Journal of Mechanical Science and Technology 26 (9) (2012) 2949~2958

www.springerlink.com/content/1738-494x
DOI 10.1007/s12206-012-0702-2

CFD analysis of fin tube heat exchanger with a pair of delta winglet
vortex generators†
Seong Won Hwang1, Dong Hwan Kim1, June Kee Min2 and Ji Hwan Jeong1,*
1
School of Mechanical Engineering, Pusan National University ,30 Jangjeon-Dong, Geumjeong-Gu, Busan 609-735, Korea
2
Rolls-Royce University Technology Center, Pusan National University, 30 Jangjeon-Dong, Geumjeong-Gu, Busan 609-735, Korea

(Manuscript Received June 19, 2011; Revised March 29, 2012; Accepted April 16, 2012)

----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Abstract

Among tubular heat exchangers, fin-tube types are the most widely used in refrigeration and air-conditioning equipment. Efforts to en-
hance the performance of these heat exchangers included variations in the fin shape from a plain fin to a slit and louver type. In the con-
text of heat transfer augmentation, the performance of vortex generators has also been investigated. Delta winglet vortex generators have
recently attracted research interest, partly due to experimental data showing that their addition to fin-tube heat exchangers considerably
reduces pressure loss at heat transfer capacity of nearly the same level. The efficiency of the delta winglet vortex generators widely varies
depending on their size and shape, as well as the locations where they are implemented. In this paper, the flow field around delta winglet
vortex generators in a common flow up arrangement was analyzed in terms of flow characteristics and heat transfer using computational
fluid dynamics methods. Flow mixing due to vortices and delayed separation due to acceleration influence the overall fin performance.
The fin with delta winglet vortex generators exhibited a pressure loss lower than that of a plain fin, and the heat transfer performance was
enhanced at high air velocity or Reynolds number.

Keywords: Fin-tube heat exchanger; Plate and fin tube; Delta winglet; Vortex generators; Flow separation
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

flow separation. The flow mixing caused by vortices enhances

1. Introduction
heat transfer, and thus researchers became interested in apply-
Generally used to exchange heat between a gas and liquid, ing vortex generators to heat exchanger fin designs. The shape
fin-tube heat exchangers are widely used in chemical process- of existing vortex generators includes a rectangular wing,
ing plants, power plants, and home appliances, including small rectangular winglet, delta wing, and delta winglet. Recent
air conditioners and refrigerators [1]. While early fin-tube heat research on the effects of vortex generators on the perform-
exchangers primarily adopted plain fins, extensive research on ance of fin-tube heat exchangers focused on delta winglet
fin shape caused substantial progress in the development of vortex generators (DWVG) rather than on other shapes.
high performance fins to produce compact heat exchangers. The vortex generators incorporated in fin-tube heat ex-
Current compact fin-tube heat exchangers adopt a slit or lou- changers can be categorized based on their configurations
ver fin if the working fluid is not significantly corrosive, such with respect to heat transfer tubes, whether ‘common flow
as in the case of a fume gas. down’ and ‘common flow up’, as illustrated in Fig. 1. The
Fin shapes for compact heat exchanger design are continu- common flow down configuration has a pair of vortex genera-
ously studied to address related global environmental concerns tors mounted behind a tube (at the downstream). Fiebig [3]
and improve the efficiency of the energy system. One notable contended that this configuration would delay flow separation
subject is the incorporation of vortex generators in fins. Early and reduce drag. In the common flow up configuration, a pair
vortex generator research focused on the boundary layer con- of vortex generators is installed in front of the tube (at the
trol of air foils [2]. Vortex generators create longitudinal vor- upstream). Torii et al. [4] speculated that this configuration
tices in the main flow to re-energize the boundary layer that is accelerates the flow between a tube and its vortex generators
developed adjacent to an air foil and, in turn, delay or prevent as well as delays flow separation; thereby reducing the wake
*
Corresponding author. Tel.: +82 51 510 3050, Fax.: +82 51 515 4038 region behind the tube.
E-mail address: [email protected] The effects of vortex generators on the performance of the
†
Recommended by Associate Editor Man-Yeong Ha
© KSME & Springer 2012
heat transfer surface have been investigated in recent decades.
2950 S. W. Hwang et al. / Journal of Mechanical Science and Technology 26 (9) (2012) 2949~2958

10% increase of the f-factor for the 2, 4, and 5-row heat ex-
changers. For the 3-row heat exchanger, however, the f-factor
dramatically decreased by 30%–50%. This was attributed to
the effects of vortices, which were produced at the first row of
the tube but did not reach the fourth and fifth rows.
Kwak et al. [8] investigated the effect of the number of
DWVG rows, as well as measured the heat transfer capacity
and pressure drop of staggered array fin-tube heat exchangers
that were identical except for the number of DWVG rows.
(a) (b) The heat exchanger with a single DWVG row experienced
10%–30% larger heat transfer capacity and 34%–55% less
Fig. 1. Vortex generator configuration: (a) common flow up; (b) com-
pressure drop than the heat exchanger without the DWVG.
mon flow down.
For the heat exchanger with two DWVG rows, however, heat
transfer capacity and the pressure drop increased by 6%–15%
Jacobi and Shah [5] reviewed previous works until the early and by 61%–117%, respectively, compared with the heat ex-
1990s. They explained physical phenomena and vortex char- changer with a single DWVG row. The DWVG in the second
acteristics associated with vortex generators on a fin as well as row obstructed and decelerated the air flow to produce an
introduced experiments and analysis on the performance of additional pressure drop.
heat exchangers with vortex generators. They categorized Recently, Joardar and Jacobi [9] investigated the effect of
active and passive methods of vortex generation. Active the number of DWVG rows in the inline array fin-tube heat
methods generate vortices using external energy, such as elec- exchangers. Single and three DWVG fin tube heat exchangers
tric or acoustic fields, mechanical device, or surface vibration. were compared with a plain fin tube heat exchanger. The
Passive methods generate vortices through structures and ad- DWVG was placed in a common flow up configuration. For
ditional fluids. Regardless of method, vortex generators en- the single DWVG, the heat transfer coefficient was enhanced
hance heat transfer and simultaneous pressure drop due to by 16.5%–44% as pressure drop increased by less than 12%.
form loss. The vortex generator has the interesting characteris- For the case of three DWVG rows, the heat transfer coeffi-
tic of having a small pressure loss compared with other inter- cient was augmented by 29.9%–68.8% as the pressure drop
rupted fin designs. As such, most previous works focused on penalty increased 26% at Re = 960 and 87.5% at Re = 220.
quantifying the improvement of heat transfer performance and DWVG arrays can significantly enhance the performance of
additional pressure drop characteristics. fin tube heat exchangers with flow depths and fin densities
Fiebig et al. [6] experimentally studied heat transfer per- typical of those used in air-cooling and refrigeration applica-
formance in a channel with vortex generators in a common tions.
flow down configuration. The DWVG increases convective Allison and Dally [10] similarly examined the performance
heat transfer by developing a boundary layer, swirl, or vortices, of the DWVG in flat fin tube heat exchangers, and compared
as well as through flow destabilization or turbulence intensifi- the performance of two fin types. The first is a plain fin with a
cation. The delta winglet enhances heat transfer more than a DWVG and the second is a louvered fin. The DWVG were
rectangular winglet. The highest heat transfer coefficient was installed in a common flow up configuration. The j- and f-
obtained at a 45° attack angle, which is defined as the angle factors of the heat exchanger with the DWVG were 87% and
between the flow direction and vortex generators. The effects 53%, respectively, of the heat exchangers with a louver fin.
of the vortex generators on fin-tube heat exchangers were Researchers also investigated the performance of DWVG fins
similarly experimentally investigated. The DWVG were in- through numerical analyses.
stalled onto 3-row fin-tube heat exchangers in a common flow Min and Xu [11] numerically evaluated and compared the
down configuration. The staggered and inline effects of tube performances of a plain, DWVG, and louver fins incorporated
arrays on the thermal-hydraulic performance of the DWVG in a flat tube heat exchanger. When compared with the heat
were compared. At nearly the same heat transfer coefficient, t exchanger with a plain fin, that with a louver fin had a heat
he f-factor of the staggered array was smaller than that of the transfer capacity of about 114.1%–139.1% with a pressure
inline array in a range of Re > 1300. The DWVG can reduce drop penalty of 163.9%–171.7%. In the case of the DWVG fin
the size and mass of a heat exchanger for a given heat load. heat exchangers, heat transfer capacity was about 46.5%–
Kwak et al. [7] measured the heat transfer coefficient and 74.1% and the pressure drop was 46.1%–49.7%. The signifi-
pressure drop in 2, 3, 4, and 5-row fin-tube heat exchangers cant variation in pressure drop could be attributed to the lou-
with a common flow up configuration DWVG in a range of ver fin undergoing flow decay at each slit, whereas the winglet
280 < Re < 2400. The DWVG were only mounted on the first fin had no flow decay.
row. Compared with heat exchangers with plain fins, those The above brief review shows that previous research mainly
with DWVG fins showed a 10% increase in heat transfer per- focused on the quantitative and qualitative effects of vortex
formance of the j-factor for all heat exchangers, and a 0%– generators on pressure drop and heat transfer performance,
 S. W. Hwang et al. / Journal of Mechanical Science and Technology 26 (9) (2012) 2949~2958 2951

through either experimental or numerical approach. Vortex Table 1. Heat exchanger geometric data.
generators were verified to enhance heat transfer. However,
Notation Meaning Value
the pressure drop penalty was less consistent, reflecting a lack
L Length (m) 0.0254
of phenomenological understanding of how vortex generators
W Width (m) 0.021
influence the heat transfer performance and pressure drop,
because few research investigated this underlying mechanism. D Tube diameter (m) 0.008
Thus, the present work aims to deepen the understanding on tth Fin thickness (m) 0.0015
H Fin pitch (m) 0.00123
the DWVG enhancement of heat transfer performance with a
Wl Winglet length (m) 0.00565
relatively small, or even reduced, pressure drop penalty. In-
formation on the fluid flow and heat transfer in a fin-tube heat Wh Winglet height (m) 0.011
exchanger with DWVG was obtained through computational a Attack angle (°) 15
fluid dynamics (CFD). b Central angle (°) 110
Distance between tube and
g 0.001
2. Numerical modeling and analysis winglet (m)

In the present work, steady-state Reynolds Averaged Na-

vier-Stokes (RANS) equations were numerically solved for
the coolant flow inside the reactor pool. Three-dimensional
continuity and momentum conservation equations can be writ-
ten as follows:
Mass conservation equation

∂
( ρ ui ) = 0 (1)
∂xi

Momentum conservation equation

(a) (b)
∂ ∂p ∂   ∂ui ∂u j  
( ρ uiu j ) = − + µ  +  − ( ρ ui u j )  (2)
∂t ∂x ∂xi   ∂x j ∂xi 
 
 

Energy conservation equation (b)

Fig. 2. Geometry definitions: (a) Plain fin-tube heat exchanger; (b)

∂ ∂  ∂T  DWVG fin-tube heat exchanger; (c) DWVG.
( ρ C pTui ) = −  − k + ρ C p ui'T '  (3)
∂x ∂xi  ∂xi 
for geometric definitions are also provided in Fig. 2. Table 1
where u, ρ, t, p, µ, T, k, and Cp represent the velocity, density, lists the geometric values. Fin thickness, fin spacing, and tube
time, pressure, dynamic viscosity, temperature, thermal con- size are similar to those of fin-tube heat exchangers that are
ductivity, and constant-pressure specific heat, respectively. Air widely used in condensers and evaporators of air-conditioners.
flow passing around a circular tube will be influenced by vari- The geometry and installed location of the delta winglet vor-
ous factors. When flow velocity is extremely low (Re << 1), tex generators are similar to those applied in the heat ex-
the flow is dominantly affected by the inertial force and main- changer experimentally investigated by Kwak et al. [7]. Con-
tains a normal stream at the back of the tube. In a range of Re jugate heat transfer was analyzed. Fig. 3 shows the solid do-
< 50, rather than the inertial force, the viscous force of the main mesh of the DWVG fin-tube heat exchanger. A cylindri-
fluid prevails on the air flow. An adverse pressure gradient cal ring represents the heat transfer tube and a flat plate repre-
may thus develop, leading to a flow separation from the tube sents the fin. The triangular structure on the flat plate is the
that no longer has a smoothly changing stream line. When DWVG. Above the meshes for the solid part are those for the
flow separation occurs, a wake develops in the rear of the tube fluid part. Fine meshes are generated in the vicinity of the
where the flow swirls and does not mix with external flow. solid surface where the velocity gradient is large. The mesh
This wake is usually detrimental in terms of heat transfer ca- size increases with the distance from the solid surface. The
pacity and pressure loss [12]. grid size was determined by testing three different fluid
Fig. 2 shows the representative geometry of a two-row fin- meshes: 4.5 million; 2.3 million; and 1.1 million.
tube heat exchanger considered for the numerical analysis The results of 4.5 million and 2.3 million meshes were con-
Delta winglet vortex generators are installed around the first sistent, whereas that of the 1.1 million meshes deviated by
row of the tube in a common flow up configuration. Notations over 10%. Finally, 2.3 million meshes in total for the fluid
2952 S. W. Hwang et al. / Journal of Mechanical Science and Technology 26 (9) (2012) 2949~2958

(Tc - Tin ) - (Tc - Tout )

∆TLM = (6)
Tc - Tin
ln
Tc - Tout

where m& , Ac, Afin, Tin, Tout, Tc, and Q represent the mass flow
rate, fin collar area, fin surface area, maximum velocity, inlet
air temperature, outlet air mean temperature, fin collar tem-
perature, and total heat transfer rate, respectively. Tin is con-
stant because we set it as a boundary condition at the entrance.
However, Tout varies over the cross-sectional plane at the exit
(a) because it is influenced by heat transfer and fluid flow. An
area average value was used in this work and is given as fol-
lows:

Tout =
∫ u TdA f
(7)
∫ u dA f

where Af represents the fluid cross sectional area at the exit.

The thermodynamic properties in the above equations are
typically temperature dependent. Application of the above
equations to temperature-varying situations use the average
values of properties evaluated at the lowest and the highest
(b) temperatures. Based on the CFD results, the average convec-
tive heat transfer coefficient can be obtained using the rela-
Fig. 3. Mesh configuration: (a) fluid mesh; (b) solid mesh.
tionships of Eqs. (4)-(7). Substituting this coefficient value for
Eqs. (8) and (9) can obtain the f-factor and the Colburn-j fac-
domain and 0.4 million meshes for the solid part were used for tor. The friction factor is given as follows:
the present computations. Figs. 3(a) and (b) show the meshes.
The semi-implicit method for pressure-linked equations ∆P D
f = 2
( c) (8)
(SIMPLE) algorithm [13] is used. (ρ × Vmax / 2) 4 L
Air flow is assumed incompressible and turbulent. RNG k-
ε model was used as a turbulence model, considering complex where ∆P and Dc represent the pressure difference and charac-
flows (wake or separation) and anisotropic turbulence. The teristic length, respectively. The diameter of the fin collar is
boundary conditions are as follows. An inlet velocity bound- used. Vmax represent the maximum velocity, which is the aver-
ary condition (0.5–20 m/s) is established for the front end of age velocity at the minimum cross sectional area (Amin) in flow
the fluid domain, whereas an outlet boundary condition is direction, defined as U(Af / Amin)
used for the rear end. Symmetric boundary condition is speci-
fied at both side-ends of the flow path. A constant temperature Cf Nu x
condition was set at the inner tube wall. The temperature of = × Pr 2/ 3 = St x Pr 2 / 3 ≡ j (9)
2 Re x Pr
the solid surface was set such that the heat flux through the
solid part balances with that through the adjacent air. Top and
bottom surfaces were set to adiabatic. The air inlet tempera- where St represents the Stanton number and j is the Colburn j-
ture was fixed at 293 K for all simulations. The heat transfer factor. This relationship is called the Reynolds-Colburn anal-
tube inner wall temperature was set at 333 K to represent a ogy, and implies that the heat transfer performance is en-
working condition of a refrigerant condenser. hanced with an increase in the frictional pressure drop [14].
The following expressions analyzed the data. The heat Heat transfer augmentation with a minimum pressure drop
transfer coefficient is related with the temperature difference, increase is preferred in enhanced heat transfer surface designs
as follows: [15].
At the beginning of the CFD analyses, a preliminary calcu-
Q = mc
ɺ p Tin - Tout (4) lation determined whether the air flow shows any transient
behavior. A fluid flowing around a tube may cause vortex
Q
h= (5) shedding, which is known as the Strouhal number defined by
( Ac + A fin )∆TLM
Eq. (10) to be 0.2 [16].
 S. W. Hwang et al. / Journal of Mechanical Science and Technology 26 (9) (2012) 2949~2958 2953

(a)
Fig. 4. Validation result with Wang et al.’s [17] correlation.

fv D
S= (10)
u

where S, fv, and D represent the Strouhal number, vortex shed-

ding frequency, and tube diameter, respectively. The vortex
shedding frequency may then be obtained by substituting the
values of S, u, and D into Eq. (10). A transient simulation was
performed for two period times, but no vortex shedding was
observed. Thus, subsequent calculations only determined the
values for the steady state.
Fig. 4 shows the j-factor of the plain fin-tube heat ex- (b)
changer evaluated by the CFD analysis, which coincides with
that estimated by the empirical correlation of Wang et al. [17]. Fig. 5. Velocity vector of air at mid-plane (Re = 1295, Vmax = 8.07
m/s): (a) Plain fin-tube heat exchanger; (b) DWVG.
This comparison proves that the present numerical models
are appropriate.
erated in counter-clockwise at the left of the DWVG. At the
same time, a secondary vortex flows at the upper vortex be-
3. Numerical result tween the DWVG and the right side wall. These two vortex
Fig. 5 shows the air flow velocity vector at the mid-plane flows combine and form one counter-clockwise vortex at the
between two fins when the frontal air velocity is 5 m/s. The air end part of the DWVG as the separation wall disappears. This
enters from the left-hand side and flows around the heat trans- vortex flow develops at the rear, and thus, influences separa-
fer tube. The air flow then separates from the tube wall and tion delays and reduces drag across the tube. Finally, the flow
develops a wake region behind the tube, and enters the fin- removes the poor heat transfer zone from the wake zone of the
tube heat exchanger where the flow slows down and circulates tube [4]. Moreover, the vortex mixes the flow, which im-
[14]. For a plain fin heat exchanger, as seen in Fig. 5(a), the proves convective heat transfer.
wake is wide and extends to the tailing edge of the fin. For a Fig. 7 shows the static pressure contour for the same case.
DWVG fin, as seen in Fig. 5(b), the flow velocity increases The arrows indicate the adverse pressure gradient points
between the DWVG and the heat transfer tube. Then the flow where the pressure on the cylinder wall is at minimum. Th is
separation point moves to the rear side of the tube, where in point mainly affects the location of flow separation in the
this case, a narrower and shorter wake region still develops. downstream. The static pressure at the tube wall decreases
The wake region appears to be extended lengthwise to the from the fore-end of the tube to the arrowhead. However,
middle of the second row of tubes. Fig. 5(b) shows that the air static pressure recovers at positions further on the rear. Fig. 7
flow accelerates between the DWVG and tube wall, conse- shows that the DWVG influences the adverse pressure gradi-
quently forming a delayed, adverse pressure gradient. Fur- ent point, which occurs at around 94.6° for the plain fin and at
thermore, the flow separation occurs further downstream. This around 100.4° for the DWVG fin from the tube fore-end. This
delayed flow separation diminishes the wake region and in- change in the adverse pressure gradient point influences the
creases the uniformity of the overall velocity profile. variation of the wake region, as shown in Fig. 5.
Fig. 6 shows the velocity vector plots of the frontal section. Fig. 8 shows the static pressure distribution at the mid-plane
The DWVG creates the vortex. In Fig. 6(b), the vortex is gen- between two adjacent fins when the frontal air velocity is 15
2954 S. W. Hwang et al. / Journal of Mechanical Science and Technology 26 (9) (2012) 2949~2958

(a) (b)

Fig. 6. Velocity vector plot of vertical section (Re = 1295, Vmax = 8.07 m/s): (a) velocity vector plots nearby tube; (b) velocity vector plot on a cross
section with the DWVG.

(a) (b)

Fig. 7. Static pressure contour at mid-plane and the adverse pressure gradient point on the cylinder (Re = 1295, Vmax = 8.07 m/s): (a) Plain fin tube
heat exchanger; (b) DWVG fin tube heat exchanger.

(a) (b)

Fig. 8. Static pressure contour at mid-plane and the adverse pressure gradient point on the cylinder (Re = 5180, Vmax = 32.3 m/s): (a) Plain fin tube
heat exchanger; (b) DWVG fin tube heat exchanger.

m/s. Fig. 8(a) shows the case of the plain fin, where the ad- larly occurs at a smaller angle compared with that of air veloc-
verse pressure gradient point shifted to the front compared ity at 5 m/s depicted in Fig. 7(b). However, the size of wake
with that for a frontal air velocity of 5 m/s. The wake region region is smaller than that of the plain fin, proving that the
thus widened and elongated. Fig. 8(b) shows the case of a effect of DWVG is still advantageous.
DWVG fin, where the adverse pressure gradient point simi- Fig. 9 shows the temperature distribution of air when its
 S. W. Hwang et al. / Journal of Mechanical Science and Technology 26 (9) (2012) 2949~2958 2955

(a) (b)

Fig. 9. Air temperature at mid-plane (Re = 1295, Vmax = 8.07 m/s): (a) Plain fin tube heat exchanger; (b) DWVG fin tube heat exchanger.

(a) (b)

Fig. 10. Fin surface temperature (Re = 1295, Vmax = 8.07 m/s): (a) Plain fin tube heat exchanger; (b) DWVG fin tube heat exchanger.

frontal velocity is 5 m/s. This plot represents the mid-plane Fig. 10 shows the surface temperature distributions of the
between two adjacent fins. Fig. 9(a) shows that in the plain fin DWVG and plain fins. Both corners of the front end show low
heat exchanger, the temperature of the wake region is signifi- temperatures due to the higher heat transfer rate, which is
cantly higher than that of the free-stream air flowing outside. caused by the higher velocity compared to the central region.
This indicates that the free-stream air and the air in the wake As the location becomes closer to the tube, the fin surface
region do not mix well, which is caused by the swirling char- temperature rises due to heat conduction through the fin itself.
acteristic of the wake as shown in the velocity vector plot of Moving away from the tube, the surface temperature declines
Fig. 5. When the DWVG are mounted on the plain fin, how- and the air velocity rises. As noted above, the heat transfer
ever, this high temperature region is reduced, as previously performance in the wake region is poor. Consequently, the fin
mentioned and shown in Fig. 9(b). The contour pattern of the surface temperature appears high at the rear of the heat trans-
air temperature distribution is quite similar to the velocity fer tubes.
vector plot shown in Fig. 5. The similarity is because the air Fig. 11 shows that this high temperature region extends to-
temperature is directly associated with convective heat trans- ward the rear end of the fin. The temperature levels around the
fer, which is controlled by the air flow itself. Early research second row tubes of the plain and DWVG fin are quite similar.
proved that the convective heat transfer is a function of Rey- However, those around the first row tubes are largely different
nolds number (Re) and Prandtl number (Pr), which explains due to the DWVG influence. The temperature level at the
the greater uniformity of the air temperature distribution of the back of the first row tube of the DWVG fin is lower than that
DWVG fin heat exchanger in Fig. 10(b), compared with that of the plain fin, because the DWVG reduced the wake region
of the plain fin heat exchanger. Fig. 11 provides a clear illus- and facilitated air mixing, causing an enhanced heat transfer.
tration through the plot of the temperature and velocity varia- Considering these observations, the DWVG can enhance con-
tion along the central line and across the second row of the vective heat transfer over a fin by reducing the wake region.
tube. Fig. 11 shows that the DWVG causes a flatter tempera- Fundamental thermodynamic variables, the values of which
ture and velocity profiles at the rear of the heat transfer tube. are obtained via CFD analyses, are related with system pa-
2956 S. W. Hwang et al. / Journal of Mechanical Science and Technology 26 (9) (2012) 2949~2958

Fig. 11. Temperature and velocity of air along the line between the Fig. 13. Comparison of f-and j-factor ratios.
tubes of the 2nd row.

DWVG would also obstruct flow to produce an additional form

loss. In the present work, however, an additional effect occurred
apart from the vortex generation: reduction of the wake region
due to flow acceleration between the tube and the DWVG itself.
This wave region reduction decreased the form loss. Therefore,
the DWVG plays a dual role of increasing and reducing form
loss due to the wake. As the Reynolds number increases, the
effect of the form loss reduction becomes dominant over that of
additional form loss. This explains how the f-factor of the
DWVG fin becomes smaller than that of the plain fin.
Fig. 13 shows the ratio of the f- and j-factors of the DWVG
fin compared with that of the plain fin. The f-factor ratio in-
creases in a low Reynolds number range, but decreases as the
Fig. 12. Variation of f-and j-factors. Reynolds number increases. The j-factor ratio suddenly in-
creases to 1.09 and do not change much thereafter. Another
interesting plot is the j-factor ratio over the f-factor ratio. The
rameters such as the heat transfer and heat loss coefficients. Reynolds analogy states that an enhanced surface would have
Fig. 12 shows the variation of the j- and f-factors with the a value of ‘1’. The value obtained in previous fin design stud-
frontal air velocity in a range of 0.5–20 m/s, and represented ies was frequently less than 1. However, Fig. 13 shows that
in terms of Reynolds number. For the conversion, the hydrau- the ratio of the present DWVG fin increases beyond ‘1’ as the
lic diameter was used as the characteristic length scale, and is Reynolds number increases, implying that the present DWVG
equivalent to double the fin pitch. The j-factor for the DWVG fin design will become more beneficial as the frontal air veloc-
fin is larger than that of the plain fin over the whole range. ity increases.
Moreover, the f-factor variation showed an interesting trend. Fig. 14 shows the variation of the area goodness factor with
The f-factor of the DWVG fin is higher than that of the plain increasing Reynolds number. The area goodness factor is rep-
fin, but the difference diminishes with the increase in Rey- resented by heat transfer rate per frontal unit area, and is given
nolds number in a range of Re < 2500. Beyond Re = 2500, the as follows:
f-factor of the DWVG fin becomes smaller than that of the
plain fin. In a range of Re > 200, the j-factor of the DWVG fin  Pr 2 / 3 NTU × m2 
j 1
remains larger than that of the plain fin. These results contra- = 2  . (11)
f Af  2ρ ∆P 
dict the Reynolds analogy, which states that an enhanced sur-  
face wall has a larger f-factor, as delineated by Eq. (9). How-
ever, this analogy should not be expected to hold in the pre- A high area goodness factor means that a much smaller
sent situation; it is applicable to laminar and turbulent flows frontal area is necessary. In Fig. 14, the heat transfer perform-
over a flat plate and a turbulent flow in a tube, but not during ance is better than that of the plain fin heat exchanger when
flow separation. If the tube and the wake region developed at the Reynolds number is over 500, coinciding with the findings
the rear of the tube are not considered, the Reynolds analogy of Joardar and Jacobi [9].
will become applicable. That is, the DWVG would generate Fig. 15 shows the variation of volume goodness factor with
longitudinal vortices and mix air to promote heat transfer. The increasing Reynolds number. The volume goodness factor is
 S. W. Hwang et al. / Journal of Mechanical Science and Technology 26 (9) (2012) 2949~2958 2957

The flow acceleration between the delta winglet vortex gen-

erator and heat transfer tube delays the flow separation from
the tube, which reduced the wake region at the rear of the tube.
Thus, form loss is reduced and heat transfer is enhanced. In-
terestingly, the pressure loss of a fin with delta winglet vortex
generators was even smaller than that of a plain fin with the
enhanced heat transfer at high air velocity or Reynolds num-
ber. Comparing the results of the f- and j-factors, the perform-
ance of the DWVG fin improved as the Reynolds number
increases in terms of enhanced heat transfer and reduced pres-
sure drop penalty.

Acknowledgement
Fig. 14. Comparison of area goodness factor.
This work was supported by a 2-year research grant from
the Pusan National University.

Nomenclature------------------------------------------------------------------------
a : Attack angle (°)
Ac : Area of fin collar (m2)
Amin : Area of minimum air flow (m2)
Af : Area of frontal air flow (m2)
Afin : Area of fin (m2)
b : Central angle (°)
Cf : Friction coefficient
Cp : Static pressure specific heat (kj/kg*K)
D : Tube diameter (m)
Fig. 15. Comparison of volume goodness factor. Dc : Characteristic length (m)
f : Fanning friction factor
represented by the heat transfer rate and pumping power per fv : Vortex shedding frequency (Hz)
unit area, and is expressed as: g : Between tube and winglet gap (m)
h : Heat transfer rate (w/m2*k)
St H : Fin pitch (m)
Volume goodness factor = . (12)
f 1/ 3 j : Colburn j factor
k : Conductivity (W/m*K)
In Fig. 15, a high volume goodness factor needs a small L : Length of heat exchanger (m)
volume to show the same performance in terms of heat trans- P : Pressure (pa)
fer rate [18]. Using the DWVG causes a high volume good- Q : Total heat flow (w)
ness factor. Consequently, a smaller volume is required for the t : Time (s)
same pumping power. tth : Thickness (m)
T : Temperature (K)
Tc : Temperature of fin collar (K)
4. Concluding remarks Tin : Inlet mean temperature (K)
In the past decade, numerous studies investigated the effect Tout : Outlet mean temperature (K)
of vortex generators on the heat transfer of fins. While the u : Velocity of x-axis (m/s)
vortex generators were consistently reported to enhance the U : Frontal fluid velocity (m/s)
heat transfer performance, the pressure loss measurements did v : Velocity of y-axis (m/s)
not yield agreement. Despite significant research in this area, a Vin : Frontal inlet velocity (m/s)
lack of understanding remains on the mechanisms contribut- Vmax : Maximum velocity (m/s)
ing to heat transfer augmentation and a limited increase in W : Width of heat exchanger (m)
pressure loss. In the present work, the air flow and heat trans- Wh : Winglet height (m)
fer in a fin-tube heat exchanger were analyzed using the CFD Wl : Winglet length (m)
to obtain a better phenomenological understanding of the ef- α : Thermal diffusivity (m2/s)
fects of the delta winglet vortex generator on fin performance. µ : Dynamic viscosity (N*s/m2)
2958 S. W. Hwang et al. / Journal of Mechanical Science and Technology 26 (9) (2012) 2949~2958

ρ : Density (kg/m3) [14] F. P. Incropera, D. P. Dewitt, T. L. Bergman and A. S. Lavine,

ν : Kinetic viscosity (m2/s) Introduction to heat transfer, 5th edition. Wiley (2007).
Nu : Nusselt number [15] S. Kakac, H. Liu, Heat exchangers, CRC Press (1998).
Pr : Prandtl number [16] K. J. Kim and P. A. Durbin, Observations of the
Re : Reynolds number (ρ*Vmax*Dh/ µ) frequencies in a sphere wake and drag increase by acoustic
S : Strouhal number excitation. Physics of Fluids, 31 (1998) 3260-3265.
St : Stanton number [17] C. C. Wang, K. Y. Chi and C. J. Chang, Heat transfer and
friction characteristics of plain fin-and-tube heat exchangers,
part Ⅱ: Correlation. International Journal of Heat and
References
Mass Transfer, 43 (2000) 2693-2700.
[1] K. Chang and P. Long, An experimental study of the airside [18] R. K. Shah and K. London, Laminar flow forced convec-
performance of slit fin-and-tube heat exchangers under dry tion in ducts, Academic Press, 1978.
and wet conditions, International Journal of Air-
Conditioning and Refrigeration, 17 (1) (2009) 7-14. Seong Won Hwang received his Mas-
[2] G. B. Schubauer and W. G. Spangenberg, Forced mixing in ter’s degree for Mechanical Engineering
boundary layers, Journal of Fluid Mechanics, 8 (1960) 10-31. at the Pusan National University in 2010.
[3] M. Fiebig, Vortices, generators and heat transfer, Trans He is currently a Ph.D. candidate. His
Institution of Chemical Engineers, 76: Part A (1998). research interests include heat exchanger
[4] K. Torii, K. M. Kwak and K. Nishino, Heat transfer en- performance tests and CFD analyses. He
hancement accompanying pressure-loss reduction with is currently working on heat exchangers
winglet-type vortex generators for fin-tube heat exchanger, with offset strip fins and vortex generators.
International Journal of Heat and Mass Transfer, 45 (2002)
3795-3801. Dong Hwan Kim received his Master’s
[5] A. M. Jacobi and R. K. Shah, Heat transfer surface en- degree for Mechanical Engineering at
hancement through the use of longitudinal vortices: A re- the Pusan National University in 2011.
view of recent progress, Experimental Thermal and Fluid He works on heat transfer enhancement
Science, 11 (1995) 295-309. and its application to energy systems.
[6] M. Fiebig, A. Valencia and N. K. Mitra, Wing-type vortex
generators for fin-and-tube heat exchangers, Experimental
Thermal and Fluid Science, 7 (1993) 287-295.
[7] K. M. Kwak, K. Torii and K. Nishino, Heat transfer and
pressure loss penalty for the number of tube rows of stag- June Kee Min studied in KAIST and
gered finned-tube bundles with a single transverse row of received his Ph.D. in 1999. He worked
winglets, International Journal of Heat and Mass Transfer, for LG Electronics and Samsung Elec-
46 (2003) 175-180. tronics as a thermo-fluid engineer for
[8] K. M. Kwak, K. Torii and K. Nishino, Simultaneous heat over 10 years. He joined the Rolls-Royce
transfer enhancement and pressure loss reduction for finned- Pusan National University Technology
tube bundles with the first or two transverse rows of built-in Center in 2008. His research interests
winglets, Experimental Thermal and Fluid Science, 29 mainly include CFD-based fluid flow and
(2005) 625-632. heat transfer problems, such as heat exchangers, HVAC&R, and
[9] A. Joardar and A. M. Jacobi, Heat transfer enhancement by the thermal management of power storage systems.
winglet-type vortex generator arrays in compact plain-fin-
and-tube heat exchanger, International Journal of refrigera- Ji Hwan Jeong received his Ph.D. from
tion, 31 (2008) 87-97. the KAIST in 1995. His post-doctoral
[10] C. B. Allison and B. B. Dally, Effect of a delta-winglet research was conducted at the Technol-
vortex pair on the performance of a tube-fin heat exchanger, ogy Centre in Aerodynamics and Heat
International Journal of Heat and Mass Transfer, 50 (2007) Transfer at Oxford University. After
5065-5072. three years of research at the Korea
[11] J. Min and W. Xu, Numerical prediction of the perform- Atomic Energy Research Institute, he
ances of the fins with punched delta winglets and the louver joined the faculty of the School of Me-
fins and their comparison, Journal of Enhanced Heat Trans- chanical Engineering at Pusan National University. His research
fer, 12 (4) (2005) 357-371. interests include multi-phase flow, nuclear safety, heat exchang-
[12] B. R. Munson, D. F. Young and T. H. Okiishi, Fundamen- ers, heat pump, and computational fluid dynamics. He has also
tals of fluid mechanics, 4th edition. Wiley (2002). consulted for the nuclear and air conditioner industries.
[13] CD-adapco, Methodology manual, STAR-CD (2006).