Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: Local heat transfer coefficients and average pressure drop measurements for turbulent flow in a circular
Received 4 June 2014 tube fitted with a nonconventional insert are presented here. The insert was fabricated with a central rod
Received in revised form 5 August 2014 on which curved delta wing vortex generators were attached on opposite sides at specific axial locations.
Accepted 15 August 2014
The influence of pitch to projected length ratio (p/pl), height to tube inner diameter ratio (e/d) and angle
Available online 7 September 2014
of attack (a) on the heat transfer performance are reported here. Air was used as the working fluid and
local heat transfer measurements for both smooth and rough surface sides of the tube are reported for
Keywords:
fully developed turbulent flow with Reynolds number (Re) varying between 10,000 and 45,000. The
Heat transfer enhancement
Vortex generators
average Nusselt number ratio with and without the insert (Nua/Nus), at constant Reynolds number (Re)
Local Nusselt number is found to be in the range of 1.3–5.0. The Nusselt number ratio (Nua/Nuc), based on equal pumping power
Turbulent flow is found to be in the range of 1.0–1.8. Empirical correlations are developed for Nusselt Number and
Circular pipe friction factor in terms of Reynolds number (Re), pitch to projected length ratio (p/pl), height to tube inner
diameter ratio (e/d) and angle of attack (a).
Ó 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2014.08.042
0017-9310/Ó 2014 Elsevier Ltd. All rights reserved.
552 P.W. Deshmukh, R.P. Vedula / International Journal of Heat and Mass Transfer 79 (2014) 551–560
Nomenclature
2.0 to 1.5 when the Reynolds number increased from 5000 to current device is manufactured using curved delta wing type vor-
13,000. tex generators to form an insert that can be used in tubes and
Another passive technique for heat transfer enhancement for pipes. The curvature is provided to ensure proper contact with
flow inside tubes well reported in the literature is the use of wire the tube wall. Experimental data for the heat transfer performance
coil inserts. Garcia et al. [2] studied the thermo-hydraulic behav- of the proposed insert is presented in this article.
iour of different configurations of helical wire coil inserts in the
laminar, transition and turbulent flow regimes. They reported that 2. Geometric details of vortex generator insert
the best helical wire coil produces a heat transfer enhancement of
2.5 at Reynolds number 10,000 and this enhancement gradually The details of the vortex generator insert used in the present
decreases with the Reynolds number. The performance ratio, R3 study are shown in Fig. 1. A 0.5 mm thick aluminum sheet
at constant pumping power was reported to vary from about 1.5 was bent in the form of a hollow cone having included angle equal
at an equivalent Reynolds number of 10,000 to almost unity at to twice the desired angle of attack of vortex generator as shown in
an equivalent Reynolds number of 100,000. Fig. 1(a). The diameter of the base of the cone was made equal to
Liu et al. [7] reviewed the various passive heat transfer the inner diameter of the tube in which the enhancement
enhancement options available in the literature. They indicated device is to be placed. The curved triangular shaped vortex
that the twisted tape inserts perform better in laminar flow condi- generator with base ‘b’, included angle ‘b’ and length ‘c’ was cut
tions whereas wire coil inserts perform better in turbulent flow. from this cone.
Discrete vortex generators have been reported for square and A 1 mm diameter steel pin as shown in Fig. 1(b), was bent such
rectangular geometries for heat transfer enhancement. Feibig that it has three parts, lying in a single plane. The first part is hor-
et al. [4] reported heat transfer and drag measurements using delta izontal, the second part is perpendicular to first and third part is at
wing type of longitudinal vortex generators for rectangular chan- an angle ‘a’ equal to desired angle of attack, to the first part. The
nel flow. They concluded that Nusselt number enhancement Nua/ delta wing shaped curved vortex generator was then glued to the
Nus is a function of the vortex generator aspect ratio, K, angle of third part in a manner such that the projected area of the curved
attack, a, and Reynolds number, Re. A higher angle of attack a vortex generator on a plane normal to that containing the steel
and aspect ratio, K results in enhancement in both heat transfer pin is maximum. These steel pins were periodically glued on a
and pressure drop. Yakut et al. [3] experimentally investigated 2 mm diameter rod to make the required insert as shown in
the dependence of geometrical parameters of delta winglet vortex Fig. 1(c). A cotton string was used to tie the part 1 to the central
generators on heat transfer, pressure drop and flow induced vibra- rod for additional strength. The length of the first part of the steel
tions characteristics in internal flow through a pipe. They used pin is not important but the length of the second part is such that
tapes with delta winglets on both sides inserted in the round tube. the base of the vortex generator is in good contact with the inner
Liou et al.[10] reported data for heat transfer enhancement in a wall of the tube. The length of third part of the steel pin is also
square duct with 12 different shaped vortex generators at not very important and is kept equal to vortex generator length
Re = 12,000. Among these configurations studied, they concluded ‘c’, shown in Fig. 1(a).
that delta wing vortex generator provide highest heat transfer aug- Initial data was obtained with the delta wing pins glued to the
mentation at constant pumping power. central rod with epoxy glue but the strength at the attachment
A new type of enhancement device for tube flow is proposed in point was very poor and therefore, the joint was fortified using a
this study. All the delta wing type vortex generators that have been string. A comparison of the data obtained with and without the
reported were used on flat surfaces and were therefore flat. The string indicated differences well within the uncertainty limits of
P.W. Deshmukh, R.P. Vedula / International Journal of Heat and Mass Transfer 79 (2014) 551–560 553
the experimentation. This was a laboratory scale insert and an pressure drop due to the insert, but large enough to provide struc-
industrial scale insert could be manufactured using joining tural rigidity to the insert. The end view of the tube with the insert
techniques like brazing or welding. A photograph of the insert along with the circumferential and flow directions, h, z respectively
being inserted into the tube is shown in Fig. 1(d). are shown in the Fig. 1(e). The circumference of the tube is divided
The diameter of the central rod and the steel pins are not into two regions for the sake of convenient data representation:
important but must be as small as possible to prevent large region with vortex generators (0° < h < 85°, 180° < h < 265°) and
554 P.W. Deshmukh, R.P. Vedula / International Journal of Heat and Mass Transfer 79 (2014) 551–560
1 – Thermocouples K type
2 – Thermal Camera
Fig. 3. Nu(z)/Nus variation for Re = 25,000 (p/pl = 12.8, a 30° i.e. e/d 0.17, K = 3.8).
(b)
Fig. 4. (a) Stream wise section of surface with vortex generators (b) circumferential distribution of local Nusselt number ratio, Nu(z, h)/Nus around the vortex generator at
Re = 35,000 (p/pl 12.8, angle of attack, a 30°, e/d 0.17).
exposed to the ambient to permit temperature measurements by Q imbalance ¼ Q in Q out Q loss ð03Þ
thermal camera. The electrical power required to maintain the test
section at a given temperature with no flow through the test section
was measured and considered equal to Qloss. The experiment was The energy imbalance calculated as percentage of the electrical
repeated at different values of electrical power and a curve fit was power input was always less than 10% for all the experiments
obtained for heat loss as a function of temperature difference performed. This imbalance is due to the uncertainties in the mea-
between the wall and ambient. surement of electrical current and voltage, flow rate, temperatures
Energy imbalance Qimbalance was calculated as: at inlet, outlet and wall.
556 P.W. Deshmukh, R.P. Vedula / International Journal of Heat and Mass Transfer 79 (2014) 551–560
Nua
Nuc ¼ 0:023Re0:8
c Pr
0:4
; R3 ¼ ð11Þ
Nuc
The experimental uncertainties, within a 99.7% confidence interval,
were calculated to be 15.6% for Nusselt number and 8.7% for friction
factor measurements.
The maximum heat transfer to the fluid due to the fin effect is
estimated to be 6.8% of the total heat input, for the case with
highest number of fins and highest heat transfer coefficients by
using the standard fin analysis for triangular fins. However, since
the fins are manually made, the physical contact between the wall
and fin is poor resulting in a high thermal contact resistance. In
addition, there is only a line and not an area contact between the
fin and the wall, the maximum effect due to the fin is, therefore
likely to be much smaller than the above estimated maximum
value and is, therefore considered insignificant.
Fig. 5. Schematic of the flow across the vortex generator. 4. Results and discussions
The thermal image was used to obtain the local temperature on The experimental data of Nusselt number and pressure drop for
the test section wall. The bulk temperature at different axial smooth tube flow at different Reynolds numbers was compared
locations was computed by assuming a linear variation between with the relations given in Eq. (10) and the mismatch between
the inlet and exit locations since a uniform heat flux at the wall average Nusselt number, Nu and friction factor, f was less than
was imposed. The local Nusselt number Nu(z, h), Circumferential 10% and 5% respectively. Additionally, since a new type of insert
averaged Nusselt number at any axial location for rough and has been proposed in the current study, the experimental method-
smooth surfaces, Nu1(z), Nu2(z) respectively, of the test section ologies were tested against the wire coil insert data reported by
and their corresponding average Nusselt number values, Nu(1) Garcia et al. [2] by replicating one of the cases giving the best per-
and Nu(2) were obtained as formance and the maximum deviation from their data was noticed
to be less than 10% – the comparison is presented in Figs. 9–11.
Q out d
Nuðz; hÞ ¼ ð04Þ These results provided the confidence in the experimental
A½T w ðz; hÞ T b ðzÞ k
R 85 methodologies and procedures adopted in the present study.
0
Nuðz; hÞ dh Fig. 3 shows the axial variation of the circumferentially
Nu1 ðzÞ ¼ R 85 ð05Þ
dh averaged Nusselt number ratio, Nu(z)/Nus along the length of test
0
R 180 section for the first four vortex generators having p/pl and e/d
85
Nuðz; hÞ dh ratios equal to 12.8 and 0.17 respectively. The exact physical
Nu2 ðzÞ ¼ R 180 ð06Þ
85
dh location of the four vortex generators inside the tube is also
RL indicated in the figure.
0
Nu1 ðzÞ dz
Nuð1Þ ¼ RL ð07Þ The flow first encounters the leading edge of the vortex
0
dz generator where the heat transfer coefficient begins to rise. This
RL
Nu2 ðzÞ dz increase continues along the stream wise direction underneath
0
Nuð2Þ ¼ RL ð08Þ the vortex generator, reaching a maximum at the trailing edge.
0
dz
R In the region downstream of the trailing edge the heat transfer
NuðzhÞdzdh coefficient values rapidly decrease but before the heat transfer
Nua ¼ R ð09Þ
dzdh coefficients drop to the smooth tube values the flow encounters
the next vortex generator and the process repeats. It can be noticed
The overall average Nusselt number, Nua was calculated as the
that the flow becomes developed, i.e. heat transfer coefficient
arithmetic average of the smooth and rough surface average Nusselt
reaches a constant value, after the first vortex generator itself.
numbers Nu(1) and Nu(2). The performance of a heat transfer
The flow became developed after the first one or two vortex gener-
augmentation device is described by Bergles et al.[9] and several
ator elements for all the geometric variations studied here.
others by using the criterion R3 (Nua/Nuc) where Nusselt number,
There is also a significant variation in the circumferential Nus-
Nuc is evaluated at the equivalent smooth tube Reynolds number
selt numbers at different stream wise (i.e. z) locations. The loca-
Rec for equal pumping power and equal heat exchange surface area.
tions are labelled as A, B, C, D, E, F, as shown in Fig. 4(a). The
The Nusselt number and friction factor for the smooth tube for pres-
locations are equally spaced between the leading edge of the one
ent study were obtained by the following well established
vortex generator and the trailing edge of the subsequent vortex
equations:
generator. The corresponding Nusselt number variations are
NuS ¼ 0:023Re0:8
S Pr
0:4
ðDittus BoelterÞ; f S ¼ 0:079Re0:25
S ðBlasiusÞ shown in Fig. 4(b). The heat transfer coefficients are highest along
ð10Þ the centreline of the vortex generator (i.e. h = 42.5°) and reduce on
either side of the centre line.
The friction factor fa, for a given augmented tube Reynolds number A schematic of the possible mechanism is shown in Fig. 5. The
Rea, was used to obtain the equivalent smooth tube Reynolds fresh mainstream fluid is directed towards the pipe wall by
number Rec evaluated using: the vortex generator and simultaneously exits from underneath
the vortex generator in the circumferential direction. This causes
Re3a
f a Re3a ¼ f s Re3c ) Re2:75
c ¼ fa high heat transfer coefficients at the centerline (h = 42.5°) with a
0:079 progressive decrease on either side. The circumferential region of
The Nusselt number for the smooth tube was evaluated at this the tube without the vortex generator (85° < h < 180°) experiences
equivalent Reynolds number. elevated heat transfer coefficients over the smooth tube values due
P.W. Deshmukh, R.P. Vedula / International Journal of Heat and Mass Transfer 79 (2014) 551–560 557
Fig. 6. Contour plot for local Nua/Nus over the tube surface (a) with VGs (b) without VGs for Re = 25,000 (p/pl = 12.8, a 30° i.e. e/d 0.17, K = 3.8).
Fig. 7. Contour plot for local Nua/Nus over the tube surface (a) with VGs (b) without Gs for Re = 45,000 (p/pl = 6.4, a 30° i.e. e/d 0.17, K = 3.8).
558 P.W. Deshmukh, R.P. Vedula / International Journal of Heat and Mass Transfer 79 (2014) 551–560
Fig. 8. Average Nua/Nus with pitch to projected length ratio, P/Pl for Re = 35,000.
Fig. 9. Average Nua/Nus v/s Reynolds number, Re for vortex generator inserts and helical wire coil.
to the cross stream flows from underneath the vortex generator due to reduction in number of vortex generators per unit length
towards this region. Mainstream fluid moving towards the wall which results in a smaller amount of fresh mainstream fluid being
beneath the vortex generators progressively enhances the heat directed towards the heated tube wall. An optimum value of p/pl
transfer coefficients from the leading edge to the trailing edge. ratio therefore exists where, heat transfer enhancement is maxi-
The region downstream of the trailing edge also has fresh main- mum. Optimum values for the e/d = 0.17 and 0.1 could be obtained
stream fluid moving towards the wall resulting in high heat trans- but experimental limitations prevented obtaining data for p/pl
fer coefficients there as well. There are likely to be longitudinal values lower than 1.6 for e/d = 0.25 in the present investigation.
vortices at low angles of attack and a mixing region due to a wake The optimum value of p/pl is noticed to reduce progressively as
at large angles of attack in the region downstream of the vortex e/d increases.
generator. The heat transfer coefficients progressively decrease as The Nusselt number and friction factor ratios with respect to
the disturbance dissipates in the zone downstream of the trailing the corresponding smooth tube values, Nua/Nus as a function of
edge. Figs. 6 and 7 correspond to a projected part of circular test Reynolds number for all different configurations are presented in
section on a two dimensional plane which contain the thermal Fig. 9. The figure also includes the data obtained from the current
camera image. Figures show the distribution of local Nusselt num- experimental apparatus for the wire coil insert geometry reported
ber enhancement ratio, Nua/Nus, of the smooth and rough surfaces. by Garcia et al. [2] to give the best performance and the compari-
Fig. 8 shows the variation of Nusselt number ratio, Nua/Nus with son can be noted to be good. The vortex generator configuration
p/pl at a given Reynolds number. At very low p/pl values, the main- having angle of attack, a 15° i.e. e/d ratio 0.09, has the lowest Nus-
stream flow is unable to penetrate the region between successive selt number augmentation. The configurations with angle of attack
vortex generators, due to the larger resistance offered by the vortex a equal to 30° and 45° show a Nusselt number augmentation,
generators resulting in reduced heat transfer coefficients. At very between 2.5 and 5.0, for the Reynolds number ranging from
high values of p/pl ratios, the heat transfer enhancement is reduced 10,000 to 45,000. The Nusselt number augmentation, Nua/Nus is
P.W. Deshmukh, R.P. Vedula / International Journal of Heat and Mass Transfer 79 (2014) 551–560 559
Fig. 10. Average fa/fs v/s Reynolds number, Re for vortex generator inserts and helical wire coil.
Fig. 11. Average performance ratio, R3 v/s equivalent Reynolds number, Rec for vortex generator inserts and helical wire coil.
Fig. 12. Comparison of empirical fitted Nusselt number, Nua with experimental values.
560 P.W. Deshmukh, R.P. Vedula / International Journal of Heat and Mass Transfer 79 (2014) 551–560
Fig. 13. Comparison of empirical fitted friction factor, fa with experimental values.
maximum for the vortex generator configuration having p/pl ratio The average heat transfer enhancement for the surface with
1.6, angle of attack, a equal to 45° i.e. e/d ratio 0.25 for entire vortex generators is found to be much higher than that for the
Reynolds number range studied. Fig. 10 shows the variation of fric- surface without vortex generators. In general, the average heat
tion factor ratio, with Reynolds number, Re. The pressure drop transfer characteristics of the vortex generator insert are found
increases with increasing angle of attack, a and decreasing p/pl. to be superior to the conventional helical wire coil insert. The vor-
Fig. 11, shows the variation of performance ratio R3, with equiv- tex generator insert induces secondary flows only at the localized
alent Reynolds number, Rec for all the configurations studied. It is areas and therefore heat transfer enhancements at specific
observed that all the configurations show the value of performance locations can be easily produced. However, large variations of heat
ratio, R3 greater than 1.0, except for the configuration of vortex transfer coefficients exist over the surface of the tube.
generator having p/pl ratio 1.4, e/d ratio 0.09, angle of attack, a
equal to 15°. Conflict of interest
Empirical correlations for Nusselt number Nu, friction factor f,
have been developed for tubes with vortex generator inserts for None declared.
Reynolds number range between 10,000 and 45,000, angle of
attack a between 15° and 45° i.e. e/d ratio 0.09–0.25, p/pl ratio of References
1.4–7.9 for aspect ratio, K of 3.8
0:125 [1] S. Chang, M. Guo, Thermal performances of enhanced smooth and spiky
0:17 0:79 2
Nu ¼ 0:46Re0:77 ðp=plÞ ðe=dÞ 0:125 ðe=dÞ ð12Þ twisted tapes for laminar and turbulent tubular flows, Int. J. Heat Mass
Transfer 55 (2012) 7651–7667.
0:7 [2] A. Garcıa, P.G. Vicente, A. Viedma, Experimental study of heat transfer
0:59 1:22 2
f ¼ 0:77Re0:14 ðp=plÞ ðe=dÞ 0:125 ðe=dÞ ð13Þ enhancement with wire coil inserts in laminar–transition–turbulent regimes
at different Prandtl numbers, Int. J. Heat Mass Transfer 48 (2005) 4640–
Figs. 12 and 13 show comparison between the predicted experi- 4651.
[3] K. Yakut, B. Sahin, C. Celik, N. Alemdarolu, A. Kurnuc, Effects of tapes with
mental data and the present predictions by the Eqs. (12) and (13) double-sided delta winglets on heat and vortex characteristics, Appl. Energy 80
above. In the figures, calculations using the present correlations (2005) 77–95.
agree reasonably (within ±20%) with experimental data for the fric- [4] M. Feibig, N. Mitra, Heat transfer enhancement and drag by longitudinal vortex
generators in channel flows, Exp. Therm. Fluid Sci. 4 (1991) 103–114.
tion factor and Nusselt number. [5] P. Sivashanmugam, S. Suresh, Experimental studies on heat transfer and
friction factor characteristics of turbulent flow through a circular tube fitted
5. Conclusions with helical screw-tape inserts, Chem. Eng. Process. 46 (2007) 1292–1298.
[6] R.M. Manglik, A.E. Bergles, Heat transfer and pressure drop correlations for
twisted tape inserts in isothermal tubes: Part II – transition and turbulent
A new insert has been proposed for enhancing heat transfer flows, J. Heat Transfer 115 (1993) 890–896.
coefficients for turbulent flows in a pipe. The local heat transfer [7] S. Liu, M. Sakr, A comprehensive review on passive heat transfer enhancements
in pipe exchangers, Renew. Sustainable Energy Rev. 19 (2013) 64–81.
and average pressure drop for turbulent flow in a circular tube fit- [8] S. Eiamsa-ard, P. Somkleang, C. Nuntadusit, C. Thianpong, Heat transfer
ted with this vortex generator insert, have been measured under enhancement in tube by inserting uniform/non-uniform twisted-tapes with
uniform heat flux conditions with air as a working fluid for differ- alternate axes: effect of rotated-axis length, Appl. Therm. Eng. 54 (2013) 289–
309.
ent geometric parameters. Vortex generator inserts of almost all [9] A.E. Bergles, R.M. Manglik, Swirl flow heat transfer and pressure drop with
the configurations show Nusselt number values about 150–500% twisted tape inserts, in: J.P. Hartnett, Thomas F. Irvine, Young I. Cho, George A.
higher than the corresponding smooth tube values. The aspect Greene (Eds.), Advances in Heat Transfer, vol. 36, Academic Press, 2002.
[10] T.M. Liou, C.C. Chen, T.E. Tsai, Heat transfer and fluid flow in a square duct with
ratio of the vortex generator was not varied in the current study. 12 different shaped vortex generators, J. Heat Transfer 122 (2000) 327–
The inserts with e/d = 0.25, p/pl = 1.6, a = 45° and e/d = 0.17, 335.
p/pl = 2.1, a = 30°, were observed to produce the best enhancement [11] S. Eiamsa-ard, P. Promvonge, Enhancement of heat transfer in a tube with
regularly-spaced helical tape swirl generators, Sol. Energy 78 (2005) 483–494.
at constant pumping power.