New Journal and we have not received input yet 26 (2021) 101132

Contents lists available at ScienceDirect

Thermal Science and Engineering Progress

journal homepage: www.sciencedirect.com/journal/thermal-science-and-engineering-progress

Multi-objective optimum design for double baffle heat exchangers

Bahador Abolpour a, *, Ramtin Hekmatkhah b, Rahim Shamsoddini c
a
Department of Chemical Engineering, Sirjan University of Technology, Sirjan, Iran
b
Department of Aerospace Engineering, The Faculty of New Sciences and Technologies, The University of Tehran, Tehran, Iran
c
Department of Mechanical Engineering, Sirjan University of Technology, Sirjan, Iran

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

Keywords: The enhancement of the heat transfer rate of heat exchangers with a lower-pressure drop in passing fluid flow
Optimum design always was a consideration and challenging purpose for researchers. Increasing industrial equipment efficiency is
Turbulent flow a crucial problem for engineers. A multi-objective optimal design in order to increase heat transfer rate and low-
Genetic algorithm
pressure drop in a two-dimensional baffle heat exchanger considering turbulent fluid flow in a channel is pre­
Baffle heat exchanger
sented. The genetic algorithm is utilized for obtaining an optimum arrangement of two flat-plate for satisfying
mentioned purposes. The image processing method is used for detecting the edges of the mounted baffles and the
appropriate mesh configuration is generated for the solution domain. Turbulent fluid flow and energy equations
are discretized and solved over the physical domain by the finite volume method. The combination of genetic
algorithm and image processing method with computational fluid dynamics provides an accurate new approach
for optimization targets that is investigated in the present study. All generated baffles dimensions by the genetic
algorithm are evaluated to achieve the design with an optimum value of a liner scaled function of two target
parameters (i.e. temperature and pressure variations of the passing fluid flow through the channel). Investigating
the results of all the generated and evaluated designs obtains the Pareto’s front in addition to the optimum design
for these target parameters. For a case study with the same importance for the pressure drop (in Pa) and the
passing fluid temperature increment (in K), the optimum baffles arrangement, with two 2 cm height baffles with
22 cm distance, shows a suitable heat transfer rate (ΔT = 32.04 K) and a low-pressure drop (ΔP = 3.007 kPa) for
the passing fluid flow. All the possible optimum designs for this double baffles heat exchanger, with different
importance of ΔT and ΔP, are obtained using the Pareto method.

Investigating newly published papers that focused on the baffled

channels, shell and tube or other types of heat exchangers, indicates that
Introduction
the main aims of those researches were improving their heat transfer
rate and decreasing the passing fluid flow pressure drop. After Patankar
Heat exchangers have wide applications in different industrial sys­
et al. [3], which investigated the forced heat convection in periodically
tems. The main purpose of utilization of this equipment is heat transfer
varying cross-section ducts, other authors [4,5] studied heat and fluid
among different fluid flows and solid bodies. Therefore, increasing the
flow in different shape designs of channels, numerically of experimen­
rate of this heat transfer rate is a considerable problem in this equip­
tally. They investigated different structures, such as a duct with a baffle
ment. Passing fluid flows through heat exchangers and also different
in the intake region [4], a rectangular channel with two baffle plates
solid parts of this equipment affect this heat transfer rate, and each one
mounted on the upper and lower walls [6] or staggered wall-mounted
of these cases that has a very high heat transfer resistance, in the com­
obstacles on these walls [7], a square duct with various arrangements
parison with the other cases, controls the overall heat transfer rate
of fins [8], a heat exchanger tube with a baffle inside the tube [9], the
within this equipment [1,2]. Attending to the controlling case, various
shell side of a shell and tube heat exchanger fitted with different ar­
mechanisms have been developed for decreasing the heat transfer
rangements of segmental baffles [10], a rectangular channel with in­
resistance of that case and improving the overall rate of heat transfer of
clined solid and perforated baffles [11], and porous block shapes inside a
the heat exchanger. Utilizing different shapes and designs of baffles is a
parallel plate channel [12]. Berner et al. [13] investigated the turbulent
well-known method for improving this rate in the fluid flow controlling
flow in a duct with several baffle plates to determine the required
cases [1].

* Corresponding author.
E-mail address: [email protected] (B. Abolpour).

https://doi.org/10.1016/j.tsep.2021.101132
Received 26 August 2021; Received in revised form 29 October 2021; Accepted 2 November 2021
Available online 9 November 2021
2451-9049/© 2021 Elsevier Ltd. All rights reserved.
B. Abolpour et al. Thermal Science and Engineering Progress 26 (2021) 101132

√̅̅̅̅
Nomenclature u* Friction velocity [35]: u* = τρw
U0 Inlet velocity (m . s− 1 )
C Constants [40]: Cμ = 0.09, Cε1 = 1.35, Cε2 = 1.8 u1,u2 Components of the fluid flow velocity vector (m . s− 1 ): u,v
*
Cf Friction coefficient:Cf = qτ∞w y+ Normal distance in wall coordinates [40]: y+ = ρuμ y
dpc Dot per centimeter w1 & w2 Weight coefficient
f The damping functions [40]: fμ = 1 − exp − 0.0115y+
H Heat transfer coefficient: h = ΔT Q Greek symbol
h1 & h2 Baffle heights ε Turbulent kinetic energy dissipation rate (J. kg− 1 )
K The turbulent kinetic energy (J. kg− 1 ) ε0 Inlet turbulent kinetic energy dissipation rate (J. kg− 1 )
k0 Inlet turbulent kinetic energy (J. kg− 1 ) μ The viscosity of the fluid (kg. m− 1 . s− 1 )
n Normal direction to the cell face μt Turbulent viscosity of the flow based on the k-ε model
2
P Mean pressure (kg. m− 1 . s− 2 ) (Pa . s)[45]: μt = Cμ fμ ρ kε
Pout Outlet pressure (kg. m− 1 . s− 2 ) μeff The effective viscosity [56]: μeff = μt + μ
Q Heat flux: Q = − k ∂∂Tn Γeff The effective thermal diffusivity [46]: Γeff = Pr
μ μt
+ Prt
(− 3)
q∞ Dynamic pressure: q∞ = 12 ρU20 ρ The density of the fluid (kg. m )
T Temperature (K) σ k, σε Turbulent Prandtl numbers [40]: σk = 1, σε = 1.3
T0 Inlet temperature (K) τw Wall shear stress (Pa)
Twall Wall temperature (K) ψ Variables indication

number of baffles for obtaining a periodic boundary condition depend­ axial velocity reduction are related to helix angle and axial overlapped
ing on Reynolds number and the geometry. They resulted that with ratio. They obtained a set of Pareto solutions to optimize the thermo­
different Reynolds numbers, different numbers of baffles are necessary dynamic and economic behaviors. The comparison of the optimization
for obtaining a periodic boundary condition, respectively. Habib et al. methods showed that the multi-objective optimization method is more
[14] studied the influence of Reynolds number and baffle height on the suitable than the single-objective optimization strategy. Honrubia et al.
heat transfer coefficient and also the pressure drop of a turbulent flow [58] developed a design based on the multi-objective tool for sizing shell
inside a rectangular duct with a uniform heat flux in walls, experi­ and tube heat exchangers. The feasibility of implementing artificial in­
mentally. They observed large recirculation regions and velocity gra­ telligence tools for the design of heat exchangers with the target of the
dients behind these baffles, and resulted that the pressure drop increases industry applications has been investigated in their work. The dataset
more rapidly than the heat transfer coefficient with the Reynolds was developed to train neural networks and was modeled in MATLAB
number. Li and Kottke [15], simulated heat transfer and pressure drop in software. The optimal configuration of the heat exchanger and design
shell-and-tube heat exchangers and resulted that at a constant Reynolds parameters (the pressure drop on the shell side, the pressure drop on the
number, increasing this distance increases the heat exchange coefficient tube side, and the heat transfer rate) have been optimized. Kaya and Ust
and also its pressure drop. [59] compared different optimization methods. They introduced Uni­
Optimization of structural design or flow parameters using compu­ variate Search Method-2 for optimization of heat exchangers using a
tational fluid dynamics is an attractive study field for researchers. In the single-objective optimization method by moving from two arms on the
fluid flow fields, Falco [16] and also Makinen et al. [17] designed op­ Pareto Front. To decrease computational time, Univariate Search
timum shapes of an airfoil, Han and Maeng [18] optimized the shapes of Method has been chosen and the Genetic algorithm and Non-dominant
cut-off in a multi-blade fan/scroll system, and Madsen et al. [19] opti­ Sorting Genetic algorithm-2 methods were also used in their study.
mized wall shapes of incompressible diffusers, using computational fluid Their results present that the Genetic algorithm has provided better
dynamic approaches. In the heat transfer fields, Fabbri [20] optimized fitness values than the Univariate Search Method. However, Univariate
heat exchange through smooth and corrugated walls, Cheng and Chang Search Method-2 has offered a more suitable design range in the Pareto-
[7] optimized the shape design of a cylinder with heat transfer, Tiwari Front. Saijal and Danish [60] developed a numerical model for investi­
et al. [21] optimized angles of attack for the delta winglets mounted on gating the effect of different parameters on the performance of the shell
the fin-surface on top of oval-shaped tubes, Matos et al. [22] maximized and tube heat exchanger with staggered baffles. A genetic algorithm is
the heat transfer rate of finned and non-finned circular and elliptic carried out to maximize the heat transfer rate while minimizing the
tubular arrangements, Kock and Herwig [23] minimized the entropy pressure drop.
production in a flow-through a heated tube with an inserted twisted tape Chen, et al. [61] designed water–water heat transfer experiment to
with different slopes, and the shapes of heat exchangers fins were compare the hydrodynamics and heat transfer characteristics of the
optimized by other researchers [24–28]. Some researchers presented three heat exchangers with triple-layer flower baffles, double-layer
Multi-objective optimizations coupled with computational fluid dy­ flower baffles, and segmental baffles. Experimental data showed that,
namics to maximize the performance of a heat sink [29] or a micro-heat- at the same flow rate, the shell-side heat transfer coefficients of the shell-
exchanger [30]. Developing the abilities of the computer made it and-tube heat exchanger with triple-layer flower baffles and double-
possible to solve complex problems in an acceptable calculative time. layer flower baffles are higher than those of traditional segmental baf­
These abilities helped researchers to optimize the shape design of fles. Thondiyil and Kodakkattu [62] presented the numerical simulation
different industrial equipment using the coupled simulation of flow and using the Taguchi method to optimize a shell and tube heat the
heat transfer with the multi-objective optimization algorithms, such as exchanger with staggered baffles corresponding to maximum heat
genetic algorithms [31–37]. transfer rate and minimum pressure drop. The heat transfer rate has
Cao and et al. [57] explored heat transfer and fluid flow character­ been calculated and the total pressure drop across the shell is obtained
istics of helical baffle heat exchangers with the aid of numerical and using computational fluid dynamics. Also, a set of experiments have
experimental analyses. They utilized a multi-objective genetic algorithm been considered as per Taguchi experimental design using five design
approach optimization design. The design variables such as; the tube parameters of the heat exchanger to reduce the number of experiments.
number, tube dimension, and helix angle have been perceived as design In the present study, a two-dimensional numerical model for a
variables. Their results investigated that the thermal enhancement and forced-convective heat transfer in a turbulent fluid flow channel with

2
B. Abolpour et al. Thermal Science and Engineering Progress 26 (2021) 101132

two flat-plate baffles has been developed in Matlab Programming Soft­ survival probability (0.5), average probability (0.33), crossover proba­
ware. The calculation mesh has been generated over the flow field by the bility (0.17), mutation probability (1), and mutation magnitude (0.3),
image processing method. The continuity, momentum, low Reynolds k- have been defined in this code. In this code, each population contains
ε, and energy equations are implemented to model steady-state forced 100 generated chromosomes, which each chromosome defines a certain
convective heat transfer turbulent flow, and the finite volume method is structural design for this double baffle channel. A JPEG format black and
utilized to discretize and solve governing equations. The association of white digital image of this channel has been imported to the developed
the finite volume method with genetic algorithm and image processing homemade MATLAB code for the image processing. All edges and cor­
method in the present study authorizes a new approach for optimization ners of this channel have been detected and then appropriate mesh cells
that has not been seen in the previous works. The predictions of the have been generated around baffles and through the channel flow,
present model are validated with the previous experimental and nu­ automatically, as shown in Fig. 2. Then, the relevant governing equa­
merical results for approving the accuracy of the calculation. The main tions have been solved in these generated calculation cells, using the
novelty of the present study is an optimization of the structural design of developed computational fluid dynamic MATLAB code in this study, for
double baffles with the purpose of maximum temperature increment and obtaining the targets (i.e. the average bulk temperature difference ΔT
minimum pressure drop of the passing airflow using the genetic algo­ and pressure difference ΔP, between the channel inlet flow and the flow
rithm approach. For this purpose, the linear weighted function of pa­ passing the second baffle) of this multi-objective optimization process in
rameters is considered as the target function for this multi-objective this channel structure.
optimization problem, according to the Pareto linear scalarization Chien’s low Reynolds k-ε model [39,40], Launder and Spalding
method. Finally, the Pareto Front and also optimum solution are ob­ approach [41], Franke and Rodi approach [42], Kato and Launder
tained, then the velocity, pressure, and temperature gradients of the approach [43,44], and Boussinesq approximation [45] have been used
passing turbulent fluid flow through the obtained optimum structure of to model this incompressible turbulent fluid flow. The steady-state
the baffles heights and distance is evaluated. governing equations for the fluid [39–45] and heat [46] flows have
been illustrated as follows according to Einstein’s notation, which has
Methodology been solved concurrently:
∂ui
Fig. 1 illustrates the physical structure and also the used boundary =0 (1)
∂xi
conditions for the modeled channel in the present study. A turbulent
airflow (Re = 75,000), with a constant inlet velocity (U0 = 7.8 m/s) and
temperature (T0 = 300 K), enters this channel. Constant temperature
(Tw = 375 K) and no-slip boundary conditions have been considered on

[ ] [ ]
∂ ∂ 1 ∂ui 2 ∂ ∂u ∂u ∂μ ∂u ∂μeff ∂uj
ρ (u u ) = − P − μeff + ρk + μ ( i + j ) + eff j − (2)
∂xi i j ∂xj 3 ∂xi 3 ∂xi eff ∂xj ∂xi ∂xj ∂xi ∂xi ∂xj

the impermeable channel’s walls and baffles. These baffle’s heights (i.e.
h1 and h2, variable between zero to 11 cm) and their distance (i.e. d, [ ][ ]
∂ [ui k] ∂ μ ∂ k ρCμ k2 ∂ui ∂uj ∂ui ∂uj
variable between 11 and 26 cm) have been considered as variable values ρ = [μ + t ]+ + − − ρε (3)
∂xi ∂xi σ k ∂xi 2ε ∂xj ∂xi ∂xj ∂xi
for optimizing by the genetic algorithm. At the inlet, constant values
[ ][ ]
have been concerned (inlet turbulent kinetic energy (k0 = 32(Tu U0 )2 ) is ∂[ui ε] ∂ μ ∂ε ρCμ k ∂ui ∂uj ∂ui ∂uj ε2
ρ = [μ + t ] + Cε1 + − − ρCε2 (4)
calculated from the equation of the isotropic turbulence, and Kolmo­ ∂xi ∂xi σ ε ∂xi 2 ∂xj ∂xi ∂xj ∂xi k
1.5
gorov expression is used for dissipation term (ε0 = Cμ0 kLε ) [38]), and at
the outlet, the gradient of variables along the x-axis is neglected. At the
first calculative iteration, the developed code detects downstream of
these baffles in the flow field as a bluff body, i.e. sets all velocity com­
ponents along downstream-direction equal to zero, to improve the
calculation time and the convergence chance of the model.
In the general structure of genetic algorithms, each problem variable
(i.e. d, h1, and h2) is called a gene, putting all these variables together
makes a chromosome, and a set of chromosomes makes up a population.
Parameters of this genetic algorithm, such as generations (1 0 0),

Fig. 2. Edge and corner detection process and mesh generation using the image
Fig. 1. Physical domain and boundary conditions for the channel. processing method.

3
B. Abolpour et al. Thermal Science and Engineering Progress 26 (2021) 101132

( )
∂ ∂ ∂T Table 1
ρ (ui T) = Γeff (5) Grid study.
∂xi ∂xj ∂xj
Mesh Mean squared error hmean (w/m2 K) Cfmean × Calculation time
The finite volume method has been used to solve the governing
(dpc) between calculated u 104 using Intel Core
equations, which have been discretized based on the Power-law differ­ and experimental i5-3470 CPU
encing scheme [47], and a staggered grid and SIMPLE algorithm have values at x = 0.525 m (min)
been utilized [48]. The under-relaxation factors have been considered [6]
equal to 0.3, 0.3, 0.5, 0.7, 0.7 and 0.7 for the u, v, P, k, ε and T, 0.5 1.828 0.983 2.199 2.703
respectively [49]. Maximum 50,000 iterations and minimum 10− 6 ac­ 1 1.054 1.222 4.201 4.996
curacy have been considered for acceptable convergence of the calcu­ 2 0.563 1.905 5.342 8.878
lated variables (i.e. u, v, P, k, ε, and T). After solving governing 4 0.211 2.572 6.972 12.551
8 0.281 2.588 6.975 29.363
equations, velocity, pressure, and temperature fields are estimated for
16 0.238 2.58 6.977 112.2
all the calculation domains.
This optimization problem consists of finding the best channel
structure to increase heat exchange while at the same time limiting the
pressure loss, which is inter-related. If the baffle heights increases, the
heat exchange will be favored and ΔT will be also enhanced. But,
simultaneously, for a given airflow rate ΔP will increase, and the heat
exchanger loses efficiency. For solving this multi-objective problem, a
linear scalarization of these objective parameters with fixed weights has
been considered, as Fit = w1 ΔT +w2 ΔP [50–52]. The choice of these
weights (i.e. w1 and w2) associated with each objective parameter
changes the solution of the optimization problem. In this study,
considering the same importance for both of these targets, i.e. high-
temperature increment (in K unit) and low-pressure drop (in kPa
unit), we defined w1 = 1 and w2 = -1 and tried to maximize the value of
the Fit function for obtaining the optimum solution using the genetic
algorithm approaches. Nevertheless, for this multi-objective optimiza­
tion problem, no single solution exists that simultaneously optimizes
each objective parameter, and there exists a (possibly infinite) number
of Pareto optimal solutions, attending to the fact that none of these Fig. 4. The convergence procedure of the present numerical solution.
targets can be improved without degrading the other target [53,54]. It
should be noted that this multi-objective optimization problem can be
studied from different viewpoints and, thus, there exist different solu­
tion philosophies and goals. Therefore, the Pareto optimal solutions and
also Pareto front have been obtained for this problem, which indicate
the set of optimum solutions with different values of w1 and w2.

Results and discussion

In previous studies, a certain structure of this double baffle, with h1

= h1 = 0.8 m and d = 0.142 m, was investigated numerically and
experimentally [6,55]. Obtained results in these studies have been used
for validating the predictions of the present model. Fig. 3 presents
comparisons among the predictions of this model (with different mesh
sizes, i.e. 1, 4, and 8 dots per centimeter (dpc) with the obtained

Fig. 5. The predicted fluid flow velocity vectors and pressure contour by the
model for the validated channel.

experimental and numerical results for the axial fluid flow velocity
profile after the second baffle plate, near the channel outlet (i.e. x =
0.525 m). The model with meshes 4 and 8 dpc have acceptable pre­
dictions for this profile. In addition, Table 1 compares the calculated
mean values of the heat transfer coefficient (hmean ) and friction coeffi­
cient (Cfmean ) at the walls of the channel by the presented model with
different mesh sizes. It is observed that these calculated values are close
together for meshes smaller than 4dpc. Therefore, attending to lower
calculation cost, the 4 dpc is a suitable mesh for the optimization process
of this problem.
For this case, Fig. 4 shows the convergence procedure of this
Fig. 3. The validation of the present model and grid study.

4
B. Abolpour et al. Thermal Science and Engineering Progress 26 (2021) 101132

Fig. 6. The predicted streamlines and temperature contour by the model for the
validated channel. Fig. 8. Searched domains of the optimization variables by genetic algorithm;
the vertical surface reflected these 3D points in a 2D-view, as a mirror for a
better view.

Fig. 7. The variations of the heat transfer coefficient and friction coefficient for
the validated channel.
Fig. 9. Predicted target values by genetic algorithm.
numerical solution, Fig. 5 shows the model predictions for the velocity
vectors and pressure contour, and Fig. 6 presents the streamlines and
temperature contour of the passing airflow through this channel. It is
clear that this channel has a high heat transfer rate and also a high-
pressure drop for the passing airflow. For this case, calculated average
values of the bulk temperature and pressure differences for the passing
fluid flow between the channel inlet and at its second baffle location (i.e.
x = 0.36 m) are ΔT = 49.843 K and ΔP = 56.498 kPa, respectively. The
high interface area of airflow-walls contact and large circulation zones
in this fluid flow field behind these baffles, resulted in these suitable heat
transfer performance and unsuitable pressure drop, in this structure.
Fig. 7 presents the variations of the calculated heat transfer coefficient
(h: from 0 to hmax , with blue to red color markers, respectively) and also
friction coefficient (Cf : from 0 to Cfmax , with green to black color markers,
respectively) at the wall boundaries of the validated channel. It is
observed that after the second baffle, at the upper wall of this channel,
we have the maximum value of these parameters attending to the high
passing flow velocity near this wall.
As previously mentioned, the main aim of this study is obtaining the
optimum design of this channel for satisfying two targets, high heat
transfer rate and low-pressure drop of the passing airflow. The genetic
algorithm approach has been used for obtaining the optimum heights (i.
e. h1 and h2) and distance (i.e. d) of these baffles within this channel for
this purpose. Fig. 8 is a three-dimensional view of searched domains of Fig. 10. The model predictions for the fluid flow velocity vectors and pressure
contour for the optimum designed channel.

5
B. Abolpour et al. Thermal Science and Engineering Progress 26 (2021) 101132

Fig. 11. The model predictions for the streamlines and temperature contour for
the optimum designed channel.

Fig. 12. The variations of the heat transfer coefficient and friction coefficient
these variables by first and final generated populations by genetic al­ for the optimum designed channel.
gorithm. Attending to the importance of the heights of these baffles in
this optimization problem, the vertical surface of this 3D plot reflects the
model. Finally, the Pareto solutions, Pareto front, and optimum solution
h1 and h2 values of these chromosomes. Fig. 9 shows the predicted values
of this multi-objective problem were obtained using the genetic algo­
of ΔT and ΔP for all the generated populations by genetic algorithm. In
rithm approach. Using the presented method in this study helps us to
these Figures, the convergence of the final generated population and
obtain the optimum structural design for any multi-objective strategies
also their calculated ΔT and ΔP to the Pareto front is clear. This is a
such as high heat transfer ratio and low-pressure drop. In addition, any
border curve between the feasible and infeasible regions in this inter-
operating performance of this equipment can be considered as an opti­
relative diagram of both of the defined target values. The located an­
mization variable to achieving any multi-objective optimization strate­
swers in this border are Pareto solutions, which indicate the best
gies using a similar combination of genetic algorithm and computational
possible solutions for this problem attending to different possible values
fluid dynamic, like the presented model in this study. The presented
of w1 and w2.
method in this study can be used for optimizing the shape or other static
Figs. 8 and 9 also show the obtained optimum solution (as one of the
or dynamic characters in different mass, heat, or fluid flow problems,
Pareto solutions with w1 = 1 and w2 = -1, which considered the same
such as drag reduction in aerodynamic designs, well mixing in any
importance for both of the objective targets, i.e. temperature increment
shapes of mixers, Nano-fluid flows, filter designs for removing solid
in K and pressure drop in kPa). Figs. 10 and 11 show the model pre­
particles, etc. Multi-objective optimization of three-dimensional prob­
dictions for the velocity vectors, streamlines, and pressure and tem­
lems also is possible using the presented method by supercomputers.
perature contours of the passing fluid flow through this optimum
designed channel. This channel has a suitable heat transfer rate (ΔT =
32.04 K) and a low-pressure drop (ΔP = 3.007 kPa) for the passing fluid CRediT authorship contribution statement
flow. Considering similar weights for these two objective targets caused
to found an optimum solution design that has a low interface area of Bahador Abolpour: Conceptualization, Software, Validation,
airflow-walls contact and small circulation zones in this fluid flow field Writing - review & editing, Supervision. Ramtin Hekmatkhah: Formal
behind these baffles, which have a medium heat transfer performance analysis, Resources, Writing – original draft, Visualization. Rahim
and low-pressure drop, consequently. Other obtained Pareto solutions in Shamsoddini: Data curation, Writing - review & editing, Supervision.
Fig. 9 indicate different optimum structures for this channel attending to
different possible strategies of the designer (i.e. deferent values of w1 and
w2). Therefore, it is possible to design optimum structures for this Declaration of Competing Interest
channel with different importance ratios of the target parameters using
the presented method. Fig. 12 shows the variations of the calculated heat The authors declare that they have no known competing financial
transfer coefficient and also friction coefficient at the wall boundaries of interests or personal relationships that could have appeared to influence
the optimum designed channel. It is observed that in this channel, these the work reported in this paper.
parameters have more uniform and lower values than the corresponding
values in the validated channel. References

[1] J.P. Holman, Heat Transfer, McGraw-hill, 2010.

Conclusions [2] T.L. Bergman, et al., Introduction to Heat Transfer, John Wiley & Sons, 2011.
[3] S.V. Patankar, C.H. Liu, E.M. Sparrow, Fully developed flow and heat transfer in
The genetic algorithm as a powerful multi-objective optimization ducts having streamwise-periodic variations of cross-sectional area, J Heat Transfer
(1977) 180–186.
approach was utilized to optimize heat transfer rate and pressure drop of [4] Z. Guo, N.K. Anand, Three-dimensional heat transfer in a channel with a baffle in
passing airflow in a double baffle channel. The image processing method the entrance region, Numer. Heat Transfer, Part A Appl. 31 (1) (1997) 21–35.
was used to detect baffles edges and generate mesh through this channel. [5] P. Sriromreun, C. Thianpong, P. Promvonge, Experimental and numerical study on
heat transfer enhancement in a channel with Z-shaped baffles, Int. Commun. Heat
Low Reynolds Chien’s model was applied to simulate this turbulent
Mass Transfer 39 (7) (2012) 945–952.
flow. Continuity, momentum, and k-ε equations were discretized using [6] L.C. Demartini, H.A. Vielmo, S.V. Möller, Numeric and experimental analysis of the
the Power Law scheme in the finite volume method, and the staggered turbulent flow through a channel with baffle plates, J. Braz. Soc. Mech. Sci. Eng. 26
grid was utilized to prevent the checker-board problem. The prepared (2) (2004) 153–159.
[7] M. Mohammadi Pirouz, M. Farhadi, K. Sedighi, H. Nemati, E. Fattahi, Lattice
homemade code was validated by previous published numerical and Boltzmann simulation of conjugate heat transfer in a rectangular channel with
experimental results approving the accuracy of the predictions of this wall-mounted obstacles, Sci. Iran. 18 (2) (2011) 213–221.

6
B. Abolpour et al. Thermal Science and Engineering Progress 26 (2021) 101132

[8] Mojtaba Mokhtari, et al., Numerical study of mixed convection heat transfer of [36] N. Ali, K. Behdinan, Optimal geometrical design of aircraft using genetic
various fin arrangements in a horizontal channel, Eng. Sci. Technol., Int. J. 20 (3) algorithms, Trans. Can. Soc. Mech. Eng. 26 (4) (2003) 373–388.
(2017) 1106–1114. [37] F. Muyl, L. Dumas, V. Herbert, Hybrid method for aerodynamic shape optimization
[9] Nasiruddin, M.H.K. Siddiqui, Heat transfer augmentation in a heat exchanger tube in automotive industry, Comput. Fluids 33 (5-6) (2004) 849–858.
using a baffle, Int. J. Heat Fluid Flow 28 (2) (2007) 318–328. [38] B.A. Younis, V.P. Przulj, Computation of turbulent vortex shedding, Comput. Mech.
[10] M. Mellal, R. Benzeguir, D. Sahel, H. Ameur, Hydro-thermal shell-side performance 37 (5) (2006) 408–425.
evaluation of a shell and tube heat exchanger under different baffle arrangement [39] V.C. Patel, W. Rodi, G. Scheuerer, Turbulence models for near-wall and low
and orientation, Int. J. Therm. Sci. 121 (2017) 138–149. Reynolds number flows-a review, AIAA J. 23 (9) (1985) 1308–1319.
[11] P. Dutta, A. Hossain, Internal cooling augmentation in rectangular channel using [40] K.-Y. Chien, Predictions of channel and boundary-layer flows with a low-Reynolds-
two inclined baffles, Int. J. Heat Fluid Flow 26 (2) (2005) 223–232. number turbulence model, AIAA J. 20 (1) (1982) 33–38.
[12] N. Guerroudj, H. Kahalerras, Mixed convection in a channel provided with heated [41] B.E. Launder, D.B. Spalding, The numerical computation of turbulent flows, in:
porous blocks of various shapes, Energy Convers. Manag. 51 (3) (2010) 505–517. Numerical Prediction of Flow, Heat Transfer, Turbulence and Combustion,
[13] C. Berner, F. Durst, DaM. McEligot, Flow around baffles, J. Heat Transfer (1984) Pergamon, 1983, pp. 96–116.
743–749. [42] R. Franke, W. Rodi, Calculation of vortex shedding past a square cylinder with
[14] M.A. Habib, et al., Experimental investigation of heat transfer and flow over baffles various turbulence models, in: Turbulent Shear Flows 8, Springer, Berlin,
of different heights, J. Heat Transfer (1994) 363–368. Heidelberg, 1993, pp. 189–204.
[15] H. Li, V. Kottke, Effect of baffle spacing on pressure drop and local heat transfer in [43] M. Kato, The modelling of turbulent flow around stationary and vibrating square
shell-and-tube heat exchangers for staggered tube arrangement, Int. J. Heat Mass cylinders, Turbulent Shear Flow 1 (1993) 10–14.
Transfer 41 (10) (1998) 1303–1311. [44] Brian Edward Launder, Neil D. Sandham (Eds.), Closure Strategies for Turbulent
[16] I. De Falco, V.P. Castellino, An Introduction to Evolutionary Algorithms and their and Transitional Flows, Cambridge University Press, 2002.
Application to the Aerofoil Design Problem–Part II: The Results, Institute for [45] B. Abolpour, et al., Study of the motion and deposition of micro particles in a
Research on Parallel Information Systems (IRSIP), National Research Council of vertical tube containing uniform gas flow, Heat Mass Transfer 53 (12) (2017)
Italy, 1997. 3517–3528.
[17] R. Mäkinen, et al., A genetic algorithm for multiobjective design optimization in [46] E. Aslan, I. Taymaz, Y. Islamoglu, Finite volume simulation for convective heat
aerodynamics and electromagnetics, Comput. Fluid Dyn. 98 (1998) 418–422. transfer in wavy channels, Heat Mass Transfer 52 (3) (2016) 483–497.
[18] S.-Y. Han, J.-S. Maeng, Shape optimization of cut-off in a multi-blade fan/scroll [47] S.V. Patankar, D.B. Spalding, Heat and Mass Transfer in Boundary Layers, Morgan-
system using neural network, Int. J. Heat Mass Transfer 46 (15) (2003) 2833–2839. Grampian, 1968.
[19] J.I. Madsen, W. Shyy, R.T. Haftka, Response surface techniques for diffuser shape [48] Henk Kaarle Versteeg, Weeratunge Malalasekera, An Introduction to
optimization, AIAA J. 38 (9) (2000) 1512–1518. Computational Fluid Dynamics: The Finite Volume Method, Pearson education,
[20] G. Fabbri, Heat transfer optimization in corrugated wall channels, Int. J. Heat Mass 2007.
Transfer 43 (23) (2000) 4299–4310. [49] Fluent, A. N. S. Y. S. ANSYS fluent theory guide 15.0. ANSYS, Canonsburg, PA 33,
[21] S. Tiwari, D. Maurya, G. Biswas, V. Eswaran, Heat transfer enhancement in cross- 2013.
flow heat exchangers using oval tubes and multiple delta winglets, Int. J. Heat [50] C.-L. Hwang, Abu Syed Md Masud, Multiple Objective Decision Making—Methods
Mass Transfer 46 (15) (2003) 2841–2856. and Applications: A State-of-the-art Survey, Springer Science & Business Media,
[22] R.S. Matos, J.V.C. Vargas, T.A. Laursen, A. Bejan, Optimally staggered finned 2012.
circular and elliptic tubes in forced convection, Int. J. Heat Mass Transfer 47 (6-7) [51] Kalyanmoy Deb, Multi-objective optimisation using evolutionary algorithms: an
(2004) 1347–1359. introduction, in: Multi-objective Evolutionary Optimisation for Product Design and
[23] F. Kock, H. Herwig, Entropy production calculation for turbulent shear flows and Manufacturing, Springer, London, 2011, pp. 3–34.
their implementation in CFD codes, Int. J. Heat Fluid Flow 26 (4) (2005) 672–680. [52] Nidamarthi Srinivas, Kalyanmoy Deb, Muiltiobjective optimization using
[24] K.-S. Lee, W.-S. Kim, J.-M. Si, Optimal shape and arrangement of staggered pins in nondominated sorting in genetic algorithms, Evol. Comput. 2 (3) (1994) 221–248.
the channel of a plate heat exchanger, Int. J. Heat Mass Transfer 44 (17) (2001) [53] Kaisa Miettinen, Jussi Hakanen, Dmitry Podkopaev, Interactive nonlinear
3223–3231. multiobjective optimization methods, in: Multiple criteria decision analysis,
[25] G. Fabbri, Heat transfer optimization in internally finned tubes under laminar flow Springer, New York, NY., 2016, pp. 927–976.
conditions, Int. J. Heat Mass Transfer 41 (10) (1998) 1243–1253. [54] Hamid Reza Hassanzadeh, Modjtaba Rouhani, A multi-objective gravitational
[26] D. Balagangadhar, S. Roy, Design sensitivity analysis and optimization of steady search algorithm. 2010 2nd International Conference on Computational
fluid-thermal systems, Comput. Methods Appl. Mech. Eng. 190 (42) (2001) Intelligence, Communication Systems and Networks, IEEE, 2010.
5465–5479. [55] Younes Menni, Ahmed Azzi, Numerical analysis of thermal and aerodynamic fields
[27] J. Bonjour, L.A.O. Rocha, A. Bejan, F. Meunier, Dendritic fins optimization for a in a channel with cascaded baffles, Periodica Polytech. Mech. Eng. 62 (1) (2018)
coaxial two-stream heat exchanger, Int. J. Heat Mass Transfer 47 (1) (2004) 16–25.
111–124. [56] Gerhard Bosch, Wolfgang Rodi, Simulation of vortex shedding past a square
[28] G. Fabbri, Effect of viscous dissipation on the optimization of the heat transfer in cylinder near a wall, Int. J. Heat Fluid Flow 17 (3) (1996) 267–275.
internally finned tubes, Int. J. Heat Mass Transfer 47 (14-16) (2004) 3003–3015. [57] Xing Cao, et al., Performance investigation and multi-objective optimization of
[29] J.A. Visser, D.J. de Kock, Optimization of heat sink mass using the DYNAMIC-Q helical baffle heat exchangers based on thermodynamic and economic analyses,
numerical optimization method, Commun. Numer. Methods Eng. 18 (10) (2002) Int. J. Heat Mass Transfer 176 (2021), 121489.
721–727. [58] de Andrés Honrubia, José Luis, et al., Development and application of a multi-
[30] Tatsuya Okabe, et al., Comparative studies on micro heat exchanger optimisation. objective tool for thermal design of heat exchangers using neural networks,
The 2003 Congress on Evolutionary Computation, 2003. CEC’03, IEEE, 2003. Mathematics 9 (10) (2021) 1120.
[31] David E. Goldberg, John Henry Holland, Genetic algorithms and machine learning, [59] Ibrahim Kaya, Yasin Ust, A new method to multi-objective optimization of shell
Sci. Programm. (1988). and tube heat exchanger for waste heat recovery, in: Energy Sources, Part A:
[32] E. Zitzler, L. Thiele, Multiobjective optimization using evolutionary algorithms—a Recovery, Utilization, and Environmental Effects, 2021, pp. 1–18.
comparative case study. International Conference on Parallel Problem Solving from [60] Kizhakke Kodakkattu Saijal, Thondiyil Danish, Design optimization of a shell and
Nature, Springer, Berlin, Heidelberg, 1998. tube heat exchanger with staggered baffles using neural network and genetic
[33] J. Lee, P. Hajela, Parallel genetic algorithm implementation in multidisciplinary algorithm, Proc. Inst. Mech. Eng., Part C: J. Mech. Eng. Sci. (2021),
rotor blade design, J. Aircraft 33 (5) (1996) 962–969. 09544062211005797.
[34] R.AE. Mäkinen, J. Périaux, J. Toivanen, Multidisciplinary shape optimization in [61] Jian Chen, et al., Experimental investigation of shell-side performance and optimal
aerodynamics and electromagnetics using genetic algorithms, Int. J. Numer. design of shell-and-tube heat exchanger with different flower baffles, Heat Transfer
Methods Fluids 30 (2) (1999) 149–159. Eng. 42 (7) (2021) 613–626.
[35] S. Obayashi, T. Tsukahara, T. Nakamura, Multiobjective genetic algorithm applied [62] Danish Thondiyil, Saijal Kizhakke Kodakkattu, Optimization of a shell and tube
to aerodynamic design of cascade airfoils, IEEE Trans. Ind. Electr. 47 (1) (2000) heat exchanger with staggered baffles using Taguchi method, Mater. Today: Proc.
211–216. (2021).