
....~ - - - b,,
II

A~RFLOWIIII1~
b
a
Collor

o __J
;-it

Figure 2. (a). The heat exchanger characteristic geometry. (b) The multipass water flow circuit. (c). A magnified view of the
tube-fin combination.

ment and was calibrated to be accurate within + 2 % of were obtained with a TSI Model 1650-1 hot-wire
the full range. The flow rate adjustment through the coil constant-temperature anemometer. The extendable probe
was accomplished by two gate valves located at the inlet wand had a sensing tip 4.7 mm in diameter. Using the
and outlet of the rotameter. sensor as a resistance thermometer, the instrument could
The water temperatures were recorded by a Sonde also be used to measure air temperature. As specified by
temperature indicator set. Measuring temperatures in the the manufacturer, the accuracy in velocity measurements
range of - 1 5 ° C to 90°C, the probes of the instrument was + 2%, and in temperature measurements +0.8% of
were 24 A W G copper-constantan (Type T) thermocouple full scale.
elements enclosed in a 10 mm O.D. stainless steel protec- A barometer indicated the ambient pressure, and a
tion tube. Vinyl-insulated lead wires terminate at the psychrometer was used to measure the dry bulb and wet
socket junction of the analog indicator. The set was cali- bulb temperatures of the room air.
brated by placing the probes in a variable-temperature
bath whose temperature was measured by a precision EXPERIMENTAL P R O C E D U R E AND DATA
thermometer. As illustrated in Fig. 3, the probes were REDUCTION
housed in wells in the exchanger headers, and their posi-
tions were fixed by fittings. To attain uniform water- The heat exchanger with specified surface geometry was
temperatures, a mixer, made of a perforated shim, was installed in the wind tunnel in such a manner that the
located upstream of each probe. horizontal position was checked with a level, and the
The airstream velocity and temperature measurements tunnel connections were sealed with epoxy. For some
266 N. Kayansayan

Table 1. Geometric Parameters of the Tested Coils

Exchanger
Tube Coil Flow Finning Tubes Number
Coil Diameter Height Length Transverse Long. Fin Spacing Factor per Row of Rows
Type dto (mm) B (ram) L (mm) Pitch s I (ram) Pitch s2 (mm) S F (m - 1) • n N
1 16.3 500 139 40 34.67 454 23.24 12 4
2 16.3 500 139 40 34.67 312 15.81 12 4
3 16.3 500 139 40 34.67 238 12.12 12 4
4 9.52 480 104 30 26 454 23.53 16 4
5 9.52 480 104 30 26 312 16.00 16 4
6 9.52 480 104 30 26 238 12.33 16 4
7 9.52 482 88 25.4 22 454 16.44 19 4
8 9.52 482 88 25.4 22 312 11.28 19 4
9 9.52 482 88 25.4 22 400 14.43 19 4
10 12.5 493 127 31.75 32 454 22.81 15 4

configurations, the exchanger height was less than the air-side mass flow rate was then determined as follows:
tunnel dimensions, and the bypass flow was eliminated by 5 4
a thin layer of foam plastic sandwiched between the edges M = ( paV)o + ¼ E ay E ( PkVk)j (|)
of the fins and the casing. Upon completion of the water- j=l k=l
side links, the coil was completely insulated with a 5 cm
thick layer of glass wool. The air in the water circuit was where the subscript k indicates the four velocity values at
purged out through the purging plugs. The upstream and a particular segmental area a .
. J

downstream valves of the rotameter were adjusted such Similarly, the air enthalpy at the exit is
that the average water velocity through the coil tubes was 5 4
approximately 0.5 m/s, and then the tunnel blower was He = ( paVcpT) o + ¼ E aj E ( PkCp,kVkT~)y (2)
turned on and the air velocity was adjusted to a desired j=l k=l
value. The water inlet and outlet temperatures were peri-
Because of the uniform temperature distributions at the
odically checked, and equilibrium was assumed to exist if
inlet of the test section, the inlet air enthalpy is
no appreciable deviation in water temperature change was
observed for the last 15 min prior to data recording. //in = Mcp,inTin (3)
As shown in Fig. 4, the tunnel cross section was divided
into six segmental areas, and in accordance with the The difference between Eqs. (2) and (3) yields the rate at
log-linear rule [8] the velocity and temperature of the which heat was gained by the air and was compared with
airstream at a total of 21 grid points were measured. The the heat loss of the water. In most experimental runs, the
heat rate difference between the two sides was within
_ 5% range of the water-side heat rate. In calculating the
exchanger overall conductance, UA, however, the arith-
metic average of the air- and water-side heat rates was
taken into account. The uncertainties in the measured
properties were estimated to be as in Table 2. With the
uncertainties given in Table 2, and over the indicated
ranges, the method of Kline and McClintock [9] was
employed to evaluate the uncertainties of the experimen-
tal results. For a typical case, the average heat flow rates
were found to be within 6.1%, the Reynolds numbers
within 8.1%, and the j factors within 11.2% of the re-
ported values.
By the Colburn analogy [10], the functional relationship
I Nu = ~b(Re, Pr, flow geometry), suggested by the govern-
ing equations becomes
Nu = C Re m Prl/3e n (4)
for Prandtl numbers in the range 0.5 < Pr < 100. For the
test cases, it was calculated that Pr ~ 0.7. In this study,
the maximum velocity, that is, the velocity at the minimum
flow area, was used for the Reynolds number characteris-
tic velocity, and as given by Eq. (A1) in the Appendix the
characteristic diameter included the collar thickness. Thus,
Figure 3. A schematic diagram of the water-side flow distri- the Reynolds number is
bution and the instrumentation. 1, Thermocouple; 2, ther-
mowell; 3, mixer; 4, header; 5, test exchanger. Re = Gmdo/lX b (5)
 Flat Plain Fins and Round Tube Exchangers 267

130 130 Teflon probe guider

as

k=l
J= 1

J:2

J=3

J=4

J=5 " T

L.
500

Figure 4. The wind tunnel cross section illustrating the velocity and temperature measurement points.

where G m = M / / A m i n • In Eq. (4), as suggested by where A T m is the logarithmic mean temperature differ-
McQuiston [6], the flow geometry effects are represented ence calculated from the measured inlet and outlet water
by the exchanger finning factor and air temperatures and F is the correction factor ap-
plied to the mean temperature difference [11]. The overall
= Ao/Ato (6)
heat transfer coefficient is related to the desired air-side
Combining the definition of the Stanton number and the film coefficient by
sensible Colburn j factor yields
j = N u / ( R e Pr 1/3) (7)
1 Ao( ) + + Rc (10)
Hence, it is apparent from Eq. (4) that
Vo hi ~o
where R c is the combined resistance of the tube wall and
j = C R e " - le" (8) the collar. As this has a value of 3.8 × 10 -s m 2. °C/W, it
A multiple linear regression analysis of the experimen- was neglected compared to the other terms of Eq. (10).
tal data permits the determination of the coefficients of The surface efficiency, r/o, is given by
Eq. (8).
Determination of h o, however, is made by first deter- no = 1 - - ~ ( 1 - ,/f) (11)
.,ri o
mining an overall heat transfer coefficient from the rela-
tionship Here, ~/f is the fin efficiency and is calculated as in Ref.
Q = UAFAT m (9) 13.
Owing to the existence of fully developed turbulent flow
inside the tubes, the water film coefficients h i were deter-
Table 2. Experimental Uncertainties mined from the Dittus-Boelter correlation [14],
Property Uncertainty Range Nui = 0.023 Re °8 Pr °'4 (12)
Water flow rate + 0.5 liter/min _< 28 liters/min Since the surface efficiency r/o depends upon h o, an
Inlet water temp. + 0.8°C 77-84°C iterative determination of ho from measured data was
Water temp. difference -1-I°C 10-41°C required.
Outlet air temp. + 0.8°C 29.2-67.8°C
Inlet air temp. + 0.5°C 7-19.5°C RESULTS AND DISCUSSION
Air velocity + 0.06 m / s 0-3 m / s Preliminary heat transfer measurements were undertaken
Air velocity :t: 0.2 m / s 2.5-12.5 m / s to check the instrumentation and methodology used in
Probe access length + 1 mm 25-475 mm this study. There are a number of finned-tube configura-
268 N. Kayansayan

Table 3. Geometric Properties of the Compared Coils ~.16 2 I I I I I I I I

Finning Hydraulic Free-Flow
Fig. Factor Diameter Area Ratio O This study [Coil t y p e 9 ]
No. Ref E d h (mm) cr [] McQuiston (}2, p 284 ]

5 This study 11.28 3.8 0.571

[12] 11.23 3.9 0.579 162 •D D
6 This study 14.43 3.0 0.560 O
o %%a
7
[12]
This study
13.88
16.44
3.1
2.6
0.572
0.553 J
f 0 0
°oo o
Oo
[3] 17.54 2.7 0.543 oo
8 This study 15.81 3.8 0.546
[3] 12.34 3.9 0.555
9 This study 23.24 2.6 0.529
[5] 21.41 2.7 0.540
1~ 3 I I , II , , , ,
10 2 103 10 4
tions for which experimental data are available that can C.~ndh
provide a basis for comparison with the results reported Jab
here. The related geometrical properties of the compared
coils are presented in Table 3. The compared sensible Figure 6. j versus Re h. Comparison of present heat transfer
heat transfer coefficients are given in Figs. 5-9 and are results with the data of McOuiston, E = 14.43.
consistent with the literature values. The coil hydraulic
diameter, as defined by Eq. (A13) of the Appendix, is used 5, Elmahdy reported an overestimation to data. In addi-
in determining the Reynolds numbers. tion, the coil tested by Elmahdy had eight rows in the flow
In general, the trends for the Colburn j factors are in direction. Then the higher j factors in his work are also
agreement with those documented in the literature. In consistent with the conclusiosn of Rich [4].
Fig. 6, owing to experimental uncertainties at low flow Reducing the measured values for a total of 110 experi-
rates, a maximum of 25% deviation in the results is noted. mental runs to j factors as defined by Eq. (7), all the data
As the Reynolds number increases, however, the discrep- points are shown in Fig. 10. The mean line through the 10
ancy decreases. A similar trend is observed in Fig. 8. In geometrical combinations of e was obtained by a least
this figure, the distinct behavior of the two compared coils squares curve fit. In the least squares treatment, the data
is attributable to a 22% discrepancy in the finning factors. points with Reynolds numbers below 500 were excluded
As given by Eq. (8), the finning factor 6, representing the because of low Reynolds number effects, conduction and
surface geometry, directly influences the j factor, and natural convection, which preclude a boundary type of
such deviations as those in Fig. 8 are expected to occur. In analysis. Accordingly the following correlation was ob-
Fig. 9, although the proper trend is exhibited, Elmahdy's tained.
correlation for his test heater consistently shows higher
values for the heat transfer. As illustrated in Fig. 5 of Ref. j = 0.15 Re 0.28~-0.362 (13)

4.162 I I 1 I I I I I I 4.10 -2 I I I I I I I I I

OThis study [Coil t y p e : 8 ) OThis study [Coil t y p e : 7 ]

D M¢ Quiston [12, p. 275 ] ORich [3, Fig:t1 ]
[] []
162
[]& 162
O []
0 D(~ ~Do~o
t %o
0
0
©
t o Oo o oo
J 0 J

163 , , I I I I I I , 1(~3102 I I t I I L I t t
10 2 103 10 4 103 104
Gmdh 6mdh
Jab .attb
Figure 5. j versus Re h. Comparison of present heat transfer Figure 7. j versus Re h. Comparison of present heat transfer
results with the data of McQuiston, e = 11.28. results with the data of Rich, E = 16.44.
 Flat Plain Fins and Round Tube Exchangers 269

4-162 ! l I I 4 -102
I I I I I I I I I I l i i

0 This study {Coil type: 2.]

..... Elmohdy 15]
[3 Rich [3,Fig: 9 ]
0 This study{Coil type:t]

[D O
162 HOlD 162
[] []
0 o 0 0
t 0 0170r-I r7
o o t o
O "~
J OOo J o o
o

, , , ,I , , ,, Id 3 i l I I I I I I I
10 2 103 104 102 10 3 10 4
Gmdh Gmdh
.,Ub J~b
Figure 8. j versus Re h. Comparison of present heat transfer Figure 9. j versus Re h. Comparison of present heat transfer
results with the data of Rich, ~ = 15.81. results with Elmahdy's correlation, • = 23.24.

in which 500 < Re < 30,000 and 11.2 _< e < 23.5. The
thermophysical properties in Eq. (13) are evaluated at the sume a value between - 0 . 4 and -0.2, where the lower
arithmetic average of the air inlet and outlet bulk temper- limit represents the tube bank and the. upper limit the
atures. channel flows. Applying a multiple regression technique to
A search of the literature revealed that attempts have the data of 16 fiat-plate heat exchangers, Gray and Webb
been made to obtain generalized correlations for the heat [16] developed a correlation in which the Reynolds num-
transfer coefficients related to the subject of the present ber exponent was -0.328. In this study, the j-factor slope
study by McQuiston [6] and more recently by Webb [15]. is determined to be -0.28. Such a slope value appears to
In McQuiston's analysis, however, the channel effect of be in agreement with the strong channel effect of fins on
the fins was neglected, and the flows over the finned tube the flow at high Reynolds numbers. Elmahdy and Biggs [5]
surface and over the bank of bare tubes were assumed to reported slopes ranging from - 0 . 3 6 to - 0 . 3 0 for several
be similar. Then, for Reynolds numbers in the range geometrically different exchangers. The slope discrepancy
100-4000, the exponent m - 1 of Eq. (8) was - 0 . 4 [14]. may be due to the distinct definition of characteristic
Because of the presence of fins, the flow along the fiat length in their work.
plate is superimposed on the flow around the tubes. The present data for coil types 8 and 4 are compared
Especially at high Reynolds numbers, the fin effect be- with the correlation stated by Webb [16, eq. (5)] in Fig. 11.
comes stronger. Hence, the exponent m - 1 should as- Starting from the first distribution for e = 11.28, which

161 I I I I i i II I I I I i I tll I I I i d_

j.E°36z : 0.15 Re -°'2a

Coil
Ty p • Symbol v ~ ~ I ~ ~l O J '(::3 ~
1 0 -- ~ ~lrAIk=l.~_ o
t 362 OO *0
2
-- 3
4 "o ° ° °°" o - "
5 GI
6
? A
8
9 V
10 D

1 0-3 I I i i I I I I I I I I I I I III I i I
102 103 104 5.10 4
G m do
J.t b

Figure 10. Average convective heat transfer as a function of Reynolds number, 11.2 _< • < 23.5.
270 N. Kayansayan

-1
10 t I l I ~ ,r] I [ [ I I l I i t] I I [ I
• McQuist on [121

• This s t u d y (Coil type:8]

Q This study [Coil type:4)

--- Webb [16.1
~This study [ecl.(13)]

,d
"•---••][.,t]t. E : 11.28

163 "'''-''--. E = 23.53

i(~3102 i , , , , J ,i'r~Iv , J L, ~ ,I,11n41v ' ' ' ' ' ''

105
IRe ,~
Figure 11. Comparison of present correlation with Ref. 16 for the data for coil types 8, 4, and McQuiston [12].

also contains the data of McOuiston [12], it should be displays a maximum of 31% error to the data for coil
noted that Webb's correlation represents the results with type 4.
reasonable accuracy at low Reynolds numbers. As the
Reynolds number increases, however, the correlation line PRACTICAL S I G N I F I C A N C E
diverges from the data points. The percentage of error in
the j-factor representation is defined as The results of this study represent the first phase of a
research program motivated by the need to develop an
Janalytical -- Jexperimental improved understanding and characterization of forced
E = × 100 (14) convection heat transfer on compact plate-fin heat ex-
Janalytical changer surfaces. In addition to covering only restricted
variations in the exchanger surface geometry, existing
where Janalytical is the j factor calculated by using any of correlations in the engineering literature are applicable to
the stated correlations. Then, typically at Re = 12,000, only a limited range of Reynolds numbers. However, heat
Webb's correlation is found to deviate by 29.5% while Eq. exchanger designers and analyts require a correlation with
(13) deviates by 8.2%. In the second distribution for reasonable accuracy validated for a wide range of Reynolds
• = 23.53, the geometric ratios for coil type 4 are S l / d o = numbers and for diversified geometrical conditions. The
3.02 and s 2 / d o = 2.62 and exceed the range of validity of present study aims to fulfill this requirement. The ex-
Webb's correlation. As shown in Fig. 11, Webb's correla- changer finning factor obtained by means of Eq. (A10) can
tion exhibits a large discrepancy with the present data. It be applied in Eq. (13) to predict the performance charac-
is quite difficult to interpret this particular manifestation. teristics of untested but geometrically similar heat ex-
However, as noted by Webb [16], the small influence of fin changers, provided they are operated in the Reynolds
spacing, especially at high flow rates, is probably mislead- number range 500-30,000.
ing. Considering a limiting case for which s J d o and
s 2 / d o are assumed to have large values and s / d o .~ 1, CONCLUSIONS
then it would not be appropriate to neglect the channel
effect of fins on the flow, and disregarding this effect may In the experiments, the geometrical parameters of the 10
lead to higher j-factors. In Fig. 11, at Re = 9000, Eq. (13) tested coils were varied in the ranges of 2.39 < s l / d o <
 Flat Plain Fins and Round Tube Exchangers 271

3.15, 2.07 < s 2 / d o < 2.67, and 0.131 < s / d o < 0.425. The The finned area per unit length and the tube outside area
Reynolds number presentation is based on the tube out- with collar per unit length are given by
side diameter including the collar thickness. As described
in the Appendix, all the geometrical properties are em- A'
~rd ] st S2 -
liSP
bodied in a single parameter, the finning factor e. The
performance of a flat fin and round tube heat exchanger is and
best expressed in terms of a Colburn j factor, and a
relation between this and the Reynolds number, the
Ato = N m r d o ( 1 - tSF) (AS)
finning factor, is then sought. Then the total outside surface area per unit length is
A strong dependence of the heat transfer coefficients do
on the finning factor • is noted. As the value of e
increases, the general behavior of the exchanger, as ex-
Ao=Nnqrdo((l-tSF)+--~[41Sl
,.o ,,.o ) -1].)
I ( s2

pected, is that the j factor decreases. The fin density s F is (Ag)

a major parameter in the representation of ~ (Eq. (A10)),
and the more dense the fins are, the more pronounced the Together with Eqs. (A9), (A8), and (A6), the definition
channel effect is. of the finning factor ~ yields
Although Eq. (13) represents the data points with a _ )I-.
correlation coefficient of 0.93, care should be exercised in sI ~ 2 # ) (AI0)
using the results. Of all the data in Fig. 10, 71.8% lie
within a + 10% dispersion band around the mean line. Furthermore, the total inside area of the coil tubes per
Out of 110 experimental data, however, five data points unit length is
are found to scatter deviations of + 30% and 14 by - 30% A i = Nmrd i (All)
which also indicate the upper and lower limits of error for Having thus described the characteristic geometry of
Eq. (13). the heat exchanger, it is apparent that the hydraulic
diameter is defined as
The work reported here is part of a research project sponsored by
Turbo-Therm Heat Exchangers Manufacturing Corporation. Their dh = 4trAfrL/A o (A12)
financial and technical support is gratefully acknowledged. In terms of the geometrical properties of the exchanger,
the following relation for the hydraulic diameter can be
APPENDIX. HEAT EXCHANGER GEOMETRY derived:

In order to relate the finning factor to the geometry of the 8& (sl/do)(s2/do)
heat exchanger, it is necessary to consider the following
definitions for the flat-fin, crossflow, staggered-tube ex-
d h ~- _ _
7]- 2
changer. Referring to Fig. 2, denoting the tube outside ~oo(I-tSv) + - 1 sF
diameter by dto and the collar thickness by tc, then the
(A13)
characteristic dimension becomes

d o = dto + 2t C (A1)
NOMENCLATURE
The relation between the number of tubes per row, n, the A surface area, m 2
transverse pitch Sl, and the exchanger height B is a segmental area of wind tunnel cross section, m 2
B exchanger height, m
B = ns I (A2)
Cp specific heat, kJ/(kg. °C)
Similarly, the number of tube rows, N, the longitudinal d diameter, m
tube pitch s 2, and the flow length L are related as E percentage of error [Eq. (14)], dimensionless
F correction to logarithmic temperature difference,
L = Ns 2 (A3) dimensionless
G m mass flux, kg/(m 2- s)
Then the minimum flow area per unit length and the H air-side enthalpy, W
exchanger frontal area per unit length are h heat transfer coefficient, W / ( m 2 • °C)
j Colburn j factor [= (ho/GmCp)pr2/3], d i m e n s i o n l e s s
Amin = n ( s 1 - do)(1 - tSF) (A4)
k thermal conductivity, W / ( m . °C)
and L flow length, m
M mass flow rate, kg/s
Afr = nsl4s (AS)
m Reynolds number exponent [Eq. (4)], dimensionless
The parameter s F is termed the fin density and has units N number of tube rows, dimensionless
of fins per unit length. The ratio of Eqs. (A4) and (A5) n number of tubes per row, dimensionless
becomes Nu Nusselt number ( = h o d o / k b ) , dimensionless
Pr Prandtl number ( = Id.bCp/kb), dimensionless
tr = (1 - d o / s x ) ( 1 - tSF) (A6) Q heat transfer rate, W
 272 N. Kayansayan

Re R e y n o l d s n u m b e r ( = Gmdo/tZb), d i m e n s i o n l e s s 3. Rich, D. G., The Effect of Fin Spacing on the Itcat Transfer and
s fin spacing, m Friction Performance of Multi-row, Smooth Plate Fin-and-Tube
s1 transverse tube pitch, m Heat Exchangers, ASHRAE Trans., 79(2), 137 145, 1973.
4. Rich, D. G., The Effect of the Number of Tube Rows on tteat
s2 longitudinal tube pitch, m
Transfer Performance of Smooth Plate Fin-and-Tube Heat Ex-
sF fin density, f i n s / m changers, ASHRAE Trans., 81(1), 307-317, 1975.
T t e m p e r a t u r e , °C 5. Elmahdy, A. H., and Biggs, R. C., Finned Tube Heat Exchanger:
AT,~ l o g a r i t h m i c m e a n t e m p e r a t u r e difference, °C Correlation of Dry Surface Heat Transfer Data, ASHRAE Trans.,
t fin thickness 85(2), 262 273, 1979.
6. McQuiston, F. C., Correlations of Heat, Mass and Momentum
tC collar thickness, m
Transport Coefficients for Plate-Fin-Tube Heat Transfer Sur-
U overall h e a t transfer coefficient, W / ( m 2. °C) faces with Staggered Tubes, ASHRAE Trans., 84(1), 294-309,
V velocity, m / s 1978.
Greek Symbols 7. Kays, W. M., and London, A. L., Heat Transfer and Flow
Friction Characteristics of Some Compact Heat Exchanger Sur-
difference
faces: Part 1. Test System and Procedure, Trans. ASME, 72,
e e x c h a n g e r finning factor [Eq. (6)], d i m e n s i o n l e s s 1075-1085, 1950.
r/ efficiency 8. British Standards, BS 1042, part 2A, 1973, pp. 38-59.
/z d y n a m i c viscosity, k g / ( m - s) 9. Kline, S. J., and McClintock, F. A., Describing Uncertainties in
p density, k g / m 3 Single-Sample Experiments, Mech. Eng., 75, 3-8, 1953.
10. Colburn, A. P., A Method of Correlating Forced Convection
cr m i n i m u m to frontal a r e a ratio, [Eq. (A13)],
Heat Transfer Data and a Comparison with Fluid Friction, Trans.
dimensionless
AIChE, 29, 174-210, 1933.
Subscripts 11. VDI-W~irmeatlas, Berechnungsbliitter fi~r den Wiirmeiibergang,
b bulk VDI-Verlag, 1963.
e exit 12. McQuiston, F. C., Heat, Mass, and Momentum Transfer Data for
f fin Five Plate-Fin-Tube Heat Transfer Surfaces, ASHRAE Trans.,
84(1), 266-293, 1978.
fr frontal
13. McQuiston, F. C., and Parker, J. D., Heating, Ventilating, andAir
h hydraulic Conditioning--Analysis and Design, 3rd ed., Wiley, New York, pp.
i inside 555-562, 1988.
in inlet 14. Incropera, F. P., and Dewitt, D. P., Fundamentals of Heat and
j, k m e a s u r e m e n t points Mass Transfer, 2nd ed., Wiley, New York, pp. 309-404, 1985.
min minimum 15. Webb, R. L., Enhancement of Single-Phase Heat Transfer, in
Handbook of Single-Phase Convective Heat Transfer, S. Kaka~, R.
o outside K. Shah, and W. Aung, Eds., Wiley, New York, Chapter 17, pp.
to tube o u t s i d e 17.16-17.17, 1987.
16. Gray, D. L., and Webb, R. L., Heat Transfer and Friction
REFERENCES Correlations for Plate Finned-Tube Heat Exchangers Having
Plain Fins, Heat Transfer 1986, 6, 2745-2750, 1986.
1. Webb, R. L., Air-Side Heat Transfer in Finned Tube Heat
Exchangers, Heat Transfer Eng., 1(3), 33-49, 1980.
2. McQuiston, F. C., Finned Tube Heat Exchangers: State of the
Art for the Air Side, ASHRAE Trans., 87, 1077-1085, 1981. Received May 20, 1992; revised October 30, 1992