You are on page 1of 12

This article was downloaded by: [Uni of Ontario Inst of Technology ]

On: 24 January 2012, At: 15:01


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Heat Transfer Engineering


Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/uhte20

Exergetic Optimization of Shell-and-Tube Heat


Exchangers Using NSGA-II
a b b
Hassan Hajabdollahi , Pouria Ahmadi & Ibrahim Dincer
a
Young Researchers Club, Islamic Azad University, Zahedan Branch, Iran
b
Department of Mechanical Engineering, University of Ontario Institute of Technology
(UOIT), Oshawa, Ontario, Canada

Available online: 17 Oct 2011

To cite this article: Hassan Hajabdollahi, Pouria Ahmadi & Ibrahim Dincer (2012): Exergetic Optimization of Shell-and-Tube
Heat Exchangers Using NSGA-II, Heat Transfer Engineering, 33:7, 618-628

To link to this article: http://dx.doi.org/10.1080/01457632.2012.630266

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,
proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.
Heat Transfer Engineering, 33(7):618–628, 2012
Copyright 
C Taylor and Francis Group, LLC
ISSN: 0145-7632 print / 1521-0537 online
DOI: 10.1080/01457632.2012.630266

Exergetic Optimization of
Shell-and-Tube Heat Exchangers
Using NSGA-II

HASSAN HAJABDOLLAHI,1 POURIA AHMADI,2 and IBRAHIM DINCER2


Downloaded by [Uni of Ontario Inst of Technology ] at 15:01 24 January 2012

1
Young Researchers Club, Islamic Azad University, Zahedan Branch, Iran
2
Department of Mechanical Engineering, University of Ontario Institute of Technology (UOIT), Oshawa, Ontario, Canada

In this article, a multi-objective exergy-based optimization through a genetic algorithm method is conducted to study and
improve the performance of shell-and-tube type heat recovery heat exchangers, by considering two key parameters, such as
exergy efficiency and cost. The total cost includes the capital investment for equipment (heat exchanger surface area) and
operating cost (energy expenditures related to pumping). The design parameters of this study are chosen as tube arrangement,
tube diameters, tube pitch ratio, tube length, tube number, baffle spacing ratio, and baffle cut ratio. In addition, for optimal
design of a shell-and-tube heat exchanger, the ε − NT U method and Bell–Delaware procedure are followed to estimate
its pressure drop and heat transfer coefficient. A fast and elitist nondominated sorting genetic algorithm (NSGA-II) with
continuous and discrete variables is applied to obtain maximum exergy efficiency with minimum exergy destruction and
minimum total cost as two objective functions. The results of optimal designs are a set of multiple optimum solutions, called
“Pareto optimal solutions.” The results clearly reveal the conflict between two objective functions and also any geometrical
changes that increase the exergy efficiency (decrease the exergy destruction) lead to an increase in the total cost and vice
versa. In addition, optimization of the heat exchanger based on exergy analysis revealed that irreversibility like pressure drop
and high temperature differences between the hot and cold stream play a key role in exergy destruction. Therefore, increasing
the component efficiency of a shell-and-tube heat exchanger increases the cost of heat exchanger. Finally, the sensitivity
analysis of change in optimum exergy efficiency, exergy destruction, and total cost with change in decision variables of the
shell-and-tube heat exchanger is also performed.

INTRODUCTION the cost of heat transfer surface area or capital investment as an


objective function [1, 2], while others assumed entropy genera-
The shell-and-tube heat exchanger, as shown in Figure 1, is tion [3–5].
widely used in many industrial power generation plants, as well Ahmadi et al. [6] conceptually discussed exergy from several
as in chemical, petrochemical, and petroleum industries. These perspectives and introduced the exergy analysis method as a
exchangers are used to transfer heat between two or more fluids, useful tool for furthering the goal of more efficient use of energy
between a solid surface and a fluid, or between solid particulates resources. Bejan [7] demonstrated the use of irreversibility as
and a fluid, at different temperatures and in thermal contact. In a criterion for evaluation of the efficiency of a heat exchanger.
heat exchangers, there are usually no external heat and work The purpose was to minimize the wasted energy by optimum
interactions. design of fluid passages in a heat exchanger.
There are some effective parameters in shell-and-tube heat Johannessen et al. [8] proved that the entropy production
exchanger design, such as tube numbers, tube length, tube ar- due to heat transfer in a heat exchanger is a minimum when
rangement, and baffle spacing. Some researchers consider only the local entropy production is constant throughout all parts
of the system. A new design strategy, involving losses due to
fluid and heat transfer irreversibilities that lead to production
Address correspondence to Professor Ibrahim Dincer, Department of Me-
of entropy, has recently been presented by Lerou et al. [9] and
chanical Engineering, Faculty of Engineering and Applied Science, University
of Ontario Institute of Technology, 2000 Simcoe St. North, Oshawa, ON L1H applied to the thermal design of a counterflow heat exchanger
7K4, Canada. E-mail: Ibrahim.Dincer@uoit.ca through minimization of entropy generation. Haseli et al. [10]

618
H. HAJABDOLLAHI ET AL. 619

Figure 1 Schematic of TEMA E shell-and-tube heat exchanger.

optimized the temperatures in a shell-and-tube condenser with ANALYSIS


respect to exergy. The optimization problem is defined subject
Downloaded by [Uni of Ontario Inst of Technology ] at 15:01 24 January 2012

to condensation of the entire vapor mass flow and it is solved Heat Transfer Analysis
based on the sequential quadratic programming (SQP) method.
Also it was concluded that optimization results revealed new An E-type TEMA shell is considered for this case study as
characteristics for the cooling water, with respect to the min- shown in Figure 1. The heat-exchanger effectiveness for this
imization of exergy destruction of the condensation process. type of shell is estimated from [18] using
Thus, these papers show that exergy analysis and optimization
based on the second law of thermodynamics is an important 2
ε=   (1)
tool for design, analysis, and improvement of heat exchangers (1 + C ∗) + (1 + C ∗2 )0.5 coth N T2 U (1 + C ∗2 )0.5
and their networks. Moreover, in reference [11] the capital cost
where the heat capacity ratio (C ∗ ) and the number of transfer
of the heat exchanger was considered as an objective function,
units (NTU) are defined as
while another group of researchers [12–16] considered the sum
of capital/investment costs (related to the heat transfer area and U At
N T Umax = (2)
energy-related costs of friction losses in the fluid flow) as an ob- Cmin
jective function to optimize shell-and-tube heat exchanger. Two
. .
objective functions (minimization of the total annualized cost Cmin min(Cc , C h ) min((m c p )c , (m c p )h )
and the amount of cooling water required) were also studied in C∗ = = = . .
Cmax max(Cc , C h ) max((m c p )c , (m c p )h )
reference [17].
In this study, after thermal modeling of an industrial shell- (3)
and-tube heat recovery heat exchanger using the ε − NT U
method and the Bell–Delaware approach [18], the exchanger where At is the total heat transfer surface area and U is the
is optimized by maximizing the exergy efficiency and minimiz- overall heat transfer coefficient, defined as
ing the total cost and exergy destruction. A genetic algorithm
technique is then employed to provide a set of the Pareto optimal At = πLdo Nt (4)
solutions. The sensitivity analysis of change in optimum values
1
of exergy efficiency (and exergy destruction) and total cost with U = do ln(do /di )
(5)
change in design variables is performed and the results are dis-
1
ho
+ Ro, f + 2kw
+ Ri, f do
di
+ 1 do
h i di
cussed. In this analysis, an earlier study [10] is greatly enhanced
Here, L, Nt , di , do , Ri, f , Ro, f , and kw are tube length, tube
by the following specific objectives:
number, tube inside and outside diameter, fouling resistance
in tube and shell side, and thermal conductivity of tube wall,
• To conduct a multi-objective optimization of shell-and-tube respectively.
heat exchangers with respect to exergy efficiency (exergy de- The tube-side heat transfer coefficient (h i ) is applied in form
struction) and total cost. of [18]:
• To find the best and optimal design parameters such as tube
h i = h t = (kt /di ) 0.024Re0.8 0.4
t Prt for 2500 < Ret < 1.24×105
configuration, tube diameter, tube pitch ratio, tube length, baf-
(6)
fle spacing ratio, and baffle cut ratio using genetic algorithm.
where kt and Prt are tube-side fluid thermal conductivity and
• To propose a closed-form equation for the total cost in term of
Prandtl number, and Ret is
exergy efficiency (exergy destruction) at the optimal design
.
point. t mdi
• To perform sensitivity analysis for the objective functions Ret = (7)
µt Ao,t
when the optimum design parameters vary.

heat transfer engineering vol. 33 no. 7 2012


620 H. HAJABDOLLAHI ET AL.

Here, ṁ t is the mass flow rate and Ao,t is the tube-side flow Table 1 The ratio of minimum free flow area to frontal area
cross-section area per pass, defined as 30◦ Triangular 45◦ Rotated square 90◦ Rotated
Square staggered array staggered array staggered array
Ao,t = 0.25πdi2 Nt /n p (8) √
2 pt − d o pt
√ f or ≥ 1.707
Also n p is the number of tube passes. The shell diameter can pt − do 2 pt do p t − do
be estimated from [19]: σ
pt 2( pt − do ) pt pt
 √ f or ≤ 1.707
2 pt do
Ds = 0.637 pt (πNt )C L/C T P (9)

where pt is tube pitch and C L is the tube layout constant, which


is unity for 45◦ and 90◦ tube arrangement and 0.87 for 30◦ and equation [18]:
60◦ tube arrangement. Also, CTP is the tube count calculation f t = 0.00128 + 0.1143(Ret )−0.311 (13)
constant, which is 0.93, 0.9, and 0.85 for a single pass, two
passes, and three passes of tube, respectively [20]. Furthermore, for 4000 < Ret < 10 range with ±2%.
7
Downloaded by [Uni of Ontario Inst of Technology ] at 15:01 24 January 2012

the Bell–Delaware method was used to compute the shell-side


heat transfer and friction factor coefficients in the form of

h o = h s = h id Jc Jl Jb Js Jr (10) Exergy Analysis

where h id is the heat transfer coefficient for the pure cross-flow Exergy is a measure of the departure of the state of a system
stream over tube bundle evaluated at a Reynolds number at or from that of the environment. It can be defined as the maxi-
near the center line of the shell in the form of [20]: mum obtainable work from the combination of the system and
 .  2/3   environment. Unlike energy, exergy is not conserved; indeed, it
ms ks µs 0.14 is destroyed by irreversibilities. The exergy destruction during
h id = js c p,s (11)
As c p,s µs µs,w a process is proportionally associated with entropy generation
due to these irreversibilities [21, 22].
Here, js is the ideal tube bank Colburn factor, As is the cross-
The steady-state exergy rate balance for a control volume can
flow area at or near the shell center line, and µs /µs,w is the ratio
be written as
 
between viscosity in average temperature and wall temperature
T0  
in the shell side. 1− Q̇ j − Ẇcv + ṁ i ei − ṁ e ee − Ė x D = 0
The shell-side pressure drop is computed as three terms in- j
Tj i e
cluding cross-flow section pressure drop, inlet and outlet pres-
sure drop and window section pressure drop. The details of (14)
computing Colburn factor, friction factor, and cross-flow area
at or near the shell center line can be found in reference [20]. where Q̇ j represents the heat transfer rate at the location
These factors depend on tube arrangement and Reynolds num- on the boundary where the instantaneous temperature is T j ,
ber. Jc is the correction factor for baffle configuration (baffle andẆcv represents the energy transfer by work, other than
cut and spacing) and takes into account the heat transfer in the flow work. The specific flow exergy, e, is evaluated as
window. Jl is the correction factor for baffle leakage effects and follows:
takes into account both the shell-to-baffle and tube-to-baffle hole V2
leakages. Jb is the correction factor for bundle and pass partition e = (h − h ◦ ) − T◦ (S − S◦ ) + + gz (15)
2
bypass streams and depends on the flow bypass area and num-
where h and S denote, respectively, enthalpy and entropy of the
ber of sealing strips. Js is the correction factor for bigger baffle
system and ho and So are the values of the same properties if the
spacing at the shell inlet and outlet sections. Jr is the correction
system was at the dead state. Also, To refers to the dead state
factor for the adverse temperature gradient in laminar flows (at
(environment) temperature.
low Reynolds numbers) [18].
Therefore, the exergy destruction for a heat exchanger is
Furthermore, the pressure drop in tube side was also esti-
calculated as follows:
mated from  
.   Ė x D,H E = ṁ i ei − ṁ e ee (16)
m 2t 4 ft L i e
Pt = + (1 − σ2 + K c ) − (1 − σ2 − K e ) n p
2ρt A2o,t di
(12) The exergy efficiency of a heat exchanger is calculated as
.
where K c and K e are entrance and exit pressure loss coefficient E x D,H E x
for a multiple circular tube core, respectively, and σ is defined ηex,A P = 1 − (17)
Ex
as the ratio of minimum free flow area to frontal area (Table 1). i,H E x
The term f t is the friction factor obtained from the following
heat transfer engineering vol. 33 no. 7 2012
H. HAJABDOLLAHI ET AL. 621

MULTI-OBJECTIVE OPTIMIZATION USING GENETIC Tournament Selection


ALGORITHM
Each two individuals compete in a tournament with randomly
A multi-objective problem consists of optimizing (i.e., min- selected individuals, a procedure that imitates survival of the
imizing or maximizing) several objectives simultaneously, with fittest in nature.
a number of inequality or equality constraints. The problem can
formally be written to find:
Find Controlled Elitism Sort

x = (xi )∀i = 1, 2, . . . , N param (18) To preserve diversity, the influence of elitism is controlled by
choosing the number of individuals from each subpopulation,
such that f i (x) is a minimum(respectively maximum) ∀i = according to the geometric distribution [28]:
1, 2, . . . , Nobj , subject to:
1 − c q−1
Sq = S c , (21)
1 − cw
Downloaded by [Uni of Ontario Inst of Technology ] at 15:01 24 January 2012

g j (x) = 0 ∀ j = 1, 2, . . . , M, (19)
to form a parent search population, Pt+1 (t denote the genera-
h k (x) ≤ 0 ∀k = 1, 2, . . . , K , (20) tion), of size S, where 0 < c < 1 and w is the total number of
ranked nondominated individuals.
where x is a vector containing the N param design parameters,
( f i )i=1,....,N obj the objective functions, and Nobj the number of
objectives. The ( f i )i=1,....,N obj returns a vector containing the Crowding Distance
set of Nobj values associated with the elementary objectives to
be optimized simultaneously. The genetic algorithms are semis- The crowding distance metric proposed by Deb and Goel
tochastic methods, based on an analogy with Darwin’s laws of [27] is utilized, where the crowding distance of an individual is
natural selection [23]. The first multi-objective genetic algo- the perimeter of the rectangle with its nearest neighbors at diag-
rithm (GA), called vector-evaluated GA (or VEGA), was pro- onally opposite corners. Thus, if individual X (a) and individual
posed by Schaffer [24]. An algorithm based on nondominated X (b) have the same rank, each one that has a larger crowding
sorting was proposed by Srinivas and Deb [25] and called non- distance is better.
dominated sorting genetic algorithm (NSGA). This was later
modified by Deb et al. [26], which eliminated higher computa-
tional complexity, lack of elitism, and the need for specifying Crossover and Mutation
the sharing parameter. This algorithm is called NSGA-II and is
coupled with the thermal modeling programming for optimiza- Uniform crossover and random uniform mutation are em-
tion. ployed to obtain the offspring population, Q t+1 . The integer-
based uniform crossover operator takes two distinct parent in-
dividuals and interchanges each corresponding binary bit with
a probability 0 < pc ≤ 1. Following crossover, the mutation
Nondominated Sorting operator changes each of the binary bits with a mutation proba-
bility, 0 < pm < 0.5.
Following the definition by Deb and Goel [27], an individual
X (a) is said to constrain-dominate an individualX (b) if any of the
following conditions are true: Objective Functions, Design Parameters, and Constraints

1. X (a) and X (b) are feasible, with (a) X (a) is no worse than X (b) In this study, the exergy efficiency (and exergy destruction)
in all objective, and (b) X (a) is strictly better thanX (b) in at and total cost are considered as two objective functions. The total
least one objective. cost includes the cost of heat transfer area and the operating cost
2. X (a) is feasible while individualX (b) is not. for the pumping power:
3. X (a) andX (b) are both infeasible, but X (a) has a smaller con-
Ctotal = Cin + Cop (22)
straint violation.
The investment cost for both the stainless steel-made shell
Here, the constraint violation  (X ) of an individual X is and tube is considered as [29]:
defined to be equal
to the sum of the violated constraint function Cin = 8500 + 409A0.85 (23)
values, (X ) = Bj=1 γ(g j (X ))g j (X ), where γ is the Heaviside t

step function. where At is the total tube outside heat transfer area.
heat transfer engineering vol. 33 no. 7 2012
622 H. HAJABDOLLAHI ET AL.

Table 2 The operating conditions of the shell-and-tube heat exchanger (input Table 4 Comparison of modeling output and the corresponding results from
data for the model) reference [18]

Properties Shell side (hot water) Tube side (cold water) Variables Unit Reference [15] Present article Difference (%)

Density (kg/m3) 980 995 — 0.1555 0.1599 2.83


Specific heat (J/kg-K) 4180 4120 C total $ 74598 74112 0.65
Viscosity (Pa-s) 0.000672 0.000695 Pt kPa 17.58 17.660 0.45
Fouling factor (m2-W/K) 0.000065 0.000074 Ps kPa 112 111.02 −0.875
Thermal conductivity (W/m-K) 0.56 0.634 q kW 393.6 404.63 2.78
ht W/m2–K 7837 7838.2 0.0153
hs W/m2–K 698.8 730.226 4.497
The total discounted operating cost related to pumping power
to overcome friction losses is computed from the following
equations [29]: tubes with definite inner and outer diameter listed in Table 3 are
considered as discrete design variables [30].

ny
Downloaded by [Uni of Ontario Inst of Technology ] at 15:01 24 January 2012

Co
Cop = (24)
k=1
(1 + i)k
RESULTS AND DISCUSSION
Co = H kel τ (25)
Modeling Verifications and Optimization Results
 
1 ṁ t ṁ s
H = Pt + Ps (26) To verify the modeling results, the simulation output were
η ρt ρs
compared with the corresponding reported results given in lit-
where ny is the equipment lifetime per year, i is annual dis- erature. The comparison of our modeling results and the corre-
count rate, and kel , τ, and η are price of electrical energy, sponding values from reference [18], for the same input values,
hours of operation per year, and pump efficiency, respectively. is shown in Table 4. Results show that the difference percentage
In this study tube arrangement, tube diameters, tube pitch ratio points of modeling output results are acceptable.
pt /do , tube length, tube number, baffle spacing ratio(L bc /Ds,i ),
and baffle cut ratio (BC/Ds,i ) were considered as seven de-
cision variables. The following constraints are introduced to Optimization Results
insure to satisfy that the ratio of L/Ds is in the range of
3 < L/Ds < 12. To maximize the exergy efficiency and minimize the total
cost, and the exergy destruction, seven design parameters
including tube arrangement, tube diameters, tube pitch ratio,
CASE STUDY tube length, tube number, baffle spacing ratio, and baffle
cut ratio were selected. The design parameters (as decision
The optimum heat exchanger configurations were obtained variables) and the range of their variations are listed in Table
for an oil cooler shell-and-tube heat recovery heat exchanger 5 according to reference [31]. The number of iterations for
in the Sarcheshmeh cupper production power plant located in finding the global optimum in the whole searching domain is
the south of Kerman city. The goals in this study were to max- 8.2 × 1015. Also, the running time of the CPU for this condition
imize exergy efficiency (minimize exergy destruction) while for 200 generations is about 10 minutes.
minimizing the total cost. The hot water (hot stream) with 8.1
kg/s mass flow rate passes through the shell side of heat ex-
changer at 78.3◦ C. The fresh water (cold stream) with 12.5 Multi-Objective Optimization of Exergy Efficiency
kg/s mass flow rate passes through the tube side at 30◦ C. The and Total Cost
operating conditions are listed in Table 2. In this study, the
equipment life assigned ny = 10yr , the rate of annual dis- The genetic algorithm optimization is performed for 100
count assigned i = 10%, price of electrical energy considered generations, using a search population size of M = 100 in-
kel = 0.15/kW h, and hours of operation and pump efficiency dividuals, crossover probability of pc = 0.9, gene mutation
considered are τ = 7500h/yr and η = 0.6, respectively. Three probability of pm = 0.035, and controlled elitism value of
tube arrangements (30◦ , 45◦ , 90◦ ) and the number of 20 standard c = 0.65. The results for a Pareto-optimal front are shown in

Table 3 Inner and outer diameters of 20 standard tubes

Inner diameter (in) 0.444 0.407 0.435 0.481 0.495 0.509 0.527 0.541 0.555 0.482 0.510 0.532 0.560 0.584 0.606 0.620 0.634 0.352 0.680 0.607
Outer diameter (in) 1/2 5/8 5/8 5/8 5/8 5/8 5/8 5/8 5/8 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 3/4 7/8

heat transfer engineering vol. 33 no. 7 2012


H. HAJABDOLLAHI ET AL. 623

Table 5 The design parameters, their range of variation, and their change step Table 6 The optimum values of exergy efficiency, total cost, and exergy
destruction for the design points A to F in Pareto-optimal fronts for input
Variables From To Step of change values given in Table 2

Tube arrangement (30◦ ,45), 90◦ ) (—) 1 A B C D E F


Tube inside diameter, m 0.0112 0.0153 —
Tube outside diameter, m 0.0126 0.022 — Exergy efficiency 0.9249 0.8889 0.7654 0.7686 0.8675 0.9246
pt /do 1.25 2 0.001 Total cost ($) 57,359 22,090 14,337 14,397 19,995 43,959
Tube length, m 3 8 0.001 Exergy destruction (kW) 12.154 17.971 37.959 37.447 21.435 12.192
Tube number 100 600 1
Baffle cut ratio 0.19 0.32 0.001
Baffle spacing ratio 0.2 1.4 0.001
exergy efficiency. The interesting point here is that considering
a numerical value for the exergy efficiency in the mentioned
range provides the minimum total cost for that optimal point
Figure 2, which clearly reveal the conflict between two ob-
along with other optimal design variables.
jectives, exergy efficiency versus total cost. Any geometrical
Downloaded by [Uni of Ontario Inst of Technology ] at 15:01 24 January 2012

change that increases the exergy efficiency leads to an increase


in the total cost, and vice versa. This shows the need for multi-
Multi-Objective Optimization of Exergy Destruction and
objective optimization techniques in optimal design of a shell-
Total Cost
and-tube heat exchanger. It is shown in Figure 2 that the max-
imum exergy efficiency appears at design point A (0.9249),
while the total cost is the biggest at this point. On the other The genetic algorithm optimization was performed for 100
hand, the minimum total cost occurs at design point C (14340 generations, using a search population size of M = 100 individu-
$), with a smallest exergy efficiency value (0.7654) at that point. als, crossover probability of pc = 0.9, gene mutation probability
Design point A is the optimal situation at which,\ exergy effi- of pm = 0.035, and controlled elitism value c = 0.55. The results
ciency is a single objective function, while design point C is for the Pareto-optimal front are shown in Figure 2, which clearly
the optimum condition at which total cost is a single objective reveal the conflict between two objectives (exergy destruction
function. versus total cost). Any geometrical change that decreases the
To provide a useful tool for the optimal design of the shell- exergy destruction leads to an increase in the total cost and vice
and-tube heat exchanger, the following equation for exergy ef- versa. It is shown in Figure 2 that the minimum exergy destruc-
ficiency versus the total cost was derived for the Pareto curve tion exists at design point F (12.19 kW), while the total cost is
(Figure 2): the biggest at this point. On the other hand, the minimum total
cost occurs at design point D ($14,397), with a biggest exergy
−94.96η3ex + 4.082η2ex + 144.4ηex − 61.56 destruction value (37.447 kW) at that point. Design point F is
Ctotal ($) = × 105
η2ex − 397.8ηex + 367.8 the optimal situation at which exergy destruction is a single ob-
(27) jective function, while design point D is the optimum condition
which is valid in the range of 0.7654 < ηex < 0.9249 for at which total cost is a single objective function.
To provide a useful tool for the optimal design of the shell-
and-tube heat exchanger, the following equation for exergy de-
struction versus the total cost was derived for the Pareto curve
(Figure 2):
5.848E 2D + 384E D − 5215
Ctotal ($) = × 103 (28)
E 2D − 3.483E D − 97.95
which is valid in the range of 12.19 < E D < 37.45 (kW )
for exergy destruction. The key point here is that consid-
ering a numerical value for the exergy destruction in the
mentioned range provides the minimum total cost for that
optimal point along with other optimal design variables. Op-
timum objectives for six typical points from A to F Pareto-
optimal fronts for input values given in Table 2 are listed in
Table 6.
The variation of optimum values of exergy efficiency with
the total cost for various values of optimum design variables
in A–C cases (Pareto front) and exergy destruction with the
Figure 2 The distribution of Pareto-optimal points solutions using NSGA-II. total cost for various values of optimum design parameters in
(Color figure available online.) D–F are shown in Figures 3a–e and Figures 4a–e, respectively.
heat transfer engineering vol. 33 no. 7 2012
624 H. HAJABDOLLAHI ET AL.
Downloaded by [Uni of Ontario Inst of Technology ] at 15:01 24 January 2012

Figure 3 The variation of exergy efficiency with total cost for five optimum design parameters in three cases of A–C where exergy efficiency and total cost are
given as objective functions.(Color figure available online.)

The effect of tube diameter is not considered because there is Effect of Decision Variables on Objective Functions
no direct relation between tube diameter and tube thickness in
the existing tubes. In addition, the optimal tube pitch for two Tube Pitch Ratio
optimization methods was fixed at 45 degrees. It was observed
that the variation of two objective functions at other points on By increasing the pt /do , both exergy efficiency and total cost
a Pareto optimal front had the same trend as the six points decreased for all design points A-C (Figure 3a). Also, increase
(A–F). The effect of design variables on objective functions are in this parameter results in increment of exergy destruction in
investigated and explained as follows. all points from D to F in Figure 4a. Therefore, variations of tube
pitch ratio cause a conflict between two objectives.

heat transfer engineering vol. 33 no. 7 2012


H. HAJABDOLLAHI ET AL. 625
Downloaded by [Uni of Ontario Inst of Technology ] at 15:01 24 January 2012

Figure 4 The variation of exergy destruction with total cost for five optimum design parameters in three cases of D–F where exergy destruction and total cost are
given as objective functions. (Color figure available online.)

Tube Length increment of this parameter leads to decrease in exergy destruc-


tion for all points from D to F in Figure 4c. Therefore, the tube
By increasing the tube length, both exergy efficiency and
number causes a conflict between these two objective functions.
total cost increases for all design points A–C (Figure 3b). In
addition, this parameter results in decrease in exergy destruction
in all points from D to F in Figure 4b. Therefore, variations of
tube length cause a conflict between two objective functions.
Baffle Spacing Ratio
As shown in Figures 3d and 4d, an increase in the baffle
Tube Number spacing ratio (L bc /Ds,i ) creates a conflict between two objective
Like tube length, both exergy efficiency and total cost in- functions (exergy efficiency and total cost, exergy destruction,
crease with increasing the tube number (Figure 3c). Moreover, and total cost).
heat transfer engineering vol. 33 no. 7 2012
626 H. HAJABDOLLAHI ET AL.

Baffle Cut Ratio CT P tube count calculation constant(−)


di inner tube diameter (m)
Increase in baffle cut ratio (BC/Ds,i ) decreases both exergy
do outer tube diameter (m)
efficiency and total cost (Figure 3e) but not directly. However
Ds shell diameter (m)
the effect is not considerable. Also, increase in this parameter
ED exergy destruction (kW)
changes exergy destruction a little for all points in Figure 4e.
f friction factor (−)
h heat transfer coefficient (W/m2-K)
CONCLUSIONS h id ideal heat transfer coefficient in shell side
hi tube-side heat transfer coefficient (W/m2-K)
In this work, a multi-objective exergy-based optimization ho shell-side heat transfer coefficient (W/m2-K)
through a genetic algorithm method has been conducted to H pumping power (W)
study and improve the performance of shell-and-tube type heat i annual discount rate (%)
recovery heat exchangers, by considering two key parameters, j Colburn number (−)
such as exergy efficiency and cost. The design parameters k thermal conductivity (W/m-K)
Downloaded by [Uni of Ontario Inst of Technology ] at 15:01 24 January 2012

(decision variables) were tube arrangement, baffle cut ratio, tube Kc entrance pressure loss coefficient (−)
pitch ratio, tube length, tube number, and baffle spacing ratio, as Ke exit pressure loss coefficient (−)
well as 20 standard tubes with definite inner and outer diameter. kel price of electrical energy ($/kWh)
In the presented optimization problem, the exergy efficiency L tube length (m)
(and exergy destruction) and the total cost were considered L bc baffle spacing (m)
.
two objective functions. Therefore exergy efficiency was max- m mass flow rate (kg/s)
imized, while minimizing the exergy destruction, and the total np number of tube pass(−)
cost was minimized. Here are some of the concluding remarks: ny equipment life (yr)
Nt number of tubes (−)
• When the exergy efficiency of a shell-and-tube heat exchanger NTU number of transfer units (−)
increases, the total cost of the heat exchanger increases respec- pt tube pitch (m)
tively. Pr Prandtl number (−)
• Increasing heat exchanger exergy efficiency leads to a more q heat transfer rate (kW)
efficient heat exchanger both thermodynamically and thermo- Ro, f shell-side fouling resistance (m2-K/W)
economically. Ri, f tube-side fouling resistance (m2-K/W)
• Decreasing exergy destruction increases the total cost. Re Reynolds number (−)
• Irreversibility, like pressure drop and high temperature differ- T temperature (◦ C)
ence between cold and hot stream, plays a key role in exergy U overall heat transfer coefficient (W/m2-K)
destruction.
• Tube pitch ratio, tube length, tube number, and baffle spacing
ratio appear to be important design parameters, while tube Greek Symbols
arrangement indicates no effect on the conflict between two
optimized objective functions. σ ratio of minimum free flow area to frontal area (−)
ε thermal effectiveness(−)
P pressure drop (Pa)
NOMENCLATURE
ρ density (kg/m3)
µ viscosity (Pa-s)
Ao,t tube side flow area per pass (m2) η pump efficiency (−)
At total tube outside heat transfer area (m2) ηex exergy efficiency (−)
As cross flow area at or near the shell center line τ hours of operation per year (h/yr)
BC baffle cut (m)
cp specific heat at constant pressure (J/kg-K)
C flow stream heat capacity rate (W/k) Subscripts
Cmin minimum of C h and Cc (W/K)
Cmax maximum of C h and Cc (W/K) c cold stream
C∗ heat capacity rate ratio (Cmin /Cmax ) h hot stream
Cin investment cost ($) i inner or inlet
Cop operational cost ($) o outer or outlet
Co annual operating cost ($/yr) s shell-side
Ctotal total cost ($) t tube side
CL tube layout constant(−) w tube wall

heat transfer engineering vol. 33 no. 7 2012


H. HAJABDOLLAHI ET AL. 627

REFERENCES Transfer Equipment Design, Computers and Chemical En-


gineering, vol. 31, pp. 1432–1448, 2007.
[1] Costa, L. H., and Queiroz, M., Design Optimization of [15] Caputo, A. C., Pelagagge, P. M., and Salini, P., Heat Ex-
Shell-and-Tube Heat Exchangers, Applied Thermal Engi- changer Design Based on Economic Optimization, Applied
neering, vol. 28, pp. 1798–1805, 2008. Thermal Engineering, vol. 28, pp. 1151–1159, 2008.
[2] Ramananda, R. K., and Shrinivasa, U. J., Synthesis of Cost [16] Ozcelik, Y., Exergetic Optimization of Shell and Tube
Optimal Shell-and-Tube Heat Exchangers, Heat Transfer Heat Exchangers Using a Genetic Based Algorithm,
Engineering, vol. 12, pp. 47–55, 1991. Applied Thermal Engineering, vol. 27, pp. 1849–
[3] Bejan, A., Tsatsaronis, G., and Moran, M., Thermal Design 1856, 2007.
and Optimization, Wiley Interscience, John Wiley & Sons, [17] Agarwal, A., and Gupta, S. K., Jumping Gene Adaptations
New York, 1996. of NSGA-II and Their Use in the Multi-Objective Opti-
[4] Johannessen, E., Nummedal, L., and Kjelstrup, S., Min- mal Design of Shell and Tube Heat Exchangers, Chemical
imizing the entropy production in heat exchange, Inter- Engineering Research and Design, vol. 86, pp. 123–139,
national Journal of Heat and Mass Transfer, vol. 45, pp. 2008.
Downloaded by [Uni of Ontario Inst of Technology ] at 15:01 24 January 2012

2649–2654, 2002. [18] Shah, R. K., and Sekulic, P., Fundamentals of Heat Ex-
[5] Sun, S., Lu, Y., and Yan, C., Optimization in calculation changer Design, John Wiley & Sons, Hoboken, NJ, 2003.
of shell-and-tube heat exchanger, International Communi- [19] Taborek, J., Industrial Heat Exchanger Design Practices
cation in Heat and Mass Transfer, vol. 20, pp. 675–685, in Boiler Evaporators and Condenser, Wiley, New York,
1993. 1991.
[6] Ahmadi, P., Hajabdollahi, H., and Dincer, I., Cost and En- [20] Kakac, S., and Liu, H., Heat Exchangers Selection
tropy Generation Minimization of a Cross Flow Plate Fine Rating, and Thermal Design, CRC Press, New York,
Heat Exchanger (PFHE) Using Multi-Objective Genetic 1997.
Algorithm, ASME Journal of Heat Transfer, vol. 133, no. [21] Dincer, I., and Rosen, M. A., Exergy, Elsevier Science,
2, DOI 1115/1.4002599 2011. Ltd., Oxford, 454 p., 2007.
[7] Bejan, A., General Criterion for Rating Heat-Exchanger [22] Orhan, M. F., Erek, A., and Dincer, I., Entropy Generation
Performance, International Journal of Heat and Mass During a Phase-Change Process in a Parallel Plate Channel,
Transfer, vol. 21, pp. 655–658, 1978. Thermochemical Acta, vol. 489, pp. 70–74, 2009.
[8] Johannessen, E., Nummedal, L., and Kjelstrup, S., Mini- [23] Goldberg, D. E., Genetic Algorithms in Search, Optimiza-
mizing the Entropy Production in Heat Exchange, Inter- tion and Machine Learning, Addison-Wesley, Reading,
national Journal of Heat and Mass Transfer, vol. 45, pp. MA, 1989.
2649–2654, 2002. [24] Schaffer, J. D., Multiple Objective Optimization With Vec-
[9] Lerou, P. P. P. M., Veenstra, T. T., Burger, J. F., Brake, H. tor Evaluated Genetic Algorithms, in Proceedings of the
J. M., and Rogalla, H., Optimization of Counter Flow Heat International Conference on Genetic Algorithm and Their
Exchanger Geometry Through Minimization of Entropy Applications, 1985.
Generation, Cryogenics, vol. 45, pp. 659–669, 2005. [25] Srinivas, N., and Deb, K., Multiobjective Optimization Us-
[10] Haseli, Y., Dincer, I., and Naterer, G. F., Optimum Tem- ing Nondominated Sorting in Genetic Algorithms, Jour-
peratures in a Shell and Tube Condenser With Respect to nal of Evolutionary Computing, vol. 2, no. 3, 1994, pp.
Exergy, International Journal of Heat and Mass Transfer, 221–248, 1994.
vol. 51, pp. 2462–2470, 2008. [26] Deb, K., Pratap, A., Agarwal, S., and Meyarivan, T. A.,
[11] Ponce-Ortega, J. M., Serna-Gonzalez, M., Salcedo- Fast and Elitist Multiobjective Genetic Algorithm: NSGA-
Estrada, L. I., and Jimenez-Gutierrez, A., Minimum- II, IEEE Trans. Evolutionary Computer, vol. 6, no. 2, pp.
Investment Design of Multiple Shell and Tube Heat Ex- 182–197, 2002.
changers Using a minlp Formulation, Chemical Engineer- [27] Deb, K., Goel, T., Controlled Elitist Non-Dominated
ing Research and Design, vol. 84, pp. 905–910, 2006. Sorting Genetic Algorithms for Better Convergence, in
[12] Fesanghary, M., Damangir, E., and Soleimani, I., De- Proceedings of The First International Conference on
sign Optimization of Shell and Tube Heat Exchangers Evolutionary Multi-Criterion Optimization, Zurich, 2001,
Using Global Sensitivity Analysis and Harmony Search 385–399.
Algorithm, Applied Thermal Engineering, vol. 29, pp. [28] Deb, K., Multi-Objective Optimization Using Evolutionary
1026–1031, 2009. Algorithms, John Wiley and Sons, Chichester, UK, 2001.
[13] Ponce-Ortega, J. M., Serna-Gonzalez, M., Rico, V., and [29] Taal, M., Bulatov, I., Klemes, J., and Stehlik, P., Cost Es-
Jiménez, A., Use of Genetic Algorithms for the Optimal timation and Energy Price Forecasts for Economic Eval-
Design of Shell-and-Tube Heat Exchangers, Applied Ther- uation of Retrofit Projects, Applied Thermal Engineering,
mal Engineering, vol. 29, pp. 203–209, 2009. vol. 23, pp. 1819–1835, 2003.
[14] Ravagnani, M. A. S. S., and Caballero, J. A., Optimal [30] Standards of the Tubular Exchanger Manufactures Asso-
Heat Exchanger Network Synthesis With the Detailed Heat ciation, 7th ed., TEMA, Tarrytown, NY, 1998.
heat transfer engineering vol. 33 no. 7 2012
628 H. HAJABDOLLAHI ET AL.

[31] Saunders, E. A. D., Heat Exchangers—Selection, Design of energy systems, renewable energies, optimization of thermal systems and
and Construction, Wiley, New York, 1988. thermodynamic modeling of integrated gas turbine SOFC power plants, ad-
vanced power plant technology, green energies, and heat exchanger design and
optimization. He has published more than 50 articles in journals and conference
Hassan Hajabdollahi received his B.Sc. degree in
proceedings as well as two books in translation and three books published for
mechanical engineering in 2006 from Sistan and
Prdazesh Isituations, Tehran, Iran. He has also been a reviewer for International
Balochestan University, Zahedan, Iran. He received
Journal of Energy Research, International Journal of Exergy, Energy Conver-
his master’s degree in mechanical engineering from
sion and Management, Applied Thermal Engineering Journal, and numerous
Iran University of Science and Technology Tehran,
national conferences in Iran.
Iran, in 2009. Now he is a Ph.D. student at Iran Uni-
versity of Science and Technology. His research in-
terests are heat exchangers optimization, power plant Ibrahim Dincer is a full professor at the Faculty of
optimization, combined heat and power generation Engineering and Applied Science at the University
optimization, central heating, nondominated sorting of Ontario Institute of Technology (UOIT), Canada.
genetic algorithm for multi-objective optimization (NSGA-II), artificial neural He is the author of more than 700 journal and confer-
networks, and fuzzy logic. He has published more than 32 articles in journals ence publications, 9 books, 12 edited books, and 33
and conference proceedings. He has also been a reviewer for Energy Conver- book contributions; editor-in-chief of International
Downloaded by [Uni of Ontario Inst of Technology ] at 15:01 24 January 2012

sion and Management, Heat Transfer Engineering, Energy and Building, and Journal of Energy Research, International Journal
International Journal of Greenhouse Gas Control. of Exergy, International Journal of Global Warming,
and The Open Environmental Engineering Journal,
Pouria Ahmadi received his B.Sc. degree in me- and is associate editor, regional editor, and has ed-
itorial board member responsibilities for a number of international reputable
chanical engineering in 2006 from Power and Water
journals. He has been keynote speaker in several prestigious conferences; he has
University of Technology (PWUT), Tehran, Iran. He
been chair of a number of international conferences, symposia, and workshops;
received his master’s degree in mechanical engineer-
served as executive/scientific committee member of various organizations and
ing from Iran University of Science and Technology,
conferences; is an inventor of various new thermal equipment configurations,
Tehran, Iran in 2009. In 2009 he was accepted for
systems and designs, and new models, correlations, and graphs for heat and mass
his Ph.D. at Sharif University of Technology, where
transfer parameters; and is the recipient of various research excellence awards,
he studied for one year as a research assistant at the
including the Premier’s Research Excellence Award in Canada more recently.
Advanced Heat Transfer Laboratory. Currently he is
His research interests cover many topics, particularly on sustainable energy
a Ph.D. student at the University of Ontario Institute
technologies.
of Technology, Canada. His research interests are second law analysis (exergy)

heat transfer engineering vol. 33 no. 7 2012

You might also like