See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/276413391

Application of genetic algorithm with multi-objective function to improve the

efficiency of glazed photovoltaic thermal system for New Delhi (India)
climatic condition

Article in Solar Energy · July 2015

DOI: 10.1016/j.solener.2015.04.025

CITATIONS READS

47 203

4 authors, including:

Sonveer Singh S. Agrawal

Technical Education Department, Government of Uttar Pradesh, India Indira Gandhi National Open University (IGNOU)
13 PUBLICATIONS 452 CITATIONS 73 PUBLICATIONS 1,863 CITATIONS

SEE PROFILE SEE PROFILE

G. N. Tiwari
Bag Energy Research Society
708 PUBLICATIONS 28,864 CITATIONS

SEE PROFILE

All content following this page was uploaded by Sonveer Singh on 06 April 2019.

The user has requested enhancement of the downloaded file.

 Available online at www.sciencedirect.com

ScienceDirect
Solar Energy 117 (2015) 153–166
www.elsevier.com/locate/solener

Application of genetic algorithm with multi-objective function

to improve the efficiency of glazed photovoltaic thermal system
for New Delhi (India) climatic condition
Sonveer Singh a,b, Sanjay Agarwal c,d,⇑, G.N. Tiwari e, Deepika Chauhan f
a
CMS Government Girls Polytechnic Daurala, Meerut 250221, India
b
S.I.T.E., S.V. Subharti University, Meerut, India
c
SMERC, University of California, Los Angeles, CA 90095, USA
d
School of Engineering and Technology, IGNOU, New Delhi 110068, India
e
Indian Institute of Technology Delhi, New Delhi, India
f
Jagannath University, Jaipur, India

Received 1 January 2015; received in revised form 7 April 2015; accepted 20 April 2015

Communicated by: Associate Editor Bibek Bandyopadhyay

Abstract

The aim of this paper is to investigate an improvement in the efficiency of photovoltaic thermal (PVT) system with the help of Genetic
Algorithm (GA) with multi-objective functions for New Delhi, India climatic condition. There are several parameters influencing effi-
ciency of PVT system which inter alia include length and depth of the channel, velocity of air fluid flowing into the channel, thickness
of the tedlar and glass, temperature of inlet fluid. All these parameters have been considered to optimize the efficiency of the PVT system.
An attempt has also been made to model and optimize the parameters of glazed hybrid single channel PVT module considering the two
objective functions separately which are: (i) the overall exergy efficiency (ii) the overall thermal efficiency. Using GA, both of the above
objective functions are separately optimized and analyzed for each of the two cases: namely, Case-I: Improvement in exergy and thermal
efficiency when overall exergy efficiency is optimized and Case-II: Improvement in exergy and thermal efficiency when overall thermal
efficiency is optimized. The variables used in GA are those that could be varied, keeping parameters like solar radiation, ambient tem-
perature unchanged in the algorithmic calculation. The electrical and thermal efficiencies after optimization were found 14.15%, 11.88%
and 14.08%, 19.48% respectively. Similarly the overall exergy and thermal efficiency are 14.87% and 56.54% respectively for both the
cases. It has been observed that there is improvement in overall exergy efficiency and overall thermal efficiency by 4.6% and 13.14%
respectively during the optimization process.
Ó 2015 Elsevier Ltd. All rights reserved.

Keywords: Genetic algorithm; PVT; Exergy; Multi objective

1. Introduction
⇑ Corresponding author at: School of Engineering and Technology,
The few works have been done on the PVT module
IGNOU, New Delhi 110068, India. Tel.: +91 9911422863, +91
using Genetic algorithm. Some related work reported in
9412895695.
E-mail addresses: [email protected] (S. Agarwal), svsingh_ the literature. The novelty of the work produced in the
dei06@rediffmail.com (S. Singh). paper is to optimize the parameter of Glazed PVT module,

http://dx.doi.org/10.1016/j.solener.2015.04.025
0038-092X/Ó 2015 Elsevier Ltd. All rights reserved.
154 S. Singh et al. / Solar Energy 117 (2015) 153–166

Nomenclature

ASC area of the solar cell, m2 b width of the channel, m

d depth of the channel, m QU useful heat, W
Cair specific heat of air, J/kg K dx small length, m
dt small time, s In incident solar intensity, W/m2
KT thermal conductivity, W/m K L length of the channel, m
NC number of channel in SCPVT module TA ambient temperature, K
U overall heat transfer coefficient, W/m2 K TAvg average temperature, K
Vair velocity of air, m/s Am area of module, m2
VF velocity of fluid (air) in channel, m/s nR number of rows in solar PVT module
b0 temperature coefficient of efficiency, 1/K TFO temperature of fluid at outlet, K
gC,power power conversion factor b packing factor
a absorptivity s transmitivity
g efficiency TFI temperature of fluid at inlet, K
q density, kg/m3
h heat transfer coefficient, W/m2 K Subscripts
hTA heat transfer coefficients from back of tedlar to A ambient
ambient, W/m2 K SC solar cell
hGA heat transfer coefficients from top glass cover to GA glass cover to ambient
ambient, W/m2 K IA insulation to ambient
hTF heat transfer coefficients from back of tedlar to Avg average
flowing air (fluid), W/m2 K R rows
hIA heat transfer coefficients from back of insulation FI fluid at inlet
to ambient, W/m2 K eff effective
USCAG an overall heat transfer coefficient from solar C channel
cell to ambient through glass cover, W/m2 K TA tedlar to ambient
USCFT an overall heat transfer coefficient from solar TF tedlar to flowing fluid
cell to fluid through tedlar, W/m2 K T thermal
UFA an overall back loss heat transfer coefficient TC test condition
from fluid to ambient, W/m2 K F fluid
mF mass flow rate of fluid (air) in channel, kg/s FO fluid at outlet
gTC efficiency at standard test condition when U useful
In = 1000 W/m2 and TA = 25 °C

of which modeling of such system is reported in literature. cost analysis) has been developed and utilized by many
There are improvements found in overall exergy and over- researchers including Rosen and Dincer (2003) for years
all thermal energy efficiencies when optimization has been together. Coventry (2005) studied the performance of a
done. Depleting fuel reserves, multiplying energy demand, concentrating PVT collector and reported overall thermal
rapid rise in the fuel import bill, and compulsions of Kyoto and electrical efficiency of concentrating PV/T system as
protocol insisting to minimize Greenhouse-gas (GHG) 58% and 11%, respectively. This implies an overall effi-
emissions exert irresistible pressure on India to harness ciency of the PV/T system as 69%. Exergy is a consistent
renewable energy resources as the more suitable alternative measure of economic value as opined in many of the
to fossil fuels for electricity generation. It is high time for researches. Tiwari et al. (2006) have validated the theoret-
the country to change country’s energy basket by shifting ical and experimental results for PV module integrated
its focus from conventional fossil fuels (coal and diesel) with air duct for composite climate of India and concluded
to renewable energy resources for electricity generation. that an overall thermal efficiency of PVT system is signifi-
Lots of theoretical and experimental work on hybrid pho- cantly increased to 18% due to the utilization of thermal
tovoltaic thermal systems is reported in the literature. energy from PV module. Chow et al. (2007) found that
These (PVT) systems preliminarily utilize thermal energy forced convective cooling under a higher coolant flow
from the sun to produce electrical energy. Zondag et al. velocity gives better results than the natural flow design.
(2002) have developed the model of a hybrid PVT air col- Nevertheless, the additional fan power consumption
lector and performed experimental studies on the systems reduces the net electrical gain of the system. Tonui and
of varying sizes. The exergy based cost analysis popularly Tripanagnostopoulos (2008) used free air convective cool-
called exergoeconomic analysis (also known as exergetic ing to remove heat from the back of the PV modules and
 S. Singh et al. / Solar Energy 117 (2015) 153–166 155

to keep the electrical efficiency at an acceptable level. Free collectors installed. Singh et al. (2015) have developed a
air convective cooling is simple and cost effective but a model for single channel unglazed PVT module and opti-
forced flow scheme provides desirable cooling rate at all mized design parameter using Genetic Algorithms and con-
times. Dubey and Tiwari (2009) have made a detailed anal- cluded that the maximum overall exergy efficiency is
ysis of thermal energy, exergy and electrical energy yield by 16.88% at optimized parameters. Lamnatou et al. (2015)
varying the number of collectors and considering four have presented a paper for critical review about solar sys-
weather conditions. Agrawal and Tiwari (2011a) have done tem modeling with emphasis on BIST configurations.
performance evaluation of hybrid modified micro-channel They also included the BI solar systems which produce
solar cell thermal tile which was an experimental valida- electricity or both electrical/thermal energy in order to pro-
tion. Agrawal and Tiwari (2011b) have also reported exper- vide a more complete view of the current literature. They
imental validation of glazed hybrid micro-channel solar cell also focused on modeling studies about the solar system
thermal tile and concluded that the glazed hybrid MCPVT itself. The results reveal that most of the investigations
module gives higher electrical efficiency in comparison with about BI configurations refer to PVT, PV or skin façades
SCPVT module by 26.7% and obtained 20.28% overall while there are few studies about BIST systems.
exergy efficiency. Dupeyrat et al. (2011) introduced a differ-
ent approach to analyze thermal and optical properties 2. System explanation
related to both PV and solar thermal functions in order
to identify an optimum combination leading to a maximum In this Paper, the proposed module comprises channel
overall efficiency. Agrawal and Tiwari (2011c) also pre- between tedlar and insulation. The glass is considered
sented the concept of series and parallel connections of above the solar cell structure. The schematic view of pho-
micro-channel solar cell thermal tiles to analyze overall tovoltaic thermal module is shown in Fig. 1. The module
energy and exergy of hybrid micro-channel PVT module. is called single channel photovoltaic thermal module
Kumar and Rosen Mark (2011) have critically reviewed (SCPVTM). The side view of SCPVTM is shown in
PVT air collectors for air heating providing useful results Fig. 2. In order to obtain maximum overall exergy effi-
relating to the practicability of these collectors for preheat- ciency and overall thermal efficiency, different parameters
ing air to suit a large variety of applications. Rajoria et al. of SCPVTM are optimized using GAs. When solar radia-
(2012) have performed overall thermal energy and exergy tion impinge on PV module, the solar energy is converted
analysis of hybrid PVT array considering four array config- into electrical energy and thermal energy. Out of these,
uration and concluded that the performance of case III is electrical energy is stored in a battery. The front surface
better than rest of the cases. Singh et al. (2012) have per- of a PV module must have a high transmission in the wave-
formed comparative study of different types of hybrid pho- lengths which can be used by the solar cells in the PV mod-
tovoltaic thermal air collectors and reported that overall ule. For silicon solar cells, the top surface must have high
annual thermal energy, exergy gain and exergy efficiency transmission of light in the wavelength range of 350–
of unglazed hybrid PVT tiles air collector was improved 1200 nm. In addition, the reflection from the front surface
by 32%, 55.9% and 53% respectively, over the conventional should be low. While theoretically this reflection could be
PVT air collectors. Hamidreza and Keith (2013) developed reduced by applying an anti-reflection coating to the top
a model for controlling the temperature of the photovoltaic surface, in practice these coatings are not robust enough
cell and keep it under a specific limit for different condi- to withstand the conditions in which most PV systems
tions they have also done genetic algorithm based opti- are used.
mization to find the optimal value of the supplied Due to thermal energy the PV cell gets heated resulting
electrical current for the thermoelectric cooling module in reduced electrical efficiency because solar cells in the
which leads to the maximum generated power by the module are made with semiconductor material.
system. Therefore, maintaining electrical efficiency of PV module
Venkateswarlu et al. (2013) have done modeling and by heat removal becomes essential. The module assumes
parameter extraction of PV modules using GAs and differ- solar radiation being absorbed by solar cell and heat energy
ential evaluation. Buonomano et al. (2014) presented an due to radiation conducting to the base of the tedlar from
analysis of a possible energy retrofit of an existing thermal heating air flowing into the channel below the ted-
University Hospital District, located in Naples (Italy), by lar as shown in Fig. 1.
using an innovative renewable poly generation system
and provided the results for energy and economic point 3. Mathematical modeling of the system
of view. The economic results show that the system can
be profitable (pay-back period around 12 years) even with- In order to write the energy balance equation of SCPVT
out any public funding. The thermo-economic optimiza- module, the following assumptions are made:
tion, based on a mixed heuristic/deterministic algorithm,
also has been showed by them and concluded that the sys- 1. There is no temperature gradient along thickness of
tem profitability can be further improved, increasing solar the solar cell.
field area and decreasing storage specific volumes for m2 of 2. Heat capacity of solar cell is negligible.
156 S. Singh et al. / Solar Energy 117 (2015) 153–166

Glass

Cell

Inlet Air Insulation

Channel
Tedlar

Fig. 1. Proposed single channel glazed photovoltaic thermal module (SCGPVTM).

h TA Solar light USCAG ½Rate of heat transfer from solar cell to flowing fluid
h GA
ðairÞ through tedlar
Glass
Cell ¼ ½The rate of heat gain by flowing fluid i:e:
Tedlar
Channel h TF USCFT air in channel
back þ ½Rate of heat transfer from flowing fluid to ambient
Insulation
h IA UFA dT F
U SCFT ðT SC  T F Þbdx ¼ mF C air dx þ U FA ðT F  T A Þbdx
Fig. 2. Side view of single channel glazed photovoltaic thermal module. dx
ð3Þ

3. Specific heat of air remains constant. where mF = qLdVF.

4. The system is in quasi-steady state throughout dur- Solving Eqs. (1)–(3), the outlet temperature (TFON),
ing observation. instantaneous thermal (gth) and electrical efficiency (g) at
5. Packing factor is unity. Nth number of glazed hybrid single channel solar cell ther-
mal tile connected in series are given by
The SCPVT cell is shown in Fig. 1. The small area of    
hp aeff NbU L L
SCPVT cell is bdx. The energy balance equation of glazed T FON ¼ In þ T A 1  exp þ T FI
UL mF C air
hybrid SCPVT cell as given by Agrawal and Tiwari  
(2011b), is as follows NbU L L
 exp ð4Þ
mF C air
½Rate of solar energy available on glazed solar cell
¼ ½Rate of heat loss from top surface of solar cell where
 1  1
to ambient through glass over Lg 1 LT 1
U SCAG ¼ þ U SCFT ¼ þ
þ ½Rate of heat transfer from solar cell to K g hGA K T hTF
flowing fluid i:e: air through tedlar  1  
Lin 1 U SCFT
U FA ¼ þ hp ¼
þ ½Rate of electrical energy produced K in hIA U SCAG þ U SCFT
 1
½ac sg In  bdx ¼ ½U SCAG ðT SC  T A Þbdx 1 1
U fa ¼ þ U L ¼ U FA þ U fa
þ ½U SCFT ðT SC  T F Þbdx þ ½sg gTC In  bdx U SCFT U SCAG
ð1Þ The thermal gain derived by Agrawal and Tiwari
Solving Eq. (1), we have (2011b) is found as:
   
aeff In þ U SCAG T A þ U SCFT T F mF C air NbU L L ðT FI  T A Þ
T SC ¼ ð2Þ gth ¼ 1  exp hp aeff  U L
U SCAG þ U SCFT U L NASC mF C air In
ð5Þ
where aeff = sg(aSC – gTC).
Energy balance for air flowing into the channel of Single The rate of useful thermal energy obtained for the nR
channel PVT for elemental area bdx is given by – row of SCPVT module is calculated as:
 S. Singh et al. / Solar Energy 117 (2015) 153–166 157
 
hp aeff Table 1
QU ;N ¼ nR mF C air In þ T A  T FI Design parameters.
UL
   S. No. Parameters Values
N C bU L L
 1  exp ð6Þ 1. Vair, m/s 1.5
mF C air 2. TFO 25 °C
3. Kg, W/m K 1.1
4. KT, W/m K 0.033
3.1. Instantaneous electrical efficiency 5. Cair, J/kg K 1012
6. hGA, W/m2 K 5.7 + 3.8 * Vair
Electrical efficiency of solar cell depends on solar cell 7. hIA, W/m2 K 2.8 + 3 * VF
temperature as given by Schott (1985) and Evans (1981) 8. b0, 1/K 0.0045
and can be written as: 9. gTC 0.15
  10. N 4
g ¼ g0 ½1  b0 T SCavg  T O  ð7Þ 11. nR 9
12. q, kg/m3 1.29
An expression for electrical efficiency of hybrid SCPVT 13. sg 0.95
which is a temperature dependent is given by Agrawal and 14. aC 0.9
15. Kin, W/m K 0.089
Tiwari (2011b) as follows:
16. hTF, W/m K 4.3
 
aeff In
g ¼ gTC 1  b0  ðT FO  T A Þ The expression for input exergy is given by Petela (2003)
U SCAG þ U SCFT
8 9 as
U SCFT hp aeff In < exp NbU LL =
mF C air "      4 #
þ 1 X 4 TA 1 TA
U L ðU SCAG þ U SCFT Þ : NbU L L ; ExIN ¼ ASC N C  In 1  þ
mF C air 3 T Sun 3 T Sun
8 9 93
<1  exp NbU L L = =
U SCFT mF C air ð14Þ
þ ðT A  T FI Þ 5 ð8Þ
U SCAG þ U SCFT : NbU L L ; ; where Am is a area of module and TSun is the sun
mF C air
temperature.
The exergy efficiency of SCPVT module is given by
3.1.1. Energy analysis Agrawal and Tiwari (2011c) as follows:
 
The energy analysis is based on the first law of thermo- ExOUT
dynamics, and total thermal gain can be represented as gEx ¼  100 ð15Þ
ExIN
follows:
P
X X QU ;EL
QUT ¼ QU ;Th þ ð9Þ 3.1.3. Constant design parameter
gC;power Design parameters of SCPVT module are given in
where Table 1.
X QU ;N 4. Optimization through genetic algorithm
QU ;Th ¼ ð10Þ
1000
Here gC,power is a electric power generation efficiency con- Optimization is the process by which maximization or
version factor of a conventional power plant. For India this minimization of real function can be done by choosing input
efficiency is in the range, 0.20–0.40 as studied by Huang values within an allowed boundary and computing the value
et al. (2001) which is based on the quality of coal. of function. This process is repeated continuously and sys-
Usually it is taken as 0.38. tematically until the desired value of the function is
obtained. Genetic Algorithms (GAs) are adaptive heuristic
search which is based on evolutionary ideas of natural selec-
3.1.2. Exergy analysis
tion and genetics. It is used to solve the problems which ter-
The general exergy balance as per Agrawal and Tiwari
minate in a finite number of steps or iterative methods that
(2011c) for a Single channel PVT module can be expressed
converge to a solution. The basic techniques of the GAs
as:
X X X are designed to simulate processes in natural systems neces-
ExOUT ¼ ExTh þ ExEL ð11Þ sary for evolution; especially those following the principles
that were first laid down by Charles Darwin of survival of
where the fittest theory. In nature, there exists competition among
X  
T A þ 273 individuals to gain access over scanty resources resulting in
ExTh ¼ QU ;N 1  ð12Þ the domination of fittest individuals over the weaker ones.
T FO þ 273
X   GAs are based on analogy with the genetic structure and
gAm In the behavior of chromosomes within a population of indi-
ExEL ¼ ð13Þ
1000 viduals using the following basic rules:
158 S. Singh et al. / Solar Energy 117 (2015) 153–166

 Individuals in a population compete for resources and 4.2.1. Roulette wheel selection
mates. In roulette selection chromosomes are selected based on
 The individuals that are most successful in each compe- their fitness relative to all other chromosomes in the popu-
tition shall produce more off springs than those perform- lation as shown in Fig. 4. Parents are selected according to
ing poorly. their fitness in a Roulette wheel every chromosome has its
 Genes from good individuals propagate throughout the place big according to its fitness function like in Fig. 4.
population implying that two good parents shall gener- Chromosome with bigger fitness will be selected more
ally produce offspring who are better equipped than times. One disadvantage of using roulette wheel is that its
those from poorly performing parents. selective pressure reduces as population converges upon a
solution, which reduces the convergence rate and may
Thus, each successive generation will become more sui- not allow finding the better solutions.
ted. The flow chart relating to the Genetic Algorithm is
shown in Fig. 3. The details of the proposed genetic algo- 4.2.2. Crossover
rithm are given as follows. In the crossover operation with crossover probability
(PC), select two parents and randomly selects a point
4.1. Coding and initialization between two genes to cut both chromosomes into two
parts. This point is called crossover point. In crossover
In order to solve the specified problem with genetic algo- operation combine the first part of first parent and second
rithms, variables are coded in some string structures. There part of second parent to get first offspring. Similarly, com-
are some of the studies in which directly variables values bine the first part of second parent and second part of first
are taken, but in the present GA work with binary coded parent to get second offspring. These off-springs belong to
variable strings. The proposed GA works with a popula- the next population. The crossover operator has three dis-
tion of binary string, not the parameters themselves. For tinct sub-steps as follows:
simplicity and convenience, binary coding is used in this
paper.  Slice each of the parent strings in two substrings.
The proposed genetic algorithm starts with initialization  Exchange a pair of corresponding substrings of
of input data and various parameters such as population parents.
size, maximum numbers of generation, number of genes
(control variables) in each chromosome, crossover proba-
bility (PC) and mutation probability (PM).

4.2. Genetic operators

The following genetic operators are adopted in the

proposed algorithm.
Fig. 4. Roulette wheel selection.

Fig. 3. Genetic algorithm flow chart.

 S. Singh et al. / Solar Energy 117 (2015) 153–166 159

 Merge the two respective substrings to form 5. Results and discussion

off-springs.
In this paper, different parameters like depth of the
channel (d), Length of the channel (L), depth of the top
4.2.3. Mutation glass cover (Lg), thickness of the tedlar (Lt), thickness of
The mutation operator selects a gene of any chromo- the insulation (Lin), velocity of the fluid i.e. air (VF) and
some of current population and changes the allele by an inlet air temperature (TFI) have been optimized for maxi-
amount called the mutation variance, this happens with a mizing efficiency of photovoltaic thermal (PVT) system.
mutation probability (PM). The parameter mutation vari- Genetic Algorithm is applied to the problem to optimize
ance and mutation frequency have a major influence on values of each parameter in order to maximize overall effi-
the quality of learning algorithms. For the proposed binary ciency. The choice of 1 h or single data is used for the opti-
coded GA mutation is equivalent of flipping a bit at any mization of the design parameters of system as given in
particular position. Since, mutation is to be used sparingly Table 2 and after optimizing the parameter further analysis
its probability is very low. has been done for a complete day and the hourly input data
for a complete day is given in Table 5. The input data con-
4.2.4. Reproduction sidered for analysis has been obtained from IMD, Pune for
Reproduction is based on the principle of survival of the a New Delhi, India for a complete year (2009) for four
better fitness. It is an operator that obtains a fixed number weather conditions like clear days, hazy days, hazy and
of copies of solutions according to their fitness value. If the cloudy days and cloudy days called as Set A to Set D
score increases, then the number of copies increases too. A respectively as considered by Agrawal and Tiwari
score value is of associated to a given solution according to (2011a). The number of clear days fall in different weather
its distance of the optimal solution (closer distances to the conditions for New Delhi are shown in Table 6.
optimal solution mean higher scores). The following procedure is adopted for this purpose.

4.2.5. Methodology adopted 5.1. Case-I: Improvement in exergy and thermal efficiency
when overall exergy efficiency is optimized
 Identify control parameters influencing overall exergy
efficiency of the proposed SCPVT module. 5.1.1. Analysis on single data at 11:00 AM
 Define upper and lower bound of each parameter value Overall exergy efficiency of single channel glazed PVT
feasible for designing the SCPVT module. module has been optimized and observations are shown
 Decide the objective function also known as fitness func- in Figs. 5–10. The input data at 11:00 AM is shown in
tion. In this paper there are two objective function one Table 2.
is; overall exergy efficiency and second is; overall ther- The variation in fitness value and overall exergy effi-
mal efficiency. ciency with respect to generation have been observed
 MATLAB program has been developed for optimizing shown in Fig. 5(a) and (b) respectively whereby the maxi-
each parameter as based on the flow chart shown in mum overall exergy efficiency of SCPVT module is found
Fig. 3. to be 14.87%. It is clear that the value of overall exergy

Fig. 5. (a) Variation in fitness value with respect to generation, while objective function is overall exergy efficiency. (b) Variation in overall exergy
efficiency with respect to generation, while objective function is overall exergy efficiency.
160 S. Singh et al. / Solar Energy 117 (2015) 153–166

Table 2 5.1.2. Analysis for a complete day from 08:00 AM to 05:00

Input data at 11:00 AM. PM
S. No. Name of input parameter Value of parameter Overall exergy efficiency is calculated for a complete day
1. Intensity of solar light (In) 680.73 W/m2 from 08:00 AM to 05:00 PM at optimized parameters. The
2. Ambient temperature (TA) 6.6 °C optimized parameters are given in Table 3. The input data
for complete day is shown in Table 5.
Figs. 6 and 7 shows variation in overall exergy efficiency
efficiency has been optimized by genetic algorithm within and Cell temperature with respect to time respectively and
1500 generations. it has been observed that the exergy efficiency is maximum
Output observed at 11:00 AM: (15.8%) when cell temperature is minimum at 8:00 AM
The maximum overall exergy efficiency of single channel while the exergy efficiency is minimum (14.6%) when the
PVT module is 14.87% which is obtained at optimized cell temperature is maximum at 1:00 PM and concluded
parameters. The optimized parameters are given in that the exergy efficiency depends on the cell temperature.
Table 3. The convergence curve (i.e. overall exergy effi- From Fig. 8, it is observed that the trend of exergy effi-
ciency Vs generation curve) is shown in Fig. 5. The results ciency is similar to the trend of overall exergy efficiency. So
of analysis done at 11:00 AM are shown in Table 4.

16 16

15.5
Overall exergy efficiency, %

Exergy efficiency, %

15.5
15

14.5

15

14

13.5
14.5 8 9 10 11 12 13 14 15 16 17
8 9 10 11 12 13 14 15 16 17 Time, h
Time, h
Fig. 8. Variation in exergy efficiency with respect to time (08:00 AM to
Fig. 6. Variation in overall exergy efficiency with respect to time (08:00 05:00 PM), while objective function is overall exergy efficiency.
AM to 05:00 PM), while objective function is overall exergy efficiency.

20
55

50
15

45
Thermal efficiency, %
Cell temperature, oC

40 10

35

5
30

25
0

20

15 -5
8 9 10 11 12 13 14 15 16 17 8 9 10 11 12 13 14 15 16 17
Time, h Time, h

Fig. 7. Variation in cell temperature with respect to time (08:00 AM to Fig. 9. Variation in thermal efficiency with respect to time (08:00 AM to
05:00 PM), while objective function is overall exergy efficiency. 05:00 PM), while objective function is overall exergy efficiency.
 S. Singh et al. / Solar Energy 117 (2015) 153–166 161

it has been observed that the effect of thermal efficiency on Fig. 9 leads to the conclusion that thermal efficiency is
overall exergy efficiency is less while overall exergy effi- regularly increasing during day time but it is not influenc-
ciency has been optimized. ing the overall exergy efficiency in the same way.
From Fig. 10 it is noted that the trend of overall thermal
60 efficiency is similar to the trend of thermal efficiency.
There is 4.6% improvement in overall exergy efficiency
58
while overall exergy efficiency has been optimized.
56
Overall thermal efficiency, %

54 5.1.3. Analysis for a complete year

52
Overall exergy efficiency is evaluated for a year from
January to December at optimized parameters (Table 3).
50
Fig. 11 shows variation in overall exergy efficiency with
48 respect to month including all weather conditions. It leads
to the conclusion that overall exergy efficiency is maximum
46

44

42 16.5

40 16
8 9 10 11 12 13 14 15 16 17
Time, h
15.5
Overall exergy efficiency, %
Fig. 10. Variation in overall thermal efficiency with respect to time (08:00
AM to 05:00 PM), while objective function is overall exergy efficiency. 15

14.5
Table 3
Optimized parameters when overall exergy efficiency is optimized. 14
Parameter Optimized value
Thickness of channel (d) 0.00082 m 13.5

Length of channel (L) 0.0341 m

Thickness of back insulation Lin 0.1200 m 13
Velocity of flowing fluid (VF) 1.0709 m/s
Thickness of tedlar (Lt) 0.0001 m 12.5
0 2 4 6 8 10 12
Thickness of top glass cover (Lg) 0.00016 m
Month, January to December
Temperature of fluid at inlet (TFI) 15 °C
Fig. 11. Variation in overall exergy efficiency with respect to months
including all weather conditions, while objective function is overall exergy
Table 4 efficiency.
Results obtained at 11:00 AM when overall exergy efficiency is optimized.
S. No. Parameters Values (%)
5
1. Overall exergy efficiency 14.87
2. Thermal efficiency 11.88 4.5
Overall thermal gain for New Delhi, kWh

3. Exergy efficiency 14.15

4
4. Overall thermal efficiency 49.11
3.5

3
Table 5
Input data for a complete day from 08:00 AM to 05:00 PM. 2.5
S. No. Time Intensity of Ambient
2
light (W/m2) temperature (°C)
1. 08:00 AM 132.99 7.90 1.5
2. 09:00 AM 355.56 7.90
3. 10:00 AM 554.69 7.90 1
4. 11:00 AM 680.73 6.60
0.5
5. 12:00 Noon 726.74 6.40
6. 01:00 PM 733.85 7.70 0
7. 02:00 PM 656.08 10.60 1 2 3 4 5 6 7 8 9 10 11 12
8. 03:00 PM 500.00 13.00 Month, January to December
9. 04:00 PM 311.46 15.00
Fig. 12. Variation in overall thermal gain with respect to month including
10. 05:00 PM 106.42 16.50
all weather conditions, while objective function is overall exergy efficiency.
162 S. Singh et al. / Solar Energy 117 (2015) 153–166

(15.68%) for the month January and minimum (12.71%) 5.1.4. Optimization for a complete year from January to
for the month of May. December
Fig. 12 shows a value of overall thermal gain for Optimization has been performed for a complete year.
different months of the year. It leads to conclusion that The input data considered for analysis has been obtained
the overall thermal gain is maximum (4.73 kW h) for the from IMD, Pune for a New Delhi, India at 11:00 AM for
month May and it minimum (2.23 kW h) for the month each month of the year (2009) as considered by Agrawal
December. and Tiwari (2011a). GA tools operate according to the algo-
Fig. 13 shows a values of overall exergy gain for rithm as shown in Fig. 3, it operates at single data at a time
different month of the year. It leads to conclusion that and give a set of optimized parameters for each month and
the overall exergy gain is maximum (1.02 kW h) for the takes next data after previous execution and it continues
month May and it minimum (0.61 kW h) for the month until the input data is over. The results obtained during
November. the optimization process are shown in Table 7 which indi-
cates distinct value of optimize parameter for each month.
It leads to conclusion that there is very slight variation
1.4 in one set of optimized parameters to another set for each
month. We have considered average values of parameters
1.2 for optimization of overall exergy efficiency as shown in
Overall exergy gain for New Delhi, kWh

Table 8. The value of overall exergy efficiency has been

1
evaluated at average value of optimized parameters and
it is found 14.862% as given in Table 8 which is similar
0.8
to the overall exergy efficiency 14.87% at single data.

0.6 5.2. Case-II: Improvement in exergy and thermal efficiency

when overall thermal efficiency is optimized
0.4
5.2.1. Analysis on single data at 11:00 AM
0.2 Overall thermal efficiency of single channel glazed pho-
tovoltaic thermal module has been optimized and observa-
0 tions are shown in Figs. 14–19. The input data at 11:00 AM
1 2 3 4 5 6 7 8 9 10 11 12
Month, January to December is shown in Table 2.
The plot of fitness function Vs generation and overall
Fig. 13. Variation in overall exergy gain with respect to month including thermal efficiency Vs generation have been observed shown
all weather conditions, while objective function is overall exergy efficiency.
in Fig. 14 whereby the maximum overall thermal efficiency

Table 6
Number of clear days fall in different weather conditions for New Delhi.
WC/Month January February March April May June July August September October November December
Set A 3 3 5 4 4 3 2 2 7 5 6 3
Set B 8 4 6 7 9 4 3 3 3 10 10 7
Set C 11 12 12 14 12 14 10 7 10 13 12 13
Set D 9 9 8 5 6 9 17 19 10 3 2 8

Table 7
Results obtained at 11:00 AM for each month when overall exergy efficiency is optimized.
S. No. Months d, m L, m Lin, m VF, m/s Lt, m Lg, m TFI, °C Overall exergy efficiency
1. January 0.0008 0.0291 0.1192 0.97 0.0001 0.0001 14.81 14.863
2. February 0.0008 0.0475 0.1180 1.41 0.0001 0.0003 15.00 14.861
3. March 0.0006 0.0289 0.1192 1.42 0.0001 0.0008 14.91 14.857
4. April 0.0009 0.0289 0.1200 0.85 0.0001 0.0002 14.98 14.863
5. May 0.0005 0.0228 0.1160 1.24 0.0001 0.0004 14.96 14.859
6. June 0.0007 0.0278 0.1200 0.90 0.0001 0.0006 14.94 14.860
7. July 0.0008 0.0190 0.1175 0.54 0.0001 0.0007 14.89 14.860
8. August 0.0007 0.0420 0.1193 1.43 0.0001 0.0001 14.94 14.862
9. September 0.0008 0.0260 0.1186 0.70 0.0001 0.0001 14.45 14.861
10. October 0.0008 0.0356 0.1161 1.21 0.0001 0.0006 14.94 14.855
11. November 0.0008 0.0266 0.1125 0.76 0.0002 0.0001 14.96 14.853
12. December 0.0008 0.0395 0.1199 1.35 0.0001 0.0005 14.99 14.861
 S. Singh et al. / Solar Energy 117 (2015) 153–166 163

Table 8
Average parameters obtained at 11:00 AM for each month when overall exergy efficiency is optimized.
S. No. Months d, m L, m Lin, m VF, m/s Lt, m Lg, m TFI, °C Overall exergy efficiency
1. January 0.0008 0.0311 0.1180 0.91 0.0001 0.0004 14.89 14.862

Fig. 14. (a) Variation in fitness value with respect to generation, while objective function is overall thermal efficiency. (b) Variation in overall thermal
efficiency with respect to generation, while objective function is overall thermal efficiency.

of SCPVT module is found to be 56.54%. It is clear that the Output observed at 11:00 AM:
value of overall thermal efficiency has been optimized by The maximum overall thermal efficiency of SCPVT
genetic algorithm within 1000 generations. module is 56.54% which is obtained at optimized parame-
The numeric values of optimized parameters are given in ters. The optimized parameters are given in Table 9. The
Tables 3 and 9 respectively when overall exergy efficiency convergence curve (i.e. overall thermal efficiency Vs gener-
(fitness value) and overall thermal efficiency (fitness value) ation curve) is shown in Fig. 14.
is optimized. The value of input parameters are given in The results of analysis done at 11:00 AM are shown in
Tables 2 and 5. Table 10.

18
100

95
16

90
Overall thermal efficiency, %

Ambient temperature, oC

14
85

80
12

75

70 10

65
8
60

55 6
8 9 10 11 12 13 14 15 16 17 8 9 10 11 12 13 14 15 16 17
Time, h Time, h

Fig. 15. Variation in overall thermal efficiency with respect to time (08:00 Fig. 16. Variation in ambient temperature with respect to time (08:00 AM
AM to 05:00 PM), while objective function is overall thermal efficiency. to 05:00 PM), while objective function is overall thermal efficiency.
164 S. Singh et al. / Solar Energy 117 (2015) 153–166

60 15.8

55 15.6

15.4
50

15.2
Thermal efficiency, %

45

Exergy efficiency, %
15
40
14.8
35
14.6
30
14.4

25 14.2

20 14

15 13.8
8 9 10 11 12 13 14 15 16 17 8 9 10 11 12 13 14 15 16 17
Time, h Time, h

Fig. 17. Variation in thermal efficiency with respect to time (08:00 AM to Fig. 19. Variation in exergy efficiency with respect to time (08:00 AM to
05:00 PM), while objective function is overall thermal efficiency. 05:00 PM), while objective function is overall thermal efficiency.

15.5 Table 9
Optimized parameters when overall thermal efficiency is optimized.
Parameter Optimized value
15
Thickness of channel (d) 0.00089 m
Length of channel (L) 0.0100 m
Overall exergy efficiency, %

Thickness of back insulation (Lin) 0.0.0919 m

14.5
Velocity of flowing fluid (VF) 1.4990 m/s
Thickness of tedlar (Lt) 0.0001 m
Thickness of top glass cover (Lg) 0.00085 m
14 Temperature of fluid at inlet (TFI) 5.0089 °C

13.5 Table 10
Results obtained at 11:00 AM when overall thermal efficiency is
optimized.
13
S. No. Parameters Values (%)
1. Overall exergy efficiency 14.18
12.5 2. Thermal efficiency 19.48
8 9 10 11 12 13 14 15 16 17 3. Exergy efficiency 14.08
Time, h 4. Overall thermal efficiency 56.54

Fig. 18. Variation in overall exergy efficiency with respect to time (08:00
AM to 05:00 PM), while objective function is overall thermal efficiency. From Fig. 17 it has been observed that the trend of ther-
mal efficiency is similar with the trend of overall thermal
5.2.2. Analysis for a complete day from 08:00 AM to 05:00 efficiency. This leads us to conclude that the effect of exergy
PM efficiency on overall thermal efficiency is less, while overall
Overall thermal efficiency is calculated for a complete thermal efficiency has been optimized.
day from 08:00 AM to 05:00 PM at optimized parameters. Fig. 18 shows that overall exergy efficiency is regularly
The optimized parameters are given in Table 9. The input decreasing during day time without much influence on
data for complete day is shown in Table 5. overall thermal efficiency. Similarly, from Fig. 19 it is
Fig. 15 and 16 shows variation in overall thermal effi- observed that the variation in exergy efficiency is not simi-
ciency and ambient temperature with respect to time lar to the variation of overall exergy efficiency as was the
respectively and it has been observed that the overall ther- case for thermal efficiency.
mal efficiency depends on the ambient temperature. When There is 13.14% improvement in overall thermal effi-
ambient temperature is maximum then the overall thermal ciency while overall thermal efficiency has been optimized.
efficiency is maximum at 5:00 PM while the overall thermal These optimized parameters have been obtained with help
efficiency is minimum at 12:00 Noon when the ambient of GA tool. Now this is challenge to industry how can they
temperature is minimum. manufactured such proposed hybrid PVT system to get
 S. Singh et al. / Solar Energy 117 (2015) 153–166 165

better electrical efficiency and the same time we are able to III. Generation: The evolution usually starts from a pop-
get thermal energy which can be useful: ulation of randomly generated individuals, and is an
iterative process, with the population in each itera-
i. To integrate to building to heat. tion called a generation. In each generation, the fit-
ii. For PVT dryer to dry medicinal plant etc. ness of every individual in the population is
evaluated; the fitness is usually the value of the objec-
6. Conclusions tive function in the optimization problem being
solved. The more fit individuals are stochastically
 While overall exergy and overall thermal efficiencies selected from the current population, and each indi-
have been optimized, separately two distinct values vidual’s genome is modified (recombined and possi-
of optimization parameters have been obtained as bly randomly mutated) to form a new generation.
indicated in Tables 3 and 9. The new generation of candidate solutions is then
 On the basis of analysis of parameters it is found used in the next iteration of the algorithm.
that the three parameters; velocity of flowing fluid, Commonly, the algorithm terminates when either a
temperature of inlet fluid and thickness of top glass maximum number of generations has been produced,
cover cause radical change in efficiency. This leads or a satisfactory fitness level has been reached for the
us to the conclusion that for optimization of the population.
overall thermal efficiency these three design parame- IV. Fitness function: Fitness function is basically a objec-
ters have dominating influence while other parame- tive function that is used to summarize, as a single fig-
ters cause negligible change in efficiency. ure of merit, how close a given design solution is to
 Changes in efficiency to the extent of 13.14% achieving the set aims.
improvement are observed in overall thermal effi-
ciency in the course of this optimization study of
overall thermal efficiency. Acknowledgements
 Changes in efficiency to the extent of 4.6% improve-
ment are observed in overall exergy efficiency in the The authors are thankful to the Department of Science
course of this optimization study of overall exergy and Technology, Government of India and the Indo-US
efficiency. Science and Technology Forum (IUSSTF) for providing
 Observations from Figs. 9–19 above lead to the con- the necessary help to give support to write a manuscript.
clusion that the cell temperature is more dominating
to the overall exergy efficiency while ambient tem- References
perature is more dominant for the overall thermal
efficiency. Agrawal, S., Tiwari, G.N., 2011a. Performance evaluation of hybrid
 Observation from Figs. 12 and 13 above lead to the modified micro-channel solar cell thermal tile: an experimental
conclusion that the overall annual thermal gain and validation. Int. J. Eng. Sci. Technol. 3, 244–254.
Agrawal, S., Tiwari, A., 2011b. Experimental validation of glazed hybrid
overall annual exergy gain is 40.75 kW h and micro-channel solar cell thermal tile. Sol. Energy 85, 3046–3056.
9.81 kW h respectively. Agrawal, S., Tiwari, G.N., 2011c. Energy and exergy analysis of hybrid
micro-channel photovoltaic thermal module. Sol. Energy 85, 356–
Glossary 370.
Buonomano, A., Calise, F., Ferruzzi, G., Vanoli, L., 2014. A novel
renewable polygeneration system for hospital buildings: design,
Some terms related to Genetic Algorithm are as follows: simulation and thermo-economic optimization. Appl. Ther. Energy
67 (1–2), 43–60.ooooo.
I. Chromosome: In Genetic Algorithm, a chromosome is Chow, T.T., He, W., Ji, J., 2007. An experimental study of facade-
a set of parameters which define a proposed solution integrated photovoltaic water-heating system. Appl. Therm. Eng. 27,
37–45.
to the problem that the genetic algorithm is trying to
Coventry, S.J., 2005. Performance of a concentrating photovoltaic/
solve. The chromosome is often represented as a sim- thermal solar collector. Sol. Energy 78, 211–222.
ple string. Sometimes it is also called as genotype. Dubey, S., Tiwari, G.N., 2009. Analysis of PVT flat plate water collectors
II. Population: A population is a summation of all the connected in series. Sol. Energy 83 (2009), 1485–1498.
chromosomes of the same group or species. In a Dupeyrat, P., Menezo, C., Rommel, M., Henning, H.M., 2011. Efficient
single glazed flat plate photovoltaic–thermal hybrid collector for
genetic algorithm, a population of candidate solu- domestic hot water system. Sol. Energy 85, 1457–1468.
tions (called individuals, creatures, or phenotypes) Evans, D.L., 1981. Simplified method for predicting PV array output. Sol.
to an optimization problem is evolved toward better Energy 27, 555–560.
solutions. Each candidate solution has a set of prop- Huang, B.J., Lin, T.H., Hung, W.C., Sun, F.S., 2001. Performance
erties (its chromosomes or genotype) which can be evaluation of solar photovoltaic/thermal systems. Sol. Energy 70 (5),
443–448.
mutated and altered; traditionally, solutions are rep-
Kumar, Rakesh., Rosen Mark, A., 2011. A critical review of photo-
resented in binary as strings of 0 s and 1 s, but other voltaic–thermal solar collectors for air heating. Appl. Energy 88
encodings are also possible. (2011), 3603–3614.
166 S. Singh et al. / Solar Energy 117 (2015) 153–166

Lamnatou, C., Mondol, J.D., Chemisan, D., Maurer, C., 2015. Modelling Singh, S., Agrwal, S., Tiwari, Arvind., Al-Helal, I.M., Avasthi, D.V.,
and simulation of building-integrated solar thermal systems: behaviour 2015. Modeling and parameter optimization of hybrid single channel
of the system. Sol. Energy 45, 36–51. photovoltaic thermal module using genetic algorithms. Sol. Energy
Petela, R., 2003. Exergy of undiluted thermal radiation. Sol. Energy 74 (6), 113, 78–87.
469–488. Tiwari, Arvind, Sodha, M.S., Chandra, Avinash, Joshi, J.C., 2006.
Rajoria, C.S., Agrawal, S., Tiwari, G.N., 2012. Overall thermal energy Performance evaluation of photovoltaic thermal solar air collector for
and exergy analysis of hybrid photovoltaic thermal array. Sol. Energy composite climate of India. Sol. Energy Mater. Solar Cells 90 (2), 175–
86, 1531–1538. 189.
Rosen, M.A., Dincer, I., 2003. Exergoeconomic analysis of power plants Tonui, J.K., Tripanagnostopoulos, Y., 2008. Performance improvement
operating on various fuels. Appl. Therm. Eng. 23, 643–658. of PVT solar collectors with natural air flow operation. Sol. Energy 82,
Schott, T., 1985. Operational temperatures of PV modules. In: 1–12.
Proceedings of 6th PV. Solar Energy Conference, pp. 392–396. Venkateswarlu, G. et al., 2013. Modeling and parameter extraction of PV
Singh, G.K., Agrawal, S., Tiwari, G.N., 2012. Analysis of different types modules using GA and differential evaluation. IOSR-JEEE 05, 37–44.
of hybrid photovoltaic thermal air collectors: a comparative study. J. Zondag, H.A., de Vries, D.W., Van Helden, W.G.J., Van Zolengen,
Fundamentals Renew. Energy Appl. http://dx.doi.org/10.4303/jfrea/ R.J.C., Steenhoven, A.A., 2002. The thermal and electrical yield of a
R120305. PV-thermal collector. Sol. Energy 72 (2), 113–128.

View publication stats