Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/259536270
CITATIONS READS
10 216
3 authors:
Hoseyn Sayyaadi
Khaje Nasir Toosi University of Technology
80 PUBLICATIONS 1,282 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Hoseyn Sayyaadi on 18 October 2014.
Chem. Eng. Technol. 2014, 37, No. 1, 8794 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
88 H. Sayyaadi et al
Nowadays, the second category that involves entropy and Table 1. Stream data for hot (fluid a) and cold (fluid b) gas
exergy has been extended among the researchers. One suitable flows.
and interesting method for optimizing heat exchangers and
Parameters Fluid a Fluid b
other energy systems is entropy generation minimization
(EGM). The concept of EGM was first introduced for model- Mass flow rate [kg s ] 1
0.8962 0.8296
ing and optimizing the technique by Bejan [26]. Later, some
Inlet temperature [K] 513 277
researchers [2730] applied this method for optimizing heat
transfer devices. Inlet pressure [kPa] 100 100
Multi-objective optimization of a cross-flow PFHE was per- 1 1
Specific heat [J kg K ] 1017.7 1011.8
formed. The value of entropy generation was divided into two
3
parts, i.e., entropy generation related to heat transfer and to Density [kg m ] 0.8196 0.9385
fluid friction. Since the entropy generations related to heat Dynamic viscosity [N s m ] 2
241 218.2
transfer and pressure drop have conflicting objectives, in which
increase in one objective leads to decrease of another and vice Prandtl number 0.6878 0.6954
versa, in order to find the best values of decision variables, Heat duty Q [kW] 160
these objectives were optimized simultaneously in a multi-
objective optimization process which balanced the two con-
flicting objectives. The multi-objective optimization of a PFHE 2.1 Thermal Calculation of a PFHE
was performed using the genetic algorithm while some geo-
metric specifications of the PFHE were considered decision The heat balance between streams can be calculated as follows:
variables and proper constraints were imposed. The Pareto
optimal frontier was obtained and a final optimal solution was Q ma ha;in ha;out mb hb;in hb;out
(1)
selected using a class of decision-making tools called LINMAP Q ma cp;a Ta;in Ta;out mb cp;b Tb;in Tb;out
(linear programming technique for multidimensional analysis
of preference) method. where Q, m, h, cp, and T denote the rate of heat transfer, mass
flow rate of fluid, heat transfer coefficient, and specific heat of
fluid and temperature, respectively. On the other hand:
2 Thermal and Hydraulic Design of the
Q UAF LMTD (2)
Cross-Flow PFHE
U, A, and F mean the overall heat transfer coefficient, heat
A PFHE consists of a block of alternating layers of various fins transfer area, and correction factor. The logarithmic mean
and flat separators, known as partitioning plates [31, 32]. A temperature difference (LMTD) is defined as:
schematic view of a simple cross-flow PFHE is given in Fig. 1.
In this heat exchanger, two gas streams flow and heat is trans- DT1 DT2
LMTD (3)
ferred from a hot gas (fluid a) into a cold gas (fluid b) stream. ln
DT1
Data of the streams for hot and cold gas flows are indicated in DT2
Tab. 1 [33].
where
DT1 Ta;in Tb;out (4a)
www.cet-journal.com 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eng. Technol. 2014, 37, No. 1, 8794
Entropy generation minimization 89
The heat transfer coefficient (h) is written in terms of the where DP, f, and q denote the pressure drop, fanning friction
Colburn factor (j) as follows: factor, and density, respectively. A new detail regarding the
h hydraulic design of a PFHE is reported in [36].
j St Pr2=3 Pr 2=3 (12)
Gcp
j, St, Pr, and G denote the Colburn factor, Stanton number, 3 Entropy Generation Minimization
Prandtl number, and mass flux velocity, respectively. The Col-
burn factor j depends on parameters such as type of fin, geom- Entropy generation minimization (EGM) is based on basic
etry and details of makeup as well as Reynolds number of the principles of thermodynamics, heat and mass transfer, and
stream [32]. fluid mechanics. EGM was introduced earlier by Bosnjakovic
Substituting h and A in the heat balance equations (Eqs. (1) in the 1930s [37, 38] and the concept was further developed in
and (2)) leads to the following expression: the 1970s by Bejan [2]. EGM is used in real applications to de-
tect irreversibilities. These irreversibilities in heat exchangers
1 Ha ta 1 na ta are related to heat transfer between streams due to tempera-
2=3 1 2na ha La ture differences and frictional pressure drop. As noted by Bejan
ja ma cp;a Pra
(13) in [39], the entropy generation rate for two streams (e.g., a and
1 Hb tb 1 nb tb FLMTD b) can be achieved in terms of temperature and pressure as:
jb mb cp;b Prb
2=3 1 2nb hb Lb Q
Ta;2 Pa;2
S ma cp;a ln Ra ln
Ta;1 Pa;1
For Re 1500, it follows [35]: (21)
Tb;2 Pb;2
mb cp;b ln Rb ln
j 0:53Re 0:5
lf =Dh 0:15
fs=H tg 0:14
(14) Tb;1 Pb;1
Chem. Eng. Technol. 2014, 37, No. 1, 8794 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
90 H. Sayyaadi et al
Pa;2 Pb;2 Ta;2 Tb;2
SDP ma Ra ln mb Rb ln Ns;DT ma cp;a ln mb cp;b ln =Cmax
Pa;1 Pb;1 Ta;1 Tb;1
Pa;1 DPa Pb;1 DPb UAF LMTD
ma Ra ln mb Rb ln ma cp;a ln 1 (28)
Pa;1 Pb;1 ma cp;a Ta;1
DPa DPb UAF LMTD
ma Ra ln 1 mb Rb ln 1 mb cp;b ln 1 =Cmax
Pa;1 Pb;1 mb cp;b Tb;1
(24b)
Non-dimensioned entropy generation related to fluid fric-
where U, A, F, LMTD, DPa, and DPb are calculated according tion:
to Sect. 2.1 and 2.2. Entropy generation is non-dimensiona-
Pa;2 Pb;2
lized through dividing by Cmax, where Cmax is the greatest mcp Ns;DP ma Ra ln mb Rb ln =Cmax
Pa;1 Pb;1
among two streams, thus it follows:
DPa DPb
ma Ra ln 1 mb Rb ln 1 =Cmax
S Pa;1 Pb;1
Ns (25)
Cmax (29)
5 Cost Analysis
4 Multi-Objective Optimization
Cost analysis was performed based on the estimation method
Multi-objective optimization problems usually exhibit a prob- of Muralikrishna and Shenoy [21]. In this formulation, the
ably uncountable set of solutions to assess the status of vectors cost of the heat exchanger is divided into capital and opera-
showing the best possible trade-offs in the objective function tional costs as follows:
space [40]. Pareto optimality is the key concept to express the Total cost = capital cost (heat exchanger core + pump a +
relationship between multi-objective optimization results in pump b) + operating cost of (pump a + pump b).
order to determine a solution which is actually one of the best
n
possible trades-offs [41, 42]. In multi-objective optimization, a Cost Af Ca Cb AcHT Ce Cf Wa d
process of decision-making is required for selection of the final
o C h (30)
optimal solution from available solutions of the Pareto fron- pow
Ce Cf Wb d Wa Wb
tier. In this paper, the LINMAP decision-making method was g
employed [43, 44]. In this method, the solution with mini-
mum distance from an imaginary solution, called the ideal where W, AHT, Ci (i {a, b, e, f, pow}, and d are pumping
point, was selected as the final desired optimal solution [44]. power, heat transfer area, and constant variables, respectively.
Details of MOEA (multiobjective evolu-
tionary algorithm) and LINMAP methods
are described in [44].
www.cet-journal.com 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eng. Technol. 2014, 37, No. 1, 8794
Entropy generation minimization 91
These parameters were calculated by thermal and hydraulic tance from the ideal point might be considered a desirable
analysis. The annual operating time of the systems was as- final solution. Since the axes in Fig. 3 did not have a similar
sumed to be 8000 h. Other coefficients and parameters in scale, it was impossible to compare the values corresponding
Eq. (33) are defined in [21] and presented in Tab. 3. to each value (i.e., Ns,DT and Ns,DP). Thus, both Ns,DT and Ns,DP
had to be normalized to become comparable. It follows:
Table 3. Cost coefficiency of a heat exchanger [21].
Ns;DT i
Parameter Value
n
Ns;DT i q P
(31)
2 i
Ns;DT
2
Af [m ] 0.322
Ca [US $] 30 000 Ns;DP i
n
Ns;DP i q
P 2 (32)
Cb 750 Ns;DP i
c 0.8
where i and n are indices for each individual solution on the
Ce 2000 Pareto frontier and normalized value, respectively. The nor-
Cf 5 malized Pareto frontier is demonstrated in Fig. 4. In the next
step, the closest point of the normalized Pareto frontier to the
d 0.68
1
Cpow [US $ Wh ] 0.00005
g 0.7
Chem. Eng. Technol. 2014, 37, No. 1, 8794 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
92 H. Sayyaadi et al
ideal solution was selected as a desirable final optimal solution. The main difference between the present approach and [33]
The procedure is schematically illustrated in Fig. 4. and [45] was related to the separation of entropy generation
The non-dimensioned Pareto frontier was curve-fitted using due to temperature difference and pressure losses in the cur-
a polynomial. The following expression denotes Ns,DP as a rent work. Values of Ns,DT and Ns,DP for the final optimal solu-
function of Ns,DT and Ns,DP for all optimal solutions located on tion were 0.017972 and 0.028655, respectively. Therefore, val-
the Pareto frontier: ues of thermal efficiency and pumping power were evaluated.
5 These values for the final optimal solution were 0.9161 and
n
Ns;DP 21:377Ns;DT
n
6 72:015 Ns;DTn 7.869 kW, respectively. As already explained, Ns,DT and Ns,DP
4 3 were separated and considered as two distinct objective func-
96:133 Ns;DT
n n
65:673 Ns;DT (33) tions. However, in [33] and [44], Ns (= Ns,DT + Ns,DP) was tak-
2 en as a single objective function. Comparison of the results
25:397 Ns;DT
n n
6:2065Ns;DT 0:9987 obtained in the present approach, in [33] and in [45] showed
that the heat transfer area of the optimized heat exchanger was
a little greater than the corresponding heat transfer area in
If Ns,DT was desired as a function of Ns,DP, then:
[33] but a little smaller than the corresponding heat transfer
5 area in [45]. The total non-dimensioned entropy generation
n
Ns;DT 15:370Ns;DP
n
6 54:776 Ns;DPn
for the optimal heat exchanger of this study was significantly
4 3 less than the corresponding total entropy generation of a simi-
78:386 Ns;DP
n n
58:145 Ns;DP (34) lar PFHE obtained in [33] and [44] (26.4 % less than the cor-
2 responding Ns in [33] and 12.1 % less than the corresponding
24:490 Ns;DP
n n
6:3477Ns;DP 1:0225 Ns in [45]).
Further pressure drop, pumping power, and operating cost
Quantities of decision variables, objective functions, and key of the optimal PFHE of the current study were much lower
parameters of the PFHE for the final solution are presented in than the corresponding values in [33] and [45]. Pumping
Tab. 4, in which they are compared with the corresponding power and operating cost of the PFHE in this study were
results obtained in the literature by other researchers [33, 45] 31.0 % and 32.9 % less than the corresponding pumping power
using different approaches. in [33] and [45], respectively (similar results valid for the
The same constraints and decision variables which were con- operating cost). The capital cost of the heat exchanger surface
sidered in [33] and [45] were taken into account in this paper. was almost similar for all cases; however, the total cost ob-
Table 4. Decision variables, objective function, and parameters for the final optimal solution.
www.cet-journal.com 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eng. Technol. 2014, 37, No. 1, 8794
Entropy generation minimization 93
tained in this work was 10.0 % and 12.1 % lower than the cor- Q [W] rate of heat transfer
responding total cost of PFHEs introduced in [33] and [45], R [J kg1K1] specific gas constant
respectively. Re [] Reynolds number
Tab. 4 indicates that the approach of the current paper with s [m] fin spacing (1/n1)
separation of the thermal entropy generation from the hydrau- S [W K1] rate of entropy generation
lic entropy generation and their consideration as two separate St [] Stanton number (h/GCp)
objective functions led to improvement in most factors of the t [m] fin thickness
PFHE compared to traditional approaches in [33] and [45]. T [K] temperature
However, the cost was not an objective of this research. Eco- U [W m2K1] overall heat transfer
nomic features of the optimized PFHE using the presented W [W] pumping power
approach were better than the corresponding economic pa-
rameters in [33] and [45]. Greek symbols
q [kg m3] density
l [N m2s1] viscosity
7 Conclusions
A multi-objective optimization of a cross-flow PFHE was pre- References
sented based on EGM. It was found that both thermal entropy
generation and pressure entropy generation of the cross-flow
[1] B. Kim, B. Sohn, Int. J. Heat Fluid Flow 2006, 27, 514521.
PFHE were in optimum state. Therefore, multi-objective opti-
[2] D. A. Reay, Heat Recovery Syst. CHP 1994, 14, 459474.
mization of this type of heat exchangers led to the best trade-
[3] A. Bejan, Int. J. Heat Mass Transfer 1978, 21, 655658.
off between entropy generation related to heat transfer and en-
[4] A. Bejan, Entropy Generation through Heat and Fluid Flow,
tropy generation related to fluid friction. It was shown that
Wiley, New York 1982.
this approach, in terms of separating entropy generation,
[5] V. den Buick, J. Heat Transfer 1991, 113, 341347.
enabled a better design of the PFHE with less total entropy
[6] K. R. Rao, Ph. D. Thesis, Indian Institute of Science, Banga-
generation, lower pressure drop and pumping power, less
lore 1991.
operating cost, and less total cost compared to the approach
[7] G. Venkatrathnam, Ph. D. Thesis, Indian Institute of Technol-
that considered total entropy generation as an objective func-
ogy, Kharagpur 1991.
tion of the single-objective optimization.
[8] B. Abramzon, S. Ostersetzer, in Proc. of the 1st Int. Conf. on
Aerospace Heat Exchanger Technology (Ed: A. Hashemi), Else-
The authors have declared no conflict of interest.
vier, New York 1993, 349368.
[9] J. E. Hesselgreaves, in Proc. of the 1st Int. Conf. on Aerospace
Heat Exchanger Technology (Ed: A. Hashemi), Elsevier, New
Symbols used York 1993, 391399.
[10] M. T. Gonzalez, N. C. Petracci, M. Urbican, Heat Transfer
A [m2] heat transfer area
Eng. 2001, 22, 1116.
Aff [m2] free flow area
[11] R. Selbas, O. Kizilkan, M. Reppich, Chem. Eng. Process. 2006,
AHT [m2] heat transfer area
4, 268275.
C [W K1] heat capacity rate
[12] B. Allen, L. Gosselin, Int. J. Energy Res. 2008, 10, 958969.
cp [W kg1K1] specific heat of fluid
[13] H. Peng, X. Ling, Appl. Therm. Eng. 2008, 5, 642650.
Dh [m] hydraulic diameter
[14] P. Wildi-Tremblay, L. Gosselin, Int. J. Energy Res. 2007, 9,
f [] fanning friction factor, correction
867885.
factor
[15] B. V. Babu, S. A. Munawar, Chem. Eng. Sci. 2007, 14, 3720
G [kg m2s1] mass flux velocity
3739.
h [W m2K1] heat transfer coefficient
[16] A. C. Caputo, P. M. Pelagagge, P. Salini, Appl. Therm. Eng.
h [m] inner height of the fin
2008, 10, 11511159.
H [m] height or outer height of the fin
[17] Y. Ozcelik, Appl. Therm. Eng. 2007, 11, 18491856.
j [] Colburn factor
[18] L. Valdevit, A. Pantano, H. A. Stone, A. G. Evans, Int. J. Heat
l [m] lance length of the fin
Mass Transfer 2006, 21, 38193830.
L [m] heat exchanger length
[19] G. N. Xie, B. Sunden, Q. W. Wang, Appl. Therm. Eng. 2008,
LMTD [] logarithmic mean temperature
8, 895906.
difference
[20] I. Ozkol, G. Komurgoz, Numer. Heat Transfer, Part A 2005, 3,
m [kg s1] mass flow rate of fluid
283296.
n [fins/m] fin frequency
[21] K. Muralikrishna, U. V. Shenoy, Chem. Eng. Res. Des. 2000,
Na, Nb [] number of fin layers of fluid a or b
58, 161167.
Ns [] dimensionless entropy generation
[22] S. Sanaye, H. Hajabdollahi, Appl. Energy 2010, 87, 1893
P [N m2] pressure
1902.
DP [N m2] pressure drop
Pr [] Prandtl number
Chem. Eng. Technol. 2014, 37, No. 1, 8794 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
94 H. Sayyaadi et al
[23] P. Ahmadi, H. Hajabdollahi, I. Dincer, J. Heat Transfer 2011, [34] The Standards of the Brazed Aluminum Plate-Fin Heat
133, 021801021810. Exchanger Manufacturers Association, ALPEMA, 2000.
[24] H. Najafi, B. Najafi, Appl. Therm. Eng. 2011, 31, 18391847. www.alpema.org
[25] R. Venkata Rao, V. Patel, Appl. Math. Model. 2013, 37, 1147 [35] H. M. Joshi, R. L. Webb, Int. J. Heat Mass Transfer 1987, 30,
1162. 6984.
[26] A. Bejan, Entropy Generation Minimization, CRC Press, Boca [36] W. Wang, S. Zhang, J. Yang, J. Zheng, X. Ding, M. Chou,
Raton, FL 1948. P. Tang, X. Zhan, Chem. Eng. Technol. 2013, 36 (4), 657664.
[27] R. T. Ogulata, F. Doba, Int. J. Heat Mass Transfer 1998, 41, [37] F. Bosnjakovic, Arch. Wrmewirtsch. Dampfkesselwes. 1938,
373381. 19 (1), 12 (in German).
[28] J. Vargas, A. Bejan, Int. J. Heat Fluid Flow 2001, 22, 657665. [38] F. Bosnjakovic, Technische Mitteilungen, Essen 1939, 32 (15),
[29] R. Culham, Y. Muzychka, IEEE Trans. Compon. Packag. 439445 (in German).
Technol. 2001, 24, 159165. [39] A. Bejan, J. Heat Transfer 1977, 99, 374380.
[30] M. Iyengar, IEEE Trans. Compon. Packag. Technol. 2003, 26, [40] Technical Assessment Guide (TAGTM), Electric Power Research
6270. Institute, Palo Alto, CA 1993.
[31] D. Jainender, Design of Compact Plate Fin Heat Exchanger, [41] C. M. Fonseca, P. J. Fleming, in Handbook of Evolutionary
National Institute of Technology, Rourkela, India 2009. Computation (Eds: T. Back, D. B. Fogel, Z. Michalewicz),
[32] M. A. Taylor, Plate-Fin Heat Exchangers Guide to their Oxford University Press, New York 1997.
Specification and Use, Heat Transfer and Fluid Flow Service, [42] D. A. Van Veldhuizen, G. B. Lamont, Evolutionary Computa-
ORA, UK 1987. tion 2000, 2, 125147.
[33] M. Mishra, P. K. Das, S. Sarangi, Appl. Therm. Eng. 2009, 29, [43] J. H. Holland, Adaptation in Natural and Artificial Systems,
29832989. University of Michigan Press, Ann Arbor, MI 1975.
[44] H. Sayyaadi, R. Mehrabipour, Energy 2012, 38, 362375.
[45] R. V. Rao, V. K. Patel, Int. J. Therm. Sci. 2010, 49, 17121721.
www.cet-journal.com 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eng. Technol. 2014, 37, No. 1, 8794