Professional Documents
Culture Documents
Chevron
Chevron
Chevron
net/publication/270472864
CITATIONS READS
10 1,004
3 authors, including:
All content following this page was uploaded by Hassan Hajabdollahi on 10 August 2016.
F
Farzaneh Hajabdollahi,1 Zahra Hajabdollahi,2
& Hassan Hajabdollahi,3,*
O
1
Mechanical Engineering Department, Ferdowsi University of Mashhad,
Mashhad, Iran
O
2
Mechanical Engineering Department, Shahid Bahonar University of Kerman,
Kerman, Iran
3 PR
Mechanical Engineering Department, Vali‐e‐Asr University of Rafsanjan,
Rafsanjan, Iran
∗Address all correspondence to H. Hajabdollahi E-mail: Hajabdollahi@iust.ac.ir
Thermal modeling and optimal design of a gasket plate heat exchanger are presented in this
R
paper. The method is applied to estimate the heat exchanger pressure drop and the effectiveness
of chevron plates. Corrugation angle, amplitude, and wavelength of chevron corrugation, plate
width, plate length, and the number of plates are considered as six design parameters. The Fast
O
and Elitist Non‐Dominated Sorting Genetic Algorithm (NSGA‐II) is applied to obtain the maxi‐
mum effectiveness and the minimum total annual cost (sum of investment and operation costs)
as two objective functions. The results of optimal designs are a set of multiple optimum solu‐
tions called ʺPareto optimal solutionsʺ. The sensitivity analysis of change in optimum effective‐
TH
ness and total annual cost with change in design parameters of the gasket plate heat exchanger
is also performed and the results are reported. It is also investigated that effectiveness is propor‐
tional to the nondimensional parameter of corrugation amplitude to wavelength in the optimum
situation.
KEY WORDS: gasket plate heat exchanger, effectiveness, total annual cost, objective
AU
1. INTRODUCTION
A gasket plate heat exchanger (GPHE) as shown in Fig. 1 is a typical heat ex-
changer that is widely used in many industrial power generation plants, chemical,
petrochemical, and petroleum industries (Shah and Sekulic, 2003). There are many
studies in design and optimization of various heat exchangers. Hajabdollahi et al.
(2011[Q1]) applied the genetic algorithm (GA) for optimization of compact heat
NOMENCLATURE
F
Cmin minimum of Ch and Cc, W/K W⋅m2/K
Cmax maximum of Ch and Cc, W/K Vt volumetric flow rate, m3/s
O
C∗ heat capacity rate ratio, Cmin/Cmax W plate width, m
Cinv investment cost, $ y depreciation time, year
Cope annual cost of operation, $/year Greek symbols
O
Ctotal total annual cost, $/year β corrugation angle, deg
Dh hydraulic diameter, m ε thermal effectiveness
f
G
h
kel
fanning friction factor
mass flux, kg/m2⋅s PR
heat transfer coefficient, W⋅m2/K
price of electrical energy, $ MW⋅h–1
η
Λ
μ
pump efficiency
chevron corrugation wavelength, mm
viscosity, Pa⋅s
ΔP pressure drop, Pa
kw plate conductivity, W/m⋅K ρ density, kg/m3
N number of chevron plates τ hours of operation per year
R
NTU number of transfer units Subscripts
Nu Nusselt number c cold
O
exchangers including the plate fin and rotary regenerator (Hajabdollahi et al.,
2011a,b; Sanaye and Hajabdollahi, 2009; Ahmadi et al., 2010). Hilbert et al.
AU
F
O
O
PR
FIG. 1: Schematic of gasket plate heat exchanger with chevron corrugated plate
as well as minimum exchanger weight and pressure loss. Gholap and Khan (2007)
also studied air cooled heat exchangers by minimizing the energy consumption by
fans and material cost as two objective functions.
TH
• Proposing a closed form equation for the total annual cost in terms of the ef-
fectiveness at the optimal design point.
• Performing sensitivity analysis of the change in the objective functions on
variation of the optimum design parameters.
2. THERMAL MODELING
The ε−NTU method is applied here for predicting the heat exchanger performance.
F
The effectiveness of the counterflow heat exchanger is calculated as follows (Shah
and Sekulic, 2003):
O
1 − e− NTU (1−C
*
)
ε= . (1)
1 − C *e −NTU (1−C
*
)
O
The number of transfer units (NTU) and heat capacity ratio (C ∗) are defined as
follows (Shah and Sekulic, 2003):
PR
NTU max =
UAtot
C min
, (2)
1 (4)
U= .
1 1 t
+ + R f ,c + R f ,h +
TH
hc hh kp
The quantity h is the heat transfer coefficient and Rf is the fouling resistance fac-
tor. Moreover, t and kp are the plate thickness and plate conductivity, respectively.
Many different correlations for the heat transfer and pressure drop in chevron
AU
plate (Fig. 1) heat exchangers have been reported in the literature. The Fanning
factor f and Nusselt number Nu (representatives of the pressure drop and of its
thermal performances, respectively), defined in Eqs. (5) and (6) by Sunders ([Q2])
and Schlunder (1998) for chevron plate cores are given as (Fig. 1)
c1 (5)
f = ,
Re c 2
hDh μ
Nu = = 0.205 Pr1/3( m )1/6 ( f . Re 2 sin 2β) 0.374 , (6)
k μw
F
,
μ
O
4a
Dh = , (8)
Φ
( )
O
1
Φ= 1 + 1 + X 2 + 4 1 + 0.5 X 2 , (9)
6
2πa
m
.
Λ
, (10)
G = , (11)
2aW
where a and Λ are the amplitude and wavelength of chevron corrugations, respec-
R
tively. Moreover, W is the plate width between the gaskets.
Furthermore, the total pressure drop by neglecting the port pressure drop is also
estimated as (Sunders, [Q2])
O
L p N p μ m −0.17
Δ P = 2 ρfG 2 ( ) , (12)
Dh μw
TH
where Np is the number of passes and Lp is the total plate length as shown in
Fig. 1.
This was later modified by Deb et al. (2002) who eliminated the higher computa-
tional complexity, lack of elitism, and the need for specifying the sharing parame-
ter. This algorithm is called NSGA-II which is coupled with the objective
functions developed in this study for optimization.
F
viduals, a procedure which imitates survival of the fittest in nature.
O
To preserve diversity, the influence of elitism is controlled by choosing the num-
ber of individuals from each subpopulation, according to the geometric distribution
(Deb and Goel, 2001),
O
1 − c q−1
Sq = S c , (13)
1 − cw
PR
to form a parent search population, Pt+1 (t denotes the generation), of size S,
where 0 < c < 1, and w is the total number of ranked non-dominated.
Uniform crossover and random uniform mutation are employed to obtain the off-
TH
spring population, Qt+1. The integer-based uniform crossover operator takes two
distinct parent individuals and interchanges each corresponding binary bits with a
probability, 0 < pc ≤ 1. Following crossover, the mutation operator changes each
of the binary bits with a mutation probability, 0 < pm < 0.5.
AU
In this study, the effectiveness and total annual cost are considered as two objec-
tive functions. The total annual cost is included with investment cost (the annual-
ized cost of the heat transfer surface area) and operating cost of a pump for
forcing the fluid flow (Hajabdollahhi et al., 2011a)
Ctotal = af × Cinv + Cope , (14)
⎛ Δ PVt ⎞ ⎛ Δ PVt ⎞
Cope = ⎜ kel τ ⎟ + ⎜ kel τ ⎟ . (16)
⎝ η ⎠c ⎝ η ⎠h
Here CA and kel are the heat exchanger investment cost per unit surface area and
the electricity unit cost, respectively; n is a constant, and τ is the operation hours
of the exchanger per year. The quantities ΔP, Vt, and η are the pressure drop, vol-
ume flow rate, and pump efficiency, respectively. Also af is the annualized factor
F
defined as
r
af = , (17)
1 − (1 + r)−y
O
where r and y are the interest rate and depreciation time, respectively. In this
study, the corrugation angle (β), chevron corrugation amplitude (a), chevron corru-
O
gation wavelength (Λ), plate width, (W), plate length (Lh), and the number of
plates (N) are considered as six design parameters.
5. CASE STUDY
PR
The hot oil with 25 kg/s mass flow rate passes through the heat exchanger at
90oC. The fresh water (cold stream) with 34 kg/s mass flow rate passes through
the exchanger at 30oC. The GPHE is made from stainless steel with the thermal
conductivity kw = 17.5 W/m⋅K. The operating conditions and the cost function
constant values are listed in Table 1. The thermophysical properties of water and
oil such as the Prandtl number, viscosity, and specific heat are considered as tem-
R
perature dependent.
TABLE 1: The operating conditions of the GPHE (input data for the case studied)
O
F
ference percentage points of the two mentioned modeling output results are within
±10%, which is acceptable for engineering problems.
O
6.2 Optimization Results
To maximize the effectiveness value and to minimize the total annual cost, six de-
sign parameters including the corrugation angle, amplitude and wavelength of
O
chevron corrugation, plate width, plate length, and the number of plates were se-
lected. The design parameters (decision variables) and the range of their variations
are listed in Table 4. The system is optimized for the depreciation time y = 10
kg/s 140
Mass flow rate of cold water kg/s 140
Inlet hot temperature o 22
C
R
Inlet cold temperature o 22
C
Chevron angle deg 45
O
Number of pass 1
Number of plates 105
TH
TABLE 3: Comparison of modeling output and the corresponding results from (Kakac and Liu,
2002)
AU
Output
Unit (Schaffer, 1985) Present paper Difference (%)
variables
fh 0.204 0.2187 7.216
fc 0.221 0.2363 6.926
2
hh W/m ⋅K 32844 32463 5.46
2
hc W/m ⋅K 27,811 30,370 9.2
Δptotal kPa 592.4 554.74 –6.357
F
O
O
PR
FIG. 2: The distribution of Pareto-optimal points solutions using NSGA-II
R
TABLE 4: The design parameters, their range of variation, and their change step
years, the interest rate r = 0.1, and five standard chevron plates with [30; 45; 50;
60; and 65] chevron angle. The genetic algorithm optimization is performed for
400 generations, using a search population size of M = 150 individuals, crossover
probability of pc = 0.9, gene mutation probability of pm = 0.035, and controlled
elitism value c = 0.55. The results for the Pareto optimal curve are shown in Fig.
2, which clearly reveals the conflict between two objectives, the effectiveness and
the total annual cost. Any geometrical change that increases the effectiveness or
heat transfer rate (ε = q ⁄ qmax) leads to an increase in the total annual cost and vice
versa. This shows the need for the multiobjective optimization techniques in the
optimal design of a GPHE. It is shown in Fig. 2 that the maximum effectiveness
exists at design point A (0.9469), while the total annual cost is the biggest at this
point. On the other hand, the minimum total annual cost occurs at design point E
(317.5 $/year), with the smallest effectiveness value (0.2071) at that point. Design
point A is the optimal situation at which the effectiveness is a single objective
function, while design point E is the optimum condition at which the total annual
F
cost is a single objective function.
The optimum values of two objectives for five typical points from A to E
(Pareto optimal fronts) for input values given in Table 1 are listed in Table 5.
O
To provide a useful tool for the optimal design of the GPHE, the following
equation for the effectiveness versus the total annual cost is derived for the Pareto
curve (Fig. 2):
O
−518.1ε 2 − 390.8 ε + 163.3
Ctotal ($ / year ) = × 2000 . (18)
ε 3 + 518.8ε 2 − 2432ε + 1902
PR
Equation (18) is valid in the range 0.2071 < ε < 0.9469 for effectiveness. The
interesting point in Eq. (18) is that considering a numerical value for the effective-
ness in the mentioned range provides the minimum total annual cost for that opti-
mal point along with other optimal design parameters.
The selection of final solution among the optimum points existing on the Pareto
front needs a process of decision-making. In fact, this process is mostly carried
R
out based on engineering experiences and importance of each objective for deci-
sion makers. In this paper, based on the information provided for designers (the
practical effectiveness values in the range 0.60 < ε < 0.81), the design points (C–
O
dotted lines. The following points could be deduced for the optimal variables in
Fig. 3:
1. The corrugation wavelength, plate width, and plate length have the values
distributed equally over its whole allowable domain.
2. The numerical values of the corrugation amplitude and number of plates are
AU
TABLE 5: The optimum values of effectiveness and the total annual cost for the design points
A to E in Pareto-optimal fronts for input values given in Table 1
A B C D E
Effectiveness 0.9469 0.8914 0.8147 0.6011 0.2071
Total annual cost ($/year) 2156 1360 1040 670.8 317.5
3. The numerical values of the corrugation angle are at their maximum level.
Since the optimum values of corrugation wavelength, plate width, and plate
length have scattered distribution in their whole allowable domains, one may pre-
dict that these three parameters have important effects on the conflict between the
higher values of effectiveness and lower amounts of total annual cost. The corru-
gation angle which is situated at its maximum value shows that this parameter has
no effect on the conflict between the two objective functions. Moreover, the cor-
F
rugation amplitude and the number of plates have the moderate effect on the con-
flict between the two objective functions.
The variation of optimum value of effectiveness with the total annual cost for
O
various values of optimum design parameters in the A–E cases (Pareto front) are
shown in Figs. 4a–f.
It is observed that the variation of two objective functions at other points on
O
Pareto optimal front has the same trend as the five points (A–E).
The effect of design variables on objective functions are investigated and ex-
plained as follows:
Corrugation wavelength PR
By increasing the corrugation wavelength (Λ), both the effectiveness (ε) and total
annual cost (Ctotal) decrease for all design points A–E (Fig. 4a). The variation of
the corrugation wavelength causes a conflict between two objective functions.
Therefore, each selection of fin pitch in the considered range of variation repre-
R
sents one point on the Pareto optimal front. The scattered distribution of the cor-
rugation wavelength in Fig. 3a approves this point too.
Corrugation amplitude and number of plates
O
With increasing the corrugation amplitude (a) (or plate number, N), both objec-
tives improved simultaneously (Figs. 4b and 4f). But there is a region in each de-
TH
sign that causes the conflicting behavior in the objective functions. Due to this
fact, the corrugation amplitude values (or plate number) obtained on the Pareto
optimal front have scattered distribution on their maximum values as is shown in
Figs. 3b and 3f.
Plate width
AU
By increasing the plate width (W), both the effectiveness and total annual cost
generally increase (Fig. 4c). The variation of the plate width causes a conflict be-
tween the two objective functions. Therefore, each selection of fin pitch in the
considered range of variation represents one point on the Pareto optimal front. The
scattered distribution of plate width in Fig. 3c approves this point too.
Plate length
By increasing the plate length Lh, both the effectiveness and total annual cost in-
crease for all design points A–E (Fig. 4d). Actually the effect of plate length is
F
O
O
PR
R
O
TH
AU
FIG. 3: Scattering of variables for the Pareto optimal front: a) corrugation wavelength; b)
corrugation amplitude; c) plate width; d) plate length; e) corrugation angle; f) number of
plates
F
O
O
PR
R
O
TH
AU
FIG. 3: (continued) Scattering of variables for the Pareto optimal front: a) corrugation
wavelength; b) corrugation amplitude; c) plate width; d) plate length; e) corrugation angle;
f) number of plates
F
O
O
PR
R
O
TH
AU
FIG. 4: Variation of effectiveness with annual cost for six optimum design parameters in
five A–E cases: a) corrugation wavelength; b) corrugation amplitude; c) plate width; d)
plate length, e) corrugation angle; f) number of plates
F
O
O
PR
R
O
TH
AU
FIG. 4: (continued) Variation of effectiveness with annual cost for six optimum design
parameters in five A–E cases: a) corrugation wavelength; b) corrugation amplitude; c) plate
width; d) plate length, e) corrugation angle; f) number of plates
the reverse of the corrugation wavelength but both of them cause conflicting of
objective functions. As a result, the plate length such as the corrugation wave-
length has the scattering distribution on the whole allowable domain as depicted in
Fig. 3d.
Corrugation angle
By increasing the corrugation angle, the effectiveness increases while the total an-
nual cost is approximately constant (Fig. 4e). As a result, the maximum of corru-
F
gation angle is suitable. The distribution of corrugation angle around its maximum
value verifies this point (Fig. 3e).
O
6.3 Effect of Two Nondimensional Parameters
The values of effectiveness versus the two nondimentional parameters (a/Λ and
O
Lh/W) for the Pareto optimal front are shown in Figs. 5 and 6, respectively. It is
seen from Fig. 5 that the effectiveness generally increases on increase of a/Λ.
There are two values of the effectiveness for any choice of the Lh/W parameter in
PR
the range 0.6–1.15. Some of them are with a high effectiveness, while the other
lead to a low value of effectiveness (Fig. 6). In addition, there is any effectiveness
in the range 0.25–0.9 for any choice of the Lh/W parameter in the range 0.38–0.6.
It is worth mentioning that the variation of the total annual cost is similar to the
effectiveness as shown on the Pareto front of Fig. 2.
R
O
TH
AU
FIG. 5: The value of effectiveness versus the ratio of corrugation amplitude and wave-
length for the Pareto optimal front
F
O
O
PR
FIG. 6: The value of effectiveness versus the ratio of plate length and plate width for the
Pareto optimal front
R
7. CONCLUSIONS
O
A gasket plate heat exchanger was optimally designed using the multiobjective op-
timization technique. The design parameters (decision variables) were corrugation
angle, amplitude and wavelength of chevron corrugation, plate width, plate length,
TH
and the number of plates. In the optimization problem presented, the effectiveness
and total annual cost were two objective functions (the effectiveness was maxi-
mized and the total annual cost was minimized). A set of the Pareto optimal
front points was shown. The results revealed the level of conflict between the
two objectives. The corrugation wavelength, plate width, and the plate length
AU
REFERENCES
Ahmadi, P., Hajabdollahi, H., and Dincer, I., Cost and entropy generation minimization of a cross
flow plate-fin heat exchanger (PFHE) using multi-objective genetic algorithm, ASME J. Heat
Transfer, DOI: 10.1115/1.4002599, 2010.
Deb, K. and Goel, T., Controlled elitist non-dominated sorting genetic algorithms for better conver-
gence, in: Proc. First Int. Conf. on Evolutionary Multi-Criterion Optimization, Zurich, pp. 385–
399, 2001.
F
Deb, K., Multi-Objective Optimization Using Evolutionary Algorithms, Chichester: John Wiley and
Sons Ltd., 2001.
Deb, K., Pratap, A., Agarwal, S., and Meyarivan, T., A fast and elitist multi-objective genetic algo-
O
rithm: NSGA-II, IEEE Trans. Evolut. Comput., vol. 6, no. 2, pp. 182–197, 2002.
Doodman, A. R., Fesanghary, M., and Hosseini, R., A robust stochastic approach for design optimi-
zation of air cooled heat exchangers, Appl. Energy, vol. 86, pp. 1240–1245, 2009.
O
Foli, K., Okabe, T., Olhofer, M., Jin, Y., and Sendhoff, B., Optimization of micro heat exchanger:
CFD, analytical approach and multi-objective evolutionary algorithms, Int. J. Heat Mass Transfer,
vol. 49, pp. 1090–1099, 2006.
PR
Gholap, A. K. and Khan, J. A., Design and multi-objective optimization of heat exchangers for re-
frigerators, Appl. Energy, vol. 84, pp. 1226–1239, 2007.
Goldberg, D. E., Genetic Algorithms in Search, Optimization and Machine Learning, Reading, MA:
Addison-Wesley, 1989.
Guo, J. F., Xu, M., and Cheng, L., The application of field synergy number in shell-and-tube heat
exchanger optimization design, Appl. Energy, vol. 86, no. 10, pp. 2079–2087, 2009.
Hajabdollahi, H., Ahmadi, P., and Dincer, I., Thermoeconomic optimization of a shell and tube con-
R
denser using evolutionary algorithm, Int. J. Refrig., vol. 34, pp. 1066–1076, 2011a.
Hajabdollahi, H., Tahani, M., and Shojaee Fard, M. H., CFD modeling and multi-objective optimiza-
tion of compact heat exchanger using CAN method, Appl. Thermal Eng., vol. 31, pp. 2597–2604,
O
2011b.
Hilbert, R., Janiga, G., Baron, R., and Thevenin, D., Multi-objective shape optimization of a heat ex-
changer using parallel genetic algorithms, Int. J. Heat Mass Transfer, vol. 49, pp. 2567–2577,
TH
2006.
Kakac, S. and Liu, H., Heat Exchangers: Selection, Rating and Thermal Design, Boca Raton, FL:
CRC Press, 2002.
Liu, Z. and Cheng, H., Multi-objective optimization design analysis of primary surface recuperator
for microturbines, Appl. Thermal Eng., vol. 28, pp. 601–610, 2008.
AU
Sahin, B., Yakut, K., Kotcioglu, I., and Celik, C., Optimum design parameters of a heat exchanger,
Appl. Energy, vol. 82, pp. 90–106, 2005.
Sanaye, S. and Hajabdollahi, H., Multi-objective optimization of rotary regenerator using genetic al-
gorithm, Int. J. Thermal Sci., vol. 48, pp. 1967–1977, 2009.
Schaffer, J. D., Multiple objective optimization with vector evaluated genetic algorithms, in: Proc.
Int. Conf. on Genetic Algorithms and Their Applications, [Q3], 1985.
Schlunder, E. U., Analogy between heat and momentum transfer, Chem. Eng. Process., vol. 37, pp.
103–107, 1998.
Shah, R. K. and Sekulic, P., Fundamental of Heat Exchanger Design, [Q4], John Wiley & Sons,
Inc., 2003.
Srinivas, N. and Deb, K., Multi-objective optimization using non-dominated sorting in genetic algo-
rithms, J. Evolut. Comput., vol. 2, no. 3, pp. 221–248, 1994.
Sunders, E. A. D., Heat Exchangers: Selection Design and Construction, New York: Wiley, [Q2].
Wang, J. J., Jing, Y. Y., and Zhang, C. F., Optimization of capacity and operation for CCHP system
by genetic algorithm, Appl. Energy, vol. 87, no. 4, pp. 1325–1335, 2010.
Xie, G. N., Sunden, B., and Wang, Q. W., Optimization of compact heat exchangers by a genetic
algorithm, Appl. Thermal Eng., vol. 28, pp. 895–906, 2008.
F
[Q1] AU: Please indicate which one: 2011a or 2011b?
O
[Q2] AU: Please indicate year
O
[Q3] AU: Please give the city, dates, volume, and page range
PR
R
O
TH
AU