Vous êtes sur la page 1sur 13

Applied Thermal Engineering 21 (2001) 535547

www.elsevier.com/locate/apthermeng

Heat transfer enhancement in dimpled tubes


Juin Chen a, Hans M
uller-Steinhagen b, Georey G. Duy a,*
a
Department of Chemical and Materials Engineering, School of Engineering, University of Auckland, Private Bag 92019,
Auckland, New Zealand
b
School of Engineering in the Environment, University of Surrey, Guildford GU2 5XH, UK
Received 28 July 1998; accepted 12 May 2000

Abstract
Heat transfer enhancement was investigated in a coaxial-pipe heat exchanger using dimples as the heat
transfer modication on the inner tube. Tube-side Reynolds numbers were in the range of 7:5  103
5:2  104 for water ow. A constant annular mass ow rate was chosen to obtain the highest possible
Reynolds number of 1:1  104 . Typically, the heating water inlet temperature was 68:1  0:1 C.
All six variants with inward-facing, raised dimples on the inner tube increased the values of heat transfer
coecient signicantly above those for the smooth tube. Heat transfer enhancement ranged from 25% to
137% at constant Reynolds number, and from 15% to 84% at constant pumping power. At a constant
Reynolds number, the relative J factor (ratio of heat transfer coecient to friction factor, relative to
smooth tube values), had values from 0.93 to 1.16, with four dimpled tube congurations having values
larger than unity. Despite the extremely simple design, this outperforms almost all heat transfer en-
hancements recommended in the literature. A correlation based on the results of the present work appears
to be suciently accurate for predicting heat transfer coecients and friction factors for the design of
dimpled-tube heat exchangers. 2001 Elsevier Science Ltd. All rights reserved.

Keywords: Heat transfer enhancement; Augmentation; Dimpled tubes; J factor; Coaxial-pipe heat exchanger

1. Introduction

The development of high-performance thermal systems has stimulated interest in methods to


improve heat transfer. The study of improved heat transfer performance is referred to as heat
transfer augmentation, enhancement, or intensication. A great deal of research has focused on
various augmentation techniques with emphasis on rough surfaces, transverse or spiral ribs,

*
Corresponding author. Tel.: +64-9-3737-999; fax: +64-9-3737-463.
E-mail address: gg.duy@auckland.ac.nz (G.G. Duy).

1359-4311/01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved.
PII: S 1 3 5 9 - 4 3 1 1 ( 0 0 ) 0 0 0 6 7 - 3
536 J. Chen et al. / Applied Thermal Engineering 21 (2001) 535547

Nomenclature

a additional heat transfer surface area, m2


A heat transfer surface area, m2
B constant in Eq. (2), m2 K/W
cp specic heat capacity, J/kg K
C constant in Eq. (3)
d tube diameter, m
e depth of dimple, m
E eectiveness of heat transfer
f friction factor
h heat transfer coecient, W/m2 K
J j-factor
k thermal conductivity, W/m K
L length, m
m exponent of Reynolds number
M mass ow rate, kg/s
N number of longitudinal dimple columns
Nu Nusselt number
Pr Prandtl number
p pitch of dimples, m
Q heat transfer rate, W
Re Reynolds number
T temperature, C
v ow velocity, m/s
W pumping power, W
Dp pressure drop, N/m2
DT logarithmic mean temperature dierence, K
d wall thickness, m
U diameter of dimples, m
Subscripts
1 inlet
2 outlet
a augmented
b bulk
i inside
o outside

transverse grooves, knurling, corrugated and spirally corrugated tubes, straight ns, and spiral
and annular ns. In this investigation, augmented surface has been achieved with dimples stra-
tegically located in a pattern along the tube of a double-pipe heat exchanger with the increased
area on the tube side. Dimpled tubes have been studied by Kalinin et al. [1], dimpled and helical
J. Chen et al. / Applied Thermal Engineering 21 (2001) 535547 537

tubes have been investigated by Giovannini et al. [2] and work on corrugated tubes has been
reported by Marto et al. [3].
Augmented surfaces can create one or more combinations of the following conditions that are
favorable for increasing the heat transfer coecient with a consequent increase in the friction
factor:

1. interruption of the development of the boundary layer and increase of the degree of turbulence,
2. eective heat transfer area increase, and
3. generation of rotating and/or secondary ows.

1.1. Performance evaluation criteria

The primary considerations for assessing the eectiveness of augmented surfaces are economic:
initial development cost, capital cost, operating cost, and maintenance cost. Reliability and safety
are also important. The relationship between the thermal and hydraulic performance must also be
considered.
Major process operational variables include the rate of heat transfer, pumping power, pressure
drop, heat ow rate and uid velocity. Webb [4] proposed a broad range of performance eval-
uation criteria for single-phase ow in tubes to obtain the ``optimum'' surface geometry. Three
performance objectives considered were increased heat duty, reduced surface area, and reduced
pump power.
A xed geometry criterion was used for one-for-one replacement (``retrot'' applications) of
smooth tubes with augmented tubes of equal length to compare the increased heat duty for
constant exchanger ow rate. The pumping power of the augmented tube exchanger would
naturally be greater for the augmented surface tube due to higher friction. Alternatively, the pump
power could be kept constant by reducing the tube-side velocity.
A xed ow area criterion for heat exchangers having constant diameter tubes, e.g. shell and
tube exchangers, was also proposed. For constant pumping power, the tube length and possibly
the ow rate would be reduced. Augmented tubes are used to obtain reduced pumping power with
constant heat duty and ow rate.
In most cases, a heat exchanger is ``sized'' for a specic thermal duty with a specied ow rate.
In these situations, the previously mentioned criteria do not apply. This is accomplished using a
greater number of tubes in parallel or by using the same number of larger diameter tubes. It must
be noted that the preferred size of a specic roughness geometry is dependent on the operational
Reynolds number. As the Reynolds number increases, the preferred roughness size becomes
smaller.

2. Experimental equipment and data reduction

2.1. Flow loop

A schematic diagram of the experimental apparatus is presented in Fig. 1. The apparatus


consists of an annular test section, a plate heat exchanger, two water tanks, mixing chambers,
538 J. Chen et al. / Applied Thermal Engineering 21 (2001) 535547

Fig. 1. Schematic diagram of the test apparatus.

rotameters and pumps. The liquid ows in a closed loop on the tube side and in an open loop on
the annulus side using two centrifugal pumps. Counter-current ow is used to obtain higher log-
mean temperature dierences, and ows are regulated with valves using calibrated rotameters.
Water is heated in the plate heat exchanger by 60 kPa steam and the temperature is controlled by
adjusting the steam valve. The pressure tappings from the test section are connected to a bank of
U-tube manometers constructed from nylon tubing. The whole apparatus is made from stainless
steel and insulated with ber glass. The main dimensions of the heat exchanger pipes are presented
in Table 1.

2.2. Dimple tubes

Six dimpled copper tubes of varying geometries were used for comparison with a standard
smooth tube. The data are presented in Table 2. A section of a four-N dimple tube used in the
investigation is shown in Fig. 2.

2.3. Data acquisition

Water ow rate, temperature and the pressure drop were measured to determine the tube-side
convective heat transfer coecient and friction factor. Inlet and outlet temperatures on both sides
J. Chen et al. / Applied Thermal Engineering 21 (2001) 535547 539

Table 1
Characteristic dimensions of heat exchangers
Double pipe heat exchanger
Pipe outside diameter 38.0 mm
Pipe inside diameter 35.0 mm
Tube outside diameter 19.1 mm
Tube inside diameter 16.6 mm
Heated length 1580 mm
Plate heat exchanger
Plate length 520.0 mm
Plate width 51.0 mm
Plate spacing 1.7 mm
Plate thickness 0.5 mm
Heated area per plate 0.031 m2

Table 2
Characteristic dimensions of the tubes
Physical Smooth Dimple tubes
quantity tube
Tube 0 Tube 1 Tube 2 Tube 3 Tube 4 Tube 5 Tube 6
aa , mm2 0.0 1510.6 2756.3 769.0 4307.6 1265.2 465.8
di , mm 16.6 16.6 16.6 16.6 16.6 16.6 16.6
d, mm 1.22 1.22 1.22 1.22 1.22 1.22 1.22
L, mm 1500 1500 1500 1500 1500 1500 1500
e, mm 1.2 1.5 0.5 1.3 0.7 0.6
/, mm 4.5 3.0 2.0 3.5 4.7 5.5
p 12.0 14.0 8.0 10.0 10.0 10.0
N 3 4 6 6 6 3
e=d 0.0723 0.0904 0.0301 0.0783 0.0422 0.0362
e=p 0.1 0.1071 0.0625 0.13 0.07 0.06
a=Ab 0.0174 0.0318 0.0089 0.0496 0.0146 0.0054
a
a N 1500 150=p/=22 e2 p1.
b
A pLdi d 86; 773:9 mm2 .

Fig. 2. Characteristic parameters of a dimple tube N 4.


540 J. Chen et al. / Applied Thermal Engineering 21 (2001) 535547

of the double-pipe heat exchanger were measured using a HewlettPackard digital multimeter,
relay multiplexer, and data were recorded, interpreted, and displayed on a personal computer.
Since tube-wall temperature was not measured directly, the tube-side heat transfer coecient hi
was determined from the overall heat transfer coecient:
1 1 Ai ln do =di Ai 1
Rf : 1
U hi 2pkLs Ao ho
In the investigation, the last three terms on the right-hand side of Eq. (1) were kept constant:
1 1
B: 2
U hi
It is well known that for turbulent ow, the tube-side heat transfer coecient is proportional to
the Reynolds number to the power m:
hi C Rem : 3

Therefore, Eq. (1) can be written as


1 1 m
m B A Re B: 4
U C Re
If the inverse overall heat transfer coecient is plotted as a function of the inverse Reynolds
number to the power of m, it will be a straight line. The parameter B is the intercept of this line
with the Y-axis. Therefore, the experimental tube-side heat transfer coecient is obtained from
1
hi 1
: 5
U
B

The overall heat transfer coecients U is determined from

Q_
U : 6
ADTlm
and the heat transfer rate is calculated by

Q_ Mc
_ p T2 T1 : 7

In this investigation, the water ow rate in the annulus and the inlet temperature were kept
constant. The level of accuracy for determining the annular heat transfer coecient therefore has
an insignicant eect on the tube-side heat transfer coecient. The tube-side thermocouple mea-
surement error was about 0:3C that is too large for an accurate energy balance. Therefore, the
heat ow rate was determined from the annulus temperature. The heating water had an inlet
temperature of 68:1  0:2 C (in most cases, the temperature varied by 0:1C), a constant mass
ow rate of 0.213 kg/s, and a constant velocity of 0.316 m/s. The tube-side water ow velocity was
set to the desired values varying from 0.39 to 2.96 m/s. This range corresponds to a range of
Reynolds number from 7:5  103 to 5:2  104 .
J. Chen et al. / Applied Thermal Engineering 21 (2001) 535547 541

3. Results and discussion

The range of investigations is presented in Table 3. The coecients A, B and m (dened in Eqs.
(2)(4)) were obtained from tting the curves of U versus Re, and U versus Rem , using a least-
squares error method. These data are presented in Table 4.

3.1. Heat transfer

Tube-side heat transfer coecients determined by Eqs. (1)(7) are plotted against Reynolds
number in Fig. 3 for the smooth tube. Experimental turbulent ow heat transfer coecients for
water in the smooth tube agree within 8% of the values predicted by the Gnielinski [5] Eq. (8):

Table 3
Range of investigations
Annulus side
Water ow velocity 0.3161 m/s
Reynolds number 11,172 < Re < 11,562
Water bulk temperature 56.9C < Tb < 68.2C
Tube side
Water ow velocity 0.385 m/s < v < 2.96 m/s
Reynolds number 7,566 < Re < 51,805
Prandtl number 5.78 < Pr < 6.69
Water bulk temperature 20.1C < Tb < 38:1C

Table 4
Values of parameters for heat transfer correlation
Tube 0 Tube 1 Tube 2 Tube 3 Tube 4 Tube 5 Tube 6
A 0.9179 0.8922 0.8409 0.8646 0.92 0.9302 1.0323
B 3:4011  104 3:3383  104 3:5361  104 3:6287  104 3:2736  104 3:4273  104 3:4545  104
m 0.849 0.895 0.889 0.889 0.934 0.889 0.883

Fig. 3. Tube-side heat transfer coecient as a function of Reynolds number for the smooth tube.
542 J. Chen et al. / Applied Thermal Engineering 21 (2001) 535547

Fig. 4. Comparison of tube-side heat transfer coecients as a function of Reynolds number for tubes 06.

"  2=3 #
k f =2Re 1000 Pr d
hi h i 1 : 8
di 1 12:7 f =21=2 Pr2=3 1 L

The heat transfer data for the six dimpled tubes are shown in Fig. 4 based on the actual heat
transfer area and tubes inside diameter. It can be seen that the values of heat transfer coecient
are signicantly larger (between 1.25 and 2.37 times) than those for the smooth tube. Tube 4 (the
largest value of e=di , e=p and N) has the maximum heat transfer coecient improvement of 2.37
times the smooth tube value. Tube 6 produced the lowest improvement of 1.25 times that for the
smooth tube. The augmentation of other dimpled tubes ranges from 1.49 to 1.65 times the smooth
tube values. All six dimpled tubes were found to give improved performance in the Reynolds
number range studied. It seems that hi increases progressively with dimple depth e and the
numbers of the dimple columns N.
An interesting feature of dimpled tubes is the fact that the Reynolds number exponent is
consistently higher than the value of 0.85 obtained for the smooth tube. The exponent is usually
below 0.8 for commonly used heat transfer enhancement devices such as wire coils or twisted
tapes. It was observed that the Reynolds number exponent for dimpled tubes in the hi Re rela-
tionship is a unique function of the dimple parameters. The Re exponent varied from 0.883 to a
maximum of 0.934. It can be observed that dimpled tubes are successful enhancement devices at
low as well as high ow rates. The following correlations closely tted the data points for the
dimpled tubes based on the tube inside diameter:
k
hd C Rem Pr 1=3 ; 9
di
where
 0:01311  0:1977  0:2055
e e e
C 0:01369 N 0:006216 ; 10
di p /
 0:07604  0:1706  0:01432
e e e
m 1:066 N 0:003262 : 11
di p /
J. Chen et al. / Applied Thermal Engineering 21 (2001) 535547 543

Eq. (9) predicts the results of the present work for the heating of water in dimpled tubes with a
standard deviation of 5.5%. In most cases, the error is less than 9% despite the dierent tube
geometries. Since the increase of inside surface area of the dimpled tubes is only 0.5% to 5% above
that of the smooth tube, the eects of the dimples on heat transfer is mainly due to the increase of
the disturbance in the laminar sublayer of the boundary layer. The same argument applies to the
increase in friction factor for the dimple tubes.

3.2. Friction factor

The smooth-tube data were in good agreement with the predicted values from the Petukhov
equation [6]:

f 141:82 log Re 1:642 : 12

Fig. 5 shows the measured and predicted friction factor as a function of the Reynolds number for
the smooth tube. The average deviation is 5.2%, with minimum and maximum deviations of
13.4% and 4.3%.
The experimental results and the comparisons for all investigated tubes are presented in Fig. 6.
For all dimpled tubes, the friction factor decreases with increasing Reynolds number in a manner
similar to that of smooth tubes. The laminar sublayer inside the dimpled tube will be disturbed by
eddies generated by the dimples, which will lead to an increase in pressure drop and hence friction
factor.
The friction factor values for the augmented dimpled tube were found to depend on the depth-
to-tube inside diameter ratio e=di , the depth-to-pitch ratio e=p, the depth-to-dimple diameter ratio
e=/, the dimple column number N, Reynolds number, and the normal ow variables. The friction
factor for the dimpled tubes was correlated using the principle of least squares, within an average
deviation of 3.35%:
fd J Rem ; 13

Fig. 5. Friction factor as a function of Reynolds number for the smooth tube.
544 J. Chen et al. / Applied Thermal Engineering 21 (2001) 535547

Fig. 6. Comparison of friction factor as a function of Reynolds number for tubes 06.

where
 0:6455  1:5434  0:5633
e e e
J 0:0667 N 0:0133 ; 14
di p /

 0:6333  1:1519  0:8132


e e e
m 0:0977 N 0:2587 : 15
di p /
The measured value of friction factor for dimpled tubes were within 14.3% to 20.2% of the
values predicted by Eq. (13) over a range of Reynolds number 7566 < Re < 51,805. The value
of friction factor for all the dimpled tubes are 1.08 to 2.35 times higher than the value for the
smooth tube. Tube 4 which has the best heat transfer performance also has the largest friction
factor. The tubes outside pressure drop measured was the same for both the dimpled tubes and
the smooth tube on the annular side.

3.3. Eectiveness of dimple tubes

One method of evaluating the eectiveness E of an enhanced tube is to compare the heat
transfer coecient with the heat transfer coecient for a smooth tube at the same pumping
power. Stated mathematically, the eectiveness E is ha =hs at equal pumping power. If E is less
than unity, there are other factors such as material cost, space limitation, etc., that may make
enhanced tubes still a viable option.
A plot of heat transfer coecient versus pumping power is presented in Fig. 7. It shows that all
six dimpled tubes substantially improve heat transfer. For any specied pumping power con-
sumption W, heat transfer coecients are augmented by a factor of 1.151.84. The best perfor-
mance is again obtained with tube 4, which has the greatest ratio of e=di and e=p of the six dimpled
tubes. The curves for the dimpled tubes are almost parallel to the smooth tube curve showing that
the performance for dimpled tubes is better over the whole range and depends only on the dimple
parameters. This result is dierent from other types of enhanced tubes where the heat transfer
improvement is strongly dependent on either Reynolds number or pumping power. More freedom
is allowed in designing heat exchangers with dimpled tubes.
J. Chen et al. / Applied Thermal Engineering 21 (2001) 535547 545

Fig. 7. Comparison of tube-side heat transfer coecients as a function of pumping power for tubes 06.

Fig. 8. Comparison of heat transfer to friction factor ratio as a function of Reynolds number for tubes 06.

The performance of an augmented tube is more clearly highlighted if the ratio of


hia =fa =his =fs (i.e., the relative J factor) at a constant Reynolds number can be obtained. A
comparison of the ratio of heat transfer coecient to friction factor hi =f as a function of
Reynolds number for the dierent tubes is shown in Fig. 8.
It is interesting to note that the mean value of the relative J factor for the dimpled tubes varies
from 0.93 to 1.16. The J factors for tube 2 (0.93) and tube 3 (0.99) are both less than 1.0. All the
other four dimpled tubes 1 and 46 give a better performance with J factors larger than unity,
namely 1.16, 1.01, 1.03 and 1.16, respectively. This means that the increase in heat transfer co-
ecient is greater than that of the increase in friction factor for specic dimpled tube designs.
These dimpled tubes may be called absolutely augmented heat transfer tubes. The result obtained
for the maximum improvement of J factor (1.16 for both tube 1 and tube 6), is dierent from that
obtained by comparing pumping power, where tube 4 gave the highest performance.

4. Comparison with other augmented surfaces

Fig. 9 shows a comparison of the performance of dimple tube 4 with other heat transfer
augmentation designs in terms of heat transfer and friction factor performance. The curves for the
546 J. Chen et al. / Applied Thermal Engineering 21 (2001) 535547

Fig. 9. J factor as a function of Reynolds number for enhanced tubes (7).

other designs have been taken from a paper by Bergles and Jensen [7]. Similar performance curves
are given by Marto [3]. Improvements in heat transfer of up to a factor of 5 are possible with some
designs. However, these are accompanied by a corresponding increase in pressure drop of up to a
factor of 200. Next to the smooth tube, tube 4 (no. 18) has the highest ratio of J-factor to friction
factor over the complete range of Reynolds numbers investigated.

5. Conclusions

Experimental investigations of six dimpled tubes used as heat transfer augmentation devices
have been carried out. The results can be summarised as follows:

1. All dimpled tubes increased the tube-side heat transfer coecient signicantly above the
smooth tube values. It was shown that the augmentation was 1.252.37 times at constant Rey-
nolds number, and 1.151.84 times at constant pumping power. Friction factor ranged from
1.08 to 2.35 times those for the smooth tube.
2. Four dimpled tubes had relative J-factors larger than 1, indicating the excellent performance of
this design.
3. Based on either constant Reynolds number or constant pumping power, the best dimpled tube
was tube 4, which had the largest dimple depth-to-tube inside diameter ratio, dimple depth-
to-pitch ratio, dimple depth-to-dimple diameter ratio, and number of dimple columns N.
J. Chen et al. / Applied Thermal Engineering 21 (2001) 535547 547

4. A correlation based on the results of the present work appears to be suciently accurate for
predicting the heat transfer coecients and friction factors.

In a heat exchanger with the controlling thermal resistance on the tube side, it would always be
possible to replace smooth tubes by selected dimpled tubes having only about one half of the
length of the smooth tubes. This could be done without changing the number of tubes, the tube
diameter, the ow rate, the pressure drop, the amount of heat transfer, the maximum wall tem-
perature or the temperature dierence between the wall and the bulk uid. Thus, as a rst ap-
proximation, the size and weight of the heat exchanger could be reduced by a factor of almost 2
without aecting any other system conditions. The shape of the dimples is such that no increased
fouling is expected, and that standard methods of chemical and mechanical cleaning should still
be applicable.

References

[1] E.K. Kalinin, G.A. Dreitser, N.V. Paramonov, A.S. Myakochin, A.I. Tikhonov, S.G. Zakirov, E.S. Levin, L.S.
Yanovsky, Comprehensive study of heat transfer enhancement in tubular heat exchangers, Thermal Fluid Engng. 4
(1991) 656666.
[2] A. Giovannini, B. Ferries, B. Lotado, Aerothermal performances of enhanced heat transfer tubes in transitional
regime for compact heat exchangers: problems associated with dimensioning criteria, Entropie, Seminar no. 9,
1991, pp. 9398.
[3] P.J. Marto, D.J. Reilly, J.H. Fenner, An experimental comparison of enhanced heat transfer condenser tubing.
Advances in Enhanced Heat Transfer, ASME, New York, 1979, pp. 19.
[4] R.L. Webb, Performance evaluation criteria for use of enhanced heat transfer surfaces in heat exchanger design,
Int. J. Heat Mass Transfer 24 (4) (1981) 715726.
[5] V. Gnielinski, New equations for heat and mass transfer in turbulent pipe and channel ow, Int. Chem. Engng. 16
(1976) 359368.
[6] B.S. Petukhov, Heat transfer and friction in turbulent pipe ow with variable physical properties, in: J.P. Hartnett,
T.F. Irvine (Eds.), Advances in Heat Transfer, Academic Press, New York, 1970, pp. 504564.
[7] A.E. Bergles, M.K. Jensen, Enhanced single-phase heat transfer for OTEC systems. In: Proceedings Fourth Annual
Conference on Ocean Thermal Energy Conversion, New Orleans, Louisiana, 1977, pp. VI-41VI-54.