Sunteți pe pagina 1din 9

Applied Thermal Engineering 31 (2011) 1839e1847

Contents lists available at ScienceDirect

Applied Thermal Engineering


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

Energy and cost optimization of a plate and fin heat exchanger using genetic
algorithm
Hamidreza Najafi a, *,1, Behzad Najafi b, Pooya Hoseinpoori c
a
Department of Mechanical Engineering, The University of Alabama, Tuscaloosa, AL 35487, USA
b
Department of Energy Engineering, Politecnico di Milano, Piacenza, Italy
c
Department of Mechanical Engineering, K.N. Toosi University of Technology, Tehran, Iran

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

Article history: In the present paper, a plate and fin heat exchanger is considered and air, as an ideal gas, is defined in
Received 17 October 2010 both sides of the heat exchanger as the working fluid. Several geometric variables within the logical
Accepted 17 February 2011 constraints are considered as optimization parameters. Two different objective functions including the
Available online 1 March 2011
total rate of heat transfer and the total annual cost of the system are defined. Since mentioned objectives
are conflicting, no single solution can well-satisfy both objective functions simultaneously. In other
Keywords:
words, any attempt to increase the value of the total rate of heat transfer leads to the higher total cost of
Plate and fin heat exchanger
the system which is certainly undesirable. Therefore, multi-objective optimization using genetic algo-
Optimal design
Genetic algorithm
rithm is utilized in order to achieve a set of optimal solutions, each of which is a trade-off between
Sensitivity analysis objectives and can satisfy both objective functions in an appropriate level. The main advantage of this
work is providing a set of optimal solutions each of which can be selected by the designer based on the
project’s limits and the available investment. A sensitivity analysis is also presented in order to inves-
tigate the effect of some geometric parameters on each objective functions.
 2011 Elsevier Ltd. All rights reserved.

1. Introduction fluids. The hot and cold streams flow through the alternating layers
of the heat exchanger and the heat is transferred mainly from the
Heat exchangers are known as one of the most essential equip- surfaces of the plates between the two streams. The main role of fins
ments in almost every industrial plant. Various types of heat is increasing the structural integrity of the heat exchanger. However,
exchangers have been designed and utilized for different purposes. they also provide the secondary heat transfer surface which raises
Plate and fin heat exchangers (PFHE) are widely used in aero- the total heat transfer area and as a result, the rate of heat transfer
space,automobile, cryogenic and chemical process plants due to between two streams. A schematic of a PFHE is shown in Fig. 1.
their low weight, compactness and high effectiveness They weigh Since the main task of a heat exchanger, regardless of its type, is
95% less than comparable conventional shell and tube exchangers transferring heat between two fluid streams, any attempt to increase
and provide 300e400 square feet of heat transfer per cubic foot of the exchanged heat, would be desirable. Optimization of heat
exchanger volume (1000e1500 m2/m3) [1,2]. Although the exchangers owing to their vital role in various industries has
construction cost of PFHEs due to the higher required details for attracted lots of interests all over the world. Several researches have
manufacturing process is remarkably more than the conventional been performed in this area for various types of heat exchangers,
shell and tube heat exchangers, the rate of heat transfer which they considering different objective functions and design parameters.
can provide can reasonably justify using this type of heat exchangers. Since PFHEs design deals with several parameters and nonlinear
PFHEs, categorized as compact heat exchangers, consist of series equations, optimization of these systems faces with rather high
of flat plates, called parting sheets and layers of corrugated fins complexity. Therefore, using conventional optimization methods is
which create series of finned chambers to transferheatbetween not appropriate for this group of problems. Genetic algorithm (GA) is
an optimization method, inspired from natural selection which has
been successfully utilized when the problem deals with numerous
* Corresponding author. variables within different constraints. In terms of thermal systems,
E-mail addresses: hamidreza.najafi@ua.edu (H. Najafi), behzad.najafi@mail. GA has been used in several cases and has been yielded impressive
polimi.it (B. Najafi), p.hoseinpoori@gmail.com (P. Hoseinpoori).
1 results in numerous researches, namely cooling channels [3], fin
Present address: No 63, 6th Mokhaberat Ave., Shahid Mokhberi St., Jannat Abad
St., 1475745481 Tehran, Iran. profiles [4], compact high performance coolers [5], shell and tube

1359-4311/$ e see front matter  2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2011.02.031
1840 H. Najafi et al. / Applied Thermal Engineering 31 (2011) 1839e1847

Nomenclature OMC operating and maintenance cost, $


P pressure, N m-2
A heat transfer area, m2 DP pressure drop, N m-2
Aff free flow area, m2 Q rate of heat transfer, W
C heat capacity rate, W K-1 Re Reynolds number
Cp specific heat, W kg-1 K-1 s fin spacing, m
Cr Cmin/Cmax t fin thickness, m
Dh hydraulic diameter, m tp operating period, year
E pumping power, W T temperature, K
ec rate of increase in energy cost, % TC total annual cost, $
f fanning friction factor U overall heat transfer coefficient, W m-2 K-1
fc fixed cost, $ uc unit cost of PFHE per area, $ m2
fe electric cost, $ kW1 h1
h heat transfer coefficient, W m-2 K-1 Greek symbols
H height of the fin, m e effectiveness
IC capital cost, $ r density, kg m-3
i interest rate, % m viscosity, Pa s
j Colburn factor hp pump efficiency
l lance length of the fin, m
L heat exchanger length, m Subscripts
m mass flow rate of fluid, kg s-1 c, h cold and hot fluids
n fin frequency, fins per meter i inlet
N number of years max maximum
Nh, Nc number of fin layers for hot and cold sides min minimum
NTU number of transfer units T total

heat exchangers [6]. GA is also used for the optimization of plate and optimized global performance by selecting finned or smooth
fin heat exchangers. Optimal design of compact recuperators for parallel plate type cross flow heat exchanger in an aircraft.
micro-turbine application is presented by [7]. Peng and Ling [8] In most of the engineering problems, one may face with more
successfully used genetic algorithm combined with back propaga- than one objective function. When these objectives are conflicting,
tion neural network for the optimal design of PFHE. They considered conventional optimization methods can not lead to global optima
the minimum total weight and total annual cost for a given con- due to the fact that optimizing with respect to a single objective
strained condition as objective functions. Mishra et al. [1] utilized leads to inappropriate results with respect to other objectives.
a genetic algorithm based optimization for cross flow plate-fin heat Multi-Objective optimization has been successfully used for opti-
exchangers in order to minimize number of entropy generation units mizing many different thermodynamic systems. Gholap and Khan
for a specified heat duty under given space restrictions. Seculic [9] [12] used multi-objective optimization in order to find the optimal
studied optimization of a cross flow compact heat exchanger using design parameters of air forced heat exchangers, considering
numerical methods in order to meet the least enthalpy exchange energy consumption and material cost as two conflicting objective
irreversibility. Bejan [11] investigated design of a gas to gas counter functions. Hilbert et al. [13] considered value of heat exchange and
flow heat exchanger with minimum and found minimum irrevers- pressure loss as objectives and performed multi-objective optimi-
ibility and design of a regenerative heat exchanger for minimum zation to find the optimum geometry which can satisfy both
heat transfer area with fixed irreversibility. Vargas and Bejan [10] objectives in an acceptable level. Najafi et al. [14] successfully
utilized multi-objective optimization using genetic algorithm in
order to achieve optimal design parameters for plate and frame
heat exchangers.In the present paper, a plate and fin heat
exchanger is modeled. Several variables including the total length
of the PFHE in both hot and cold side, fin height, fin frequency, lance
length of the fin, fin thickness and the number of fin layers are
considered as optimization parameters within reasonable
constraints. The total rate of heat transfer and the total annual cost
of the PFHE system are considered as two conflicting objective
functions. Multi-objective optimization using genetic algorithm is
utilized in order to achieve set of optimal solutions, called Pareto
solutions, each of which is a trade-off between objective functions.
The user can choose any of the obtained solutions depending on the
project limits and available investment. In other words, for any
specific required rate of heat transfer, the designer can select the
optimal values for the geometric parameters which lead to the least
possible total annual cost. In view of the fact that several geometric
variables have been considered and also a wide possible range has
also been taken into account for each parameter, the gener-
ated result have sufficient generality to be used for different
Fig. 1. A schematic of a plate and fin heat exchanger (PFHE). configurations.
H. Najafi et al. / Applied Thermal Engineering 31 (2011) 1839e1847 1841

The total heat transfer area is the sum of the heat transfer areas
at both sides:

AT ¼ Ah þ Ac (5)
The rate of the heat transfer can be determined as:
 
Q ¼ eCmin Th;i  Tc;i (6)

Where the effectiveness (e) can be found as [15]:


  n h i o
1
e ¼ 1  exp NTU 0:22 exp  Cr:NTU 0:78  1 (7)
Cr

Where Cr ¼ Cmin =Cmax and the value of NTU calculated as follows:


 
1 C 1 1
¼ min ¼ þ (8)
NTU UA ðhAÞh ðhAÞc
By substituting h, the heat transfer coefficient, the above
equation can be expressed as:
" #
1 1 Aff ;h 1 Aff ;c
¼ Cmin 2=3
 þ 2=3

NTU jh Cph Prh mh Ah jc Cpc Prc mc Ac
(9)
Where j is the thermal performance of the surface of the PFHE and
the empirical correlation for this term can be given as [16]:

j ¼ 0:53ðReÞ0:5 ð1=Dh Þ0:15 fs=H  tg0:14 ðfor Re  1500Þ


Fig. 2. The detailed scheme of a PFHE and the fins geometry.
(10)

j ¼ 0:21ðReÞ0:4 ð1=Dh Þ0:24 ðt=Dh Þ0:02 ðfor Re > 1500Þ (11)

In addition, a sensitivity analysis is also carried out in order to Where Dh, the hydraulic diameter and the Reynolds number can be
investigate the effect of some of the geometric variables on the calculated as follow:
objective functions.
2ðs  tÞðH  tÞ
Dh ¼ (12)
ðH  tÞt
2. Modeling of the system fs þ ðH  tÞg þ
l

In this section, the equations for calculating the total rate of heat GDh mDh
transfer, pressure drop and the total annual cost of the system are Re ¼ ¼ (13)
m Aff m
presented.
s ¼ ð1=n  tÞ (14)
2.1. Heat transfer equations

The eNTU method is utilized in order to determine the value of


2.2. Pressure drop calculations
the total rate of heat transfer for a cross flow PFHE. It should be
noted that in this work, the heat exchanger is working under steady
The friction performance of the surface of the PFHE can be found
state condition and both fluids are assumed to be air in the ideal gas
by the following equations for the laminar and turbulent flow [16]:
condition.
In Fig. 2, the geometry of the PFHE and fins are shown in detail.
f ¼ 8:12ðReÞ0:74 ð1=Dh Þ0:41 fs=H  tg0:02 ðfor Re  1500Þ
Heat transfer areas for the hot side and the cold side can be found as
below [1]: (15)

Ah ¼ Lh Lc Nh ½1 þ 2nh ðHh  th Þ (1) f ¼ 1:12ðReÞ0:36 ð1=Dh Þ0:65 ðt=Dh Þ0:17 ðfor Re > 1500Þ (16)

Ac ¼ Lc Lh Nc ½1 þ 2nc ðHc  tc Þ (2)


Therefore, the frictional pressure drop for the hot and cold
And the flow free areas can be calculated as: streams can be found as:

Aff ;h ¼ ðHh  th Þð1  nh th ÞLc Nh (3)


4fh Lh G2h 2fh m2h Lh
DPh ¼ ¼ 
2rh Dh;h rc Dh;h L2c N 2 ðHh  th Þ2 ð1  nh th Þ2
h
Aff ;c ¼ ðHc  tc Þð1  nc tc ÞLh Nc (4)
(17)
1842 H. Najafi et al. / Applied Thermal Engineering 31 (2011) 1839e1847

effort to improve Pareto solution with respect to any objective will


4fc Lc G2c 2fc m2c Lc lead to a decrease in the value ofat least one other objective. The
DPc ¼ ¼  (18)
2rc Dh;c rc Dh;c L2 Nc2 ðHc  tc Þ2 ð1  nc tc Þ2 corresponding objective function values for a given Pareto optimal
h
set are called the Pareto front.

2.3. Cost estimation


3.2. Genetic algorithm
The total annual cost is the sum of the capital cost, operational
cost and the maintenance cost of the system during a year. The cost Evolutionary algorithms, specifically genetic algorithm, owing
of the PFHE system can be found by using the following equations to their desirable characteristics are an appropriate method to deal
[8] [17e19],: with multi-objective optimization problems [14].
Genetic algorithm (GA) is a stochastic numerical search method,
inspired by evolutionary processes, which was first conceived by
IC ¼ fc þ A  uc  C1 (19)
Holland [21]. A comprehensive discussion about GA can be found in
[21] and [22].
1þi In GA terminology, a solution vector x˛X is called an indi-
C1 ¼ (20) vidual or a chromosome. GA operates with a collection of chro-
tp
mosomes called population. Chromosomes are made of discrete
units called genes. Each gene controls one or more features of the
ðEc þ Eh ÞAH  fe chromosome [20].
OMC ¼ (21)
3600  1000 The initial population is usually randomly generated. After-
ward, parents are selected based on their fitness values. The
Where E is the pumping power and can be calculated as: higher fitness of an individual, leads to higher possibility of being
selected as parent for the reproduction. In the next step, repro-
DPm duction, two chromosomes (parents), combine together and form
E ¼ (22)
rhp new chromosomes, called offspring. Since individuals with higher
fitness have more chance for being selected and produce
offspring, the new population, generated after reproduction,
Therefore, the total annual cost can be found by:
posses more qualified genes and consequently higher fitness.
Therefore, using crossover operator iteratively leads to conver-
TC ¼ IC þ OMC (23)
gence to a general good solution. The mutation operator, which is
generally applied in the genes level, causes random changes in
The assumed values for the fixed parameters including uc,fc,AH, characteristics of chromosomes. Since the mutation rate is small
fe, i, ec and tp are given in Table 1 [8]. and depends on length of the chromosome, the new chromosome
produced by mutation will not be much different from the earlier
3. Optimization one. Hence, crossover leads the population to converge by making
the chromosomes in the population alike. Mutation reintroduces
3.1. Multi-objective optimization genetic diversity back into the population and assists the search to
escape from local optima [20]. In multi-objective genetic algo-
There are indeed rare problems in the real world which only rithm (MOGA), the procedure is the same except that there are
deal with a single objective. When these objectives are conflicting more than one fitness for each individual regarding different
no single solution can fully satisfy all objective functions simul- considered objectives.
taneously. In these cases, a trade-off between two or more In the present work, multi-objective genetic algorithm is
objective functions is needed to make the optimal decision. utilized in order to obtain optimal geometric parameters of a PFHE
Hence, a reasonable solution to a multi-objective problem is to system which lead to the minimum total annual cost and the
investigate a set of solutions, each of which satisfies the objectives maximum total rate of heat transfer. Since the value of the total
at an acceptable level without being dominated by any other heat transfer rate must be maximized, e Q is considered as one of
solution [20]. the objective functions and is minimized through the optimization
If all objective functions are for minimization, a feasible solution procedure which in turn leads to maximization of the total rate of
x is said to dominate another feasible solution y (x > y), if and only heat transfer, Q. Several geometric variables within logical
if Zi (x)  Zi (y), for i ¼ 1,.,K and Zj (x) < zj (y) for least one constraints are considered as optimization parameters which are
objective function j. A solution is said to be Pareto optimal if it is not given in Table 2. In each iteration, the values of optimization
dominated by any other solution in the solution space [20]. Any parameters (including the total length of the PFHE in both hot and

Table 1 Table 2
Constant values for cost evaluation of the system. Optimization parameters and constant values for the case study.

Parameter Value Parameter Constant values for the case study Specified range
AH Annual operating period (second/year) 21,600,000 Lh (m) 0.5 0.1e1.5
ec, rate of increase of energy cost (%) 30 Lc (m) 0.35 0.1e1.1
fc, fixed cost ($) 187.5 H (mm) 9.5 2e20
fe, electric cost ($/kW h) 0.065 n (fins/m) 535 200e1000
i, interest rate (%) 40 t (mm) 0.146 0.1e0.3
tp, total operating period (year) 5 l (mm) 6 3e10
uc, unit cost of heat exchanger per area ($/m2) 25 Nh 8 2e40
H. Najafi et al. / Applied Thermal Engineering 31 (2011) 1839e1847 1843

Table 3 corresponding upper and lower limits for each parameter are
Operating conditions for the case study. presented in Table 2.
Parameters Hot fluid Cold fluid In this section, a sensitivity analysis regarding the geometric
Mass flow rate, m (kg.s1) 0.9 1 variables is demonstrated and discussed. Afterward, multi-objec-
Inlet temperature, Ti (k) 473 303 tive optimization via genetic algorithm is carried out and the cor-
Density, r (kg.m3) 0.746 1.166 responding results are presented.
Specific heat, Cp (J.kg1.k1) 1026 1005
Viscosity,m (N.s.m2) 2.623  105 1.8845  105
Prandtl number, Pr 0.685 0.712

4.1. Sensitivity analysis

In this section, considering the operating condition and constant


cold sides, fin height, fin frequency, lance length of the fin, fin values given in Table 3, the effect of some geometric variables on
thickness and the number of fin layers) are given new values within objective functions is investigated. In each section, the values of all
their specified constraints and the considered objective functions parameters, except the one which is selected for the investigation,
are evaluated based on the new values. The initial population size, are kept constant. By varying the value of the selected parameter,
the number of individuals in each generation, is considered to the sensitivity of each objective function with respect to that
be110. The crossover fraction which specifies the fraction of the parameter can be discussed.
next generation, other than elite children that are produced by
crossover is selected to be 0.8 and the possibility of the mutation, 4.1.1. Effect of the fin height
which is the probability of each vector entry in the individual to be The effect of the fin height on the total rate of heat transfer (Q)
mutated is chosen to be 0.01. The optimization is terminated after and the total annual cost are presented in Fig. 3 and Fig. 4 respec-
140 generation when the average change in the spread of Pareto tively. Increasing the fin height results in an increment in both free
solutions was less than the considered termination tolerance on the flow area and the total heat transfer area which nonlinearly affects
fraction value which is 106. the value of the NTU. As can be seen in Fig. 3, when the value of H
Considering the given operating conditions in Table 3, some of varies from 2 mm to 20 mm, the value of Q increases almost 8 kW
the selected optimal design parameters obtained by utilizing GA are which is a remarkable amount.
given in Table 4. From another point of view, increment in H leads to the
reduction of pressure drops which in turn decrease the values of
the pumping power and the operational cost. However, as already
4. Results and discussions mentioned, any raise in Hresults in increasing the total heat
transfer area which leads to increment in the initial cost of the
The operating conditions for the considered case are given in PFHE. Therefore, the fin height has two conflicting effects on the
Table 3. Both fluids are considered as air (with ideal gas assump- total annual cost of the system. However, as can be seen in Fig. 4,
tion) and the PFHE is mainly made of Aluminum. This is noteworthy the variation of the total annual cost regarding the fin height is
that the change of the thermal properties of air regarding the descending which shows that by increasing the fin height, the
temperature is taken into account. The list of the geometric vari- reduction in the operational cost is more than the raise in the
ables which are considered as optimization parameters and the capital cost.

Table 4
Some of the selected optimal results.

Lh (m) Lc (m) H (m) n t (m) l (m) Nh Q (W) TAC ($)


0.109038 0.110786 0.018724 302.4028 0.000116 0.009951 36 61987.3 611.3285
0.109065 0.110863 0.018756 302.4029 0.000155 0.01 36 62625.5 626.3794
0.103668 0.138988 0.016896 311.5889 0.000276 0.00997 36 66020.9 746.5186
0.131764 0.114721 0.018862 347.674 0.000208 0.007498 35 79434.5 828.8629
0.115484 0.175607 0.018524 322.2846 0.000198 0.009901 35 95665.2 841.0823
0.141702 0.247614 0.019048 331.9817 0.000202 0.009359 36 108786 1219.689
0.134954 0.280309 0.017232 331.124 0.000267 0.007873 35 109263 1324.175
0.140199 0.194417 0.018849 445.3507 0.000235 0.006359 33 117152 1371.012
0.184437 0.21291 0.019091 449.2851 0.000209 0.009199 35 122427 1668.722
0.193571 0.251092 0.019719 484.5435 0.000202 0.008484 34 126316 2110.709
0.234117 0.206783 0.018539 604.8335 0.000284 0.005177 33 132922 2807.374
0.113355 0.451576 0.017357 897.8511 0.000203 0.004985 33 135099 5539.146
0.36289 0.199444 0.018425 677.9377 0.000205 0.0076 33 137316 4078.445
0.382856 0.670864 0.018524 453.4827 0.000295 0.007755 34 138259 8338.536
0.59678 0.218212 0.019041 651.476 0.000221 0.008256 34 141679 6855.37
0.274351 0.524699 0.019394 844.3528 0.000245 0.004906 34 143553 9698.804
0.84112 0.245421 0.018866 652.9682 0.000235 0.005667 33 145077 10451.97
0.911661 0.287125 0.019178 627.7016 0.000272 0.004226 31 146064 11936.24
0.768093 0.66683 0.018122 694.2699 0.000283 0.006344 30 148065 21199.46
1.165348 0.440329 0.019507 900.537 0.00025 0.006211 32 151499 31800.76
1.294303 0.586106 0.019617 920.7268 0.000286 0.004069 29 152999 45530.96
1.288773 0.668546 0.019556 918.1537 0.000293 0.003656 29 153262 50495.11
1.310729 0.758679 0.019523 922.2589 0.000292 0.004011 31 153518 59228.45
1.345425 0.97389 0.019932 929.4124 0.000297 0.003753 29 154037 77220.33
1844 H. Najafi et al. / Applied Thermal Engineering 31 (2011) 1839e1847

Fig. 3. Effect of the fin height on the total rate of heat transfer.

4.1.2. Effect of the fin frequency given equations can be found out, any increment in the number of
Fig. 5 shows the variation of the total rate of heat transfer versus fin layers leads to an increase in both the free flow area and the total
the fin frequency. The increment in the fin frequency causes heat transfer area. Where they have a conflicting effect on the value
a reduction in the free flow area and raise in the total heat transfer of NTU and consequently Q. As it is shown in Fig. 7, increasing the
area which in turn increases the value of NTU. As it shows in Fig. 5, number of fin layers resulted in raising the total rate of heat
when the value of n raises from 200 to 1000, Q is increased more transfer. There is a jump in the diagram, where Nh ¼ 20 which is
than 35 kW. owing to the nonlinear relationship between the areas, NTU and
Increasing the number of fins per meter results in increasing the the total rate of heat transfer. As the Nh ¼ 20 in changing from
pressure drop and consequently the operational cost. By raising the 2e40, the total rate of heat transfer increased about 27 kW which
total heat transfer area, the initial cost is also increased. These effects shows this parameter can highly affect the value of Q.
resulted in increment in the total annual cost, as it shows in Fig. 6. When the value of Nh is increased, the values of pressure
drops reduced which in turn leads to less operational cost. In the
4.1.3. Effect of the number of fin layers other side, the total heat transfer area is also raised which in turn
It should be noted that the relation between the number of fin cause increment in the value of the initial cost. As can be seen in
layers for the cold side and the hot side is Nc ¼ Nhþ1. From the Fig. 8, the total annual cost is constantly decreased since the

Fig. 4. Effect of the fin height on the total annual cost.


H. Najafi et al. / Applied Thermal Engineering 31 (2011) 1839e1847 1845

Fig. 5. Effect of the fin frequency on the total rate of heat transfer.

effect of Nh on the initial cost is dominated by its effect on the


operational cost.
Fig. 7. Effect of the number of fin layers on the total rate of heat transfer.

4.2. Optimization results

Considering the optimization parameters presented in Table 2, Pareto front can give the least possible total annual cost for any
and the operating conditions which are given in Table 3, the multi- specific required total rate of heat transfer. As can be seen, the value of
objective genetic algorithm is utilized in order to achieve design the total heat transfer Q varies from 0.6  105 to 1.54  105 (W) where
parameters which lead to the highest possible total rate of heat the range of the total annual cost is from 500 to 80000 $.
transfer and the least total annual cost. The result is a Pareto front Some of the selected optimal values of the design parameters
which is a set of optimal solutions each of which is a trade-off and the corresponding values of the objective functions are pre-
between objective functions. The generated Pareto front is shown in sented in Table 4 to provide a nice output scheme of the results.
Fig. 9. The designer can select each of the optimal solutions regarding These data are selected to cover the generated range of the total
the project limits and the available investment. In other words, for rate of heat transfer and the total annual cost of the PFHE in the
any given mass flow rates and inlet temperatures, the generated Pareto front as depicted in Fig. 9.

Fig. 6. Effect of the fin frequency on the total annual cost.


1846 H. Najafi et al. / Applied Thermal Engineering 31 (2011) 1839e1847

Fig. 8. Effect of the number of fin layers on the total annual cost.

Fig. 9. Pareto front.

5. Conclusion geometric variables on each objective function. Since increasing


the total rate of heat transfer necessitates raising the heat
In the present work, a plate and fin heat exchanger (PFHE) is transfer area and any increment in the area of the system,
optimized considering two objective functions including the results in a more expensive design, the considered objectives are
total rate of heat transfer and the total annual cost of the system. conflicting and no single solution can satisfy both objectives
Several geometric variables including the total length of the hot simultaneously. The multi-objective genetic algorithm is utilized
and cold side of the heat exchanger, fin height, fin frequency, for optimization of the system and achieving set of optimal
lance length of the fin, fin thickness and the number of fin layers solutions each of which is a trade-off between the highest total
are considered as optimization parameters. A sensitivity analysis rate of heat transfer and the least total annual cost. The principal
is carried out in order to investigate the effect of some advantage of this work is providing a wide range of optimal
H. Najafi et al. / Applied Thermal Engineering 31 (2011) 1839e1847 1847

solutions which allows the user to choose the best design [10] A. Bejan, The concept of irreversibility in heat exchanger design: counter-flow
heat exchangers for gas-to-gas applications, ASME Journal of Heat Transfer 99
parameters regarding the application and the total annual cost
(1977) 374e380.
of the system. [11] J.V.C. Vargas, A. Bejan, Thermodynamic optimization of finned cross‐flow heat
exchangers for aircraft environmental control systems, International Journal
References of Heat and Fluid Flow, Volume 22 (Issue 6) (December 2001) 657e665.
[12] A.K. Gholap, J.A. Khan, Design and multi-objective optimization of heat
exchangers for refrigerators, Applied Energy Volume 84 (Issue 12) (December
[1] M. Mishra, P.K. Das, S. Sarangi, Second law based optimisation of crossflow 2007) 1226e1239.
plate-fin heat exchanger design using genetic algorithm, Applied Thermal [13] R. Hilbert, G. Janiga, R. Baron, D. Thévenin, Multi-objective shape optimization
Engineering 29 (2009) 2983e2989. of a heat exchanger using parallel genetic algorithms, International Journal of
[2] K. Aminian, D. Patchen, Thermoacoustic liquefaction of coal mine methane to Heat and Mass Transfer Volume 49 (Issues 15e16) (July 2006) 2567e2577.
produce LNG for heavy vehicle applications, U.S. Department of Energy Report [14] H. Najafi, B. Najafi, Multi-objective optimization of a plate and frame heat
(1999). exchanger via genetic algorithm, Heat and Mass Transfer Volume 46 (Issue 6)
[3] J.V. Wolfersdorf, E. Achermann, B. Weigand, Shape optimization of cooling (May 2010) 639e647.
channels using genetic algorithms, ASME Journal of Heat Transfer 119 (1997) [15] F.P. Incropera, D.P. DeWitt, Fundamentals of heat and mass transfer. John
380e388. Wiley and Sons, Inc, 1998.
[4] G. Fabbri, A genetic algorithm for fin profile optimisation, International [16] H.M. Joshi, R.L. Webb, Heat transfer and friction in the offset strip-fin heat
Journal of Heat and Mass Transfer 40 (9) (1997) 2165e2172. exchanger, International Journal of Heat and Mass Transfer 30 (1) (1987)
[5] T.S. Schmit, A.K. Dhingra, F. Landis, G. Kojasoy, A genetic algorithm optimi- 69e84.
zation technique for compact high intensity cooler design, Journal of [17] F.O. Jegede, G.T. Polley, Optimum heat exchanger design, Trans IChemE, Part A
Enhanced Heat Transfer 3 (4) (1996) 281e290. 70 (1992) 133e141.
[6] M.C. Tayal, Y. Fu, U.M. Diwekar, Optimum design of heat exchangers: [18] K. Muralikrishna, U.V. Shenoy, Heat exchanger design targets for minimum
a genetic algorithm framework, Industrial and Engineering Chemistry area and cost, Trans. IChemE, Part A 78 (2000) 161e167.
Research 38 (1999) 456e467. [19] O. Genceli, Heat Exchangers. Birsen Book Company, Istanbul, Turkey, 1999.
[7] A. Traverso, A.F. Massardo, Optimal design of compact recuperators for micro- [20] A. Konak, D. Coit, A. Smith, Multi-objective optimization using genetic algo-
turbine application, Applied Thermal Engineering 25 (2005) 2054e2071. rithm: a tutorial, Reliability Engineering and System Safety 91 (2006)
[8] Hao Peng, Xiang Ling, Optimal design approach for the plate-fin heat 992e1007.
exchangers using neural networks cooperated with genetic algorithms, [21] J. Holland, Adaptation in Natural and Artificial System. University of Michigan
Applied Thermal Engineering 28 (2008) 642e650. Press, Ann Arbor, 1975.
[9] D.P. Seculic, C.V. Herman, One approach to irreversibility minimization in [22] D.E. Goldberg, Genetic Algorithms in Search, Optimization, and Machine
compact crossflow heat exchanger, International Communications in Heat and Learning. Addison-Wesley Longman, Inc, 2000.
Mass Transfer 13 (1986) 23e32.

S-ar putea să vă placă și