International Communications in Heat and Mass Transfer 143 (2023) 106709

Contents lists available at ScienceDirect

International Communications in Heat and Mass Transfer

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

Optimization of H-type finned tube heat exchangers with combinations of

longitudinal vortex generator, dimples/protrusions and grooves by
Taguchi method
Yongshi Feng a, Ruyan Xu b, Yi Cao b, Xin Wu a, *, Cai Liang a, Liyuan Zhang a
a
Key Laboratory of Energy Thermal Conversion and Control of Ministry of Education, School of Energy and Environment, Southeast University, Nanjing 210096, China
b
China Tobacco Jiangsu Industrial Co., Ltd., Nanjing 210019, China

A R T I C L E I N F O A B S T R A C T

Keywords: In this paper, some novel structures aimed to optimize the airside performance of H-type finned tube heat ex­
H-type fin changers, such as longitudinal vortex generators, dimples/protrusions and grooves, are examined numerically.
Heat transfer enhancement The effects of geometric parameters on the heat transfer, flow resistance and comprehensive thermal–hydraulic
Longitudinal vortex generator
performances are investigated by means of Taguchi method. The optimum configuration of compound designs is
Dimple
Groove
determined on the basis of the optimizations of these structures. According to the results, for longitudinal vortex
generators, dimples/protrusions and grooves, the heat transfer characteristics are significantly affected by the
winglet type, vertical distance and groove diameter while the attack angle, dimple depth and groove type have
prominent impacts on the flow resistance characteristics. The winglet type, vertical distance and groove number
make the most contributions to the comprehensive thermal–hydraulic performance, respectively. In compound
designs, the longitudinal vortex generator plays the most important role in the performance of heat exchangers.
The comprehensive thermal–hydraulic characteristic of the optimal compound design exerts the improvements
of 0.9–23.8%, 24.5–57.1% and 8.3–37.5% than those of LVGs, dimples/protrusions and grooves, respectively,
when Reynolds number ranges from 4650 to 28,300. The enhancement of 4.5–16.6% can be achieved compared
with the case built based on references.

thermal and flow resistance performance of heat exchangers with

different types of fins, including the plate fins [9], annular fins [10],
1. Introduction
herringbone wavy fins [11], crimped spiral fins [12], serrated spiral fins
[13] and slotted fins [14]. Besides, the longitudinal vortex generator has
Environmental pollution and global warming have posed various
been a common structure adopted by many researchers to investigate
challenges for the world in the area of ecological sustainability as a
the heat transfer augmentation. Different types of winglets are
result of an abuse of fossil energy [1,2]. China, emitting the most carbon
compared experimentally by Fiebig et al. [15], and it was found that the
dioxide in the world, is committed to being carbon neutral by 2060
delta winglet had more effective heat transfer performance. Hwang et al.
[3,4]. The boom of renewable energy sources in China places great de­
[16] investigated numerically the effect of the vortex generator with
mands on coal-fired power units to adjust the load more quickly and
delta winglet on the performance of finned tubes. They found that the fin
flexibly [5–7]. However, the low-load operation makes the units off
with delta winglet had less pressure loss than the plain fin. The effect of
design conditions, and causes definitely different parameters of the flue
the winglet location on the thermal and flow performance was experi­
gas [8], which require more efficient recovery of industrial waste heat to
mentally studied by Pesteei [17]. The results indicated that the winglets
guarantee the boiler efficiency. Therefore, exploring more effective
placed in the downstream were more effective. Wu et al. [18] carried out
methods to enhance the thermal performance is of great importance,
both the experiments and numerical simulations to evaluate the impact
especially in the load regulation of coal-fired power units.
of the attack angle of winglet pairs on the heat transfer characteristic.
As a simple and effective way to promote the effect of heat transfer in
They observed that the average Nusselt number augmented with the
heat exchangers, extending surface has become widespread. Various
increasing attack angle. Naik et al. [19] investigated numerically the
numerical and experimental studies were conducted to investigate the

* Corresponding author.
E-mail address: [email protected] (X. Wu).

https://doi.org/10.1016/j.icheatmasstransfer.2023.106709

Available online 8 March 2023

0735-1933/© 2023 Elsevier Ltd. All rights reserved.
Y. Feng et al. International Communications in Heat and Mass Transfer 143 (2023) 106709

Nomenclature S modulus of the mass rate-of-stress tensor, s− 1

Sij mean stress rate, s− 1
A fin width, m SNR signal-to-noise ratio
cp specific heat capacity at constant pressure (J⋅kg− 1⋅K− 1) SS square sum
C1εC2ε Cμ turbulence model constants T temperature, K
dd diameter of the dimple, m u velocity, m⋅s− 1
dg width of the groove, m w1 vertical distance (distance from the X-axis), m
do tube outer diameter, m w2 horizontal distance (distance from the Y-axis), m
di tube inner diameter, m w3 dimple/grooves pitch, m
df degree of freedom x, y, z Cartesian coordinates
Eu Euler number
f friction factor Greek symbols
F cross-sectional area, m2 α attack angle of winglets, ◦
g1 winglet length, m β angle of obliquity of winglets, ◦
g2 distance from the fin surface, m δ fin thickness, m
G fin slit width, m ηf fin efficiency
h heat transfer coefficient, W⋅m− 2⋅K− 1 θ log mean temperature difference, K
H fin height, m λ fin thermal conductivity, W⋅m− 1⋅K− 1
I turbulent intensity, % μ dynamic viscosity, kg⋅m− 1⋅s− 1
k turbulent kinetic energy, m2⋅s− 2 ν kinematic viscosity, m2⋅s− 1
j Colburn factor ρ density, kg⋅m− 3
JF JF factor ε turbulent energy dissipation rate, m2⋅s− 3

n number of tube rows Subscripts

Nu average Nusselt number eff effective
p pressure, Pa i the ith factor
Δp pressure drop, Pa in inlet
P fin pitch, m fin fin surface
Pr Prandtl number max maximum
qm mass flow rate of fluid, kg/s min minimum
Q total heat transfer capacity, W sum total
R range of average SNRs w wall
Re Reynolds number based on hydraulic diameter wo outer wall of the tube
S1 transverse tube pitch, m out outlet
S2 longitudinal tube pitch, m

thermal performance of curved winglet vortex generators. The results comprehensive performance.
revealed that the curved winglet vortex generators with arc angle of 75◦ However, finned tube heat exchangers often work in the medium
had the heat transfer enhancement of 22% than plain winglet type. that hardly belongs to clean gases. The flue gas produced by the coal-
Furthermore, some novel structures have been designed on plain fins fired power plants contains particles and acid gases, which exert nega­
to improve the thermodynamic characteristics. Experiments were con­ tive impacts on finned tubes, including wear, scaling and corrosion on
ducted by Wang et al. [20] to study the effects of the plain or semi- the surface [26,27]. In recent years, H-type finned tube heat exchangers
dimple vortex generator on the performance of finned tube heat ex­ have showed the potential to augment the heat transfer more efficiently
changers. They observed that the semi-dimple vortex generator on account of their simple structure conducive to abilities of self-
enhanced the heat transfer by 17% with the flow resistance penalty of cleaning and wear resistance [28,29]. Li et al. [30], Wu et al. [31] and
30%. The effect of different types of dimpled heat sink on the heat Chen et al. [32] conducted extensive experiments to evaluate the per­
transfer and flow resistance characteristics was investigated experi­ formance of plain H-type finned tube heat exchangers at different Rey­
mentally by Gupta et al. [21]. The results indicated that the heat transfer nolds numbers and geometric parameters, and thus summarized the
could be enhanced by decreased pitch ratio or increased depth ratio. correlations of Nusselt and Euler number. By means of numerical sim­
Sangtarash et al. [22] investigated the effect of multilouvered fins with ulations, Jin et al. [27,33] and Wang et al. [34] found that the number of
dimples and perforated dimples by experiments and numerical simula­ tube rows, fin pitch and spanwise tube pitch should be considered firstly
tions. The simulations revealed that the dimple-perforation in staggered when optimizing the heat transfer characteristic. On this basis, some
had the best heat transfer characteristic. A trailing edge cooling struc­ novel structures have been designed on plain H-type fins in order to
ture was optimized by Luo et al. [23] by adding the pin fins and dimples/ further improve the performance of the heat exchangers. The longitu­
protrusions. The results suggested that higher Nusselt number and dinal vortex generators (LVGs) studied by Tang et al. [35] improved the
friction factor were obtained by the pin fin channel with protrusion performance evaluation index by 48–55%. Zhao et al. [36,37] investi­
instead of the dimpled channel. Liu et al. [24] studied numerically the gated the characteristics of heat transfer and erosion of H-type finned
convective heat transfer in the square channels with different-shaped oval tubes with LVGs and dimples through numerical simulations and
grooves. They observed that the rounded transition groove out­ experiments. The simulation results indicated that the H-type fin with
performed the square fins in both enhancing heat transfer and reducing compound LVG-dimple could both control fouling and improve the
flow resistance. Several types of microchannel heat sinks were designed thermal performance. The H-type fin with LVGs was the most conducive
by Zhu et al. [25] to evaluate the effect of groove geometry on the to the anti-wear performance. The numerical simulations on double H-
comprehensive performance. They observed that all types of grooves type finned tubes with LVGs and grooves were conducted by Li et al.
except rectangles achieved the significant improvement of the [38] to evaluate the thermal-hydraulic-fouling performance. The

2
Y. Feng et al. International Communications in Heat and Mass Transfer 143 (2023) 106709

improvement in heat transfer of 12.92–16.63% with pressure loss of five and five parameters are chosen as control factors, respectively,
21.75–30.72% and reduction in fouling rate of 2.4–23.7% were obtained whose ranges are subdivided into four levels. The optimization objec­
by the fin structure with LVGs and grooves. tives of the best heat transfer and comprehensive performance together
The previous studies have devoted a good deal of effort into the with the smallest flow resistance are achieved via compounding each
performance optimization of finned tube heat exchangers, which pro­ optimal level of control factors. The effects of geometric parameters are
vide the significant guidelines. However, only a single factor was often quantified by their contributions to the overall performance. The opti­
studied by most researchers in each test, which caused a significant mum configuration of compound designs is determined on the basis of
amount of experimental material waste. More importantly, they hardly the optimizations of three innovative structures. In order to verify the
took the interactions between various control factors into account. additivity of the interaction among control factors, a confirmatory test is
Hence, in the process of exploring the reliable optimal parametric carried out. The comprehensive thermal-hydraulic performance of the
combinations, it is necessary to take some reasonable methods to opti­ optimum compound design is compared with that of other cases to
mize the process and reduce the number of tests. Taguchi method has justify the reliability of the optimization results. This paper can provide
been built on the basis of the statistical principle by Genichi Taguchi, reference for designing and optimizing heat exchangers with novel H-
which ensures the rationality and simplicity of the process of optimi­ type fins. Meanwhile, it can be a source of inspiration for the applica­
zation, because each level of all factors has the same probability to tions of other industries, such as the heat transfer enhancement for the
participate in combinations [39,40]. There have been some in­ trailing side of gas turbine and effective heat sinks for integrated elec­
vestigations adopting Taguchi method to optimize the design of heat tronic circuits.
exchangers. In the investigation of Etghani et al. [41] about the heat
transfer coefficient and exergy loss, Taguchi method was adopted to 2. Physical model and numerical method
obtain the optimum levels of the design factors considering four design
parameters. Zhang et al. [42] discussed the heat transfer and flow 2.1. Model description
resistance characteristics of finned tubes with different structural pa­
rameters, and used Taguchi method to improve the heat transfer effi­ Fig. 1 displays the diagrammatic sketch of plain H-type finned tube
ciency. Based on the numerical simulations, Wang et al. [34] adopted banks with multiple rows. Considering the periodicity and symmetry,
Taguchi method to optimize the structure of plain H-type finned tube the computation domain is delineated by the dashed line, whose top
heat exchangers to promote the heat transfer characteristic. The in­ view and side view are given in Fig. 1 (b) and Fig. 1 (c). In order to
fluences of various parameters on the performance of heat exchangers ensure the uniform inlet velocity and prevent the outlet backflow, the
with vortex generators were analyzed by Taguchi method in the in­ zones are extended at the inlet and outlet, which are 3 times and 9 times
vestigations of Tang et al. [43] and Zeng et al. [44], and the optimal of the outer diameter of the tube, respectively. The geometric parame­
structural parameters were obtained. However, there are few studies on ters of plain H-type finned tube banks applied in this paper are listed in
the comprehensive optimization of modified H-type finned tube heat Table 1, which are references to the authors' previous study [45]. The
exchangers with several innovative structures via Taguchi method. additional modifications will be based on this structure.
The purpose of this study is to propose the optimal designs by Three types of modified H-type finned tubes and their combination
imposing some additional modifications (longitudinal vortex genera­ investigated in this paper are shown in Fig. 2. Fig. 2 (a) displays LVGs on
tors, dimples/protrusions and grooves) on the plain H-type finned tube the H-type finned tube (LVGs-HT). LVGs are positioned on the fin by the
heat exchangers. In present study, 3D numerical simulations are vertical distance w1 and horizontal distance w2. The winglet length g1
executed for optimizing the geometric parameters of innovative struc­ and winglet width g2 represent the size of LVGs. The attack angle α is the
tures on plain H-type finned tubes by means of Taguchi method. For angle between the long side of LVGs and the direction of flow. The angle
longitudinal vortex generators, dimples/protrusions and grooves, seven, of obliquity β is the angle between the winglets and the plane of H-type

u ,T

H S

A
S

G
3d d d 9d

Fig. 1. Diagrammatic sketch of the heat exchanger with plain H-type fins.

3
Y. Feng et al. International Communications in Heat and Mass Transfer 143 (2023) 106709

Table 1
Geometrical parameters.
H /mm A /mm G /mm δ /mm do /mm di /mm P /mm S1 /mm S2 /mm n

84 84 10 3 38 28 14 96 96 10

Cμ ρη3 (1 − η/η0 )ε2

w R= ( ) (6)
w
1 + β η3 k
w w w
g
g dd g where η = Sk/ε, η0 = 4.38 and β = 0.012.
g
S2 = 2Sij Sij (7)
( )
∂ui ∂u
where Sij = 12 + ∂xji . C1ε and C2ε are 1.42 and 1.68.
w
w ∂xj

dg g 2.3. Parameter definition

Reynolds number : Re = umax do /ν (8)

Fig. 2. Schematic diagram of different modified structures on H-type fin­
ned tubes.
Nusselt number : Nu = hdo /λ (9)

fins. Fig. 2 (b) presents the H-type finned tube with dimples (D-HT). The 2Δp
dimples/protrusions are positioned on the fin by the vertical distance w1 Euler number : Eu = (10)
ρu2max n
and spacing between dimples/protrusions w3. The horizontal distance
w2 is determined by the spacing between dimples/protrusions w3 (w2 = where, umax is the average velocity at the minimum section at the shell-
H/2- w3). The depth g2 represents the distance between the top of side, m/s. Δp is the pressure drop between the inlet and outlet, Pa. n is
dimples/protrusions and the plate of H-type fins. In Fig. 2 (c), grooves the tube row number. ν, λ, ρ are the physical properties of the shell-side
are set on the H-type fin (G-HT). The value of horizontal distance w2 is fluid, where the average temperature is specified as the qualitative
fixed at 36 mm. The spacing between grooves w3 is the distance between temperature. For Re, Nu and Eu, the outer diameter of the tube is
the center lines of two grooves. The depth g2 represents the distance specified as their characteristic length.
between the bottom of grooves and the plate of H-type fins. Fig. 2 (d)
presents the H-type finned tube with compound designs (L-D-G-HT). umax =
Fin (uin + uout )
(11)
2Fmin
2.2. Governing equations and boundary conditions
Δp = pin − pout (12)
The fluid is assumed to have constant properties and be viscous, where, Fin is the cross-sectional area at the inlet, m2. Fmin is the minimum
incompressible and steady turbulent flow. The renormalization group sectional area at the shell-side, m2. uin and uout are the flow velocities at
(RNG) k-ε turbulence model is adopted. The equations, boundary con­ the inlet and outlet, m/s. pin and pout are the pressures at the inlet and
ditions and conjugated computations adopted in this paper are the same outlet, Pa.
to authors' previous study [45]. Thus, for the sake of brevity and clarity, The Colburn factor j and friction factor f are applied to evaluate the
this paper will not repeat the demonstrations but just list the equations. heat transfer and flow resistance characteristics of heat exchangers. The
∂(ρui ) nondimensional number JF concluded from Yun and Lee [46] can
Continuity equation : =0 (1) describe and act as a criterion to evaluate the comprehensive thermal­
∂xi
–hydraulic performance.
[ ( )]
∂ ( ) ∂ ∂uj ∂ui ∂p ( )
Momentum equation : ρui uj = μ + − (2) Nu h
∂xi ∂xi ∂xi ∂xj ∂xj j= = Pr2/3 (13)
RePr 1/3 ρumax cp
( )
∂ ∂ λ ∂T
Energy equation : (ρui T) = (3) Δp
∂xi ∂xi cp ∂xi f = ρu2 (14)
max Fsum
2 Fmin
where xi, xj (i, j = 1,2,3) represent the x, y, z in Cartesian coordinates. ui,
uj (i, j = 1,2,3) are the components of the fluid velocity in x, y and z JF =
j/j0
(15)
direction.
1/3
(f /f0 )
RNG k-ε turbulence model:
[ ] where, Fsum is the total heat exchange area at the shell-side, m2. j0 and f0
Dk ∂ ∂k are obtained from the results of plain H-type finned tube banks.
ρ = αp μeff + μt S2 − ρε (4)
Dt ∂xi ∂xi The total heat transfer Q is expressed as
[ ]
Dε ∂ ∂ε ε ε2 Q = cp qm (Tout − Tin ) (16)
ρ = αp μeff + C 1 ε μt S 2 − C 2 ε ρ − R (5)
Dt ∂xi ∂xi k k ( )
Q = h Fwo + Ffin ηf θ (17)
2
k
where μeff = μ + μt, and μt = ρCμ ε with Cμ = 0.0845.
where, cp is constant pressure heat capacity of fluid, J/(kg⋅K). qm is the
mass flow rate of fluid, kg/s. Fwo and Ffin are the heat exchange area of
the tube and fin, m2. ηf is the fin efficiency derived from Feng et al. [47].

4
Y. Feng et al. International Communications in Heat and Mass Transfer 143 (2023) 106709

The log mean temperature difference θ is defined as [33].

Nu - Correlation from
Tout − Tin
θ= (18) Nu - Present simulation
lnTTwowo−− TTout
in
Eu - Correlation from
Therefore, the average heat transfer coefficient can be derived. Eu - Present simulation

cp qm (Tout − Tin )
h=( ) (19)
Fwo + Ffin ηf θ

Nu

Eu
2.4. Grid independence and validation of the computational model

ANSYS ICEM CFD 19.0 is adopted to establish the grids. Because the
additional modifications are designed based on the plain H-type fins in
the following simulations, in the case of the basic structure presented in
Table 1, the validation of the grid independence is performed. This
paper selects Nu to evaluate the grid independence, and Fig. 3 displays Re
the variation of Nu with grid numbers. It can be concluded that the grid Fig. 4. Comparisons of Nu and Eu between experimental correlations and
number of 1,552,500 is appropriate, and its grid settings will be applied present simulations.
in the following simulations.
In order for the validation of the computational model, results of the 3. Optimization with Taguchi method
case of the basic structure presented in Table 1 are compared with those
from the correlations obtained by experiments of Li et al. [30] (Eqs. (20) 3.1. General procedure
and Eq. (21)).
( )− 0.145 ( )− 0.031 ( )0.2 ( )− 0.85 ( )0.3921
H G P S1 S2 The number of experiments is depended on the number of the ele­
Nu = 0.399Re0.635 Pr0.33 ments which influence the system. Taguchi method is a simpler and
do do do do do
more effective method to obtain the optimal design by means of the
(20)
orthogonal arrays and robustness. In the design of heat exchangers,
( )0.927 ( )0.913 ( )− 0.572 ( )− 2.326 ( )− 0.351 Taguchi method can be utilized to obtain the optimal levels of control
H G P S1 S2
Eu = 2.743Re− 0.189
factors which have effects on the thermodynamic performance. This
do do do do do
paper pays more attention to how Taguchi method is applied to design
(21) the optimal heat exchangers, but explains few detailed mathematical
Fig. 4 compares Nu and Eu obtained from numerical simulations and backgrounds.
experimental correlations. It can be observed that the results of present The optimization process by Taguchi method consists of several
simulations are generally smaller than those from the experiments. The phases, namely system design, parameter design and tolerance. Taguchi
difference between Nu increases while it descends between Eu with the method employs the main-effect plots and analysis of variance to illus­
increasing Re. The maximum deviation in Nu is around − 7%, and the trate the impacts of control factors and determine the optimal level of
average difference in Eu is approximately − 10%. There are mainly two each factor. It is necessary to perform confirmatory tests to demonstrate
factors attributing to these differences, namely the radiation heat and the additivity of the interactions among control factors. After obtaining
manufacture precision. The radiation heat transfer is ignored in simu­ the optimal levels of each factor, the case with combined optimal levels
lations, but it is merged in the calculation of the convection heat transfer is required to be validated that it outperforms other cases.
in experiments, thus leading to the smaller Nu of present simulations.
The rough and uneven surfaces of models in experiments can contribute
3.2. Orthogonal array
to stronger turbulence and thus better heat transfer and more pressure
penalty. Overall, the results have high goodness of agreement and
Orthogonal array is the foundation of Taguchi method, which evenly
acceptable differences in engineering.
compounds the levels of all control factors to reduce the number of tests
effectively. It is expressed as Ld (ak), where d, a and k represent the
numbers of tests, levels for each factor and control factors, respectively.
For example, the orthogonal array of L32 (47) applied in the investigation
of LVGs has seven factors with four levels. Test times have been reduced
by the orthogonal array from 47 (=16,384) to 32, where every level of
each factor has the same number of repeats. Thus, the main-effect plots
are proper to compare the effects of all control factors and obtain the
optimum structure by compounding the best values of all factors.
Nu

3.3. Signal-to-noise analysis

The signal-to-noise ratios (SNR) are used to logarithmically trans­

form the raw data from the simulations, and the promoted statistical
properties are conducive to the optimization [40]. In Taguchi method,
the performance statistics include “the largest is the best” and “the
smallest is the best”, namely maximizing the expected objective and
minimizing the unexpected objective. It is generally known that opti­
Grid number mizing heat exchangers requires the maximum j and JF but the mini­
Fig. 3. Changes of Nu at different grid numbers (Re = 18,900). mum f. Therefore, the following equation “the largest is the best” is

5
Y. Feng et al. International Communications in Heat and Mass Transfer 143 (2023) 106709

applied to j and JF Table 3

( ) Raw results and SNRs for the cases of LVGs.
1∑ n
1
SNRL = − 10log (dB) (22) Case no. j SNR-j f SNR-f JF SNR-JF
n i=1 Yi
1–1 0.004062 − 23.91 0.01498 18.24 0.9939 − 0.02650
1–2 0.003938 − 24.05 0.01459 18.36 0.9722 − 0.1223
while “the smallest is the best” equation is utilized to transform f 1–3 0.004014 − 23.96 0.01475 18.31 0.9872 − 0.05614
( ) 1–4 0.003926 − 24.06 0.01466 18.34 0.9675 − 0.1437
1∑ n
1–5 0.004015 23.96 0.01510 18.21 0.9796 − 0.08960
(23)
−
SNRS = − 10log Yi (dB)
n i=1 1–6 0.003940 − 24.05 0.01424 18.47 0.9804 − 0.08578
1–7 0.004043 − 23.93 0.01582 18.01 0.9715 − 0.1254
1–8 0.003941 − 24.04 0.01472 18.32 0.9701 − 0.1319
where SNR is the criterion for the optimal objective and the largest one is 1–9 0.003966 − 24.02 0.01481 18.30 0.9742 − 0.1133
preferred. Y is the raw data and n counts the number of repetitions per 1–10 0.004022 − 23.96 0.01482 18.29 0.9877 − 0.05381
test whose value is one. 1–11 0.004675 − 23.30 0.01860 17.30 1.064 0.2708
SNR analysis is required for each factor, and the intuitive analysis 1–12 0.004004 − 23.97 0.01474 18.31 0.9850 − 0.06551
1–13 0.004159 23.81 0.01559 18.07 1.004 0.01820
needs to obtain the average SNR firstly, which arithmetically averages
−
1–14 0.004123 − 23.85 0.01630 17.88 0.9808 − 0.08416
SNRs of all levels and factors. R is the deviation between the maximum 1–15 0.004259 − 23.71 0.01750 17.57 0.9894 − 0.04614
and minimum average SNRs of each factor. 1–16 0.004036 − 23.94 0.01576 18.02 0.9709 − 0.1281
1–17 0.005196 − 22.84 0.02175 16.63 1.123 0.5031
R = SNRmax,i − SNRmin,i (24) 1–18 0.004002 − 23.98 0.01491 18.27 0.9807 − 0.08470
1–19 0.004223 − 23.74 0.01608 17.94 1.009 0.03936
where i indexes the factor. 1–20 0.004740 − 23.24 0.01935 17.13 1.065 0.2731
1–21 0.005090 22.93 0.02124 16.73 1.109 0.4481
The contribution of each factor is depended on R, defined by −
1–22 0.004129 − 23.84 0.01566 18.05 0.9955 − 0.01955
SNRmax,i − SNRmin,i 1–23 0.004229 − 23.74 0.01912 17.19 0.9540 − 0.2043
Contribution ratio = ∑
n ( ) (25) 1–24 0.004040 − 23.94 0.01682 17.74 0.9511 − 0.2179
SNRmax,i − SNRmin,i 1–25 0.004303 − 23.66 0.01675 17.76 1.015 0.06256
i=1 1–26 0.003862 − 24.13 0.01486 18.28 0.9473 − 0.2350
1–27 0.004134 − 23.84 0.01646 17.84 0.9802 − 0.08698
4. Results and discussion 1–28 0.004747 − 23.24 0.01997 17.00 1.055 0.2338
1–29 0.004186 − 23.78 0.01657 17.81 0.9903 − 0.04230
1–30 0.004044 − 23.93 0.01817 17.41 0.9279 − 0.3251
4.1. Longitudinal vortex generators on H-type fins 1–31 0.004041 − 23.93 0.01529 18.16 0.9821 − 0.07859
1–32 0.007548 − 21.22 0.04866 13.13 1.247 0.9587
4.1.1. Factors and orthogonal array for LVGs
The control factors of LVGs on H-type finned tubes included in this
mixing, the attack angle is used to judge the dominance of longitudinal
study are seven parameters in geometric, i.e. attack angle α, vertical
vortex over transverse vortex generated by LVGs [19]. The attack angle
distance w1, horizontal distance w2, angle of obliquity β, winglet length
increases with larger and larger dominance of longitudinal vortices,
g1, winglet width g2 and winglet types. Table 2 displays the levels of
which results in the increment in j. Increasing attack angle contributes to
control factors, which refer to Ref. [35] and are arranged evenly in the
the reduction in the wake region, rise in the flow velocity near wall and
wide ranges. Although the ranges of levels are required to be large
impact of fluid on the tube wall, which are beneficial to enhance the heat
enough to cover as many working conditions as possible, the selections
transfer significantly. Besides, for the LVGs with increasing winglet
of levels also take the actual processing conditions into consideration.
length, the flow velocity increases between the LVGs and the tube. The
Other parameters are set as the constant values listed in Table 1. In this
delay of flow separation results in the narrower and shorter wake region
paper, the investigation of LVGs adopts an orthogonal array of L32 (47) at
and more uniformity of the overall velocity profile. Furthermore,
Re = 14,080.
different winglet types can have different impacts on the flow and
vortices. The rectangle winglets, by contrast, enhance the fluid mixing
4.1.2. SNR analysis for LVGs
and transport and remove the zone with poor heat transfer from the
The raw results (j, f and JF) and their transformed SNRs are presented
wake region, which is generally more efficient than other winglet types
in Table 3. After the calculations, the effects and contribution ratios of
in enhancing heat transfer. The contribution ratio of these three pa­
all factors and levels to the results for LVGs are displayed in Fig. 5 more
rameters totals approximately 60%. Hence, the promotion in the heat
visually.
transfer characteristic is supposed to give priority to the attack angle α,
As Fig. 5 (a) presents, the contribution ratios of control factors to j are
winglet length g1 and winglet type.
ordered as G > A > E > C > F > B > D. The attack angle α (A), winglet
As displayed in Fig. 5 (b), the control factors are ordered as A > G >
length g1 (E) and winglet type (G) exert dominant impacts on j with
E > B > F > C > D according to their contributions. SNR-f is most heavily
contribution ratios of 18.31%, 16.17% and 22.76% respectively. In fluid
affected by the attack angle α (A) with a contribution ratio of 20.60%.
The larger attack angle of winglets makes the fluid path more obstructed
Table 2
and thus leads to the added pressure drop penalty. In addition, the
Levels of the control factors of LVGs.
winglet length g1 (E) and winglet type (G) also play primary roles in
Name Factor Level 1 Level 2 Level 3 Level 4 SNR-f, accounting for 18.86% and 15.71%. The size of wake regions
A attack angle α (◦ ) 0 30 60 90 behind LVGs with longer winglet length or rectangle winglets is bigger
vertical distance w1 than others because of the larger area facing the main flow, leading to
B 11 13 15 17
(mm)
greater pressure drop. The total contribution ratio of these three factors
horizontal distance
C
w2 (mm)
16 18 20 22 comes to 55.16%, which implies their prominent impacts on the flow
D trailing angle β (◦ ) 45 60 75 90 resistance characteristic. It can also be found that the angle of obliquity β
E
winglet length g1
8 12 16 20
shows the least effect.
(mm) As for the criterion to evaluate the thermodynamic performance, the
winglet width g2
F
(mm)
1.68 2.52 3.36 4.20 control factors are in order of G > C > A > E > D > F > B according to the
G winglet types triangle hemisphere trapezoid rectangle contribution ratios to SNR-JF. Fig. 5 (c) indicates that SNR-JF is most

6
Y. Feng et al. International Communications in Heat and Mass Transfer 143 (2023) 106709

SNR-JF, the optimal condition (A3B2C1D4E4F4G4) can be achieved by

compounding each best level.
By comparisons of the main-effect plots, the impact of control factors
on SNR-JF is reasonable and explicable. In Fig. 5 (c), SNR-JF has fast
growth when the attack angle (A) augments from A1 to A3 because of
the quick increase in SNR-j. Although SNR-f also declines rapidly, the
variation of SNR-JF is more affected by j factor. When the attack angle
further increases to A4, the improvement in SNR-j is too slight to
counteract the sharply decreasing SNR-f, so SNR-JF descends. Similarly,
when horizontal distance w1 (B) increases from B3 to B4, the aug­
menting SNR-j and the sharper fall in SNR-f cause the minor increase of
SNR-JF. In general, the trend of SNR-JF is more similar to that of SNR-j
rather than SNR-f. There is no denying that f factor has a certain impact
on SNR-JF, but in contrast, the effect of j factor is more significant. By
the analogies, similar conclusions can be made for other factors.

4.2. Dimples/protrusions on H-type fins

4.2.1. Factors and orthogonal array for dimples/protrusions

The control factors of dimples/protrusions on H-type finned tubes
included in this study are five parameters in geometric, i.e. vertical
distance w1, spacing between dimples/protrusions w3, diameter dd,
depth g2 and types. Table 4 lists the levels of control factors, which are
arranged evenly in the wide ranges with reference to Ref. [21]. An
orthogonal array of L16 (45) is adopted for dimples/protrusions at Re =
14,080.

4.2.2. SNR analysis for dimples/protrusions

The raw results (j, f and JF) and their transformed SNRs are presented
in Table 5. After the calculations, the effects and contribution ratios of
all factors and levels to the results for dimples/protrusions are displayed
in Fig. 6 more visually.
Fig. 6 (a) reveals that the contribution ratios of control factors to j are
ordered as H > L > K > I > J. The vertical distance w1 (H) and dimple
type (L) play primary roles in j accounting for 52.24% and 21.72%,
while j is minimally affected by the dimple diameter (J). The increasing
vertical distance w1, namely the distance from the center of the tube,
makes the flow more turbulent. The flow near the edge of fins enhances
the heat transfer and therefore promotes j factor. Furthermore, different
types of dimples/protrusions can cause different impacts on the flow and
vortices. By contrast with other types, the dimples are generally more
efficient for the reason of its enhancement in the fluid mixing and
transport vertical to the direction of flowing. The total contribution ratio
of these two parameters is larger than 70%. Hence, when considering the
promotion in the heat transfer characteristic, the vertical distance w1
and types of dimples/protrusions are supposed to be addressed as
priorities.
As Fig. 6 (b) presents, the contribution ratios of control factors to
SNR-f are ranked as K > H > L > J > I, which sets it apart from that for j.
SNR-f is mainly influenced by the depth of dimples/protrusions g2 (K)
with a contribution ratio of 33.57%. The increasing depth of dimples/
protrusions results in bigger friction factor since deeper vortices are
formed in the fluid near dimpled wall, which enlarge the pressure loss.
Besides, vertical distance w1 (H) has the second contribution to SNR-f
with the ratio of 25.86%. The added dimples/protrusions on fin surfaces
strengthen the flow turbulence and increase the flow resistance. There is

Fig. 5. Main-effect plots and contribution ratios of control factors for LVGs (Re Table 4
= 14,080). Levels of the control factors for dimples/protrusions.
Name Factor Level 1 Level 2 Level 3 Level 4
affected by the winglet type while the horizontal distance w1 and
H vertical distance w1 (mm) 26 30 34 38
winglet width g2 are the least important factors. The contribution ratio
I spacing w3 (mm) 8 10 12 14
of the winglet type, vertical distance w2 and attack angle α totals J diameter dd (mm) 1.8 2.2 2.6 3.0
approximately 60%, which suggests their prominent impacts when K depth g2 (mm) 1.4 1.8 2.2 2.6
designing and optimizing the heat exchangers. From the perspective of L types dimple hexagon circle square

7
Y. Feng et al. International Communications in Heat and Mass Transfer 143 (2023) 106709

Table 5
Raw results and SNRs for the cases of dimples/protrusions.
Case no. j SNR-j f SNR-f JF SNR-JF

2–1 0.004033 − 23.94 0.01537 18.13 0.9785 − 0.09456

2–2 0.003846 − 24.15 0.01705 17.68 0.9012 − 0.4518
2–3 0.003805 − 24.20 0.01735 17.61 0.8865 − 0.5231
2–4 0.003616 − 24.42 0.01991 17.01 0.8047 − 0.9437
2–5 0.003809 − 24.19 0.01938 17.13 0.8553 − 0.6789
2–6 0.004007 − 23.97 0.01796 17.46 0.9228 − 0.3487
2–7 0.004050 − 23.93 0.01719 17.65 0.9464 − 0.2392
2–8 0.004154 − 23.81 0.01658 17.80 0.9826 − 0.07620
2–9 0.004275 − 23.69 0.02127 16.72 0.9306 − 0.3123
2–10 0.004286 − 23.68 0.01721 17.64 1.001 0.002349
2–11 0.004257 − 23.71 0.01819 17.40 0.9764 − 0.1038
2–12 0.004176 − 23.79 0.01686 17.73 0.9824 − 0.07723
2–13 0.004559 − 23.41 0.02027 16.93 1.008 0.03638
2–14 0.004395 − 23.57 0.01874 17.27 0.9982 − 0.007977
2–15 0.004677 − 23.30 0.02219 16.54 1.004 0.01741
2–16 0.004299 − 23.67 0.01849 17.33 0.9807 − 0.08485

no significant difference found in the effects on the flow resistance

characteristic among the remaining three factors with the average ratio
of about 13%.
Fig. 6 (c) indicates that SNR-JF is most affected by the vertical dis­
tance w1 (H) while the spacing between dimples/protrusions w3 (I) and
dimple diameter dd (J) are the least important factors, with the order of
H > L > K > J > I. The total contribution ratio of the vertical distance w1,
dimple type and depth of dimples/protrusions g2 is 83.49%, which are
supposed to be given priorities when designing and optimizing heat
exchangers. From the perspective of SNR-JF, the optimal condition
(H4I2J1K1L1) can be achieved by compounding each best level.
By comparing the main-effect plots, it can be observed that the
variations of both heat transfer and pressure loss contribute to the trend
of comprehensive performance. Although SNR-j and SNR-f present the
opposite tendencies when the vertical distance (H) augments from H1 to
H4, SNR-JF keeps increasing because of the more significant effect of j
factor. Because SNR-j descends slightly while SNR-f increases quickly
when the spacing (I) increases from I1 to I2, SNR-JF augments. But when
spacing (I) further increases from I3 to I4, the deterioration in heat
transfer impairs the benefit of the reduction in flow resistance. In
conclusion, JF is determined by j and f factor together while there is a
difference between the influence of j and f on JF for different factors.

4.3. Grooves on H-type fins

4.3.1. Factors and orthogonal array for grooves

The control factors of grooves on H-type finned tubes included in this
study are five parameters in geometric, i.e. spacing between grooves w3,
diameter dg, depth g2, number of grooves and groove types. Table 6
displays the levels of control factors, which refer to Ref. [24,25] and are
arranged evenly in the wide ranges. Due to the same numbers of levels
and factors with those of dimples/protrusions, the orthogonal array of
L16 (45) for grooves is similar to that for dimples/protrusions.

4.3.2. SNR analysis for grooves

The raw results (j, f and JF) and their transformed SNRs are presented
in Table 7. After the calculations, the effects and contribution ratios of
all factors and levels to the results for grooves are displayed in Fig. 7.
Fig. 6. Main-effect plots and contribution ratios of control factors for dimples/
It reveals from Fig. 7 (a) that the sequence of the contribution ratios
protrusions (Re = 14,080).
to j is N > P > O > Q > M. The groove diameter dg (N), depth g2 (O) and
number of grooves (P) play more important roles in j with contribution
ratios of 32.12%, 20.37% and 21.53%, respectively. The fluid crosses bigger vortexes formed in the grooves, thus enhances the heat transfer.
the grooves is divided into two parts. One part makes the formation of Besides, the grooves with increasing depths not only augment the length
vortexes in grooves, taking away the heat through flushing the fin wall. of flow path, but also interrupt periodically the thermal boundary layer
Then it mixes with the other part of the fluid, which stays out of the by the boundary layer vortex shedding, thereby enhancing the heat
vortexes and flows directly to the region of grooved-fins. The larger transfer. As noted previously, the formed periodic vortexes in grooves
groove diameter results in the stronger intensity of fluid mixing and the contribute to strengthening the fluid mixing and heat transfer. In

8
Y. Feng et al. International Communications in Heat and Mass Transfer 143 (2023) 106709

Table 6
Levels of the control factors for grooves.
Name Factor Level 1 Level 2 Level 3 Level 4

M spacing w3 (mm) 8 10 12 14
N diameter dg (mm) 3 4 5 6
O depth g2 (mm) 1.4 1.8 2.2 2.6
P number 2 3 4 5
Q types cylindrical trapezoid rectangle triangle

Table 7
Raw results and SNRs for the cases of grooves.
Case no. j SNR-j f SNR-f JF SNR-JF

3–1 0.004308 − 23.66 0.01769 17.52 0.9974 − 0.01110

3–2 0.005082 − 22.94 0.02316 16.35 1.075 0.3154
3–3 0.004337 − 23.63 0.03255 14.88 0.8194 − 0.8650
3–4 0.004339 − 23.63 0.03249 14.88 0.8202 − 0.8609
3–5 0.004196 − 23.77 0.02407 16.19 0.8765 − 0.5726
3–6 0.004162 − 23.81 0.02356 16.28 0.8758 − 0.5761
3–7 0.004406 − 23.56 0.02054 16.87 0.9706 − 0.1298
3–8 0.005831 − 22.34 0.02674 15.73 1.176 0.7041
3–9 0.004091 − 23.88 0.02759 15.59 0.8166 − 0.8799
3–10 0.004327 − 23.64 0.02798 15.53 0.8597 − 0.6565
3–11 0.004664 − 23.31 0.02147 16.68 1.012 0.05325
3–12 0.004824 − 23.17 0.02194 16.59 1.039 0.1676
3–13 0.003889 − 24.10 0.02610 15.83 0.7909 − 1.019
3–14 0.004736 − 23.25 0.02441 16.12 0.9847 − 0.06694
3–15 0.004666 − 23.31 0.02607 15.84 0.9492 − 0.2266
3–16 0.004573 − 23.40 0.02398 16.20 0.9567 − 0.1923

particular, more disturbances can be introduced to mix the fluid by

larger number of grooves, which attributes to the interrupted and
redeveloping thermal boundary layer. The total contribution ratio of
these three parameters is larger than 74%, revealing that SNR-j is
minimally affected by other factors. Hence, the diameter dg, depth g2
and number of grooves are supposed to be prioritized in the promotion
of the heat transfer characteristic.
As illustrated in Fig. 7 (b), the contribution ratios of control factors to
SNR-f are ordered as P > O > N > Q > M. The number of grooves (P)
plays the primary role in SNR-f accounting for 33.46%. The overall
pressure drop across the region of grooved-fins comes from one along
the fin passage and one resulted from grooves. When air runs through
the dilated flow passage of grooved-fins, the decreasing fluid velocity
transforms to the increasing pressure. Because of the higher pressure
drop in the vortex regions, the increasing number of grooves (P) leads to
more vortexes formed in the grooves, and thereby more pressure drop.
In addition, SNR-f is also affected by groove depth g2 (Q) and groove
diameter (N) with contribution ratios of 32.36% and 12.85%, respec­
tively. The bigger vortexes formed inside the grooves lead to more
pressure drop penalty. The total contribution ratio of those three factors
comes to 78.67%, indicating their prominent impacts on optimizing the
flow resistance characteristic of heat exchangers.
For the criterion to evaluate the thermodynamic performance of heat
exchangers, the control factors are in order of P > O > N > Q > M ac­
cording to the contribution ratios to SNR-JF. It can be observed from
Fig. 7 (c) that SNR-JF is most affected by the number of grooves while
the spacing between grooves w3 is the least important factor. The
contribution ratio of more than 20% can be achieved with all control
factors except the spacing between grooves w3. The contribution ratio of Fig. 7. Main-effect plots and contribution ratios of control factors for grooves
the groove number, depth g2 and diameter dg totals approximately 70%, (Re = 14,080).
which play important roles in designing and optimizing heat ex­
changers. From the perspective of SNR-JF, the optimal condition transfer enhancement. When the number of grooves further increases
(M2N4O2P2Q1) can be achieved by compounding each best level. from P2 to P4, both SNR-j and SNR-f descend, resulting in sharper
The impact of control factors on SNR-JF can be explained according decrease in SNR-JF. Similarly, when the groove depth g2 (O) augments
to the comparison of main-effect plots. In Fig. 7 (c), SNR-JF increases from O1 to O2, the larger increase in SNR-j and fall in SNR-f lead to the
minorly when the number of grooves (P) augments from P1 to P2 increasing SNR-JF. In general, the trend of SNR-JF is more influenced by
because the quick decline in SNR-f counteracts most benefit of heat SNR-j though SNR-f has a certain impact on SNR-JF.

9
Y. Feng et al. International Communications in Heat and Mass Transfer 143 (2023) 106709

4.4. Combination of improvement methods Table 9

Raw results and SNRs for the cases of compound designs.
4.4.1. Factors and orthogonal array for compound designs Case no. j SNR-j f SNR-f JF SNR-JF
The control factors of compound designs on H-type finned tubes
4–1 0.007280 − 21.38 0.03510 14.55 1.341 1.275
included in this study are three parameters, i.e. attack angle α of LVGs, 4–2 0.007402 − 21.31 0.03655 14.37 1.345 1.288
vertical distance w1 of dimples/protrusions and depth g2 of grooves. The 4–3 0.007487 − 21.26 0.03814 14.19 1.342 1.276
attack angle α of LVGs is the major factor influencing the performances 4–4 0.006317 − 21.99 0.02696 15.69 1.271 1.041
after the types of LVGs are determined to be rectangle (G4), and the 4–5 0.007541 − 21.23 0.04036 13.94 1.326 1.226
4–6 0.007480 21.26 0.03912 14.08 1.329 1.235
depth g2 of grooves has the great effect after the number of grooves are
−
4–7 0.006566 − 21.83 0.03151 15.02 1.254 0.9825
fixed to be three (P2). Table 8 lists the levels of control factors. These 4–8 0.007689 − 21.14 0.04204 13.76 1.334 1.251
levels are refined since the ranges in previous sections are not detailed 4–9 0.007701 − 21.13 0.04568 13.40 1.299 1.137
enough to specify the real optimums. Other parameters are designed as 4–10 0.006596 − 21.81 0.03368 14.73 1.232 0.9056
4–11 0.007551 21.22 0.04244 13.72 1.306 1.159
the optimal levels obtained. “None” represents the absence of this −
4–12 0.007693 − 21.14 0.04380 13.58 1.316 1.193
structure, which is aimed to probe whether the corresponding structure 4–13 0.004160 − 23.81 0.01562 18.06 1.004 0.01638
has prominent influence. An orthogonal array of L16 (43) is adopted for 4–14 0.005727 − 22.42 0.02584 15.88 1.168 0.6759
compound designs at Re = 14,080. 4–15 0.005502 − 22.59 0.02411 16.18 1.149 0.6023
4–16 0.005141 − 22.89 0.02183 16.61 1.109 0.4512

4.4.2. SNR analysis for compound designs

The raw results (j, f and JF) and their transformed SNRs are presented and optimizing the heat exchangers. According to the main-effect plot
in Table 9. After the calculations, the effects and contribution ratios of for SNR-JF, the optimal combination (A1H2O3) can be achieved by
all factors and levels to the results for compound designs are displayed in compounding each optimal level, whose structure is displayed in Fig. 9.
Fig. 8. It should be noticed that in compound designs, the optimal values of
It reveals from Fig. 8 (a) that the contribution ratios of control factors attack angle α of LVGs and depth g2 of grooves differ from those in the
to j are ordered as A > O > H with contribution ratios of 60.22%, 32.72% separate studies. Two reasons attribute to this difference. One is the
and 7.06% respectively. The attack angle α of LVGs exerts the most comprehensive impact of the flow in compound designs, and another is
significant influence on j, and the bigger attack angle results in the better that the ranges investigated previously are not detailed enough to
heat transfer performance. Because the longitudinal vortices can intro­ specify the real optimums.
duce stronger disturbances to the flow than grooves and dimples/pro­
trusions, which are favorable for the significant enhancement in heat 5. Verification of the optimization
transfer. The grooves contribute to strengthening the heat transfer due
to the periodical interruption of the thermal boundary layer and vortex 5.1. Analysis of variance (ANOVA)
shedding. The deeper the grooves, the larger vortices formed in the
grooves are, thereby enhancing the heat transfer. The parametric The analysis of variance (ANOVA) is a method to measure the
effectiveness for SNR-j of H-type fins with LVGs/grooves is much higher importance of the relative effects of control factors on JF in statistical.
than those without LVGs/grooves. In contrast, the effect of dimples/ Because the principle and the process are similar, in the interest of
protrusions on enhancing heat transfer is not as obvious as LVGs or saving space, the ANOVA is only performed on three parameters of the
grooves since the introduced disturbance to the flow only occurs near compound designs in this section. Table 10 presents the results of
the edge of H-type fins. There is an optimal depth of dimples/protrusions ANOVA for JF. Variance of a factor measures the effectiveness of the
(H2) at which better heat transfer characteristic can be achieved. target response to the corresponding factor, which can be used to indi­
As demonstrated in Fig. 8 (b), the attack angle α of LVGs (A) has the cate the contribution ratio to the objective. It can be found that the
most contribution of 61.16% to SNR-f while the contribution ratios for contribution of factors has the same order with those in intuitive anal­
the vertical distance w1 of dimples/protrusions (H) and depth g2 of ysis. The attack angle α of LVGs (A) plays the most important role in JF
grooves (O) are 4.89% and 33.95%, respectively. The trends of SNR-f at accounting for 78.77% while the vertical distance w1 of dimples/pro­
the corresponding factors are opposite to those of SNR-j. Since the LVGs trusions (H) is the least significant factor.
obstruct the fluid path, the pressure drop across the heat exchanger rises F test is performed to validate the results at the 95% confidence level.
obviously. The bigger vortexes can be formed inside the deeper grooves, By comparing with the standard tables of proper F test, if the calculated
leading to the increasing pressure drop penalty. The dimples/pro­ result of F is larger, the corresponding factors are identified to have great
trusions can augment the flow resistance because of the introduced contributions to the objective. The results indicate that the attack angle
turbulence. The parametric effectiveness for SNR-f of H-type fins with α of LVGs (A) and depth g2 of grooves (O) play prominent parts in the
LVGs/grooves is much less than those without LVGs/grooves, while the comprehensive thermal–hydraulic performance. Therefore, in innova­
effect of dimples/protrusions on flow resistance is minor. tive structures, these parameters must be given priorities. ANOVA is a
As for the criterion to evaluate the thermodynamic performance of further proof of the correctness of the intuitive analysis.
heat exchangers, the control factors are also in order of A > O > H ac­
cording to the contribution ratios of 62.93%, 28.02% and 9.05% to SNR- 5.2. Additivity by confirmatory test
JF, respectively. It indicates from Fig. 8 (c) that SNR-JF is most affected
by the attack angle α of LVGs (A) while the vertical distance w1 of Only on the premise of the insignificant interaction among control
dimples/protrusions (H) is the least important factor when designing factors, the optimal geometric parameters can be achieved by com­
pounding the optimal levels of all control factors. In order to confirm the
additivity, the effect of interactions needs to be evaluated by comparing
Table 8
Levels of the control factors for compound designs. the predicted optimal SNR-JF with the real tested results. When the
deviation is within ±2 dB, the premise of the insignificant interaction is
Name Factor Level 1 Level 2 Level 3 Level 4
determined. The predicted value is figured out grounded on the attack
A LVG-attack angle α (◦ ) 50 60 70 None angle α of LVGs (A) and depth g2 of grooves (O), which are of statistical
Dimple-vertical distance w1
H 36 38 40 None importance for JF.
(mm)
O Groove-depth g2 (mm) 1.6 1.8 2.0 None

10
Y. Feng et al. International Communications in Heat and Mass Transfer 143 (2023) 106709

Fig. 9. Schematic diagram of the optimal combination (A1H2O3).

Table 10
ANOVA for JF.
Factors Degree of Sum of Variance F-test F0.05 Contribution
freedom square (V) ratio (%)
(df) (SS)

A 3 3.237 1.618 56.39a 4.76 78.77

H 3 0.311 0.078 2.711 4.76 3.79
O 3 0.659 0.330 11.58a 4.76 16.05
Error 6 0.057 0.029 1.40
Total 15 4.265
a
Significant at 95% confidence level.

interactions among control factors demonstrates that Taguchi method

has the reasonability and dependability for optimizing the heat
exchangers.

5.3. Validation of the optimal combination

The compound design of A1H2O3 is identified as the optimal com­

bination based on the structure of plain H-type fins in this paper. As this
optimal combination is chosen under the constant inlet velocity of 3 m/s
(Re = 14,080), further validation for its reliability in a wide range of Re
is required. To be more convincing, case 4–2 (A1H2O2) with the
maximum JF among all 16 cases is selected for the comparison. The
cases A2H2O4 (L-D) and A2H4O2 (L-G) are chosen for the sake of
generalizability, where the LVGs are combined with dimples/pro­
trusions and grooves and these random level combinations are not
embraced in the 16 cases. Furthermore, the optimal combinations for
LVGs (A3B2C1D4E4F4G4), dimples/protrusions (H4I2J1K1L1) and
grooves (M2N4O2P2Q1) are taken into the comparison as well. As dis­
played in Fig. 10, the JF of several cases with different combined levels
varies with Re. It can be observed that JF of the optimal combination
(A1H2O3) is overall larger than that of all other cases. To be specific,
when Re ranges from 4650 to 28,300, JF of case A1H2O3 increases by
2.8–4.6% compared with the largest JF of case 4–2 (A1H2O2) among the
16 cases. It also gains the improvements of 0.9–23.8%, 24.5–57.1% and
8.3–37.5% than those of LVGs, dimples/protrusions and grooves,
respectively. The difference of JF between case A2H4O2 and case 4–2 is
Fig. 8. Main-effect plots and contribution ratios of control factors for com­ minor because the levels of the most influential factor (attack angle of
pound designs (Re = 14,080). LVGs) and the second most influential factor (depth of grooves) for JF
are consistent. This tendency further demonstrates the fact that the
SNRp = SNR + (SNRA -SNR) + (SNRO -SNR) (26) attack angle of LVGs and depth of grooves are primary factors for the
comprehensive performance. The optimal combination is supported to
where SNR averages all results, and SNRA and SNRO are adopted at the be reliable by its outstanding performance.
best level, respectively. Besides, in order to demonstrate the improvement of the optimal
According to the results of simulations, the SNR-JF for the prediction combination, further comparisons with references are required. The
and optimum condition are 1.323 dB and 1.367 dB, whose deviation of Ref. [35] with LVGs, Ref. [21] with dimples/protrusions and Ref. [24]
only 0.0441 dB is smaller than the criterion. The additivity of the with grooves are selected to compare the comprehensive

11
Y. Feng et al. International Communications in Heat and Mass Transfer 143 (2023) 106709

18.31%, while the vertical distance w1 and angle of obliquity β have

minor influences. In terms of flow resistance characteristic, the
attack angle α, winglet type and winglet length g1 exert significant
impacts with a total contribution ratio of 55.16% while the effect of
angle of obliquity β is minimum. In the analysis about the compre­
hensive thermal–hydraulic performance, the prominent factors are
the winglet types, vertical distance w2, attack angle α and winglet
JF

length g1, whose total contribution ratio is 74.82%.

2. For dimples/protrusions, the vertical distance w1 is the primary
factor on the thermal performance amounting for 52.24%, while the
dimple diameter contributes little. All factors affect the flow resis­
tance characteristic, but the depth of dimples/protrusions g2 and
vertical distance w1 are near the top of the list with contribution
ratios of 33.57% and 25.86%, respectively. The comprehensive
thermal–hydraulic performance is most affected by the vertical dis­
tance w1, dimple type and depth of dimples/protrusions g2. The total
Re contribution of these three parameters is larger than 80%.
3. For grooves, the diameter dg, depth g2 and number of grooves play
Fig. 10. Changes of JF for compound designs with different parametric
combinations. more important roles in j with contribution ratios of 32.12%, 20.37%
and 21.53%, respectively. The flow resistance characteristic is most
affected by the groove depth g2 and diameter dg with a total contri­
thermal–hydraulic performance with the optimal combination
bution ratio of over 45%. The number of grooves has the strongest
(A1H2O3). The comprehensive case (LDG) is also been built by
influence on the comprehensive thermal–hydraulic performance,
combining the structures from above three references. The comparison
and the contribution ratio of more than 20% can be achieved by all
is presented in Fig. 11. It is revealed that the optimal combination
the factors of grooves except the spacing between grooves w3.
(A1H2O3) gains the improvements of 5.9–29.5%, 16.0–50.5% and
4. The compound designs consider the factors comprehensively, which
10.2–43.9% than LVGs, dimples and grooves of references, respectively.
are the key elements affecting performances of different innovative
The enhancement of 4.5–16.6% in JF can be achieved by the optimal
structures. The results indicate that the heat transfer characteristic is
combination (A1H2O3) by comparison with the case (LDG). The supe­
mainly depended by LVGs with the contribution ratio of 60.22%,
rior performance of A1H2O3 demonstrates the significance of the
while the LVGs also lead to the most pressure drop penalty. The
compound design for designing and optimizing heat exchangers.
dimples/protrusions are minorly effective to enhance heat transfer or
reduce flow resistance.
6. Conclusion
5. After the multifactorial evaluation on the ground of JF, the com­
pound design of A1H2O3 is identified as the optimal combination on
In this paper, under the condition of the constant basic structure
the basis of the optimization of three innovative structures by means
parameters of plain H-type finned tube heat exchangers, the effects of
of Taguchi method. The additivity of the minor interactions among
different geometric parameters and types of novel designs on the heat
control factors proves that Taguchi method has the reasonability for
transfer, flow resistance and comprehensive thermal–hydraulic perfor­
optimization. When Re ranges from 4650 to 28,300, the optimal
mances are investigated numerically. The innovative structures
compound design gains the improvements of 0.9–23.8%,
considered include LVGs, dimples/protrusions, grooves and their com­
24.5–57.1% and 8.3–37.5% than those of LVGs, dimples/protrusions
pound designs. The contributions of different parameters to the objec­
and grooves, respectively. The enhancement of 4.5–16.6% in JF can
tives (j, f and JF) are assessed and the optimal geometric structures of
be achieved by the optimal combination (A1H2O3) by comparison
these innovative designs are obtained via Taguchi method. The
with the case (LDG) built based on references.
following conclusions are drawn from the investigations:
This paper contributes to getting deeper insights into the effects of
1. For LVGs, the winglet type and attack angle α are the most remark­
different novel structures and innovative compound structure on the
able factors to the thermal performance accounting for 22.76% and
characteristics of flow and heat transfer in H-type finned tube heat ex­
changers, and therefore can be guidelines for designing and optimizing
heat exchangers. Furthermore, these results also have implications in
similar devices for strengthening heat transfer in the microelectronics
industry.

CRediT authorship contribution statement

Yongshi Feng: Investigation, Methodology, Software, Validation,

JF

Data curation, Writing – original draft, Writing – review & editing.

Ruyan Xu: Software, Validation, Funding acquisition. Yi Cao: Formal
analysis, Visualization. Xin Wu: Conceptualization, Methodology,
Writing – review & editing, Supervision. Cai Liang: Software, Project
administration, Funding acquisition. Liyuan Zhang: Software, Investi­
gation, Visualization.

Declaration of Competing Interest

Re
Fig. 11. Comparison of JF for compound designs with cases from references. The authors declare that there is no conflict of interest.

12
Y. Feng et al. International Communications in Heat and Mass Transfer 143 (2023) 106709

Data availability [22] F. Sangtarash, H. Shokuhmand, Experimental and numerical investigation of the
heat transfer augmentation and pressure drop in simple, dimpled and perforated
dimpled louver fin banks with an in-line or staggered arrangement, Appl. Therm.
No data was used for the research described in the article. Eng. 82 (2015) 194–205, https://doi.org/10.1016/j.applthermaleng.2015.02.073.
[23] L. Luo, W. Du, S.T. Wang, W.L. Wu, X.H. Zhang, Multi-objective optimization of the
Acknowledgements dimple/protrusion channel with pin fins for heat transfer enhancement, Int. J.
Numer. Methods Heat Fluid Flow 29 (2019) 790–813, https://doi.org/10.1108/
hff-05-2018-0194.
This work is supported by the National Key R&D Program of China, [24] J. Liu, G.N. Xie, T.W. Simon, Turbulent flow and heat transfer enhancement in
grant number 2018YFB0605303; this work is also supported by Con­ rectangular channels with novel cylindrical grooves, Int. J. Heat Mass Transf. 81
(2015) 563–577, https://doi.org/10.1016/j.ijheatmasstransfer.2014.10.021.
struction and application of cigarette combustion cone model with [25] Q.F. Zhu, H.X. Xia, J.J. Chen, X.M. Zhang, K.P. Chang, H.W. Zhang, H. Wang, J.
different circumferences, grant number H202205. F. Wan, Y.Y. Jin, Fluid flow and heat transfer characteristics of microchannel heat
sinks with different groove shapes, Int. J. Therm. Sci. 161 (2021), 106721, https://
doi.org/10.1016/j.ijthermalsci.2020.106721.
References [26] Y. He, H. Han, S. Tang, T. Zhou, Sulfuric acid deposition characteristics of H-type
finned tube bank with 10 rows, Int. J. Heat Mass Transf. 81 (2015) 137–141,
[1] Y.L. Guan, H.W. Lu, Y.L. Jiang, P.P. Tian, L.H. Qiu, P. Pellikka, J. Heiskanen, https://doi.org/10.1016/j.ijheatmasstransfer.2014.10.013.
Changes in global climate heterogeneity under the 21st century global warming, [27] Y. Jin, Z. Yu, G. Tang, Y. He, W. Tao, Parametric study and multiple correlations of
Ecol. Indic. 130 (2021), 108075, https://doi.org/10.1016/j.ecolind.2021.108075. an H-type finned tube bank in a fully developed region, Numer. Heat Transf. Part A
[2] Unravelling global patterns of irreversible climate change, Nat. Clim. Chang. 12 Appl. 70 (2016) 64–78, https://doi.org/10.1080/10407782.2016.1173433.
(2022) 787–788, https://doi.org/10.1038/s41558-022-01456-9. [28] P. Bhale, M. Kaushik, J. Liaw, C. Wang, Airside performance of H-type finned tube
[3] H. Yang, X.J. Huang, J.L. Hu, J.R. Thompson, R.J. Flower, Achievements, banks with surface modifications, Energies 12 (2019) 584, https://doi.org/
challenges and global implications of China' s carbon neutral pledge, Front. 10.3390/en12040584.
Environ. Sci. Eng. 16 (2022) 111, https://doi.org/10.1007/s11783-022-1532-9. [29] T. Han, C. Wang, Q. Cao, W. Chen, D. Che, Investigation on heat transfer
[4] S. Zhang, W.Y. Chen, China’s energy transition pathway in a carbon neutral vision, characteristics of the H-type finned tube in flue gas with high content of ash,
Engineering 14 (2022) 64–76, https://doi.org/10.1016/j.eng.2021.09.004. Energy Procedia 105 (2017) 4680–4684, https://doi.org/10.1016/j.
[5] S.D. Feng, X.P. Zhang, H.N. Zhang, Multi-objective optimization of coal-fired egypro.2017.03.1014.
power units considering deep peaking regulation in China, Environ. Sci. Pollut. [30] X. Li, D. Zhu, J. Sun, X. Mo, Y. Yin, Air side heat transfer and pressure drop of H
Res. (2022), https://doi.org/10.1007/s11356-022-22628-2 accessed Sep. 09. type fin and tube bundles with in line layouts, Exp. Thermal Fluid Sci. 96 (2018)
[6] J.J. Wang, J.K. Huo, S. Zhang, Y. Teng, L. Li, T.Y. Han, Flexibility transformation 146–153, https://doi.org/10.1016/j.expthermflusci.2018.02.029.
decision-making evaluation of coal-fired thermal power units deep peak shaving in [31] X. Wu, Y. Shang, J. Wang, G. Ren, Experimental research heat transfer
China, Sustainability 13 (2021) 1882, https://doi.org/10.3390/su13041882. characteristics of H-type finned tube bundles, J. Southeast Univ. 43 (2013) 88–93,
[7] L. Lin, B.Q. Xu, S.W. Xia, Multi-angle economic analysis of coal-fired units with https://doi.org/10.3969/j.issn.1001-0505.2013.01.017. (in Chinese).
plasma ignition and oil injection during deep peak shaving in China, Appl. Sci. 9 [32] H. Chen, Y. Wang, Q. Zhao, H. Ma, Y. Li, Z. Chen, Experimental investigation of
(2019) 5399, https://doi.org/10.3390/app9245399. heat transfer and pressure drop characteristics of H-type finned tube banks,
[8] Y. Zhao, P. Fan, C. Wang, M. Liu, D. Chong, J. Yan, Fatigue lifetime assessment on a Energies 7 (2014) 7094–7104, https://doi.org/10.3390/en7117094.
high-pressure heater in supercritical coal-fired power plants during transient [33] Y. Jin, G. Tang, Y. He, W. Tao, Parametric study and field synergy principle
processes of operational flexibility regulation, Appl. Therm. Eng. 156 (2019) analysis of H-type finned tube bank with 10 rows, Int. J. Heat Mass Transf. 60
196–208, https://doi.org/10.1016/j.applthermaleng.2019.04.066. (2013) 241–251, https://doi.org/10.1016/j.ijheatmasstransfer.2012.11.043.
[9] C.W. Lu, J.M. Huang, W.C. Nien, C.C. Wang, A numerical investigation of the [34] H. Wang, Y. Liu, P. Yang, R. Wu, Y. He, Parametric study and optimization of H-
geometric effects on the performance of plate finned-tube heat exchanger, Energy type finned tube heat exchangers using Taguchi method, Appl. Therm. Eng. 103
Convers. Manag. 52 (2011) 1638–1643, https://doi.org/10.1016/j. (2016) 128–138, https://doi.org/10.1016/j.applthermaleng.2016.03.033.
enconman.2010.10.026. [35] S.Z. Tang, F.L. Wang, Y.L. He, Y. Yu, Z.X. Tong, Parametric optimization of H-type
[10] Farouk Tahrour, Hijaz Ahmad, Houari Ameur, Tareq Saeed, Hanaa Abu-Zinadah, finned tube with longitudinal vortex generators by response surface model and
Younes Menni, 3D numerical study and comparison of thermal-flow performance genetic algorithm, Appl. Energy 239 (2019) 908–918, https://doi.org/10.1016/j.
of various annular finned-tube designs, J. Ocean Eng. Sci. (2022), https://doi.org/ apenergy.2019.01.122.
10.1016/j.joes.2022.02.009 accessed Feb. 15. [36] X.B. Zhao, G.H. Tang, X.W. Ma, Y. Jin, W.Q. Tao, Numerical investigation of heat
[11] S. Wongwises, Y. Chokeman, Effect of fin pitch and number of tube rows on the air transfer and erosion characteristics for H-type finned oval tube with longitudinal
side performance of herringbone wavy fin and tube heat exchangers, Energy vortex generators and dimples, Appl. Energy 127 (2014) 93–104, https://doi.org/
Convers. Manag. 46 (2005) 2216–2231, https://doi.org/10.1016/j. 10.1016/j.apenergy.2014.04.033.
enconman.2004.09.011. [37] X.B. Zhao, G.H. Tang, Y.T. Shi, Y.K. Li, Experimental study of heat transfer and
[12] P. Pongsoi, S. Pikulkajorn, C.C. Wang, S. Wongwises, Effect of number of tube rows pressure drop for H-type finned oval tube with longitudinal vortex generators and
on the air-side performance of crimped spiral fin-and-tube heat exchanger with a dimples under flue gas, Heat Transf. Eng. 39 (2018) 608–616, https://doi.org/
multipass parallel and counter cross-flow configuration, Int. J. Heat Mass Transf. 10.1080/01457632.2017.1325658.
55 (2012) 1403–1411, https://doi.org/10.1016/j.ijheatmasstransfer.2011.09.064. [38] X.L. Li, S.Q. Wang, D.L. Yang, G.H. Tang, Y.C. Wang, Thermal-hydraulic and
[13] H. Zhou, D. Liu, Q.Y. Sheng, M.Y. Hu, Y. Cheng, K.F. Cen, Research on gas side fouling performances of enhanced double H-type finned tubes for residual heat
performance of staggered fin-tube bundles with different serrated fin geometries, recovery, Appl. Therm. Eng. 189 (2021), 116724, https://doi.org/10.1016/j.
Int. J. Heat Mass Transf. 152 (2020), 119509, https://doi.org/10.1016/j. applthermaleng.2021.116724.
ijheatmasstransfer.2020.119509. [39] G. Taguchi, Taguchi on Robust Technology Development. Bring Quality
[14] H.Z. Li, H.J. Wang, M.Y. Yao, L.X. Zhang, H.F. Gu, J.P. Nie, PIV and thermal-vision Engineering (QE) Upstream, ASME, 1991.
experimental and numerical investigation on the airside performance of slotted fin [40] T. Mori, Taguchi Method: Benefits, Impacts, Mathematics, Statistics and
surfaces, Int. J. Heat Mass Transf. 82 (2015) 568–580, https://doi.org/10.1016/j. Applications, ASME Press, 2011.
ijheatmasstransfer.2014.11.039. [41] M.M. Etghani, S.A.H. Baboli, Numerical investigation and optimization of heat
[15] M. Fiebig, P. Kallweit, N.K. Mitra, Wing-type vortex generators for heat transfer transfer and exergy loss in shell and helical tube heat exchanger, Appl. Therm. Eng.
enhancement, IHTC 6 (1986) 2909–2913, https://doi.org/10.1615/IHTC8.1460. 121 (2017) 294–301, https://doi.org/10.1016/j.applthermaleng.2017.04.074.
[16] S.W. Hwang, D.H. Kim, J.K. Min, J.H. Jeong, CFD analysis of fin tube heat [42] J.N. Zhang, M. Cheng, Y.D. Ding, Q. Fu, Z.Y. Chen, Influence of geometric
exchanger with a pair of delta winglet vortex generators, J. Mech. Sci. Technol. 26 parameters on the gas-side heat transfer and pressure drop characteristics of three-
(2012) 2949–2958, https://doi.org/10.1007/s12206-012-0702-2. dimensional finned tube, Int. J. Heat Mass Transf. 133 (2019) 192–202, https://
[17] S.M. Pesteei, P.M.V. Subbarao, R.S. Agarwal, Experimental study of the effect of doi.org/10.1016/j.ijheatmasstransfer.2018.12.118.
winglet location on heat transfer enhancement and pressure drop in fintube heat [43] L.H. Tang, S.C. Tan, P.Z. Gao, M. Zeng, Parameters optimization of fin-and tube
exchangers, Appl. Therm. Eng. 25 (2005) 1684–1696, https://doi.org/10.1016/j. heat exchanger with a novel vortex generator fin by Taguchi method, Heat Transf.
applthermaleng.2004.10.013. Eng. 37 (2016) 369–381, https://doi.org/10.1080/01457632.2015.1052715.
[18] J.M. Wu, W.Q. Tao, Effect of longitudinal vortex generator on heat transfer in [44] M. Zeng, L.H. Tang, M. Lin, Q.W. Wang, Optimization of heat exchangers with
rectangular channels, Appl. Therm. Eng. 37 (2012) 67–72, https://doi.org/ vortex-generator fin by Taguchi method, Appl. Therm. Eng. 30 (2010) 1775–1783,
10.1016/j.applthermaleng.2012.01.002. https://doi.org/10.1016/j.applthermaleng.2010.04.009.
[19] H. Naik, S. Harikrishnan, S. Tiwari, Numerical investigations on heat transfer [45] Y.S. Feng, C. Liang, X. Wu, H. Fang, Parametric study on thermal-hydraulic
characteristics of curved rectangular winglet placed in a channel, Int. J. Therm. Sci. characteristics of H-type finned tube heat exchangers, J. Thermophys. Heat Transf.
129 (2018) 489–503, https://doi.org/10.1016/j.ijthermalsci.2018.03.028. (2022), https://doi.org/10.2514/1.T6689 accessed Sep. 26.
[20] C.C. Wang, K.Y. Chen, Y.T. Lin, Investigation of the semi-dimple vortex generator [46] J.Y. Yun, K.S. Lee, Influence of design parameters on the heat transfer and flow
applicable to fin-and-tube heat exchangers, Appl. Therm. Eng. 88 (2015) 192–197, friction characteristics of the heat transfer with slit fins, Int. J. Heat Mass Transf. 43
https://doi.org/10.1016/j.applthermaleng.2014.09.054. (2000) 2529–2539, https://doi.org/10.1016/s0017-9310(99)00342-7.
[21] A. Gupta, M. Kumar, A.K. Patil, Enhanced heat transfer in plate fin heat sink with [47] Y.S. Feng, X. Wu, C. Liang, Z.P. Sun, A convenient method for the accurate
dimples and protrusions, Heat Mass Transf. 55 (2019) 2247–2260, https://doi.org/ calculation of fin efficiency of H-type fins based on linear nomograms and fitting
10.1007/s00231-019-02561-w. formulae, Energies 15 (2022) 456, https://doi.org/10.3390/en15020456.

13