A study of the air-side heat transfer and pressure drop characteristics

of tube-n no-frost evaporators

Jader R. Barbosa Jr.
*
, Cludio Melo, Christian J.L. Hermes, Paulo J. Waltrich
POLO National Institute of Science and Technology of Refrigeration and Thermophysics, Federal University of Santa Catarina, 88040-900, Florianopolis, SC, Brazil
a r t i c l e i n f o
Article history:
Received 21 May 2008
Received in revised form 22 November 2008
Accepted 25 November 2008
Available online 31 December 2008
Keywords:
Tube-n evaporator
Experimental analysis
Correlation
Frost-free refrigerator
a b s t r a c t
A study is presented on the inuence of the air ow rate and surface geometry on the thermal-hydraulic
performance of commercial tube-n no-frost evaporators. A specially constructed wind-tunnel calorim-
eter was used in the experiments from which data on the overall thermal conductance, pressure drop,
Colburn j-factor and Darcy friction factor, f, were extracted. Eight different evaporator samples with dis-
tinct geometric characteristics, such as number of tube rows, number of ns and n pitch were tested.
Semi-empirical correlations for j and f are proposed in terms of the air-side Reynolds number and the n-
ning factor. A discussion is presented on the performance of the evaporators with respect to specic cri-
teria such as the pumping power as a function of heat transfer capacity and the volume of material in
each evaporator.
2008 Elsevier Ltd. All rights reserved.
1. Introduction
Compartment cooling in the so-called no-frost refrigerators
relies on external forced convection heat transfer to a tube-n
evaporator. Despite the abundant literature on air-side thermal-
hydraulic performance of tube-n heat exchangers, there is a lack
of specic experimental data for the class of evaporators used in
no-frost household appliances. In relation to more conventional
tube-n heat exchangers, there are several distinguishing aspects
of no-frost evaporators which make them unique. Firstly, the
geometry of no-frost evaporators is such that the frontal (or face)
area is smaller and the evaporator length is larger than more con-
ventional heat exchanger geometries. Thus, in no-frost evapora-
tors, the number of tube rows through which the air ows may
be considerably larger (see Fig. 1). In addition, to avoid ow
obstruction due to frost formation on the air-side heat transfer sur-
face, n spacing may be non-uniform along the coil and is signi-
cantly larger ($0.40.8 ns per cm) than those found in
conventional tube-n exchangers ($510 ns per cm). Electric
heaters are also utilized to eliminate the frost build-up on the heat
exchanger surface. The most usual heaters are aluminiumsheathed
coils (5 mm OD) mounted on the outer edge of the ns, parallel to
the tubes. During normal operation of the refrigerator, at pre-
determined time intervals, the heater is switched on to melt the
frost accumulated on the surface. Another feature of no-frost
evaporators is that the range of air ow rates ($50 m
3
/h) tends
to be lower than in conventional tube-n heat exchangers (10
2

10
3
m
3
/h).
The literature on the heat transfer and pressure drop behaviour
of plain nned-tube heat exchangers is profuse. Detailed reviews of
the experimental studies, data reduction procedures and correla-
tions methods have been presented recently by Wang et al. [1], Ja-
cobi et al. [2] and Webb and Kim [3]. Nevertheless, despite the
substantial amount of information on hand, this is somewhat re-
stricted to values of n pitch lower than 5.0 mm and numbers of
tube rows less than six. Wang et al. [4,5] developed one of the most
complete and accurate sets of correlations for the air side heat
transfer and pressure drop of plain nned-tube heat exchangers.
They have used data from 74 samples from eight different sources
evaluated over a wide range of ow conditions. However, only
around 3% of the data points have been collected for values of n
pitch larger than 5.0 mm. Therefore, there is a justied need for
more experimental data and specic correlations for the air side
heat transfer and pressure drop under conditions of large n pitch
and large number of tube rows in the direction of the ow.
As far as the research on household refrigeration is concerned,
Karatas et al. [6] carried out an experimental study of the air-side
heat transfer and pressure drop in no-frost evaporators. They
tested four evaporators and assessed the effect of non-uniformities
in the temperature and velocity distributions of the inlet air ow.
Although the number of ns (and hence the n pitch and the n-
ning factor, e) was different for each evaporator, their basic charac-
teristics (tube pitches, number of tube rows, number of tubes per
row and face dimensions) were kept xed. Correlations were
proposed for the Colburn j-factor and for the friction factor as a
0306-2619/$ - see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apenergy.2008.11.027
* Corresponding author. Tel.: +55 48 3234 5691; fax: +55 48 3234 5166.
E-mail address: jrb@polo.ufsc.br (J.R. Barbosa).
Applied Energy 86 (2009) 14841491
Contents lists available at ScienceDirect
Applied Energy
j our nal homepage: www. el sevi er. com/ l ocat e/ apenergy
function of the air-side Reynolds number (based on the outside
tube diameter), the Prandtl number and the nning factor as
follows:
j 0:138Re
0:281
do
e
0:407
; 300 6 Re
do
6 1000 and 1 6 e 6 6; 1
f 0:152Re
0:164
do
e
0:331
300 6 Re
do
6 1000 and 3:5 6 e 6 6: 2
They concluded that the heat transfer correlation was equally
valid for the non-uniform ow cases (typical operating condition
of Combi refrigerators, where air is drawn from both the freezer
and fresh-food compartments) if mass ow averaged values of
temperature and velocity were used at the evaporator inlet.
Lee et al. [7] investigated experimentally the behaviour of the
air-side heat transfer coefcient for the same evaporator congura-
tion, but with three different n geometries (discrete at plate ns,
continuous at plate ns and spine ns). Despite the lower evapo-
rator length and smaller heat transfer area, the spine nned tube
evaporator exhibited the best thermal-hydraulic performance un-
der dry conditions (no condensate or frost formation). The authors
put forward empirical heat transfer correlations for the three evap-
orators as follows (where the subscripts 1, 2 and 3 stand for the
discrete at plate, continuous at plate and spine n heat exchang-
ers, respectively),
j
1
0:170Re
0:37
do
; 3
j
2
0:162Re
0:39
do
; 4
j
3
0:148Re
0:33
do
: 5
The range of validity of the above correlations was not indicated
and detailed pressure drop data were not provided for the three
evaporator types investigated by Lee et al. [7]. Nevertheless, it
was mentioned that the pressure drop of the spine nned evapora-
tor was signicantly lower than those of the discrete and continu-
ous at plate n evaporators.
Kim and Kim [8,9] presented experimental data on the air side
heat transfer characteristics of at plate nned tube heat exchang-
Nomenclature
A
f
face area (D W) (m
2
)
A
i
inside tube area (m
2
)
A
min
minimum free ow area (m
2
)
A
o
outside area (ns and tubes) (m
2
)
A
to
outside area (tubes only) (m
2
)
c
P
specic heat capacity (J/kg K)
D heat exchanger height (m)
d
h
hydraulic diameter (m)
d
i
inside tube diameter (m)
d
o
outside tube diameter (m)
f friction factor ()
F pure counterow correction factor ()
F
P
n pitch (m)
G
a,max
maximum air mass ux (kg/m
2
s)
h heat transfer coefcient (W/m
2
K)
j Colburn j-factor ()
k thermal conductivity (W/m K)
L heat exchanger length (m)
_ m mass ow rate (kg/s)
N number of tube rows ()
N
P
number of data points ()
P
l
longitudinal tube pitch (m)
P
t
transverse tube pitch (m)
Pr Prandtl number ()
_
Q heat transfer rate (W)
R tube radius (m)
Re Reynolds number ()
t thickness (m)
T temperature (C)
UA overall thermal conductance (W/C)
W heat exchanger width (m)
Greek symbols
Dp
a
air-side pressure drop (Pa)
e nning factor (= A
o
/A
to
) ()
g
n
n efciency ()
g
o
surface efciency ()
q density (kg/m
3
)
l viscosity (kg/m s)
Subscripts
a air
f face
n relative to the ns
i inside
in inlet
o outer
out outside
t tube
w wall
Fig. 1. Evaporator geometry: (a) a structural n showing the grooves where the defrost heater coil is mounted and (b) principal dimensions and n distribution.
J.R. Barbosa et al. / Applied Energy 86 (2009) 14841491 1485
ers with values of n pitch between 7.5 and 15 mm. Heat exchan-
ger samples with inline and staggered tube arrangements were
evaluated and the number of tube rows were varied between 1
and 4. They concluded that the heat transfer coefcient increases
with an increase of the n pitch and is inversely proportional to
the number of tube rows. They proposed a correlation for the Col-
burn j-factor as follows:
j 0:170N
0:141
F
p
d
o
_ _
0:384
Re
0:349
d
h
600 6 Re
d
h
6 2000: 6
More recently, Melo et al. [10] carried out in situ tests in a real
refrigerator maintaining all of the original characteristics of the air
distribution system. Three evaporators with nearly identical geo-
metrical characteristics, but with distinct ow arrangements (par-
allel-ow, counter-ow and standard, i.e., 2-pass), were evaluated.
As expected, the ow arrangement did not show any signicant ef-
fect on heat transfer performance for refrigerant outlet superheat
lower than 5 C. However, the counter-ow evaporator exhibited
the highest performance at 10 C superheat.
It is important to note that in their studies Karatas et al. [6] and
Lee et al. [7] kept the number of tube rows xed (13 and 10 rows,
respectively), thus neglecting the importance of such design
parameter in the heat exchanger performance. The assessment of
the effect of the number of tube rows on the evaporator thermal-
hydraulic performance is, therefore, the main focus of this study,
where a systematic investigation of the inuence of several geo-
metric parameters (e.g., number of tube rows, n pitch, number
of ns) and air ow rate on the air-side thermal-hydraulic perfor-
mance of tube-n no-frost evaporators under dry conditions (i.e.,
no humidity condensate nor frost formation) is reported. Eight
evaporator samples were evaluated and the variation of geometric
parameters between the samples resulted in distinct values of
evaporator length, heat transfer surface area and volume of mate-
rial (mass of aluminium). A purpose-built wind-tunnel calorimeter
was used in the experiments from which overall thermal conduc-
tance, pressure drop, Colburn j-factor and Darcy friction factor, f,
data were extracted. The experimental data expressed in terms
of j and f were reproduced by empirical correlations within 7%
error bands.
2. Experimental work
2.1. Facility and equipment
The wind tunnel was designed according to ANSI/ASHRAE Stan-
dards 37 [11], 41.2 [12] and 51 [13]. The facility was constructed
from a double layer of ordinary steel plates. In between the plates,
a 100 mm thick layer of glass wool was inserted to provide thermal
insulation. The test section dimensions are illustrated in Fig. 2.
Screens are employed to make the ow uniform in the inlet and
exit sections and also upstream of the air ow nozzles. In this re-
gion, the height of the test section was increased to accommodate
the nozzle array used for measuring the air ow rate.
The following components comprise the wind tunnel air-side
instrumentation: a 51-W speed controlled fan, a 400-W (max.)
PID controlled electrical heater for air inlet temperature setting, a
set of ve aluminium nozzles with diameters ranging from
19.05 mm (0.75
00
) to 31.75 mm (1.25
00
), and two differential pres-
sure transducers to measure the air pressure drop across the evap-
orator and the nozzles. The evaporator pressure drop is measured
by mounting perforated hoses (spacing between adjacent holes of
50 mm) on two grooves machined on the bottom wall of the test
section (one upstream and the other downstream of the evapora-
tor). The grooves are perpendicular to the main ow direction
and their depth is such that the pressure taps are at the same level
as the bottom wall. The adjoining surfaces are leveled with silicone
glue to avoid disturbing the ow in the vicinity of the pressure
taps. One end of each hose is connected to the differential pressure
transducer while the other end is sealed. The accuracy of each
pressure transducer is 0.5% of the full scale ($25 Pa for the evap-
orator transducer and $995 Pa for the nozzle transducer). Opera-
tion limits and design conditions are as follows; air ow rate:
min 17 m
3
/h (10 cfm), max 102 m
3
/h (60 cfm), nominal 51 m
3
/h
(30 cfm); and heat transfer rate: min 40 W, max 200 W, nominal
120 W. The maximum evaporator dimensions for the test section
are: length (L) 250 mm, width (W) 580 mm, height (D) 80 mm.
The main function of the water loop is to circulate water at con-
trolled temperatures and ow rates through the evaporator. The
following components make up the water circuit: a 1.58 L/min
(max.) speed-controlled rotary pump; a 100 C (max.), 0.1 C
accuracy, thermostatic bath; and a 2.46 L/min (max.), 1.4% full
scale accuracy, turbine ow meter. The loop is thermally insulated
and T-type immersion thermocouples (0.2 C) are placed immedi-
ately upstream and downstream of the evaporator. Data acquisi-
tion is performed with a PC integrated 40 channel system. This
system, in conjunction with a purpose-built control panel, moni-
tors and records pressure, temperature, relative humidity and
water ow rate signals.
2.2. Procedure
The apparatus is switched on, the inlet water temperature
(approximately 32 C in all cases) is set and approximately
10 min are required for it to stabilize. The desired air ow rate is
adjusted and the inlet air temperature is set (approximately
28 C in all cases). The water ow rate is set to provide roughly a
0.5 C temperature variation between the inlet and outlet the ex-
changer. Approximately 5080 min depending on the ow rates
are required to reach steady-state. The steady-state criterion has
been established as follows: each parameter is averaged over a
30 min interval and the standard deviation associated with its sig-
nal is calculated. If the difference between the reading at the begin-
ning of the sampling interval (t = 0) and at the end (t = 30 min) is
less than three times the standard deviation, the test is considered
stable and the average values of each experimental parameter
(temperature, pressure, etc.) are stored. After data collection for
one experimental condition, the ow rates are altered so that a
new experimental condition is achieved.
2.3. Evaporator samples
Eight sample evaporators made from aluminium (both n and
tubes) were tested. The inner and outer diameters of the tubes
in all evaporators were 6.6 mm and 7.9 mm, respectively. The
number of tubes per row is two in all exchangers. A staggered tube
array is used in all samples and the transverse and longitudinal
tube pitches, P
t
and P
l
, are 23 mm and 22 mm, respectively. The
width, W, and height, D, in all evaporators are 340 mm and
59 mm. The geometry of the discrete at ns is such that their
length (in the direction of the ow) and height are xed at
35 mm and 59 mm and they are mounted so as to span two con-
secutive tube rows (for example, in an evaporator with N tube
rows there are N/2 n rows). The n thickness is 0.127 mm. As
shown in Fig. 1, the evaporator coil is assembled on two 1 mm
thick structural ns of length L, which also provide support for
mounting the electrical resistances for defrosting. The coil length,
LL
b
, is dened as the distance between the tips of the ns at the
inlet and outlet sections (L
b
= 55 mm). The number of tube rows,
the evaporator length, the number of ns per meter per row, the
surface area of the ns and of the tubes, the nning factor, e,
and the amount of material (evaporator mass) may vary between
the samples and their values are summarized in Table 1. The tests
1486 J.R. Barbosa et al. / Applied Energy 86 (2009) 14841491
were conducted with the defrosting electrical resistances mounted
on the heat exchangers.
3. Data reduction
3.1. Heat transfer
The heat transfer rate from the exchanger is taken as the arith-
metic mean of the air and water loops heat transfer rates (Eq. (7)).
It is worth noting that in all experimental runs, the difference be-
tween
_
Q
a
and
_
Q
w
was within 8% for all runs.
_
Q
1
2

_
Q
a

_
Q
w

1
2
_ m
a
c
P;a
T
a;out
T
a;in
_ m
w
c
P;w
T
w;in
T
w;out

7
The overall thermal conductance is calculated using the log-
mean temperature difference approach [14], as follows:
UA
_
Q
FDT
lm

_
Q lnT
w;out
T
a;in
=T
w;in
T
a;out

T
w;out
T
a;in
T
w;in
T
a;out

8
where F was assumed equal to unity because of the high water ow
rate which is set to provide a negligible water temperature drop
across the exchanger ($0.5 C). The air-side thermal resistance,
1250 400 300 500
SCREEN
Dimensions in [mm]
8
0
0
1
0
0
TEMPERATURE
MEASUREMENT
EVAPORATOR
200 400 400 300 500 100
SCREEN
TEMPERATURE
MEASUREMENT
SCREEN
WATER BATH
DIFFERENTIAL
MANOMETER
FLOWMETER
4350
NOZZLE
Fig. 2. A schematic view of the section of the wind-tunnel.
Table 1
Geometric parameters of evaporator samples.
Evaporator sample No. of tube rows Coil length (LL
b
) (mm) No. of ns Area (m
2
) Finning factor (e) Fin density (cm
1
) Mass (g)
#1 4 74 61 Fins: 0.20 3.86 1st n row: 0.79 380.6
Tube: 0.07 2nd Fin row: 1.00
Total: 0.27
#2 6 112 128 Fins: 0.41 5.10 1st Fin row: 0.79 496.4
Tube: 0.10 2nd Fin row: 1.00
Total: 0.51 3rd Fin row: 1.97
#3 8 151 194 Fins: 0.61 5.36 1st Fin row: 0.79 607.1
Tube: 0.14 2nd Fin row: 1.00
Total: 0.75 3rd Fin row: 1.97
4th Fin row: 1.94
#4 10 189 129 3.47 1st Fin row: 0.79 677.3
Fins: 0.42 2nd n row: 1.00
Tube: 0.17 3rd Fin row: 1.00
Total: 0.59 4th Fin row: 0.50
5th Fin row: 0.50
#5 8 151 95 Fins: 0.31 3.93 1st Fin row: 0.41 560.0
Tube: 0.14 2nd Fin row: 0.50
Total: 0.55 3rd Fin row: 0.91
4th Fin row: 0.97
#6 10 189 261 5.82 1st n row: 0.79 776.6
Fins: 0.82 2nd Fin row: 1.00
Tube: 0.17 3rd Fin row: 1.97
Total: 0.99 4th Fin row: 1.94
5th n row: 1.97
#7 4 74 31 Fins: 0.11 2.57 1st n row: 0.41 350.3
Tube: 0.07 2nd Fin row: 0.50
Total: 0.18
#8 6 112 65 Fins: 0.22 3.20 1st Fin row: 0.41 503.3
Tube: 0.10 2nd Fin row: 0.50
Total: 0.32 3rd Fin row: 1.00
J.R. Barbosa et al. / Applied Energy 86 (2009) 14841491 1487
g
o
h
o

1
, is calculated neglecting the thermal resistance of the tube
wall
g
o
h
o

A
o
UA
_ _

A
o
A
i
h
i
_ _ _ _
1
9
where the water-side heat transfer coefcient, h
i
, was estimated
using the Gnielinski correlation given by [14]
h
i
d
i
k
i

f
i
=8Re
d
i
1000Pr
i
1 12:7

f
i
=8
_
Pr
2=3
i
1
10
where
f
i
1:82log
10
Re
d
i
1:64
2
11
Because of the way in which the ns are designed and mounted
on the tube coil the coil is made from a single tube bent into
shape and the ns are mounted from the sides of the coil and slid
into place with a certain degree of mechanical interference the
contact line between the ns and the perimeter of the tubes is
not uniform (see Fig. 1a). This complicates substantially the calcu-
lation of the surface efciency and hinders the utilization of stan-
dard methods such as that by Schmidt [15], which assumes radial
symmetry. Karatas et al. [6,16,17] circumvented this problem by
calculating the n efciency of a no-frost evaporator via a nite
element analysis. However, their results are specic for the n
geometry and tube pitches employed in their study. In order to en-
sure that the present data are compared with correlations available
in the literature without being inuenced by the n efciency cor-
relation utilized in each study, the Colburn j-factor has been de-
ned based on the air-side thermal resistance as follows:
j
g
o
h
o
G
a;max
c
P;a
Pr
2=3
a
; 12
where
G
a;max
_ m
a
=A
min
; 13
A
min
DW N
fin
Dt
fin
2d
o
W 2d
o
t
fin
: 14
3.2. Pressure drop
Based on the measured air-side pressure drop, the friction fac-
tor is calculated as follows [18]:
f
A
min
A
o
q
a
q
a;in
2Dp
a
q
a;in
G
2
a;max

q
a;in
q
a;out
1
_ _
1
A
2
min
A
2
f
_ _ _ _
; 15
where q
a
is the average air density evaluated at the average tem-
perature between the inlet and outlet.
A series of repeatability tests was performed and it was found
that pressure drop, friction factors, overall thermal conductance
and j-factors are reproducible within the uncertainty levels corre-
sponding to 5%, 8%, 9% and 11%, respectively.
4. Results and discussion
The air-side pressure drop as a function of air ow rate is pre-
sented in Fig. 3. The expected behaviour of increasing pressure
drop with increasing ow rate is observed. The effect of n number
on pressure drop is quite pronounced but, as will be seen, this is
less than the effect of n number on thermal conductance. For
example, evaporator #4 exhibits a somewhat higher pressure drop
than evaporator #3, even though the former has fewer ns (and
also a lower total heat transfer surface). In this case, the reduction
in pressure drop due to the lower number of ns in sample #4 is
offset by an increase in pressure drop due to the greater number
of tube rows in sample #4 compared to sample #3. The more dis-
tinct inuence of the number of tube rows on pressure drop can
also be observed by comparing the pressure drop curves for sam-
ples #2 and #4, which have almost the same number of ns, but
a different number of tube rows.
Fig. 4 presents the overall thermal conductance of the eight
evaporator samples as a function of the air ow pumping power,
given by the product of the air volume ow rate and the air-side
pressure drop. As expected, UA increases with air ow rate and
pumping power in all cases. The total number of ns seems to be
the most important parameter controlling the magnitude of UA,
since the lowest conductance at a give ow rate is that of sample
#7 and this increases progressively as the total number of ns,
and hence the heat transfer surface area increases (sample #6
5 10 15 20 25 30 35
AIRFLOW RATE [L/s]
0
4
8
12
16
20
24
A
I
R
-
S
I
D
E

P
R
E
S
S
U
R
E

D
R
O
P

[
P
a
]
#1
#2
#3
#4
#5
#6
#7
#8
Fig. 3. Air-side pressure drop.
0 200 400 600 800
PUMPING POWER [mW]
0
10
20
30
40
O
V
E
R
A
L
L

T
H
E
R
M
A
L

C
O
N
D
U
C
T
A
N
C
E

(
U
A
)

[
W
/
K
]
#1
#2
#3
#4
#5
#6
#7
#8
Fig. 4. Overall thermal conductance.
1488 J.R. Barbosa et al. / Applied Energy 86 (2009) 14841491
exhibits the highest conductance at all ow rates). The number of
tube rows also contributes for an increase in UA, but the increase in
surface area due to a larger number of tube rows is, in some cases,
overcome by that associated with more ns distributed over a low-
er number of tube rows (as in the cases of evaporators #2 and #5).
Curves of Colburn j-factor and friction factor as a function of the
air-side Reynolds number are presented in Figs. 5 and 6, respec-
tively. The air-side Reynolds number is calculated based on the
maximum air mass ux through the evaporator
Re
a

d
o
G
a;max
l
a
; 16
where l
a
is the average air viscosity evaluated at the average tem-
perature between the inlet and outlet. The expected behaviour of
decreasing j and f with Reynolds number is observed. The shift be-
tween the curves characterizes the effect of n number, n distribu-
tion and evaporator length on heat transfer and pressure drop.
Empirical correlations for the j-factor and for the friction factor
were devised based on the whole experimental data set and are
presented below. These include the air-side Reynolds number,
the nning factor and the number of n rows as follows:
j 0:6976 Re
0:4842
a
e
0:3426
; 17
f 5:965 Re
0:2948
a
e
0:7671
N=2
0:4436
: 18
The form of the correlations for j and f reects the experimental
observations that the number of tube rows (and hence of n rows)
exerts a stronger inuence on pressure drop than on the overall
heat conductance [19]. For this reason, this parameter was consid-
ered only in the f correlation. The constants were determined
through minimization of the RMS error given by Eq. (19) using a
Quasi-Newton method.
RMS

1
N
P

Np
i1
g
cal;i
g
exp;i
g
exp;i
_ _
2

_
19
where g can be either j or f. Figs. 7 and 8 show comparisons between
calculated and experimental j-factor and friction factors, respec-
tively. As can be seen, the models can predict the experimental data
within 7% error bands. The correlations are valid for 320 6 Re
a
6
1200, 2.6 6 e 6 5.8 and 2 6 (N/2) 6 5.
Predictions of existing heat transfer and friction factor correla-
tions for large n pitch tube-n heat exchangers are also shown
in Figs. 7 and 8. The correlations of Karatas et al. [6], Lee et al.
[7] and Kim and Kim [8] under predict the Colburn j-factor by as
much as 40%. In order that the methodologies are compared on a
common basis, the inuence of the surface efciency employed
in each correlation has been eliminated by dening the Colburn
j-factor as in Eq. (12). Therefore, the calculated Colburn j-factors
in Fig. 7 are the products of Eqs. 7, 9, 12 and the surface efciencies
1000
2000 900 800 700 600 500 400 300 200
AIR-SIDE REYNOLDS NUMBER [-]
0.01
0.02
0.03
0.04
0.05
0.06
C
O
L
B
U
R
N

j
-
F
A
C
T
O
R

[
-
]
#1
#2
#3
#4
#5
#6
#7
#8
Fig. 5. Colburn j-factor as a function of the Reynolds number.
1000
2000 900 800 700 600 500 400 300 200
AIR-SIDE REYNOLDS NUMBER [-]
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.09
F
R
I
C
T
I
O
N

F
A
C
T
O
R

[
-
]
#1
#2
#3
#4
#5
#6
#7
#8
Fig. 6. Friction factor as a function of the Reynolds number.
0 0.01 0.02 0.30 0.04
EXPERIMENTAL j-FACTOR
0
0.01
0.02
0.03
0.04
C
A
L
C
U
L
A
T
E
D

j
-
F
A
C
T
O
R
+7%
-7%
This work (Eq. 17)
Karatas et al. [6]
Lee et al.[7]
Kim and Kim [8]
Fig. 7. Evaluation of the existing correlations for the Colburn j-factor (Eq. (12)) in
the light of the present experimental data.
J.R. Barbosa et al. / Applied Energy 86 (2009) 14841491 1489
associated with the works of Karatas et al. [6], Lee et al. [7] and Kim
and Kim [8], respectively. The n efciency utilized by Karatas
et al. [6] is given by [16,17]
g
fin
3:429m
4
6:457m
3
4:308m
2
0:736m 0:949 20
where
m

2h
0
=k
fin
t
fin
_
21
where the coefcients were obtained from a least-squares polyno-
mial t of the n temperature and heat transfer distribution calcu-
lated via a nite element method. Lee et al. [7] employed the
classical Schmidt [15] methodology to correlate the n efciency
g
fin
tanh/mr
o
=/mr
o
22
where m is given by Eq. (21) and
/ r
eq
=r
o
11 0:35 ln r
eq
=r
o
23
r
eq
=r
o
1:28X
M
=r
o

X
L
=X
M
0:3
_
Staggered tube array 24
X
M
P
t
=2 25
X
L

P
t
=2
2
P
l
=2
2
_
26
Kim and Kim [8] did not specify in detail which correlation was
used to calculate the n efciency. Therefore, for the sake of com-
pleteness in the present calculation, the n efciency was also esti-
mated from Eqs. (21)(26). In all correlations, the surface efciency
has been calculated from [14]
g
o
A
o
A
to
g
fin
A
fin
27
The friction factor correlation of Karatas et al. [6] severely under
predicts the experimental data. Although it has not been ascer-
tained in their paper, one possible explanation for this behaviour
may be due to the inuence of the defrost heaters, which were
present in all of our heat exchangers and were veried experimen-
tally to be responsible for an increase of the order of 40% in the
pressure drop. Karatas et al. [6] did not mention if the defrost heat-
ers were mounted on the evaporators during their tests. Neverthe-
less, even if it does take into account the effect of the defrost
heaters, the Karatas et al. [6] friction factor correlation provides
values of pressure drop ($0.5 Pa) which are at least one order of
magnitude below those usually found in real household appliances
($5 Pa).
Fig. 9 presents a comparison between the performances of each
evaporator sample. The pumping power as a function of the heat
transfer capacity, as calculated from the empirical correlations, is
plotted for each evaporator. At low heat transfer capacities, all
evaporators seem capable of providing the desired cooling capacity
with at a low pumping power. At high heat transfer capacities, on
the other hand, the pumping power for the small evaporators be-
comes prohibitive.
The above information, together with the fact that the mass of
aluminium in the evaporator decreases signicantly with a de-
crease in the number of tube rows (as can be seen from Table 1)
represents an important aspect to be taken into account in the
appliance design and cost assessment. This matter should, how-
ever, be regarded as an issue to be dealt with in conjunction with
a more thorough analysis considering the mechanism of frost for-
mation and defrosting operations.
5. Conclusions
This paper presented experimental data on the thermal-hydrau-
lic performance of evaporators used in no-frost household appli-
ances. An open wind-tunnel test facility specially designed and
built for testing this type of evaporator was utilized and data on
overall thermal conductance, pressure drop, j and f were collected
for eight evaporators with different geometric characteristics. Cor-
relations for the Colburn j-factor and the friction factor have been
proposed. A performance evaluation of the pumping power as a
function of the heat transfer capacity shows that evaporators with
lower length (i.e., fewer tube rows) and lower surface area perform
equally well as some of their counterparts at low heat transfer
rates, indicating that the last tube rows contribute less effectively
to heat transfer, whilst still exert some inuence on pressure drop.
Moreover, the analysis indicates clearly that cost savings can be
0 0.1 0.2 0.3 0.4
EXPERIMENTAL FRICTION FACTOR
0
0.1
0.2
0.3
0.4
C
A
L
C
U
L
A
T
E
D

F
R
I
C
T
I
O
N

F
A
C
T
O
R
+7%
-7%
This work (Eq. 18)
Karatas et al. [6]
Fig. 8. Evaluation of the existing correlations for the friction factor (Eq. (15)) in the
light of the present experimental data.
40 80 120 160 200 240
HEAT TRANSFER CAPACITY [W]
0.01
0.1
1
10
100
1000
10000
P
U
M
P
I
N
G

P
O
W
E
R

[
W
]
`
#1
#2
#3
#4
#5
#6
#7
#8
Fig. 9. Calculated pumping power as a function of heat transfer capacity.
1490 J.R. Barbosa et al. / Applied Energy 86 (2009) 14841491
achieved by using lighter evaporators with equivalent (in some
cases improved) performance characteristics.
Acknowledgements
The authors are grateful to Messrs. L.G. Pereira, J. Boeng, R.O.
Piucco, B. N. Borges, A. Berwanger, A.G.U. Souza and G.A. Rodrigues
for their valuable support in the experiments and data correlation.
Financial support from Whirlpool S.A. is duly acknowledged.
References
[1] Wang CC, Webb RL, Chi KY. Data reduction for air-side performance of n-and-
tube heat exchangers. Exp Thermal Fluid Sci 2000;21:21826.
[2] Jacobi AM, Park Y, Tafti D, Zhang X, 2001. An assessment of the state of the art,
and potential design improvements, for at-tube heat exchangers in air
conditioning and refrigeration applications phase I. Final report, ARTI-21CR/
20020-01.
[3] Webb RL, Kim NH. Principles of enhanced heat transfer, 2nd ed. NY: Taylor and
Francis.
[4] Wang CC, Chi KY. Heat transfer and friction characteristics of plain n-and-
tube heat exchangers, part I: new experimental data. Int J Heat Mass Trans
2000;43:268191.
[5] Wang CC, Chi KY, Chang CJ. Heat transfer and friction characteristics of plain
n-and-tube heat exchangers, part II: correlation. Int J Heat Mass Trans
2000;43:2693700.
[6] Karatas H, Dirik E, Derbentil T, 1996. An experimental study of air-side heat
transfer and friction factor correlations on domestic refrigerator nned-tube
evaporator coils. In: 8th international refrigeration and air conditioning
conference at Purdue, West Lafayette, IN, July 2528.
[7] Lee TH, Lee JS, Oh SY, Lee MY, Lee KS, 2002. Comparison of air-side heat
transfer coefcients of several types of evaporators of household freezer/
refrigerators. In: 9th international refrigeration and air conditioning
conference at Purdue, West Lafayette, IN, July 1619.
[8] Kim Y, Kim Y. Heat transfer characteristics of at plate nned-tube heat
exchangers with large n pitch. Int J Refrig 2005;28:8518.
[9] Kim Y, Kim Y. Erratum. Int J Refrig 2006;29:336.
[10] Melo C, Piucco RO, Duarte, POO, 2006. In-Situ performance evaluation of No-
Frost evaporators. In: 11th international refrigeration and air conditioning
conference at Purdue, West Lafayette, IN, July 1720, paper R076.
[11] ANSI/ASHRAE 37. Methods of testing for rating unitary air-conditioning and
heat pump equipment, Atlanta, GA; 1988.
[12] ANSI/ASHRAE 41.2, (RA 92). Standard methods for laboratory airow
measurement, Atlanta, GA; 1987.
[13] ANSI/ASHRAE 51. Laboratory methods of testing fans for aerodynamic
performance rating, Atlanta, GA; 1999.
[14] Incropera FP, DeWitt DP, Fundamentals of heat and mass transfer, 3rd ed. NY:
Wiley.
[15] Schmidt TE. Heat transfer calculation for extended surfaces. Refrig Eng
1949:351.
[16] Seker D, Karatas H, Egrican N. Frost formation on n-and-tube heat
exchangers. Part I modeling of frost formation on n-and-tube heat
exchangers. Int J Refrig 2004;27:36774.
[17] Seker D, Karatas H, Egrican N. Frost formation on n-and-tube heat
exchangers. Part II experimental investigation of frost formation on n-
and-tube heat exchangers. Int J Refrig 2004;27:3757.
[18] Kays WM, London AL. Compact heat exchangers. Malabar, FL: Krieger; 1998.
third ed..
[19] Shah RK, Sekulic D. Fundamentals of heat exchanger design. NY: Wiley; 2003.
J.R. Barbosa et al. / Applied Energy 86 (2009) 14841491 1491