Anda di halaman 1dari 10

Experimental Thermal and Fluid Science 33 (2009) 818–827

Contents lists available at ScienceDirect

Experimental Thermal and Fluid Science


journal homepage: www.elsevier.com/locate/etfs

Experimental and numerical investigation on air-side performance


of fin-and-tube heat exchangers with various fin patterns
L.H. Tang, M. Zeng, Q.W. Wang *
State Key Laboratory of Multiphase Flow in Power Engineering, Xi’an Jiaotong University, Xi’an, Shaanxi 710049, PR China

a r t i c l e i n f o a b s t r a c t

Article history: Air-side heat transfer and friction characteristics of five kinds of fin-and-tube heat exchangers, with the
Received 10 July 2008 number of tube rows (N = 12) and the diameter of tubes (Do = 18 mm), have been experimentally inves-
Received in revised form 16 January 2009 tigated. The test samples consist of five types of fin configurations: crimped spiral fin, plain fin, slit fin, fin
Accepted 25 February 2009
with delta-wing longitudinal vortex generators (VGs) and mixed fin with front 6-row vortex-generator
fin and rear 6-row slit fin. The heat transfer and friction factor correlations for different types of heat
exchangers were obtained with the Reynolds numbers ranging from 4000 to 10000. It was found that
Keywords:
crimped spiral fin provides higher heat transfer and pressure drop than the other four fins. The air-side
Fin-and-tube heat exchanger
Evaluation criteria
performance of heat exchangers with the above five fins has been evaluated under three sets of criteria
GA optimization and it was shown that the heat exchanger with mixed fin (front vortex-generator fin and rear slit fin) has
better performance than that with fin with delta-wing vortex generators, and the slit fin offers best heat
transfer performance at high Reynolds numbers. Based on the correlations of numerical data, Genetic
Algorithm optimization was carried out, and the optimization results indicated that the increase of VG
attack angle or length, or decrease of VG height may enhance the performance of vortex-generator fin.
The heat transfer performances for optimized vortex-generator fin and slit fin at hand have been com-
pared with numerical method.
Ó 2009 Elsevier Inc. All rights reserved.

1. Introduction number of tube rows and the outside diameter of tubes are large.
Therefore, it is essential to investigate the heat transfer and friction
Fin-and-tube heat exchangers are extensively employed in characteristics of fin-and-tube heat exchangers with large number
chemical engineering, refrigeration, and HVAC (heating, ventilation of tube rows and large tube diameter. In the present study, the air-
and air conditioning) applications such as compressor intercoolers, side heat transfer and friction performance of fin-and-tube heat
air-coolers and fan coils. The dominant thermal resistance is usu- exchangers with large number of tube rows (N = 12) and large tube
ally on the air side in practical applications, and therefore the diameter (Do = 18 mm) was experimentally studied and compared
use of finned surfaces on the air side is very common to effectively under three sets of criteria.
improve the overall thermal performance of heat exchangers. In recent years, more and more applications of Genetic Algo-
These finned surfaces include crimped spiral fin, plain fin, slit fin rithms (GAs) into heat exchangers optimizations have suggested
and fin with delta-wing longitudinal vortex generator and so on. that GAs have a strong ability of auto-search and combination opti-
Recently, the air-side performance was addressed by Nuntaphan mization and can successfully optimize and predict thermal prob-
et al. with respect to the crimped spiral fin [1,2]. For plain fin lems [18–20]. At this point, the GA technique may be used in the
geometry, McQuiston [3], Gray and Webb [4] and Wang et al. [5– geometrical optimization of heat exchangers in order to obtain
7] established extensive data and correlations. During the past optimal results under specified design objectives within the allow-
few years, there are also many investigations on the slit fin pattern able pressure drops. In this paper, a Genetic Algorithm based opti-
[8–11] and on the longitudinal vortex-generator fins [12–17]. mization technique has been conducted for heat exchanger having
However, these studies focused only on small numbers of tube vortex generators, and the heat transfer performances for opti-
rows (mostly being <4) or small tube diameter (mostly covering mized vortex-generator fin and slit fin at hand was also compared.
range of 8–13 mm), which are usually used in the condensers or
evaporators in refrigeration engineering. However, in some appli- 2. Experimental system and procedure
cations, such as the intercooler of multi-stage compressor, the
The experiments are conducted in an open wind tunnel. The
* Corresponding author. Tel./fax: +86 029 8266 3502. system consists of two loops: air loop and steam loop. The air loop
E-mail address: wangqw@mail.xjtu.edu.cn (Q.W. Wang). is provided to blow air across the finned bundles of test core, and

0894-1777/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.expthermflusci.2009.02.008
L.H. Tang et al. / Experimental Thermal and Fluid Science 33 (2009) 818–827 819

Nomenclature

c1, c2, c3, c4 coefficient of formulation v velocity, m  s1


A area, m2 vfr air frontal velocity, m  s1
Afr air frontal area, m2 V12 abbreviation of vortex-generator fin with N = 12
Amin minimum flow area, m2 Vh height of vortex generator, mm
Ao total surface area, m2 Vl length of vortex generator, mm
cp specific heat at constant pressure, J  kg1  K1 VSM12 abbreviation of mixed fin with front 6-row vortex-gen-
Dc fin collar outside diameter, Dc = Do + 2d, mm erator fin and rear 6-row slit fin
Di inside diameter of tube, mm
Do outside diameter of tube, mm Greek symbols
f friction factor, f ¼ 2DpDc =ðqv 2max LÞ q density, kg  m3
Fp fin pitch, mm k thermal conductivity, W  m1  K1
Fs fin spacing, mm l dynamic viscosity of fluid, kg  m1  s1
g gravity acceleration, m  s2 g fin efficiency
h heat transfer coefficient, W  m2  K1 go surface efficiency
H length of tube, mm d fin thickness, mm
j Colburn factor U heat transfer rate, W
L length of fin, mm a angle of attack, degree
m mass flow rate, kg  s1 r contraction ratio of fin array
N number of tube rows
Nu Nusselt number, Nu = h Dc/k Subscripts
Dp pressure drop, Pa air air side
P12 abbreviation of plain fin with N = 12 ave average value
Pl longitudinal tube pitch, mm b base surface
Pt transverse tube pitch, mm c compared fin
Q heat transfer rate, W f fin surface
r latent heat of vaporization, J  kg1 i tube inside
rc fin collar outside radius, Dc/2, mm in air-side inlet
Re Reynolds number, Re = q v Do/l max maximum value
ReDc Reynolds number, ReDc = q v Dc/l o tube outside
sh height of slit or spiral fin, mm out air-side outlet
sl length of slit, mm p plain fin
sw width of slit, mm r reference fin
S12 abbreviation of slit fin with N = 12 s saturated steam
SP12 abbreviation of spiral fin with N = 12 w tube wall
t temperature, K water condensation water
Dt logarithmic-mean temperature difference, K steam steam side
U overall heat transfer coefficient, W  m2  K1

the steam loop is designed to supply slightly superheated stream gauges (at small flow rate) or U-tube water column manometer
through the tubes of test core. The extended (finned) surfaces are (at high flow rate).
prepared for the test core, which are placed in the test section. Before recording the data in the experiments, much attention is
The steam–air system is employed for accomplishment of steam- paid on the thermal equilibrium to ensure the steady state of heat
to-air heat exchange, as schematically shown in Fig. 1. exchange. The steam gauge pressure is maintained around
Air is induced to the wind tunnel by a centrifugal blower. The 200 mmHg (mercury column), and the potential of thermal couple
inlet air temperature and the temperature difference between in- varies around 0.01 mV. In the data-acquisition procedure, each
let and outlet through the test core are measured by two sets of measured value is read five times in 10 min, and the arithmetic
multi-point T-type copper-constantan thermocouple grids. Each mean of the recorded data is used for checking the heat balance be-
set contains 16 calibrated thermocouples within the accuracy of tween the energy gain of air and the energy reduction of the con-
0.1 °C, and the junctions of thermocouples are connected in series densing steam. In all tests, the thermal equilibrium between air
to give a single reading. Steam is generated in an electrically side and steam side is within 5%.
heated boiler, of which the power can be adjusted by six trans- Five kinds of fin-and-tube heat exchangers are tested in this
formers. After flowing through the over-heater, the steam is study. The slit fin pattern in this paper was popularly investi-
superheated to provide the desired 1–3 °C of superheat. The gated, and the geometry of slit fin is based on the available re-
superheated steam temperature is directly read with an ethanol ported results by Wang et al. [5–7]. The geometry of VGs is
thermometer, which is inserted in the hole that designed in the based on the available reported results by Fiebig [21]. These re-
steam inlet header. sults [21] show that the winglet-type VGs are superior to wing-
The wall static pressures before and after the test cores are type VGs, and suggest that the geometry of winglet-type VG
measured by a U-tube water column manometer, and the super- should be right triangle with the ratio of base edge to height
heated steam pressure is read by a U-tube mercury column equals 2. And Pesteei et al. [15] indicated that the best attack an-
manometer. The air velocity is measured by a Pitot-tube meter, gle of the winglet-type VGs is 45°. The detailed geometrical
which is located in the flow metering duct far downstream of the parameters are tabulated in Table 1 and schematically shown in
test core. The Pitot-tube meter is connected to the inclined draft Fig. 2. All tubes and fins are made of copper. The test samples
820 L.H. Tang et al. / Experimental Thermal and Fluid Science 33 (2009) 818–827

1-Entrance; 2-Transition section; 3-Contraction section; 4-Straightening section; 5-Test heat exchanger;
6-Flow metering duct; 7-Blower; 8-Boiler; 9-steam over-heater; 10-Inclined tube draft gauge;
11-Volumetric meter; 12-Digital voltmeter; 13-Ther mocouples grid; 14-Ice bath; 15-U tube manometer;
16- valve

Fig. 1. Schematic of experimental facility.

Table 1
Geometric dimensions of fin-and-tube heat exchangers.

Name Fin type Di Do Dc Fp Pt Pl sw sl sh Vl Vh a N


P12 Plain 16 18 18.6 3.1 42 34 – – – – – – 12
S12 Slit 16 18 18.6 3.1 42 34 2.2 16 1.0 – – – 12
V12 Vortex 16 18 18.6 3.1 42 34 – – – 5 2.55 45 12
VSM12 Mixed 16 18 18.6 3.1 42 34 2.2 16 1.0 5 2.55 45 12
SP12 Spiral 16 18 18 3.1 42 34 – – 10 – – – 12

Note: (1) Fin thicknesses of all test samples are 0.3 mm; (2) VSM12 is mixed by front 6-row vortex-generator fin and rear 6-row slit fin.

are covered by the foam insulation of about 20 mm thick to re- where Dtm is the logarithmic-mean temperature difference, defined
duce the heat loss to the surroundings. by

ðt s  t in Þ  ðt s  t out Þ
3. Data reduction Dt m ¼ ð5Þ
ln ttsst
t in
out

The main purpose of data reduction is to determine the air side


heat transfer and friction characteristics, Nusselt number (Nu) or where tin is the inlet temperature of air, tout is the outlet tempera-
Colburn factor (j) and friction factor (f) of the heat exchangers from ture, and ts is the saturated temperature of steam at the correspond-
the experimental data, which are recorded at steady-state condi- ing pressure.
tions during each test run, and to find out corresponding power- The overall heat transfer resistance can be defined as
law correlations of Nu vs. Re, and f vs. Re for each case. The average 1 1 1 Do Do 1
¼ þ ln þ : ð6Þ
value of the inlet and outlet temperatures of air side is used to UA go ho Ao 2 kw Aw Di hi Ai
evaluate the thermal properties of air, and the thermal properties
of steam are determined by the steam pressure. The steam-side heat transfer coefficient, hi, is evaluated from
The saturated steam gauge pressure in the tubes is around the Nusselt correlation [22]
200 mm Hg. The steam-side heat transfer rate Q steam is given as !1=4
gr steam qwater ðqwater  qsteam Þk3water
Q steam ¼ msteam rsteam ð1Þ hi ¼ 1:13 : ð7Þ
lwater ðts  tw ÞH
where msteam is the vapor mass flowrate, and r steam is the latent heat
In Eq. (6), go is the finned surface efficiency, which may be writ-
of steam at the corresponding pressure. The air-side heat transfer
ten in terms of the fin efficiency g, fin surface area Af and total sur-
rate Q air is given as
face area Ao, as follows:
Q air ¼ mair cp;air Dtair ð2Þ
Af
go ¼ 1  ð1  gÞ ð8Þ
where mair is the air mass flowrate. Ao
The total heat transfer rate is defined as the average of the air
side and the steam-side heat transfer rates where Ao = Af + Ab, Af and Ab are the areas of the fin and base surface,
respectively. g is calculated by the approximation method de-
Q ave ¼ ðQ steam þ Q air Þ=2: ð3Þ scribed by Schmidt [23]
The total heat transfer coefficient, UA product, is calculated tanhðmr c xÞ
from the following relationship: g¼ ð9Þ
mr c x
UA ¼ Q ave =Dt m ð4Þ where
L.H. Tang et al. / Experimental Thermal and Fluid Science 33 (2009) 818–827 821

Z
Y
o
X

(a) Crimped spiral fin (b) Plain fin

Z Z
Y Y
o o
X X

(c) Slit fin (d) Vortex-generator fin

Z
Y
o
X

(e) Mixed fin


Fig. 2. Fin configurations.

Table 2
sffiffiffiffiffiffiffiffi
Summary of estimated uncertainties.
2ho
m¼ ð10Þ Primary measurements Derived quantities
kf df
Parameter Uncertainties Parameter Uncertainties (%)
   
R R mair 4.24% Qair 5.83
x ¼ eq  1 1 þ 0:35 loge eq ð11Þ msteam 3.6% Qsteam 4.69
rc rc
tin 0.1 °C Re 4.59
 0:5 tout 0.1 °C Nu 6.48
Req XM XL Dp 3% f 8.69
¼ 1:27  0:3 ð12Þ
rc rc X M
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
XL ¼ ðPt =2Þ2 þ P 2l =2 ð13Þ

X M ¼ Pt =2: ð14Þ 1000


Δp, 9-row bare tube heat exchanger
The air-side heat transfer coefficient ho and the surface effi-
Δp, Jakob's relation[25]
ciency go can be acquired through solving Eqs. (9)–(14) with an
Δp / Pa

iterative method. Δp
The heat transfer and friction characteristics of the heat exchan-
ger are presented in the following dimensionless forms:
100
Nu ¼ ho Dc =k ð15Þ

ReDc ¼ qv max Dc =l
Nu

ð16Þ
Nu
  Nu, 9-row bare tube heat exchanger
j ¼ Nu= ReDc Pr1=3 ð17Þ Nu, Zukauskas's relation[24]

2Dp Dc 10
f ¼ ð18Þ 5000 6000 7000 8000 9000 10000
qv 2max L
ReDc
where vmax is the velocity at the minimum free flow area, vmax = vfr/
r. The term r is the ratio of the minimum flow area to frontal area. Fig. 3. Nu and Dp vs. ReDc for validation.
822 L.H. Tang et al. / Experimental Thermal and Fluid Science 33 (2009) 818–827

The accuracies of the measurements and the uncertainties in de- 10000


rived experimental values are given in Table 2. P12
S12
4. Experimental results for five fin patterns V12
VSM12
4.1. Experimental system validation 1000 SP12
Δp

Δp / Pa
In order to verify the experimental system, a 9-row bare tube
heat exchanger without fins is firstly run through the above-men-
tioned experimental system. Fig. 3 shows the variation of Nusselt
number and Dp with Reynolds number. The maximum deviation 100
of the Nusselt number between the present data and well-known Nu

Nu
correlation of Zukauskas [24] is 9%, and the average deviation is
5.8%. The maximum deviation of Dp between the present data
and Jakob’s correlation [25] is 14.1%, and the average deviation is
11.8%. According to the comparison results, the present experi-
10
mental system and method are reliable. 2.5 3 3.5 4 4.5 5 5.5
-1
vfr / m.s
4.2. Comparison of heat transfer performance between small tube
diameter and large tube diameter Fig. 5. Nu and Dp for 12-row tested samples.

The experimental correlations for air-side heat transfer and fric-


tion performance of plain fin-and-tube heat exchanger with small
10 P12
number of tube rows (N 6 4) and small tube diameter
S12
(Do 6 12 mm) was compared with the present data of plain fin in
V12
this paper. Fig. 4 shows the comparisons between the experimental
VSM12
correlations and the present data of plain fin. The maximum devi-
SP12
ation and the minimum deviation of the Nusselt number between
the present data and the experimental correlations [26] are 16.4%
1
f

and 13.1% respectively, while those between the present data f


and the experimental correlations [7] are 25.1% and 4.1%, respec-
tively, and those between the present data and the experimental
correlations [27] are 10.1% and 11.1%, respectively. The average
deviations of Dp between the present data and the experimental
correlations in above references [7,26,27] are 26.5%, 19.9% and
j

14.5%, respectively. That is to say, the experimental correlations 0.1


j
of plain fin-and-tube heat exchanger with small tube diameter
and small number of tube rows can not be directly extended to de- 2.5 3 3.5 4 4.5 5 5.5
sign the fin-and-tube heat exchanger with large tube diameter and -1
vfr / m.s
larger number of tube rows.
Fig. 6. j and f for 12-row tested samples.
4.3. The effect of fin patterns

Figs. 5 and 6 show Nu, Dp, j-factor and f-factor with air frontal
velocity for various fin patterns. In these figures, ‘‘P12” stands for plain fin with ‘‘N = 12”, ‘‘S12” stands for slit fin with ‘‘N = 12”,
‘‘V12” stands for vortex-generator fin with ‘‘N = 12”, ‘‘VSM12”
stands for mixed fin with front 6-row vortex-generator fin and rear
6-row slit fin with ‘‘N = 12”, and ‘‘SP12” stands for spiral fin with
present experimental results of P12 ‘‘N = 12”, respectively. Both Nusselt number and Dp increase with
1000 experimental correlation [26] increasing air frontal velocity, and both j-factor and f-factor de-
experimental correlation [7] crease with the increase of air frontal velocity. In Fig. 5, the Nusselt
Δp / Pa

experimental correlation [27] Δp number of SP12 is the highest among the five fins at the same fron-
tal velocity, and that of S12 takes the second place, while that of
P12 is the lowest. On the other hand, Fig. 6 shows that j-factor of
SP12 is the highest among these fins at the same frontal velocity,
100 and that of P12 are the lowest. From Fig. 5, Dp of SP12 is the high-
est, while Dp of P12 is the lowest. The f-factor of SP12 is obvious
Nu the highest among these fins at the same frontal velocity in
Nu

Fig. 6. In a sense, these results can be expected, because the


enhancement of heat transfer is usually penalized by the increase
of pressure drop.
10 The correlations for heat transfer and friction factors can be ex-
4000 6000 8000 10000
pressed as follows:
ReDc
Nu ¼ c1 RecD2c ;
Fig. 4. Nu and Dp comparison between small tube diameter and large tube
diameter.
f ¼ c3 RecD4c ð19Þ
L.H. Tang et al. / Experimental Thermal and Fluid Science 33 (2009) 818–827 823

Table 3
The correlations of Nusselt number and friction factor.

Name Range of ReDc Nu = f(Re) Maximum relative error (%) f = f(Re) Maximum relative error (%)
0.71 0.36
P12 4000–10000 Nu = 0.080 ReDc 6.4 f = 12.83 ReDc 4.9
S12 4000–10000 Nu = 0.057 ReDc0.77 4.6 f = 13.28 ReDc0.35 2.5
V12 4000–10000 Nu = 0.094 ReDc0.71 4.0 f = 10.68 ReDc0.34 3.3
VSM12 4000–10000 Nu = 0.076 ReDc0.74 3.6 f = 6.15 ReDc0.27 2.9
SP12 4000–10000 Nu = 0.069 Re0.76 1.6 f = 4.31 Re0.13 1.1

The corresponding correlations for five different fin patterns in


1.3
the present study are shown in Table 3. These correlations can be S12
referred to engineering applications or further researches such as V12
optimization or prediction. 1.2 VSM12
SP12
4.4. Heat transfer performance comparison

Φc/Φp
1.1
As mentioned above, the heat exchanger with crimped spiral fin
provides the highest Nusselt number associated with the highest
pressure drop. Accordingly, it is essential to compare the heat
transfer enhancement performance of the test heat exchangers. 1
In the present study, the identical mass flow rate criteria, the iden-
tical pumping power criteria and the identical pressure drop crite-
ria are used. These criteria were successfully adopted by Yu et al.
[28] and Wang et al. [29]. Based on the constant properties 0.9
assumption, the formulations of these criteria are given as: 4000 6000 8000 10000
ReDc
(a) identical mass flow rate criteria: (a) identical mass flow rate criteria
ðReA=Dc Þc ¼ ðReA=Dc Þp ð20Þ 1.2
(b) identical pumping power criteria:
3 3
ðfRe A=D4c Þc ¼ ðfRe A=D4c Þp ð21Þ

(c) identical pressure drop criteria: 1


S12
Φc/Φp

2 2
ðfRe =D3c Þc ¼ ðfRe =D3c Þp ð22Þ V12
VSM12
where the subscripts of ‘p’ and ‘c’ refer to plain finned tube (P12) SP12
and enhanced finned tubes (S12, V12, VSM12, or SP12) respec-
tively. Under the condition of same temperature difference 0.8
between the fluid and the wall, the ratio of heat transfer rate
between the enhanced finned tubes and the plain finned tube
may be formulated as follows:
4000 6000 8000 10000
Uc ðNuðReÞA=Dc Þc ReDc
¼ ð23Þ
Up ðNuðReÞA=Dc Þp (b) identical pumping power criteria
where Nu(Re) represents the experimental correlation of Nusselt 1.2
number versus Reynolds number.
The comparison results are shown in Fig. 7, where the Reynolds
number of plain fin heat exchanger is taken as the x-coordinate. It
1
can be clearly seen that S12, V12 and VSM12 have better heat
transfer performances than P12, while SP12 has worse perfor-
S12
mance than P12. V12 offers the best heat transfer performance
Φc/Φp

V12
when ReDc is less than about 4500 under identical mass flow rate VSM12
criteria, or when ReDc is less than about 4500 under identical 0.8 SP12
pumping power criteria, or when ReDc is less than about 5000 un-
der identical pressure drop criteria. However, when ReDc is larger
than about 4500 under identical mass flow rate criteria, or when
ReDc is larger than about 6000 under identical pumping power cri-
teria, or when ReDc is larger than about 6700 under identical pres-
0.6
sure drop criteria, the performance of S12 is the best. At low
4000 6000 8000 10000
Reynolds numbers, the boundary layer of airflow is thick. The ReDc
structure of winglet-type VGs makes VGs be able to destroy the (c) identical pressure drop criteria
boundary layer more effectively and provide better airflow mixed
than the slit fin, therefore, the winglet-type VGs perform better Fig. 7. Uc/Up comparisons for 12-row tested samples.
824 L.H. Tang et al. / Experimental Thermal and Fluid Science 33 (2009) 818–827

at low Reynolds numbers. But at high Reynolds numbers, the dent of the number of tube rows when N P 6. The schematic dia-
boundary layer of airflow becomes thin, and the slit fin disturbs gram of the fin with vortex generators is shown in Fig. 8 with six
the airflow stronger and provides better airflow mixed than that rows of tubes in the flow direction. Table 1 lists all the geometric
of the winglet-type VGs. So with the increase of Re, the ratio Uc/ dimensions of the heat exchanger. The air flow direction is x-direc-
Up for vortex generator decreases, while that for slit fin is quite tion, fin span-wise direction is y-direction and fin thickness direc-
contrary.Although it has been reported that those fins with vortex tion is z-direction. The actual length of the computational domain
generators have good performance, but for the slit fin, the fin with is 7.5 times of the length of vortex-generator fin. That is, the do-
delta-wing vortex generators, and the mixed fin (front vortex-gen- main is extended 1.5 times of the length of vortex-generator fin
erator fin and rear slit fin) at hand, the mixed fin has better perfor- for the entrance section to ensure the inlet uniformity, and at the
mance than the fin with delta-wing vortex generators, and the slit exit, the domain is extended 5 times of the length of vortex-gener-
fin offers best heat transfer performance at high Reynolds num- ator fin in order to make sure that the exit flow boundary has no
bers. That is to say, the fin with vortex generators should be opti- flow recirculation.
mized in order to further enhance heat transfer. In the present Due to the geometrical symmetry of the flow domain, only one-
study, the attack angle, length and height of vortex generators half of the heat exchanger element has been computed. The flow is
are considered as the optimization variables. The correlations of assumed to be steady, turbulent and no viscous dissipation. The
heat transfer and friction including above three optimization vari- fluid is considered incompressible ideal gas with constant physical
ables are obtained by numerical method. properties. The dimensionless equations for continuity, momen-
tum and energy may be expressed in tensor rotation as
5. Numerical simulation for vortex-generator fin
r  ðq~
V/Þ ¼ r  ðC/ grad/Þ þ S/ : ð24Þ
5.1. Physical model In the above equation, the dependent variable, /; stands for the
velocity components, temperature, k and e, C/ and S/ represent the
In our previous study [30], we indicated that heat transfer and appropriate diffusion coefficients and the source terms,
friction performance for the fin with vortex generators is indepen- respectively.

(a) physical model


Vh

α Vl

(b) vortex generator

(c) computational domain


Fig. 8. Schematic diagram of vortex-generator fin.
L.H. Tang et al. / Experimental Thermal and Fluid Science 33 (2009) 818–827 825

5.2. Mathematic model Table 4


Simulated geometrical parameters of vortex generator.

The commercial code FLUENT 6.3 is used for the numerical solu- a (°) Vl (mm) Vh (mm)
tion of the Navier–Stokes and energy equations. FLUENT uses a 30 4 1.7
control-volume-based technique to convert the governing equa- 45 5 2.1
tions to algebraic equations that can be solved numerically. This 60 6 2.5
involves subdividing the region in which the flow is to be solved
into individual cells or control volumes so that the equations can
be integrated numerically on a cell-by-cell basis to produce dis- dicted and tested results show the reliability of the present physi-
crete algebraic (finite volume) equations [31,32]. All variables, cal model and numerical method.
including velocity components, pressure and temperature, are The most important factors affecting the heat transfer perfor-
averaged to a control volume. The coupling between pressure mance of vortex-generator fin are the attack angle of vortex gener-
and velocity is implemented by the SIMPLEC algorithm. The QUICK ator, the length of vortex generator and the height of vortex
scheme is used for the space discretization of the momentum, tur- generator. The simulated geometrical parameters of vortex gener-
bulence kinetic energy, turbulence dissipation rate and energy ator are tabulated in Table 4. A total of 27 models are simulated for
equations in the simulations. The residuals are less than 105 the development of correlations for Coburn j-factors and friction
and 1010 for the continuity and energy equations, respectively, factors f.
to ensure the convergence of computations. Base on the numerical simulation, the correlations of Coburn j-
At the channel inlet boundary, the flow velocity uin is assumed factors and friction factors f are given as follows:
to be uniform, and the temperature tin is taken to be constant. At  0:0014
the downstream end of the computational domain, stream-wise Vh
j ¼ 0:9856Re0:571 ðsin aÞ0:0086 ð26Þ
gradient (Neumann boundary conditions) for all the variables are Vl
set to zero, on the solid surfaces (tube, fin), no-slip conditions are
 0:04
used and uniform tube inside-wall temperature tw are specified. Vh
A symmetrical match is established on the boundary planes (X–Y f ¼ 43:28Re0:501 ðsin aÞ0:0143 : ð27Þ
Vl
plane and X–Z plane). The punched part from the fin is employed
by periodical boundary conditions. Additionally, at the solid–fluid As shown in Figs. 10 and 11, Eqs. (24) and (25) can describe all
interface simulated data within 10%. The mean error of j-factor correlation is
1.5%, and the mean error of f-factor correlation is 2.2%.
ts ¼ tf ; ks @t s =@n ¼ kf @t f =@n: ð25Þ
6. Genetic Algorithm optimization
5.3. Code validation and correlations
For heat exchanger optimization, the trade-off between heat
In order to validate the reliability of numerical simulation pro- transfer and pressure drop may be considered. In this paper, a Ge-
cedure, numerical simulations are carried out at the same operat- netic Algorithm based optimization technique has been conducted
ing conditions as the experimental vortex-generator finned heat for heat exchanger having vortex generators. The attack angle a,
exchangers with six rows [30]. Fig. 9 shows the comparisons be- length Vl and height Vh of the vortex generator are varied as the
tween the simulated results of three turbulence models (standard optimization parameters.
k  e; RNG k  e and realizable k  e) and the experimental results. Binary string is adopted for encoding the variables of a given
From Fig. 9, the simulated results of standard k  e model agree model. The search ranges of the three parameters are 30°–60°, 4–
best with the experimental results. Compared to the experimental 6 mm and 1.7–2.5 mm. Considering the ability of computer han-
results, the mean deviation of Nu number is 7.1%, and the mean dling bit operation and engineering application, the computational
deviation of Dp is 12.6%. Such good agreements between the pre- precision is set to three decimals (0.001). An individual of a model
needs 31-character binary: 10 for attack angle, 11 for lengths and

100 500 0.010


experimental results of vortex-generator fin [28]
simulation results of standardk-ε model 0.009
80 simulation results of realizablek-ε model
0.008
simulation results of RNGk-ε model
0.007
60
0%
+1

0.006
Δp / Pa
Nu

0%
jc

Δp 0.005
-1

100
0.004
40
0.003

0.002

Nu 0.001

0.000
4000 6000 8000 10000 0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.010
ReDc jp

Fig. 9. Comparison of Nu and Dp between simulation and experimental results. Fig. 10. Comparison of the simulated data and the j-factor correlation.
826 L.H. Tang et al. / Experimental Thermal and Fluid Science 33 (2009) 818–827

1.0 Table 5
Optimized results for maximum JF.
0.9
a (°) Vl (mm) Vh (mm) Nu Dp (Pa) JF
0.8 Reference 45 5.00 2.5 31.88 82.87 0.976
Optimized results by GA 60 6.00 1.7 32.87 82.08 1.014
0.7

0%
0.6

+1
fc

0.5 500

0%
-1
numerical simulation results of slit fin
0.4
numerical simulation results of optimized VG fin

Δp / Pa
0.3

0.2

0.1 Δp
0.0 100
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
fp
Nu

Nu
Fig. 11. Comparison of the simulated data and the f-factor correlation.

10 for height, respectively. The selection of upper bound and lower 4000 6000 8000 10000
bound for shape parameters seems somewhat arbitrary, however,
ReDc
their bound values are around the corresponding points from ref-
erences, and one/user can freely change the bound values based Fig. 13. Nu and Dp comparison between slit fin and optimized vortex-generator fin.
on practical constraints. The size of population and maximum evo-
lution generation are set to 20 and 1000, respectively. Probability by about 1.0%, in turn leading to the increased JF being larger than
of crossover and mutation are set to 0.5 and 0.005, respectively. unity. In this case, the optimized geometrical parameters will be
For the first optimization objective, the evaluation criteria, arrived within the bounding values. It is clearly shown that the lar-
JF ¼ ðj=f 1=3 Þc =ðj=f 1=3 Þr is chosen, where the subscripts of ‘c’ and ‘r’ ger attack angle and the higher length may be leading to the higher
refer to compared vortex-generator fin and reference vortex-gen- local heat transfer, and smaller height may be leading to smaller
erator fin respectively, the parameters of reference vortex-genera- flow resistance.
tor fin are given as: a = 45°, Vl = 5 mm and Vh = 2.5 mm. It is noted
that the performance j and f calculations are based on the above
7. Comparison of heat transfer performance between slit fin
correlations. The evolution process for maximum JF is shown in
and optimized vortex-generator fin
Fig. 12. At the beginning of the evolution process (less than 50 gen-
erations), the individuals with higher fitness are saved, and the
As mentioned above, the optimized vortex-generator fin offers
individuals with small fitness are removed. After certain genera-
better heat transfer performance than the reference vortex-gener-
tion (larger than 200 generations), the differences between every
ator fin. On the other hand, the slit fin also offers better heat trans-
individual are relatively large, in turn the variation of fitness for
fer performance than the reference vortex-generator fin, as shown
minimum weight is small, finally a level off value is found. Com-
in Fig. 7. Therefore, it is necessary to compare the heat transfer per-
pared with results from the reference case with VGs (Re = 4500),
formance between slit fin and optimized vortex-generator fin.
which is the original one before optimization, the optimized results
Fig. 13 shows the numerical simulation results (Nu and Dp)
for maximum JF are listed in Table 5. It is seen that the Nusselt
with Re numbers for slit fin and optimized vortex-generator fin.
number increases by about 3.1% while the pressure drop decreases
As can be seen, the Nu number of optimized vortex-generator fin
is larger than that of slit fin at the same Re number, while Dp of
1.015 optimized vortex-generator fin is smaller than that of slit fin.
Therefore, it can be concluded that after proper optimal design,
the vortex-generator fin can enhance the heat transfer perfor-
mance of heat exchanger and provide better performance than
the slit fin.

8. Conclusions
1.010
JF

In the present study, the air-side heat transfer of five kinds of


fin-and-tube heat exchangers have been experimentally investi-
gated with the Reynolds number ranging from 4000 to 10000,
and the optimization of heat exchanger with VGs is also addressed.
The main conclusions are summarized as follows:

(1) Before optimization, at high Reynolds numbers, the heat


1.005
exchanger with slit fin offers best heat transfer performance.
0 200 400 600 800 1000
Generation (2) The larger attack angle, higher length and smaller height of
vortex generators will lead to better overall performance of
Fig. 12. Evolution process for maximum JF. heat exchangers with VGs.
L.H. Tang et al. / Experimental Thermal and Fluid Science 33 (2009) 818–827 827

(3) The optimized vortex-generator fin can offer better heat exchangers, International Journal of Heat and Mass Transfer 45 (2002) 3795–
3801.
transfer performance than slit fin.
[14] J.S. Leu, Y.H. Wu, J.Y. Jang, Heat transfer and fluid flow analysis in plate-fin and
tube heat exchangers with a pair of block shape vortex generators,
International Journal of Heat and Mass Transfer 47 (2004) 4327–4338.
Acknowledgments [15] S.M. Pesteei, P.M.V. Subbarao, R.S. Agarwal, Experimental study of the effect of
winglet location on heat transfer enhancement and pressure drop in fin-tube
heat exchangers, Applied Thermal Engineering 47 (2005) 1684–1696.
This work is supported by National Nature Science Foundation [16] Y.H. Zhang, X. Wu, L.B. Wang, K.W. Song, Y.X. Dong, S. Liu, Comparison of heat
of China (No. 50821064) and Program for New Century Excellent transfer performance of tube bank fin with mounted vortex generators to tube
bank fin with punched vortex generators, Experimental Thermal and Fluid
Talents in University (No. NCET-04-0938). Science 33 (2008) 58–66.
[17] L.B. Wang, F. Ke, S.D. Gao, Y.G. Mei, Local and average characteristics of heat/
References mass transfer over flat tube bank fin with four vortex generators per tube,
ASME Journal of Heat Transfer 124 (3) (2002) 546–552.
[18] G.N. Xie, Q.W. Wang, B. Sunden, Application of a genetic algorithm for thermal
[1] A. Nuntaphan, T. Kiatsiriroat, C.C. Wang, Heat transfer and friction
design of fin-and-tube heat exchangers, Heat Transfer Engineering 29 (2008)
characteristics of crimped spiral finned heat exchangers with
597–607.
dehumidification, Applied Thermal Engineering 25 (2005) 327–340.
[19] G.N. Xie, B. Sunden, Q.W. Wang, Optimization of compact heat exchangers by a
[2] A. Nuntaphan, T. Kiatsiriroat, C.C. Wang, Air side performance at low Reynolds
genetic algorithm, Applied Thermal Engineering 28 (2008) 895–906.
number of cross-flow heat exchanger using crimped spiral fins, International
[20] Q.W. Wang, H.X. Liang, G.N. Xie, L.Q. Luo, Z.P. Feng, Genetic algorithm
Journal of Heat and Mass Transfer 32 (2005) 151–165.
optimization for primary surfaces recuperator of microturbine, ASME Journal
[3] F.C. McQuiston, Correlation of heat. Mass and momentum transport
of Engineering for Gas Turbines and Power 129 (2007) 436–442.
coefficients for plate fin-and-tube heat transfer surfaces with staggered
[21] M. Fiebig, Vortex generators for compact heat exchangers, Enhanced Heat
tubes, ASHRAE Transactions 84 (1978) 294–308.
Transfer 2 (1–2) (1995) 43–61.
[4] D.L. Gray, R.L. Webb, Heat transfer and friction correlations for plate fin-and-
[22] J.P. Holman, Heat Transfer, eighth ed., McGraw-Hill, 1997.
tube heat exchangers having plain fins, in: 8th International Heat Transfer
[23] T.E. Schmidt, Heat transfer calculations for extended surfaces, Refrigerating
Conference, San Francisco, California, 1986, pp. 2745–2750.
Engineering 57 (1949) 351–357.
[5] C.C. Wang, Y.J. Chang, Y.C. Hsieh, Y.T. Lin, Sensible heat and friction
[24] A.A. ZukausKas, Heat transfer from tubes in cross flow, Advances in Heat
characteristics of plate fin-and-tube heat exchangers having plain fins,
Transfer 8 (1972) 93–160.
International Journal of Refrigeration 19 (1996) 223–230.
[25] M. Jakob, Heat transfer and flow resistance in cross flow of gases over tube
[6] C.C. Wang, Y.T. Lin, C.J. Lee, An air side correlation for plain fin-and-tube heat
banks, Transactions of ASME 60 (1938) 384–386.
exchangers in wet conditions, International Journal of Heat and Mass Transfer
[26] H.J. Kang, W. Li, H.Z. Li, R.C. Xin, W.Q. Tao, Experimental study on heat transfer
43 (2000) 1869–1872.
and pressure drop characteristics of four types of plate fin-and-tube heat
[7] C.C. Wang, K.Y. Chi, C.J. Chang, Heat transfer and friction characteristics of plate
exchanger surfaces, Journal of Thermal Science 3 (1994) 34–42.
fin-and-tube heat exchangers, Part II: correlation, International Journal of Heat
[27] N.H. Kim, B. Youn, R.L. Webb, Air-side heat transfer and friction correlations for
and Mass Transfer 43 (2000) 2693–2700.
plain fin-and-tube heat exchangers with staggered tube arrangements, Journal
[8] W. Nakayama, L.P. Xu, Enhanced fins for air-cooled heat exchangers-heat
of Heat Transfer 121 (1999) 662–667.
transfer and friction correlations, in: 1st ASME/JSME Thermal Engineering Joint
[28] B. Yu, W.Q. Tao, Pressure drop and heat transfer characteristics of turbulent
Conference, Honolulu, Hawaii, 1983, pp. 495–502.
flow in annular tubes with internal wave-like longitudinal fins, Heat and Mass
[9] C.C. Wang, W.H. Tao, C.J. Chang, An investigation of the air side performance of
Transfer 40 (2004) 643–651.
the slit fin-and-tube heat exchangers, International Journal of Refrigeration 22
[29] L.B. Wang, W.Q. Tao, Q.W. Wang, Experimental study of developing turbulent
(1999) 595–603.
flow and heat transfer in ribbed convergent/divergent square ducts,
[10] C.C. Wang, W.S. Lee, W.J. Sheu, A comparative study of compact enhanced fin-
International Journal of Heat and Fluid Flow 22 (2001) 603–613.
and-tube heat exchangers, International Journal of Heat and Mass Transfer 44
[30] L.H. Tang, M. Zeng, G.N. Xie, Q.W. Wang, Fin pattern effects on air side heat
(2000) 3565–3573.
transfer and friction characteristics of fin-and-tube heat exchangers with large
[11] Y.J. Du, C.C. Wang, An experimental study of the air side performance of the
number of large-diameter tube rows, Heat Transfer Engineering 30 (2009)
superslit fin-and-tube heat exchangers, International Journal of Heat and Mass
171–180.
Transfer 43 (2000) 4475–4482.
[31] S.V. Patanker, Numerical Heat Transfer and Fluid Flow, Hemisphere Publishers,
[12] K.M. Kwak, K. Torli, K. Nishino, Heat transfer and flow characteristics of fin-
1980.
tube bundles with and without winglet-type vortex generators, Experiments
[32] H.K. Versteeg, W. Malalasekera, An Introduction to Computational Fluid
in Fluids 33 (2002) 696–702.
Dynamics – The Finite Volume Method, Prentice-Hall, 1995.
[13] K. Torli, K.M. Kwak, K. Nishino, Heat transfer enhancement accompanying
pressure-loss reduction with winglet-type vortex generators for fin-tube heat

Anda mungkin juga menyukai