Anda di halaman 1dari 5

2015 1st International conference on futuristic trend in computational analysis and knowledge management

(ABLAZE 2015)

Multi-objective Optimization of two stage


series connected Thermo-electric generator
using genetic algorithm
Ranjana Hans#1, S.C. Kaushik*1, S. Manikandan*2
#

Renewable Energy Department,


Amity University Haryana,
Gurgaon-122413, India
*
Centre for Energy Studies, IIT Delhi,
New Delhi-16
1
ranjana1219@rediffmail.com
noise generated in thermoelectric devices makes
them fit for versatile usage with no environmental
deprivation concerns [3-4]. Moreover, system
investigation and analysis is a key step while
designing a high performance thermoelectric
device. To investigate the performance of single
stage or multi-stage thermoelectric generators,
conventional non-equilibrium thermodynamics is
commonly used [5-9]. Multi-stage thermoelectric
generators give much improved performance with
range of advantages over the single stage
configuration. Optimization analysis of two stage
thermoelectric generator is done by Zhou [10].
Performance analysis and single objective
optimization of two stage thermoelectric generator
is done by Chen et al. [11]. Multi-objective
optimization of different thermal energy conversion
systems has been investigated by number of
researchers [12-13] and various research gaps are
found in this area. There is no similar study of
performance
analysis
and
multi-objective
optimization
of
two-stage
exo-reversible
thermoelectric generators.
In the proposed work, an optimum value of power
output (Pout) and thermal efficiency () has been
attained for two-stage thermoelectric generator by
using multi-objective genetic algorithm and Pareto
frontier along with Pareto optimal set of optimum
values has been attained in MATLAB environment.
After that Fuzzy Bellman-Zadeh decision making
method has been executed to extract the best
optimal values of above objectives for the proposed
system.

Abstract- Two-stage thermoelectric generator in exoreversible mode is thermodynamically optimized in


view of finite time thermodynamics (FTT) and multiobjective genetic algorithm (MOGA). Electrically
series configuration with only internal irreversibilities
of thermoelectric generator has been considered for
investigation in the current research. Power output
and thermal efficiency of the aforementioned system
are considered as dual objective functions obtained
from FTT technique. These two objectives are
maximized at the same time with the use of multi
objective genetic algorithm. For MOGA optimization
of thermoelectric generator, five decision variables as
number of thermoelectric element pairs at the bottom
and top stage as m and n, heat source temperature
and heat sink temperature and working electric
current has been considered in the current study.
MATLAB environment is used to obtain the Pareto
Optimal frontier between power output and thermal
efficiency and their best optimal values are selected
by Fuzzy Bellman-Zadeh decision making technique.
A comparative analysis of single objective and dual
objective optimization of aforementioned objectives
has been carried out and discussed
.
Keywords: Multi-objective optimization, MOGA, Finite
time thermodynamics, Two-stage semiconductor
thermoelectric
generator,
Non
equilibrium
thermodynamics.

I. INTRODUCTION
The thermoelectric technology is emerging as a
forthcoming area of research with the invention of
new fabrication techniques and materials [1] in
various application fields such as aerospace,
instrumentation
and
defense
etc.
The
thermoelectric generator based on See beck
principle provides range of advantages over
conventional power generators [2], such as longer
life span, no moving parts and requires least
maintenance and its good performance makes it fit
for wide range of applications in industrial,
commercial and domestic vicinity. In addition to it,
the absence of working fluid and lesser amount of
978-1-4799-8433-6/15/$31.002015IEEE

II. SYSTEM DESCRIPTION


Two-stage exo-reversible thermoelectric generator
in electrically series configuration is shown in
fig.1. It has m pairs of thermoelectric elements in
its bottom stage and n pairs of thermoelectric
elements in the top stage with M be the total
number of thermoelectric element pairs of

165

2015 1st International conference on futuristic trend in computational analysis and knowledge management
(ABLAZE 2015)

thermoelectric
generator,
M=m+n.
All
thermoelectric elements are composed of p-type
and n-type semiconductor legs. The system is
assumed to have only internal irreversibilities as
Conduction heat loss and Joulean electrical
resistive loss (I2R) between the hot and the cold
junctions. K (Th-Tm) is the heat loss by conduction
for the top stage and K (Tm-Tc) is the heat loss by
conduction for the bottom stage respectively. In the
proposed system, only internal irreversibilities are
being considered and system is implied to have no
interaction with the surroundings.

Qc1 = Qh 2
1

n 1Tm I + I 2 R1 + K1 (Th Tm ) =
2

1 2

2Tm I 2 I R1 + K 2 (Tm Tc ) m

(5)

Putting the values of I, , K and R in equation (5),


As same current is flowing in two cascaded stages
I1 = I 2 = I
(6)
Substituting from Eqn (6),
when See beck Coefficients, thermal conductivity
and Resistance of two stages can be equated as

if 1 = 2 ; K1 = K 2 ; R1 = R2 ; then

Tm [(m n) I + (n + m) K ] =

(7)

(m + n) I 2 R + nKTh + mKTc
Tm =

(8)

The power output of generator can be computed as


Pout = Qh1 Qc 2
(9)

Fig.1 Two stage electrically series connected TEG

III. THERMODYNAMIC ANALYSIS

for M=m+n
The voltage and working electric current of the first
stage can be calculated as
V1 = n1 (Th Tm ) nIR1
(10)

The two stage exo-reversible thermoelectric


generator equations with the use of Newtons law
at three junctions are

1
Qh1 = [1Th I I 2 R1 + K1 (Th Tm )]n
2
1
Qc1 = [1Tm I + I 2 R1 + K1 (Th Tm )]n
2
1
Qh 2 = [ 2Tm I I 2 R2 + K 2 (Tm Tc )]m
2
1
Qc 2 = [ 2Tc I + I 2 R2 + K 2 (Tm Tc )]m
2

(m + n)0.5I 2 R + nKTh + mKTc


(m n) I + (n + m) K

P1 = Qh1 Qc1 = n1I (Th Tm ) nI 2 R1

(1)

(11)

The voltage and working electric current of the first


stage can be calculated as
V2 = m 2 (Tm Tc ) mIR2
(12)

(2)

P2 = Qh 2 Qc 2 = m 2 I (Tm Tc ) mI 2 R2
(3)

(13)

The efficiency of thermoelectric generator is

(4)

Where 1 and 2 are the See beck coefficients for


thermoelectric generator elements and Qh is the rate
of heat flow between the two stages.
Thermodynamic modeling of the thermoelectric
generator has been done with following
assumptions
In the proposed system, only internal
irreversibilities are being considered and
system is implied to have no interaction
with the surroundings.
Non consideration of Thomson effect for
the analysis.
The
exo-reversible
thermoelectric
generator system which is having no
interaction with the surroundings.

Pout
Qh1

(14)

IV. MULTIOBJECTIVE OPTIMIZATION


Multi-objective optimization does simultaneous
optimization of two or more objective function for
carrying out maximization or minimization of
various performance parameters of the system,
which not the case in single objective optimization.
By using multi-objective genetic algorithm for
solving optimization problems, number of optimal
solutions can be obtained from Pareto optimal set
and out of those solutions the best one can be
evaluated with the help of appropriate decision
making technique. In the current study, MOGA is
applied on two stage series connected
thermoelectric heat pump while taking dual
objectives as power output and thermal efficiency.
Pareto frontier and Pareto optimal set of optimum
values has been obtained for the aforementioned

Now combining Eqn. (2) and (3) gives,

166

2015 1st International conference on futuristic trend in computational analysis and knowledge management
(ABLAZE 2015)

system and the final solution is obtained using


Fuzzy Bellman-Zadeh decision making technique.

Detailed comprehensive study has been presented


by Sayyadi et al. [13]. The fuzzy set or objective
function can be defined as

a) Objective function, decision variables and


constraints:
Power output (Pout) and thermal efficiency () of
the above mentioned system are taken as dual
objective functions as derived by equations [1-14]
and five decision variables as number of
thermoelectric element pairs at the bottom and top
stage as m and n, heat sink temperature (Tc) and
heat source temperature (Th) and working electric
current (I) has been selected for multi-objective
optimization process. The following constraints are
applied on decision variables as
0 I 15
(15)
0 n 30
(16)
0 m 30
(17)
(18)
200 Th 500
(19)
200 Tc 400
The constant parameters for the optimization are

Aj = X , Aj ( X ) , X L, j = 1, 2 k

Here

(X) are membership functions of matrix

A j which is denoted by Fj ( X ) in fuzzy. Final


decision is represented by the intersection of fuzzy
restrictions and criteria which is accessible by the
means of membership functions.
A fuzzy route D can be formed by the fuzzy
collection of membership functions as

D = kj =1 A j

D ( X ) = kj =1 A ( X ) = min A ( X ), X L
j

j =1 k

The maximum limit for the outcome to be verified


is

max D ( X ) = max min Aj ( X )

R= 2.88510
K= 0.02772W

/ km
6
= 42510 V / K

X L

j =1 k

X = arg max min Aj ( X )


0

j =1 k

X L

To achieve above equation output one has to make


membership functions A j ( X ) ; j=1,2--k in

The following multi-objective genetic algorithm


parameters are considered in the current analysis.
Parameter
Population type
Population Size
Selection
Generation
Cross over
Mutation function

Fj ( X )
X L; j = 1, 2 k .
The matrix table {T } has the diagonal

Type
Double Vector
100
Tournament
250
0.8
Constraint Dependent

elements as ideal data points.

F (X 0) F (X 0)
2
1
1 1
%
{T } = F1 ( X 20 )

0
F2 ( X n0 )
F1( X n )

b) Decision making methods in multi objective


optimization:

Fn ( X10 )
Fn ( X 20 )

Fn ( X n0 )

Where maximum and minimum limits can be


defined as

Fi min = min Fj ( X 0j ), i = 1, 2 n

The decision making procedure is used to get the


optimum value of the design parameters to placate
the accomplishing condition of multi-objective
optimization. It gives distinctive solution from
Pareto optimal set of all possible solutions. The
authors have applied Fuzzy Bellman-Zadeh
decision making technique in the current study.
The Bellman-Zadeh (1970) projected an approach
which defines final decision with the intersection of
fuzzy constraints and criteria by its membership
functions whose matrix is given below.
Membership matrix column comprise of fuzzy
membership function for every defined objective
and row incorporates membership function value
for every solution obtained on Pareto frontier.
Hence in membership matrix number of rows
reflects number of solutions of Pareto frontier and
number of objectives chosen of membership
functions is revealed by number of columns.

Fi

max

= max Fj ( X 0j ), i = 1, 2 n
j

V. RESULTS AND DISCUSSIONS


Fig. (2) shows the original Pareto frontier for dual
objective (power output and thermal efficiency)
optimization plotted using multi-objective genetic
algorithm technique along with various optimal
values of dual objectives of the TEG system. Fuzzy
Bellman-Zadeh decision making practice has
specified final optimization solution on Pareto
optimal set of values. The following III degree
polynomial is obtained while carrying out the
optimal design of thermoelectric heat generator
based on MOGA

167

2015 1st International conference on futuristic trend in computational analysis and knowledge management
(ABLAZE 2015)

= 1.8P3 + 14 P 2 36P + 31

Average Distance Between Individuals


6

Pareto front
0.0591

= - 1.8*P + 14*P - 36*P + 31

Avergae Distance

4
Ideal Solution

Efficiency ( )

0.059
Fuzzy Solution

Non-Ideal Solution
0.059

Optimal Values
cubic

0.0589
2.58

100

200

300

400
500
Generation

600

700

800

Fig 3 Average distance between individuals


2.585

2.59
2.595
2.6
Power Output, P (Watts)

2.605

2.61

Average Spread: 0.065216


1

Fig. 2 Original Pareto front and optimal solutions for dualobjective optimization

0.9
0.8

The ideal, non-ideal and Fuzzy solution obtained


through MOGA are mentioned in Table1

Average Spread

0.7

Table 1

0.6
0.5
0.4
0.3

Parameter
Power output
(W)
Thermal
efficiency

Ideal
Solution
2.607

Non-ideal
Solution
2.582

Fuzzy
Solution
2.598

0.0598

0.05899

0.0590

0.2
0.1
0

100

200

300

400
500
Generation

600

700

800

Fig 4 Average spread


Score Histogram
25

Fig 3 shows average distance between individuals


while performing multi objective optimization
using genetic algorithm which is found out to be
0.0037. The average spread is shown in Fig. 4
which is found out to be 0.11074. Fig. 5 to Fig. 8
shows various graphs obtained through multiobjective genetic algorithm. The algorithm stops
when generation reaches 238 as specified by Fig. 7.

Number of individuals

20

15

10

Table 2
Optimal values of various parameters obtained through MOGA:

Parameter
P

I
N
M
Th
Tc

Value
2.598 W
0.059
5A
15
15
450.58 K
299.27 K

0.5

1.5
Score (range)

2.5

Fig. 5 Number of children versus individuals


Rank histogram
25

20
Number of individuals

Sr. No.
1
2
3
4
5
6
7

The power output and thermal efficiency obtained


through proposed algorithm found in the following
ranges.
2.582 W P 2.607 W
0.05899 0.0598

15

10

1.5

2.5

3.5
Rank

4.5

Fig 6 Rank Histogram

168

5.5

2015 1st International conference on futuristic trend in computational analysis and knowledge management
(ABLAZE 2015)

REFERENCES
Stopping Criteria

[1]
[2]

Time

[3]

[4]

Generation

10

20

30

40
50
60
% of criteria met

70

80

90

[5]

100

Fig 7 Percentage of criteria met


-3

2.5

[6]

Distance of individuals

x 10

[7]

[8]
Distance

1.5

[9]
1

0.5

[10]
0

10

20

30
Individuals

40

50

60

[11]

Fig 8 Distance of individuals

CONCLUSION

[12]

The optimization technique using multi-objective


genetic algorithm has been used to simultaneous
optimize power output and thermal efficiency of
two stage exo-reversible thermoelectric generator.
The constraints are specified in equations [15-19]
and the fitness function is thermodynamically
developed and optimized. The Pareto frontier
between power output and thermal efficiency is
obtained and the best optimal values are chosen by
Fuzzy Bellman-Zadeh decision making method.
The ideal values of power output and efficiency are
2.607W and 0.0598 respectively where as their
non-ideal values are 2.582W and 0.0589
respectively, as shown in Pareto frontier given
above. Fuzzy decision making method is used to
get best optimal values of power output and
thermal efficiency of proposed model which comes
out to be 2.598W and 0.0590 respectively. Hence
the optimum designing of thermoelectric generator
given in the current paper can be considered as the
standard model for designing and analysis of real
thermoelectric generator in the view of
thermodynamics.

[13]

D.M. Rowe, CRC handbook of thermoelectric,


CRC press, 1995.
X. Ma, S.B. Riffat, Thermoelectric: a review of
present and potential applications, Applied Thermal
Engg., vol. 23(8), pp. 913-35, 2003.
J. Chen, Z. Yan, The influence of the Thomson
effect on the maximum power-output and maximum
efficiency of a thermoelectric generator, J. Appl.
Phys., vol. 79 (11), pp. 8823-8, 1996.
P.G. Lau, R.J. Buist, Calculation of thermoelectric
power generation performance using finite element
analysis, In. 16th int. Conf. On thermoelectric.
Germany Dresden, August 26-29, pp. 563-566, 1997.
J. Chen, Z. Yan, L. Wu, Non equilibrium
thermodynamic analysis of a thermoelectric device,
Energy: the Int J., vol. 22(10), pp. 979-985, 1997.
D.M. Rowe, G. Min, Evaluation of thermoelectric
modules for power generation and power sources,
vol. 73 (2), pp. 193-198, 1998.
M. Naji, M. Alata, M.A. Al-Nimr, Transient
behaviour of a thermoelectric device, Proc. Instrum
Mech Eng Part A, J Power Energy, vol. 217 (6A):,
pp. 615-621, 2003.
J. Chen, Thermodynamic analysis of a solar driven
thermoelectric generator, J. Appl. Phys., vol. 79(5),
pp. 2717-2721, 1996.
J. Chen, B. Andersen, New bounds on the
performance parameters of a thermoelectric
generator, Int. J. Power Energy Syst., vol. 17(1), pp.
23-27, 1997.
Y.H. Zhou, Optimal design of a new type of
semiconductor thermoelectric generator, J. Xiamen
University, vol. 40(4), pp. 882-887, 2001.
L. Chen, Jun Li, F. Sun, C. Wu, Performance
optimization of a two stage semiconductor
thermoelectric generator, Applied Energy, vol. 82,
pp. 300-312, 2005.
A. Konak, D.W. Coit and A.E. Smith, MultiObjective optimization using genetic algorithms: A
tutorial,ReliabilityEngineering and SystemSafety,
vol. 91, pp. 992-1007, 2006.
H. Sayyadi, R. Mehrabipour, Efficiency enhancement
of gas turbine cycle using an optimized tubular
recuperative heat exchanger, Energy, vol. 38 (1), pp.
362-375, 2012.

Nomenclature:
Pout= Power output
= Efficiency of thermoelectric generator
Q = Rate of heat transfer
M= Total number of pairs of thermoelectric elements of
thermoelectric generator
n= pairs of thermoelectric elements in top stage of
thermoelectric generator
m= pairs of thermoelectric elements in bottom stage of
thermoelectric generator
R= Total internal electrical resistance of semiconductor
thermocouple ()
I=working electrical current (A)
T= Temperature (K)
V= Voltage (V)
K = Semiconductor couples thermal conductance
TEG=Thermoelectric generator
Greek letters:
=See beck coefficient (V/K)
Subscripts:
H= Hot side / heat source
C= Cold side / sink side
1= Top stage of thermoelectric heat pump
2= Bottom stage of thermoelectric heat pump
Tm = Junction temperature between two stages

169

Anda mungkin juga menyukai