Green Energy and Technology

Venkata Rao Ravipudi

Hameer Singh Keesari

Design Optimization
of Renewable
Energy Systems
Using Advanced
Optimization
Algorithms
Green Energy and Technology
Climate change, environmental impact and the limited natural resources urge scien-
tific research and novel technical solutions. The monograph series Green Energy
and Technology serves as a publishing platform for scientific and technological
approaches to “green”—i.e. environmentally friendly and sustainable—technolo-
gies. While a focus lies on energy and power supply, it also covers “green” solu-
tions in industrial engineering and engineering design. Green Energy and Tech-
nology addresses researchers, advanced students, technical consultants as well as
decision makers in industries and politics. Hence, the level of presentation spans
from instructional to highly technical.
**Indexed in Scopus**.
**Indexed in Ei Compendex**.

More information about this series at https://link.springer.com/bookseries/8059

Venkata Rao Ravipudi · Hameer Singh Keesari

Design Optimization
of Renewable Energy
Systems Using Advanced
Optimization Algorithms
Venkata Rao Ravipudi Hameer Singh Keesari
Department of Mechanical Engineering Department of Mechanical Engineering
S. V. National Institute of Technology Sreenidhi Institute of Science
Surat, Gujarat, India and Technology
Hyderabad, Telangana, India

ISSN 1865-3529 ISSN 1865-3537 (electronic)

Green Energy and Technology
ISBN 978-3-030-95588-5 ISBN 978-3-030-95589-2 (eBook)
https://doi.org/10.1007/978-3-030-95589-2

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature
Switzerland AG 2022
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

Global warming is a significant concern that raises a need for cleaner production
of energy. Renewable energy resources are significant sources of contribution to
such clean energy demands. Taking advantage of these renewable energy sources
provides significant opportunities for handling energy-related problems. In the last
few decades, researchers have focused on renewable energy resources like solar
energy, bioenergy, wave energy, ocean thermal energy, tidal energy, geothermal
energy, and wind energy. This has resulted in the development of new techniques and
tools that could harvest energy from renewable energy sources. However, to meet
energy demands and reduce investment, a rigorous study of energy extraction systems
is required. Identifying, analyzing, and optimizing the effect of various parame-
ters of a renewable energy system contribute significantly to assessing the system
performance. Furthermore, it is always not preferable to present the optimum system
parameters considering only a single objective as these systems have multiple objec-
tives such as power output, system efficiency, investment cost, and economic and
ecological factors. Hence, researchers have developed various optimization models
of these systems and presented optimum system parameters through single- and
multi-objective optimization using advanced optimization algorithms.
All evolutionary computation and swarm intelligence-based optimization algo-
rithms are population-based algorithms and have control parameters such as popula-
tion size, crossover probability, mutation probability, scaling factor, inertia weight,
social and cognitive parameters, among others. Appropriate adjustment of the control
parameters dictates the algorithm convergence toward the global optimum. Inappro-
priate adjustment of the control parameters leads to premature convergence and
increased computational efforts. Also, selecting an appropriate population size for
different optimization applications is a tedious job. Additionally, in multi-objective
optimization, selecting the most suitable solution from non-dominated solution set
is difficult.
In this book, recently developed Jaya and Rao (Rao-1, Rao-2, and Rao-3) algo-
rithms are described for single- and multi-objective optimization of selected renew-
able energy systems. In addition, variants of the Jaya and Rao algorithms are
presented to show the improvement in performances. Furthermore, variants of the

v
vi Preface

Jaya algorithm, namely multi-team perturbation guiding Jaya (MTPG-Jaya) algo-

rithm and adaptive multi-team perturbation guiding Jaya (AMTPG-Jaya) algorithm,
and variants of the Rao algorithms, namely elitist Rao (ERao-1, ERao-2, and ERao-3)
algorithms and self-adaptive population Rao (SAP-Rao) algorithm, are demonstrated
for optimization of the selected renewable energy systems. These algorithms have
no algorithm-specific parameters and require only the common control parameters.
Additionally, the applicability of multi-attribute decision-making methods in multi-
objective optimization problems is discussed, and a decision-making procedure is
recommended based on average rank to identify the best solution in a Pareto-front.
The Jaya and Rao algorithms and their variants are developed by our team, and
these are gaining wide acceptance in the optimization research community. After its
introduction in 2016, the Jaya algorithm is finding many applications in different
fields of science and engineering. The major applications, as of December 2021, are
found in the fields of electrical engineering, mechanical design, thermal engineering,
manufacturing engineering, civil engineering, structural engineering, computer engi-
neering, electronics engineering, physics, chemistry, biotechnology, and economics.
Many research papers have been published in various reputed international journals
of Elsevier, Springer-Verlag, Taylor & Francis, and IEEE Transactions, in addition
to those published in the proceedings of international conferences. The number of
research papers is continuously increasing at a faster rate. The Jaya algorithm and its
variants have carved a niche in the field of advanced optimization, and many more
researchers may find this as a potential optimization algorithm. The Rao algorithms
are developed by our team in 2020, and these are also gaining acceptance in the
optimization research community.
This book presents a comprehensive review on latest research and development
trends at international level for parameter optimization of various renewable energy
systems. Using examples of various renewable energy systems, the possibilities for
parameter optimization with Jaya and Rao algorithms including their variants are
demonstrated. The book presents real case studies, results of applications of the basic
Jaya and Rao algorithms, and their variants and comparison with other advanced
optimization techniques and highlights the best optimization technique to achieve
best performance. The book also includes the validation of different variants of the
Jaya and Rao algorithms through application to complex single- and multi-objective
unconstrained benchmark functions. The algorithms and computer codes of different
versions of Jaya and Rao algorithms are included in the book that will be very much
useful to the readers. This book is expected to become a valuable reference for those
wishing to do research on the use of advanced optimization techniques for solving
single-/multi-objective combinatorial optimization problems related to the renewable
energy systems.
We are grateful to Dr. Anthony Doyle and his team of Springer for their support and
help in producing this book. We wish to thank various researchers and the publishers
of international journals for publishing the research works of our team. Our special
thanks are due to the director and the colleagues at Sardar Vallabhbhai National
Institute of Technology, Surat, India. While every attempt has been made to ensure
that no errors (printing or otherwise) enter the book, the possibility of these creeping
Preface vii

into the book is always there. We will be grateful to the readers if these errors are
pointed out. Suggestions for further improvement of the book will be thankfully
acknowledged.

Surat, India Venkata Rao Ravipudi

Hyderabad, India Hameer Singh Keesari
December 2021
Contents

1 Introduction to Renewable Energy Systems . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Solar Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Wind Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Hydroenergy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Ocean Thermal Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Geothermal Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.6 Bioenergy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.7 Nuclear Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.8 Other Emerging Renewable Energy Technologies . . . . . . . . . . . . . . . 7
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2 Selected Renewable Energy Systems and Formulation of Their
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1 Wind Farm Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.1 Wake Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.2 Power Generation Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1.3 Cost Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1.4 Objective Function of the Wind Farm Layout
Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2 Solar Assisted Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.1 Solar-Assisted Brayton Heat Engine System . . . . . . . . . . . . . 14
2.2.2 Solar-Assisted Stirling Heat Engine System . . . . . . . . . . . . . . 17
2.2.3 Solar-Assisted Carnot-Like Heat Engine System . . . . . . . . . . 22
2.3 Bio-Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.3.1 Single-Cylinder Direct-Injection Diesel Engine . . . . . . . . . . . 24
2.3.2 Turbocharged DI Diesel Engine . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3.3 Compression Ignition Biodiesel Engine with an EGR
System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.3.4 Microalgae-Based Biomass Cultivation Process . . . . . . . . . . 28
2.4 Hydro Energy and Geothermal Energy Systems . . . . . . . . . . . . . . . . . 29
2.4.1 Hydropower Generation and Reservoir Operation . . . . . . . . . 29

ix
x Contents

2.4.2 Ground Source Heat Pump-Radiant Ceiling Air

Conditioning System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3 Advanced Engineering Optimization Techniques and Their
Role in Energy Systems Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4 Working of Jaya and Rao Optimization Algorithms and Their
Variants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.1 Working of the Jaya Algorithm and Its Modified Versions . . . . . . . . 53
4.1.1 Jaya Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.1.2 Multi-team Perturbation-Guiding Jaya (MTPG-Jaya)
Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.1.3 Adaptive Multi-team Perturbation-Guiding Jaya
(AMTPG-Jaya) Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.1.4 Multi-objective Jaya and Multi-objective
AMTPG-Jaya Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.2 Working of the Rao Algorithms and Their Modified Versions . . . . . 69
4.2.1 Rao Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.2.2 Multi-objective Rao Algorithms . . . . . . . . . . . . . . . . . . . . . . . . 70
4.2.3 Elitist Rao Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.2.4 Multi-objective Elitist Rao Algorithms . . . . . . . . . . . . . . . . . . 74
4.2.5 Self-Adaptive Population Rao (SAP-Rao) Algorithm . . . . . . 74
4.2.6 Multi-objective SAP-Rao Algorithms . . . . . . . . . . . . . . . . . . . 79
4.3 Performance Indicators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.3.1 Coverage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.3.2 Spacing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.3.3 Hypervolume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.3.4 Inverted Generational Distance (IGD) . . . . . . . . . . . . . . . . . . . 81
4.4 Computational Results Analysis on Single-objective
Optimization of Unconstrained Benchmark Problems . . . . . . . . . . . . 82
4.4.1 Computational Results Analysis on 30 Unconstrained
Standard Benchmark Problems . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.4.2 Computational Results Analysis on Unconstrained
Unimodal and Multimodal Standard Benchmark
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
4.5 Computational Results Analysis on Multi-objective
Optimization Benchmark Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5 Multi-attribute Decision-Making Methods and Their
Implementation in Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
5.1 Simple Additive Weighing (SAW) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
5.2 Weighted Product Method (WPM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
Contents xi

5.3 Preference Ranking Organization Method for Enrichment

Evaluation (PROMETHEE) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
5.4 Technique for Order Preference by Similarity to Ideal
Solution (TOPSIS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
5.5 Modified TOPSIS (MTOPSIS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
5.6 Compromise Ranking Method (VIKOR) . . . . . . . . . . . . . . . . . . . . . . . 135
5.7 Complex Proportional Assessment (COPRAS) . . . . . . . . . . . . . . . . . . 136
5.8 Gray Relational Analysis (GRA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
6 Optimization of Wind Farm Layouts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
6.1 Problem Definition and Wind Scenarios of the Wind Farm
Layout Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
6.2 Case-I: Fixed Wind Speed and Fixed Direction . . . . . . . . . . . . . . . . . 142
6.3 Case-II: Fixed Wind Speed and Variable Wind Direction . . . . . . . . . 145
6.4 Case-III: Fixed Wind Speed and Variable Wind Direction . . . . . . . . 147
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
7 Optimization of the Selected Solar-Assisted Energy Systems . . . . . . . . 159
7.1 Optimization of a Solar-Assisted Brayton Heat Engine System . . . . 159
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System . . . . 176
7.2.1 Case Study-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
7.2.2 Case Study-2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
7.2.3 Case Study-3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
7.3 Optimization of a Solar-Assisted Carnot-like Heat Engine
System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
8 Optimization of the Selected Bio-Energy Systems . . . . . . . . . . . . . . . . . . 249
8.1 Design Optimization of the Single-Cylinder Direct-Injection
Diesel Engine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
8.2 Design Optimization of a Turbocharged DI Diesel Engine . . . . . . . . 273
8.3 Design Optimization of a Compression Ignition Biodiesel
Engine with an EGR System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
8.4 Process Optimization of a Microalgae-Based Biomass
Cultivation Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
9 Optimization of Hydroenergy and Geothermal Energy Systems . . . . . 329
9.1 Optimization of a Hydropower Generation System . . . . . . . . . . . . . . 329
9.2 Optimization of a Ground Source Heat Pump-Radiant Ceiling
Air Conditioning System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343

Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
Chapter 1
Introduction to Renewable Energy
Systems

Abstract Global warming is a significant concern that raises a need for cleaner
energy production. In the last few decades, researchers have focused on exploiting
renewable energy resources to meet the clean energy demands. This chapter presents
a brief introduction to different renewable energy generation systems and the
associated problems.

1.1 Solar Energy Systems

Solar energy has the greatest potential of all the sources of renewable energy. Solar
power is vastly available and is a low-grade renewable energy. However, the use of
this energy on commercial basis is somewhat facing difficulties due to the following
reasons:
• The solar energy is very dilute and spread out.
• The intensity of radiation is weather dependent and season dependent. The vari-
ation due to the weather and season affects the quantum of radiation received in
a place.
• The solar radiation is available only in daytime. The day and night cycle interrupts
the continuous flow of energy.
Solar thermal energy and solar electric energy are the two ways in which solar
energy can be used. A solar thermal system produces hot water or hot air, cooks
food, dries materials, etc. Solar photovoltaic uses solar radiation in a solar electric
energy system to produce electricity for household appliances and commercial and
industrial buildings. Using a solar-assisted heat engine is a best way of extracting
solar energy. The solar concentrator focuses the sunlight at the concentrator’s focus
point. The receiver located at the focus point of the concentrator absorbs the heat
energy and transfers it to the heat engine that produces the useful work.
Another best way to extract solar energy is through solar photovoltaic energy
systems. Electrical power is generated by converting solar radiation into the flow
of electrons by using photovoltaic (PV) to obtain direct current electricity using
semiconductors that exhibit the photovoltaic effect. Many solar panels comprising

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 1

V. R. Ravipudi et al., Design Optimization of Renewable Energy Systems Using Advanced
Optimization Algorithms, Green Energy and Technology,
https://doi.org/10.1007/978-3-030-95589-2_1
2 1 Introduction to Renewable Energy Systems

many cells containing a semiconducting material are needed to generate direct current
(DC) electricity.
The successful applications of solar energy are: cooling and heating of residential
buildings, solar cookers, solar drying of animal and agricultural products, solar water
heating, evaporation of seawater to produce salt, solar refrigeration and air condi-
tioning, food refrigeration, solar distillation on a small scale, solar engines for water
pumping, solar television, radio and tube lights, solar calculators, solar furnaces,
solar drying, photovoltaic conversion, solar thermal power station, etc.
Solar energy systems are safe and have a well-established technology now
compared to the other technologies related to electricity generation. Solar energy
systems generally have a complex structure, making it a significant task to identify
all possible environmental impacts for life cycle analysis and assessment. Compara-
tively lower ecological risks are associated with photovoltaic cell technologies. Many
characteristics of these systems have been studied by the researchers. The perfor-
mance measures such as economic and ecological aspects, system efficiency, and
output power have been considered and the optimal design and process parameters
are presented (Santos et al., 2017; Kajela & Manshahia, 2017; Khare et al., 2019a;
Al-Shahri et al., 2021).

1.2 Wind Energy Systems

Wind energy can be economically used for the electrical energy generation. The
winds are caused by two factors:
• Heating and cooling of earth’s atmosphere which generates conversion currents.
Heating and cooling are caused by the day and night solar energy cycle. In fact,
it is the sun’s energy which causes the wind.
• The rotation of the earth and its motion around the sun. Due to this, seasonal
changes take place which affects the wind.
The differences in heating of the ground surface by the sun cause the movement of
large air. If the wind speed is between 5 and 25 m/s, then this can be used for electricity
generation. Wind energy conversion systems do electricity generation through wind.
Wind energy conversion systems convert the wind’s kinetic energy into electricity
or other forms of energy.
Nowadays, wind power is a thoroughly established and sustainable branch of
electricity generation. The horizontal axis wind turbine and vertical axis wind turbine
are the two types of wind turbines used in wind energy conversion systems. Many
types of windmills have been designed and developed. However, only a few have
been found to be practically suitable and useful. Some of them are horizontal axis
propeller-type windmills, sail-type horizontal axis windmill, multi-blade type of
horizontal axis windmill, Savonius-type vertical axis windmill, and Darrieus-type
vertical windmill. Among the different systems proposed, the proven and established
1.2 Wind Energy Systems 3

wind energy systems are: (i). simple on-site domestic unit with storage battery and
(ii). simple system with mechanical and solar storage.
The wind speed distribution and the average wind speed affect the power produced
by a wind turbine. The strength of the wind varies from zero to storm force. The
important criteria of power generation are the design of wind turbines and rotor
blades. Each wind turbine design is related to a particular wind velocity to generate
electrical energy (Khare et al., 2019a).
The noise pollution from turbines is significantly reduced in the modern wind
turbine designs. The noise reflects the loss of energy and decreases in efficiency.
Recent advancements in wind turbine design locate the blades upwind in place of
downwind to reduce the level of infrasound and noise pollution (Khare et al., 2019a).

1.3 Hydroenergy Systems

The hydel power being a renewable perennial source of energy plays a vital role
in generation of power. The quantity of hydroelectricity depends on the volume of
water flow and the head created by the water reservoir. Once a dam is constructed, the
electricity can be produced at a constant rate for many decades. During the operation,
greenhouse gases and the atmospheric pollution are not created. However, the natural
environment may be destroyed due to flooding of large land areas. The hydroelectric
dams must be built in areas with the suitable conditions. The major difficulty in
the development of hydropower projects is the relatively longer time required for
its hydrological, topological, and geological investigations. Hydropower projects
generate power at low cost. It is renewable, easy to manage, and pollution free.
However, the major drawback is that it is dependent on the mercy of the nature.
These plants are very suitable for peak load operation along with thermal plants
being operated at base load. Depending on the load, the hydropower stations can be
low head (less than 30 m), medium head (between 30 and 300 m), and high head
(more than 300 m). Depending upon the plant capacity, the hydropower stations
can be microhydel (less than 5 MW), medium capacity (5–100 MW), high capacity
(101–1000 MW), and super plants (greater than 1000 MW).
A hydroenergy system is considered a renewable as well as a non-renewable
energy system. Small hydroprojects have a relatively low environmental impact
compared to a large hydroproject because they frequently have small reservoirs and
civil construction work (Khare et al., 2019a).
The hydropower scheduling is a complex optimization problem that includes
nonlinear dynamical hydraulic heads, nonlinear flows, and the interactional relation-
ships of nonlinear input and output variables. The objective is to obtain optimal
hydroresources available for maximum hydroelectric generation given a set of
starting conditions and many complex constraints. The researchers have discussed
the operation of reservoir and interconnected systems and developed the mathemat-
ical models for the optimum operation of reservoir for the power generation. The
4 1 Introduction to Renewable Energy Systems

parameters such as reservoir level, time of release water, and flow of the stream are
generally included in the mathematical models.

1.4 Ocean Thermal Energy Systems

A large amount of solar energy is collected and stored in tropical oceans. The heat
contained in the oceans is converted into electricity by utilizing the fact that the
temperature difference between the warm surface waters of the oceans and the water
in the depths is about 20-25 °C. The basis of ocean thermal energy conversion
(OTEC) systems lies on proper utilization of this storage with its associated temper-
ature difference and its conversion into work. The surface water which is at higher
temperature could be used to heat some low boiling organic fluid, the vapors of which
would run a heat engine. The exit vapor would be condensed by pumping cold water
from the deeper regions. The amount of energy available for ocean thermal power
generation is enormous and is replenished continuously.
An OTEC plant operates based on a Rankine cycle, often with a refrigerant as the
working fluid. The required heat exchanger areas of the evaporator and condenser
and the water mass flow rates are very large and reduce the overall efficiency of the
OTEC systems. The disadvantages of OTEC include low thermodynamic efficiency
due to very low temperature difference, the size of boilers and condensers becomes
big, capital cost goes up, and onshore installation requires long and big piping which
add up to the cost. The materials suggested for heat exchangers are titanium or alloy
of copper and nickel which are resistant to corrosion, but then the cost adds up.
The efficiency of an OTEC plant can be increased by using multiple cascaded
stages instead of only utilizing a single stage. Upshaw et al. (2011) provided a
methodology for analyzing the major plant design parameters and determining the
optimal design of OTEC facilities. To take advantage of the potential of the ocean as
an unlimited energy source, many researchers are working on developing new and
efficient solutions. The optimization of condensers, vaporizers, heat exchangers, and
turbines is very important. The optimization methodology can be developed from
thermodynamic theory, new materials, and engineering designs.

1.5 Geothermal Energy Systems

Energy present as heat in the earth’s crust (about 10 km depth) is potentially useful.
This heat is apparent from increase in temperature of the earth with increasing depth
below the surface. Although high and low temperatures occur, the average temper-
ature at a depth of 10 km is present underneath the earth’s surface at depths greater
than about 200 °C. Hot molten or partially molten rock, called magma, with temper-
ature of about 3000 °C is present underneath the earth’s surface at depths greater than
24–40 km. The crust ranging in thickness from 15 to 159 km insulates the surface
1.5 Geothermal Energy Systems 5

of the earth and protects it. However, at certain locations, where thickness is small
and the crust is cracked, the subsoil water comes in contact with the hot matter and
gets converted into steam or hot water. In some places, the steam/hot water issues
under pressure. The steam coming under high pressure or water at high tempera-
ture can be used in conventional steam turbines for power generation or for heating
purposes. Four different geothermal systems have been developed/proposed, and
they are: (i). dry steam system, (ii). wet steam system, (iii). hot dry rack system, and
(iv). magmatic system. Some of the advantages of geothermal energy are: versatility
in use, continuous availability, less pollution, and low capital and generation cost
compared to nuclear and coal plants. Some of the disadvantages include overall low
efficiency, noisy drilling operations, corrosion of components due to salt, thermal
pollution due to effluent if not reinjected, nuisance due to gaseous effluents, less
life span compared to nuclear and coal plants, and requirement of large areas for
exploitation of geothermal energy.
The heat energy can be used either for heating or for making electricity, depending
on the temperature of the hydrothermal resource. Low-temperature geothermal
energy is used to heat and cool homes and public apartments, and grow crops and dry
lumber, vegetables, and fruits. To generate electrical energy, hydrothermal resources
at high temperatures can be used.
Various designs have been proposed to improve the energy conversion efficiency
of geothermal energy systems and their optimization (Lee et al., 2019). The principles
of basic and state-of-the-art technologies are essential for developing the advanced
energy systems. A comprehensive review of the geothermal energy systems must be
carried out from systems design, analysis, and optimization.
The geothermal energy systems can be expected to proliferate soon to meet the
demand of increasing worldwide primary energy demands coupled with the need for
fossil fuel replacement. Improving the conversion efficiency is an important issue
of the geothermal energy systems for the optimum use of geothermal resources.
For efficient geothermal energy systems development, a deep understanding of the
conversion principles and configurations of the existing technologies is helpful.

1.6 Bioenergy Systems

Biomass is a biological material derived from living or recently lived organisms.

Electricity generation through a biomass system is a renewable energy source. Biofuel
is considered as an environmentally friendly and sustainable alternative to petroleum-
based fuels. Biofuels are produced from sugar, starch, vegetable oil, animal fats,
switchgrass, cereals, agricultural waste, and microalgae. Biodiesel engine system
uses the blend of biofuel and petrodiesel as the fuel. The performance of the biodiesel
engine has been considerably affected by the blending proportions. Using biodiesel
as an alternative to diesel can reduce the wear of the engine parts and the emissions
such as carbon oxides, hydrocarbons, and particulate matter. However, there is a
6 1 Introduction to Renewable Energy Systems

loss of power, lower fuel efficiency, and increased nitrogen oxides emission with
biodiesels.
Food sources such as animal fats, sugar, starch, and vegetable oil are the first-
generation biofuels. The second-generation biofuels are derived from cellulosic
biomass, such as agricultural waste, switchgrass, and cereals, using thermochem-
ical processes. Producing biofuels from microalgae is gaining popularity nowadays
(Agarwal et al., 2017; Banerjee et al., 2016; Dickinson et al., 2017). Reducing exhaust
emissions and enhancing engine performance of diesel engines is a crucial area in
biodiesel research (Damanik et al., 2018).
Biomass energy plants are like the conventional power plants because both involve
the combustion of a feedstock to generate electrical energy. Biomass energy systems
are like conventional power plants in groundwater use and air emissions. Valuable
biomass resources include energy crops that do not compete with food crops for land,
portions of crop residues such as wheat straw or corn stover, clean municipal and
industrial wastes, and sustainably harvested wood and forest residues (Khare et al.,
2019a).

1.7 Nuclear Energy Systems

The use of nuclear energy is considered as a source of power and as an appropriate

alternative to meet the ever-increasing demand. Opposition to the installation of
this type of plants is mainly due to the fear of radiation hazards. 92 U235 is one of
the elements whose nucleus easily fissions. Other naturally available elements are
stable, and the nucleus of them is not split easily. When a neutron enters the nucleus of
92 U235, the nucleus splits into two fragments and releases 2 to 3 neutrons per fission.
The difference in binding energy between the products of fission and the original
nucleus is involved during the fission reaction. This process is called as nuclear
fission. The neutrons released during the fission cause further fission reactions of
uranium nucleus, and a chain reaction is maintained.
The chain reaction under uncontrolled condition is known as the atomic explosion.
The breaking of U235 nucleus by neutron absorption can take place in several ways.
Each way of splitting U235 nucleus ejects different number of neutrons. On an average,
2.5 neutrons are ejected per neutron absorbed. Nearly 0.3 neutron is lost due to
escape at the surface out of 2.5 neutrons, and the remaining 2.2 neutrons are allowed
to continue chain reaction. Enormous amount of heat energy will be evolved as the
reaction rate will increase exponentially. Such reaction is known as uncontrolled
chain reaction. It is called controlled chain reaction when only one neutron after
every fission is allowed to continue to cause fission reaction. This type of reaction is
used in nuclear reactors, and the energy evolved remains at a constant level. Usually
out of 2.5 neutrons per fission, about 0.8 neutron is absorbed by U238 converting into
a fissionable material Pu239 , 0.5 neutron is absorbed by the control rod, 0.2 neutron
is lost from the reaction, and the chain reaction is maintained by the remaining 1
neutron. If the number of neutrons for next reaction is less than 1 and absorption is
1.7 Nuclear Energy Systems 7

more than above, then the fission reaction will die out very quickly (Dubey et al.,
2013).
The nuclear fuels are mainly classified as fissile fuels such as 92 U235, 92 U233,
and 94 Pu239 , and fertile fuels such as 92 U238 and 90 Th232 . The basic components of
a nuclear reactor of a power plant include fuel element such as natural or enriched
uranium, moderator such as ordinary water or heavy water or graphite or beryllium,
reflector, coolant such as ordinary water or heavy water or carbon dioxide, control
rods to start and maintain the nuclear chain reaction or to shut down the reactor
automatically under emergency conditions, shielding, and a reactor vessel. Different
types of reactors are in use such as ordinary water-cooled reactor (either boiling
water reactor or pressurized water reactor) and heavy water-cooled and moderated
reactor.
The advantages of nuclear energy systems include less pollution, no contribution
to global warming, very low fuel costs, long lifetime of the plant, and generation of
large amount of electricity with a very small amount of nuclear fuel. The disadvan-
tages include disposal of radioactive waste (hazardous to humans and the environ-
ment), high costs of building and safe decommissioning, catastrophic accidents, and
local thermal pollution from wastewater that may affect the marine life.

1.8 Other Emerging Renewable Energy Technologies

Wave energy is a renewable energy system that extracts energy directly from surface
waves. There is a huge amount of energy in ocean waves to provide up to 3 TW of
electricity. However, the big limitation is that the wave energy cannot be harnessed
everywhere.
Wave energy can be measured as an intense form of solar and wind energy. The
differences in heating of the earth by the sun produce winds, and when air passes
over open waterbodies, the wind is converted into waves. Such a mechanism is used
to produce electrical energy with the help of wave energy. The major advantage of
wave power is that it is easily predictable and can be used to calculate the amount of
power that it can produce. However, wave power is in the early stages of development
and costs involved may be higher (Khare et al., 2019a).
Tidal power or tidal energy is a type of hydropower that changes the energy
acquired from tides in valuable types of energy. The tides in the ocean are due to
the gravitational effect of the sun and moon on the earth. The effect of this force is
apparent in the motion of water which shows a periodic rise and fall in levels which
is in rhythms with the daily cycle of rising and setting of sun and moon. This periodic
rise and fall of the water level of sea is called tide. These tides can be used to produce
electric power which is known as tidal power.
The use of tides for electric power generation is practical in a few favorable
situated sites where the geography of an inlet or bay favors the construction of a
large-scale hydroelectric plant. A dam is to be built across the mouth of the bay to
harness the tides. Water streams from the ocean into the tidal bowl through the water
8 1 Introduction to Renewable Energy Systems

turbine at high tide. The stature of the tide is over that of the tidal bowl. Consequently,
the turbine unit works, and the generator coupled to the turbine produces the power
(Khare et al., 2019b).
The tidal energy has significant potential to meet the future power requirements.
Tidal energy offers many advantages such as availability throughout the year irre-
spective of weather conditions, clean energy without pollution, requirement of small
land area, and meeting the power requirements of local coastal areas. However, tidal
energy is also linked to some disadvantages such as highly variable tidal range, limited
power availability, suitable only for selected locations, and corrosion of machinery
due to seawater.
The ocean currents are highly predictable. This makes it easier to construct tidal
energy systems with the correct dimensions as the kind of power the equipment will
be exposed to its known. It is effective at low speeds because of the higher density
of water. However, it still has some environmental effects, and the tidal energy is
considered reliable only when adequate energy storage solutions are available.
In the future, the number of renewable sources will continually increase with
the increase in the demand for more energy and that may reduce the price of the
renewable energy.

References

Agarwal, A. K., Agarwal, R. A., Gupta, T., & Gurjar, B. R. (2017). Introduction to Biofuels. In
Biofuels, Springer. https://doi.org/10.1007/978-981-10-3791-7_1
Al-Shahri, O. A., Ismail, F. B., Hannan, M. A., Lipu, M. S. H., Al-Shetwi, A. Q., Begum, R. A.,
Al-Muhsen, N. F. O., & Soujeri, E. (2021). Solar photovoltaic energy optimization methods,
challenges and issues: A comprehensive review. Journal of Cleaner Production, 284, 125465.
https://doi.org/10.1016/j.jclepro.2020.125465
Banerjee, A., Guria, C., & Maiti, S. K. (2016). Fertilizer assisted optimal cultivation of microalgae
using response surface method and genetic algorithm for biofuel feedstock. Energy, 115, 1272–
1290. https://doi.org/10.1016/j.energy.2016.09.066
Dickinson, S., Mientus, M., Frey, D., Amini-Hajibashi, A., Ozturk, S., Shaikh, F., & El-Halwagi,
M. M. (2017). A review of biodiesel production from microalgae. Clean Technologies and
Environmental Policy, 19(3), 637–668. https://doi.org/10.1007/s10098-016-1309-6
Damanik, N., Ong, H. C., Tong, C. W., Mahlia, T. M. I., & Silitonga, A. S. (2018). A review on the
engine performance and exhaust emission characteristics of diesel engines fueled with biodiesel
blends. Environmental Science and Pollution Research, 25(16), 15307–15325. https://doi.org/10.
1007/s11356-018-2098-8
Dubey, S. P., Memon, A. A., & Bhatt, M. K. (2013). A basic course in mechanical engineering.
New Popular Prakashan, Surat, India.
Kajela, D., & Manshahia, M. S. (2017). Optimization of renewable energy systems : A review.
International Journal of Scientific Research in Science and Technology, 3(8), 769–795.
Khare, V., Khare, C., Nema, S., & Baredar, P. (2019a). Introduction to energy sources. Tidal Energy
Systems. https://doi.org/10.1016/b978-0-12-814881-5.00001-6
Khare, V., Khare, C., Nema, S., & Baredar, P. (2019b). Introduction of tidal energy. Tidal Energy
Systems.https://doi.org/10.1016/b978-0-12-814881-5.00002-8
Lee, I., Tester, J. W., & You, F. (2019). Systems analysis, design, and optimization of geothermal
energy systems for power production and polygeneration: State-of-the-art and future challenges.
References 9

Renewable and Sustainable Energy Reviews, 109, 551–577. https://doi.org/10.1016/j.rser.2019.

04.058
Santos, S. F., Fitiwi, D. Z., Shafie-khah, M., Bizuayehu, A. W., & Catalão, J. P. S. (2017). Introduction
to renewable energy systems. Optimization in Renewable Energy Systems: Recent Perspectives,
1–26. https://doi.org/10.1016/B978-0-08-101041-9.00001-6
Upshaw, C. R., & Webber, M. E. (2011). Integrated thermal-fluids system modeling of an ocean
thermal energy conversion power plant for analysis and optimization. In ASME 2011 5th Inter-
national Conference on Energy Sustainability, Parts A, B, and C, (pp. 1255–1264). https://doi.
org/10.1115/ES2011-54595
Chapter 2
Selected Renewable Energy Systems
and Formulation of Their Problems

Abstract This chapter presents the details of the case studies of selected renew-
able energy systems along with their problem formulations. The selected renewable
energy system case studies include wind farm layout optimization for different wind
scenarios, a case study of solar-assisted Brayton heat engine system, three case studies
of solar-assisted Stirling heat engine system, a case study of solar-assisted Carnot-
like heat engine system, three case studies of biodiesel engine, a case study of the
microalgae cultivation process, a case study of hydropower energy generation, and
a case study of a ground source heat pump-radiant ceiling air conditioning system.

2.1 Wind Farm Layout

A wind farm is a large area where several wind turbines are installed as a cluster. The
wind farm layout optimization problem (WFLO) aims to minimize the cost of energy.
It can be accomplished by optimally placing the number of wind turbines available
such that maximum possible energy can be extracted. In the process of extracting
the kinetic energy of wind, wind turbines decrease the wind speed at the back of the
rotor and swirl the airflow; this is called wind turbine wake effect. Consequently,
wind turbines in the wake effect of another turbine receive a modified wind inflow
in terms of mean velocity and turbulence, which causes lesser energy production.

2.1.1 Wake Model

The Jensen wake model is used to determine the growth of the wake effect created
by an upstream turbine on a downstream turbine. The wind velocity in the wake is
affected by the distance between the upstream turbines to the downstream turbine
(Shakoor et al., 2016). The assumptions made by Mosetti et al. (1994) and Grady
et al. (2005) are followed in this book. According to Jensen wake model, the incident
wind velocity (u) of a downstream wind turbine at a distance x as shown in Fig. 2.1
is given by the following equations (Moorthy & Deshmukh, 2013):

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 11

V. R. Ravipudi et al., Design Optimization of Renewable Energy Systems Using Advanced
Optimization Algorithms, Green Energy and Technology,
https://doi.org/10.1007/978-3-030-95589-2_2
12 2 Selected Renewable Energy Systems …

Fig. 2.1 Wake behind a wind turbine


  
kx 2
u/u 0 = 1 − 2a/ 1 + (2.1)
r1


1−a
r1 = rd (2.2)
1 − 2a

C T = 4a(1 − a) (2.3)

Where u0 is the incident wind speed at the upstream turbine, a is axial induction
factor, k is Entrainment constant, r 1 is downstream rotor radius, C T is turbine thrust
coefficient, and r d is the rotor radius. Let z be the hub height of the turbine and z0
be the surface roughness, then k is given by:
z
k = 0.5/ ln (2.4)
z0
2.1 Wind Farm Layout 13

2.1.2 Power Generation Model

The algorithm proposed by Chowdhury et al. (2012) to calculate the total power
produced by the wind farm is used. The power generated by turbine j for an incident
velocity u is then given by:

1
P j = η ρ Au 3 (2.5)
2

Taking efficiency of wind turbine (η) as 40%, air density (ρ) as 1.2 kg/m3 and
swept area of wind turbine,

P j = 0.3u 3 kW (2.6)

The total power produced by the farm (Ptotal ) consisting of N turbines is the
summation of power produced by the individual wind turbines.


N
Ptotal = Pj (2.7)
j=1

Farm efficiency ( ηfarm ) is given by Eq. 2.8. Let the power produced by a turbine
when it has encountered wind with full velocity be P0 j .

Ptotal
ηfarm =  N (2.8)
j=1 P0 j

2.1.3 Cost Model

The total cost of the wind farm is the same as that used by Mosetti et al. (1994). For
a wind farm consisting of N turbines, the total cost per year is given by the following
equation:

2 1
Cost total = N ( + e−0.00174N )
2
(2.9)
3 3
14 2 Selected Renewable Energy Systems …

2.1.4 Objective Function of the Wind Farm Layout

Optimization

The objective function of this WFLO problem has been chosen to minimize the cost
of unit power generated, which is given by:

Objective Function = Costtotal /Ptotal (2.10)

It can be observed from the objective of the WFLO model considered in this book,
that the objective function is independent of the time variable.

2.2 Solar Assisted Energy Systems

Six multi-objective optimization case studies of solar assisted energy systems are
considered for optimization to see if there can be any improvement in the perfor-
mances of the selected systems. The subsequent section presents the solar-assisted
Brayton heat engine system.

2.2.1 Solar-Assisted Brayton Heat Engine System

The schematic diagram of the solar-assisted regenerative Brayton heat engine system
considered in this book is shown in Fig. 2.2. This system is driven by the combination
of solar energy and fossil fuel. The heat absorbed at the receiver is used to heat the
working fluid before entering the combustion chamber. Hence, less fuel is burned to
heat the working fluid. The solar energy from the receiver is used as the supplement

Fig. 2.2 Schematic diagram of a solar-assisted Brayton heat engine system

2.2 Solar Assisted Energy Systems 15

Fig. 2.3 T-S diagram of a solar-assisted Brayton heat engine system

to the fuel of a solar-assisted Brayton heat engine. Furthermore, the waste heat of
working fluid exhausted from the gas turbine is used to preheat the working fluid
coming out of the compressor. The working fluid from the combustion chamber
expands rapidly and drives the turbine, which in turn drives the alternator Li et al.
(2015).
The T-S diagram of the solar assisted regenerative Brayton engine is shown in
Fig. 2.3. The Brayton cycle (with ideal regenerator) consists of four processes.
Process 1-2: an isobaric heat addition. During this process, the working fluid is
heated by the regenerator (state 1-5) and then the heat source at T H temperature (state
5-2).
Process 2-3: an isentropic expansion process, in which the hot working fluid drives
a turbine.
Process 3-4: an isobaric heat rejection process in which heat released to the
regenerator (state 3-6) and then to heat sink at T L temperature (state 6-4).
Process 4-1 is an isentropic compression process. During this process, the working
fluid at state 4 is compressed to state 1 by the compressor.
Let, T 1 be the temperature of the working fluid at state 1, T 0 be the ambient
temperature, εH be the hot side heat exchange effectiveness, εL be the cold side heat
exchange effectiveness, εR be the regenerator effectiveness, and C wf be the working
fluid heat capacity rate, then the power output is given by the following equation:
16 2 Selected Renewable Energy Systems …

P = Cw f ε H [TH − (1 − ε R )T1 − ε R a8 ]
− Cw f ε L [(1 − ε R )a8 + ε R T1 − TL ] (2.11)

where,

a1 = (1 − ε H )(1 − ε L )(1 − ε R )ε R (2.12)

a2 = a4 T1 + a5 (2.13)

a3 = a1 T12 + a6 T1 + a7 (2.14)

a4 = (1 − ε H )(1 − ε L )ε2R + (1 − ε H )(1 − ε L )(1 − ε R )2 − 1 (2.15)

a5 = ε H (1 − ε R )(1 − ε L )TH + (1 − ε H )ε L ε R TL (2.16)

a6 = ε H (1 − ε L )ε R TH + (1 − ε H )ε L (1 − ε R )TL (2.17)

a 7 = ε H ε L T H TL (2.18)

a8 = −a2 − a22 − 4a1 a3 /2a1 (2.19)

Now, the Brayton engine thermal efficiency (ηb ) and solar concentrator efficiency
(ηs ) are given by:

ε H [TH − (1 − ε R )T1 − ε R a8 ] − ε L [(1 − ε R )a8 + ε R T1 − TL ]

ηb = (2.20)
ε H [TH − (1 − ε R )T1 − ε R a8 ] + ξ (TH − TL )
1
ηs = η0 − [h c THavg − T0 + eC δ(TH4avg − T04 )] (2.21)
I RC

where, THavg is average absorber temperature, RC is the collector concentrating ratio,

eC is the emissivity factor of the collector, hc is the convection heat transfer coef-
ficient, I is the solar irradiance, δ is the Stefan-Boltzmann constant, and η0 is the
collector optical efficiency. Now, the thermal efficiency of the system is given by:

ηm = ηs ηb (2.22)

The non-dimensional thermo-economic performance function is given by:

2.2 Solar Assisted Energy Systems 17

ε H [TH − (1 − ε R )T1 − ε R a8 ] − ε L [(1 − ε R )a8 + ε R T1 − TL ]

F=   (2.23)
H) ln(1−ε L )
ε H [TH − (1 − ε R )T1 − ε R a8 ] − k ln(1−ε
hH
+ hL

2.2.2 Solar-Assisted Stirling Heat Engine System

As shown in Fig. 2.4, the warm region of the Stirling engine is placed at the focal
point of the concentrator. The concentrator reflects solar energy towards the focal
point. The heat exchanger placed at the warm region of the Stirling engine transmits
this reflected heat energy to the working fluid. The work output from the Stirling
cycle is then utilized to drive a generator that produces electric power. Three case
studies of the solar-assisted heat engine system are considered for optimization in
this book. These case studies are presented in the subsequent sections.

Fig. 2.4 Schematic diagram

of the solar-assisted stirling
heat engine
18 2 Selected Renewable Energy Systems …

2.2.2.1 System Description of Case Study-1

Thermodynamic based optimization of the solar-assisted Stirling heat engine is

considered in case study-1. The schematic diagram and TS diagram of the solar-
assisted Stirling heat engine system considered in this book are shown in Fig. 2.5.
The Stirling cycle (with ideal regenerator) consists of four processes. Process 1–2: an
isothermal (at T c temperature) compression involving heat rejection to the heat sink
(at T L temperature). Process 2–3: an isochoric (volume V 2 = V 3 ) heating process,
in which the temperature of the working fluid rises to T h by the regenerator. Process
3–4: an isothermal (at T h temperature) expansion process, in which heat added to the
working fluid from the heat source (at T H temperature). Process 4–1: an isochoric
(volume V 4 = V 1 ) process in which the regenerator absorbs heat from the working
fluid; thus, the temperature falls from T h to T c . The detailed description and finite-
time thermodynamic analysis of the considered system were presented by Ahmadi
et al. (Ahmadi, Mohammadi, et al., 2013).
Let, n be the number of moles of the working fluid, C v be the specific heat capacity
of the working fluid during the regenerative process, R be the universal gas constant,
λ be the ratio of volume during expansion and compression, M 1 be the regenerative
time constant at the heating region, M 2 be the regenerative time constant at the
cooling region, T H1 and T H2 be the heat source temperature before and after heat
transferred to the working fluid during the isothermal expansion, and T L1 and T L2 be
the heat sink temperature before and after heat transferred from working fluid during
isothermal compression, then the power output of a solar-assisted Stirling system is
given by the following equation:

n R(Th − Tc )lnλ
P= n RTh lnλ+nCv (1−ε R )(Th −Tc ) n RTc lnλ+nCv (1−ε R )(Th −Tc )
C H ε R (TH 1 −Th )+ζ C H ε H (TH4 1 −Th4 )
+ C L ε L (Tc −TL1 )
+ 1
M1
+ 1
M2
(Th − Tc )
(2.24)

Fig. 2.5 a T-S diagram and b schematic representation of the solar-assisted stirling heat engine
2.2 Solar Assisted Energy Systems 19

Now, QL be the net heat absorbed by the heat sink, QH be net heat released by the
heat source, T Havg and T Lavg be the average temperatures of heat source and sink,
ηt be the thermal efficiency of the Stirling engine, and ηs is the thermal efficiency
of the dish collector, then the rate of entropy generation (σ ) and thermal efficiency
(ηm ) of the system are given by the following equations respectively.
 
1 QL QH
σ = − (2.25)
t TL avg THavg

ηm = ηs ηt (2.26)
n R(Th − Tc )lnλ
ηt =
n RTh lnλ + nCv (1 − ε R )(Th − Tc ) + 0.5K 0 {(2 − ε H )TH 1 − (2 − ε L )TL1 + (ε H Th − ε L Tc )}t
(2.27)
1   
ηs = η0 − h THavg − T0 + εδ TH4avg − T04 (2.28)
IC

where T 0 is the ambient temperature, C is the collector concentrating ratio, ε is the

emissivity factor of the collector, h is the convection heat transfer coefficient, I is the
solar irradiance, δ is the Stefan-Boltzmann constant, and η0 is the collector optical
efficiency.

2.2.2.2 System Description of Case Study-2

Thermo-economic based optimization of the solar-assisted Stirling heat engine is

considered in case study-2. The TS diagram of the solar-assisted Stirling heat engine
system considered in this case study is shown in Fig. 2.5. The Stirling cycle consists
of four processes. First: isothermal heat removal process, in which heat is rejected by
the working fluid (at T c temperature) to the heat sink (at T L temperature). Second:
an isochoric heat addition process, in which the temperature of the working fluid is
increased to T h by the regenerator. Third: isothermal heat addition process, in which
heat added to the working fluid (at T h temperature) from the heat source (at T H
temperature). Fourth: an isochoric heat removal process, in which the regenerator
absorbs heat from the working fluid. In this case study, a MOO case is considered
to find the optimal thermo-economic parameters to maximize dimensionless power
and thermal efficiency.
In this MOO case, three objective functions are considered. Detailed thermody-
namic and thermo-economic analysis of the solar-assisted Stirling heat engine is
presented by Ahmadi et al. (2013). The objective functions of the optimization are
thermal efficiency of solar-assisted Stirling system (ηth ), the thermo-economic objec-
tive function (F), and the dimensionless power output (P). All objective functions
are to be maximized. The decision variables of these objective functions are internal
irreversibility parameter (φ), heat transfer area ratio (Ar ), temperature ratio (χ ), the
temperature of the heat source (T H ), the temperature of the working fluid in the
20 2 Selected Renewable Energy Systems …

high-temperature isothermal process (T h ). The thermo-economic objective function

is given by the following equation:

(1 − φχ )
F=      
 
1+ 1−f f Ar φh h χ 1+ 1−f f Ar
TL (TH −Th )
+ Ar h c (χ Th −TL )
+ ψh h A H (1 − χ )(1 + 1− f
f
Ar )
(2.29)
1 1 1
ψ= ( + ) (2.30)
n Rlnλ M1 M2

where, hh is convection heat transfer coefficient for the high-temperature side, hc is

convection heat transfer coefficient for the low-temperature side, AH is heat transfer
area for the heat exchanger of the hot side, f is relative investment cost parameter
of the heat exchanger hot side. Let n be the working fluid mole number, R is the
universal gas constant, λ be the ratio of volume during the regenerative processes,
M 1 be the regenerative time constant at the heating region, M 2 be the regenerative
time constant at the cooling region. Now the dimensionless power output (P) and
thermal efficiency (ηth ) of the Stirling system are expressed using the following
equations:

(1 − φχ )
P=   (2.31)
φh h χ
1
TL (TH −Th)
+ Ar h c (χ Th −TL )
+ ψh h A H (1 − χ )

ηth = ηt ηs (2.32)

where ηt is the thermal efficiency of the Stirling engine, and ηs is the thermal
efficiency of the dish collector, expressed by

(1 − φχ )
ηt =   (2.33)
φχ Th
Th + k0 (TH − TL ) 1
h h A H (TH −Th )
+ Ar A H h h (χ Th −TL )
+ ψ(1 − χ )

1
ηs = η 0 − [h(TH − T0 ) + εδ(TH4 − T04 )] (2.34)
IC

where C is the collector concentrating ratio, h is the convection heat transfer coef-
ficient, T 0 is the ambient temperature, T H is the absorber temperature, ε is the
emissivity factor of the collector, δ is the Stefan-Boltzmann constant, I is the solar
irradiance, and η0 is the collector optical efficiency.
2.2 Solar Assisted Energy Systems 21

2.2.2.3 System Description of Case Study-3

The multi-objective optimization problem considered in this case study was formu-
lated by Dai et al. (2018) for determining the optimal design parameters of a solar-
assisted Stirling heat engine system considering conductive thermal bridging losses,
finite heat transfer rate, and regenerative heat loss. The solar-assisted Stirling heat
engine system considered in this case study consists of two isothermal and two
isochoric processes, as shown in Fig. 2.5. From state-1 to state-2 is an isothermal
compression process at constant temperature T c . During this process, the working
medium rejects heat energy (QL ) to the heat sink at temperature T L . From state-
2 to state-3 is an isochoric heat addition process at constant volume V 1 . During
this process, the working medium absorbs heat energy from the regenerator to attain
temperature T h . From state-3 to state-4 is an isothermal expansion process at constant
temperature T h . During this process, the working medium absorbs heat energy (QH )
from the heat source at temperature T H . From state-4 to state-1 is an isochoric heat
rejection process at constant volume V 2 . During this process, the working medium
rejects heat energy to the regenerator to attain temperature T c .
The detailed thermodynamic analysis and description of the solar-assisted Stirling
engine considered in this case study were given in Dai et al. (2018). By implementing
finite-time thermodynamic analysis presented a theoretical model of this system to
achieve optimum power output (P), system efficiency (η), and ecological coefficient
of performance (COPE ). The proposed model considers finite-rate heat transfer,
thermal bridging, and regenerative losses. Let C v be the constant volume specific heat
capacity of the working medium during the regenerative process; λ be the volumetric
ratio during expansion and compression, n be the number of moles of the working
medium, k 1 and k 2 be the rates of temperature change in hot heating and cooling
regions respectively, R be the universal gas constant, μ2-3 and μ4-1 be the regeneration
coefficients during heat energy addition and heat energy rejection process, t be the
Stirling cycle time, α h and α c be the convective heat transfer coefficient during
expansion and compression process. The power output (P) is given by the following
equation:

n R(Th − Tc )lnλ
P= (2.35)
t
Let σ be the entropy generated, αleak be the coefficient of heat leak, and then the
following equations give the system efficiency (η) and COPE :

n R(Th − Tc )lnλ
η= (2.36)
αleak (TH − TL )t + μ2−3 mCv (Th − Tc ) + n RTh lnλ
n R(Th − Tc )lnλ
EC O P = (2.37)
T0 σ
 
1 − μ2−3 1 − μ4−1 n RTc ln λ
t= + (Th − Tc ) +
k1 k2 αc (Tl − TL )
22 2 Selected Renewable Energy Systems …

mCv TH − μ2−3 Tc − (1 − μ2−3 )Th

+ ln
αh TH − Th
mCv μ4−1 Th − TL + (1 − μ4−1 )Tc n RTh ln λ
+ ln + (2.38)
αc Tc − TL αh (TH − Th )

αleak (TH − TL )t + μ4−1 mCv (Th − Tc ) + nRTl ln λ
σ =
TL

αleak (TH − TL )t + μ2−3 mCv (Th − Tc ) + nRTh ln λ
− (2.39)
TH

2.2.3 Solar-Assisted Carnot-Like Heat Engine System

The detailed description and thermodynamic analysis of the considered solar-driven

heat engine system were presented by Sayyaadi et al. (2015). In this Carnot-like
heat engine system, finite temperature difference for working and maximum ecolog-
ical function conditions are considered to achieve optimum non-dimensional power
output (W ), non-dimensional ecological function (E), and thermal efficiency (η).
The T-S diagram of the solar-assisted Carnot heat engine is shown in Fig. 2.6. From
state-1 to state-2 is an isentropic heat addition process. During this process, working
medium temperature rises to T h from T c . From state-2 to state-3 is an isothermal
expansion process at constant temperature T h . During this process, the working
medium absorbs heat energy (QH ) from the heat source at temperature T H . From
state-3 to state-4 is an isentropic heat rejection process. During this process, working
medium temperature falls from T h to T c . From state-4 to state-1 is an isothermal
compression process at constant temperature T c . During this process, the working
medium releases heat energy (QL ) to the heat sink at temperature T L .
Let α Hr and α Hc be the radiation and convection heat transfer coefficients respec-
tively at the heat source, and α Lc be the convection heat transfer coefficient at the heat
sink, then the design variables of the system are as follows: the temperature ratio is
γ = Th /TH , the convective coefficient ratio is φ = α H,c /α L ,c , the operating temper-
ature ratio is τ = TL /TH , and the heat source to the heat sink allocation parameter is
β = (α H,r /α L ,c )TH3 . Now, the non-dimensional power output (W ), non-dimensional
ecological function (E), and the thermal efficiency (η) are given by the following
equations:
 
W = η β 1 − γ 4 − φ(1 − γ ) (2.40)
 
  β 1 − γ 4 + φ(1 − γ ) − γ + 1
E = β 1 − γ + φ(1 − γ ) × η − τ
4
γ − β 1 − γ 4 − φ(1 − γ )
(2.41)
2.2 Solar Assisted Energy Systems 23

Fig. 2.6 T-S diagram of a solar-assisted Carnot-like heat engine system

 
η = 1 − τ/ γ − β 1 − γ 4 − φ(1 − γ ) (2.42)

2.3 Bio-Energy Systems

Biofuel can be considered as a sustainable and environmentally friendly alternative

to petroleum-based fuels. The performance of the biodiesel engine has considerably
affected by the blending proportions. Using biodiesel as an alternative to diesel
can reduce the wear of the engine parts and the emissions such as carbon oxides,
hydrocarbons, and particulate matter. However, there is a loss of power, lower fuel
efficiency, and increased nitrogen oxides emission with biodiesels.
Three multi-objective optimization case studies of biodiesel engine design and a
multi-objective optimization case study of microalgae cultivation process optimiza-
tion are considered for optimization to see if there can be any improvement in the
performances of the selected systems. The subsequent section presents the design
optimization of a biodiesel engine system.
24 2 Selected Renewable Energy Systems …

2.3.1 Single-Cylinder Direct-Injection Diesel Engine

This case study was presented by Dhingra et al. (2014). In this case study, a
single-cylinder direct-injection compression ignition engine that runs using Jatropha
biodiesel blends was considered. The seven objectives of this case study are: mini-
mizing the combustion parameters such as brake specific fuel consumption (BSFC—
kg/kWh) and peak cylinder pressure (Pmax –bar); maximizing the performance in
terms of brake-thermal efficiency (BTE–N-m); and minimizing the emissions such
as carbon mono oxide emission (CO–%), nitrogen oxides emission (NOx –ppm),
hydrocarbon emission (HC–ppm), and smoke emission opacity (S m ). The design
variables are biodiesel blending ratio (X 1 ), load torque (X 2 ), and compression ratio
(X 3 ) and their ranges are 11.25 ≤ X 1 ≤ 33.75 (% V/V ), 7.5 ≤ X 2 ≤ 12.5 (N-m),
and 13.5 ≤ X 3 ≤ 16.5 (V/V ). The regression models of this case study’s objectives
proposed by Dhingra et al. (2014) are as given below:

B S FC = −46.68493 + 0.13685X 1 + 1.32378X 2 + 5.31712X 3

− 3.21186 × 10−3 X 12 − 0.056480X 22 − 0.16911X 32 + 2.33817 × 10−3 X 1 X 2
− 1.13137 × 10−3 X 1 X 3 − 0.022062X 2 X 3 (2.43)

BT E = −2400.88522 + 10.28829X 1
+ 71.43483X 2 + 259.72937X 3 − 0.18203X 12
− 4.07476X 22 − 9.07056X 32 − 0.19143X 1 X 2
− 4.19026 × 10−3 X 1 X 3 + 1.26091X 2 X 3 (2.44)

Pmax = −3669.50268 + 10.41183X 1

+ 129.7155X 2 + 396.37136X 3 − 0.28479X 12
− 5.68691X 22 − 12.90809X 32 − 0.037712X 1 X 2
+ 0.18856X 1 X 3 − 1.03709X 2 X 3 (2.45)

Sqrt(C O) = −102.47076 + 0.41566X 1 + 1.86746X 2

+ 11.94069X 3 − 7.55818 × 10−3 X 12 − 0.11396X 22
− 0.40724X 32 − 3.75222 × 10−3 X 1 X 2 − 2.57927
× 10−3 X 1 X 3 + 0.035673X 2 X 3 (2.46)

Sqrt(N Ox ) = −765.06345 + 3.33115X 1 + 12.59961X 2

+ 91.17741X 3 − 0.062945X 12 − 0.70774X 22 − 3.08084X 32
− 0.026616X 1 X 2 − 0.01625X 1 X 3 + 0.1666X 2 X 3 (2.47)
2.3 Bio-Energy Systems 25

log10 (H C) = −182.12527 + 0.57186X 1 + 3.99162X 2

+ 21.03397X 3 − 0.012787X 12 − 0.19348X 22 − 0.70153X 32
+ 8.11844 × 10−5 X 1 X 2 + 2.71724 × 10−4 X 1 X 3 + 1.69324 × 10−3 X 2 X 3
(2.48)

Sqrt(Sm ) = 133.78384 − 0.22401X 1 − 3.48682X 2 − 15.04873X 3

+ 0.010848X 12 + 0.18082X 22 + 0.49577X 32 + 3.59009 × 10−3 X 1 X 2
− 5.06832 × 10−3 X 1 X 3 + 0.041469X 2 X 3 (2.49)

2.3.2 Turbocharged DI Diesel Engine

This case study was presented by Shirneshan et al. (2016). In this case study, a
turbocharged DI diesel engine using biodiesel and diesel blends was considered.
The waste vegetable cooking oil was considered as the source of biodiesel. The
specific fuel properties and engine specifications were presented in Shirneshan et al.
(2016). By investigating the effect of design variables such as biodiesel blending
ratio, engine speed, and engine load on the combustion, performance, and emissions
of the considered biodiesel engine, Shirneshan et al. (2016) presented the regression
models. The six objectives of this case study are: maximizing the brake power (P)
and brake torque (T ); minimizing the brake specific fuel consumption (BSFC) and
emissions such as carbon mono oxide emission (CO), nitrogen oxides emission
(NOx ), and hydrocarbon emission (HC). The design variables were the percentage
of biodiesel in fuel (X1 ), engine speed (X2 ), and engine load (X3 ) and their ranges
were 0 ≤ X 1 ≤ 100 (%), 1000 ≤ X 2 ≤ 2800(rpm), and 25 ≤ X 3 ≤ 100 (%). The
regression models of this case study’s objectives are as follows:

P = −47.32 − 0.08X 1 + 0.056X 2 + 0.205X 3 + 0.0002X 12

− 1.4 × 10−5 X 22 − 5.499 × 10−4 X 32 + 0.0004X 1 X 3 + 0.0002X 2 X 3 (2.50)

T = −299.277 − 0.524X 1 + 0.302X 2 + 4.654X 3 + 0.00362X 12

− 7.401 × 10−5 X 22 − 0.00637X 32 − 3.771 × 10−4 X 2 X 3 (2.51)

B S FC = 298.74 + 0.5X 1 − 0.088X 2 − 0.236X 3 + 0.0014X 12

+ 2.67 × 10−5 X 22 − 0.00018X 1 X 2 − 0.00336X 1 X 3 (2.52)

CO = 0.109 − 3.43 × 10−4 X 1 − 3.96 × 10−5 X 2 − 7.53 × 10−4 X 3

+ 6.48 × 10−7 X 12 + 6.32 × 10−9 X 22 + 2.93 × 10−6 X 32
26 2 Selected Renewable Energy Systems …

+ 1.69 × 10−6 X 1 X 3 + 7.33 × 10−8 X 2 X 3 (2.53)

N Ox = 216.71 − 0.264X 1 + 0.158X 2 + 0.755X 3 + 0.0114X 12

− 5.37 × 10−5 X 22 + 0.0558X 32 − 0.00188X 2 X 3 (2.54)

H C = 56.38 − 0.028X 1 + 0.019X 2 − 0.554X 3

− 5.19 × 10−4 X 12 − 2.1 × 10−6 X 22 + 0.0026X 32 + 3.14 × 10−5 X 1 X 2
− 0.0011X 1 X 3 + 8.38 × 10−5 X 2 X 3 (2.55)

2.3.3 Compression Ignition Biodiesel Engine with an EGR

System

This case study was presented by Jaliliantabar et al. (2019). In this case study, a
compression ignition biodiesel engine with an exhaust gas recirculation system is
considered for multi-objective optimization to see if there can be any improvement
in the considered system performance. The design variables of this case study are
the exhaust gas recirculation rate (ER), engine load percentage (EL), engine speed
(ES) in rpm, and biodiesel percentage (BP). The ranges of design variables are as
follows: 0 ≤ E R ≤ 30, 25 ≤ E L ≤ 75, 1800 ≤ E S ≤ 2400, and 0 ≤ B P ≤ 15.
The objective functions of this case study are maximization of power output (P), and
minimization of the brake specific fuel consumption (BSFC) and emissions such as
carbon mono oxide emission (CO), nitrogen oxides emission (N O x ), hydrocarbon
emission (HC), and smoke opacity (Sm ). The regression models of the objectives of
this case study are as follows:

P = 7.08 + (0.04 × EL) − 7.33 × 10−3 × ES

− (0.01 × ER) + (0.11 × BP)
+ 4.73 × 10−6 × EL × ES + 3.7 × 10−5 × EL × ER
− 1.13 × 10−4 × EL × BP + 5.87 × 10−6 × ES × ER
− 4.77 × 10−5 × ES × BP + 6.69 × 10−5 × ER × BP
− 7.88 × 10−5 × EL2 + 1.8 × 10−6 × ES2
+ 5.9 × 10−5 × ER2
− 7.93 × 10−5 × BP2 (2.56)

CO = 5.55 − (0.08 × EL) − 4.3 × 10−3 × ES

− (0.02 × ER) − 1.36 × BP × 10−6
2.3 Bio-Energy Systems 27

+ 2.59 × 10−5 × EL × ES + 5.48 × 10−5 × EL × ER

− 6.2 × 10−4 × EL × BP
+ 9.2 × 10−6 × ES × ER + 3.66 × 10−5 × ES × BP
− 2.35 × 10−4 × ER × BP
+ 5.2 × 10−4 × EL2 + 9.02 × 10−7 × ES2
− 4.28 × 10−5 × ER2
− 6.08 × 10−4 × BP2 (2.57)

NOx = 1514.26 + (17.21 × EL) − (1.61 × ES)

+ (0.51 × ER)
+ (3.73 × BP) − 1.55 × 10−3 × EL × ES
− (0.05 × EL × ER)
− 8.8 × 10−3 × EL × BP + 4.73 × 10−5 × ES × ER
− 1.48 × 10−3 × ES × BP
− (0.07 × ER × BP) − 0.1 × EL2
+ 3.86 × 10−4 × ES2
+ 0.03 × ER2 − 0.07 × BP2 (2.58)

HC = 8.76 − (7.49 × EL) + (0.096 × ES) − (0.17 × ER)

− (0.18 × BP) + 2.33 × 10−3 × EL × ES + (0.01 × EL × ER)
− (0.06 × EL × BP) + 4.69 × 10−4 × ES × ER
− 1.61 × 10−4 × ES × BP
− (0.03 × ER × BP) + 0.05 × EL2
− 2.8 × 10−5 × ES2
− 0.02 × ER2 + 0.09 × BP2 (2.59)

BSFC = −631.23 − (28.02 × EL)

+ (1.5 × ES) + (2.92 × ER)
− (2.12 × BP) + 2.5 × 10−3 × EL × ES
+ 7.07 × 10−3 × EL × ER
− (0.02 × EL × BP) − 2.02 × 10−3 × ES × ER
− 1.6 × 10−3 × ES × BP
− 4.17 × 10−3 × ER × BP + 0.19 × EL2
28 2 Selected Renewable Energy Systems …

− 3.38 × 10−4 × ES2

+ 0.03 × ER2 + 0.37 × BP2 (2.60)

Sm = 15.09 − (0.36 × EL)

− 8.69 × 10−3 × ES − (0.05 × ER)
+ 6.72 × 10−3 × BP + 8.06 × 10−5 × EL × ES
− 1.45 × 10−4 × EL × ER
− 3.8 × 10−3 × EL × BP + 5.57 × 10−5 × ES × ER
− 2.74 × 10−6 × ES × BP
− 5.06 × 10−4 × ER × BP + 4.17 × 10−3 × EL2 − 1.9 × 10−6 × ES2
− 1.04 × 10−3 × ER2 + 5.2 × 10−3 × BP2 (2.61)

2.3.4 Microalgae-Based Biomass Cultivation Process

Nowadays, producing biofuels from microalgae is gaining popularity. This case

study’s objectives include the maximization of biomass production, eicosapentaenoic
acid (EPA), and lipid productions. Banerjee et al. (2016) presented the mathe-
matical models of these objectives considering the cultivation light intensity (X 1
in μmol/m2 /s), temperature (X 2 in °C), and concentrations of NaCl (X 3 in M),
NaHCO3 (X 4 in g/L) and NPK-10:26:26 fertilizer (X 5 in g/L) as the design variables.
The regression models for the biomass-production (BMP), total lipid-production
(TLP), and EPA generations in terms of the considered design variables are as follows:

BMP (g/L) = 0.069 + 0.029X1 − 0.0039X2 − 0.0419X3

+ 0.0014X4 + 0.002X5 + 0.0028X1 X2 − 0.00037X1 X3
+ 0.00068X1 X4 + 0.0026X1 X5 + 0.00037X2 X3 − 0.0014X2 X4
+ 0.00031X2 X5 + 0.0015X3 X4 + 0.0036X3 X5 − 0.00068X4 X5
− 0.0046X21 − 0.0022X22 − 0.023X23 − 0.012X24 − 0.0072X25 (2.62)

TLP (% ) = 40.87 + 5.07X1 + 0.081X2 − 3.49X3 + 1.46X4

+ 2.11X5 − 0.26X1 X2 − 0.67X1 X3 − 0.31X1 X4 − 0.029X1 X5
− 0.25X2 X3 − 0.29X2 X4 − 0.24X2 X5 − 0.19X3 X4 − 0.24X3 X5
− 0.28X4 X5 − 1.61X21 − 0.47X22 − 2.39X23 − 1.35X24 − 1.45X25 (2.63)

EPA (% ) = 19.66345 + 1.50675X1 − 1.26475X2 − 2.04X3 + 0.11X4

2.3 Bio-Energy Systems 29

+ 0.51X5 + 0.081X1 X2 − 0.009X1 X3 + 0.0078X1 X4 + 0.04X1 X5

+ 0.14X2 X3 − 0.096X2 X4 − 0.066X2 X5 + 0.0115X3 X4 − 0.033X3 X5
+ 0.089X4 X5 − 1.42X21 + 0.0219X22 − 0.228X23 − 0.059X24 − 0.474X25 (2.64)

2.4 Hydro Energy and Geothermal Energy Systems

2.4.1 Hydropower Generation and Reservoir Operation

Optimization of the Nigerian Jebba hydropower plant performance characteristics

to enhance its electricity generation case study is considered to demonstrate the
application of advanced optimization algorithms. This case study was presented by
Onokwai et al. (2020). This case study investigates the influence of pressure head and
discharge with other constraints such as the temperature of the turbine, oil pressure
for lubrication and cooling of the bearings, on estimated power generation (E P w ). By
using the response surface methodology (RSM), the effect of the selected variables
on the estimated power generation was measured, and presented regression models
for the optimization of the plant E P w . The detailed plant’s description including
technical specifications and validation of the modal was presented in Onokwai et al.
(2020).
The input variables considered in this case study are the discharge (D), pressure
drop between the head and tail-water (Pd ), the stator temperature (Tst ), the water
pressure (Pwater ) for cooling the generator and the oil pressure (Poil ) for cooling
and lubrication of the bearings. The objective of the RSM based optimization is to
maximize the power generation (E P w ) of the power plant. The RSM based regression
model of the objective is as follows (Onokwai et al. 2020):

E Pw = 5173 + 1.58D + 1636Pd − 1092Tst − 1033Pwater + 884Poil

− 0.000130D 2 − 42.7Pd2 + 3.8Tst2 − 1.3Pwater
2
− 9.76Poil
2
− 0.136Pd D
+ 0.0742Tst D + 0.0106Pwater D − 0.0064Poil D − 2.3Pd Tst + 31.6Pd Pwater
+ 28.7Pd Poil + 12.9Pwater Tst − 8.1Poil Tst − 16.1Pwater Poil (2.65)

2.4.2 Ground Source Heat Pump-Radiant Ceiling Air

Conditioning System

The ground source heat pump-radiant ceiling (GSHP-RC) air conditioning system
case study was presented by Xie et al. (2020). This case study is to investigate
30 2 Selected Renewable Energy Systems …

the coupling mechanism between the system and building thermal environment
and optimize the whole system globally. By using the response surface method-
ology (RSM), the effect of the controllable variables on the output variables was
measured, and presented regression models for global optimization of the system. The
detailed description including technical specifications and validation of the modal
was presented in Xie et al. (2020).
The controllable variables considered in this case study are the water supply
temperature of radiant ceiling (a), the indoor set temperature (b) and the water supply
temperature of the heat pump (c). The objective functions of the system are seasonal
performance factor (SPF), predicted mean value (PMV) and the operating cost (OC).
The seasonal performance factor is the energy consumption evaluation index of the
system, in which the power consumptions of ventilation units, heat pump units and
pumps are all taken into account.
PMV is a thermal comfort evaluation index based on the human body heat balance
equation and the subjective thermal sensation of the human body. The average value
of PMV in the air conditioned rooms is used as an indicator for the thermal comfort
of the system. In order to ensure the thermal comfort of the air-conditioned room,
it is desirable that the average value of the PMV in the air-conditioned room varies
between −0.5 and 0.5. It is better for the value close to 0. The system operating cost
is used as the indicator of the system economy and it is the product of the electricity
price and the total power consumption in the whole cooling season. The RSM based
regression models of these objectives are as follows (Xie et al. 2020):

S P F = 2.422 − 0.01573a + 0.01933b + 0.2157c − 0.01102c2 (2.66)

P M V = −3.75 + 0.4103a − 0.303b + 0.01642b2 − 0.01449ab (2.67)

OC = 77076 + 2412a − 5066b + 1364c + 47.7a 2

+ 120.3b2 − 106.5c2 − 135ab + 0.2bc (2.68)

The following chapter presents a review on the application of advanced engi-

neering optimization techniques to the various renewable energy systems.

References

Ahmadi, M. H., Mohammadi, A. H., Dehghani, S., & Barranco-Jiménez, M. A. (2013). Multi-
objective thermodynamic-based optimization of output power of solar dish-Stirling engine by
implementing an evolutionary algorithm. Energy Conversion and Management, 75, 438–445.
https://doi.org/10.1016/j.enconman.2013.06.030
References 31

Ahmadi, M. H., Sayyaadi, H., Mohammadi, A. H., & Barranco-Jimenez, M. A. (2013). Thermo-
economic multi-objective optimization of solar dish-Stirling engine by implementing evolu-
tionary algorithm. Energy Conversion and Management, 73, 370–380. https://doi.org/10.1016/j.
enconman.2013.05.031
Banerjee, A., Guria, C., & Maiti, S. K. (2016). Fertilizer assisted optimal cultivation of microalgae
using response surface method and genetic algorithm for biofuel feedstock. Energy, 115, 1272–
1290. https://doi.org/10.1016/j.energy.2016.09.066
Chowdhury, S., Zhang, J., Messac, A., & Castillo, L. (2012). Unrestricted wind farm layout optimiza-
tion (UWFLO): Investigating key factors influencing the maximum power generation. Renewable
Energy, 38(1), 16–30. https://doi.org/10.1016/j.renene.2011.06.033
Dai, D., Yuan, F., Long, R., Liu, Z., & Liu, W. (2018). Performance analysis and multi-objective opti-
mization of a Stirling engine based on MOPSOCD. International Journal of Thermal Sciences,
124, 399–406. https://doi.org/10.1016/j.ijthermalsci.2017.10.030
Dhingra, S., Bhushan, G., & Dubey, K. K. (2014). Multi-objective optimization of combustion,
performance and emission parameters in a jatropha biodiesel engine using non-dominated sorting
genetic algorithm-II. Frontiers of Mechanical Engineering, 9(1), 81–94. https://doi.org/10.1007/
s11465-014-0287-9
Grady, S. A., Hussaini, M. Y., & Abdullah, M. M. (2005). Placement of wind turbines using genetic
algorithms. Renewable Energy, 30(2), 259–270. https://doi.org/10.1016/j.renene.2004.05.007
Jaliliantabar, F., Ghobadian, B., Najafi, G., Mamat, R., & Carlucci, A. P. (2019). Multi-objective
NSGA-II optimization of a compression ignition engine parameters using biodiesel fuel and
exhaust gas recirculation. Energy, 187, 115970. https://doi.org/10.1016/j.energy.2019.115970
Li, Y., Liao, S., & Liu, G. (2015). Thermo-economic multi-objective optimization for a solar-dish
Brayton system using NSGA-II and decision making. International Journal of Electrical Power
and Energy Systems, 64, 167–175. https://doi.org/10.1016/j.ijepes.2014.07.027
Mosetti, G., Poloni, C., & Diviacco, B. (1994). Optimization of wind turbine positioning in
large windfarms by means of a genetic algorithm. Journal of Wind Engineering and Industrial
Aerodynamics, 51(1), 105–116. https://doi.org/10.1016/0167-6105(94)90080-9
Moorthy, C. B., & Deshmukh, M. K. (2013). A new approach to optimise placement of wind turbines
using particle swarm optimisation. International Journal of Sustainable Energy, 34(6), 396–405.
https://doi.org/10.1080/14786451.2013.860140
Onokwai, A. O., Owamah, H. I., Ibiwoye, M. O., Ayuba, G. C., & Olayemi, O. A. (2020). Application
of response surface methodology (RSM) for the optimization of energy generation from Jebba
hydro-power plant, Nigeria. ISH Journal of Hydraulic Engineering. https://doi.org/10.1080/097
15010.2020.1806120
Sayyaadi, H., Ahmadi, M. H., & Dehghani, S. (2015). Optimal design of a solar-driven heat engine
based on thermal and ecological criteria. Journal of Energy Engineering, 141(3), 4014012. https://
doi.org/10.1061/(asce)ey.1943-7897.0000191
Shakoor, R., Hassan, M. Y., Raheem, A., & Wu, Y. K. (2016). Wake effect modeling: A review of
wind farm layout optimization using Jensen’s model. Renewable and Sustainable Energy Reviews,
58, 1048–1059. https://doi.org/10.1016/j.rser.2015.12.229
Shirneshan, A., Samani, B. H., & Ghobadian, B. (2016). Optimization of biodiesel percentage
in fuel mixture and engine operating conditions for diesel engine performance and emission
characteristics by artificial bees colony algorithm. Fuel, 184, 518–526. https://doi.org/10.1016/j.
fuel.2016.06.117
Xie, Y., Hu, P., Zhu, N., Lei, F., Xing, L., & Xu, L. (2020). Collaborative optimization of ground
source heat pump-radiant ceiling air conditioning system based on response surface method and
NSGA-II. Renewable Energy, 147, 249–264. https://doi.org/10.1016/j.renene.2019.08.109
Chapter 3
Advanced Engineering Optimization
Techniques and Their Role in Energy
Systems Optimization

Abstract This chapter presents a brief literature review on the optimization of the
selected renewable energy systems and the role of advanced engineering optimization
techniques in energy systems.

Global warming is a significant concern that raises a need for cleaner production
of energy. In the last few decades, researchers have focused on renewable energy
resources like solar energy, bio-energy, wave energy, ocean thermal energy, tidal
energy, geothermal energy, and wind energy. This has resulted in the development of
new techniques and tools that could harvest energy from renewable energy sources.
Among all these sources, the solar, wind, and bio-energy sources have become major
resources upon which the focus is made.
However, to meet energy demands and reduce investment, a rigorous study of
energy extraction systems is required. Identifying, analyzing, and optimizing the
effect of various parameters of a renewable energy system contribute significantly
in assessing the system performance. Furthermore, it is always not preferable to
present the optimum system parameters considering only a single objective as these
systems have multiple objectives such as power output, system efficiency, invest-
ment cost, and economic and ecological factors. Hence, researchers have developed
various optimization models of these systems and presented optimum system param-
eters through single- and multi-objective optimization using advanced optimization
algorithms.
Wind farm layout optimization
Researchers have made several attempts for prediction and optimization of turbines
location in the wind farm to maximize the net power generation. Mosetti et al.
(1994) attempted the genetic algorithm (GA) to find the optimal location of turbines
on a wind farm. Grady et al. (2005) improved the location of turbines by making
proper improvements in the selection of parameters used in the GA. González et al.
(2010) employed a variable-length genetic algorithm with novel crossover proce-
dures to obtain the optimal positioning of wind turbines. Ituarte-Villarreal et al.
(2011) determined the optimal positioning of turbines using viral-based optimiza-
tion (VBO) algorithm. Chowdhury et al. (2012) presented the unrestricted wind farm

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 33

V. R. Ravipudi et al., Design Optimization of Renewable Energy Systems Using Advanced
Optimization Algorithms, Green Energy and Technology,
https://doi.org/10.1007/978-3-030-95589-2_3
34 3 Advanced Engineering Optimization Techniques …

layout optimization (UWFLO) framework that determines the position of turbines

in a wind farm and appropriate turbines installed at the same time. Also, using
the constrained particle swarm optimization (PSO) algorithm, the UWFLO model
was optimized. Chowdhury et al. (2013) enhanced the UWFLO methodology for
designing commercial-scale wind farms and used an advanced mixed-discrete PSO
algorithm to optimize the proposed model. Pookpunt and Ongsakul (2013) proposed
a binary particle swarm optimization (BPSO) for determining the optimal placement
of wind turbines.
Sessarego et al. (2014) used a hybrid non-dominated sorting genetic algorithm
(NSGA-II) to optimize yearly energy production, flap wise root-bending moment,
and wind turbine blade mass concurrently. Similarly, Turner et al. (2014) proposed
mixed-integer linear and quadratic optimization formulations, Moorthy et al. (2014)
proposed GA, and Moorthy and Deshmukh (2013) used PSO to optimize the place-
ment of turbines in a wind farm. Shakoor et al. (2015) proposed modeling of WFLO
to reduce wake effect losses and obtain optimized layout for wind farms. In finding
the optimal arrangement of wind turbines under unidirectional and uniform wind
speed scenario, Patel et al. (2015) have implemented the teaching–learning-based
optimization (TLBO) algorithm, and Ogunjuyigbe et al. (2017) used GA.
DuPont et al. (2016) have presented a modeling system that includes precise
cost and power modeling, partial wake interface, and unstable atmosphere effects.
It is validated by employing within an extended pattern search (EPS) multi-agent
system (MAS) optimization approach. Furthermore, DuPont and Cagan (2016)
proposed hybrid EPS/GA (HEPS/GA) and a multi-objective extended pattern search
for optimizing the multi-stage wind farms layout. Li et al. (2017) proposed the
selection hyper-heuristics technique to solve a multi-objective WFLO problem.
Patel et al. (2017) proposed enhanced versions of the basic teaching–learning-based
optimization algorithm named PAL and AL to optimize the WFLO problem.
Biswas et al. (2018) proposed a decomposition-based multi-objective evolutionary
algorithm for simultaneous optimization of the power output and farm efficiency of
four case studies of the WFLO problem. Pillai et al. (2018) proposed a WFLO
framework for the levelized cost of energy that includes the wind farm’s electrical
infrastructure, annual energy production, and cost as functions of the wind farm
layout and also demonstrated the implementation of the proposed model to three
different cases of the WFLO problems and optimized using the PSO algorithm.
Wang et al. (2019) proposed an optimization model for the WFLO problems
based on mean wind farm power output and variability. Also, the proposed models
are optimized through weighted optimization and confidence interval approaches
using the GA. By making the wake effect on individual turbines uniform, Yang et al.
(2019) proposed and tested a new objective function to increase the power output
and reducing the wake losses. The simulated annealing (SA) algorithm was used for
layout optimization, and its performance was validated for an actual wind farm.
Wu et al. (2020) presented a WFLO model considering wind turbine noise along
with the power output and wake effect. Also, two strategies (strict noise control
strategy and economically compensated control strategy) are presented for designing
3 Advanced Engineering Optimization Techniques … 35

the wind farms, and by applying the PSO algorithm, the proposed models are opti-
mized. Reddy (2020) proposed a framework that provides a large set of analytical
wake models and wake superposition schemes, by considering terrain elevation and
the ambient wind velocity profile. Also, an empirical relation for cost in terms of
turbine rotor diameter and height was presented. So far, it can be observed that
advanced optimization meta-heuristics like GA, enhanced GA, VBO, PSO, BPSO,
hybrid NSGA-II, TLBO, EPS-MAS, HEPS/GA, BBO, SH, AL, and PAL algorithms
have been implemented in finding optimal layout for WFLO.
Optimization of solar-assisted energy systems
Many researchers and scientists have focused upon developing and optimizing the
solar aided heat engines in order to extract and utilize solar energy more efficiently.
Ahmadi et al. (2013a) presented an optimization model of an irreversible solar-driven
heat engine based on thermodynamic and thermo-economic criteria. Also, optimum
designs were obtained using the NSGA-II algorithm to achieve maximum power
and thermal efficiency of the system. Ahmadi et al. (2013b) investigated the Stirling
engine’s application powered by solar energy and proposed a mathematical model
considering regenerative losses, thermal bridging losses, and finite rate heat transfer.
Also, Ahmadi et al. (2013c) studied the effect of the various parameters such as regen-
erator effectiveness, volumetric ratio, absorber temperature, and temperature ratio on
the performance of the solar-powered Stirling system. Furthermore, Pareto-optimal
design points were obtained through the multi-objective optimization (MOO) using
NSGA-II algorithm to achieve maximum power and thermal efficiency. Similarly,
Ahmadi et al. (2013d) presented the optimum design of a solar-powered Stirling
system through the MOO of the thermo-economic function, thermal efficiency, and
dimensionless power output.
Li et al. (2015) presented optimum design parameters of an integrated solar-
assisted irreversible open Brayton engine system with thermal bridging losses and
an imperfect solar collector using the NSGA-II algorithm. Furthermore, the NSGA-II
algorithm was employed by Sadatsakkak et al. (2015) for thermo-economic optimiza-
tion of an irreversible closed regenerative Brayton cycle by taking the power output,
ecological, and thermo-economic functions as MOO goals. Ahmadi et al. (2016d)
presented a thermodynamic analysis of combined Stirling and Otto cycle for power
production, and MOO was performed for optimal power output and thermal effi-
ciency of the Stirling engine using the NSGA-II algorithm. Ahmadi et al. (2016c)
optimized a finite speed thermodynamic analysis of Stirling heat engine considering
internal irreversibilities such as fluid friction and imperfect regeneration.
Ahmadi et al. (2016a) presented a thermodynamic model of solar-driven regen-
erative irreversible Brayton cycle system and identified optimal process parame-
ters of the system to achieve maximum power and thermal efficiency using the
NSGA-II algorithm. Ahmadi et al. (2016b) presented the thermodynamic analysis
of a solar-assisted Stirling system with high-temperature differential, and simulta-
neous optimization was carried out using the NSGA-II algorithm. Arora et al. ()
performed multi-objective optimization of a solar-driven Stirling system with regen-
erative losses by employing the NSGA-II algorithm. Bellos et al. (2017) investigated
36 3 Advanced Engineering Optimization Techniques …

the solar energy supported gas turbine system and optimized to attain optimum
collector area, electricity production, and fuel consumption.
Ahmadi et al. (2017) optimized a solar-driven small-scale transcritical CO2 power
cycle system with the liquid natural gas heat sink using the NSGA-II algorithm. Arora
et al. (2017) performed multi-objective optimization using the NSGA-II algorithm to
solar energy-driven Stirling heat engine system with regenerative heat losses, finite
rate heat transfer, thermal bridging losses, and regeneration process time. Dai et al.
(2018) proposed and optimized the finite-time thermodynamic model of Stirling heat
engine by considering finite rate heat transfer, conductive thermal bridging losses,
and regenerative heat losses.
Kim et al. (2018) presented a solar-assisted heat pump system for hot water supply.
The proposed system uses a hybrid solar collector. The hybrid solar collector uses
a solar thermal receiver and air source–receiver to collect solar energy. Further-
more, presented a thermodynamic model of the proposed system and performed
a parametric analysis of the system’s critical parameters. Sanaye and Taheri (2018)
proposed a hybrid liquid desiccant heat pump system and considering energy, exergy,
environmental, and economic aspects optimized using the MOGA. Total annual cost
and exergy efficiency were considered as the objectives to identify the optimum
parameters of the system.
Furthermore, solar-assisted systems applications related to several integrated
energy systems have been researched. Jing et al. (2018) presented a combined MOO
framework and multi-criteria decision-making for optimal designing and planning of
a solar-assisted solid oxide fuel cell distributed power system. Rashidi and Khorshidi
(2018) presented a mathematical model for a solar-based multi-generation system
consisting of desalination unit, water heater, organic Rankine cycle evaporator, PV
solar collectors, and a single effect absorption chiller unit by performing energy and
exergy analysis of the system. Also, taking total cost rate and energy efficiency as
the objectives performed MOO using multi-objective differential evolution (MODE)
algorithm. Habibollahzade et al. (2018) presented an integrated power generation
system which consists of a biomass-based solid oxide fuel cell (BSOFC), a Stir-
ling engine, and an electrolyzer. In this system, the waste heat of the BSOFC unit
was utilized to run the Stirling engine, and the excess power generated by the Stir-
ling engine was then used for hydrogen production by proton exchange membrane
electrolyzer.
Behzadi et al. (2019) proposed a solar-based integrated energy system and
presented optimal design parameters which maximize exergy efficiency and
hydrogen production rate and minimize total cost rate, through MOO of the proposed
system using the genetic algorithm. Khanmohammadi et al. (2020) proposed a solar-
assisted integrated power and refrigeration system with the thermoelectric generator
(TEG) and investigated the effect of various parameters on the system’s performance
through energy, exergy, and exergo-economic analyses. Song et al. (2020) presented
optimal parameters of a solar hybrid combined cooling, heating and power system
using the NSGA-II algorithm.
3 Advanced Engineering Optimization Techniques … 37

It can be observed that the effect of various parameters of a solar-driven energy

system had been researched in assessing the performance of the system. The perfor-
mances of these systems had been studied and optimized simultaneously on thermo-
dynamic, thermo-economic, ecological, and economic aspects. The multi-objective
optimization considering these aspects was performed by advanced optimization
algorithms such as NSGA-II, MOPSO, and MODE. For a MOO problem, these
algorithms suggest a set of non-dominated Pareto-optimal solutions.
Optimization of bio-energy systems
Biofuel can be considered as a sustainable and environmentally friendly alternative
to petroleum-based fuels. Biofuels can be produced from sources such as sugar,
starch, vegetable oil, animal fats, agricultural waste, switchgrass, cereals, microalgae.
Biodiesel engine system uses the blend of the biofuel and petrodiesel as the fuel.
The performance of the biodiesel engine has considerably affected by the blending
proportions. Using biodiesel as an alternative to diesel can reduce the wear of the
engine parts and the emissions such as carbon oxides, hydrocarbons, and particulate
matter. However, there is a loss of power, lower fuel efficiency, and increased nitrogen
oxides emission with biodiesels.
First-generation biofuels are derived from food sources such as sugar, starch,
vegetable oil, and animal fats. The second-generation biofuels are derived from
cellulosic biomass, such as agricultural waste, switchgrass, and cereals, using thermo-
chemical processes. Nowadays, producing biofuels from microalgae is gaining popu-
larity. The microalgae combine the CO2 with water using solar energy and create
biomass more rapidly (Agarwal et al., 2017).
Microalgae consist of lipids, carbohydrates, and proteins and can be used to
produce biofuels like biodiesel, biogas, and bioethanol. Microalgae consist of high
oil and lipid content and have a faster growth rate. Nannochloropsis species is a
popularly used microalgae for biofuel production. This species, in addition to gener-
ating relatively high-density cells and lipids, it contains the eicosapentaenoic acid
(EPA), which is used as an additive in human diets. Nannochloropsis species is culti-
vated in artificial seawater (ASW) containing concentrated f/2 medium and vitamins
(Banerjee et al., 2016).
Kumari et al. (2014) proposed a culture medium using NPK-10:26:26 fertil-
izer and flue gas for cultivating the Spirulina Platensis. In another work, Kumari
et al. (2015) investigated the effect of different light sources on the growth of
the Spirulina Platensis using NPK-10:26:26 fertilizer assisted cultivation medium.
Kumar et al. (2015) presented optimal nutrients and environmental parameters of
NPK-10:26:26 fertilizer-based cultivation medium for the cultivation of Dunaliella
tertiolecta species. The response surface methodology (RSM), GA, and NSGA-
II algorithms were employed to identify the optimal parameters for the maximum
production of biomass and lipid.
Dharma et al. (2016) presented the optimal parameters of biodiesel production
from Jatropha curcus and Ceiba pentandra feedstocks using RSM. Rahman et al.
(2017) proposed a two-step process and presented optimal operating conditions for
maximizing biodiesel production from Spirulina maxima microalgae. Nayak et al.
38 3 Advanced Engineering Optimization Techniques …

(2018) proposed an optimal production process and presented optimum process

parameters for maximum biomass production using Scenedesmus species, using
flue gas as the source of CO2 and domestic wastewater as the culture medium.
The optimum process parameters were identified using an artificial neural network
(ANN)-assisted genetic algorithm. Yusuff et al. (2019) investigated Leucaena leuco-
cephala seed oil as a feedstock for biodiesel generation. Santya et al. (2019) inves-
tigated the catfish oil, waste cooking oil, and chicken eggshells as feedstock for
biodiesel production. Also, process-optimization was performed on the sensitivity
of the molar ratio of oil to methanol and catalyst.
Enhancing engine performance and reducing exhaust emissions of diesel engines
is another crucial area in biodiesel research. The major biodiesel engine perfor-
mance characteristics are engine torque, brake specific fuel consumption (BSFC),
exhaust gas temperature, brake-thermal efficiency, power output, and brake power.
The exhaust gas emissions are carbon monoxide (CO), carbon dioxide (CO2 ),
unburned hydrocarbons (HC), nitrogen oxides (NOx ), smoke opacity, and partic-
ulate matter (PM). Enhancing the performance and reducing the emissions can be
achieved by modifying combustion chamber geometry, the timing of fuel injec-
tion, and compression ratio, installing exhaust gas recirculation (EGR) systems,
and formulating biodiesel blends that have appropriate physicochemical properties
(Damanik et al., 2018).
Ong et al. (2014) investigated Calophyllum inophyllum oil as the feedstock for
biodiesel production. Wong et al. (2015) presented the process parameters that have
improved the biodiesel engine performance using the cuckoo search (CS) algorithm.
Shirneshan et al. (2016) determined optimal engine speed, engine load, and biodiesel
ratio using the ABC algorithm to achieve optimum engine performance and emission
characteristics. Bharadwaz et al. (2016) presented optimum parameters of a variable
compression ratio biodiesel engine to improve the engine performance. Using the
DoE based on RSM experiments, the effect of the compression ratio, fuel blend, and
load on engine performance and emissions are modeled. Finally, optimal parameters
are identified using the derringers desirability approach.
Khoobbakht et al. (2016) investigated the effect of engine load, speed, and fuel
blend levels on the performance of the biodiesel direct-injection diesel engine. Also,
optimum input parameters were determined using RSM, which has resulted in the
decrement of CO, HC, NOx , and smoke opacity. Hasni et al. (2017) presented the
optimum process parameters to produce biodiesel from Brucea javanica seeds using
the RSM.
Yatish et al. (2018) presented the optimal biodiesel production process parame-
ters using RSM based on CCD for extracting biodiesel from bauhinia variegata oil.
Jaliliantabar et al. (2019) identified optimum operating parameters of a biodiesel
CI engine with an EGR system through MOO using the NSGA-II algorithm. Dey
et al. (2020a) investigated the performance and emission characteristics of a signal
cylinder, four-stroke direct-injection CI engine considering different diesel, ethanol,
and palm oil biodiesel blends. Dey et al. (2020b) investigated the application of artifi-
cial neural network (ANN) model for predicting the brake specific energy consump-
tion (BSEC) and emissions such as NOx , HC, and CO2 of a single-cylinder diesel
3 Advanced Engineering Optimization Techniques … 39

engine which operates with diesel-palm biodiesel-ethanol blends. Also, using a fuzzy
interface system, optimum engine operating parameters were identified from the
experimental and ANN predicted data. Hirkude and Belokar (2020) investigated
the effect of compression ratio, fuel blend, and engine load on the performance
of single-cylinder water-cooled naturally aspirated direct-injection diesel engine.
Also, predicted the BSEC, BTE, exhaust gas temperature, and smoke opacity were
predicted using the RSM.
From the above literature, it can be observed that researchers had focused on
the optimum nutrient concentration and environmental factors, which result in an
increase in the productivity of lipid, biomass, and EPA and reduce the cost of
cultivation. Furthermore, optimum process parameters have been identified using
response surface methodology and various optimization algorithms. Furthermore,
researchers are focused on finding the optimum operating parameters of biodiesel
engine systems, which would improve engine performance and reduce emissions.
Also, optimum system parameters had been identified using various methods such
as RSM, GRA, and Taguchi and various optimization algorithms such as SA, PSO,
CS, ABC, GA, and NSGA-II.
Hydropower generation and reservoir operation
Hydroelectricity is a clean and renewable energy whose quantity depends on the
volume of water flow and the amount of head created by the water reservoir. In
cascade reservoirs, the reservoir water level of a downstream plant is influenced by the
generation of the upstream plant. Therefore, hydropower scheduling is a complicated
nonlinear dynamical optimization problem that includes nonlinear flows, nonlinear
dynamical hydraulic heads, and the interactional relationships of nonlinear input
and output variables. The objective is to obtain the optimal utilization of the hydro
resources available for maximum hydroelectric generation given a set of starting
conditions and many complex constraints in the hydropower system.
Optimizing hydropower reservoirs are complex because of their nonlinear objec-
tive function and constraints. Sharma et al. (2015) developed a decision support
system (DSS) for the operation scheduling and optimization of Sewa Hydro Electric
Plant to improve the flexibility of decision support system for reservoir operation
under availability-based tariff (ABT) regime. Upon implementation on the actual
project site, the developed DSS demonstrated to be stable and managed to give
reliable decisions.
Li and Qiu (2016) proposed a long-term multi-objective optimization model for
integrated hydro/PV power system considering the smoothness of power output
process and the total amount of annual power generation of the system simulta-
neously. The PV power output was firstly calculated by hourly solar radiation and
temperature data, which was then taken as the boundary condition for reservoir opti-
mization. For hydropower, due to its great adjustable capability, a month was taken
as the time step to balance the simulation cost. This multi-objective optimization
problem was then solved by using the modified version of NSGA-II algorithm.
40 3 Advanced Engineering Optimization Techniques …

Shang et al. (2017) presented a practical genetic algorithm based solution for
solving the economic load dispatch problem (ELDP) and compared the perfor-
mance with that of dynamic programming (DP). Specifically, their performance
was comprehensively evaluated in terms of addressing the ELDP through a case
study of 26 turbines in the Three Gorges Hydropower Plant with a focus on calcu-
lation accuracy, calculation time, and algorithm stability. Kumar and Yadav (2018)
proposed the TLBO and Jaya algorithms for solving the discrete time four-reservoir
operation (DFRO), the continuous time four-reservoir operation (CFRO), and the ten-
reservoir operation (TRO). The results obtained by the TLBO and JA were compared
with different approaches from the literature. It was found that both Jaya algorithm
and TLBO algorithm provided a satisfactory solution. Niu et al. (2018) presented a
parallel multi-objective PSO to resolve the cascade hydropower reservoirs operation
balancing benefit and firm output of Lancang cascade hydropower system in south-
west China. The results indicated that PMOPSO can provide satisfying scheduling
results in different cases.
Wang et al. (2019) proposed a double-layer model for coordinating the operations
of cascaded hydropower and neighboring wind and photovoltaic (PV) facilities, in
order to reduce the output fluctuation and maximize the combined power generation
of multi-energy system. In this model, a local complementary hydropower (LCH)
model, hydropower with less regulating ability was dispatched to alleviate short-term
fluctuations, whereas in the global complementary hydropower, (GCH), hydropower
with greater regulating capacity was operated to alleviate long-term fluctuations. The
results of a case study of a hydro-wind-PV cluster project in southwestern China
showed that most short-term fluctuations can be smoothed by LCH with a minimum
impact on the capacity factor. In contrast, alleviation of long-term fluctuations can
be achieved with GCH, but this causes a large capacity factor reduction. The findings
from the case study demonstrated that clustering of hydropower cascades with wind
and solar generation facilities was a promising avenue for the de-carbonization of
electricity systems.
Onokwai et al. (2020) presented optimum performance characteristics of Jebba
hydropower plant, Nigeria, using response surface methodology (RSM) to enhance
the estimated power generated from the plant. Results obtained showed that elec-
tricity generated, turbine speed, and efficiency of the hydropower plant were influ-
enced by discharge, pressure drop, stator temperature, and water pressure. Also, the
power output was optimum when the discharge and pressure drop were kept high.
The RSM simulation of the experimental data showed that the optimal values for
discharge, pressure drop, stator temperature, and water pressure were 8783.37 m3 ,
28.59 m, 58.0 °C, and 31.00 N/m2 , respectively.
Hatamkhani et al. (2020) developed a simulation–optimization model for optimal
design of hydropower systems with a systematic view of the basin. Then, PSO algo-
rithm, linked to the simulation model and the developed optimization-simulation
model, was used to solve the problem of optimal design of Garsha, Kuran Buzan,
Sazbon and Tange mashoore power plant projects in Karkhe river basin. The objec-
tive function was maximization of net benefits of hydropower projects considering all
resources, uses, and demands in the basin. Normal water level, minimum operation
3 Advanced Engineering Optimization Techniques … 41

level, installed capacity of dams, and power plants were chosen as decision variables
of the problem. Finally, the results of optimal design of hydropower system were
discussed for both economic objective functions.
Chong et al. (2021) investigated the Jaya algorithm as an optimization method
for reservoir operation. Also, when deriving the optimal operational rule a hedging
strategy is introduced to attenuate the impact of reduced water supply. This strategy
can effectively counterbalance the lack of water supply with reservoir storage require-
ments. The Jaya algorithm solution reported a higher amount of hydropower gener-
ated compared to other algorithms. Several reservoir performance indices, such
as total hydropower generation, reliability, and resilience, were used to judge the
proposed algorithm and other algorithms efficiency.
Ahmadianfar et al. (2021) developed an adaptive differential evolution with
particle swarm optimization (A-DEPSO) algorithm to derive optimal operating rules
for multi-reservoir systems with the purpose of hydropower production. The appli-
cation of the proposed method was verified by solving a complex four-reservoir
hydropower generation system, located in the southwest of Iran. The results achieved
indicate that the A-DEPSO reduced the value of the objective function by an average
of 57% compared with other well-known optimizers in the literature. Furthermore,
the total power generated from the four-reservoir system using the proposed method
was increased by an average of 11% compared to the other methods.
From the literature, it can be found that focus was on the operation of reservoir
system as well as cascade system. These systems mainly discussed the operation of
reservoir as well as interconnected system and developed the mathematical model
for the optimum operation of reservoir for the power generation. These mathematical
models generally considered the parameters such as reservoir level, time of release
water, and flow of stream. The small hydropower plants are mostly run of river, and
thus, further study is required for optimum operation of such plants.
Geothermal energy conversion
Geothermal energy has significant potential to reduce fossil fuel consumptions and
environmental impacts. To improve energy conversion efficiency of geothermal
energy systems, numerous systems designs have been proposed, and their optimiza-
tion sought (Lee et al., 2019). At this point, it is worth reviewing current developed
geothermal energy systems because understanding configurations and principles of
basic and state-of-the-art technologies is important for developing advanced energy
systems. A comprehensive review of the geothermal energy systems is carried out
from the perspective of systems analysis, design, and optimization.
Tugcu and Arslan (2017) presented parametric analysis of an absorption refrig-
eration system designed using Simav geothermal resources. Also, a novel two-stage
artificial neural network (ANN) model was built in which the outcome of the first
network was comprised of the input of the second network and using the weights
of the resultant networks and bias values, optimum system was determined. The
optimum system determined with ANN was redesigned analytically, and the results
were determined to be statistically consistent with ANN results.
42 3 Advanced Engineering Optimization Techniques …

Boyaghchi and Safari (2017) presented a quadruple energy production system

integrated with geothermal energy involving a cascade organic Rankine cycle, lique-
fied natural gas vaporization process, proton exchange membrane electrolyzer to
produce four types of energies, namely electricity power, heating load for vaporizing
liquefied natural gas, cooling effect, and hydrogen. The proposed system parameters
were then optimized using NSGA-II to achieve the maximum improvement potential
for desired system. Optimization results showed that total avoidable exergy destruc-
tion rate and total avoidable exergy destruction cost rate get 3.27 and 4.9 times,
respectively, and total avoidable investment cost rate was improved within 17.4%
relative to the base point.
Cao et al. (2018) proposed a combined cooling and power system driven by
geothermal energy for ice-making and hydrogen production. The proposed system
combines geothermal flash cycle, Kalina cycle, ammonia-water absorption refriger-
ation cycle, and electrolyzer. Also, investigated the exergy destruction of different
components and analyzed the effect of key parameters on system performance.
Furthermore, an optimization was carried out with Jaya algorithm and genetic algo-
rithm (GA) to achieve maximum exergy efficiencies under three different geothermal
water temperatures. The results showed that the optimum exergy efficiencies calcu-
lated by the Jaya algorithm are about 23.59, 25.06 and 26.25% for geothermal water
temperature of 150, 160 and 170 °C, respectively.
Miglani et al. (2018) presented a methodology for the optimization of a building
energy system including a detailed thermal model of a borehole heat exchanger-
based GSHP and solar thermal collectors. This approach is a bi-level, multi-objective
optimization approach that minimizes the total costs and CO2 emissions.
Ren et al. (2019) presented a hybrid combined cooling heating and power (CCHP)
system integrated with natural gas, solar energy, geothermal energy, and electricity
grid. Also, a multi-objective optimization model was proposed to optimize the config-
urations of the hybrid CCHP system in different operation strategies. The multi-
objective optimization from the energy, economic, and environmental aspects was
carried out using the NSGA-II algorithm. The results demonstrated that the config-
uration of the hybrid system that operates using the following electric load strategy
achieves better performances than other modes.
Tian et al. (2019) proposed a novel superstructure of the geothermal binary power
systems considering multiple heat source temperatures, working fluid selection, and
alternative heat rejection systems. Based on the superstructure, a life cycle optimiza-
tion model was formulated as a mixed-integer nonlinear fractional program (MINFP)
to determine the optimal design, and a tailored global optimization algorithm was
employed to optimize the model.
Yu et al. (2020) presented a novel combined system integrated with a Kalina
and a transcritical CO2 (T-CO2 ) cycle and investigated for making effective use of
geothermal resources. In the proposed system, the geothermal energy is first utilized
by Kalina sub-cycle and then deeply recovered by T-CO2 sub-cycle. A steady-state
mathematical model was developed to further study the novel combined system.
The result showed that under the given conditions, the system net power output
was higher than both the single Kalina cycle and single T-CO2 cycle. Furthermore, a
3 Advanced Engineering Optimization Techniques … 43

multi-objective optimization considering both thermodynamic and economic aspects

was carried out using the NSGA-II algorithm, and the optimal design condition was
found by means of the TOPSIS method based on entropy weight.
Xie et al. (2020) presented a parametric collaborative optimization method for
ground source heat pump-radiant ceiling (GSHP-RC) system to find the optimum
parameters by maximizing system performance and reducing operating costs while
ensuring indoor thermal comfort. This method integrates response surface method
(RSM) and fast non-dominated sorting genetic algorithm (NSGA-II) to search the
nonlinearity relationship between the controllable factors and the response factors
and execute multi-objective optimization progressively.
Soltani et al. (2021) developed a new method for studying whole geothermal
heating/cooling system on the basis of optimization of a ground heat pump (GHP)
system and energy analysis of the optimized system using various circulating fluids
for geothermal heat exchanger (GHE). Also, established the optimum configura-
tion of the GHE using the genetic algorithm and investigated the effect of various
circulating fluids on the system’s energy consumption.
From the above literature, it can be observed that the geothermal energy systems
can be expected to grow rapidly soon to meet the demand of growing worldwide
primary energy demands coupled with need for fossil fuel replacement. For the
better use of geothermal resources, improving conversion efficiency is the key issue
of the advancement of geothermal energy systems. Understanding the configurations
and conversion principles of the existing technologies would be helpful for efficient
geothermal energy systems development.
Ocean thermal energy conversion
An advantage of ocean thermal energy conversion (OTEC) is that it can be used as
a baseline power source due to constant ocean water temperatures and currents, a
capability that both solar and wind power generation lack. Also, OTEC might be
a good fit for island or coastal communities that have high electricity prices, are
proximate to good OTEC resources, or are isolated from large-scale infrastructure.
An OTEC plant operates based on a Rankine cycle often with a refrigerant as the
working fluid. Unlike traditional Rankine cycles, an OTEC plant operates at a very
low temperature difference, usually about 20 °C. Therefore, the required water mass
flow rates, as well as the heat exchanger areas of the evaporator and condenser, are
very large. These requirements reduce the overall efficiency of OTEC systems.
To increase the efficiency of an OTEC plant, multiple cascaded stages can be
added instead of only utilizing a single stage. The benefit of multiple stages is that
they can use the available temperature difference more effectively, thus capturing
more thermal energy per unit of pumped water. However, the diminishing returns of
adding successive stages can eventually be outweighed by the added pumping power
required to move the water though the added heat exchangers and piping. Therefore,
an optimal amount of stages for a 20MWe plant depends on the design parameters
of the system and is 2–5 depending on the thermal performance and pressure drop
in the heat exchangers. Many engineering obstacles must be overcome in order to
construct a large-scale OTEC plant, from manufacturing components to installation
44 3 Advanced Engineering Optimization Techniques …

logistics. Upshaw et al. (2011) provided a methodology for analyzing major plant
design parameters and a means for determining the optimal design of OTEC facilities.
Ahmadi et al. (2015) presented a comprehensive thermodynamic analysis and
multi-objective optimization of hybrid solar OTEC system to produce hydrogen
using electrolysis. Also, a fast and elitist non-dominated sorting genetic algorithm
(NSGA-II) was applied to determine the best design parameters for the system. The
objectives of the multi-objectives were minimization of total cost rate of the system
and maximization of the cycle exergy efficiency. In addition, a closed form equation
for the relationship between exergy efficiency and total cost rate was presented.
Their results showed that system performance is notably affected by the mass flow
rate of warm ocean surface water, solar radiation intensity, condenser temperature,
and evaporator pinch point temperature difference.
Kim et al. (2016) developed fundamental theory and a software package for a
dual-used open cycle OTEC plant with selective generation of electric power and
desalinated water. Phenomenological equations were suggested to estimate steam
generation and condensation rates, which were used to optimize the overall perfor-
mance of the dual-used open cycle OTEC operations. Also, effects of system scales
from the conventional 207 kW to designed 1000 kW plants were scrutinized in terms
of steam and cold water flow rates and fractions, temperatures of input warm and
cold seawater, and preset vacuum pressure of the evaporator.
Wang et al. (2018) presented a multi-objective optimization model for an ORC
of an OTEC system. The multi-objective particle swarm optimization (MOPSO)
algorithm has been used to solve the proposed model considering the levelized cost of
energy (LCOE) and exergy efficiency as the objectives of optimization. Furthermore,
the LINMAP technique has been used to identify Pareto-optimal solution. From the
computational results, it was observed that, with the increase of cool seawater depth,
LCOEs of six working fluids will all first decrease and then keep stable or slightly
increase. The ranking of six working fluids in terms of LCOE and exergy efficiency
was as follows: R717 and R601 have the best performance, followed by R152a,
while R134a and R600a have relatively poor performance, and R227ea was the least
desirable.
Wu et al. (2019) presented a constructal design of an evaporator in OTEC system
by taking the minimum total dimensionless pumping power (DPP) as optimization
objective and the effective volume and heat transfer rate as the constraints. The
effects of the structure and flow parameters of the evaporator on the minimum DPP
and optimal heat transfer plate width are analyzed. The optimal constructal results
of the evaporator with eight different working fluids are compared. Their results
showed that the DPP after constructal optimization reduces by 37.1% compared
with that of the initial design point. The minimum DPP dramatically augments with
the increases of heat transfer plate effective length, corrugation angle, and mass
flow rate of the warm seawater and dramatically diminishes with the increases of
corrugation wavelength and effective volume of the evaporator.
Wu et al. (2020) presented a constructal optimization of a plate condenser with
fixed heat transfer rate and effective volume in an OTEC system based on constructal
theory. Considering the entropy generation rate in heat transfer process and total
3 Advanced Engineering Optimization Techniques … 45

pumping power due to friction loss as the objectives for the optimization was carried
out using the multi-objective genetic algorithm. The results demonstrated that the
effective volume of the plate condenser has a positive impact on the considered
objectives.
Alawadhi et al. (2020) presented optimized design of the turbine geometry for an
OTEC system by maximizing the efficiency of the turbine at low inlet temperature.
Firstly, meanline design was computed that was later optimized using the response
surface methodology. Furthermore, an improvement of 4.7% in power and 4.2% in
efficiency is achieved through the optimization process.
Li et al. (2020) investigated the optimized thermodynamic performance of closed
off-shore OTEC system based on Rankine cycle. To approach the ideal Lorenz
cycle and obtain better thermal matching, a Rankine cycle using CO2-based binary
zeotropic mixtures was considered. Pure working fluids, including NH3 and CO2 ,
were also comparatively investigated with mixtures. Overall optimization to maxi-
mize the system thermal efficiency and specific work to warm seawater flow rate
are carried out. The results indicate that CO2 -based binary zeotropic mixtures could
improve thermodynamic coupling of cycle and external seawater. The performance
of Rankine cycle at mixture composition has evaporating temperature glide of 7–8 °C
is recommended.
Wu et al. (2020) presented a constructal thermodynamic optimization model for an
OTEC system with a dual-pressure organic Rankine cycle (DPORC) by combining
constructal theory with finite-time thermodynamics. The net power output of the
OTECS is chosen as the optimization objective, and the results showed that the net
power outputs after the primary, twice, triple, and sextuple constructal thermody-
namic optimizations were improved by 2.80, 4.66, 9.95, and 14.95%, respectively,
compared with the initial net power output.
Assareh et al. (2021) presented an energy system consisting of a combination
of sub-systems, including a flat panel solar collector, OTEC system, wind turbine,
organic Rankine cycle (ORC), and a thermoelectric. The proposed system was
designed and evaluated based on the average yearly electricity consumption required
by an Iranian household. The R227ea refrigerant was applied as an organic fluid in the
ORC and water for the OTEC system. Engineering equation solver (EES) software
was utilized to model the system and obtain thermodynamic results. After sensitivity
analysis, the most significant and influential parameters were proved to be wind
speed, ORC pump inlet temperature, solar irradiance, and collector area. Eventually,
the system was optimized via NSGA-II.
From the above literature review, it can be observed that there is enormous poten-
tial for OTEC technology. Many researchers are working on developing new and
efficient solutions, to take advantage of the potential of the ocean as an unlimited
energy source. The optimization of heat exchangers, condensers, vaporizers, and
turbines, among other devices, is an active line of research that must be developed
from various points of view, including thermodynamic theory, new materials, and
engineering designs. The following chapter presents the working of Jaya algorithm,
Rao algorithms, and their modified versions.
46 3 Advanced Engineering Optimization Techniques …

References

Agarwal, A. K., Agarwal, R. A., Gupta, T., & Gurjar, B. R. (2017). Introduction to biofuels. Biofuels,
Springer. https://doi.org/10.1007/978-981-10-3791-7_1
Ahmadi, M. H., Ahmadi, M. A., & Feidt, M. (2016a). Performance optimization of a solar-driven
multi-step irreversible Brayton cycle based on a multi-objective genetic algorithm. Oil and Gas
Science and Technology—Revue d’IFP Energies Nouvelles, 71(1), 16. https://doi.org/10.2516/
ogst/2014028
Ahmadi, M. H., Ahmadi, M. A., Mellit, A., Pourfayaz, F., & Feidt, M. (2016). Thermodynamic
analysis and multi objective optimization of performance of solar dish Stirling engine by the
centrality of entransy and entropy generation. International Journal of Electrical Power and
Energy Systems, 78, 88–95. https://doi.org/10.1016/j.ijepes.2015.11.042
Ahmadi, M. H., Ahmadi, M. A., Pourfayaz, F., Bidi, M., Hosseinzade, H., & Feidt, M. (2016c). Opti-
mization of powered Stirling heat engine with finite speed thermodynamics. Energy Conversion
and Management, 108, 96–105. https://doi.org/10.1016/j.enconman.2015.11.005
Ahmadi, M. H., Ahmadi, M. A., Pourfayaz, F., Hosseinzade, H., Acikkalp, E., Tlili, I., & Feidt,
M. (2016). Designing a powered combined Otto and Stirling cycle power plant through multi-
objective optimization approach. Renewable and Sustainable Energy Reviews, 62, 585–595.
https://doi.org/10.1016/j.rser.2016.05.034
Ahmadi, M. H., Dehghani, S., Mohammadi, A. H., Feidt, M., & Barranco-Jimenez, M. A. (2013a).
Optimal design of a solar driven heat engine based on thermal and thermo-economic criteria.
Energy Conversion and Management, 75, 635–642. https://doi.org/10.1016/j.enconman.2013.
07.078
Ahmadi, M. H., Mehrpooya, M., Abbasi, S., Pourfayaz, F., & Bruno, J. C. (2017). Thermo-economic
analysis and multi-objective optimization of a transcritical CO2 power cycle driven by solar energy
and LNG cold recovery. Thermal Science and Engineering Progress, 4, 185–196. https://doi.org/
10.1016/j.tsep.2017.10.004
Ahmadi, M. H., Mohammadi, A. H., Dehghani, S., & Barranco-Jiménez, M. A. (2013b). Multi-
objective thermodynamic-based optimization of output power of solar dish-Stirling engine by
implementing an evolutionary algorithm. Energy Conversion and Management, 75, 438–445.
https://doi.org/10.1016/j.enconman.2013.06.030
Ahmadi, M. H., Sayyaadi, H., Mohammadi, A. H., & Barranco-Jimenez, M. A. (2013c). Thermo-
economic multi-objective optimization of solar dish-Stirling engine by implementing evolu-
tionary algorithm. Energy Conversion and Management, 73, 370–380.https://doi.org/10.1016/
j.enconman.2013.05.031
Ahmadi, M. H., Sayyaadi, H., Dehghani, S., & Hosseinzade, H. (2013). Designing a solar powered
Stirling heat engine based on multiple criteria: Maximized thermal efficiency and power. Energy
Conversion and Management, 75, 282–291. https://doi.org/10.1016/j.enconman.2013.06.025
Ahmadi, P., Dincer, I., & Rosen, M. A. (2015). Multi-objective optimization of an ocean thermal
energy conversion system for hydrogen production. International Journal of Hydrogen Energy,
40(24), 7601–7608. https://doi.org/10.1016/j.ijhydene.2014.10.056
Ahmadianfar, I., Kheyrandish, A., Jamei, M., & Gharabaghi, B. (2021). Optimizing operating rules
for multi-reservoir hydropower generation systems: An adaptive hybrid differential evolution
algorithm. Renewable Energy, 167, 774–790. https://doi.org/10.1016/j.renene.2020.11.152
Alawadhi, K., Alhouli, Y., Ashour, A., & Alfalah, A. (2020). Design and optimization of a radial
turbine to be used in a rankine cycle operating with an otec system. Journal of Marine Science
and Engineering, 8(11), 1–22. https://doi.org/10.3390/jmse8110855
Ali, M. Z., Awad, N. H., Suganthan, P. N., & Reynolds, R. G. (2017). An adaptive multipopulation
differential evolution with dynamic population reduction. IEEE Transactions on Cybernetics,
47(9), 2768–2779. https://doi.org/10.1109/TCYB.2016.2617301
Arora, R., Kaushik, S. C., & Kumar, R. (2016). Multi-objective thermodynamic optimization of
solar parabolic dish stirling heat engine with regenerative losses using NSGA-II and decision
making. Applied Solar Energy, 52(4), 295–304. https://doi.org/10.3103/S0003701X16040046
References 47

Arora, R., Kaushik, S. C., & Kumar, R. (2017). Multi-objective thermodynamic optimisation of solar
parabolic dish Stirling heat engine using NSGA-II and decision making. International Journal
of Renewable Energy Technology, 8(1), 64. https://doi.org/10.1504/IJRET.2017.080873
Arora, R., Kaushik, S. C., Kumar, R., & Arora, R. (2016). Multi-objective thermo-economic opti-
mization of solar parabolic dish Stirling heat engine with regenerative losses using NSGA-II
and decision making. International Journal of Electrical Power and Energy Systems, 74, 25–35.
https://doi.org/10.1016/j.ijepes.2015.07.010
Assareh, E., Assareh, M., Alirahmi, S. M., Jalilinasrabady, S., Dejdar, A., & Izadi, M. (2021).
An extensive thermo-economic evaluation and optimization of an integrated system empow-
ered by solar-wind-ocean energy converter for electricity generation—case study: Bandar Abas,
Iran. Thermal Science and Engineering Progress, 25, 100965. https://doi.org/10.1016/j.tsep.2021.
100965
Banerjee, A., Guria, C., & Maiti, S. K. (2016). Fertilizer assisted optimal cultivation of microalgae
using response surface method and genetic algorithm for biofuel feedstock. Energy, 115, 1272–
1290. https://doi.org/10.1016/j.energy.2016.09.066
Behzadi, A., Gholamian, E., Ahmadi, P., Habibollahzade, A., & Ashjaee, M. (2018). Energy, exergy
and exergoeconomic (3E) analyses and multi-objective optimization of a solar and geothermal
based integrated energy system. Applied Thermal Engineering, 143(August), 1011–1022. https://
doi.org/10.1016/j.applthermaleng.2018.08.034
Behzadi, A., Habibollahzade, A., Ahmadi, P., Gholamian, E., & Houshfar, E. (2019). Multi-objective
design optimization of a solar based system for electricity, cooling, and hydrogen production.
Energy, 169, 696–709. https://doi.org/10.1016/j.energy.2018.12.047
Bellos, E., Tzivanidis, C., & Antonopoulos, K. A. (2017). Parametric analysis and optimization of
a solar assisted gas turbine. Energy Conversion and Management, 139, 151–165. https://doi.org/
10.1016/j.enconman.2017.02.042
Bharadwaz, Y. D., Rao, B. G., Rao, V. D., & Anusha, C. (2016). Improvement of biodiesel methanol
blends performance in a variable compression ratio engine using response surface methodology.
Alexandria Engineering Journal, 55(2), 1201–1209. https://doi.org/10.1016/j.aej.2016.04.006
Biswas, P. P., Suganthan, P. N., & Amaratunga, G. A. J. (2018). Decomposition based multi-objective
evolutionary algorithm for windfarm layout optimization. Renewable Energy, 115, 326–337.
https://doi.org/10.1016/j.renene.2017.08.041
Boyaghchi, F. A., & Safari, H. (2017). Parametric study and multi-criteria optimization of total
exergetic and cost rates improvement potentials of a new geothermal based quadruple energy
system. Energy Conversion and Management, 137, 130–141. https://doi.org/10.1016/j.enconman.
2017.01.047
Cao, L., Lou, J., Wang, J., & Dai, Y. (2018). Exergy analysis and optimization of a combined
cooling and power system driven by geothermal energy for ice-making and hydrogen production.
Energy Conversion and Management, 174(May), 886–896. https://doi.org/10.1016/j.enconman.
2018.08.067
Chong, K. L., Lai, S. H., Ahmed, A. N., Wan Jaafar, W. Z., & El-Shafie, A. (2021). Optimization
of hydropower reservoir operation based on hedging policy using Jaya algorithm. Applied Soft
Computing, 106, 107325. https://doi.org/10.1016/j.asoc.2021.107325
Chowdhury, S., Zhang, J., Messac, A., & Castillo, L. (2012). Unrestricted wind farm layout optimiza-
tion (UWFLO): Investigating key factors influencing the maximum power generation. Renewable
Energy, 38(1), 16–30. https://doi.org/10.1016/j.renene.2011.06.033
Chowdhury, S., Zhang, J., Messac, A., & Castillo, L. (2013). Optimizing the arrangement and the
selection of turbines for wind farms subject to varying wind conditions. Renewable Energy, 52,
273–282. https://doi.org/10.1016/j.renene.2012.10.017
Dai, D., Yuan, F., Long, R., Liu, Z., & Liu, W. (2018). Performance analysis and multi-objective opti-
mization of a Stirling engine based on MOPSOCD. International Journal of Thermal Sciences,
124, 399–406. https://doi.org/10.1016/j.ijthermalsci.2017.10.030
Damanik, N., Ong, H. C., Tong, C. W., Mahlia, T. M. I., & Silitonga, A. S. (2018). A review on the
engine performance and exhaust emission characteristics of diesel engines fueled with biodiesel
48 3 Advanced Engineering Optimization Techniques …

blends. Environmental Science and Pollution Research, 25(16), 15307–15325. https://doi.org/10.

1007/s11356-018-2098-8
Dey, S., Reang, N. M., Deb, M., & Das, P. K. (2020a). Study on performance-emission trade-off
and multi-objective optimization of diesel-ethanol-palm biodiesel in a single cylinder CI engine:
ATaguchi-fuzzy approach. Energy Sources, Part A: Recovery, Utilization and Environmental
Effects. https://doi.org/10.1080/15567036.2020.1767234
Dey, S., Reang, N. M., Majumder, A., Deb, M., & Das, P. K. (2020b). A hybrid ANN-Fuzzy
approach for optimization of engine operating parameters of a CI engine fueled with diesel-palm
biodiesel-ethanol blend. Energy, 202, 117813. https://doi.org/10.1016/j.energy.2020.117813
Dharma, S., Masjuki, H. H., Ong, H. C., Sebayang, A. H., Silitonga, A. S., Kusumo, F., & Mahlia,
T. M. I. (2016). Optimization of biodiesel production process for mixed Jatropha curcas–Ceiba
pentandra biodiesel using response surface methodology. Energy Conversion and Management,
115, 178–190. https://doi.org/10.1016/j.enconman.2016.02.034
Dhingra, S., Bhushan, G., & Dubey, K. K. (2014). Multi-objective optimization of combustion,
performance and emission parameters in a jatropha biodiesel engine using non-dominated sorting
genetic algorithm-II. Frontiers of Mechanical Engineering, 9(1), 81–94. https://doi.org/10.1007/
s11465-014-0287-9
DuPont, B., & Cagan, J. (2016). A hybrid extended pattern search/genetic algorithm for multi-stage
wind farm optimization. Optimization and Engineering, 17(1), 77–103. https://doi.org/10.1007/
s11081-016-9308-3
DuPont, B., Cagan, J., & Moriarty, P. (2016). An advanced modeling system for optimization of
wind farm layout and wind turbine sizing using a multi-level extended pattern search algorithm.
Energy, 106, 802–814. https://doi.org/10.1016/j.energy.2015.12.033
González, J. S., Gonzalez Rodriguez, A. G., Mora, J. C., Santos, J. R., & Payan, M. B. (2010).
Optimization of wind farm turbines layout using an evolutive algorithm. Renewable Energy,
35(8), 1671–1681. https://doi.org/10.1016/j.renene.2010.01.010
Grady, S. A., Hussaini, M. Y., & Abdullah, M. M. (2005). Placement of wind turbines using genetic
algorithms. Renewable Energy, 30(2), 259–270. https://doi.org/10.1016/j.renene.2004.05.007
Habibollahzade, A., Gholamian, E., Houshfar, E., & Behzadi, A. (2018). Multi-objective optimiza-
tion of biomass-based solid oxide fuel cell integrated with Stirling engine and electrolyzer. Energy
Conversion and Management, 171, 1116–1133. https://doi.org/10.1016/j.enconman.2018.06.061
Hasni, K., Ilham, Z., Dharma, S., & Varman, M. (2017). Optimization of biodiesel production from
Brucea javanica seeds oil as novel non-edible feedstock using response surface methodology.
Energy Conversion and Management, 149, 392–400. https://doi.org/10.1016/j.enconman.2017.
07.037
Hatamkhani, A., Moridi, A., & Yazdi, J. (2020). A simulation—optimization models for multi-
reservoir hydropower systems design at watershed scale. Renewable Energy, 149, 253–263.
https://doi.org/10.1016/j.renene.2019.12.055
Hirkude, J., & Belokar, V. (2020). Investigations on performance of CI engine with waste palm
oil biodiesel-diesel blends using response surface methodology. In S. Singh & V. Ramadesigan
(Eds.), Advances in Energy Research, Vol. 2 (pp. 505–514). Springer.
Ituarte-Villarreal, C. M., & Espiritu, J. F. (2011). Optimization of wind turbine placement using a
viral based optimization algorithm. Procedia Computer Science, 6, 469–474. https://doi.org/10.
1016/j.procs.2011.08.087
Jaliliantabar, F., Ghobadian, B., Najafi, G., Mamat, R., & Carlucci, A. P. (2019). Multi-objective
NSGA-II optimization of a compression ignition engine parameters using biodiesel fuel and
exhaust gas recirculation. Energy, 187, 115970. https://doi.org/10.1016/j.energy.2019.115970
Jing, R., Zhu, X., Zhu, Z., Wang, W., Meng, C., Shah, N., & Zhao, Y. (2018). A multi-objective
optimization and multi-criteria evaluation integrated framework for distributed energy system
optimal planning. Energy Conversion and Management, 166, 445–462. https://doi.org/10.1016/
j.enconman.2018.04.054
References 49

Khanmohammadi, S., Musharavati, F., Kizilkan, O., & Duc Nguyen, D. (2020). Proposal of a
new parabolic solar collector assisted power-refrigeration system integrated with thermoelec-
tric generator using 3E analyses: Energy, exergy, and exergo-economic. Energy Conversion and
Management, 220, 113055. https://doi.org/10.1016/j.enconman.2020.113055
Khoobbakht, G., Najafi, G., Karimi, M., & Akram, A. (2016). Optimization of operating factors and
blended levels of diesel, biodiesel and ethanol fuels to minimize exhaust emissions of diesel engine
using response surface methodology. Applied Thermal Engineering, 99, 1006–1017. https://doi.
org/10.1016/j.applthermaleng.2015.12.143
Kim, A. S., Kim, H. J., Lee, H. S., & Cha, S. (2016). Dual-use open cycle ocean thermal energy
conversion (OC-OTEC) using multiple condensers for adjustable power generation and seawater
desalination. Renewable Energy, 85, 344–358. https://doi.org/10.1016/j.renene.2015.06.014
Kim, T., Choi, B. I., Han, Y. S., & Do, K. H. (2018). A comparative investigation of solar-assisted
heat pumps with solar thermal collectors for a hot water supply system. Energy Conversion and
Management, 172(April), 472–484. https://doi.org/10.1016/j.enconman.2018.07.035
Kumar, A., Pathak, A. K., & Guria, C. (2015). NPK-10:26:26 complex fertilizer assisted optimal
cultivation of Dunaliella tertiolecta using response surface methodology and genetic algorithm.
Bioresource Technology, 194, 117–129. https://doi.org/10.1016/j.biortech.2015.06.082
Kumar, R. S., & Prasad, A. K. V. (2019). Environment friendly butyl ester biodiesel production
from mahua oil: Optimization and characterization. SN Applied Sciences, 1(8), 872. https://doi.
org/10.1007/s42452-019-0913-6
Kumar, V., & Yadav, S. M. (2018). Optimization of reservoir operation with a new approach in evolu-
tionary computation using TLBO algorithm and jaya algorithm. Water Resources Management,
32(13), 4375–4391. https://doi.org/10.1007/s11269-018-2067-5
Kumari, A., Pathak, A. K., & Guria, C. (2015). Effect of light emitting diodes on the cultivation
of Spirulina platensis using NPK-10:26:26 complex fertilizer. Phycological Research, 63(4),
274–283. https://doi.org/10.1111/pre.12099
Kumari, A., Sharma, V., Pathak, A. K., & Guria, C. (2014). Cultivation of Spirulina platensis using
NPK-10:26:26 complex fertilizer and simulated flue gas in sintered disk chromatographic glass
bubble column. Journal of Environmental Chemical Engineering, 2(3), 1859–1869. https://doi.
org/10.1016/j.jece.2014.08.002
Lee, I., Tester, J. W., & You, F. (2019). Systems analysis, design, and optimization of geothermal
energy systems for power production and polygeneration: State-of-the-art and future challenges.
Renewable and Sustainable Energy Reviews, 109(April), 551–577. https://doi.org/10.1016/j.rser.
2019.04.058
Li, C., Pan, L., & Wang, Y. (2020). Thermodynamic optimization of Rankine cycle using CO2-based
binary zeotropic mixture for ocean thermal energy conversion. Applied Thermal Engineering, 178,
115617. https://doi.org/10.1016/j.applthermaleng.2020.115617
Li, F. F., & Qiu, J. (2016). Multi-objective optimization for integrated hydro-photovoltaic power
system. Applied Energy, 167, 377–384. https://doi.org/10.1016/j.apenergy.2015.09.018
Li, W., Özcan, E., & John, R. (2017). Multi-objective evolutionary algorithms and hyper-heuristics
for wind farm layout optimisation. Renewable Energy, 105, 473–482. https://doi.org/10.1016/j.
renene.2016.12.022
Li, Y., Liao, S., & Liu, G. (2015). Thermo-economic multi-objective optimization for a solar-dish
Brayton system using NSGA-II and decision making. International Journal of Electrical Power
and Energy Systems, 64, 167–175. https://doi.org/10.1016/j.ijepes.2014.07.027
Liao, X., Zhou, J., Ouyang, S., Zhang, R., & Zhang, Y. (2014). Multi-objective artificial bee colony
algorithm for long-term scheduling of hydropower system: A case study of China. Water Utility
Journal, 7, 13–23.
Miglani, S., Orehounig, K., & Carmeliet, J. (2018). Integrating a thermal model of ground source
heat pumps and solar regeneration within building energy system optimization. Applied Energy,
218(February), 78–94. https://doi.org/10.1016/j.apenergy.2018.02.173
50 3 Advanced Engineering Optimization Techniques …

Moorthy, C. B., & Deshmukh, M. K. (2013). A new approach to optimise placement of wind turbines
using particle swarm optimisation. International Journal of Sustainable Energy, 34(6), 396–405.
https://doi.org/10.1080/14786451.2013.860140
Moorthy, C. B., Deshmukh, M. K., & Mukherejee, D. (2014). New approach for placing wind
turbines in a wind farm using genetic algorithm. Wind Engineering, 38(6), 633–642. https://doi.
org/10.1260/0309-524X.38.6.633
Mosetti, G., Poloni, C., & Diviacco, B. (1994). Optimization of wind turbine positioning in
large windfarms by means of a genetic algorithm. Journal of Wind Engineering and Industrial
Aerodynamics, 51(1), 105–116. https://doi.org/10.1016/0167-6105(94)90080-9
Nayak, M., Dhanarajan, G., Dineshkumar, R., & Sen, R. (2018). Artificial intelligence driven process
optimization for cleaner production of biomass with co-valorization of wastewater and flue gas
in an algal biorefinery. Journal of Cleaner Production, 201, 1092–1100. https://doi.org/10.1016/
j.jclepro.2018.08.048
Niu, W. jing, Feng, Z. kai, Cheng, C. tian, & Wu, X. yu. (2018). A parallel multi-objective particle
swarm optimization for cascade hydropower reservoir operation in southwest China. Applied Soft
Computing Journal, 70, 562–575. https://doi.org/10.1016/j.asoc.2018.06.011
Ogunjuyigbe, A. S. O., Ayodele, T. R., & Bamgboje, O. D. (2017). Optimal placement of wind
turbines within a wind farm considering multi-directional wind speed using two-stage genetic
algorithm. Frontiers in Energy. https://doi.org/10.1007/s11708-018-0514-x
Ong, H. C., Masjuki, H. H., Mahlia, T. M. I., Silitonga, A. S., Chong, W. T., & Leong, K. Y.
(2014). Optimization of biodiesel production and engine performance from high free fatty acid
Calophyllum inophyllum oil in CI diesel engine. Energy Conversion and Management, 81, 30–40.
https://doi.org/10.1016/j.enconman.2014.01.065
Onokwai, A. O., Owamah, H. I., Ibiwoye, M. O., Ayuba, G. C., & Olayemi, O. A. (2020). Application
of response surface methodology (RSM) for the optimization of energy generation from Jebba
hydro-power plant Nigeria. ISH Journal of Hydraulic Engineering, 00(00), 1–9. https://doi.org/
10.1080/09715010.2020.1806120
Patel, J., Savsani, V., & Patel, R. (2015). Maximizing energy output of a wind farm using
teaching–learning-based optimization. Volume 2: Photovoltaics; Renewable-Non-Renewable
Hybrid Power System; Smart Grid, Micro-Grid Concepts; Energy Storage; Solar Chemistry;
Solar Heating and Cooling; Sustainable Cities and Communities, Transportation; Symposium
on Integrated/Sustainable Buil. https://doi.org/10.1115/ES2015-49164
Patel, J., Savsani, V., Patel, V., & Patel, R. (2017). Layout optimization of a wind farm to maximize
the power output using enhanced teaching learning based optimization technique. Journal of
Cleaner Production, 158, 81–94. https://doi.org/10.1016/j.jclepro.2017.04.132
Pillai, A. C., Chick, J., Johanning, L., & Khorasanchi, M. (2018). Offshore wind farm layout opti-
mization using particle swarm optimization. Journal of Ocean Engineering and Marine Energy,
4(1), 73–88. https://doi.org/10.1007/s40722-018-0108-z
Pookpunt, S., & Ongsakul, W. (2013). Optimal placement of wind turbines within wind farm
using binary particle swarm optimization with time-varying acceleration coefficients. Renewable
Energy, 55, 266–276. https://doi.org/10.1016/j.renene.2012.12.005
Rahman, M. A., Aziz, M. A., Al-khulaidi, R. A., Sakib, N., & Islam, M. (2017). Biodiesel production
from microalgae Spirulina maxima by two step process: Optimization of process variable. Journal
of Radiation Research and Applied Sciences, 10(2), 140–147. https://doi.org/10.1016/j.jrras.2017.
02.004
Rashidi, H., & Khorshidi, J. (2018). Exergoeconomic analysis and optimization of a solar based
multigeneration system using multiobjective differential evolution algorithm. Journal of Cleaner
Production, 170, 978–990. https://doi.org/10.1016/j.jclepro.2017.09.201
Reddy, S. R. (2020). Wind Farm Layout Optimization (WindFLO): An advanced framework for
fast wind farm analysis and optimization. Applied Energy, 269, 115090. https://doi.org/10.1016/
j.apenergy.2020.115090
References 51

Ren, F., Wang, J., Zhu, S., & Chen, Y. (2019). Multi-objective optimization of combined cooling,
heating and power system integrated with solar and geothermal energies. Energy Conversion and
Management, 197, 111866. https://doi.org/10.1016/j.enconman.2019.111866
Sadatsakkak, S. A., Ahmadi, M. H., & Ahmadi, M. A. (2015). Thermodynamic and thermo-
economic analysis and optimization of an irreversible regenerative closed Brayton cycle. Energy
Conversion and Management, 94, 124–129. https://doi.org/10.1016/j.enconman.2015.01.040
Sanaye, S., & Taheri, M. (2018). Modeling and multi-objective optimization of a modified hybrid
liquid desiccant heat pump (LD-HP) system for hot and humid regions. Applied Thermal
Engineering, 129, 212–229. https://doi.org/10.1016/j.applthermaleng.2017.09.116
Santya, G., Maheswaran, T., & Yee, K. F. (2019). Optimization of biodiesel production from high
free fatty acid river catfish oil (Pangasius hypothalamus) and waste cooking oil catalyzed by waste
chicken egg shells derived catalyst. SN Applied Sciences, 1(2), 152. https://doi.org/10.1007/s42
452-018-0155-z
Sessarego, M., Dixon, K. R., Rival, D. E., & Wood, D. H. (2014). A hybrid multi-objective
evolutionary algorithm for wind-turbine blade optimization. Engineering Optimization, 47(8),
1043–1062. https://doi.org/10.1080/0305215x.2014.941532
Shakoor, R., Hassan, M. Y., Raheem, A., & Rasheed, N. (2015). The modelling of wind farm layout
optimization for the reduction of wake losses. Indian Journal of Science and Technology, 8(17).
https://doi.org/10.17485/ijst/2015/v8i17/69817
Shang, Y., Lu, S., Gong, J., Liu, R., Li, X., & Fan, Q. (2017). Improved genetic algorithm for
economic load dispatch in hydropower plants and comprehensive performance comparison with
dynamic programming method. Journal of Hydrology, 554, 306–316. https://doi.org/10.1016/j.
jhydrol.2017.09.029
Sharma, R. N., Chand, N., Sharma, V., & Yadav, D. (2015). Decision support system for operation,
scheduling and optimization of hydro power plant in Jammu and Kashmir region. Renewable and
Sustainable Energy Reviews, 43, 1099–1113. https://doi.org/10.1016/j.rser.2014.11.005
Shirneshan, A., Samani, B. H., & Ghobadian, B. (2016). Optimization of biodiesel percentage
in fuel mixture and engine operating conditions for diesel engine performance and emission
characteristics by artificial bees colony algorithm. Fuel, 184, 518–526. https://doi.org/10.1016/j.
fuel.2016.06.117
Soltani, M., Farzanehkhameneh, P., Moradi Kashkooli, F., Al-Haq, A., & Nathwani, J. (2021).
Optimization and energy assessment of geothermal heat exchangers for different circulating
fluids. Energy Conversion and Management, 228(November 2020), 113733. https://doi.org/10.
1016/j.enconman.2020.113733
Song, Z., Liu, T., & Lin, Q. (2020). Multi-objective optimization of a solar hybrid CCHP system
based on different operation modes. Energy, 206, 118125. https://doi.org/10.1016/j.energy.2020.
118125
Tian, X., Meyer, T., Lee, H., & You, F. (2020). Sustainable design of geothermal energy systems for
electric power generation using life cycle optimization. AIChE Journal, 66(4). https://doi.org/10.
1002/aic.16898
Tugcu, A., & Arslan, O. (2017). Optimization of geothermal energy aided absorption refrigeration
system—GAARS: A novel ANN-based approach. Geothermics, 65, 210–221. https://doi.org/10.
1016/j.geothermics.2016.10.004
Turner, S. D. O., Romero, D. A., Zhang, P. Y., Amon, C. H., & Chan, T. C. Y. (2014). A new
mathematical programming approach to optimize wind farm layouts. Renewable Energy, 63,
674–680. https://doi.org/10.1016/j.renene.2013.10.023
Upshaw, C. R., & Webber, M. E. (2011). Integrated thermal-fluids system modeling of an ocean
thermal energy conversion power plant for analysis and optimization. In ASME 2011 5th Inter-
national Conference on Energy Sustainability, Parts A, B, and C, (pp. 1255–1264). https://doi.
org/10.1115/ES2011-54595
Wang, J., Huang, W., Ma, G., & Chen, S. (2015). An improved partheno genetic algorithm for
multi-objective economic dispatch in cascaded hydropower systems. International Journal of
Electrical Power and Energy Systems, 67, 591–597. https://doi.org/10.1016/j.ijepes.2014.12.037
52 3 Advanced Engineering Optimization Techniques …

Wang, L., Zuo, M. J., Xu, J., Zhou, Y., & Tan, A. C. (2019). Optimizing wind farm layout by
addressing energy-variance trade-off: A single-objective optimization approach. Energy, 189,
116149. https://doi.org/10.1016/j.energy.2019.116149
Wang, M., Jing, R., Zhang, H., Meng, C., Li, N., & Zhao, Y. (2018). An innovative organic rankine
cycle (ORC) based ocean thermal energy conversion (OTEC) system with performance simulation
and multi-objective optimization. Applied Thermal Engineering, 145, 743–754. https://doi.org/
10.1016/j.applthermaleng.2018.09.075
Wang, X., Virguez, E., Xiao, W., Mei, Y., Patiño-Echeverri, D., & Wang, H. (2019). Clustering and
dispatching hydro, wind, and photovoltaic power resources with multiobjective optimization of
power generation fluctuations: A case study in southwestern China. Energy, 189, 116250. https://
doi.org/10.1016/j.energy.2019.116250
Wu, X., Hu, W., Huang, Q., Chen, C., Jacobson, M. Z., & Chen, Z. (2020). Optimizing the layout
of onshore wind farms to minimize noise. Applied Energy, 267, 114896. https://doi.org/10.1016/
j.apenergy.2020.114896
Wu, Z., Feng, H., Chen, L., & Ge, Y. (2020). Performance optimization of a condenser in ocean
thermal energy conversion (OTEC) system based on constructal theory and a multi-objective
genetic algorithm. Entropy, 22(6). https://doi.org/10.3390/E22060641
Wu, Z., Feng, H., Chen, L., Tang, W., Shi, J., & Ge, Y. (2020). Constructal thermodynamic opti-
mization for ocean thermal energy conversion system with dual-pressure organic Rankine cycle.
Energy Conversion and Management, 210, 112727. https://doi.org/10.1016/j.enconman.2020.
112727
Wu, Z., Feng, H., Chen, L., Xie, Z., & Cai, C. (2019). Pumping power minimization of an evaporator
in ocean thermal energy conversion system based on constructal theory. Energy, 181, 974–984.
https://doi.org/10.1016/j.energy.2019.05.216
Xie, Y., Hu, P., Zhu, N., Lei, F., Xing, L., & Xu, L. (2020). Collaborative optimization of ground
source heat pump-radiant ceiling air conditioning system based on response surface method and
NSGA-II. Renewable Energy, 147, 249–264. https://doi.org/10.1016/j.renene.2019.08.109
Yang, K., Kwak, G., Cho, K., & Huh, J. (2019). Wind farm layout optimization for wake effect
uniformity. Energy, 183, 983–995. https://doi.org/10.1016/j.energy.2019.07.019
Yatish, K. V., Lalithamba, H. S., Suresh, R., & Hebbar, H. R. H. (2018). Optimization of bauhinia
variegata biodiesel production and its performance, combustion and emission study on diesel
engine. Renewable Energy, 122, 561–575. https://doi.org/10.1016/j.renene.2018.01.124
Yu, Z., Su, R., & Feng, C. (2020). Thermodynamic analysis and multi-objective optimization of
a novel power generation system driven by geothermal energy. Energy, 199. https://doi.org/10.
1016/j.energy.2020.117381
Yusuff, A. S., Lala, M. A., Popoola, L. T., & Adesina, O. A. (2019). Optimization of oil extraction
from Leucaena leucocephala seed as an alternative low-grade feedstock for biodiesel production.
SN Applied Sciences, 1(4), 357. https://doi.org/10.1007/s42452-019-0364-0
Chapter 4
Working of Jaya and Rao Optimization
Algorithms and Their Variants

Abstract This chapter presents the details of the working of the Jaya algorithm,
Rao algorithms and their modified versions, namely multi-team perturbation-guiding
Jaya algorithm, adaptive multi-team perturbation-guiding Jaya algorithm, elitist Rao
algorithms and self-adaptive population Rao algorithm. Also, it presents the imple-
mentation of these algorithms to multi-objective optimization problems. In addition,
multi-objective optimization performance indicators such as hypervolume, spacing,
coverage and inverted generational distance are described. Furthermore, the results
of application of the Jaya and Rao algorithms along with their modified versions
to 53 unconstrained single-objective optimization benchmark problems and five
multi-objective optimization benchmark problems are presented.

4.1 Working of the Jaya Algorithm and Its Modified

Versions

The working of the Jaya algorithm and its modified versions in solving single- and
multi-objective optimization problems is described in this section.

4.1.1 Jaya Algorithm

The Jaya algorithm (Rao, 2016) is a robust and easily applicable optimization
approach that always tries to become victorious by reaching the best solution. Jaya
algorithm has no algorithm-specific control parameters. It requires tuning of common
control parameters only and has only one stage for updating the solutions. Thus, it is
easier to employ in engineering applications. The Jaya algorithm uses both the best
and the worst solutions to update the candidate solutions.
The flowchart of the Jaya algorithm is presented in Fig. 4.1. Let the objective
be to minimize or maximize the function O(u) with m number of design variables
(i.e., v = 1, 2, …, m). Begin the Jaya algorithm with the initialization of a random
population of size n (i.e., p = 1, 2, …, n). Next, identify the best and worst candidate

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 53

V. R. Ravipudi et al., Design Optimization of Renewable Energy Systems Using Advanced
Optimization Algorithms, Green Energy and Technology,
https://doi.org/10.1007/978-3-030-95589-2_4
54 4 Working of Jaya and Rao Optimization Algorithms …

Fig. 4.1 Flowchart of the Jaya algorithm in single-objective optimization

solutions from the population based on the objective function value. Then, update
all the candidate solutions using the following equation (Rao, 2016):
     
u v, p,i = u v, p,i + r1,v,i u v,best,i − u v, p,i  − r2,v,i u v,worst,i − u v, p,i  (4.1)

During any iteration i, u v, p,i is the value of the vth variable for the pth candidate,
u v,best,i is the value of vth variable for the best candidate solution, and u v,worst,i is
the value of the vth variable for the worst candidate solution. u v, p,i is the updated
value of u v, p,i , and r1,v,i and r2,v,i are the two random numbers for the vth variable
during the ith iteration in the range [0, 1]. Accept updated candidate solution if it
is giving better objective value else keep the old candidate solution. At the end of
the iteration, the accepted candidate solutions are moved to the subsequent iteration.
This procedure is repeated until the termination criterion is satisfied. The best and
worst solutions are updated during every iteration.
4.1 Working of the Jaya Algorithm and Its Modified Versions 55

4.1.2 Multi-team Perturbation-Guiding Jaya (MTPG-Jaya)

Algorithm

In the MTPG-Jaya algorithm (Rao & Keesari, 2018), the following modifications to
the Jaya algorithm are implemented:
1. During the search process, the proposed algorithm modifies perturbation or
movement equation based on solution quality produced by the movement equa-
tion. The movement equations considered here are the modified versions of
the basic Jaya algorithm. The capability of these modified versions in finding
the optimal solutions has already been proven by previous researchers (Ocłoń
et al., 2018; Rao & Rai, 2017; Yu et al., 2017). Thus, these movement equations
have been considered to take advantage of these modified versions of the Jaya
algorithm.
2. The proposed algorithm uses multiple teams over the entire population to explore
the search space. Unlike the other multi-population algorithms, the population
is not divided into multiple teams. These teams are not multiple sets of the
population searching in individual regions of the search space. Each team uses
the same set of the population and has a different movement equation. As each
team has a different movement equation, and in the search process, each team
moves the current set of solutions to a different region in the search space. It
means that although all teams are moving from the same candidate solutions,
they may get different new candidate solutions, but may overlap in searching
regions.
3. When exploiting search space, the population may generate almost identical or
identical solutions. If the fitness value remains the same after a certain number
of iterations, then the solution is said to be stagnated. Then, non-improved
solutions with certain probability are selected to allow further exploration. Thus,
the candidates resulting in lower-quality solutions can be inserted into the search
plan, replacing higher-quality solutions. Convergence avoidance mechanisms
may help the population to escape from local optima. The uses of stagnation
treatments enable moving to regions far from the local neighborhood. While this
may slow down the attainment of the optimum solution and reduce its accuracy,
it can expand the algorithm’s exploration capability to locate promising regions
in the search space.

4.1.2.1 Perturbation Equations

The proposed algorithm uses six perturbation equations, including the basic Jaya
algorithm movement equation, to find the new solution in the search space. The first
perturbation equation is the same as that of the Jaya algorithm equation presented
in Eq. 4.1. The second perturbation equation is similar to that of the Jaya algorithm.
However, the random numbers r 1 and r 2 have been replaced by the chaotic random
number. The chaotic sequence used in this study is the well-known logistic map
56 4 Working of Jaya and Rao Optimization Algorithms …

defined by Eq. 4.2. Also, the effectiveness of this chaotic sequence generator is
presented by Yu et al. (2017). Let cm be a chaotic random number in the iteration i;
then, the second perturbation equation is given in Eq. 4.3.

cm+1 = 4cm (1 − cm ) (4.2)

     
u v, p,i = u v, p,i + cm,v,i u v,best,i − u v, p,i  − cm,v,i u v,worst,i − u v, p,i  (4.3)

The third perturbation equation is taken from Rao and Rai (2017). This movement
equation considers the concept of opposition-based learning. A quasi-opposite value
of a variable of a candidate solution is a value randomly chosen between the center
of the search space and the mirror point of the variable. Let LBv and UBv be the
lower and upper bounds of variable v, respectively. Then, the quasi-opposite value
of the candidate solution variable u v, p,i is given by Eqs. 4.4–4.6.
q
u v, p,i = rand(a, b) (4.4)

a = (LBv + UBv )/2 (4.5)

b = LBv + UBv − u v, p,i (4.6)

The fourth movement equation is taken from Ocłon et al. (2018). In the original
Jaya, the movement equation considers only the best candidate solution, whereas
in this movement equation, instead of the best candidate, any one of the top three
solutions based on the fitness value (best-1, best-2, and best-3) is considered. Let rb
be a random integer among {1, 2, 3}, and then, the movement equation is as follows:
     
u v, p,i = u v, p,i + r1,v,i u v,best(r b),i − u v, p,i  − r2,v,i u v,worst,i − u v, p,i  (4.7)

The fifth movement equation is taken from Yu et al. (2017). Let cm be a chaotic
random number generated by Eq. 4.2 and rand be a random number between [0, 1],
and then, the new candidate solution is generated by the Eq. 4.8. The sixth movement
equation is shown in the Eq. 4.9, which is inspired by the fifth movement equation.
In this, the chaotic random number is replaced by a random number between [0, 1].

u v, p,i = u v,best,i + rand(2cm − 1) (4.8)

u v, p,i = u v,best,i + rand(2rand − 1) (4.9)

4.1 Working of the Jaya Algorithm and Its Modified Versions 57

4.1.2.2 Boundary Violation and Corrections

Boundary violations of the variables in the candidate solutions are of two types. First
is a variable below the lower boundary. The second is a variable above the upper
boundary. In the proposed approach, violations are handled in three ways in both
cases. For each team, along with the corrections, violations are also calculated. These
violations are later used in finding the quality of each team. At any iteration i, let u v, p,i
be the value of the vth variable for the pth candidate. Violation of individual team is
calculated using the following equation for all u v, p,i,Team < LBv or u v, p,i,Team > UBv .


N 
D
   
ViolationTeam = u p,v,i − UBv  + LBv − u p,v,iTeam  (4.10)
Team
p=1 v=1

If u v, p,i < LBv , then it is corrected in any one of the following methods randomly.
1. Method-1: u v, p,i = LBv
2. Method-2: u v, p,i = LBv + 0.1r1 (UBv − LBv ), where r 1 is a random number
between 0 and 1.   
3. Method-3: u v, p,i = LBv + remainder u v, p,i /(UBv − LBv ) 
Similarly, if u v, p,i > UBv , then it is corrected in any one of the following methods
randomly.
4. Method-1: u v, p,i = UBv
5. Method-2: u v, p,i = UBv + 0.1r1 (UBv − LBv ), where r 1 is a random number
between 0 and 1.   
6. Method-3: u v, p,i = UBv + remainder u v, p,i /(UBv − LBv ) .

4.1.2.3 Steps in Multi-team Perturbation-Guiding Jaya Algorithm

Figure 4.2 presents the flowchart of the MTPG-Jaya algorithm. The necessary steps
of the proposed approach are as follows:
Step 1: Initialize control parameters such as population size (N), number of teams
(N_T ), number of design variables (D), lower boundary (LB), upper boundary (UB),
the probability of accepting a worse solution if stagnated (P_aws), the maximum
number of movement_iterations (i.e., MaxMI), and termination criteria. The termi-
nation criteria can be limited function evaluations (Max_Fevs) or the desired level
of accuracy of the objective function value (Termination_Value).
Step 2: Initialize the candidate solutions of size N, and calculate the initial function
values for the generated candidate solutions. Identify the global best solution based
on the function values.
Step 3: For each team, assign a movement equation randomly to guide the current
candidate solutions to new candidate solutions. All movement equations available
to each team have an equal probability of being selected. Start the counter for the
movement_iterations (MI).
58 4 Working of Jaya and Rao Optimization Algorithms …

Fig. 4.2 Flowchart of the MTPG-Jaya algorithm

Step 4: Find the new solutions for all the candidate solutions in each team based
on the movement equation of the respective team. If the new solutions violate
the boundary conditions, then apply the boundary corrections, as explained in the
previous section.
Step 5: Evaluate the function values for each solution in each team. Rank the
teams according to the fitness of function values. The team whose function value for
4.1 Working of the Jaya Algorithm and Its Modified Versions 59

a candidate solution is better than the remaining teams’ function value of respective
candidate solution is ranked number one, and the team whose function value for
a candidate solution is worse than remaining teams’ function value of respective
candidate solution is ranked number N_T.
Step 6: Update the old solutions with new solutions in the following way. For
each old candidate solution, each team generates a new candidate solution. Find the
best candidate solution out of all teams for each old candidate solution, respectively.
If this best candidate is better than the old candidate, substitute the old candidate
with the best candidate. In this manner, update all old candidate solutions.
Step 7: Update the global best solution if the best solution for updated candidate
solutions is better than the previous global best solution.
Step 8: If the global best solution is stagnated, select non-improved solutions with
P_aws probability.
Step 9: Update the perturbation or movement equation by the quality of the team.
Based on the average teams ranking and boundary condition violations, the quality
of a team is measured. Thus, for i = 1 … N_T, the team quality (T_Q) is calculated
by T_Qi = T_ARi + T_V i . Where T_AR is an average team ranking, and T_V is a
team boundary violation. Sometimes just by chance, good solutions may be gener-
ated by correction procedures. So, violations are also considered in finding team
quality. Consider the team with minimum T_Q as the best team and the team with
maximum T_Q as the worst team. Team quality is calculated during every iteration
and accumulated until movement_iterations (MI) reaches the MaxMI. Here, MI is
the iterations after selecting or changing the movement equation for any team. As
soon as the MI reaches the MaxMI, the worst team movement equation is randomly
replaced by a new movement equation.
Step 10: Verify the termination criterion. If the search process satisfies the termi-
nation criterion, then stop the program by reporting a global optimal solution, else
repeat the procedure from Step 4. The general framework of the proposed approach
is described in Algorithm 1.
60 4 Working of Jaya and Rao Optimization Algorithms …

Algorithm 1: framework of the multi-team perturbation

guiding Jaya algorithm

BEGIN

Initialize the population size 'N', Number of design variables 'D',

Teams 'N_T' probability of accepting worst solution 'P_aws', max-
imum movement iterations 'M_Itr' and termination criterion
'FE_max'. Maximum iterations Max_I = FE_max / (N×N_T);

Generate the initial population, Identify global best solution

'G_Best' and set current iteration i=0

For each team assign a movement equation randomly and set Move_Itr=1

While i < Max_I

i = i +1;

For j = 1→N_T

For k = 1→N

, → Update the variables of solutions us-

ing the equation assigned for the jth team.

, → Check the boundary condition violations

and apply corrections.

End For

End For

For j = 1→N

Rank the teams based on objective function fitness

value.

, Find the best candidate solution for the

candidate solution out of all teams.

If better than

Else

End If
4.1 Working of the Jaya Algorithm and Its Modified Versions 61

End For

Find the best solution from updated solutions.

If better than G_Best

G_Best =

Else If (stagnation reached)

Select non-improved solutions with P_aws

probability.

End If

If (Move_Itr=M_Itr)

For each team assign a movement equation randomly and

set Move_Itr=1

Else

Move_Itr=Move_Itr+1

End

End While

END

4.1.3 Adaptive Multi-team Perturbation-Guiding Jaya

(AMTPG-Jaya) Algorithm

In the MTPG-Jaya algorithm, the population size and number of teams need to be
tuned according to the optimization problem. Also, the MTPG-Jaya algorithm uses
stagnation treatment, which includes the probability of accepting the worst solutions
(Pa ) and the criterion for stagnation. This stagnation treatment method enables the
population to move away from the local optima but may lead to more computational
efforts and slow down the attainment of the optimum solution. Initial studies of
the MTPG-Jaya had revealed that the value of Pa has a significant effect on the
stagnation treatment, which in turn affects the algorithm’s performance. However,
as the complexity of the problem increases, tuning of the population size, Pa , and
the number of teams concurrently can become a tiresome work. These parameters
require appropriate tuning and may make the algorithm time-consuming. Hence, by
implementing a few modifications to the MTPG-Jaya algorithm, the AMTPG-Jaya
algorithm is proposed.
62 4 Working of Jaya and Rao Optimization Algorithms …

The AMTPG-Jaya algorithm (Rao et al., 2019; 2020; Rao & Keesari, 2019) is
proposed by eliminating the adjustment of the number of teams and the probability
of accepting the worst solutions in the MTPG-Jaya algorithm. In the AMTPG-Jaya
algorithm, an autonomously updating scheme for the number of teams is imple-
mented. Additionally, the stagnation treatments used in the MTPG-Jaya algorithm
are removed and adopted a new scheme for stagnation treatment. During the search
process, in line with the improvement of the best-so-far candidate solution, the
AMTPG-Jaya adapts the number of teams. The AMTPG-Jaya algorithm increases
the number of teams if the solution gets stagnated and removes the worst performing
team if there is an improvement in the best-so-far solution fitness value.
The perturbation schemes used in the AMTPG-Jaya algorithm are the same as
those used in the MTPG-Jaya algorithm. The stepwise procedure is also the same.
However, Step-8 of the MTPG-Jaya algorithm is removed, and an additional step
to update the number of teams is added. The AMTPG-Jaya updates the number
of teams when movement iterations (MI) become equal to the maximum number
of movement iterations (MaxMI). After updating the number of teams, it allows
the teams to explore the search space until MI reaches MaxMI. If the global best
solution recorded at MaxMI is better than that recorded at the beginning of movement
iterations, then the number of teams is reduced by one, else the number of teams is
increased by one. After updating the number of teams, it allows the teams to explore
search space until MI reaches MaxMI. The flowchart of the AMTPG-Jaya algorithm
is presented in Fig. 4.3.
Generally, the number of function evaluations is taken as the algorithm-
computational expenditure. Let TI be the total number of iterations; then, the compu-
tational expense of the AMTPG-Jaya algorithm can be calculated by N × N_T × TI.
However, the number of teams of the AMTPG-Jaya algorithm will be varying as the
search process is in progress. Thus, in the computational experiments, computational
expense in iterations can be calculated by multiplying the number of population with
the number of teams in the respective iteration, and the summation of computational
expense of all iterations will give the total computational expense.
  
 
In  the proposed  approach, the terms r1,v,i u v,best,i − u v, p,i and

cm1,v,i u v,best,i − u v, p,i  in Eqs. 4.1 and 4.3 represent the movement of
the solution
 towards
 the
 best solution,
 respectively.
 Similarly,
 the terms
−r2,v,i u v,worst,i − u v, p,i  and −cm2,v,i u v,worst,i − u v, p,i  in Eqs. 4.1 and
4.3 represent the movement of the solution to avoid the worst solution, respectively.
The amount of the movements towards the best solution and away from the worst
solution is controlled by the random numbers present in the respective terms. In
Eq. 4.1, the random numbers are selected from the uniform random distribution
between 0 and 1, whereas in Eq. 4.3, the random numbers are selected from a
chaotic sequence generated using Eq. 4.2. This chaotic sequence has randomicity
and ergodicity features, which will introduce more diversity in the population.
The quasi-oppositional movement (Eq. 4.4) further diversifies the population and
improves the convergence rate by simultaneously exploring the quasi-oppositional
region of the current population. During the search process, sometimes the best
solution in the current iteration may be the local optimum solution, and subsequent
4.1 Working of the Jaya Algorithm and Its Modified Versions 63

Fig. 4.3 Flowchart of the AMTPG-Jaya algorithm in single-objective optimization

best solutions may be located near to global optima. Thus, to further enhance
the capability to examine the search space, multiple best solutions concept is
employed in Eq. 4.7. Furthermore, to improve the exploitation capability, local
search movement is introduced in Eqs. 4.8 and 4.9. These movements will improve
the quality of solutions by producing solutions around it.
The proposed algorithm uses a single set of the population during the search
process. Multiple movement equations simultaneously guide this single set of the
population. The single set of population with a selected movement equation is consid-
ered as a team in this book. As each team has its own movement equation, each team
will move the population towards different regions of the search space. Furthermore,
based on the improvement in the best candidate, the number of teams will be updated.
If there is an improvement in the candidate solution, then the proposed AMTPG-Jaya
approach reduces the number of teams, thereby reducing the computational expense.
64 4 Working of Jaya and Rao Optimization Algorithms …

Similarly, if there is no improvement in the candidate solution, then the proposed

approach increases the teams, which in turn introduces more diversity in the popula-
tion. This way, the proposed algorithm updates the candidate solutions and achieves
the convergence. The general framework of the proposed approach is described in
Algorithm 2.

Algorithm 2: the framework of the AMTPG-Jaya algorithm

BEGIN

Initialize: N, D, LB, UB, N_T, MaxMI, and set the termination crite-
rion as MaxFE.

Generate the initial population

GlobalBest = best candidate solution,

Function evaluations FE = 0; Iterations i=0;

Allocate perturbation equation arbitrarily for each team and set MI

=1, GBold = GlobalBest

While FE < MaxFE

i=i+1

For k= 1→N_T

For j= 1→N

→ Update variables of , candidate using

equation assigned to kth team.

→ Check the boundary condition violations

and apply corrections.

End For

FE=FE+N

End For
4.1 Working of the Jaya Algorithm and Its Modified Versions 65

For j= 1→N

Rank the teams based on objective function value.

Find the best candidate out of all teams.

If better than

Else

End If
End For

Find the best candidate solution from the up

dated candidate solutions.

If better than GlobalBest

GlobalBest =

End If

GBnew = GlobalBest

If MI=MaxMI

For each team assign a movement equation randomly and

MI=1

If GBnew is better than GBold

N_T=N_T-1

Else

N_T=N_T+1

End If

GBold = GBnew

Else

MI= MI+1

End If

End While

END
66 4 Working of Jaya and Rao Optimization Algorithms …

4.1.4 Multi-objective Jaya and Multi-objective AMTPG-Jaya

Algorithms

Identifying a global best solution that gives optimum values for all the objectives
of the multi-objective optimization problem is difficult when the objectives are
conflicting in nature. The values of decision variables that produce the most optimum
value for one objective may produce non-optimal values for the other objectives.
In such situations, the decision-maker can convert the multi-objective optimization
problem into a single-objective optimization problem using the priority information
among the objectives and can find the global best solution respective to the priorities.
If the priorities of the objectives are unknown, then Pareto-optimal solutions can be
found using a posteriori approach (Deb et al., 2002) for multi-objective optimiza-
tion problems. The Pareto-optimal solutions are the solutions in which at least one
objective function value of each solution is better than the other solutions in the
Pareto-front.
The flowcharts of the multi-objective Jaya and AMTPG-Jaya algorithms are
presented in Figs. 4.4 and 4.5, respectively. In multi-objective optimization using
the Jaya algorithm and its variant AMTPG-Jaya algorithm, a posteriori articulation
of preferences is followed to handle multiple objectives simultaneously. However,
the flow of the algorithms remains the same as those in the single-objective opti-
mization scenario. In the multi-objective optimization scenario, solutions are ranked
using dominance principles and crowding distance measurements (Deb et al., 2002)
during the iterative process of the algorithms. The candidate solution with the best
rank value (i.e., 1) and maximum crowding distance (ξ ) is considered as the best
candidate. On the other hand, the solution having the worst rank value and lowest
ξ value is considered as the worst candidate. Such a selection scheme is adopted so

Fig. 4.4 Flowchart of the Jaya algorithm in multi-objective optimization

4.1 Working of the Jaya Algorithm and Its Modified Versions 67

Fig. 4.5 Flowchart of the AMTPG-Jaya algorithm in multi-objective optimization

that the solution in the less populous region of the objective space may guide the
search process.
68 4 Working of Jaya and Rao Optimization Algorithms …

4.1.4.1 Ranking Based on Non-dominance Principles

Let N be a set of candidate solutions to be ranked for an optimization problem

consisting of M number of objectives. For a minimization objective, a candidate
solution x k is considered as dominating another candidate solution x l if and only if
Gj (x k ) ≤ Gj (x l ) ∀ 1 ≤ j ≤ M and Gj (x k ) < Gj (x l ) for at least one j, where j ∈ {1, 2,
…, M}. A candidate solution x k in population N is considered as a non-dominated
candidate if there is no candidate solution x l in population N, which dominates x k . In
this manner, every candidate solution in N is compared with other candidate solutions
in N, and the non-dominated candidates are ranked one. The remaining (excluding
the ranked candidates) candidates are also ranked in the same manner and ranked
two. This procedure is continued until all the candidate solutions are ranked. All the
candidate solutions with the same rank will be considered as the front (F).

4.1.4.2 Crowding Distance (ξ) Calculation

The crowding distance (ξ k ) is an approximate concentration of the candidate solutions

in the neighborhood of a particular candidate solution k. For a selected front F, the
number of solutions in front l = |F| is determined, and ξ k = 0 ∀ k ∈ l is assigned.
Then, for each objective function j = 1, 2, …, M, the set is sorted in the worst order
of Gm . The largest crowding distance is assigned to boundary solutions in the sorted
list (ξ 1 = ξ l = ∞), and for all the other candidates in the sorted list k = 2 to (l − 1),
the crowding distance is assigned as follows:
 max 
ξk = ξk + (G k+1
j − G k−1
j )/ G j − G min
j (4.11)

where Gj is the objective function value of jth objective, G max j and G min
j are the
population-maximum and population-minimum values of the jth objective function.
In the multi-objective optimization cases, more than one optimal candidate solu-
tion exists. Thus, to find efficient Pareto-frontier using the proposed algorithm, a
candidate from the remote area of the search space is given more priority than the
candidate in the packed area of the search area. In the proposed algorithm, the candi-
date having a better rank is given higher priority, and among the two competing
candidates having an equal rank, the solution with a higher ξ value is preferred.
This will avoid converging towards a single optimum candidate solution and ensure
diversity among the candidate solutions.
Furthermore, in multi-objective optimization through the AMTPG-Jaya algo-
rithm, the scheme used for updating the number of teams is changed for multi-
objective optimization. Because after a few iterations, both new global best and old
global best solutions become non-dominated solutions. Due to this, calculating the
improvement in the global best solutions becomes difficult. Hence, in the multi-
objective version of the AMTPG-Jaya algorithm, the number of teams is updated
based on the available number of function evaluations. Initially, the search starts
4.1 Working of the Jaya Algorithm and Its Modified Versions 69

with a maximum number of teams. As the search progresses, the number of teams is
randomly reduced or increased by one for every time MI = MaxMI. After completing
60% of function evaluations, the number of teams is reduced by one whenever MI
= MaxMI and maintained a minimum number of teams (two). After updating the
number of teams, it allows the teams to explore and exploit search space until MI
reaches MaxMI. The next section presents the working of the Rao algorithms and
their modified versions.

4.2 Working of the Rao Algorithms and Their Modified

Versions

The working of the Rao algorithms (Rao, 2020) and their modified versions in solving
single- and multi-objective optimization problems is described in this section.

4.2.1 Rao Algorithms

The Rao algorithms (Rao-1, Rao-2, and Rao-3) are population-based algorithms and
are simple and easy to implement for optimization applications. These algorithms
have no algorithm-specific parameters and have no metaphorical explanation. During
the iterative process, these algorithms use iteration best solution, iteration worst solu-
tion, and random interactions among the population to explore and exploit the search
region. The flow of these three algorithms is similar, but the movement equation used
is different for each algorithm. The steps of the Rao algorithms are given below:
Step 1: Define the quantity of population (P); define the quantity design variables
(N) and their boundaries: Lower (LB), Upper (UB); termination criterion: it can be
the number of function evaluations or iterations.
Step 2: Randomly initialize the population of size P, and evaluate the objective
function Z for all the population.
Step 3: Select the best and worst solutions from the current population based on
their Z value. If the Z is a minimization function, then the solution with the smallest
Z value is the best, and the solution with the largest Z value is the worst solution and
vice versa if the Z is a maximization function.
Step 4: Locate new solutions for all the population (m = 1, 2, …, P): During the
ith iteration, let sn,m,i be the value of the nth variable of mth solution, sn,b,i be the
value of the nth variable of the best solution, sn,w,i be the value of the nth variable

of the worst solution, and sn,m,i be the newly located value of sn,m,i . Then,
As per the Rao-1 algorithm, the new solutions are found using the following
equation:

 
sn,m,i = sn,m,i + r1,n,i sn,b,i − sn,w,i (4.12)
70 4 Working of Jaya and Rao Optimization Algorithms …

As per the Rao-2 algorithm, the new solutions are located using the following
equation:
     

sn,m,i = sn,m,i + r1,n,i sn,b,i − sn,w,i + r2,n,i sn,m,i or sn,l,i  − sn,l,i or sn,m,i 
(4.13)

As per the Rao-3 algorithm, the new solutions are located using the following
equation:
      

sn,m,i = sn,m,i + r1,n,i sn,b,i − sn,w,i  + r2,n,i sn,m,i or sn,l,i  − sn,l,i or sn,m,i
(4.14)

where r1,n,i and r2,n,i are random numbers in the range [0, 1] for the nth variable
during the ith iteration. In Eqs. 4.13 and 4.14, the third term on the right-hand
side represents the interaction between the current solution (mth) and a random
solution (lth) selected from the current population. These two terms are depen-
dent on the Z values of the current (mth) and randomly selected (lth) solutions.
If the current solution Z value is superior to the randomly
  selected solution Z value,
then the third term in Eq. 4.13 becomes r sn,m,i  − sn,l,i  , and the third term
   
2,n,i
in Eq. 4.14 becomes r2,n,i sn,m,i  − sn,l,i . Similarly, if the randomly selected
solution Z value is superior  to the current  solution Z value, then the third term in
Eq. 4.13 becomes r sn,l,i  − sn,m,i  , and the third term in Eq. 4.14 becomes
   
2,n,i
r2,n,i sn,l,i  − sn,m,i .
Step 5: Evaluate Z values for the new population, and apply a greedy selection

process. If the Z value corresponding to the new solution (sn,m,i ) is superior to that
of the old solution (sn,m,i ), then replace the old solution with the new solution, if not
discard the new solution.
Step 6: Verify the stopping criterion. If the termination criterion is satisfied, report
the optimum solution from the final population, else go to Step 3. Figure 4.6 presents
the flowchart of Rao algorithms.

4.2.2 Multi-objective Rao Algorithms

In this book, a posteriori version of Rao algorithms is proposed for solving multi-
objective optimization problems (Rao & Keesari, 2020a). In this version of the Rao
algorithms, the new solutions are located in the same manner as in Rao algorithms.
However, the superiority among the solutions is identified based on non-dominance
rank and crowding distance evaluation approach (Deb et al., 2002; Rao et al., 2017).
In this version of Rao algorithms, the set of N solutions are ranked using domi-
nance principles, and proximity of the solutions with each other is calculated using
crowding distance measurement. The solution with the best rank (rank = 1) and
largest crowding distance is regarded as the best solution. On the other hand, the
4.2 Working of the Rao Algorithms and Their Modified Versions 71

Fig. 4.6 Flowchart of the Rao algorithms in single-objective optimization

solution with the worst rank and least crowding distance is regarded as the worst
solution. After identifying the best and worst solutions, a new set of N solutions
are located using movement equations of the respective Rao algorithm. Now, the set
of new solutions is combined with the set of earlier solutions forming a set of 2N
solutions. Then, the combined solutions are again ranked using dominance princi-
ples, and the crowding distance is calculated for every solution. Based on the new
ranking and crowding distance value, a set of N solutions will be selected for the
next iteration. Figure 4.7 presents the flowchart of Rao algorithms for multi-objective
optimization through a posteriori approach.

4.2.3 Elitist Rao Algorithms

Three elitist Rao algorithms based on the elitism concept for solving optimization
problems are proposed (Rao et al., 2020b). The elitist Rao (ERao-1, ERao-2, and
ERao-3) algorithms have no algorithm-specific parameters. The key features added
in the elitist Rao algorithms are elitism concept and duplicate solutions removal. In
the proposed algorithms to avoid premature convergence, the worst solutions in the
current population are replaced by elite solutions in the current population. However,
after several iterations, it may lead to the replacement of the entire population with
elite solutions, which may result in trapping into a local optimum solution. Hence,
72 4 Working of Jaya and Rao Optimization Algorithms …

Fig. 4.7 Flowchart of the Rao algorithms in multi-objective optimization

to avoid local trapping, duplicate solutions are replaced by randomly generated solu-
tions. Like the Rao algorithms, these algorithms use iteration best solution, iteration
worst solution, and random interactions among the population to exploit and explore
the solution space.
Similar to the Rao algorithms, the flow of these three algorithms is similar, but the
perturbation equation used is different for each algorithm. Let Z be a minimization
(or maximization) objective function. During any iteration i of the search process,
the ERao-1 algorithm updates the solution (s) in the current population using the
Eq. 4.12. Similarly, the ERao-2 and ERao-3 algorithms update the solutions in the
current population using the Eqs. 4.13, 4.14, respectively. The flowchart of the elitist
Rao algorithms is presented in Fig. 4.8. The essential steps of elitist Rao algorithms
for optimization are as follows:
Step 1: Define the population size (P), elite size (ES), design variables (N), and
their boundaries (Lower: LB; Upper: UB) and termination criterion.
Step 2: Randomly initialize the population of size P, and calculate the fitness
value (Z) for all the population.
Step 3: Select the best and worst solutions from the current population based on
their fitness value.
4.2 Working of the Rao Algorithms and Their Modified Versions 73

Fig. 4.8 Flowchart of the elitist Rao algorithms in single-objective optimization

Step 4: Update the solutions in the current population using the perturbation
equation. Equation 4.12 is used for the ERao-1 algorithm, Eq. 4.13 is used for the
ERao-2 algorithm, and Eq. 4.14 is used for the ERao-3 algorithm.
Step 5: Evaluate the fitness values of the updated solutions, and apply a greedy
selection process. If the fitness value corresponding to the updated solution (s  ) is
superior to that of the corresponding old solution (S), then replace the old solution
with the updated solution, if not discard the new solution.
74 4 Working of Jaya and Rao Optimization Algorithms …

Step 6: Identify the duplicate solutions, and randomly generate new solutions in
place of the duplicate solutions.
Step 7: Select the worst solutions of size ES based on the fitness values, and
replace them with the elite solutions of size ES.
Step 8: Verify the termination criterion. If the termination criterion is satisfied,
report the optimum solution from the final population, else go to Step 3.

4.2.4 Multi-objective Elitist Rao Algorithms

Similar to the Rao algorithms in multi-objective optimization, a posteriori version

of elitist Rao algorithms is proposed for solving multi-objective optimization prob-
lems. Figure 4.9 presents the flowchart of elitist Rao algorithms for multi-objective
optimization through a posteriori approach. In this version of the elitist Rao algo-
rithms, the best and worst solutions are located using the same procedure used in
the multi-objective Rao algorithms. Furthermore, after applying the non-dominated
sorting to the current solutions, the solutions at the top of the sorted solutions are
considered as the elite solutions. Correspondingly, the solutions at the bottom of
the sorted solutions are considered as the worst solutions. After identifying the elite
populations as per the elite size, the worst solutions are replaced by elite solutions.

4.2.5 Self-Adaptive Population Rao (SAP-Rao) Algorithm

The SAP-Rao algorithm (Rao & Keesari, 2021) has no algorithm-specific control
parameters, and it does not require adjustment of the population size. The key features
of the proposed algorithm are as follows:
7. In addition to the Rao algorithms perturbation equations, one more perturbation
equation is proposed in the SAP-Rao algorithm. Inspired by the Rao algo-
rithms perturbation equations, a new equation is proposed, and it is given by the
following equation:
  
sn,m,i = sn,m,i + r1,n,i sn,b,i − sn,w,i
      
+ 0.5 r2,n,i sn,w,i − sn,m,i + r3,n,i sn,b,i − sn,m,i − r4,n,i sn,b,i − sn,m,i
(4.15)

8. where, r1,n,i , r2,n,i , r3,n,i , and r4,n,i are random numbers in the range [0, 1] for
the nth variable during the ith iteration.
9. During the iterative search process, the population is randomly divided into
four sub-population groups. For each sub-population, a unique perturbation
equation from Eqs. 4.12–4.15 is allocated. As these equations have different
performance characteristics, each sub-population is moved towards a different
region of search space. Furthermore, in all iterations, different solutions will
4.2 Working of the Rao Algorithms and Their Modified Versions 75

Fig. 4.9 Flowchart of the elitist Rao algorithms in multi-objective optimization

enter into different sub-population sets, and perturbation equations will also
change. This will ensure diversity in the exploration and exploitation of the
search space.
10. During the search process, the proposed algorithm adapts the population size
based on the improvement in the fitness value. Let Z be a minimization objec-
tive function, Z best be the minimum value of the objective function in the

 be the minimum
previous iteration, and Z best value of the objective function
in the current iteration. If modulo Z best , Z best
  < 0.1, then the current popu-
lation is reduced by 10%, else the current population is increased by 10%.
However, as these equations use the best, worst, and random solutions to move
the population in search space, a minimum of population size 20 is maintained
in all iterations. This is because when the population gets divided into four
sub-population groups, each sub-population can at least have a population of
76 4 Working of Jaya and Rao Optimization Algorithms …

size 5. Then, the best, worst, and random solutions can be selected from these
five solutions. If the population size falls below 20, at least one sub-population
will have less than five solutions. In such situations, there is more probability of
selecting the best solution or the worst solution as the random solution, which
will reduce the algorithm’s effectiveness. Hence, it is suggested to maintain
a minimum population size of 20. Figure 4.10 presents the flowchart of the
SAP-Rao algorithm.

The essential steps for optimization of an objective function Z are as follows:

Step 1: Define the quantity of population (P = Pold ); define the quantity design
variables (N) and their boundaries: Lower (LB), Upper (UB); termination criterion:
it can be the number of function evaluations or iterations.
Step 2: Randomly initialize the population of size P, and evaluate the objective
function Z for all the population. If Z is a minimization function, then take the
minimum value of the objective function as Z best . If Z is a maximization function,
then take the maximum value of the objective function as Z best .
Step 3: Randomly divide the population into four groups. Assign a unique equation
for each group from Eqs. 4.12–4.15.

Fig. 4.10 Flowchart of the SAP-Rao algorithm in single-objective optimization

4.2 Working of the Rao Algorithms and Their Modified Versions 77

Step 4: For each group, select the best and worst solutions from the group
population based on their Z value.
Step 5: Locate new solutions for each group population (m = 1, 2… (P/4)) using
the equation assigned to it.
Step 6: Combine the population of all groups, evaluate Z values for the new
population, and apply a greedy selection process. If the Z value corresponding to the

new solution sn,m,i is superior to that of the old solution, then replace the old solution
with the new solution, if not discard the new solution.

Step 7: Similar to the Z best value presented in Step-II, identify
 the Z best value from

 
the new population. Then, compare the Z best with Z best . If modulo Z best , Z best  <
0.1, then reduce the current population by 10% (Pnew = Pold − round(0.1 × Pold )),
else increase the current population by 10% (Pnew = Pold + round(0.1 × Pold )).
Step 8: If Pnew < Pold , then select Pnew amount of best solutions for the next
iteration from the current population Pold . If Pnew > Pold , then the extra solutions
(Pextra = Pnew − Pold ) needed are selected from the current population itself. The
Pextra amounts of elite solutions are duplicated in the current population, and the total
population will be considered for the next iteration. Else the current population will
be considered in the next iteration.
Step 9: Verify the stopping criterion. If the stopping criterion is satisfied, report
the best solution from the final population, else go to Step 3. The general framework
of the SAP-Rao algorithm is given in Algorithm 3.

Algorithm 3: the framework of the SAP-Rao algorithm

BEGIN

Initialize P= , N, LB, UB, and FE_max- termination criterion;

Generate the initial candidate solutions and find fitness values ( ),

find , FE = 0, i=0;

While FE<FE_Max

i=i +1; FE = FE +P;& Divide the population into 4 groups

ly: , , , and

Assign a unique equation to each subpopulation.

For SP= →

For m=1→

For n= 1→N

Identify and solutions from the subpopula-

tion group SP

→ Update the variables of using the as-

signed equation to group SP
78 4 Working of Jaya and Rao Optimization Algorithms …

End For

End For

End For

Merge the updated solutions of all groups ( , , , and )

into P

- Evaluate the fitness values for the updated solutions

For m= 1→P

If better than

End If

End For
′
Find the best fitness value from the updated solutions -

′
If | ( , )| < 0.1

= − (0.1 × )

Else

= + (0.1 × )

End If

If <

Select number of best solutions from the current population

to the next iteration

Else If >

= Select number of best solutions from the current

population

= +

Else

Solutions in the current population will be considered for the

next iteration

End If

End While

END
4.2 Working of the Rao Algorithms and Their Modified Versions 79

4.2.6 Multi-objective SAP-Rao Algorithms

Figure 4.11 presents the flowchart of the SAP-Rao algorithm for solving multi-
objective optimization problems
After identifying the best solution and worst solution for each group, a new set of
P/4 solutions are located for each group. Then, these new solutions are combined and
evaluated for objective function values. Then, the non-dominance ranks are found,
and the crowding distances are calculated for each solution. Now, the population size
is updated based on the number of solutions ranked one. Let S R1 be the number of
solutions ranked one in the combined population before updating the solutions and

S R1 be the number of solutions ranked one in the combined population after updating

the solutions. If S R1 > S R1 , then increase the current population by 10% (Pnew =
Pold + round(0.1 × Pold )), else reduce the current population by 10% (Pnew = Pold −
round(0.1 × Pold )). Now, the set of new solutions is combined with the set of earlier

Fig. 4.11 Flowchart of the SAP-Rao algorithm in multi-objective optimization

80 4 Working of Jaya and Rao Optimization Algorithms …

solutions forming solutions set of size 2P. Then, these combined solutions are again
ranked using dominance principles, and crowding distance values are calculated for
all solutions. Based on the new ranks and crowding distance values, a set of Pnew
solutions will be selected for the next iteration. The next subsection presents a brief
description of the multi-objective optimization performance indicators considered in
this book.

4.3 Performance Indicators

In multi-objective optimization, the algorithms obtain a diverse set of non-dominated

solutions called as a Pareto-front. Hence, to assess the performances of the algorithms
based on the Pareto-fronts achieved, four performance indicators are adopted in this
book, and they are as follows.

4.3.1 Coverage

Zitzler et al. (2000) proposed that this performance indicator to compare and calculate
the percentage of solutions in a Pareto-front (B) is dominated by the solutions in
another Pareto-front (A). That is the coverage of the Pareto-front A over the Pareto-
front B is given by the following equation:

|{b ∈ B|∃a ∈ A : a <= b}|

Cov(A, B) = (4.16)
|B|

where a <= b means a dominates b or is equal to b.

If all the solutions of Pareto-front B are either dominated or equal to all the
solutions of the Pareto-front A then the value of Cov(A, B) is equal to 1. Similarly,
if none of the solutions in Pareto-front B are covered by the Pareto-front A then the
value of Cov(A, B) is equal to zero.

4.3.2 Spacing

This performance indicator proposed by Schott (1995) quantifies the distribution

of solutions along the Pareto-front. The spacing of a Pareto-front consisting of n
non-dominated solutions is given by the following equation:

1 n  2
S= d − di (4.17)
|n − 1| i=1
4.3 Performance Indicators 81

min  i 
k
di = f m − f mj , i, j = 1, 2, . . . , n (4.18)
i, i = j
m=1


n
d= (di /n) (4.19)
i=1

where k is the number of objectives, and f m is the objective function value of the mth
objective. The spacing indicator determines the uniformity of the distribution of the
solutions along the Pareto-front. For a Pareto-front, zero spacing value indicates that
the solutions in the Pareto-front are equidistantly spread.

4.3.3 Hypervolume

The hypervolume indicator (Rao, 2019) for a Pareto-front provides the search space
volume, which is dominated by it with respect to a specified reference point. There-
fore, in multi-objective optimization when two or more algorithm-performances are
compared, a higher value of hypervolume indicates superior performance. For a
Pareto-front consisting of N solutions, the hypervolume indicator is given by the
following equation:

N
Hypervolume = volume vi (4.20)
i=1

where vi is the hypervolume constructed with reference point Ref, and the solution i
as the diagonal corners of the hypercube.

4.3.4 Inverted Generational Distance (IGD)

The IGD indicator measures the convergence and diversity performance of the algo-
rithms in objective space. The IGD values are given by the following equation Zhou
et al. (2009):

  1   
IGD P ∗ , P = min f − f   (4.21)
|P ∗ | f ∈P ∗ f 

where P ∗ is the number of solutions in the true Pareto-front,

 P is the number of
Pareto solutions obtained by the algorithm, and  f − f   is the Euclidean distance
82 4 Working of Jaya and Rao Optimization Algorithms …

between the two solutions from the Pareto solutions sets of P ∗ and P. The algorithm
with the least IGD value can be considered as the best.

4.4 Computational Results Analysis on Single-objective

Optimization of Unconstrained Benchmark Problems

The Jaya algorithm, Rao algorithms, and their modified versions are coded in
MATLAB-R2016b, and computational tests are performed by using a CPU with
3.40 GHz Intel (R) Core i5-7500 processor and 8 GB RAM. The performance of
the Jaya and Rao algorithms and their modified versions are compared with various
optimization algorithms: genetic algorithm (GA) and its variants, particle swarm
optimization (PSO) and its variants, grasshopper optimization algorithm (GOA), arti-
ficial bee colony (ABC), dragonfly algorithm (DA), differential evolution (DE), ant
lion optimizer (ALO), and teaching–learning-based optimization (TLBO) algorithm.
The performance of the proposed modified algorithms is evaluated on 53 uncon-
strained single-objective optimization benchmark problems (including unimodal and
multimodal functions), and five multi-objective optimization benchmark problems
considered from literature.

4.4.1 Computational Results Analysis on 30 Unconstrained

Standard Benchmark Problems

The performances of the proposed modified algorithms are tested using 30 uncon-
strained standard benchmark problems. A detailed description, including the number
of independent variables and their ranges, is presented in Appendix A1 (Rao, 2016).
Firstly, the performance of the proposed modified algorithms in solving the 25 uncon-
strained problems is compared with the performances of the basic Jaya and Rao
algorithms. Then, the performance of proposed modified algorithms is compared
with those of other algorithms from the literature for all 30 unconstrained problems.
In computational experiments on these test problems, all the algorithms compared
from the literature are tested by taking 500,000 function evaluations as the termina-
tion criterion. The statistical results are presented for 30 consecutive runs. Hence, for
a fair comparison of the performances of the proposed modified algorithms with other
algorithms, the proposed algorithms are tested for 500,000 function evaluations, and
the statistical results are presented for 30 consecutive runs. In computational exper-
iments, different population sizes ranging from 10 to 100 are used to solve these
problems by keeping function evaluations fixed. Also, different elite sizes are used
for elitist Rao algorithms based on population size. The elite size is varied from 10
to 40% of the population size.
4.4 Computational Results Analysis on Single-objective … 83

In this optimization scenario, the performances of the proposed modified algo-

rithms are compared with those of the PSO, DE, GA, ABC, TLBO, Jaya, and Rao
algorithms in solving 30 unconstrained standard benchmark problems. Friedman’s
statistical test is conducted for all the algorithms’ results to judge the performances
of the proposed algorithms with those of the other algorithms. The algorithms are
ranked based on their statistical results achieved in solving the unconstrained bench-
mark problems. In ranking the algorithms for each problem, the priority is given to
the best value achieved. If the best value achieved is the same for two or more algo-
rithms, then the mean values achieved are considered for the ranking. Similarly, the
next priority is given to the worst value achieved and then standard deviation value.
If all these values are the same for the algorithms, then mean function evaluations
value is considered for ranking the algorithms. However, the mean function evalu-
ations taken by the PSO, DE, GA, ABC, TLBO, and Jaya algorithms to reach the
global optimum solution are not available. However, the mean function evaluations
taken by the Rao algorithms are available (Rao, 2020). Hence, the performances
of the proposed modified algorithms are compared with Rao algorithms separately.
In comparison with the Rao algorithms’ results, if the statistical values (best, worst,
mean, and standard deviation) are the same for two or more algorithms, then the algo-
rithms are ranked based on the mean function evaluations. In comparison with the
results of PSO, DE, GA, ABC, TLBO, and Jaya algorithms, if the statistical values
(best, worst, mean, and standard deviation) are same for two or more algorithms,
then the algorithms are given average ranks as per the Friedman’s test.
The algorithms are ranked based on their statistical results achieved in solving
the unconstrained benchmark problems. In ranking the algorithms for each problem,
the priority is given to the best value achieved. If the best value achieved is the same
for two or more algorithms, then the mean values achieved are considered for the
ranking. Similarly, the next priority is given to the worst value achieved and then
standard deviation value. If all these values are the same for the algorithms, then
mean function evaluations value is considered for ranking the algorithms.
Table 4.1 presents a comparison of the statistical results achieved by the proposed
modified algorithms with Rao algorithms. In all the tables, the term ‘B’ refers to the
best solution, the term ‘W’ refers to the worst solution, the term ‘M’ refers to the
mean value of the final objective function values in 30 runs, the term ‘SD’ refers
to the standard deviation of the final objective function values in 30 runs, and the
term ‘MFE’ refers to the mean of the function evaluations taken by the algorithm
to reach the final reported solution in each run. Also, the MFE taken by the Jaya
algorithm was not available in Rao (2016). Hence, the results of the Jaya algorithm
are not considered in Table 4.1. However, the Jaya algorithm’s computational results
are compared with the proposed modified algorithms in Table 4.3. Here, the perfor-
mance of the SAP-Rao algorithm is compared with the performances of the Rao
algorithms. The performance of the ERao-1 algorithm is compared with that of the
Rao-1 algorithm. Similarly, the ERao-2 and ERao-3 algorithms’ performances are
compared with those of the Rao-2 and Rao-3 algorithms, respectively. Similarly, the
performance of the AMTPG-Jaya algorithm is compared with that of the MTPG-Jaya
algorithm.
Table 4.1 Comparison of the results of the basic and improved versions in the optimization of the unconstrained standard benchmark problems
84

S. No Function Rao-1 Rao-2 Rao-3 ERao-1 ERao-2 ERao-3 SAP-Rao MTPG-Jaya AMTPG-Jaya
(Optimum)
F1 Sphere (0) B 0 0 0 0 0 0 0 0 0
W 0 0 0 0 0 0 0 0 0
M 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0
MFE 499,976 499,791 277,522 381,309 499,935 199,177 306,759 416,661 390,316
F2 Sum Squares (0) B 0 0 0 0 0 0 0 0 0
W 0 0 0 0 0 0 0 0 0
M 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0
MFE 499,975 499,851 276,556 378,232 499,968 234,084 304,779 416,401 380,789
F3 Beale (0) B 0 0 0 0 0 0 0 0 0
W 0 0 0 0 0 0 0 0 0
M 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0
MFE 9805 7612 7325 9591 5124 5152 3653 7632 5588
F4 Easom (−1) B −1 −1 −1 −1 −1 −1 −1 −1 −1
W 0 −1 −1 −1 −1 −1 −1 −1 −1
M −0.5667 −1 −1 −1 −1 −1 −1 −1 −1
SD 0.504 0 0 0 0 0 0 0 0
MFE 3010 11,187 14,025 1261 1209 1152 1998 4989 2870
F5 Matyas (0) B 0 0 0 0 0 0 0 0 0
(continued)
4 Working of Jaya and Rao Optimization Algorithms …
Table 4.1 (continued)
S. No Function Rao-1 Rao-2 Rao-3 ERao-1 ERao-2 ERao-3 SAP-Rao MTPG-Jaya AMTPG-Jaya
(Optimum)
W 0 0 0 0 0 0 0 0 0
M 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0
MFE 77,023 110,544 143,088 22,511 24,948 20,349 35,274 13,147 8556
F6 Colville (0) B 0 0 0 0 0 0 0 0 0
W 0 5.4E−23 1E−25 0 2E−30 1E−29 0 0 0
M 0 1.8E−24 7E−27 0 2E−31 8E−31 0 0 0
SD 0 9.8E−24 2E−26 0 4E−31 2.E−30 0 0 0
MFE 385,066 477,753 488,127 369,779 103,949 106,320 145,962 239,087 93,134
F7 Trid 6 (−50) B −50 −50 −50 −50 −50 −50 −50 −50 −50
W −50 −50 −50 −50 −50 −50 −50 −50 −50
M −50 −50 −50 −50 −50 −50 −50 −50 −50
4.4 Computational Results Analysis on Single-objective …

SD 0 0 0 0 0 0 0 0 0
MFE 17,485 37,209 34,796 12,866 17,628 16,690 4916 26,531 16,670
F8 Trid 10 (−210) B −210 −210 −210 −210 −210 −210 −210 −210 −210
W −210 1171 −210 −210 −210 −210 −210 −210 −210
M −210 −30.858 −210 −210 −210 −210 −210 −210 −210
SD 0 413 0 0 0 0 0 0 0
MFE 48,231 144,156 142,253 8336 37,091 30,062 31,684 55,708 39,770
F9 Zakharov (0) B 0 0 0 0 0 0 0 0 0
W 0 0 0 0 0 0 0 0 0
(continued)
85
Table 4.1 (continued)
86

S. No Function Rao-1 Rao-2 Rao-3 ERao-1 ERao-2 ERao-3 SAP-Rao MTPG-Jaya AMTPG-Jaya
(Optimum)
M 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0
MFE 345,615 499,767 258,451 271,356 472,115 316,339 217,552 380,925 378,262
F10 Schwefel 1.2 (0) B 0 0 0 0 0 0 0 0 0
W 0 0 0 0 0 0 0 0 0
M 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0
MFE 301,513 499,849 144,367 198,157 343,190 102,493 159,572 99,119 75,501
F11 Rosenbrock (0) B 9.0E−26 1.9E−16 1.40E−14 3.0E−29 0 0 6.7E−29 0 1.3E−28
W 3.9866 2.2E+01 2.22E+01 1.1E−27 2.0E−29 1.5E−26 9.1E−26 0 1.4E−25
M 0.6644 7.4E−01 7.40E−01 2.5E−28 1.5E−30 7.7E−28 4.4E−27 0 7.6E−27
SD 1.51 4.05 4.05E+00 2.4E−28 4.3E−30 2.7E−27 1.7E−26 0 2.5E−26
MFE 489,811 478,410 478,420 57,680 124,044 495,743 97,659 94,401 84,985
F12 Dixon-Price (0) B 0.667 2.8E−30 0.667 1.5E−29 1.8E−30 9.8E−29 1.4E−29 0.667 0.667
W 0.667 0.667 0.667 0.667 0.667 0.667 0.667 0.667 0.667
M 0.667 0.289 0.667 0.644 0.311 0.644 0.644 0.667 0.667
SD 0 0.336 7.4E−05 1.2E−01 3.4E−01 1.2E−01 1.2E−01 0 0
MFE 75,427 113,638 159,231 40,471 129,255 35,742 73,259 82,335 95,163
F13 Shekel’s Foxholes B 0.998 0.998 0.998 0.998 0.998 0.998 0.998 0.998 0.998
(0.998004) W 0.998 0.998 0.998 0.998 0.998 0.998 0.998 0.998 0.998
M 0.998 0.998 0.998 0.998 0.998 0.998 0.998 0.998 0.998
(continued)
4 Working of Jaya and Rao Optimization Algorithms …
Table 4.1 (continued)
S. No Function Rao-1 Rao-2 Rao-3 ERao-1 ERao-2 ERao-3 SAP-Rao MTPG-Jaya AMTPG-Jaya
(Optimum)
SD 0 0 0 0 0 0 0 0 0
MFE 18,839 95,983 243,748 21,530 24,357 154,277 12,933 25,664 20,071
F14 Branin (0.397887) B 0.397887 0.397887 0.39788 0.39788 0.39788 0.39788 0.39788 0.39788 0.39788
W 0.397931 0.397933 0.39789 0.39788 0.39788 0.39788 0.39788 0.39788 0.39788
M 0.397892 0.397891 0.39788 0.39788 0.39788 0.39788 0.39788 0.39788 0.39788
SD 1.05E−05 1.03E−05 1.44E−07 0 0 0 0 0 0
MFE 102,785 41,263 80,683 67,660 13,930 16,213 1354 28,740 29,489
F15 Bohachevsky 1 (0) B 0 0 0 0 0 0 0 0 0
W 0 0 0 0 0 0 0 0 0
M 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0
MFE 3129 4751 3435 1036 2719 878 1669 1831 1222
4.4 Computational Results Analysis on Single-objective …

F16 Booth (0) B 0 0 0 0 0 0 0 0 0

W 0 0 0 0 0 0 0 0 0
M 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0
MFE 5583 4485 4312 1860 1847 1824 2893 4481 3688
F17 Michalewicz-2 B −1.8013 −1.8013 −1.8013 −1.8013 −1.8013 −1.8013 −1.8013 −1.8013 −1.8013
(−1.8013) W −1.8013 −1.8013 −1.8013 −1.8013 −1.8013 −1.8013 −1.8013 −1.8013 −1.8013
M −1.8013 −1.8013 −1.8013 −1.8013 −1.8013 −1.8013 −1.8013 −1.8013 −1.8013
SD 0 0 0 0 0 0 0 0 0
(continued)
87
Table 4.1 (continued)
88

S. No Function Rao-1 Rao-2 Rao-3 ERao-1 ERao-2 ERao-3 SAP-Rao MTPG-Jaya AMTPG-Jaya
(Optimum)
MFE 3863 2694 2751 419 871 437 1965 2921 1650
F18 Michalewicz-5 B −4.6877 −4.6877 −4.6877 −4.6877 −4.6877 −4.6877 −4.6877 −4.6877 −4.6877
(−4.6877) W −4.5377 −3.1168 −3.4959 −4.6536 −4.4959 −4.4959 −3.5992 −4.6459 −4.6459
M −4.6743 −4.4299 −4.4922 −4.6865 −4.6370 −4.6466 −4.5376 −4.6835 −4.6799
SD 0.0309 0.3600 0.2790 0.0062 0.0514 0.0412 0.2045 0.0127 0.0161
MFE 39,710 67,252 58,401 247,154 164,310 138,840 34,031 84,480 203,453
F19 Bohachevsky-2 (0) B 0 0 0 0 0 0 0 0 0
W 0 0 0 0 0 0 0 0 0
M 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0
MFE 2963 4272 3191 1081 1589 931 1733 1141 1093
F20 Bohachevsky-3 (0) B 0 0 0 0 0 0 0 0 0
W 0 0 0 0 0 0 0 0 0
M 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0
MFE 4725 12,337 6821 1331 1673 1289 2207 3164 1459
F21 GoldStein-Price B 3 3 3 3 3 3 3 3 3
(3) W 3 84 3 3 3 3 3 3 3
M 3 5.7 3 3 3 3 3 3 3
SD 0 14.8 0 0 0 0 0 0 0
MFE 180,121 176,933 353,893 2185 1881 301,287 22,158 16,768 6391
(continued)
4 Working of Jaya and Rao Optimization Algorithms …
Table 4.1 (continued)
S. No Function Rao-1 Rao-2 Rao-3 ERao-1 ERao-2 ERao-3 SAP-Rao MTPG-Jaya AMTPG-Jaya
(Optimum)
F22 Perm (0) B 0 0 0 0 0 0 0 0 0
W 3.7E−09 0 0 8.9E−31 0 0 0 0 0
M 1.5E−10 0 0 1.1E−31 0 0 0 0 0
SD 6.8E−10 0 0 2.9E−31 0 0 0 0 0
MFE 82,792 3139 4453 3192 18,004 8325 3749 22,072 17,524
F23 Hartmann-3 (0) B −3.863 −3.863 −3.863 −3.863 −3.863 −3.863 −3.863 −3.863 −3.863
W −3.863 −3.863 −3.863 −3.863 −3.863 −3.863 −3.863 −3.863 −3.863
M −3.863 −3.863 −3.863 −3.863 −3.863 −3.863 −3.863 −3.863 −3.863
SD 0 0 0 0 0 0 0 0 0
MFE 4459 3022 3271 4256 1938 1900 85,136 2859 1557
F24 Ackley (0) B 1.5E−14 8.0E−15 4.4E−15 8.0E−15 8.9E−16 4.4E−15 4.4E−15 0 8.9E−16
W 2.2E+00 1.5E−14 1.5E−14 6.2E00 2.0E+01 4.4E−15 4.4E−15 0 8.9E−16
4.4 Computational Results Analysis on Single-objective …

M 5.7E−01 1.0E−14 6.7E−15 1.3E+00 4.0E+00 4.4E−15 4.4E−15 0 8.9E−16

SD 7.4E−01 3.1E−15 2.4E−15 1.3E+00 8.1E+00 0 0 0 0
MFE 129,392 417,741 76,352 232,012 267,183 73,578 81,369 2463 3612
F25 Penalized-2 (0) B 1.4E−32 1.4E−32 1.4E−32 7.3E−17 7.7E−17 8.0E−17 6.3E−17 0 7.6E−17
W 1.1E−02 1.6E+00 1.4E−01 1.0E−16 1.0E−16 1.0E−01 1.0E−16 0 1.0E−16
M 1.5E−03 5.8E−02 1.6E−02 9.4E−17 9.2E−17 9.4E−03 8.7E−17 0 9.2E−17
SD 3.8E−03 2.9E−01 3.5E−02 6.2E−18 6.2E−18 2.9E−02 9.6E−18 0 6.4E−18
MFE 173,661 115,593 55,637 219,860 120,748 61,684 127,018 39,767 67,741
Source Rao-1, Rao-2, Rao-3: Rao (2020)
Result in boldface indicates better values
89
90 4 Working of Jaya and Rao Optimization Algorithms …

From Table 4.1, it can be observed that the SAP-Rao algorithm is converging
faster than the Rao algorithms. In 24 out of 25 test problems, the SAP-Rao algorithm
has achieved global optimum solutions in fewer MFEs when compared to those of the
Rao-1 algorithm. Similarly, the SAP-Rao algorithm consumed fewer MFEs for 23 out
of 25 test problems compared to the Rao-2 algorithm, and in 19 out of 25 problems
compared to the Rao-3 algorithm. Furthermore, the SAP-Rao algorithm has attained
better or the same results in terms of the best, worst, mean, and standard deviation
values in 23 problems compared to those of the Rao algorithms. For the test problems
11, 24, and 25, the SAP-Rao algorithm has achieved better performance compared
to that of Rao algorithms. Furthermore, it can be observed that the AMTPG-Jaya
algorithm is converging faster than the MTPG-Jaya algorithm. In 19 out of 25 test
problems, the AMTPG-Jaya algorithm achieved global optimum solutions in fewer
MFEs compared to those of the MTPG-Jaya algorithm in all the test problems.
The performance of the ERao-1 algorithm is superior or the same when compared
to the performance of the Rao-1 algorithm in solving 24 out of 25 problems in terms of
better, worst, mean, and standard deviation values. However, in all the test problems
except for the benchmark problems 13, 18, 24, and 25, the ERao-1 algorithm has
consumed relatively lesser function evaluations than the Rao-1 algorithm to achieve
better or the same solutions. For the benchmark problems 4, 11, 12, 14, 18, 22, and
25, the ERao-1 algorithm has achieved superior performance compared to that of the
Rao-1 in terms of best, worst, mean, and standard deviation values.
Similarly, the performance of the ERao-2 algorithm is better or the same in solving
24 out of 25 problems when compared to the performance of the Rao-2 algorithm
in terms of better, worst, mean, and standard deviation values. However, the ERao-2
algorithm has consumed relatively lesser function evaluations than the Rao-2 algo-
rithm to achieve better or the same solutions in all the test problems except for the
benchmark problems 1, 2, 12, 18, 22, and 25. Also, for the benchmark problems 6,
8, 11, 12, 14, 18, 21, and 25, the ERao-2 algorithm has achieved superior perfor-
mance in terms of best, worst, mean, and standard deviation values than that of the
Rao-2 algorithm. The performance of the ERao-3 algorithm is superior or the same
in solving all benchmark problems when compared to the performance of the Rao-3
algorithm in terms of better, worst, mean, and SD values. However, the elitist Rao-3
algorithm has consumed relatively lesser function evaluations than the Rao-3 algo-
rithm to achieve better or the same solutions in all the test problems except for the
benchmark problems 9, 11, 18, 22, and 25. For the benchmark problems 6, 11, 12,
14, 18, 24, and 25, the ERao-3 algorithm has achieved superior performance in terms
of best, worst, mean, and standard deviation values.
From the above analysis, it can be observed that the proposed modified algorithms
are faster in converging and more effective in achieving global optimum solutions.
However, the performances of these algorithms in solving the unconstrained bench-
mark problems are evaluated using the Friedman statistical test. The algorithms are
ranked based on their statistical results achieved in solving the unconstrained bench-
mark problems. In ranking the algorithms for each problem, the priority is given to
the best value achieved. If the best value achieved is the same for two or more algo-
rithms, then the mean values achieved are considered for the ranking. Similarly, the
4.4 Computational Results Analysis on Single-objective … 91

next priority is given to the worst value achieved and then standard deviation value.
If all these values are the same for the algorithms, then mean function evaluations
value is considered for ranking the algorithms. After assigning the ranks to algo-
rithms based on their statistical results, each algorithm’s average rank is calculated.
As nine algorithms are compared in this test, the χ 2 distribution with eight degrees
of freedom is considered. The average ranks, χ 2 value, and p-value achieved in the
Friedman test are presented in Table 4.2. The ranks of these algorithms and their
average rank values are plotted in Fig. 4.12.
Here, an observation can be made that the p-value of the Friedman test is much less
than 0.01, which indicates that the results are highly significant. The least value of the
average rank indicates superior performance. The ERao-3 algorithm has achieved the
least average rank of 3.40. The ranking of these algorithms in solving unconstrained
benchmark problems is ERao-3, AMTPG-Jaya, SAP-Rao, ERao-1, ERao-2, MTPG-
Jaya, Rao-3, Rao-1, and Rao-2. Here, an observation can be made that the proposed
modified algorithms are converging faster than the respective basic algorithms.
In addition, the performance of the proposed modified algorithms is compared
with those of the PSO, DE, GA, ABC, and TLBO algorithms in solving 30 uncon-
strained standard benchmark problems. The computational results are presented in
Table 4.3 in terms of mean and standard deviation values achieved in 30 runs. The
results of the TLBO algorithm are taken from Rao and Patel (2013), and the results of
the ABC, DE, PSO, and GA algorithms are taken from Karaboga and Akay (2009).
Here, an observation can be made that the performances of the modified versions
are better or competitive compared to the basic Jaya and Rao algorithms as well
as the GA, PSO, DE, ABC, and TLBO algorithms. In 24 out of 30 problems, the
SAP-Rao algorithm has achieved better or the same performance. The AMTPG-Jaya
algorithm has achieved better or the same performance in 22 out of 30 problems.
Similarly, the MTPG-Jaya algorithm has achieved better or the same performance in
25 out of 30 problems. In 19 out of 30 problems, the elitist Rao algorithms (ERao-1,
ERao-2, and ERao-3) have achieved better or the same performance.
However, to judge the performance of the proposed algorithms in comparison with
the other algorithms, the Friedman statistical test is conducted. The mean values are
given priority and then to the standard deviation values for ranking the algorithms in
all the problems. Here, 12 algorithms are compared in this test with 11° of freedom
for the χ 2 distribution. The average ranks, χ 2 value, and p-value achieved in the
Friedman test are presented in Table 4.4. The ranks of these algorithms and their
average rank values are plotted in Fig. 4.13.
The p-value of the Friedman test is 8.46E−06, which indicates that the results are
highly significant. The MTPG-Jaya algorithm has achieved the least average rank
of 5.20. The ranking of these algorithms in solving the unconstrained benchmark
problems is MTPG-Jaya, SAP-Rao, AMTPG-Jaya, Jaya, ERao-1, ERao-2, ERao-
3, ABC, TLBO, DE, PSO, and GA. Here, an observation can be made that the
proposed algorithms have better performance than those of the ABC, PSO, DE, GA,
and TLBO algorithms. The next subsection presents the analysis of the proposed
modified algorithms’ computational results in solving some additional unimodal,
multimodal, and fixed dimension multimodal optimization problems.
 92

Table 4.2 Friedman statistical test results for the results presented in Table 4.1
Algorithm Rao-1 Rao-2 Rao-3 ERao-1 ERao-2 ERao-3 SAP-Rao MTPG-Jaya AMTPG-Jaya
Avg. Rank 6.88 7.16 6.36 3.96 4.56 3.40 3.88 4.84 3.84
χ2 51.035
p-value The p-value is 3.09519e−8. The result is significant at p < 0.01
4 Working of Jaya and Rao Optimization Algorithms …
4.4 Computational Results Analysis on Single-objective … 93

Fig. 4.12 Average ranks of Rao algorithms and modified versions in the Friedman statistical test
for the unconstrained standard benchmark problems

4.4.2 Computational Results Analysis on Unconstrained

Unimodal and Multimodal Standard Benchmark
Problems

In addition to the unconstrained benchmark test problems, modified algorithms’

performance is tested using 23 unimodal and multimodal optimization test problems.
The algorithms’ performance in the unimodal test problems reveals the exploitation
capability of the algorithm, and the performance in the multimodal problems reveals
the algorithm’s exploration capability. Firstly, the performances of the proposed
modified algorithms in solving 23 unimodal and multimodal test problems are
compared with the performances of the Jaya and Rao algorithms. A detailed descrip-
tion, including the number of independent variables and their ranges of these test
problems, is presented in Appendix A2 (Rao, 2020). Then, the performances of the
proposed modified algorithms are compared with GSA, RGA, PSO, and its variants,
Jaya algorithm, and its variant using ten unimodal and multimodal optimization test
functions. A detailed description, including the number of independent variables and
their ranges of these test functions, is presented in Appendix A3 (Rao & Saroj, 2017).
In computational experiments, different population sizes ranging from 10 to 100 are
used for solving these problems by keeping function evaluations as constant. Also,
different elite sizes are used for elitist Rao algorithms based on population size. The
elite size is varied from 10 to 40% based on the population size.
Table 4.3 Comparison of the results of improved versions with the results of other algorithms achieved in the optimization of the unconstrained standard
94

benchmark problems
Function GA PSO DE ABC TLBO Jaya ERao-1 ERao-2 ERao-3 MTPG-Jaya AMTPG-Jaya SAP-Rao
F1 M 1110 0 0 0 0 0 0 0 0 0 0 0
SD 74.214474 0 0 0 0 0 0 0 0 0 0 0
F2 M 148 0 0 0 0 0 0 0 0 0 0 0
SD 12.409289 0 0 0 0 0 0 0 0 0 0 0
F3 M 0 0 0 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0 0 0 0
F4 M −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1
SD 0 0 0 0 0 0 0 0 0 0 0 0
F5 M 0 0 0 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0 0 0 0
F6 M 0.014938 0 0.0409122 0.0929674 0 0 0 2.72732E−31 8.12321E−31 0 0 0
SD 0.007364 0 0.081979 0.066277 0 0 0 4.82641E−31 2.20373E−30 0 0 0
F7 M −49.9999 −50 −50 −50 −50 −50 −50 −50 −50 −50 −50 −50
SD 2.25E–5 0 0 0 0 0 0 0 0 0 0 0
F8 M 0.193417 0 0 0 0 −210 −210 −210 −210 −210 −210 −210
SD 0.035313 0 0 0 0 0 0 0 0 0 0 0
F9 M 0.013355 0 0 0.0002476 0 0 0 0 0 0 0 0
SD 0.004532 0 0 0.000183 0 0 0 0 0 0 0 0
F10 M 7400 0 0 0 0 0 0 0 0 0 0 0
SD 1140 0 0 0 0 0 0 0 0 0 0 0
F11 M 196,000 15.088617 18.203938 0.0887707 0.0000162 0 2.5149E−28 1.47911E−30 7.6729E−28 0 7.61941E−27 4.41602E−27
(continued)
4 Working of Jaya and Rao Optimization Algorithms …
Table 4.3 (continued)
Function GA PSO DE ABC TLBO Jaya ERao-1 ERao-2 ERao-3 MTPG-Jaya AMTPG-Jaya SAP-Rao
SD 38,500 24.170196 5.036187 0.07739 0.0000364 0 2.3951E−28 4.27474E−30 2.73293E−27 0 2.50556E−26 1.66875E−26
F12 M 1220 0.6666667 0.6666667 0 0.6666667 0 0.64444444 0.311111111 0.644444444 0.666667 0.666666667 0.644444444
SD 266 1E–8 1E–9 0 0 0 0.12171612 0.338277509 0.121716124 0 0 0.121716124
F13 M 0.998004 0.998004 0.998004 0.998004 0.998004 0.998004 0.998004 0.998004 0.998004 0.9980038 0.998004 0.998004
SD 0 0 0 0 0 0 0 0 0 0 0 0
F14 M 0.397887 0.3978874 0.3978874 0.3978874 0.3978874 0.397887 0.397887 0.397887 0.397887 0.397887 0.397887 0.397887
SD 0 0 0 0 0 0 0 0 0 0 0 0
F15 M 0 0 0 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0 0 0 0
F16 M 0 0 0 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0 0 0 0
F17 M −1.8013 −1.572869 −1.801303 −1.801303 −1.801303 −1.801303 −1.801303 −1.801303 −1.801303 −1.801303 −1.801303 −1.801303
SD 0 0.11986 0 0 0 0 0 0 0 0 0 0
F18 M −4.64483 −2.490872 −4.683482 −4.687658 −4.672657 −4.68013 −4.6865216 −4.63703182 −4.64660672 −4.683482 −4.67987586 −4.53757116
SD 0.09785 0.256952 0.012529 0 0.0474 0.0158 0.00622509 0.051374466 0.041245533 0.012743 0.016094139 0.204546577
4.4 Computational Results Analysis on Single-objective …

F19 M 0.06829 0 0 0 0 0 0 0 0 0 0 0
SD 0.078216 0 0 0 0 0 0 0 0 0 0 0
F20 M 0 0 0 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0 0 0 0
F21 M 5.870093 3 3 3 3 3 3 3 3 3 3 3
SD 1.071727 0 0 0 0 0 0 0 0 0 0 0
(continued)
95
Table 4.3 (continued)
96

Function GA PSO DE ABC TLBO Jaya ERao-1 ERao-2 ERao-3 MTPG-Jaya AMTPG-Jaya SAP-Rao
F22 M 0.302671 0.0360516 0.0240069 0.0411052 0.0006766 0 1.1176E−31 0 0 0 0 0
SD 0.193254 0.048927 0.046032 0.023056 0.0007452 0 2.9037E−31 0 0 0 0 0
F23 M −3.86278 −3.633352 −3.86278 −3.86278 −3.86278 −3.86278 −3.86278 −3.86278 −3.86278 −3.86278 −3.86278 −3.86278
SD 0 0.116937 0 0 0 0 0 0 0 0 0 0
F24 M 14.67178 0.1646224 0 0 0 0 1.28742222 3.992006819 4.44089E−15 0 8.88178E−16 4.44089E−15
SD 0.178141 0.493867 0 0 0 0 1.31175382 8.120502433 0 0 0 0
F25 M 125.0613 0.0076754 0.0021975 0 2.34E−08 0 9.4079E−17 9.23047E−17 0.009419631 0 9.18662E−17 8.73247E−17
SD 12.0012 0.016288 0.004395 0 0 0 6.206E−18 6.15941E−18 0.02878998 0 6.37641E−18 9.56381E−18
F26 M −1.08094 −0.67927 −1.08094 −1.08094 −1.08094 −1.08094 −1.08094 −1.08094 −1.08094 −1.08094 −1.08094 −1.08094
SD 0 0.274621 0 0 0 0 0 0 0 0 0 0
F27 M 0.287548 0.213626 0 0.000208 0.000016 −1.2405 −1.4357000 −1.35794285 −1.39999731 −1.343692 −1.49359886 −1.49571339
SD 0.052499 0.039003 0 0.000038 0.000003 0.312691 0.17171383 0.249983625 0.212284697 0.266023 0.002454328 0.325329651
F28 M −0.63644 −0.002566 −1.0528 −0.446093 −0.64906 −0.62050 −0.6631870 −0.76402499 −0.73180160 −1.003922 −1.49425443 −1.4485531
SD 0.374682 0.003523 0.302257 0.133958 0.172862 0.319588 0.2236297 0.32784579 0.236382006 0.352104 0.163604635 0.162849789
F29 M 0.004303 1457.8834 5.988783 0.17355 2.203813 0.000159 5.5974E−05 0.000823461 0.000356285 0 6.9003E−08 0
SD 0.009469 1269.3624 7.334731 0.068175 4.386321 0.00052 8.7041E−05 0.001987679 0.000677345 0 3.27883E−07 0
F30 M 29.57348 1364.4556 781.55028 8.23344 35.971004 0.000543 9.6657E−05 0.000667392 0.000199627 0 0 0
SD 16.02108 1325.3797 1048.8135 8.092742 71.284369 0.000987 0.00023549 0.001469405 0.000449935 0 0 0

Source TLBO: Rao and Patel (2013); ABC, DE, PSO, and GA: Karaboga and Akay (2009), Jaya: Rao (2016)
Bold values indicate best values
4 Working of Jaya and Rao Optimization Algorithms …
Table 4.4 Friedman statistical test results for the results presented in Table 4.3
Algorithm GA PSO DE ABC TLBO Jaya ERao-1 ERao-2 ERao-3 SAP-Rao MTPG-Jaya AMTPG-Jaya
Avg. rank 9.6 8.35 7 6.82 6.87 5.48 5.8 6.15 6.2 5.23 5.2 5.27
χ2 43.621
p-value The p-value is 8.46E−06. The result is significant at p < 0.01
4.4 Computational Results Analysis on Single-objective …
97
98 4 Working of Jaya and Rao Optimization Algorithms …

Fig. 4.13 Average ranks of various algorithms in the Friedman statistical test for the unconstrained
standard benchmark problems

4.4.2.1 Experiment-1

In the first test, the Rao algorithms were tested on 23 unimodal and multimodal
optimization test problems by taking 30,000 function evaluations as the termination
criterion, and the statistical results were presented for 30 consecutive runs (Rao,
2020). Hence, for a fair comparison among the performances of the basic and the
modified algorithms, the modified algorithms along with the basic Jaya algorithm are
tested by taking 30,000 function evaluations as termination criterion, and statistical
results are presented for 30 consecutive runs. In Table 4.5, the performances of
the proposed modified algorithms are compared with those of the Jaya and Rao
algorithms in terms of the mean (M) values, standard deviation (SD), and the mean
of the function evaluations (MFE) taken by the algorithm to reach the global optimum
solution. The problems P1 to P7 are unimodal test problems, P8 to P13 are multimodal
test problems, and P14 to P23 are multimodal fixed dimension test problems. Here,
the performance of the SAP-Rao algorithm is compared with the performances of
the Rao algorithms. The performance of the ERao-1 algorithm is compared with
that of the Rao-1 algorithm. Similarly, the performances of the ERao-2 and ERao-3
algorithms are compared with those of the Rao-2 and Rao-3 algorithms, respectively.
Similarly, the performances of the MTPG-Jaya and AMTPG-Jaya algorithms are
compared with the Jaya algorithm’s performance.
Here, an observation can be made that the SAP-Rao algorithm has better explo-
ration capability when compared to that of the Rao algorithms. The SAP-Rao algo-
rithm has achieved better mean values in multimodal problems (P9 to P19 and P21
to P23) compared to those of the Rao algorithms. Similarly, the standard deviation
values of the SAP-Rao algorithm in multimodal problems are better or competitive
when compared to those of the Rao algorithms. Furthermore, the exploitation capa-
bility of the SAP-Rao algorithm is better or competitive to the Rao algorithms. In all
unimodal test problems, the SAP-Rao algorithm has achieved better mean and stan-
dard deviation values when compared to those of the Rao-1 and Rao-2 algorithms.
Table 4.5 Comparison of the results of the basic and improved algorithms in the experiment-1 of the unimodal and multimodal benchmark problems
Problem (Optimum Rao-1 Rao-2 Rao-3 ERao-1 ERao-2 ERao-3 SAP-Rao Jaya MTPG-Jaya AMTPG-Jaya
Number Value)
P1 (0) M 3.59E−22 3.57E−12 6.71E−42 7.1653E−21 8.31983E−13 1.21844E−42 2.497E−27 1.93353E−11 0 0
SD 7.33E−22 7.95E−12 1.56E−41 2.72917E−20 2.01641E−12 4.26759E−42 8.947E−27 9.42658E−11 0 0
MFE 29,998 29,953 29,991 30,000 29,972.66667 29,991 29,985 29,990 16,075 8659
P2 (0) M 4.07E−12 6.78E−01 9.33E−21 3.36004E−12 0.334779 2.27072E−19 1.315E−18 0.001589666 1.0535E−266 0
SD 1.4E−11 2.53E + 00 3.84E−20 1.23803E−11 1.825483173 1.1745E−18 2.678E−18 0.007652932 0 0
MFE 29,994 29,882 29,983 29,995 29,923.4 29,974.93333 29,981 29,990 29,855 16,378
P3 (0) M 8.34E−40 1.27E−16 1.68E−53 2.5186E−39 1.03267E−21 6.17941E−73 9.878E−55 2.10485E−16 0 0
SD 2.9E−39 6.93E−16 9.12E−53 1.00807E−38 5.42299E−21 3.3432E−72 4.350E−51 1.15055E−15 0 0
MFE 29,993 29,975 29,959 29,995 29,965 29,982.33333 29,991 29,991 9481 5015
P4 (0) M 2.19522 16.56395 0.081469 2.196461584 14.96336474 0.070440448 0.032234 35.13809811 8.4471E−156 3.5266E−296
SD 1.150517 5.632224 0.078402 1.011003951 6.393191671 0.102924171 0.020639 6.830014968 4.6266E−155 0
MFE 29,882 28,845 29,899 29,933 29,623.2 29,873.33333 29,925 29,991 29,988 17,122
P5 (0) M 31.6044 11.47408 29.2063 29.17288578 11.34218421 23.48199954 0.886832 25.0963 24.22616415 7.360013893
SD 28.4067 16.68387 29.0933 26.22552675 14.93837212 28.410953 1.163335 0.6335 0.830964354 6.581159362
MFE 29,609 28,925 28,922 29,477 29,716 29,822.83333 29,908 29,992 29,991 29,653
4.4 Computational Results Analysis on Single-objective …

P6 (0) M 2.63E−21 1.09E−07 2.919904 2.12382E−21 5.53027E−09 2.556397649 5.923E−12 42.89921891 1.74742E−06 1.2626E−09
SD 7.87E−21 3.09E−07 0.39977 2.94981E−21 1.5451E−08 0.341602663 1.383E−11 39.59748517 5.49956E−06 6.7466E−09
MFE 29,993 29,945 20,023 29,989 29,984 18,843.5 29,981 24,427 29,949 29,981
P7 (0) M 0.058328 0.087804 0.01577 0.070803102 0.079307038 0.013706853 0.011698 0.287414385 0.286096194 9.72796E−05
SD 0.027453 0.044495 0.008669 0.026594612 0.04049604 0.008316235 0.004410 0.145627618 0.163397722 0.00016837
MFE 26,785 25,355 24,044 26,331.66667 25,270 24,839 26,448 28,552 28,580 17,103
(continued)
99
Table 4.5 (continued)
100

Problem (Optimum Rao-1 Rao-2 Rao-3 ERao-1 ERao-2 ERao-3 SAP-Rao Jaya MTPG-Jaya AMTPG-Jaya
Number Value)
P8 M −8685.17 −8757.58 −9664.70 −9168.326455 −9738.589736 −9745.532039 −9416.37 −5644.26 −4430.431501 −10,877.30491
(−12,569) SD 1690.55 1896.34 1544.66 1647.69389 923.4525188 1629.643643 589.57 1181.28 928.8317877 1592.909786
MFE 21,166 22,377 28,385 29,090 27,573.33333 28,282.8 29,292 24,072 20,786 27,577
P9 (0) M 87.0136 148.9495 84.1229 91.65656892 98.91672012 34.53251765 41.8233 76.7882 0 0
SD 32.3175 41.5267 38.1792 44.50224678 49.82896302 33.7673293 14.1592 24.0204 0 0
MFE 26,015 24,754 27,934 29,021.5 28,196.66667 28,948 29,771 28,965 2963 1857
P10 (0) M 0.619739 0.170688 7.97E−08 1.08289E−10 7.03105E−06 2.13755E−14 2.718E−14 4.238427857 8.88178E−16 8.88178E−16
SD 0.695792 0.31832 8.69E−08 2.45761E−10 3.00679E−05 1.17898E−14 3.695E−14 6.20468532 0 0
MFE 29,929 29,881 29,919 29,991 29,951 21,731.5 28,758 27,512 3003 1761
P11 (0) M 0.011455 0.044885 0.028906 0.00886436 0.028550199 0.007132844 5.884E−06 0.341604681 0 0
SD 0.014397 0.066572 0.042806 0.009729053 0.027959171 0.01247169 1.227E−05 0.974177478 0 0
MFE 29,971 29,406 21,654 29,961 29,892.5 20,597.33333 29,885 29,743 2004 1444
P12 (0) M 1.549523 6.222186 0.791997 1.29548785 0.632098973 0.32481914 0.005735 45.30554917 31.84073202 0.003214338
SD 1.49792 7.075035 0.372832 2.02404146 0.757373886 0.082793378 0.019997 52.69414276 10.70135226 0.013707242
MFE 29,957 28,537 26,432 29,979 29,977.2 24,858.16667 29,692 28,012 27,704 29,405
P13 (0) M 0.024281 0.458132 0.009724 1.351972689 0.771735 0.008753439 0.002564 4.394105597 0.984793828 0.494615064
SD 0.078964 0.638728 0.026098 3.692304456 1.244696081 0.026151805 0.008503 7.706212251 1.276655337 1.029536115
MFE 29,927 29,996.33333 29,947 29,655.83333 29,998 29,984.16667 29,946 29,596 27,959 29,927
P14 (0.998) M 0.998004 0.998004 0.998116 0.998004 0.998004 0.998100837 0.998004 0.99828498 1.064140548 1.097275325
SD 8.25E−17 2.43E−08 0.000251 7.1417E−17 0 0.000162056 0 0.000955316 0.362245683 0.399531422
MFE 12,013 24,069 14,583 6474 16,239 14,333.33333 8979 18,213 24,152 22,596
(continued)
4 Working of Jaya and Rao Optimization Algorithms …
Table 4.5 (continued)
Problem (Optimum Rao-1 Rao-2 Rao-3 ERao-1 ERao-2 ERao-3 SAP-Rao Jaya MTPG-Jaya AMTPG-Jaya
Number Value)
P15 M 0.0014295 0.0006656 0.0004858 0.001237618 0.00053442 0.000438635 0.000418 0.0010676 0.000720215 0.00042325
(0.0003) SD 0.0035890 0.0005148 0.0003264 0.003618665 0.000428917 0.000179217 0.000157 0.0036482 0.001477502 0.000248732
MFE 21,826.66667 23,386 21,737 25,300 20,273.33333 20,101.66667 24,297 24,158 19,112 25,975
P16 M −1.031627 −1.031626 −1.031628 −1.031628453 −1.031628453 −1.03162813 −1.031628 −1.031626471 −1.031628403 −1.031628205
(−1.0316) SD 0.00000436 0.00000739 8.39E−08 4.87871E−16 5.83118E−16 5.11231E−07 6.775E−16 5.21578E−06 5.18055E−13 1.04625E−06
MFE 2577 4612 20,283 1035.666667 4483 18,786 3731 16,643 15,087 14,024
P17 M 0.397887 0.397887 0.397887 0.397887 0.397887 0.397887 0.397887 0.397887 0.397887 0.397887
(0.397887) SD 0 0 0 0 0 0 0 0 0 0
MFE 995 695 692 791 773 766 682 3387 2972 2342
P18 (3) M 3 3 3.000021 3 3 3.000017892 3 3.000043566 5.700060063 3
SD 9E−16 6.06E−16 0.000033 1.21E−16 5.71336E−16 3.08598E−05 1.293E−15 4.22765E−05 14.78849771 1.80296E−15
MFE 10,031 18,098 22,145.6 9774.5 9419 20,921.2 9398 20,723 18,815 16,730
P19 M −3.86278 −3.86278 −3.86278 −3.86278 −3.86278 −3.86278 −3.86278 −3.86278 −3.86278 −3.86278
(−3.86) SD 1.56E−15 3.11E−15 3.06E−15 3.17786E−15 3.16177E−15 3.16177E−15 3.148E−15 3.14776E−15 3.16177E−15 3.16177E−15
MFE 575 4093 6680 1173 4048 3207.5 2094 3296 3236 3191
4.4 Computational Results Analysis on Single-objective …

P20 M −3.286657 −3.29792 −3.278659 −3.286487924 −3.282632647 −3.286487924 −3.296739 −3.242896976 −3.246870819 −3.258630447
(−3.32) SD 0.05664 0.05719 0.058427 0.055748152 0.057154947 0.055748152 0.060487 0.057155167 0.058426783 0.060646486
MFE 8003 2799 6916 4120 5043 3713.2 10,065 8204 8105 8692
P21 M −7.566177 −8.405803 −8.168698 −8.263266404 −9.485500142 −9.016394549 −9.463294 −5.121234807 −5.832555732 −7.135195843
(−10.1532) SD 2.413688 2.391694 2.693478 2.81732706 1.499163896 2.102883561 1.676048 3.216961611 3.418373019 2.507042827
MFE 11,371 11,016 13,321 10,927 26,130 20,521.66667 20,252 3159 4879 13,767
(continued)
101
Table 4.5 (continued)
102

Problem (Optimum Rao-1 Rao-2 Rao-3 ERao-1 ERao-2 ERao-3 SAP-Rao Jaya MTPG-Jaya AMTPG-Jaya
Number Value)
P22 M −8.760775 −10.108301 −9.976039 −10.40294056 −10.40294057 −10.40108248 −10.402941 −6.275389632 −10.08284131 −10.36102021
(−10.4029) SD 2.146664 1.004131 0.626313 8.55657E−09 3.29861E−16 0.009767544 1.189E−15 3.538623556 1.418549935 0.129930888
MFE 13,592 17,633 22,713 29,208 20,097 29,420 17,914 5794 15,048 14,672.33333
P23 M −9.570118 −10.470286 −10.486057 −10.53640982 −10.53640982 −10.53639413 −10.536410 −6.253818899 −10.53042573 −10.536410
(−10.5364) SD 1.598056 0.212811 0.275792 2.84401E−12 2.21278E−15 8.59026E−05 1.776E−15 3.661325977 0.032776206 2.28535E−15
MFE 16,652 26,983 18,602 25,374 15,089.33333 29,745 15,853 7077.333333 14,370.66667 16,153

Source Rao-1, Rao-2, Rao-3: Rao (2020)

Result in boldface indicates better values
4 Working of Jaya and Rao Optimization Algorithms …
4.4 Computational Results Analysis on Single-objective … 103

For the test problems P4, P5, P7, P9, P10, P11, P12, P13, P15, P16, P21, P22, and
P23, the SAP-Rao algorithm has achieved better mean and standard deviation values
when compared to those of the Rao algorithms.
For the problems P14, P17, P18, and P19, the SAP-Rao algorithm has achieved
better or the same performance in terms of the mean and standard deviation when
compared to those of the Rao algorithms. For the test problems P1, P2, P3, P6, P8,
and P20, the SAP-Rao algorithm has achieved better or competitive performance
in terms of the mean and standard deviation when compared to those of the Rao
algorithms. In addition, the AMTPG-Jaya algorithm has achieved better results in 21
out of 23 problems when compared to those of the MTPG-Jaya and Jaya algorithms.
Similarly, the MTPG-Jaya algorithm has achieved better results in 20 out of 23
problems compared to those of the Jaya algorithm.
Furthermore, the elitist Rao algorithms have better exploration and exploitation
capabilities than those of the Rao algorithms. In unimodal test problems P1 to P7,
except for the problem P2, the ERao-3 algorithm has outperformed the Rao-3 algo-
rithm in terms of the mean values and MFE. Similarly, in all unimodal test problems
except for the problem P5, the ERao-2 algorithm has achieved better results in terms
of the mean and MFE values when compared to those of the Rao-2 algorithm. The
ERao-1 algorithm has achieved better or competitive results compared to those of
the Rao-1 algorithm in unimodal test problems.
Similarly, in multimodal test problems P8 to P13, the ERao-3 algorithm has
achieved better results than those achieved by the Rao-3 algorithm in terms of the
mean and MFE values. Similarly, the ERao-2 algorithm has achieved better results in
terms of mean values when compared to those of the Rao-2 algorithm for the multi-
modal problems P9 to P12. The ERao-1 algorithm has achieved better or competitive
results when compared to those of the Rao-1 algorithm in solving multimodal test
problems.
Furthermore, the ERao-3 algorithm has outperformed the Rao-3 algorithm in
solving the multimodal fixed dimension problems P14 to P23. For all the multi-
modal fixed dimension problems, the ERao-3 algorithm has achieved superior results
in terms of the mean and MFE values compared to those of the Rao-3 algorithm.
Similarly, in all multimodal fixed dimension test problems except for the problem
P20, the ERao-1 algorithm has achieved better results in terms of best and mean when
compared to those of the Rao-1 algorithm. The ERao-2 algorithm has achieved better
results in terms of best and mean when compared to those of the Rao-2 algorithm in
solving all multimodal fixed dimension test problems except for the problems P15
and P20.
Here, an observation can be made that the modified versions are converging faster
than the respective basic algorithms. However, the performances of these algorithms
in solving these benchmark problems are evaluated using the Friedman statistical
test. The algorithms are ranked based on their statistical results achieved in solving
these problems. Here, while ranking the algorithms, priority is given to the mean
value achieved, then the MFEs value achieved, and then the standard deviation value.
Ten algorithms are compared in this test, with nine degrees of freedom for the χ 2
distribution. The average ranks, χ 2 value, and p-value achieved in the Friedman test
104 4 Working of Jaya and Rao Optimization Algorithms …

are presented in Table 4.6. The ranks of these algorithms and their average rank
values are plotted in Fig. 4.14.
Here, an observation can be made that the p-value of the Friedman test is much
lesser than 0.01, which indicates that the results are highly significant. The least
value of the average rank indicates superior performance. The SAP-Rao algorithm
has achieved the least average rank of 2.57. The ranking of these algorithms in
solving unimodal and multimodal benchmark problems is SAP-Rao, AMTPG-Jaya,
ERao-3, ERao-1, ERao-2, Rao-3, MTPG-Jaya, Rao-1, Rao-2, and Jaya algorithms.
Furthermore, from the mean function evaluations (MFE) consumed by algorithms, it
can be observed that the proposed algorithms are converging faster than the respective
basic algorithms.

4.4.2.2 Experiment-2

In the second experiment, the performances of the proposed modified algorithms

are evaluated with respect to the performances of the real-coded genetic algorithm
(RGA), gravitational search algorithm (GSA), PSO and its variants, Jaya, and self-
adaptive multi-population Jaya (SAMP-Jaya) algorithms in solving ten unimodal and
multimodal test functions. The description, along with the ranges of design variables
of these functions, is presented in Appendix A3. For all the functions, 30 design
variables are considered. The computational results presented by various algorithms
are with either 50,000 function evaluations or 200,000 function evaluations. Hence,
the proposed modified algorithms are tested separately by taking 50,000 and 200,000
function evaluations as the termination criterion. The algorithms are executed for 30
consecutive runs, and statistical results are presented in terms of mean (M) and
standard deviation (SD).
The test functions TF1 to TF7 are unimodal test functions, and TF8 to TF10
are multimodal test functions. All these test functions are minimization problems.
The performances are compared with PSO (Eberhart & Kennedy, 1995), cooperative
PSO (CPSO) (van den Bergh and Engelbrecht, 2004), fully informed particle swarm
(FIPS) (Mendes et al., 2004), RGA (Haupt and Haupt, 2003), comprehensive learning
PSO (CLPSO) (Liang et al., 2006), Frankenstein’s PSO (F-PSO) (Montes de Oca
et al., 2009), GSA (Rashedi et al., 2009), adaptive inertia weight PSO (AIWPSO)
(Nickabadi et al., 2011), extraordinariness particle swarm optimizer (EPSO) (Ngo
et al., 2016), Jaya, and its variant SAMP-Jaya (Rao & Saroj, 2017) algorithms.
Table 4.7 presents the comparison of the results achieved by the proposed modified
algorithms with those of other algorithms in 50,000 function evaluations.
From Table 4.7, it can be observed that the AMTPG-Jaya algorithm has outper-
formed all the algorithms compared in solving 9 out of 10 (TF1–TF7, TF9, and
TF10) functions. Similarly, the ERao-3, SAP-Rao, and MTPG-Jaya algorithms have
achieved better or the same results in solving 8 out of 10 problems when compared to
the performances of the other algorithms from the literature. The ERao-1 and ERao-
2 algorithms have achieved better or the same performances in solving 6 out of 10
problems compared to other algorithms from the literature. Similar to experiment-1,
Table 4.6 Friedman statistical test results for the results presented in Table 4.5
Algorithm Rao-1 Rao-2 Rao-3 ERao-1 ERao-2 ERao-3 SAP-Rao Jaya MTPG-Jaya AMTPG-Jaya
Avg. Rank 6.26 6.96 5.96 5.17 5.26 4.13 2.57 9.13 6.04 3.48
χ2 76.89249012
p-value The p-value is <0.00001. The result is significant at p < 0.01
4.4 Computational Results Analysis on Single-objective …
105
106 4 Working of Jaya and Rao Optimization Algorithms …

Fig. 4.14 Average ranks of Jaya and Rao algorithms along with their modified versions in the
Friedman statistical test of the unimodal and multimodal benchmark problems of experiment-1

to judge the performance of the proposed modified algorithms with the other algo-
rithms, the Friedman statistical test is conducted. The mean values are given priority
and then to the standard deviation values for ranking the algorithms in all the prob-
lems. Here, 11 algorithms are compared in this test with ten degrees of freedom
for the χ 2 distribution. The average ranks, χ 2 value, and p-value achieved in the
Friedman test are presented in Table 4.8. The ranks of these algorithms and their
average rank values are plotted in Fig. 4.15.
The p-value of the Friedman test is 8.04989e−7, which indicates that the results are
highly significant. The AMTPG-Jaya algorithm has achieved the least average rank
of 2.85. The ranking of these algorithms in solving the unimodal and multimodal test
functions is AMTPG-Jaya, MTPG-Jaya, ERao-3, SAP-Rao, ERao-1, SAMP-Jaya,
EPSO, ERao-2, Jaya, RGA, and GSA. Here, an observation can be made that the
proposed algorithms have better or competitive performance compared to those of
the GSA, RGA, EPSO, Jaya, and SAMP-Jaya algorithms.
Table 4.9 presents the comparison of the results of the proposed modified algo-
rithms with those of other algorithms in 200,000 function evaluations. Here, an
observation can be made that the AMTPG-Jaya, SAP-Rao, and MTPG-Jaya algo-
rithms have achieved better or the same results in 8 out of 10 problems. Similarly,
the elitist Rao algorithms have achieved better or the same results for the problems
TF1 to TF7 and achieved competitive results for the problems TF8 to TF10, when
compared to the results achieved by the other algorithms from the literature. In addi-
tion, to judge the performance of the proposed modified algorithms with the other
algorithms, the Friedman statistical test is conducted. The mean values are given
priority and then to the standard deviation values for ranking the algorithms in all the
problems. Here, 15 algorithms are compared in this test with 14 degrees of freedom
for the χ 2 distribution. The average ranks, χ 2 value, and p-value achieved in the
Friedman test are presented in Table 4.10. The ranks of these algorithms and their
average rank values are plotted in Fig. 4.16.
The p-value of the Friedman test is 4.049E−05, which indicates that the results are
highly significant. The AMTPG-Jaya algorithm has achieved the least average rank
of 4.3. The ranking of these algorithms in solving the unimodal and multimodal test
Table 4.7 Comparison of the results of proposed modified algorithms with the results of other algorithms achieved in 50,000 function evaluations for the
optimization of the unimodal and multimodal test functions
Test RGA GSA EPSO Jaya SAMP-Jaya ERao-1 ERao-2 ERao-3 MTPG-Jaya AMTPG-Jaya SAP-Rao
function
TF1 M 23.13 6.8000E−17 7.7760E−18 7.3215E−19 2.7998E−19 1.1515E−38 3.6207E−21 1.2821E−73 0 0 5.9000E−46
SD 12.15 2.1800E−17 6.0100E−18 2.8206E−18 8.2707E−19 2.6622E−38 1.3790E−20 5.3552E−73 0 0 2.2825E−45
TF2 M 1.073 6.0600E−08 6.7870E−12 4.5329E−11 1.3439E−12 5.3155E−17 4.6506E−02 2.1707E−36 0 0 1.0639E−29
SD 0.2666 1.1900E−08 3.0080E−12 1.0544E−10 2.1634E−11 1.5376E−16 1.1723E−01 1.0270E−35 0 0 2.6690E−29
TF3 M 561.7 942.7 0.2121 5.6234E−26 1.5464E−35 8.6098E−72 1.2778E−40 1.438E−138 0 0 3.6007E−90
SD 125.6 246.6 0.5461 1.6346E−25 4.4840E−35 4.5044E−71 4.0768E−40 7.875E−138 0 0 1.8095E−89
TF4 M 11.78 4.207 9.9410E−03 5.060152 1.152804 0.304880134 8.394887097 1.19E−03 1.86E−296 0 8.3423E−04
SD 1.576 1.122 9.8550E−03 7.352134 2.808808 0.154216849 4.899670188 2.85E−03 0 0 1.8778E−03
TF5 M 1180 47.95 1.7850E−02 1.7527E−06 1.8006E−09 2.2083E−20 1.1122E−12 1.9122E−19 1.0092E−07 1.1459E−12 1.2442E−12
SD 548.1 3.956 2.1360E−02 1.7527E−05 1.8443E−08 9.2372E−20 4.4456E−12 4.2324E−19 3.9963E−07 2.8337E−12 4.3990E−12
TF6 M 24.01 0.931 0 0 0 0 0 0 0 0 0
SD 10.17 2.51 0 0 0 0 0 0 0 0 0
TF7 M 0.0675 0.0782 6.4700E−04 7.4146E−19 3.0770E−28 1.1053E−23 5.4179E−13 1.0485E−54 5.233E−197 0 8.7612E−36
4.4 Computational Results Analysis on Single-objective …

SD 0.0287 0.041 4.5420E−04 4.0529E−18 1.1412E−27 2.7420E−23 1.5510E−12 3.0905E−54 0 0 3.9283E−35

TF8 M −12,480 −3604 −12,570.00 −11,466.34 −11,763.96 −9124.64 −9224.39 −10,343.10 −6599.25 −6234.76 −10,136.59
SD 53.26 564.1 3.85E−12 1288.14 838.95 1297.42 1557.99 1008.49 1244.76 1073.75 1042.49
TF9 M 5.902 29.4 2.2740E−14 116.247 68.517 89.413 166.264 86.420 0 0 28.139
SD 1.171 4.727 2.8320E−14 58.292 32.544 21.211 38.819 45.287 0 0 8.828
TF10 M 2.14 4.8000E−09 1.2840E−09 9.3528E−11 7.0737E−11 5.9626E−14 1.0171E−10 1.2257E−14 8.8818E−16 8.8818E−16 6.9278E−15
(continued)
107
Table 4.7 (continued)
108

Test RGA GSA EPSO Jaya SAMP-Jaya ERao-1 ERao-2 ERao-3 MTPG-Jaya AMTPG-Jaya SAP-Rao
function
SD 0.4014 5.4200E−10 7.2800E−10 1.4691E−10 6.7168E−11 2.2867E−14 5.1388E−10 3.5404E−15 0 0 1.6559E−15
Source RGA, GSA, EPSO, Jaya, and SAMP-Jaya: Rao and Saroj (2017)
Result in boldface indicates better values
4 Working of Jaya and Rao Optimization Algorithms …
Table 4.8 Friedman statistical test results for the results presented in Table 4.7
Algorithm RGA GSA EPSO Jaya SAMP-Jaya ERao-1 ERao-2 ERao-3 MTPG-Jaya AMTPG-Jaya SAP-Rao
Avg. rank 9.2 9.6 6.6 7.4 5.9 5.5 7.4 4 3.25 2.85 4.3
χ2 47.377
p-value The p-value is 8.04989e−7. The result is significant at p < 0.01
4.4 Computational Results Analysis on Single-objective …
109
110 4 Working of Jaya and Rao Optimization Algorithms …

Fig. 4.15 Average ranks of various algorithms in the Friedman statistical test for the unimodal and
multimodal test functions results with 50,000 function evaluations

functions is AMTPG-Jaya, MTPG-Jaya, SAP-Rao, ERao-3, SAMP-Jaya, ERao-1,

ERao-2, Jaya, EPSO, AIWPSO, CLPSO, PSO, FIPSO, CPSO, and F-PSO. Here,
an observation can be made that the proposed algorithms have better or competitive
performance when compared to those of the EPSO, AIWPSO, CLPSO, PSO, FIPSO,
CPSO, F-PSO, Jaya, and SAMP-Jaya algorithms.
For all the unimodal and multimodal test functions, the proposed algorithms have
attained better or competitive results in terms of mean and standard deviation values.
It indicates that the exploration and exploitation capability of the proposed modified
algorithms are better or competitive than that of the other algorithms compared.
In addition, the Friedman statistical test confirms that the performances of the
proposed modified algorithms are better or competitive than the performances of
basic algorithms as well as the other algorithms compared.
From the computational results of the proposed modified algorithms in single-
objective optimization benchmark problems, it can be observed that the proposed
modified algorithms are more robust in finding better solutions than the basic Jaya
and Rao algorithms. Also, the modified versions are converging faster than the basic
Jaya and Rao algorithms. In addition, the performances of the modified versions are
better than or competitive to those of the other algorithms compared from the litera-
ture. Hence, the modified versions of the Jaya and Rao algorithms can be considered
as the improved versions of the Jaya and Rao algorithms, respectively. Here, it can
be noted that the AMTPG-Jaya algorithm is an improved version of the MTPG-Jaya
algorithm. Furthermore, the performance of the AMTPG-Jaya algorithm precedes
the performance of the MTPG-Jaya algorithm in solving almost all the problems.
Hence, the MTPG-Jaya algorithm is excluded in solving the multi-objective opti-
mization problems and case studies. The next subsection presents the analysis of the
computational results of the proposed modified algorithms in solving multi-objective
optimization benchmark problems.
Table 4.9 Comparison of the results of proposed modified algorithms with the results of other algorithms achieved in 200,000 function evaluations for the
optimization of the unimodal and multimodal test functions
Test PSO CPSO CLPSO FFIPS F-PSO AIWPSO EPSO Jaya SAMP-Jaya ERao-1 ERao-2 ERao-3 MTPG-Jaya AMTPG-Jaya SAP-Rao
function
TF1 M 5.2E−70 5.1E−13 4.9E−39 4.6E−27 2.4E−16 3.3E−133 1.7E−74 7.1E−90 4.9E−89 2.5E−167 2.5E−144 3.9E−279 0 0 2E−203
SD 1.1E−74 7.8E−25 6.8E−39 2.0E−53 2.0E−31 5.2E−267 2.8E−75 3.4E−89 1.9E−88 0 7.7E−144 0 0 0 0
TF2 M 2.1E−25 1.3E−07 8.9E−24 2.3E−16 1.6E−11 1.4E−54 1.9E−47 6.9E−69 9.5E−96 6.2E−124 1.2E−109 3.2E−155 0 0 2E−124
SD 1.4E−49 1.2E−14 7.9E−49 1.1E−32 1.0E−22 7.4E−119 2.2E−47 3.8E−68 4.4E−95 1.7E−123 3.2E−109 1.3E−154 0 0 1E−123
TF3 M 1.458 1889 192.2 9.463 173.2 1.9E−10 0.002014 1.2E−132 0 0 0 0 0 0 0
SD 1.78 9,911,000 384.3 25.98 9158 1.2E−19 0.001934 2.9E−132 0 0 0 0 0 0 0
TF4 M − − − − − − − 7.16E−05 6.91E−08 3.07E−12 8.48E−09 6.46E−15 1E−296 0 2E−17
SD − − − − − − − 2.00E−04 6.91E−08 3.38E−10 9.18E−08 2.24E−14 0 0 4E−17
TF5 M 25.4 0.8265 13.22 26.71 28.16 2.5 2.8E−05 1.8E−17 2.3E−25 1.9E−30 2.5E−31 1.2E−29 1E−07 1.1E−12 9E−27
SD 590.3 2.345 214.8 200.3 231.3 16 3.7E−05 9.7E−17 6.2E−25 6.5E−30 1.4E−30 2.4E−29 4E−07 2.8E−12 4E−26
TF6 M 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
SD 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
TF7 M 1.2E−02 1.1E−02 4.1E−03 3.3E−03 4.2E−03 5.5E−03 2.6E−04 1.0E−106 2.2E−111 9.8E−125 1.6E−115 9.7E−221 5E−197 0 1E−143
SD 2.3E−05 2.8E−05 9.6E−07 8.4E−03 2.4E−06 1.5E−05 1.9E−04 5.7E−106 1.2E−110 3.7E−124 6.9E−113 0 0 0 8E−143
TF8 M −11,000 −12,130 −12,550 −11,050 −11,220 −12,570 −12,570 −11,225 −12,276 −9467 −10,732 −11,199 −6599 −6235 −10,432
4.4 Computational Results Analysis on Single-objective …

SD 137,500 33,800 4257 944,200 222,700 1.1E−25 2.5E−12 1070 570 1189 1684 749 1245 1074 1131
TF9 M 34.76 3.6E−13 0 58.5 73.84 0.1658 0 76.982 59.246 92.410 116.493 53.707 0 0 24.01
SD 106.4 1.5E−24 0 191.9 370.6 0.2105 0 26 24 43 43 23 0 0 9
TF10 M 1.5E−14 1.6E−07 9.2E−15 1.4E−14 2.2E−09 7.0E−15 1.2E−14 5.2E−15 4.4E−15 3.1E−14 2.4E−14 1.4E−14 8E−16 8.9E−16 4E−15
(continued)
111
Table 4.9 (continued)
112

Test PSO CPSO CLPSO FFIPS F-PSO AIWPSO EPSO Jaya SAMP-Jaya ERao-1 ERao-2 ERao-3 MTPG-Jaya AMTPG-Jaya SAP-Rao
function
SD 1.9E−29 7.9E−14 6.6E−30 2.3E−29 1.7E−18 4.2E−31 3.1E−15 2.7E−15 0.0E + 00 9.6E−15 1.3E−14 5.4E−15 0 0 0

Source PSO, CPSO, CLPSO, FFIPS, FPSO, AIWPSO, EPSO, Jaya, and SAMP-Jaya: Rao and Saroj (2017)
Result in boldface indicates better values
4 Working of Jaya and Rao Optimization Algorithms …
Table 4.10 Friedman statistical test results for the results presented in Table 4.9
Algorithm PSO CPSO CLPSO FFIPS F-PSO AIWPSO EPSO Jaya SAMP-Jaya ERao-1 ERao-2 ERao-3 MTPG-Jaya AMTPG-Jaya SAP-Rao
Avg. 11.00 11.40 9.35 11.00 12.10 8.25 8.00 7.80 6.55 7.40 7.50 5.40 4.60 4.30 5.35
Rank
χ2 45.020
p-value The p-Value is 0.0000404896. The result is significant at p < 0.01
4.4 Computational Results Analysis on Single-objective …
113
114 4 Working of Jaya and Rao Optimization Algorithms …

Fig. 4.16 Average ranks of various algorithms in the Friedman statistical test for the unimodal and
multimodal test functions results with 200,000 function evaluations

4.5 Computational Results Analysis on Multi-objective

Optimization Benchmark Problems

The performances of the proposed modified algorithms are tested using five chal-
lenging multi-objective optimization test problems known as ZDT (Zitzler, Deb, and
Thiele’s) test problems (Zitzler et al., 2000). These problems are ZDT-1, ZDT-2, ZDT-
3, ZDT-1 with linear Pareto-front (ZDT-1L), and ZDT-2 with 3 objectives (ZDT2-
3O). The ZDT-1 problem has a convex-shaped Pareto-front. The ZDT-2 problem has
a concave-shaped Pareto-front, and the ZDT-3 problem has a discrete convex Pareto-
front. A detailed description, including the number of independent variables and
their ranges of the considered multi-objective optimization problems, is presented in
Appendix B1 (Mirjalili, 2016).
In multi-objective optimization, a posteriori approach is implemented to handle all
the objectives simultaneously. In computational experiments on these test problems,
the proposed algorithms are tested for 10,000 function evaluations, and the statistical
results are presented for ten consecutive runs. In computational experiments, popula-
tion size is taken as 100 for solving these problems. In the execution of the SAP-Rao
algorithm, a minimum of 100 population size is preserved to maintain consistency
with other algorithms. Also, different elite sizes are used for elitist Rao algorithms
based on population size. The elite size is varied from 10 to 40% of the population
size. The computational results of the proposed algorithms are presented in terms
of IGD and spacing values. The performances of the proposed modified algorithms
in terms of IGD values are compared with those of the basic algorithms as well
as the multi-objective dragonfly algorithm (MODA), multi-objective grasshopper
optimization algorithm (MOGOA), multi-objective ant lion optimizer (MOALO),
multi-objective particle swarm optimization (MOPSO), and non-dominated sorting
genetic algorithm (NSGA-II). The spacing values achieved by the MODA, MOGOA,
MOALO, MOPSO, and NSGA-II were not available in the literature. Hence, the
performances of the proposed modified algorithms in terms of spacing values are
compared with those of the basic algorithms.
4.5 Computational Results Analysis on Multi-objective … 115

Table 4.11 compares the statistical results achieved by the proposed modified
algorithms with other algorithms to solve the ZDT-1 problem. The statistical results
in all the tables are presented in terms of best, worst, mean values, median, and
standard deviation (SD). Here, an observation can be made that the proposed modified
algorithms and basic Jaya and Rao algorithms have outperformed the other algorithms
from the literature in terms of IGD values. The SAP-Rao algorithm has achieved the
least mean IGD and spacing values. The next best mean IGD values are achieved by
ERao-3, ERao-1, AMTPG-Jaya, and ERao-2 algorithms, respectively. The MOPSO
algorithm has the best IGD value. However, the proposed modified algorithms have
achieved better mean values than the other algorithms. It indicates that the proposed
algorithms are consistent in finding the optimal Pareto-front.
Similarly, the comparison of the statistical results achieved by the proposed modi-
fied algorithms with other algorithms in solving the ZDT-2 problem is presented in
Table 4.12. Here, an observation can be made that the proposed modified algo-
rithms and basic Jaya and Rao algorithms have outperformed the MOGOA, NSGA-
II, MODA, and MOALO algorithms in terms of IGD values. The performances of
the proposed algorithms are competitive with the MOPSO algorithm. The SAP-Rao
algorithm has achieved the least mean spacing value. The MOPSO algorithm has
achieved the least mean IGD value. The next best mean IGD values are achieved by
ERao-3, ERao-2, SAP-Rao, and AMTPG-Jaya algorithms, respectively.
Table 4.13 presents the comparison of the statistical results achieved by the
proposed modified algorithms with other algorithms in solving the ZDT-3 problem.
Here, an observation can be made that the proposed modified algorithms and basic
Jaya and Rao algorithms have outperformed the other algorithms from the literature
in terms of IGD values. The ERao-3 algorithm has achieved the least mean IGD and
spacing values. The next best mean IGD values are achieved by SAP-Rao, ERao-2,
ERao-2, and AMTPG-Jaya algorithms.
Similarly, the comparison of the statistical results achieved by the proposed modi-
fied algorithms with other algorithms in solving the ZDT-1L problem is presented
in Table 4.14. For this problem also, the proposed modified algorithms and basic
Jaya and Rao algorithms have outperformed the other algorithms compared in terms
of mean IGD values. The SAP-Rao algorithm has achieved the least mean spacing
value. The ERao-2 algorithm has achieved the least mean IGD value. The next best
mean IGD values are achieved by ERao-1, AMTPG-Jaya, ERao-3, and SAP-Rao
algorithms, respectively. Similarly, Table 4.15 presents the comparison of the statis-
tical results achieved by the proposed modified algorithms with other algorithms in
solving the ZDT2-3O problem. In this problem also, the proposed modified algo-
rithms and basic Jaya and Rao algorithms have outperformed the other algorithms
from the literature in terms of IGD values. The SAP-Rao algorithm has achieved the
least mean IGD value. The next best mean IGD values are achieved by AMTPG-
Jaya, ERao-3, ERao-2, and Jaya algorithms, respectively. The ERao-1 algorithm has
achieved the least mean spacing value.
In addition, for qualitative assessment of the algorithms’ performances, the Pareto-
fronts achieved by the proposed modified algorithms, along with Jaya and Rao
algorithms, are presented in Figs. 4.17, 4.18, 4.19, 4.20 and 4.21.
 116

Table 4.11 Comparison of the results achieved by various algorithms in multi-objective optimization of the ZDT-1 problem
Algorithm IGD Spacing
Best Worst M Median SD Best Worst M Median SD
MOGOA 0.0028 0.0822 0.0121 0.0046 0.0247 − − − − −
MOPSO 0.0015 0.0101 0.00422 0.0037 0.003103 − − − − −
NSGA-II 0.0546 0.0702 0.05988 0.0574 0.005436 − − − − −
MODA 0.0024 0.0096 0.00612 0.0072 0.002863 − − − − −
MOALO 0.0061 0.0209 0.01524 0.0166 0.005022 − − − − −
Jaya 0.0022848 0.0036293 0.0028700 0.0027842 0.0004846 0.0030862 0.0097681 0.0054556 0.0045533 0.0023200
Rao-1 0.0022577 0.0031300 0.0026479 0.0025347 0.0003014 0.0030372 0.0095579 0.0049653 0.0047373 0.0020980
Rao-2 0.0023985 0.0040621 0.0032697 0.0032697 0.0006065 0.0037399 0.0086020 0.0055581 0.0049339 0.0017361
Rao-3 0.0022729 0.0025932 0.0024277 0.0024228 0.0001149 0.0030360 0.0037921 0.0033705 0.0033788 0.0002450
ERao-1 0.0022155 0.0025364 0.0023957 0.0023957 0.0000959 0.0032215 0.0037840 0.0034621 0.0034621 0.0001688
ERao-2 0.0022931 0.0025271 0.0024221 0.0024221 0.0000846 0.0028235 0.0038067 0.0032328 0.0031770 0.0003324
ERao-3 0.0022214 0.0024956 0.0023560 0.0023916 0.0000957 0.0029571 0.0036882 0.0033311 0.0033311 0.0002319
AMTPG-Jaya 0.0023440 0.0026824 0.0024228 0.0024110 0.0001099 0.0031013 0.0039536 0.0034919 0.0034595 0.0002897
SAP-Rao 0.0021807 0.0024752 0.0023407 0.0023407 0.0000869 0.0028501 0.0032986 0.0030606 0.0030606 0.0001578
Source MOGOA, MOPSO, NSGA-II: Mirjalili et al. (2018); MODA: Mirjalili (2016); MOALO: Mirjalili et al. (2017)
Result in boldface indicates better values
4 Working of Jaya and Rao Optimization Algorithms …
Table 4.12 Comparison of the results achieved by various algorithms in multi-objective optimization of the ZDT-2 problem
Algorithm IGD Spacing
Best Worst M Median SD Best Worst M Median SD
MOGOA 0.0016 0.0273 0.007 0.0049 0.009 − − − − −
MOPSO 0.0013 0.0017 0.00156 0.0017 0.000174 − − − − −
NSGA-II 0.1148 0.1834 0.13972 0.1258 0.026263 − − − − −
MODA 0.0023 0.006 0.00398 0.0033 0.001604244 − − − − −
MOALO 0.005 0.0377 0.01751 0.0165 0.010977 − − − − −
Jaya 0.0030770 0.0038014 0.0033411 0.0031579 0.0002801 0.0038416 0.0052066 0.0044104 0.0042834 0.0004967
Rao-1 0.0024935 0.0038934 0.0029074 0.0027294 0.0004580 0.0031799 0.0041586 0.0035683 0.0035279 0.0002996
Rao-2 0.0023552 0.0041087 0.0029883 0.0027617 0.0006216 0.0031329 0.0064620 0.0042046 0.0039475 0.0011133
Rao-3 0.0030140 0.0036245 0.0032858 0.0032330 0.0002315 0.0035532 0.0049834 0.0042331 0.0042331 0.0004602
4.5 Computational Results Analysis on Multi-objective …

ERao-1 0.0024032 0.0032612 0.0026477 0.0025724 0.0002764 0.0029941 0.0051762 0.0036644 0.0035615 0.0006950
ERao-2 0.0022341 0.0026368 0.0024325 0.0024291 0.0001173 0.0028565 0.0035270 0.0031859 0.0031859 0.0002121
ERao-3 0.0022758 0.0026128 0.0024245 0.0024093 0.0001046 0.0026497 0.0035915 0.0032201 0.0032441 0.0003372
AMTPG-Jaya 0.0023580 0.0029221 0.0025997 0.0025184 0.0001928 0.0032547 0.0036704 0.0034484 0.0034484 0.0001430
SAP-Rao 0.0023364 0.0029362 0.0025754 0.0025504 0.0002138 0.0028400 0.0037287 0.0031847 0.0031474 0.0002785
Source MOGOA, MOPSO, NSGA-II: Mirjalili et al. (2018); MODA: Mirjalili (2016); MOALO: Mirjalili et al. (2017)
Result in boldface indicates better values
117
 118

Table 4.13 Comparison of the results achieved by various algorithms in multi-objective optimization of the ZDT-3 problem
Algorithm IGD Spacing
Best Worst M Median SD Best Worst M Median SD
MOGOA 0.0224 0.0345 0.0306 0.0313 0.0034 − − − − −
MOPSO 0.0308 0.0497 0.03782 0.0362 0.006297 − − − − −
NSGA-II 0.0315 0.0557 0.04166 0.0403 0.008073 − − − − −
MODA 0.02 0.0304 0.02794 0.0302 0.004021 − − − − −
MOALO 0.0303 0.033 0.03032 0.0323 0.000969 − − − − −
Jaya 0.0051872 0.0057729 0.0053927 0.0052673 0.0002199 0.0044896 0.0064756 0.0053651 0.0052044 0.0006667
Rao-1 0.0050920 0.0058165 0.0053540 0.0053084 0.0002343 0.0040658 0.0063537 0.0048799 0.0045751 0.0007317
Rao-2 0.0048858 0.0055722 0.0051537 0.0051575 0.0002144 0.0037246 0.0050765 0.0044082 0.0044386 0.0004175
Rao-3 0.0050084 0.0054534 0.0052065 0.0052065 0.0001564 0.0039955 0.0053726 0.0046838 0.0046838 0.0004642
ERao-1 0.0025187 0.0027268 0.0026053 0.0026053 0.0000626 0.0019391 0.0024701 0.0021656 0.0021563 0.0001654
ERao-2 0.0025197 0.0027046 0.0025945 0.0025803 0.0000609 0.0019289 0.0025008 0.0021647 0.0020792 0.0002345
ERao-3 0.0024503 0.0026772 0.0025445 0.0025275 0.0000752 0.0018835 0.0021931 0.0020180 0.0020180 0.0001020
AMTPG-Jaya 0.0029888 0.0033379 0.0031541 0.0031573 0.0001151 0.0022910 0.0027968 0.0025368 0.0025343 0.0001531
SAP-Rao 0.0024656 0.0027676 0.0025747 0.0025347 0.0001038 0.0018454 0.0023592 0.0020464 0.0019970 0.0001700
Source MOGOA, MOPSO, NSGA-II: Mirjalili et al. (2018); MODA: Mirjalili (2016); MOALO: Mirjalili et al. (2017)
Result in boldface indicates better values
4 Working of Jaya and Rao Optimization Algorithms …
Table 4.14 Comparison of the results achieved by various algorithms in multi-objective optimization of the ZDT-1L problem
Algorithm IGD Spacing
Best Worst M Median SD Best Worst M Median SD
MOGOA 0.0017 0.0498 0.0091 0.0023 0.0148 − − − − −
MOPSO 0.0012 0.0165 0.00922 0.0098 0.005531 − − − − −
NSGA-II 0.0773 0.0924 0.08274 0.0804 0.005422 − − − − −
MODA 0.0022 0.0163 0.00616 0.0038 0.005186 − − − − −
MOALO 0.0106 0.033 0.01982 0.0196 0.007545 − − − − −
Jaya 0.0029703 0.0033261 0.0031069 0.0030627 0.0001278 0.0039527 0.0064543 0.0046406 0.0044064 0.0008148
Rao-1 0.0028968 0.0033283 0.0030612 0.0030361 0.0001454 0.0035196 0.0051419 0.0045532 0.0046276 0.0005193
Rao-2 0.0028631 0.0034530 0.0030898 0.0030741 0.0001802 0.0037438 0.0044214 0.0041032 0.0041032 0.0002069
Rao-3 0.0027155 0.0033786 0.0028957 0.0027991 0.0002204 0.0033409 0.0047274 0.0040933 0.0041095 0.0004013
4.5 Computational Results Analysis on Multi-objective …

ERao-1 0.0021659 0.0024248 0.0023073 0.0023155 0.0000824 0.0030625 0.0036670 0.0033255 0.0033232 0.0002051
ERao-2 0.0022049 0.0024593 0.0023063 0.0023063 0.0000821 0.0027937 0.0034583 0.0031784 0.0031795 0.0002059
ERao-3 0.0021698 0.0025456 0.0023266 0.0022982 0.0001205 0.0032572 0.0036294 0.0034229 0.0034302 0.0001192
AMTPG-Jaya 0.0022559 0.0023868 0.0023089 0.0023089 0.0000445 0.0028441 0.0040103 0.0033502 0.0032842 0.0003506
SAP-Rao 0.0022487 0.0024537 0.0023403 0.0023392 0.0000711 0.0028657 0.0035153 0.0030986 0.0030265 0.0002279
Source MOGOA, MOPSO, NSGA-II: Mirjalili et al. (2018); MODA: Mirjalili (2016); MOALO: Mirjalili et al. (2017)
Result in boldface indicates better values
119
 120

Table 4.15 Comparison of the results achieved by various algorithms in multi-objective optimization of the ZDT-3O problem
Algorithm IGD Spacing
Best Worst M Median SD Best Worst M Median SD
MOGOA 0.0068 0.0289 0.0114 0.0081 0.0079 − − − − −
MOPSO 0.0189 0.0225 0.02032 0.0203 0.001278 − − − − −
NSGA-II 0.0371 0.0847 0.0626 0.0584 0.017888 − − − − −
MODA 0.0048 0.0191 0.00916 0.0063 0.005372 − − − − −
MOALO 0.0191 0.0315 0.02629 0.0288 0.004451 − − − − −
Jaya 0.0020173 0.0022966 0.0021641 0.0021641 0.0000889 0.0456603 0.0595403 0.0528256 0.0519847 0.0044812
Rao-1 0.0020965 0.0033595 0.0024693 0.0022880 0.0004120 0.0406990 0.0658425 0.0540621 0.0546690 0.0083277
Rao-2 0.0020227 0.0027209 0.0023362 0.0023137 0.0002245 0.0475266 0.0609942 0.0535145 0.0535145 0.0048116
Rao-3 0.0022324 0.0033942 0.0026145 0.0024506 0.0003970 0.0462863 0.0692203 0.0574161 0.0561472 0.0075947
ERao-1 0.0020711 0.0024332 0.0022214 0.0022214 0.0001195 0.0444766 0.0558539 0.0512404 0.0512518 0.0036667
ERao-2 0.0020921 0.0022527 0.0021572 0.0021572 0.0000617 0.0482222 0.0743055 0.0570622 0.0538116 0.0093994
ERao-3 0.0019836 0.0024472 0.0021456 0.0020888 0.0001500 0.0438590 0.0629687 0.0527477 0.0530921 0.0060305
AMTPG-Jaya 0.0020546 0.0022480 0.0021270 0.0021115 0.0000646 0.0452289 0.0587363 0.0539866 0.0549457 0.0044246
SAP-Rao 0.0018341 0.0021865 0.0020453 0.0020763 0.0001176 0.0446194 0.0604126 0.0521410 0.0525578 0.0053660
Source MOGOA, MOPSO, NSGA-II: Mirjalili et al. (2018); MODA: Mirjalili (2016); MOALO: Mirjalili et al. (2017)
Result in boldface indicates better values
4 Working of Jaya and Rao Optimization Algorithms …
4.5 Computational Results Analysis on Multi-objective … 121

Fig. 4.17 True Pareto-front (PF) and the obtained Pareto-fronts in the objective space of the ZDT-1
problem

Furthermore, to judge the performance of the proposed modified algorithms, the

Friedman statistical test is conducted based on the results achieved for all the prob-
lems. For ranking the algorithms, the mean values are given priority, and then, the
priorities are given to the best, median, worst, and standard deviation values, respec-
tively. Here, 14 algorithms are compared in this test with 13° of freedom for the χ 2
distribution. The average ranks, χ 2 value, and p-value achieved in the Friedman test
are presented in Table 4.16. The ranks of these algorithms and their average rank
values are plotted in Fig. 4.22.
The p-value of the Friedman test is 5.12859E−7, which indicates that the results
are highly significant. The ERao-3 algorithm has achieved the least average rank of
2.4. The ranking of these algorithms in solving the ZDT test functions is ERao-3,
SAP-Rao, ERao-2, ERao-1, AMTPG-Jaya, Rao-3, Rao-1, Rao-2, Jaya, MOPSO,
MODA, MOGOA, MOALO, and NSGA-II. Here, an observation can be made that
the proposed algorithms have achieved better average ranks compared to those of the
122 4 Working of Jaya and Rao Optimization Algorithms …

Fig. 4.18 True Pareto-front (PF) and the obtained Pareto-fronts in the objective space of the ZDT-2
problem

MOPSO, MODA, MOGOA, MOALO, and NSGA-II algorithms. From the compu-
tational results, it can be observed that the performances of the proposed modified
algorithms are superior or competitive to that of the other algorithms compared.
Furthermore, mean IGD values are relatively closer to the best values. It indi-
cates that the proposed algorithms are consistent in finding the optimal Pareto-front.
Also, the proposed algorithms Pareto-fronts are consistent and along with the true
Pareto-fronts. The performances of the proposed modified algorithms are superior
or competitive to those of the other algorithms compared.
From the computational results in single-objective and multi-objective optimiza-
tion benchmark test problems, it can be observed that the performance of the
AMTPG-Jaya algorithm is superior to other proposed modified algorithms when
the function evaluations are higher. The performances of the SAP-Rao and elitist
Rao algorithms are superior to that of the AMTPG-Jaya algorithm for the prob-
lems where fewer function evaluations are considered as a termination criterion.
4.5 Computational Results Analysis on Multi-objective … 123

Fig. 4.19 True Pareto-front (PF) and the obtained Pareto-fronts in the objective space of the ZDT-3
problem

However, in both scenarios, the performances of the proposed algorithms are supe-
rior or competitive to those of the other algorithms compared. Hence, the modified
versions of the Jaya and Rao algorithms can be considered as the improved versions
of the Jaya and Rao algorithms, respectively.
The next chapter presents the multi-attribute decision-making methods and their
implementation in multi-objective optimization scenarios.
124 4 Working of Jaya and Rao Optimization Algorithms …

Fig. 4.20 True Pareto-front (PF) and the obtained Pareto-fronts in the objective space of the ZDT-1L
problem
4.5 Computational Results Analysis on Multi-objective … 125

Fig. 4.21 True Pareto-front (PF) and the obtained Pareto-fronts in the objective space of the ZDT2-
3O problem
 126

Table 4.16 Friedman statistical test results for the results presented in Tables 4.11, 4.12, 4.13, 4.14 and 4.15 corresponding to the unconstrained multi-objective
benchmark problems
Algorithm MOGOA MOPSO NSGA-II MODA MOALO Jaya Rao-1 Rao-2 Rao-3 ERao-1 ERao-2 ERao-3 AMTPG-Jaya SAP-Rao
Avg. Rank 11.6 9.6 14 10.4 12.6 8.2 7.4 7.6 7.2 4.2 3 2.4 4.2 2.6
χ2 54.44
p-value The p-value is 5.12859E−7. The result is significant at p < 0.01
4 Working of Jaya and Rao Optimization Algorithms …
4.5 Computational Results Analysis on Multi-objective … 127

Fig. 4.22 Average ranks of various algorithms in the Friedman statistical test for the ZDT test
functions results

References

Deb, K., Pratap, A., Agarwal, S., & Meyarivan, T. (2002). A fast and elitist multiobjective genetic
algorithm: NSGA-II. IEEE Transactions on Evolutionary Computation, 6(2), 182–197. https://
doi.org/10.1109/4235.996017
Eberhart, R., & Kennedy, J. (1995). A new optimizer using particle swarm theory. MHS’95. Proceed-
ings of the Sixth International Symposium on Micro Machine and Human Science, 39–43. https://
doi.org/10.1109/MHS.1995.494215
Haupt, R. L., & Haupt, S. E. (2003). Practical Genetic Algorithms. 2004 John Wiley & Sons, Inc.
https://doi.org/10.1002/0471671746
Karaboga, D., & Akay, B. (2009). A comparative study of artificial bee colony algorithm. Applied
Mathematics and Computation, 214(1), 108–132. https://doi.org/10.1016/j.amc.2009.03.090
Liang, J. J., Qin, A. K., Suganthan, P. N., & Baskar, S. (2006). Comprehensive learning particle
swarm optimizer for global optimization of multimodal functions. IEEE Transactions on
Evolutionary Computation, 10(3), 281–295. https://doi.org/10.1109/TEVC.2005.857610
Mendes, R., Kennedy, J., & Neves, J. (2004). The fully informed particle swarm: simpler, maybe
better. IEEE Transactions on Evolutionary Computation, 8(3), 204–210. https://doi.org/10.1109/
TEVC.2004.826074
Mirjalili, S. (2016). Dragonfly algorithm: a new meta-heuristic optimization technique for solving
single-objective, discrete, and multi-objective problems. Neural Computing and Applications,
27(4), 1053–1073. https://doi.org/10.1007/s00521-015-1920-1
Mirjalili, S., Jangir, P., & Saremi, S. (2017). Multi-objective ant lion optimizer: a multi-objective
optimization algorithm for solving engineering problems. Applied Intelligence, 46(1), 79–95.
https://doi.org/10.1007/s10489-016-0825-8
Mirjalili, S. Z., Mirjalili, S., Saremi, S., Faris, H., & Aljarah, I. (2018). Grasshopper optimiza-
tion algorithm for multi-objective optimization problems. Applied Intelligence, 48(4), 805–820.
https://doi.org/10.1007/s10489-017-1019-8
Montes de Oca, M. A., Stutzle, T., Birattari, M., & Dorigo, M. (2009). Frankenstein’s PSO:
a composite particle swarm optimization algorithm. IEEE Transactions on Evolutionary
Computation, 13(5), 1120–1132. https://doi.org/10.1109/TEVC.2009.2021465
Ngo, T. T., Sadollah, A., & Kim, J. H. (2016). A cooperative particle swarm optimizer with
stochastic movements for computationally expensive numerical optimization problems. Journal
of Computational Science, 13, 68–82. https://doi.org/10.1016/j.jocs.2016.01.004
Nickabadi, A., Ebadzadeh, M. M., & Safabakhsh, R. (2011). A novel particle swarm optimization
algorithm with adaptive inertia weight. Applied Soft Computing, 11(4), 3658–3670. https://doi.
org/10.1016/j.asoc.2011.01.037
128 4 Working of Jaya and Rao Optimization Algorithms …

Ocłoń, P., Cisek, P., Rerak, M., Taler, D., Rao, R. V., Vallati, A., & Pilarczyk, M. (2018). Thermal
performance optimization of the underground power cable system by using a modified Jaya
algorithm. International Journal of Thermal Sciences, 123, 162–180. https://doi.org/10.1016/j.
ijthermalsci.2017.09.015
Rao, R. V. (2016). Jaya: A simple and new optimization algorithm for solving constrained and uncon-
strained optimization problems. International Journal of Industrial Engineering Computations,
7(1), 19–34. https://doi.org/10.5267/j.ijiec.2015.8.004
Rao, R. V. (2019). Jaya: An Advanced Optimization Algorithm And Its Engineering Applications.
Springer International Publishing AG, part of Springer Nature. (2019). Springer. Cham. https://
doi.org/10.1007/978-3-319-78922-4
Rao, R. V. (2020). Rao algorithms: Three metaphor-less simple algorithms for solving optimization
problems. International Journal of Industrial Engineering Computations, 11, 107–130. https://
doi.org/10.5267/j.ijiec.2019.6.002
Rao, R. V., & Keesari, H. S. (2018). Multi-team perturbation guiding Jaya algorithm for optimization
of wind farm layout. Applied Soft Computing Journal, 71, 800–815. https://doi.org/10.1016/j.
asoc.2018.07.036
Rao, R. V., & Keesari, H. S. (2019). Solar assisted heat engine systems: multi-objective optimisation
and decision making. International Journal of Ambient Energy, 0(0), 1–27.https://doi.org/10.
1080/01430750.2019.1636870
Rao, R. V., & Keesari, H. S. (2020). Rao algorithms for multi-objective optimization of selected
thermodynamic cycles. Engineering with Computers. https://doi.org/10.1007/s00366-020-010
08-9
Rao, R. V., & Keesari, H. S. (2021). A self-adaptive population Rao algorithm for optimization of
selected bio-energy systems. Journal of Computational Design and Engineering, 8(1), 69–96.
https://doi.org/10.1093/jcde/qwaa063
Rao, R. V., Keesari, H. S., Oclon, P., & Taler, J. (2019). Improved multi-objective Jaya optimization
algorithm for a solar dish Stirling engine. Journal of Renewable and Sustainable Energy, 11(2),
25903. https://doi.org/10.1063/1.5083142
Rao, R. V., Keesari, H. S., Oclon, P., & Taler, J. (2020). An adaptive multi-team perturbation-
guiding Jaya algorithm for optimization and its applications. Engineering with Computers, 36(1),
391–419. https://doi.org/10.1007/s00366-019-00706-3
Rao, R. V., Keesari, H. S., Taler, D., Taler, J., & Ocłoń, P. (2020b). Multi-objective optimization of
a solar-assisted Stirling heat engine system using elitist Rao algorithms. In B. Samojeden (Ed.),
Energy Fuels Environment 2020, Kraków (Poland), 1–4 December 2020 : book of abstracts
(p. 77). Kraków: Faculty of Energy and Fuels, AGH University of Science and Technology.
Rao, R. V., & Patel, V. (2013). Comparative performance of an elitist teaching-learning-based
optimization algorithm for solving unconstrained optimization problems. International Journal
of Industrial Engineering Computations, 4(1), 29–50. https://doi.org/10.5267/j.ijiec.2012.09.001
Rao, R. V., & Rai, D. P. (2017). Optimisation of welding processes using quasi-oppositional-based
Jaya algorithm. Journal of Experimental & Theoretical Artificial Intelligence, 29(5), 1099–1117.
https://doi.org/10.1080/0952813x.2017.1309692
Rao, R. V., Rai, D. P., & Balic, J. (2017). A multi-objective algorithm for optimization of modern
machining processes. Engineering Applications of Artificial Intelligence, 61(August 2015), 103–
125. https://doi.org/10.1016/j.engappai.2017.03.001
Rao, R. V., & Saroj, A. (2017). A self-adaptive multi-population based Jaya algorithm for engi-
neering optimization. Swarm and Evolutionary Computation, 37, 1–26. https://doi.org/10.1016/
j.swevo.2017.04.008
Rashedi, E., Nezamabadi-pour, H., & Saryazdi, S. (2009). GSA: A gravitational search algorithm.
Information Sciences, 179(13), 2232–2248. https://doi.org/10.1016/j.ins.2009.03.004
Schott, J. R. (1995). Fault Tolerant Design Using Single and Multicriteria Genetic Algorithm
Optimization. Thesis, Massachusetts Institute of Technology, Boston, MA.
References 129

vanden Bergh, F., & Engelbrecht, A. P. (2004). A cooperative approach to particle swarm optimiza-
tion. IEEE Transactions on Evolutionary Computation, 8(3), 225–239. https://doi.org/10.1109/
TEVC.2004.826069
Yu, K., Liang, J. J., Qu, B. Y., Chen, X., & Wang, H. (2017). Parameters identification of photovoltaic
models using an improved JAYA optimization algorithm. Energy Conversion and Management,
150, 742–753. https://doi.org/10.1016/j.enconman.2017.08.063
Zhou, A., Zhang, Q., & Jin, Y. (2009). Approximating the set of pareto-optimal solutions in both
the decision and objective spaces by an estimation of distribution algorithm. IEEE Transactions
on Evolutionary Computation, 13(5), 1167–1189. https://doi.org/10.1109/TEVC.2009.2021467
Zitzler, E., Deb, K., & Thiele, L. (2000). Comparison of multiobjective evolutionary algorithms:
Empirical Results. Evolutionary Computation, 8(2), 173–195. https://doi.org/10.1162/106365
600568202
Chapter 5
Multi-attribute Decision-Making
Methods and Their Implementation
in Energy Systems

Abstract This chapter presents multi-attribute decision-making methods and their

implementation in energy systems’ optimization for identifying the best compromise
solution from the Pareto-fronts.

Decision-making in the presence of multiple conflicting criteria can be considered as

multiple criteria decision-making (MCDM). Based on the problem type, the MCDM
problems can be divided into multiple attribute decision-making (MADM) problems
and multiple objective decision-making (MODM) problems. The MODM problems
involve finding the best-suited solution from an infinitive or a large number of choices,
which satisfies problem constraints and priorities. The MADM problems involve
selecting the best choice from a limited number of alternatives. A MADM method
specifies how the attribute information is to be processed in order to arrive at a choice.
MADM methods require both inter- and intra-attribute comparisons and involve
appropriate explicit trade-offs. This chapter presents the multi-attribute decision-
making methods and their implementation in multi-objective optimization.
Identifying the best Pareto-optimal solution from a Pareto-front is treated as a
multiple attribute decision-making problem (Rao & Keesari, 2019) in this book.
MADM methods are a class of MCDM methods which are employed when a limited
number of alternatives are available. These methods suggest the best alternative
using the attribute information of various alternatives. The set of Pareto-optimal
solutions are regarded as alternatives (Ai ∀ alternatives i = 1, 2…N), and the objective
functions are regarded as attributes (Bj ∀ attributes j = 1, 2…M). The objective value
of a non-dominated solution i corresponding to an objective or attribute j can be
considered as the performance measure (mij ). The decision-maker can appraise the
relative importance or weights (wj ) of the objectives such that the summation of
weights of all the attributes together is equal to 1. The objective functions considered
in this book have different units. Thus, all the objective values must be normalized
(nmij ) to the same units before applying the MADM methods. Based on the potential
to find the best alternative, the following methods have been considered in this book
(Rao, 2007, 2013).

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 131
V. R. Ravipudi et al., Design Optimization of Renewable Energy Systems Using Advanced
Optimization Algorithms, Green Energy and Technology,
https://doi.org/10.1007/978-3-030-95589-2_5
132 5 Multi-attribute Decision-Making Methods …

5.1 Simple Additive Weighing (SAW)

Fishburn (1967) developed the SAW method. For the given weights of the attributes,
the cumulative performance (Pi ) of alternative i is given by the following equation:


M
 
Pi = W j nmij (5.1)
j=1

where nmij is the normalized measure of performance, and the best alternative is the
one with the highest Pi value.

5.2 Weighted Product Method (WPM)

Miller and Starr (1969) developed the WPM method. For the given weights of the
attributes, the cumulative performance (Pi ) of alternative i is given by the following
equation:


M
 w
Pi = nmij j (5.2)
j=1

where nmij is the normalized measure of performance, and the best alternative is the
one with the highest Pi value.

5.3 Preference Ranking Organization Method

for Enrichment Evaluation (PROMETHEE)

Brans et al. (1986) developed the PROMETHEE method, which belongs to the class
of outranking methods. In this method, for each objective function, the Pareto-optimal
solutions are compared pair-wise to determine the strength of a solution ai over the
solution ak . For the given weights of the objectives, in pair-wise comparison of the
solutions usual preference function (Pj, aiak ) is considered in this book. The usual
preference function is the difference between the values of solutions ai and ak for an
objective bj . Now, the objective preference (Π aiak ) index is given by:


M
ajak = w j P j, aiak (5.3)
j=1
5.3 Preference Ranking Organization Method … 133

The value of ajak varies between zero and 1, and it denotes the intensity of the
inclination toward the solution ai over ak , when concurrently all the objectives are
compared. Now, for a solution ai , the outranking relations are calculated using the
following equations:
  
Leaving flow Øi+ = ax ai (5.4)
x∈A
  
Entering flow Øi− = ai ax (5.5)
x∈A
     
Net flow Ø(ai ) = Øi+ − Øi− (5.6)

The leaving flow denotes the supremacy of solution ai over all other solutions.
The entering flow denotes the degree to which all other solutions are dominating the
solution ai . The net flow denotes the outranking relationship of the solutions, and
higher values of the net flow mean the best Pareto-optimal solution. For more details
about the PROMETHEE method, the readers may refer to Rao (2013).

5.4 Technique for Order Preference by Similarity to Ideal

Solution (TOPSIS)

The TOPSIS method developed by Hwang and Yoon (1981) provides the relative
similarity index (SI) near to an ideal and non-ideal solution. The ideal and non-
ideal solutions represent the hypothetical best and worst scenarios of all the objec-
tives, respectively. The normalized values (nmij ) of the objective function values are
calculated by using the following equation:
⎛ ⎞

 M
⎝
nmij = m ij / m 2ij ⎠ (5.7)
j=1

Let Rij be the matrix of normalized values and w j be the weights of the objective
functions, and then the weighted normalized matrix can be calculated by using the
following equation:

Vij = w j Rij (5.8)

Now, identify the ideal (O ideal

j ) and non-ideal (O non
j
- ideal
) solutions with respect to
all objective functions using the following criteria, respectively. For a maximization
objective function (j), the solution with maximum Vij value will be considered as
an ideal solution and the solution with minimum Vij value will be considered as the
134 5 Multi-attribute Decision-Making Methods …

non-ideal solution. Similarly, for minimization objective function (j), the solution
with minimum Vij value will be considered as an ideal solution and the solution with
maximum Vij value will be considered as the non-ideal solution. Then, calculate the
Pareto-optimal solution distances (di ) from the ideal and non-ideal solution using
the following equations respectively:

 M 2

diideal = Vij − O ideal
j (5.9)
j=1

 M 2

dinon - ideal = Vij − O non
j
- ideal
(5.10)
j=1

where i = 1, 2… N (i.e., number of Pareto-optimal candidate solutions or alternatives

in a Pareto frontier), j = 1, 2… M (i.e., the number of objectives or attributes of the
Pareto frontier). Now, the similarity index can be calculated using the following
equation:

dinon - ideal
SI = (5.11)
diideal + dinon - ideal

If the similarity index of a Pareto-optimal solution is higher than that of all other
solutions in the Pareto frontier, then it is the nearest solution to the ideal solution.
Similarly, if the similarity index of a Pareto-optimal solution is lower than that of
other solutions in the Pareto frontier, then it is the nearest solution to a non-ideal
solution.

5.5 Modified TOPSIS (MTOPSIS)

Deng et al. (2000) proposed a modification to the TOPSIS method by suggesting

weighted Euclidean distances. In this modified version, the ideal and non-ideal solu-
tions are selected from the normalized matrix (Rij ) and the weighted Euclidean
distances are calculated from the ideal and non-ideal solutions. According to the
modified TOPSIS (MTOPSIS) method, the Pareto-optimal solution distances (di )
from the ideal and non-ideal solution are calculated by the following equations:

 M 2

diideal = w j nmij − nmideal
j (5.12)
j=1
5.5 Modified TOPSIS (MTOPSIS) 135

 M 2

dinon - ideal = w j nmij − nmnon
j
- ideal
(5.13)
j=1

Now, the similarity index is calculated using Eq. (5.11). The solution with the
highest similarity index will be considered as the best solution, and the solution with
the lowest similarity index will be considered as the relatively worst solution.

5.6 Compromise Ranking Method (VIKOR)

The idea of compromise ranking was introduced by Yu (1973) and Zeleny (1982) and
later advanced by Opricovic and Tzeng (2002, 2003). This method is also called as
VIšekriterijumsko KOmpromisno Rangiranje (VIKOR). The procedure for VIKOR
is described below:
Step 1: Identify the ideal (m ideal
ij ) and non-ideal (m ij
non - ideal
) solutions for all the
attributes based on the objective function values of the solutions for respective
attributes using the following criteria. For a maximization objective function (j),
the solution with maximum m ij value will be considered as the ideal solution and
the solution with minimum m ij value will be considered as the non-ideal solution.
Similarly, for minimization objective function (j), the solution with minimum m ij
value will be considered as the ideal solution and the solution with maximum m ij
value will be considered as the non-ideal solution.
Step 2: Calculate E i and Fi values for all Pareto-optimal solutions in the Pareto-
front.


M  
Ei = w j m ideal
ij − m ij /m ideal
ij − m non
ij
- ideal
(5.14)
j=1

For Pareto-optimal solution i,

   
Fi = maximum of w j m ideal
ij − m ij /m ideal
ij − m non - ideal
ij | j = 1, 2 . . . M
(5.15)

Step 3: Calculate the performance index (Pi ) using the following equation:
       
Pi = V E i − E i− / E i+ − E i− + (1 − V ) Fi − Fi− / Fi+ − Fi− (5.16)

where E i+ and E i− are the maximum and minimum values of E i , respectively.

Similarly, Fi+ and Fi− are the maximum and minimum values of Fi , respectively.
V is taken as 0.5.
136 5 Multi-attribute Decision-Making Methods …

Step 4: The solution with the least Pi will be considered as the best alternative if
it satisfies the following conditions (Tzeng et al., 2005).
1. Condition-1 (Acceptable advantage):Pi min − Pi next min ≥ (1/(N − 1)).
2. Condition-2 (Acceptable stability in decision-making): The solution with least
Pi must also be the best solution according to the E i and/or Fi values. The
solution with least E i values is considered best according to E i values, and the
solution with least Fi value is considered as best according to Fi values.
3. Condition-3: If these conditions failed, then a set of compromise solutions is
proposed based on the following criteria.
– The solutions Pi min ,Pi next min … Pik are compromise solutions if condition-1
failed, where Pik − Pi next min ≈ (1/(N − 1)).
– The solutions Pi min and Pi next min are compromise solutions if only condition-
2 fails.

5.7 Complex Proportional Assessment (COPRAS)

The complex proportional assessment method was introduced by Edmundas

Kazimieras Zavadskas in 1994 (Hajiagha et al., 2013). This method is simple to
use and stepwise procedure for ranking and evaluating the alternatives. The stepwise
procedure of COPRAS method is as follows:
The normalized values (nmij ) of the objective function values are calculated by
using the following equation:
⎛ ⎞
N
nmij = m ij /⎝ m ij ⎠ (5.17)
j=1

Let Rij be the matrix of normalized values and w j be the weights of the objective
functions. Then, the weighted normalized matrix can be calculated by using the
following equation:

Vij = w j Rij (5.18)

Now, calculate the summation Pi , only for the maximization objectives using the
following equation:


k
Pi = Vij (5.19)
j=1

where k is the number of maximization objectives. Now, calculate summation Ri ,

only for minimization objectives using the following equation:
5.7 Complex Proportional Assessment (COPRAS) 137


M
Ri = Vij (5.20)
j=k+1

Now, calculate the relative weight of each solution (Qi ) using the following
equation:
 M   
 
M
Q i = Pi + Ri / Ri (1/Ri ) (5.21)
i i

Now, calculate the degree of utility (U i ) for each solution using the following
equation:

Ui = (Q i /Q i max ) × 100 % (5.22)

The solution with a high degree of utility will be considered as the best solution.

5.8 Gray Relational Analysis (GRA)

Gray relational analysis is one of the derived evaluation methods based on the concept
of gray relational space (GRS). The normalized values (nmij ) of the objective function
values are calculated by using the following equations:
 
mij − min mij
nmij =     For maximization (5.23)
max mij − min mij
 
max mij − m ij
nmij =     For minimization (5.24)
max m ij − min m ij

Now, calculate the gray relational coefficient (GR) using the following equation:

  min − ξ max
GR nm0 j , nmij = (5.25)
i j − ξ max

where  
ij = nm0 j − nmij 
nm0 j ∈ reference
 sequence{nm01 , nm02 , . . . , nm0M }
min = min ij , i = 1, 2, . . . , N ; j = 1, 2, . . . , M 
max = max ij , i = 1, 2, . . . , N ; j = 1, 2, . . . , M
Distinguishing coefficient (ξ) = 0.5
Now, the gray relational grade can be calculated by the following equation:
138 5 Multi-attribute Decision-Making Methods …


M
 
(nm0 , nmi ) = w j GR nm0 j , nmij for i = 1, 2, . . . , N (5.26)
j=1

If a solution has the highest gray relational grade with the reference sequence,
it means that the comparability sequence is most similar to the reference sequence,
and that solution would be the best choice.
In the implementation of the MADM methods, the case studies’ objective func-
tions are considered as attributes and Pareto-optimal solutions as alternatives. In
all the multi-objective optimization case studies, equal weights are taken for all the
objectives. The MADM methods use different characteristics to rank the alternatives,
which indicate that these methods rank the solutions differently. Therefore, to iden-
tify the final best solution, average ranks have been calculated. The average rank for
a solution is nothing but the average value of the ranks given by different MADM
methods.
Furthermore, to assess the correlation and strength of the ranking given by the
MADM methods, Spearman’s correlation coefficient is calculated between different
pairs of methods. The Spearman correlation coefficient is 1 when the ranking given
by two MADM methods is identical and -1 when the ranking given by two MADM
methods is the opposite. If the Spearman correlation coefficient between two MADM
methods is near 1, the ranking of those two methods is similar to each other. However,
in this book, the ranking given by different MADM methods is considered for
calculating average ranks based on the following conditions:
1. If a MADM method has a positive Spearman correlation coefficient greater
than or equal to 0.5 with all other methods, then it is considered for calculating
average ranks.
2. If a pair of MADM methods have the Spearman correlation coefficient value
lesser than 0.5 (including negative values), then the MADM method, which has a
better Spearman correlation coefficient value with more methods, is considered
for the average ranking. For example, the SAW-WPM pair has a Spearman
correlation less than 0.5. The SAW method has better correlation values (greater
than or equal to 0.5) with TOPSIS, MTOPSIS, VIKOR, PROMETHEE, and
GRA methods. The WPM method has better correlation values (greater than
or equal to 0.5) with TOPSIS, MTOPSIS, VIKOR, and COPRAS methods.
Then, the ranks given by the SAW method will be considered for calculating
the average rank, and the ranks given by the WPM method will not be considered
for calculating the average rank because the SAW method has better correlation
values with five methods, whereas the WPM method has a better correlation
value with four methods. Furthermore, if both methods have better correlation
values (greater than or equal to 0.5) with the same number of methods, then the
correlation values are compared, and the method with better correlation values
will be considered.
The best Pareto-optimal solution identified based on the average rank is then
similarly compared with the solutions identified for the other algorithms. The next
5.8 Gray Relational Analysis (GRA) 139

chapter presents the applications of the Jaya and Rao algorithms, along with their
improved versions, to wind farm layout optimization problems.

References

Brans, J. P., Vincke, P., & Mareschal, B. (1986). How to select and how to rank projects: The
Promethee method. European Journal of Operational Research, 24, 228–238. https://doi.org/10.
1016/0377-2217(86)90044-5
Deng, H., Yeh, C. H., & Willis, R. J. (2000). Inter-company comparison using modified TOPSIS
with objective weights. Computers and Operations Research, 27, 963–973. https://doi.org/10.
1016/S0305-0548(99)00069-6
Fishburn, P. C. (1967). Additive utilities with incomplete product sets: Application to priorities and
assignments. Operations Research, 15(3), 537–542. https://doi.org/10.1287/opre.15.3.537
Hajiagha, S. H. R., Hashemi, S. S., & Zavadskas, E. K. (2013). A complex proportional assess-
ment method for group decision making in an interval-valued intuitionistic fuzzy environment.
Technological and Economic Development of Economy, 19(1), 22–37. https://doi.org/10.3846/
20294913.2012.762953
Hwang, C.-L., & Yoon, K. (1981). Multiple Attribute Decision Making: Methods and Applications
A State-of-the-Art Survey. https://doi.org/10.1007/978-3-642-48318-9
Miller, D. W., & Starr, M. K. (1969). Executive decisions with operations research. Prentice Hall,
Englewood Cliffs.
Opricovic, S., & Tzeng, G. H. (2002). Multicriteria planning of post-earthquake sustainable recon-
struction. Computer-Aided Civil and Infrastructure Engineering, 17, 211–220. https://doi.org/
10.1111/1467-8667.00269
Opricovic, S., & Tzeng, G.-H. (2003). Fuzzy multicriteria model for postearthquake land-use
planning. Natural Hazards Review, 4, 59–64. https://doi.org/10.1061/(asce)1527-6988(2003)4:
2(59)
Rao, R. V. (2007). Decision making in the manufacturing environment using graph theory and
fuzzy multiple attribute decision making methods. Springer Series in Advanced Manufacturing,
Springer-Verlag, London, UK. https://doi.org/10.1007/978-1-84628-819-7
Rao, R. V. (2013). Decision making in the manufacturing environment using graph theory and
fuzzy multiple attribute decision making methods (Volume 2). Springer Series in Advanced
Manufacturing, Springer-Verlag, London, UK. https://doi.org/10.1007/978-1-4471-4375-8_1
Rao, R. V., & Keesari, H. S. (2019). Solar assisted heat engine systems: multi-objective optimisation
and decision making. International Journal of Ambient Energy, 0(0), 1–27. https://doi.org/10.
1080/01430750.2019.1636870.
Tzeng, G. H., Lin, C. W., & Opricovic, S. (2005). Multi-criteria analysis of alternative-fuel buses for
public transportation. Energy Policy, 33, 1373–1383. https://doi.org/10.1016/j.enpol.2003.12.014
Yu, P. L. (1973). A class of solutions for group decision problems. Management Science, 19,
936–946. https://doi.org/10.1287/mnsc.19.8.936
Zeleny, M. (1982). Multiple criteria decision making. McGraw Hill.
Chapter 6
Optimization of Wind Farm Layouts

Abstract This chapter presents the applications of different versions of Jaya and
Rao algorithms to wind farm layout optimization. Three scenarios of wind farm
layout optimization (WFLO) problem are considered, which are fixed wind speed
and direction (case-I); fixed wind speed with changing wind direction (case-II);
changing wind speed with changing direction (case-III). The computational results
of the proposed algorithms are compared with those of the GA and its variant, TLBO
and its variants, ABC algorithm, and other algorithms from the literature.

6.1 Problem Definition and Wind Scenarios of the Wind

Farm Layout Optimization

The optimal placing of wind turbines results in better mean incident velocity for a
wind farm; thus, more power is produced with the same investment. In the optimiza-
tion of the WFLO problem, each wind turbine location is taken as an independent
variable for a specified farmland size and number of turbines. The characteristics of
the wind turbines considered in this book are shown in Table 6.1.
The present work considers a square region of 2 × 2 km2 in which a wind turbine
can be placed at any position in the specified area, maintaining the minimum distance
(5d = 200 m) between two adjacent turbines. The surface roughness of the region is
assumed as 0.3 m. The C T is considered constant throughout the processes (Grady
et al., 2005). In optimizing the WFLO problem, three different cases of WFLO
are considered, which are: fixed wind speed and fixed direction (case-I), fixed wind
speed and variable direction (case-II), and variable wind speed and variable direction
(case-III). In each case, optimal layouts are obtained by minimizing the cost per unit
power. Along with the minimum value of the objective function, total power (KW)
and farm efficiency are also calculated in computational experiments.
Computational experiments are performed by the Jaya algorithm and its modified
versions (MTPG-Jaya and AMTPG-Jaya) and Rao algorithms and their modified
versions (elitist Rao and SAP-Rao algorithms). In the implementation of the Jaya
algorithm, the wind farm area considered is divided into 100 possible turbine loca-
tions as a square grid. A wind turbine can be placed at the center of every cell in

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 141
V. R. Ravipudi et al., Design Optimization of Renewable Energy Systems Using Advanced
Optimization Algorithms, Green Energy and Technology,
https://doi.org/10.1007/978-3-030-95589-2_6
142 6 Optimization of Wind Farm Layouts

Table 6.1 Wind turbine

Property Value
properties
Hub height (Z) 60 m
Rotor diameter (d) 40 m
Thrust coefficient (C T ) 0.88
Ground roughness (Z 0 ) 0.3 m
Wind velocity (u0 ) 12 m/s
Axial induction factor (a) 0.33
Entrainment constant (k) 0.094

the square grid. In the implementation of the remaining algorithms, the turbines
are placed in the wind farm, restricting the minimum distance between the adjacent
turbines to 200 m.
The population size is varied between 10 and 100 for different algorithms in
the computational experiments. For the MTPG-Jaya and AMTPG-Jaya algorithms
after performing few trials with different configurations, the probability of accepting
the worst solution and maximum movement iterations is chosen as 0.35 and 10,
respectively. Four teams are taken for the MTPG-Jaya algorithm. The elite size for
the elitist Rao algorithms is varied from 10 to 40%. The number of teams and the
population size are selected suitable for each case. Each candidate solution gives
one layout for the WFLO problem. The computational results are compared with
the well-established algorithms such as GA, lazy greedy algorithm (LGA), ABC,
TLBO, TLBO enhanced with ABC (AL), and TLBO enhanced with ABC and PSO
(PAL). The GA results are taken from Grady et al. (2005), and the results of LGA are
taken from Changshui et al. (2011). The TLBO results in case-I are taken from Patel
et al. (2015). In case-II and case-III, results for the TLBO, ABC, AL, and PAL are
taken from Patel et al. (2017). In the next sub-section, the analysis of computational
results related to WFLO case-I is presented.

6.2 Case-I: Fixed Wind Speed and Fixed Direction

It is the case of unidirectional wind at a steady speed. The incident wind speed is taken
as 12 m/s in the normal direction. The wind speed inside the wake effect of a turbine
is reduced. Thus, the incident wind speed for a turbine, which is placed behind an
upstream turbine, will be modified. The incident wind speed of a turbine will not be
affected by the wake effect of its adjacent wind turbine. In this book, optimization is
carried out with the 30 turbines configuration. The comparative performance results
are presented in the form of the total power (kW), farm efficiency, and objective
value (cost per unit power generated). The TLBO algorithm results are presented for
9000 function evaluations, which are least when compared to GA (180,000). Thus,
Jaya and Rao algorithms, including their modified versions, are tested with 9000
6.2 Case-I: Fixed Wind Speed and Fixed Direction 143

Table 6.2 Results obtained by the various algorithms in case-I with 30 turbines
Algorithm Total power (kW/year) Farm efficiency Objective value
GA 14,310 92.015 0.0015436
LGA 14,310 92.01 0.0015436
TLBO 14,310 92.01 0.0015436
Jaya 14,310 92.01 0.0015436
MTPG-Jaya 14,601.0174 93.88514275 0.001512825
AMTPG-Jaya 14,606.10621 93.91786401 0.001512298
Rao-1 14,503.42196 93.25760006 0.001523005
Rao-2 14,470.30126 93.04463259 0.001526491
Rao-3 14,514.64737 93.32977991 0.001521828
ERao-1 14,528.11154 93.41635505 0.001520417
ERao-2 14,544.37374 93.52092171 0.001518717
ERao-3 14,549.54668 93.55418388 0.001518177
SAP-Rao 14,689.3961 94.45342116 0.001503724
Source GA-Grady et al. (2005), LGA-Changshui et al. (2011), TLBO- Patel et al. (2015)
Result in boldface indicates a better performing algorithm

function evaluations. Computational results of various algorithms for the case-I with
30 turbines are summarized in Table 6.2.
Here, an observation can be made that the Jaya algorithm and its modified
versions and Rao algorithms and their modified versions have obtained better layouts
compared to the layouts achieved by the GA, LGA, and TLBO algorithms. The graph-
ical representation of the total power output of the solutions achieved by various
algorithms in this case study is presented in Fig. 6.1. The SAP-Rao algorithm has
achieved the most efficient layout, producing 14,689.39 kW power with 94.45%

Fig. 6.1 Total power produced by the layouts achieved by different algorithms in case-I with 30
turbines
144 6 Optimization of Wind Farm Layouts

efficiency. Furthermore, the layout achieved by the SAP-Rao algorithm has resulted
in a 2.7% increment in total power when compared to that of the solutions of the
GA, LGA, TLBO, and Jaya algorithms. Similarly, the total power of the SAP-Rao
algorithm is 1.3%, 1.5%, 1.2%, 1.1%, 1%, and 1% higher when compared to that
of the Rao-1, Rao-2, Rao-3, ERao-1, ERao-2, and ERao-3 algorithms, respectively.
The solutions of the AMTPG-Jaya, MTPG-Jaya, ERao-3, ERao-2, ERao-1, Rao-3,
Rao-1, and Rao-2 algorithms are the next best solutions. The GA, LGA, TLBO, and
Jaya algorithms have achieved identical layouts.
Also, to draw attention to the computational cost of the Jaya and Rao algorithms
along with their modified versions, convergence plots are presented in Fig. 6.2. In
the graph, curves represent the variation of the objective function with respect to
the function evaluations presented. It can be observed that the SAP-Rao algorithm
is converged faster than the other algorithms. Also, the convergence of the modified
versions is faster than the basic algorithms.
The next sub-section presents the analysis of the computational results related to
WFLO case-II.

Fig. 6.2 Convergence plots of the basic Jaya and Rao algorithms along with their modified versions
in case-I with 30 turbines
6.3 Case-II: Fixed Wind Speed and Variable Wind Direction 145

6.3 Case-II: Fixed Wind Speed and Variable Wind

Direction

It is the case of multi-directional wind with a fixed speed. In case-II, unlike case-
I, wind direction is varied from 0 to 360° with equal probability of occurrence.
Figure 6.3 shows the wind profile for the case-II.
A total of 36 directions are considered with 10° intervals. The incident wind speed
of the wind farm is taken as 12 m/s in all directions. The performances of proposed
algorithms are tested for 39 turbines configuration in case-II. Similar to the case-I, in
this case, also 9000 function evaluations are taken as termination criterion for all the
algorithms. In this case, the computational performances of the proposed algorithms
are compared with GA, LGA, ABC, TLBO, and TLBO variants (AL and PAL).
The computational results of various algorithms for the case-II with 39 turbines
are summarized in Table 6.3. In this case also, the Jaya algorithm and its modified
versions, Rao algorithms and their modified versions have obtained better layouts
in comparison to the layouts achieved by the GA, LGA, TLBO, ABC, AL, and
PAL algorithms. The AMTPG-Jaya algorithm has achieved the most efficient layout,
which produces 18,552.48 kW power with 91.76% efficiency. The SAP-Rao, ERao-
3, MTPG-Jaya, ERao-1, ERao-2, Rao-1, Rao-2, Rao-3, and Jaya algorithms have
achieved the next best solutions, respectively. The graphical representation of the
total power output of the solutions achieved by various algorithms in this case is
presented in Fig. 6.4. Here, an observation can be made that the layout achieved
by the AMTPG-Jaya algorithm has resulted in 7.7%, 5.4%, 1%, 2.7%, and 1.4%
increment in total power when compared to that of the solutions of the GA, LGA,
TLBO, ABC, and AL algorithms, respectively. In addition, the convergence plots of

Fig. 6.3 Wind profile for

WFLO problem case-II
146 6 Optimization of Wind Farm Layouts

Table 6.3 Results obtained by the various algorithms in case-II with 39 turbines
Algorithm Total power (kW/year) Farm efficiency Objective value
GA 17,220 85.17 0.001567
LGA 17,611 87.11 0.0015318
TLBO 18,401 91.01 0.001463
ABC 18,062 89.34 0.00149
AL 18,305 90.53 0.001466
PAL 18,441 91.21 0.001459
Jaya 18,448.3 91.24 0.0014589
MTPG-Jaya 18,478.89 91.40 0.001456902
AMTPG-Jaya 18,552.48 91.764 0.001451108
Rao-1 18,463.98 91.326 0.001468063
Rao-2 18,460.92 91.311 0.001468305
Rao-3 18,450.64 91.260 0.001459117
ERao-1 18,468.07 91.347 0.00145774
ERao-2 18,466.30 91.338 0.00145788
ERao-3 18,494.23 91.476 0.001455678
SAP-Rao 18,541.83 91.711 0.001451941
Source GA- Grady et al. (2005), LGA- Changshui et al. (2011), TLBO- Patel et al. (2015), ABC-,
AL-, and PAL- Patel et al. (2017)
Result in boldface indicates a better performing algorithm

Fig. 6.4 Total power produced by the layouts achieved by different algorithms in case-II with 39
turbines

the Jaya and Rao algorithms, along with their modified versions, are presented in
Fig. 6.5.
It can be observed that the convergence of the modified versions is faster than
the basic algorithms. The ERao-3 and AMTPG-Jaya algorithms have converged
6.3 Case-II: Fixed Wind Speed and Variable Wind Direction 147

Fig. 6.5 Convergence plots of the basic Jaya and Rao algorithms along with their modified versions
in case-II with 39 turbines

faster than the other algorithms. The next sub-section presents the analysis of the
computational results related to the WFLO case-III problems.

6.4 Case-III: Fixed Wind Speed and Variable Wind

Direction

Case-III is the case of multi-directional wind with varying wind speed. The variation
of wind direction is the same as that in case-II. In this case, the wind flows from 36
rotational directions with an unequal probability of occurrence for all wind velocity
in each direction. The incident wind speed of the wind farm is taken as 17, 12, and
8 m/s. The fraction of the occurrence of these velocities in 36 rotational directions
is presented in Fig. 6.6. The probability distribution for various wind speeds in all
considered directions is taken from Patel et al. (2017). In this case, 39 turbines wind
farm configuration is considered. The layouts obtained by the proposed algorithms
are compared with GA, LGA, ABC, and TLBO variants (AL and PAL). The proposed
algorithms are tested with 50,000 function evaluations as the termination criterion,
148 6 Optimization of Wind Farm Layouts

Fig. 6.6 Wind direction and velocity profile bar graph for WFLO case-III

which is the same as that of ABC, AL, and PAL algorithms. The computational
results of various algorithms are summarized in Table 6.4.

Table 6.4 Results obtained by the various algorithms in case-III with 39 turbines
Algorithm Total power (kW/year) Farm efficiency Objective value
GA 32,038 86.619 0.00080314
LGA 33,553 90.5 0.00080236
ABC 33,652 90.97 0.0008
AL 33,732 91.18 0.000798
PAL 33,810 91.4 0.000796
Jaya 34,147.92 92.3141 0.000788383
MTPG-Jaya 34,710.90 93.836 0.000775596
AMTPG-Jaya 34,703.54 93.816 0.000775761
Rao-1 34,089.86 92.157 0.000789726
Rao-2 34,093.57 92.167 0.00078964
Rao-3 34,108.13 92.207 0.000789303
ERao-1 34,614.30 93.575 0.000777761
ERao-2 34,708.46 93.829 0.000775651
ERao-3 34,647.53 93.665 0.000777015
SAP-Rao 34,665.86 93.714 0.000776604
Source GA- Grady et al. (2005), LGA- Changshui et al. (2011), TLBO- Patel et al. (2015), ABC-,
AL-, and PAL- Patel et al. (2017)
Result in boldface indicates a better performing algorithm.
6.4 Case-III: Fixed Wind Speed and Variable Wind Direction 149

Fig. 6.7 Total power produced by the layouts achieved by different algorithms in case-III with 39
turbines

In this case also, the Jaya algorithm and its modified versions, Rao algorithms
and their modified versions have obtained better layouts in comparison with the
layouts achieved by the GA, LGA, TLBO, ABC, AL, and PAL algorithms. The
graphical representation of the total power output of the solutions achieved by various
algorithms is presented in Fig. 6.7. The MTPG-Jaya algorithm has achieved the most
efficient layout, which produces 34,710.89 kW power with 93.836% efficiency. The
ERao-2, AMTPG-Jaya, SAP-Rao, ERao-3, ERao-1, Jaya, Rao-3, Rao-2, and Rao-1
algorithms have achieved the next best solutions, respectively. Here, an observation
can be made that the layout achieved by the MTPG-Jaya algorithm has resulted in
8.34%, 3.5%, 3.14%, 2.9%, 2.7%, 1.65%, 1.82%, 1.81%, and 1.77% increment in
total power when compared to that of the solutions achieved by the GA, LGA, ABC,
AL, PAL, Jaya, Rao-1, Rao-2, and Rao-3 algorithms, respectively.
In addition, the convergence plots of the Jaya and Rao algorithms, along with
their modified versions, in this case, are presented in Fig. 6.8. For this problem also,
the convergence of the modified versions is faster than the basic algorithms. The
ERao-2, ERao-3, and MTPG-Jaya algorithms have converged relatively faster than
the other algorithms.
The wind farm layouts achieved by the basic and proposed modified algorithms in
case-I with 30 turbines configuration are presented in Figs. 6.9 and 6.10. Similarly,
the wind farm layouts achieved by the basic and proposed modified algorithms in
case-II with 39 turbines configuration are presented in Figs. 6.11 and 6.12. The
wind farm layouts achieved by the basic Jaya and Rao algorithms and their modified
versions in case-III with 39 turbines configuration are presented in Figs. 6.13 and
6.14. From the visual representations of the layouts obtained by Jaya and its modified
versions, Rao and its modified versions, it can be observed that most of the turbines
are placed at the outer edge by both proposed algorithms.
For the considered wind farm characteristics, it can be observed that the Jaya algo-
rithm’s optimal layout is the same as that obtained by GA, and TLBO algorithms
150 6 Optimization of Wind Farm Layouts

Fig. 6.8 Convergence plots of the basic Jaya and Rao algorithms along with their modified versions
in case-III with 39 turbines

for case-I with 30 turbines problem. In the implementation of Jaya and TLBO algo-
rithms to the WFLO problem, the wind farm area is divided into 100 discrete grid
locations, and a turbine can be placed at the center of every cell in the square grid.
However, in the implementation of modified versions of the Jaya algorithm, Rao
algorithms, and modified versions of the Rao algorithms, it is not discrete. Instead,
it is considered as continuous space. In this implementation, turbine locations are
considered as two-dimensional Cartesian coordinates within the specified area, and
these variables are considered as continuous variables.
In general, an upstream wind turbine wake effect will be gradually expanding
behind the turbine. As the distance from the upstream turbine increases, the diameter
of the wake effect will increase. Wake radius will be crossing 100 m approximately
after moving 851 m behind the upstream turbine. Thus, there will be a greater number
of alternative locations in case-I with continuous space layout when compared to a
discrete grid layout, as in the case of TLBO and Jaya algorithms.
Here, it can be observed that, in all three cases, the performances of the proposed
modified algorithms are superior or competitive to those of the basic Jaya and Rao
algorithms. Hence, the modified versions of the Jaya and Rao algorithms can be
considered as the improved versions of the Jaya and Rao algorithms respectively.
6.4 Case-III: Fixed Wind Speed and Variable Wind Direction 151

Fig. 6.9 Wind farm layouts achieved by the Jaya, MTPG-Jaya, AMTPG-Jaya, and SAP-Rao
algorithms for case-I with 30 turbines

In all three cases, the results of the improved versions are superior to the results
presented by other algorithms in terms of farm efficiency, the total power generated,
and cost per unit power generated. In case-I with 30 turbines configuration, the layout
achieved by the SAP-Rao algorithm has resulted in a 2.7% increment in total power
when compared to that of the solutions of the GA, LGA, TLBO, and Jaya algorithms.
Similarly, the total power of the SAP-Rao algorithm is 1.3%, 1.5%, 1.2%, 1.1%, 1%,
and 1% higher when compared to that of the Rao-1, Rao-2, Rao-3, ERao-1, ERao-
2, and ERao-3 algorithms, respectively. In case-II with 39 turbines configuration,
the layout obtained by the AMTPG-Jaya has increased the total power by 7.7%,
5.4%, 1%, 2.7%, and 1.4% compared to that of GA, LGA, TLBO, ABC, and AL,
respectively. In case-III with 39 turbines configuration, the layout obtained by the
MTPG-Jaya has increased the total power by 8.34%, 3.5%, 3.14%, 2.9%, 2.7%,
1.65%, 1.82%, 1.81%, and 1.77% when compared to that of the GA, LGA, ABC,
AL, PAL, Jaya, Rao-1, Rao-2, and Rao-3 algorithms, respectively.
The next chapter presents the Jaya algorithm and Rao algorithms’ application
along with their modified versions to the solar-assisted energy systems.
152 6 Optimization of Wind Farm Layouts

Fig. 6.10 Wind farm layouts achieved by the Rao and elitist Rao algorithms for case-I with 30
turbines
6.4 Case-III: Fixed Wind Speed and Variable Wind Direction 153

Fig. 6.11 The wind farm layouts achieved by the Rao and elitist Rao algorithms for case-II with
39 turbines
154 6 Optimization of Wind Farm Layouts

Fig. 6.12 Wind farm layouts achieved by the Jaya, MTPG-Jaya, AMTPG-Jaya, and SAP-Rao
algorithms for case-II with 39 turbines
6.4 Case-III: Fixed Wind Speed and Variable Wind Direction 155

Fig. 6.13 Wind farm layouts achieved by the Jaya, MTPG-Jaya, AMTPG-Jaya, and SAP-Rao
algorithms for case-III with 39 turbines
156 6 Optimization of Wind Farm Layouts

Fig. 6.14 Wind farm layouts achieved by the Rao and elitist Rao algorithms for case-III with 39
turbines

References

Changshui, Z., Guangdong, H., & Jun, W. (2011). A fast algorithm based on the submodular property
for optimization of wind turbine positioning. Renewable Energy, 36(11), 2951–2958. https://doi.
org/10.1016/j.renene.2011.03.045
Grady, S. A., Hussaini, M. Y., & Abdullah, M. M. (2005). Placement of wind turbines using genetic
algorithms. Renewable Energy, 30(2), 259–270. https://doi.org/10.1016/j.renene.2004.05.007
Patel, J., Savsani, V., Patel, V., & Patel, R. (2017). Layout optimization of a wind farm to maximize
the power output using enhanced teaching learning based optimization technique. Journal of
Cleaner Production, 158, 81–94. https://doi.org/10.1016/j.jclepro.2017.04.132
References 157

Patel, J., Savsani, V., & Patel, R. (2015). Maximizing energy output of a wind farm using
teaching–learning-based optimization. Volume 2: Photovoltaics; Renewable-Non-Renewable
Hybrid Power System; Smart Grid, Micro-Grid Concepts; Energy Storage; Solar Chemistry;
Solar Heating and Cooling; Sustainable Cities and Communities, Transportation; Symposium
on Integrated/Sustainable Built. https://doi.org/10.1115/ES2015-49164
Chapter 7
Optimization of the Selected
Solar-Assisted Energy Systems

Abstract This chapter presents the applications of different versions of Jaya and
Rao algorithms to the problems of solar-assisted heat engine systems. In this chapter,
five multi-objective optimization case studies of solar-assisted energy systems which
includes a case study of solar-assisted Brayton heat engine system, three case studies
of solar-assisted Stirling heat engine system, a case study of solar-assisted Carnot-like
heat engine system, are considered for optimization. The optimization is carried out
using the Jaya algorithm, Rao algorithms, and their modified versions. Furthermore,
the solutions obtained in multi-objective optimization scenarios are non-dominated
in nature due to the conflicting nature of the objectives. Hence, to identify the best
solutions from the Pareto-fronts, the average rank method described in Chap. 5 is
used. Computational results revealed that the performances of the modified Jaya and
Rao algorithms are superior to those of the other algorithms. Also, the performances
of the selected systems are improved by the solutions of the proposed algorithms.

7.1 Optimization of a Solar-Assisted Brayton Heat Engine

System

The description of the selected solar-assisted Brayton heat engine system is presented
in Sect. 2.2.1. The detailed description and thermodynamic analysis of the solar-
assisted Brayton heat engine system considered in this book were presented by Li
et al. (2015). The three objective functions considered in this case study are thermal
efficiency (ηm ), power output (P), and non-dimensional thermo-economic perfor-
mance function (F) of the solar-assisted Brayton heat engine system. These three
objective functions are maximization functions which are given in Eq. 2.11, Eq. 2.22,
and Eq. 2.23. The decision variables of these objective functions are the temperature
of the hot reservoir (T H ), the temperature of the cold reservoir (T L ), the temperature
of the working fluid at state 1 of Brayton cycle (T 1 ), hot side heat exchange effective-
ness (εH ), cold side heat exchange effectiveness (εL ), and regenerator effectiveness
(εR ). The lower and upper boundaries of the decision variables are as follows:

700 K ≤ TH ≤ 1000 K (7.1)

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 159
V. R. Ravipudi et al., Design Optimization of Renewable Energy Systems Using Advanced
Optimization Algorithms, Green Energy and Technology,
https://doi.org/10.1007/978-3-030-95589-2_7
160 7 Optimization of the Selected Solar-Assisted Energy Systems

400 K ≤ TL ≤ 500 K (7.2)

TL ≤ T1 ≤ TH (7.3)

0.5 ≤ ε H ≤ 0.7 (7.4)

0.5 ≤ ε L ≤ 0.7 (7.5)

0.5 ≤ ε R ≤ 0.8 (7.6)

The characteristics of the solar-assisted Brayton heat engine system taken are as
follows: I = 1000Wm−2 , eC = 0.9, RC = 1300, η0 = 0.85, k = 4, ξ = 0.02,
δ = 5.67 × 10−8 WK−4 m−2 , h C = 20Wm−2 K−1 , Cw f = 1050 WK−1 , T0 = 300 K,
and h H = h L = 2000 WK−1 m−2 .
Li et al. (2015) had reported optimal solutions by employing the NSGA-II algo-
rithm and reported three optimal solutions using the TOPSIS, Shannon’s entropy, and
LINMAP methods. The computations of the NSGA-II algorithms were performed
by taking 125,000 function evaluations as the termination criterion. However, in this
book, the proposed algorithms are executed using only 10,000 function evaluations as
the termination criterion and reported the Pareto-optimal solutions using the average
rank based on the multiple decision-making methods. Firstly, single-objective opti-
mization is performed independently for each objective function, and then MOO is
performed in this book. In all the computational experiments, the population size is
maintained as 25 for all the algorithms. The elite size for the elitist Rao algorithms
is taken as 20%.
Table 7.1 presents the single-objective optimization results of this case study. The
Jaya algorithm, Rao algorithms, and their improved versions have obtained identical
solutions in all the single-objective optimization scenarios of this case study. Hence,
only one solution is presented here. Also, the solutions obtained by the improved
algorithms in all the single-objective optimization scenarios are superior to those
obtained by the NSGA-II algorithm. The maximum power output, thermal efficiency,
and thermo-economic function obtained by the proposed algorithms are 71.60 kW,
23.77%, and 0.3144, respectively.
In the multi-objective optimization, to demonstrate the proposed average rank
method for identifying the best compromise solution, the Pareto-front achieved by
the SAP-Rao algorithm is considered. The best compromise solution of the SAP-Rao
algorithm is found to be the best compromise solution among the solutions of the other
algorithms compared. Hence, the SAP-Rao algorithm Pareto-front is considered for
the demonstration of the proposed average rank method. Table 7.2 presents the Pareto-
optimal solutions obtained by the SAP-Rao algorithm in multi-objective optimization
of this case study. These solutions are non-dominated in nature, and all the solutions
can be considered as equivalent. However, the compromise among the objectives
Table 7.1 Results obtained by the NSGA-II and proposed algorithms in single-objective optimization scenarios of the solar-assisted Brayton engine system
case study
Objective Algorithm εH εL εR T H (K) T L (K) T 1 (K) P (kW) ηm F
Maximize-P NSGA-II 0.69 0.69 0.8 999.83 400.19 604.92 71.42 0.2203 0.2932
Proposed algorithms 0.7 0.7 0.8 1000 400 612.2243 71.6037 0.2276918 0.29949823
Maximize-ηm NSGA-II 0.7 0.69 0.79 999.84 400.15 567.97 68.06 0.2376 0.3124
Proposed algorithms 0.7 0.7 0.8 1000 400 566.3031 67.86533 0.23776 0.31442866
Maximize-F NSGA-II 0.69 0.69 0.79 999.78 400.11 564.21 67.42 0.2372 0.3144
Proposed algorithms 0.7 0.7 0.8 1000 400 563.2638 67.32262 0.2377065 0.3145
Source NSGA-II—Li et al. (2015)
7.1 Optimization of a Solar-Assisted Brayton Heat Engine System

Result in boldface indicates a better performing algorithm

161
162 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.2 Pareto-optimal solutions obtained by the SAP-Rao algorithm for the solar-assisted
Brayton engine system case study
Solution εH εL εR T H (K) T L (K) T 1 (K) P (kW) ηm F
1 0.7 0.7 0.8 1000 400 612.224 71.60 0.22769 0.29950
2 0.7 0.7 0.8 1000 400 566.306 67.87 0.23776 0.31443
3 0.7 0.7 0.8 1000 400 563.262 67.32 0.23771 0.31450
4 0.7 0.7 0.8 1000 400 608.822 71.59 0.22901 0.30133
5 0.7 0.7 0.8 1000 400 596.782 71.21 0.23304 0.30700
6 0.7 0.7 0.8 1000 400 605.909 71.54 0.23008 0.30282
7 0.7 0.7 0.8 1000 400 610.018 71.60 0.22856 0.30070
8 0.7 0.7 0.8 1000 400 603.221 71.47 0.23101 0.30413
9 0.7 0.7 0.8 1000 400 601.597 71.42 0.23155 0.30489
10 0.7 0.7 0.8 1000 400 586.947 70.52 0.23550 0.31059
11 0.7 0.7 0.8 1000 400 589.335 70.72 0.23498 0.30981
12 0.7 0.7 0.8 1000 400 599.993 71.36 0.23207 0.30562
13 0.7 0.7 0.8 1000 400 594.434 71.08 0.23370 0.30795
14 0.7 0.7 0.8 1000 400 590.984 70.85 0.23459 0.30924
15 0.7 0.7 0.8 1000 400 570.475 68.54 0.23766 0.31411
16 0.7 0.7 0.8 1000 400 605.053 71.52 0.23038 0.30325
17 0.7 0.7 0.8 1000 400 567.980 68.15 0.23775 0.31433
18 0.7 0.7 0.8 1000 400 592.524 70.96 0.23420 0.30868
19 0.7 0.7 0.8 1000 400 598.676 71.30 0.23247 0.30620
20 0.7 0.7 0.8 1000 400 582.310 70.08 0.23637 0.31192
21 0.7 0.7 0.8 1000 400 583.643 70.21 0.23614 0.31156
22 0.7 0.7 0.8 1000 400 585.205 70.37 0.23585 0.31112
23 0.7 0.7 0.8 1000 400 580.282 69.85 0.23669 0.31242
24 0.7 0.7 0.8 1000 400 593.177 71.00 0.23403 0.30843
25 0.7 0.7 0.8 1000 400 571.973 68.77 0.23758 0.31393

is different for each solution. Moreover, the solution that has the best compromise
among the objectives can be considered as the best solution. Hence, to identify the
solution with the best compromise, MADM methods can be used.
In the multi-attribute decision-making scenario, the MADM methods are used
to rank various alternatives based on their values of attributes. By considering each
solution as an alternative, and each objective function as an attribute, these solutions
are ranked using decision-making methods. Here, each method uses different prin-
ciples to rank the solutions. Thus, eight MADM methods are considered for ranking
these solutions instead of depending on a single method. Now, the average rank is
calculated based on the ranks suggested by these methods to the solutions. The solu-
tion with the least average rank value is considered as the best solution. The ranks
obtained by each solution using different decision-making methods are shown in
7.1 Optimization of a Solar-Assisted Brayton Heat Engine System 163

Table 7.3. Furthermore, to see if there is any correlation between the ranks suggested
by different pairs of MADM methods, Spearman’s correlation coefficients are calcu-
lated for different pairs of MADM methods. For calculating the average rank values,
only the ranks of the methods with a correlation coefficient value greater than 0.5
with the other methods are considered. Spearman’s correlation for different pairs of
rankings given by decision-making methods is shown in Table 7.4.
Spearman’s correlation coefficients for all the pairs of decision-making methods
are positive, which indicates that there is some similarity among the ranks suggested
by the MADM methods. Also, the pairs SAW-COPRAS and TOPSIS-MTOPSIS
have Spearman’s correlation coefficient equal to 1, which indicates that the ranking
of these pairs is identical. It can be confirmed by the ranks shown in Table 7.3. The
ranks suggested by the SAW and COPRAS methods are identical, and the ranks
suggested by the TOPSIS and MTOPSIS methods are also identical. If we consider
the SAW-WPM pair, Spearman’s correlation value is 0.997. It indicates that these
methods have a higher similarity in ranks suggested, and it can be observed in Table
7.3.
Similarly, for the PROMETHEE and GRA methods, Spearman’s correlation coef-
ficient values are lesser with other methods. It indicates that these methods have
lesser similarities in the ranks suggested to the solutions. Table 7.3 shows that the
ranks of these methods are different from those suggested by the other methods
for the first 19 solutions, and for the bottom six solutions, the suggested ranks
are the same as those given by other methods. If Spearman’s correlation coef-
ficient value is negative, then the ranks can be considered dissimilar and cannot
be used to calculate the average ranks. However, in this book, the lower threshold
value for Spearman’s correlation is taken as 0.5. Spearman’s correlation values for
the PROMETHEE-TOPSIS, PROMETHEE-MTOPSIS, GRA-TOPSIS, and GRA-
MTOPSIS pairs are less than 0.5. Also, for these four methods, Spearman’s corre-
lation value is greater than 0.5 with the remaining methods. If we compare the
pair PROMETHEE-TOPSIS, Spearman’s correlation values of TOPSIS method with
SAW, WPM, VIKOR, MTOPSIS, COPRAS, and GRA methods are 0.893, 0.913,
1, 0.966, 0.893, and 0.357, respectively. Moreover, Spearman’s correlation values
of the PROMETHEE method with SAW, WPM, VIKOR, MTOPSIS, COPRAS, and
GRA methods are 0.671, 0.64, 0.342, 0.541, 0.671, and 0.997, respectively. Here,
the TOPSIS method has better correlation values with five (SAW, WPM, VIKOR,
MTOPSIS, and COPRAS) methods, whereas the PROMETHEE method has a better
correlation value with one (GRA) method. Hence, the ranks suggested by the TOPSIS
method are considered for calculating average ranks and the ranks suggested by the
PROMETHEE method will be unused. Similarly, by comparing the PROMETHEE-
MTOPSIS, GRA-TOPSIS, and GRA-MTOPSIS pairs, the TOPSIS and MTOPSIS
methods are selected for calculating the average ranks.
Now, the corrected average ranks are calculated, excluding the ranks suggested
by the PROMETHEE and GRA methods, and presented in Table 7.3 as the corrected
ranks. Solution 20, with an average rank of 2.33, can be considered as the best
solution, which has the best compromise among the power output, system efficiency,
Table 7.3 Ranks suggested by the MADM methods for the SAP-Rao algorithm solutions for the solar-assisted Brayton engine system case study
164

Solution SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA AR CR FR

20 1 1 5 5 1 7 1 7 3.5 2.33 1
21 3 2 3 3 1 8 3 8 3.875 2.5 2
22 4 4 2 2 1 9 4 9 4.375 2.83 3
10 5 5 1 1 1 10 5 10 4.75 3 4
23 2 3 7 7 6 6 2 6 4.875 4.5 5
11 6 6 4 4 1 11 6 11 6.125 4.5 5
14 7 7 6 6 7 12 7 12 8 6.67 7
18 9 8 8 8 8 13 9 13 9.5 8.33 8
24 10 10 9 9 9 14 10 14 10.625 9.5 9
25 8 9 14 14 11 5 8 4 9.125 10.67 10
13 12 12 10 10 10 15 12 15 12 11 11
5 14 13 11 11 12 16 14 16 13.375 12.5 12
15 11 11 16 16 13 4 11 3 10.625 13 13
19 16 15 12 12 14 17 16 17 14.875 14.17 14
17 13 14 18 18 15 2 13 1 11.75 15.17 15
12 17 17 13 13 16 18 17 18 16.125 15.5 16
9 18 18 15 15 17 19 18 19 17.375 16.83 17
2 15 16 21 21 18 1 15 2 13.625 17.67 18
8 20 20 17 17 19 20 20 20 19.125 18.83 19
16 21 21 19 19 21 21 21 21 20.5 20.33 20
3 19 19 23 23 20 3 19 5 16.375 20.5 21
(continued)
7 Optimization of the Selected Solar-Assisted Energy Systems
Table 7.3 (continued)
Solution SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA AR CR FR
6 22 22 20 20 22 22 22 22 21.5 21.33 22
4 23 23 22 22 23 23 23 23 22.75 22.67 23
7 24 24 24 24 24 24 24 24 24 24 24
1 25 25 25 25 25 25 25 25 25 25 25
AR average rank, CR corrected rank, FR final rank
7.1 Optimization of a Solar-Assisted Brayton Heat Engine System
165
166 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.4 Spearman’s rank correlation coefficients between different pairs of MADM method’s
ranking for the SAP-Rao algorithm solutions
Method SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA
SAW 1 0.997 0.893 0.893 0.970 0.671 1 0.687
WPM 0.997 1 0.913 0.913 0.980 0.640 0.997 0.655
TOPSIS 0.893 0.913 1 1 0.966 0.342 0.893 0.357
MTOPSIS 0.893 0.913 1 1 0.966 0.342 0.893 0.357
VIKOR 0.970 0.980 0.966 0.966 1 0.541 0.970 0.556
PROMETHEE 0.671 0.640 0.342 0.342 0.541 1 0.671 0.997
COPRAS 1 0.997 0.893 0.893 0.970 0.671 1 0.687
GRA 0.687 0.655 0.357 0.357 0.556 0.997 0.687 1

and thermo-economic function values. In this way, the best solution from the Pareto-
optimal solutions is identified for all the MOO scenarios in this report.
The Pareto-optimal solutions achieved by the Jaya algorithm, Rao algorithms, and
their modified versions in this case study are presented in Tables 7.5, 7.6, 7.7, 7.8,
7.9, 7.10, 7.11 and 7.12. Here an observation can be made that the system thermal
efficiency is varied between 0.2276 and 0.2378, power output is varied between
67.31 and 71.60 kW, and non-dimensional thermo-economic function is varied from
0.2994 to 0.315. Similar to the SAP-Rao algorithm’s best solution, the best solutions
to these algorithms are identified. Solution 18 of the Jaya algorithm, Solution 6 of
the Rao-1 algorithm, Solution 16 of the Rao-2 algorithm, Solution 9 of the Rao-3
algorithm, Solution 5 of the AMTPG-Jaya algorithm, Solution 20 of the ERao-1
algorithm, Solution 10 of the ERao-2, and Solution 7 of the ERao-3 algorithm are
identified as the best solutions from the respective algorithm Pareto-front. Now, these
best solutions are compared with those of the NSGA-II algorithm in Table 7.13.
In Table 7.13, the solutions with the best average rank from the Pareto-optimal
solutions of the various algorithms are compared with those reported by the NSGA-
II algorithm. Figure 7.1 presents the Pareto-fronts achieved by the AMTPG-Jaya,
SAP-Rao, and elitist Rao algorithms, including the optimal solutions reported for
the NSGA-II, Jaya, and Rao algorithms. Here, an observation can be made that
the NSGA-II algorithm solutions selected by employing TOPSIS, Shannon-entropy,
and LINMAP methods belong to a lower level Pareto-front, which is dominated
by the proposed algorithm’s Pareto-fronts. The thermal efficiency, thermo-economic
function, and power output of the proposed algorithms solutions are better than
those reported for the NSGA-II algorithm. Furthermore, power output of the system
for the ERao-2 algorithm solution is highest, which is 3%, 2.35%, 2.4%, 0.34%,
0.4%, 0.18%, 0.5%, 0.33%, 0.3%, 0.2%, and 0.14% higher than that of the NSGA-II
(Shannon-entropy, LINMAP, and TOPSIS), Jaya, Rao-1, Rao-2, Rao-3, AMTPG-
Jaya, SAP-Rao, ERao-1, and ERao-3 algorithms solutions, respectively.
Similarly, the thermal efficiency for the Rao-3 algorithm solution is best, which
is 1.17%, 1.51%, 1.47%, 0.07%, 0.07%, 0.15%, 0.07%, 0.1%, 0.2%, 0.2%, and
0.2% higher when compared to that of the NSGA-II (Shannon-entropy, LINMAP,
7.1 Optimization of a Solar-Assisted Brayton Heat Engine System 167

Table 7.5 Pareto-optimal solutions obtained by the Jaya algorithm for the solar-assisted Brayton
engine system case study
Solution εH εL εR T H (K) T L (K) T 1 (K) P (kW) ηm F
1 0.7 0.7 0.8 1000 400 566.305 67.87 0.23776 0.31443
2 0.7 0.7 0.8 1000 400 563.267 67.32 0.23771 0.31450
3 0.7 0.7 0.8 1000 400 612.225 71.60 0.22769 0.29950
4 0.7 0.7 0.8 1000 400 596.314 71.19 0.23317 0.30719
5 0.7 0.7 0.8 1000 400 591.744 70.90 0.23440 0.30896
6 0.7 0.7 0.8 1000 400 580.194 69.84 0.23671 0.31244
7 0.7 0.7 0.8 1000 400 588.914 70.69 0.23507 0.30995
8 0.7 0.7 0.8 1000 400 606.486 71.55 0.22987 0.30253
9 0.7 0.7 0.8 1000 400 610.046 71.60 0.22855 0.30068
10 0.7 0.7 0.8 1000 400 564.353 67.52 0.23774 0.31449
11 0.7 0.7 0.8 1000 400 598.489 71.29 0.23253 0.30628
12 0.7 0.7 0.8 1000 400 583.930 70.24 0.23609 0.31148
13 0.7 0.7 0.8 1000 400 587.566 70.58 0.23537 0.31039
14 0.7 0.7 0.8 1000 400 602.978 71.46 0.23110 0.30425
15 0.7 0.7 0.8 1000 400 569.780 68.44 0.23769 0.31418
16 0.7 0.7 0.8 1000 400 586.322 70.47 0.23563 0.31078
17 0.7 0.7 0.8 1000 400 585.158 70.36 0.23586 0.31113
18 0.7 0.7 0.8 1000 400 581.822 70.02 0.23645 0.31204
19 0.7 0.7 0.8 1000 400 590.691 70.83 0.23466 0.30934
20 0.7 0.7 0.8 1000 400 599.285 71.33 0.23229 0.30593
21 0.7 0.7 0.8 1000 400 571.966 68.77 0.23758 0.31393
22 0.7 0.7 0.8 1000 400 568.977 68.31 0.23772 0.31425
23 0.7 0.7 0.8 1000 400 611.535 71.60 0.22797 0.29988
24 0.7 0.7 0.8 1000 400 576.678 69.42 0.23716 0.31318
25 0.7 0.7 0.8 1000 400 575.305 69.24 0.23731 0.31343

and TOPSIS), Jaya, Rao-1, Rao-2, AMTPG-Jaya, SAP-Rao, ERao-1, ERao-2, and
ERao-3 algorithms solutions, respectively. In addition, thermo-economic function of
the system for the Rao-3 algorithm solution is best, which is 1.1%, 1.49%, 1.46%,
0.08%, 0.08%, 0.2%, 0.09%, 0.12%, 0.2%, 0.3%, and 0.2% higher when compared
to that of the NSGA-II (Shannon-entropy, LINMAP, and TOPSIS), Jaya, Rao-1,
Rao-2, AMTPG-Jaya, SAP-Rao, ERao-1, ERao-2, and ERao-3 algorithms solutions,
respectively. However, these solutions may not be the best as they have lesser fitness
values in other objectives. Hence, to identify the best solution, which has the best
compromise among the objectives, the MADM methods based average ranks are
calculated and presented in Table 7.14.
The decision-making methods ranking of the Pareto-optimal solutions obtained
by different algorithms is shown in Table 7.14. Spearman’s correlation coefficients
168 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.6 Pareto-optimal solutions obtained by the Rao-1 algorithm for the solar-assisted Brayton
engine system case study
Solution εH εL εR T H (K) T L (K) T 1 (K) P (kW) ηm F
1 0.7 0.7 0.8 1000 400 612.214 71.60 0.22770 0.29950
2 0.7 0.7 0.8 1000 400 566.303 67.87 0.23776 0.31443
3 0.7 0.7 0.8 1000 400 563.264 67.32 0.23771 0.31450
4 0.7 0.7 0.8 1000 400 585.056 70.35 0.23588 0.31116
5 0.7 0.7 0.8 1000 400 590.844 70.84 0.23462 0.30929
6 0.7 0.7 0.8 1000 400 581.727 70.01 0.23647 0.31206
7 0.7 0.7 0.8 1000 400 605.101 71.52 0.23037 0.30322
8 0.7 0.7 0.8 1000 400 583.929 70.24 0.23609 0.31148
9 0.7 0.7 0.8 1000 400 606.784 71.56 0.22977 0.30238
10 0.7 0.7 0.8 1000 400 568.159 68.18 0.23774 0.31432
11 0.7 0.7 0.8 1000 400 579.597 69.78 0.23679 0.31257
12 0.7 0.7 0.8 1000 400 591.817 70.91 0.23438 0.30894
13 0.7 0.7 0.8 1000 400 573.878 69.04 0.23744 0.31366
14 0.7 0.7 0.8 1000 400 593.790 71.04 0.23387 0.30820
15 0.7 0.7 0.8 1000 400 611.429 71.60 0.22801 0.29993
16 0.7 0.7 0.8 1000 400 600.333 71.37 0.23196 0.30547
17 0.7 0.7 0.8 1000 400 610.164 71.60 0.22850 0.30062
18 0.7 0.7 0.8 1000 400 572.734 68.88 0.23753 0.31383
19 0.7 0.7 0.8 1000 400 576.185 69.35 0.23722 0.31328
20 0.7 0.7 0.8 1000 400 607.992 71.57 0.22932 0.30176
21 0.7 0.7 0.8 1000 400 563.572 67.38 0.23772 0.31450
22 0.7 0.7 0.8 1000 400 594.373 71.07 0.23371 0.30797
23 0.7 0.7 0.8 1000 400 609.250 71.59 0.22885 0.30110
24 0.7 0.7 0.8 1000 400 570.771 68.59 0.23765 0.31408
25 0.7 0.7 0.8 1000 400 600.983 71.40 0.23175 0.30517

for different pairs of rankings given by decision-making methods for different algo-
rithm’s solutions are shown in Table 7.15. The ranks given by the COPRAS, WPM,
and SAW methods are equal for each solution. Also, the TOPSIS and MTOPSIS
methods ranks for each solution are identical. Furthermore, Spearman’s correlation
for all the pairs formed by all the MADM methods (except for the pairs GRA-TOPSIS,
GRA-MTOPSIS, PROMETHEE-TOPSIS, and PROMETHEE-MTOPSIS) is greater
than 0.5. Also, Spearman’s correlation values for the GRA and PROMETHEE
methods with other methods (SAW, WPM, and COPRAS) is lesser when compared to
those of the TOPSIS and MTOPSIS methods. Thus, the ranks suggested by the GRA
and PROMETHEE methods cannot be considered for calculating the average ranks.
Hence, the corrected average ranks are calculated, excluding the ranks suggested by
7.1 Optimization of a Solar-Assisted Brayton Heat Engine System 169

Table 7.7 Pareto-optimal solutions obtained by the Rao-2 algorithm for the solar-assisted Brayton
engine system case study
Solution εH εL εR T H (K) T L (K) T 1 (K) P (kW) ηm F
1 0.7 0.7 0.8 1000 400 563.255 67.32 0.23771 0.31450
2 0.7 0.7 0.8 1000 400 566.297 67.86 0.23776 0.31443
3 0.7 0.7 0.8 1000 400 612.231 71.60 0.22769 0.29949
4 0.7 0.7 0.8 1000 400 602.434 71.45 0.23128 0.30450
5 0.7 0.7 0.8 1000 400 598.845 71.31 0.23242 0.30612
6 0.7 0.7 0.8 1000 400 603.701 71.49 0.23085 0.30390
7 0.7 0.7 0.8 1000 400 608.095 71.58 0.22928 0.30171
8 0.7 0.7 0.8 1000 400 592.338 70.94 0.23425 0.30875
9 0.7 0.7 0.8 1000 400 605.678 71.53 0.23016 0.30294
10 0.7 0.7 0.8 1000 400 606.290 71.55 0.22994 0.30263
11 0.7 0.7 0.8 1000 400 611.407 71.60 0.22802 0.29995
12 0.7 0.7 0.8 1000 400 610.149 71.60 0.22851 0.30063
13 0.7 0.7 0.8 1000 400 584.901 70.34 0.23591 0.31120
14 0.7 0.7 0.8 1000 400 609.331 71.59 0.22882 0.30106
15 0.7 0.7 0.8 1000 400 578.971 69.70 0.23688 0.31271
16 0.7 0.7 0.8 1000 400 582.905 70.14 0.23627 0.31176
17 0.7 0.7 0.8 1000 400 594.171 71.06 0.23377 0.30805
18 0.7 0.7 0.8 1000 400 570.213 68.50 0.23767 0.31414
19 0.7 0.7 0.8 1000 400 597.990 71.27 0.23268 0.30649
20 0.7 0.7 0.8 1000 400 573.696 69.02 0.23745 0.31369
21 0.7 0.7 0.8 1000 400 577.233 69.49 0.23710 0.31308
22 0.7 0.7 0.8 1000 400 586.318 70.47 0.23563 0.31078
23 0.7 0.7 0.8 1000 400 596.827 71.21 0.23302 0.30698
24 0.7 0.7 0.8 1000 400 589.028 70.70 0.23505 0.30991
25 0.7 0.7 0.8 1000 400 590.672 70.83 0.23466 0.30935

the GRA and PROMETHEE methods and presented in Table 7.14 as the corrected
ranks.
The SAP-Rao and Rao-2 algorithms solutions have achieved the least average
rank of 2. However, the SAP-Rao algorithm’s solution ranked one by four (SAW,
WPM, VIKOR, and COPRAS) methods, whereas the Rao-2 algorithm ranked one
by three (TOPSIS, MTOPSIS, and VIKOR) methods. Hence, the SAP-Rao and Rao-
2 algorithms solutions are assigned with the final ranks of 1 and 2, respectively.
The AMTPG-Jaya, ERao-3, and ERao-1 algorithms solutions have the next best
average ranks, which are 2.83, 3.17, and 3.33, respectively. The SAP-Rao algorithm
solution ranked one by four methods (SAW, WPM, VIKOR, and COPRAS) and
achieved a better average rank value. Furthermore, the power output of the system
for the SAP-Rao algorithm solution is 2.72%, 2.1%, and 2.1% higher than that of
170 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.8 Pareto-optimal solutions obtained by the Rao-3 algorithm for the solar-assisted Brayton
engine system case study
Solution εH εL εR T H (K) T L (K) T 1 (K) P (kW) ηm F
1 0.7 0.7 0.8 1000 400 563.259 67.32 0.23771 0.31450
2 0.7 0.7 0.8 1000 400 566.298 67.86 0.23776 0.31443
3 0.7 0.7 0.8 1000 400 612.229 71.60 0.22769 0.29950
4 0.7 0.7 0.8 1000 400 583.172 70.17 0.23623 0.31169
5 0.7 0.7 0.8 1000 400 586.873 70.52 0.23552 0.31061
6 0.7 0.7 0.8 1000 400 574.808 69.17 0.23736 0.31352
7 0.7 0.7 0.8 1000 400 578.664 69.67 0.23692 0.31278
8 0.7 0.7 0.8 1000 400 605.353 71.53 0.23028 0.30310
9 0.7 0.7 0.8 1000 400 580.778 69.91 0.23662 0.31230
10 0.7 0.7 0.8 1000 400 607.885 71.57 0.22936 0.30182
11 0.7 0.7 0.8 1000 400 569.338 68.37 0.23771 0.31422
12 0.7 0.7 0.8 1000 400 609.849 71.59 0.22862 0.30079
13 0.7 0.7 0.8 1000 400 611.078 71.60 0.22815 0.30013
14 0.7 0.7 0.8 1000 400 593.913 71.05 0.23384 0.30815
15 0.7 0.7 0.8 1000 400 592.007 70.92 0.23433 0.30887
16 0.7 0.7 0.8 1000 400 601.867 71.43 0.23146 0.30477
17 0.7 0.7 0.8 1000 400 594.656 71.09 0.23364 0.30786
18 0.7 0.7 0.8 1000 400 608.581 71.58 0.22910 0.30146
19 0.7 0.7 0.8 1000 400 572.332 68.82 0.23755 0.31389
20 0.7 0.7 0.8 1000 400 588.495 70.66 0.23517 0.31009
21 0.7 0.7 0.8 1000 400 604.496 71.51 0.23058 0.30352
22 0.7 0.7 0.8 1000 400 596.224 71.18 0.23320 0.30723
23 0.7 0.7 0.8 1000 400 571.704 68.73 0.23760 0.31397
24 0.7 0.7 0.8 1000 400 600.684 71.39 0.23185 0.30531
25 0.7 0.7 0.8 1000 400 565.190 67.67 0.23775 0.31447

the NSGA-II (Shannon-entropy, LINMAP, and TOPSIS), respectively. Similarly,

the system’s thermal efficiency for the SAP-Rao algorithm solution is 1.1%, 1.4%,
and 1.34% higher than that of the NSGA-II algorithm (Shannon-entropy, LINMAP,
and TOPSIS) solutions, respectively. The system’s thermo-economic function for
the SAP-Rao algorithm solution is 1%, 1.4%, and 1.34% higher when compared to
that of the NSGA-II algorithm (Shannon-entropy, LINMAP, and TOPSIS) solutions,
respectively.
Now, the performances of the Jaya algorithm, Rao algorithms, and their modified
versions in the MOO scenario are evaluated based on the hypervolume, coverage, and
spacing indicators. Table 7.16 presents the hypervolume and spacing values of the
Pareto-fronts obtained by different algorithms in the MOO scenario. The Pareto-front
obtained by the NSGA-II algorithm was not reported by Li et al. (2015). Thus, the
7.1 Optimization of a Solar-Assisted Brayton Heat Engine System 171

Table 7.9 Pareto-optimal solutions obtained by the AMTPG-Jaya algorithm for the solar-assisted
Brayton engine system case study
Solution εH εL εR T H (K) T L (K) T 1 (K) P (kW) ηm F
1 0.7 0.699999991 0.8 1000 400 563.262 67.32 0.23771 0.31450
2 0.7 0.7 0.8 1000 400 566.311 67.87 0.23776 0.31443
3 0.7 0.699999991 0.8 1000 400 612.244 71.60 0.22768 0.29949
4 0.7 0.699999993 0.8 1000 400 606.777 71.56 0.22977 0.30238
5 0.7 0.7 0.8 1000 400 581.877 70.03 0.23644 0.31203
6 0.7 0.699999994 0.8 1000 400 575.446 69.26 0.23729 0.31341
7 0.7 0.7 0.8 1000 400 608.815 71.59 0.22901 0.30133
8 0.7 0.699999999 0.8 1000 400 610.683 71.60 0.22830 0.30034
9 0.7 0.699999989 0.8 1000 400 600.180 71.37 0.23201 0.30553
10 0.7 0.699999998 0.8 1000 400 604.345 71.50 0.23063 0.30359
11 0.7 0.7 0.8 1000 400 597.888 71.27 0.23271 0.30654
12 0.7 0.699999993 0.8 1000 400 583.956 70.25 0.23609 0.31147
13 0.7 0.699999993 0.8 1000 400 573.810 69.03 0.23744 0.31367
14 0.7 0.7 0.8 1000 400 578.762 69.68 0.23691 0.31276
15 0.7 0.699999992 0.8 1000 400 586.778 70.51 0.23553 0.31064
16 0.7 0.7 0.8 1000 400 594.830 71.10 0.23359 0.30779
17 0.7 0.699999991 0.8 1000 400 596.870 71.21 0.23301 0.30696
18 0.7 0.699999973 0.8 1000 400 565.280 67.69 0.23776 0.31447
19 0.7 0.699999993 0.8 1000 400 593.262 71.01 0.23401 0.30840
20 0.7 0.69999999 0.8 1000 400 571.915 68.76 0.23758 0.31394
21 0.7 0.7 0.8 1000 400 578.820 69.68 0.23690 0.31275
22 0.7 0.699999992 0.8 1000 400 595.617 71.15 0.23337 0.30748
23 0.7 0.699999996 0.8 1000 400 588.576 70.66 0.23515 0.31006
24 0.7 0.699999992 0.8 1000 400 589.841 70.76 0.23486 0.30964
25 0.7 0.699999988 0.8 1000 400 603.016 71.47 0.23108 0.30423

performance metric values for the NSGA-II algorithm were not presented here. The
SAP-Rao algorithm has a better spacing value, which is 49%, 6.2%, 17.64%, 10.32%,
28.43%, 43.44%, 18.34%, and 29.25% lesser value when compared to that of the
Jaya, Rao-1, Rao-2, Rao-3, AMTPG-Jaya, ERao-1, ERao-2, and ERao-3 algorithms,
respectively. The ERao-3 algorithm Pareto-front has a relatively higher hypervolume
value. The ERao-1, ERao-2, and SAP-Rao algorithms have achieved the next better
hypervolume values, respectively. Here, an observation can be made that the values of
hypervolume achieved by the algorithms compared have relatively small differences
with each other. It indicates that the Pareto-fronts of these algorithms are at the same
level and spread over the maximum volume, which can be observed in Fig. 7.1.
Besides, the distribution of the solutions along the Pareto-front is more uniform for
the SAP-Rao algorithm.
172 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.10 Pareto-optimal solutions obtained by the ERao-1 algorithm for the solar-assisted
Brayton engine system case study
Solution εH εL εR T H (K) T L (K) T 1 (K) P (kW) ηm F
1 0.7 0.7 0.8 1000 400 566.303 67.87 0.23776 0.31443
2 0.7 0.7 0.8 1000 400 612.242 71.60 0.22768 0.29949
3 0.7 0.7 0.8 1000 400 563.237 67.32 0.23771 0.31450
4 0.7 0.7 0.8 1000 400 601.768 71.43 0.23150 0.30481
5 0.7 0.7 0.8 1000 400 610.657 71.60 0.22831 0.30035
6 0.7 0.7 0.8 1000 400 606.756 71.56 0.22978 0.30240
7 0.7 0.7 0.8 1000 400 597.449 71.24 0.23284 0.30672
8 0.7 0.7 0.8 1000 400 604.745 71.51 0.23049 0.30340
9 0.7 0.7 0.8 1000 400 580.658 69.90 0.23664 0.31233
10 0.7 0.7 0.8 1000 400 594.270 71.07 0.23374 0.30801
11 0.7 0.7 0.8 1000 400 578.646 69.66 0.23692 0.31278
12 0.7 0.7 0.8 1000 400 596.705 71.21 0.23306 0.30703
13 0.7 0.7 0.8 1000 400 572.759 68.89 0.23753 0.31383
14 0.7 0.7 0.8 1000 400 592.204 70.94 0.23428 0.30880
15 0.7 0.7 0.8 1000 400 590.089 70.78 0.23480 0.30955
16 0.7 0.7 0.8 1000 400 585.153 70.36 0.23586 0.31113
17 0.7 0.7 0.8 1000 400 592.080 70.93 0.23431 0.30884
18 0.7 0.7 0.8 1000 400 588.618 70.67 0.23514 0.31005
19 0.7 0.7 0.8 1000 400 576.251 69.36 0.23721 0.31326
20 0.7 0.7 0.8 1000 400 583.024 70.15 0.23625 0.31173
21 0.7 0.7 0.8 1000 400 595.159 71.12 0.23350 0.30766
22 0.7 0.7 0.8 1000 400 583.121 70.16 0.23623 0.31170
23 0.7 0.7 0.8 1000 400 605.895 71.54 0.23008 0.30283
24 0.7 0.7 0.8 1000 400 587.024 70.53 0.23548 0.31056
25 0.7 0.7 0.8 1000 400 575.366 69.25 0.23730 0.31342

The coverage metric values of the Pareto-fronts obtained by the Jaya algorithm,
Rao algorithms, and their modified versions are presented in Table 7.17. The coverage
of all algorithms compared versus Rao-1 and Jaya algorithms is zero (column Jaya
and Rao-1 in Table 7.17), which indicates that n the other methods dominated no
solution of the Rao-1 and Jaya algorithms. In contrast, the coverage values of Rao-1
versus Rao-2, Rao-3, AMTPG-Jaya, SAP-Rao, ERao-1, ERao-2, and ERao-3 algo-
rithms are 4, 4, 8, 4, 8, 4, and 8, respectively. It implies that the Rao-1 algorithm’s
solutions dominate 4% of the Rao-2 and Rao-3 algorithms solutions, 8% of AMTPG-
Jaya algorithm solutions, 4% of the SAP-Rao algorithm solutions, 8% of ERao-1
algorithm solutions, 4% of ERao-2 algorithm solutions, and 8% of ERao-3 algo-
rithm solutions. Similarly, the coverage values of all compared (except for Rao-1)
algorithms versus the SAP-Rao algorithm are zero. The Jaya, Rao-1, and SAP-Rao
7.1 Optimization of a Solar-Assisted Brayton Heat Engine System 173

Table 7.11 Pareto-optimal solutions obtained by the ERao-2 algorithm for the solar-assisted
Brayton engine system case study
Solution εH εL εR T H (K) T L (K) T 1 (K) P (kW) ηm F
1 0.7 0.7 0.8 1000 400 563.245 67.32 0.23771 0.31450
2 0.7 0.7 0.8 1000 400 612.198 71.60 0.22770 0.29951
3 0.7 0.7 0.8 1000 400 566.302 67.87 0.23776 0.31443
4 0.7 0.7 0.8 1000 400 609.502 71.59 0.22875 0.30097
5 0.7 0.7 0.8 1000 400 605.098 71.52 0.23037 0.30322
6 0.7 0.7 0.8 1000 400 607.990 71.57 0.22932 0.30176
7 0.7 0.7 0.8 1000 400 593.544 71.02 0.23394 0.30829
8 0.7 0.7 0.8 1000 400 574.206 69.09 0.23741 0.31361
9 0.7 0.7 0.8 1000 400 594.693 71.09 0.23363 0.30785
10 0.7 0.7 0.8 1000 400 584.131 70.26 0.23605 0.31142
11 0.7 0.7 0.8 1000 400 598.798 71.31 0.23244 0.30614
12 0.7 0.7 0.8 1000 400 596.386 71.19 0.23315 0.30716
13 0.7 0.7 0.8 1000 400 586.394 70.48 0.23561 0.31076
14 0.7 0.7 0.8 1000 400 567.970 68.15 0.23775 0.31433
15 0.7 0.7 0.8 1000 400 602.189 71.44 0.23136 0.30462
16 0.7 0.7 0.8 1000 400 597.390 71.24 0.23286 0.30675
17 0.7 0.7 0.8 1000 400 579.851 69.80 0.23676 0.31252
18 0.7 0.7 0.8 1000 400 600.904 71.39 0.23178 0.30521
19 0.7 0.7 0.8 1000 400 581.666 70.01 0.23648 0.31208
20 0.7 0.7 0.8 1000 400 599.975 71.36 0.23207 0.30563
21 0.7 0.7 0.8 1000 400 606.915 71.56 0.22972 0.30231
22 0.7 0.7 0.8 1000 400 591.555 70.89 0.23445 0.30903
23 0.7 0.7 0.8 1000 400 588.627 70.67 0.23514 0.31005
24 0.7 0.7 0.8 1000 400 590.363 70.80 0.23474 0.30946
25 0.7 0.7 0.8 1000 400 576.956 69.45 0.23713 0.31313

algorithms have achieved better coverage values than the other algorithms compared.
One out of 25 Pareto solutions of the SAP-Rao and ERao-2 algorithms are dominated
by the solutions of the other algorithms compared. Similarly, two out of 25 Pareto
solutions are dominated from the Pareto-fronts of the Rao-2, Rao-3, AMTPG-Jaya,
ERao-1, and ERao-3 algorithms.
From the computational results of the solar-assisted Brayton engine case study,
it can be observed that the performance of the considered system can be modified
with the solutions achieved by the proposed algorithms. The proposed algorithms
have achieved better solutions than those achieved by the NSGA-II algorithm in
fewer function evaluations. The solution of the SAP-Rao and Rao-2 algorithm has
better compromise among the power output, thermal efficiency, and non-dimensional
thermo-economic function. By the SAP-Rao algorithm’s solution, the power output
174 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.12 Pareto-optimal solutions obtained by the ERao-3 algorithm for the solar-assisted
Brayton engine system case study
Solution εH εL εR T H (K) T L (K) T 1 (K) P (kW) ηm F
1 0.70000000 0.7 0.8 1000 400 612.237 71.60 0.22769 0.29949
2 0.69999993 0.7 0.8 999.9998262 400 566.308 67.87 0.23776 0.31443
3 0.69999994 0.7 0.8 999.9999246 400 563.211 67.31 0.23770 0.31450
4 0.69999998 0.7 0.8 999.99998 400 601.765 71.43 0.23150 0.30481
5 0.69999996 0.7 0.8 1000 400 575.366 69.25 0.23730 0.31342
6 0.70000000 0.7 0.8 999.9998695 400 605.294 71.53 0.23030 0.30313
7 0.69999998 0.7 0.8 999.9997704 400 583.180 70.17 0.23622 0.31169
8 0.70000000 0.7 0.8 1000 400 585.136 70.36 0.23586 0.31114
9 0.69999995 0.7 0.8 999.9999623 400 586.709 70.50 0.23555 0.31066
10 0.69999998 0.7 0.8 999.9999731 400 588.535 70.66 0.23516 0.31008
11 0.69999999 0.7 0.8 999.9999602 400 594.840 71.10 0.23359 0.30779
12 0.69999997 0.7 0.8 1000 400 572.018 68.78 0.23758 0.31393
13 0.70000000 0.7 0.8 999.9997307 400 610.956 71.60 0.22819 0.30019
14 0.69999993 0.7 0.8 999.9999285 400 608.760 71.58 0.22903 0.30136
15 0.69999988 0.7 0.8 999.9998717 400 580.594 69.89 0.23665 0.31234
16 0.69999991 0.7 0.8 1000 400 597.238 71.23 0.23290 0.30681
17 0.69999998 0.7 0.8 999.999813 400 592.112 70.93 0.23431 0.30883
18 0.69999995 0.7 0.8 1000 400 577.360 69.50 0.23708 0.31305
19 0.69999989 0.7 0.8 999.9999166 400 600.813 71.39 0.23181 0.30525
20 0.69999995 0.7 0.8 999.9997208 400 593.865 71.04 0.23385 0.30817
21 0.69999994 0.7 0.8 1000 400 607.084 71.56 0.22966 0.30223
22 0.69999985 0.7 0.8 999.9997966 400 567.982 68.15 0.23774 0.31433
23 0.69999981 0.7 0.8 999.9997094 400 578.461 69.64 0.23695 0.31282
24 0.69999983 0.7 0.8 999.9999883 400 606.634 71.55 0.22982 0.30246
25 0.69999992 0.7 0.8 999.9998098 400 608.556 71.58 0.22911 0.30147

is increased by 2.72%, the thermal efficiency is increased by 1.4%, and the thermo-
economic function of the system is increased by 1.4% when compared to that of the
NSGA-II (Shannon-entropy, LINMAP, and TOPSIS) algorithm solutions. Further-
more, the spacing value achieved by the SAP-Rao algorithm is much better than that
achieved by other algorithms. Also, the SAP-Rao algorithm’s performance in terms
of hypervolume and coverage metrics is better or competitive to that of the other
algorithms. In addition, the performances of the modified versions in this case study
are better or competitive to those of the basic algorithms as well as the NSGA-II
algorithm. The next section presents the application of the Jaya and Rao algorithms
along with their modified versions, to the solar-assisted Stirling heat engine case
studies.
Table 7.13 Best solutions obtained by various algorithms in MOO scenario of the solar-assisted Brayton engine system case study
Algorithm εH εL εR T H (K) T L (K) T 1 (K) P (kW) ηm F
Entropy NSGA-II 0.69 0.7 0.79 999.99 400.06 582.33 68.22 0.23390 0.30890
LINMAP 0.69 0.69 0.79 999.99 400.01 586.87 68.65 0.23310 0.30770
TOPSIS 0.7 0.7 0.8 999.99 400.01 586.69 68.64 0.23320 0.30780
Jaya 0.7 0.7 0.8 1000 400 581.8217 70.02 0.23645 0.31204
Rao-1 0.7 0.7 0.8 1000 400 581.7274 70.01 0.23647 0.31206
Rao-2 0.7 0.7 0.8 1000 400 582.9045 70.14 0.23627 0.31176
Rao-3 0.7 0.7 0.8 1000 400 580.7785 69.91 0.23662 0.31230
AMTPG-Jaya 0.7 0.7 0.8 1000 400 581.8773 70.03 0.23644 0.31203
SAP-Rao 0.7 0.7 0.8 1000 400 582.3098 70.08 0.23637 0.31192
ERao-1 0.7 0.7 0.8 1000 400 583.0244 70.15 0.23625 0.31173
7.1 Optimization of a Solar-Assisted Brayton Heat Engine System

ERao-2 0.7 0.7 0.8 1000 400 584.1309 70.26 0.23605 0.31142
ERao-3 0.699999976 0.7 0.8 999.99 400 583.1799 70.17 0.23622 0.31169
Result in boldface indicates better values
175
176 7 Optimization of the Selected Solar-Assisted Energy Systems

Fig. 7.1 Plot of Pareto-fronts of the proposed algorithms and solutions of the methods compared
in solar-assisted Brayton engine system case study

7.2 Optimization of a Solar-Assisted Stirling Heat Engine

System

The description of the solar-assisted Stirling heat engine system is presented in

Sect. 2.2.2. Three case studies of the solar-assisted heat engine system are considered
for optimization in this book. These case studies are presented in the subsequent
sections.

7.2.1 Case Study-1

The description of this case study is presented in Sect. 2.2.2.1. The detailed descrip-
tion and thermodynamic analysis of the solar-assisted Stirling heat engine system
considered in this case study were presented by Ahmadi et al. (2013a). The perfor-
mance criteria of this case study are to maximize output power (P) and thermal
efficiency (ηm ), and minimize the rate of entropy generation (σ ). These are given in
Eqs. 2.24–2.26. The decision variables and their boundaries are presented in Table
7.18. The characteristics of the solar-assisted Stirling heat engine system taken are
as follows: λ = 2, ε = 0.9, I = 1000Wm−2 , K 0 = 2.5WK−1 , Cv = 15Jmol−1 k−1 ,
ζ = 2×10−10 , R = 4.3Jmol−1 K−1 , δ = 5.67×10−8 WK−4 m−2 , h = 20Wm−2 K−1 ,
η0 = 0.9 (1/M1 + 1/M2 ) = 2 × 10−5 sK−1 , TH 1 = 1300K, TL1 = 290K,
T0 = 288K, C = 1300, and n = 1mol.
Table 7.14 Ranks suggested by the MADM methods for different algorithm solutions presented in Table 7.13
Algorithm SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA AR CR FR
Entropy 10 10 12 12 1 10 10 10 9.4 9.17 10
LINMAP 12 12 10 10 12 12 12 12 11.5 11.3 12
TOPSIS 11 11 11 11 11 11 11 11 11 11 11
Jaya 7 7 7 7 1 3 7 3 5.3 6 7
Rao-1 6 6 8 8 1 2 6 2 4.9 5.83 6
Rao-2 3 3 1 1 1 6 3 6 3 2 2
Rao-3 8 8 9 9 1 1 8 1 5.6 7.17 9
AMTPG-Jaya 2 2 5 5 1 4 2 4 3.1 2.83 3
SAP-Rao 1 1 4 4 1 5 1 5 2.8 2 1
ERao-1 5 5 2 2 1 7 5 7 4.3 3.33 5
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System

ERao-2 9 9 6 6 1 9 9 9 7.3 6.67 8

ERao-3 4 4 3 3 1 8 4 8 4.4 3.17 4
AR average rank, CR corrected rank, FR final rank
177
178 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.15 Spearman’s rank correlation coefficients between different pairs of MADM method’s
rankings presented in Table 7.14
Method SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA
SAW 1 1 0.811 0.811 0.650 0.545 1 0.545
WPM 1 1 0.811 0.811 0.650 0.545 1 0.545
TOPSIS 0.811 0.811 1 1 0.511 0.266 0.811 0.266
MTOPSIS 0.811 0.811 1 1 0.511 0.266 0.811 0.266
VIKOR 0.650 0.650 0.511 0.511 1 0.650 0.650 0.650
PROMETHEE 0.545 0.545 0.266 0.266 0.650 1 0.545 1
COPRAS 1 1 0.811 0.811 0.650 0.545 1 0.545
GRA 0.545 0.545 0.266 0.266 0.650 1 0.545 1

Table 7.16 Hypervolume and spacing values of the Pareto-fronts obtained by the proposed
algorithms in MOO for the solar-assisted Brayton engine case study
Algorithm Hypervolume Spacing
Jaya 0.0265084 0.0562112
Rao-1 0.0265094 0.0305907
Rao-2 0.0265131 0.0348516
Rao-3 0.0265120 0.0320051
AMTPG-Jaya 0.0265110 0.0401055
SAP-Rao 0.0265148 0.0287033
ERao-1 0.026521 0.050744
ERao-2 0.026517 0.035148
ERao-3 0.026533 0.040567
Results in the bold figure indicate better values

Table 7.17 Coverage (%) values of the Pareto-fronts obtained by the proposed algorithms in MOO
for the solar-assisted Brayton engine case study
Algorithm Jaya Rao-1 Rao-2 Rao-3 AMTPG-Jaya SAP-Rao ERao-1 ERao-2 ERao-3
Jaya – 0 8 8 8 0 8 4 8
Rao-1 0 – 4 4 8 4 8 4 8
Rao-2 0 0 – 0 4 0 8 4 8
Rao-3 0 0 8 – 4 0 8 4 8
AMTPG-Jaya 0 0 0 0 – 0 4 4 4
SAP-Rao 0 0 8 8 8 – 8 4 8
ERao-1 0 0 0 0 4 0 – 0 4
ERao-2 0 0 0 0 4 0 4 – 4
ERao-3 0 0 0 0 4 0 4 0 –
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System 179

Table 7.18 Decision

Decision variables Boundaries
variables and their boundaries
of the Stirling heat engine Regenerator effectiveness (εR ) 0.4 ≤ εR ≤ 0.9
case study-1 High-temperature heat exchanger 0.4 ≤ εH ≤ 0.8
effectiveness (εH )
Low-temperature heat exchanger 0.4 ≤ εL ≤ 0.8
effectiveness (εL )
Heat capacitance rate of the heat source 300 ≤ C H ≤ 1800
(C H )
Heat capacitance rate of the heat sink 300 ≤ C L ≤ 1800
(C L )
Working fluid temperature during the 800 ≤ T h ≤ 1000
isothermal expansion (T h )
Working fluid temperature during 400 ≤ T c ≤ 510
isothermal compression (T c )

Ahmadi et al. (2013a) solved this multi-objective optimization problem using

the NSGA-II algorithm and reported three Pareto-optimal solutions. These three
solutions were documented based on the similarity to the ideal solutions calcu-
lated by employing TOPSIS, LINMAP, and Fuzzy Bellman–Zadeh decision-making
methods. The number of function evaluations taken by the NSGA-II is 400,000.
However, in this book, the Jaya algorithm, Rao algorithms, and their modified
versions are tested by taking 10,000 function evaluations as the termination crite-
rion. The best Pareto-optimal solutions are reported using the average rank based
on multiple decision-making methods. Similar to the previous case studies, in this
case study also, first single-objective optimization is performed for each objective
function independently, and then multi-objective optimization is performed by the
proposed algorithms. In all the computational experiments, the population size is
maintained as 25 for all the algorithms. The elite size for the elitist Rao algorithms
is taken as 20%.
Table 7.19 presents the single-objective optimization results of the Stirling heat
engine case study-1. The Jaya algorithm, Rao algorithms, and their modified versions
have obtained identical solutions in all the single-objective optimization scenarios,

Table 7.19 Results obtained by the proposed algorithms for single-objective optimization scenarios
of the solar-assisted Stirling heat engine case study-1
Objective εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)
Maximize-P 0.9 0.8 0.8 1800 1800 1000 510 70.3417 31.5795 139.2852
Maximize- ηm 0.9 0.8 0.8 1800 1800 1000 400 61.1952 36.5151 106.3605
Minimize- σ 0.9 0.4 0.4 300 300 1000 400 8.1778 32.2944 21.9193
180 7 Optimization of the Selected Solar-Assisted Energy Systems

which are presented in Table 7.19. The maximum power output and thermal effi-
ciency obtained by the proposed algorithms are 70.3417 kW and 36.5151%, respec-
tively. Similarly, the minimum entropy obtained by the proposed algorithms is
21.9193 W/ K.
The Pareto-optimal solutions achieved by the Jaya algorithm, Rao algorithms,
and their modified versions in this case study are presented in Tables 7.20, 7.21,
7.22, 7.23, 7.24, 7.25, 7.26, 7.27 and 7.28. The best solutions from the Pareto-fronts
obtained by the proposed algorithms are identified based on the average rank method.
Solution 13 of the Jaya algorithm, Solution 15 of the Rao-1 algorithm, Solution 17 of
the Rao-2 algorithm, Solution 8 of the Rao-3 algorithm, Solution 7 of the AMTPG-
Jaya algorithm, Solution 1 of the SAP-Rao algorithms, Solution 3 of the ERao-1
algorithm, Solution 18 of the ERao-2, and Solution 4 of the ERao-3 algorithm are
identified as the best solutions from the respective algorithm Pareto-front and are
compared in Table 7.29.
In Table 7.29, the solutions with the best average rank from the Pareto-optimal
solutions of the proposed algorithms are compared with those reported by the NSGA-
II algorithm. Figure 7.2 presents the Pareto-fronts achieved by the AMTPG-Jaya,
SAP-Rao, and elitist Rao algorithms, including the optimal solutions reported for
the NSGA-II, Jaya, and Rao algorithms. The conflicting nature of these objectives
can be observed in Fig. 7.2. Any change in the design variables that will result in
the increment of output power also leads to the increment of the generated entropy,
which is not desirable. Also, the variation of thermal efficiency is relatively less when
compared to the other objectives. It can also be observed from the Pareto-optimal
solutions obtained by the algorithms. It indicates that the effect of the design variables
on thermal efficiency is relatively less.
The decision-making methods ranking of the Pareto-optimal solutions obtained
by different algorithms is shown in Table 7.30. Spearman’s correlation coefficients
for different pairs of rankings given by decision-making methods for different
algorithm’s solutions are shown in Table 7.31. The ranks given by the TOPSIS
and MTOPSIS methods for each solution are identical. Furthermore, Spearman’s
correlation for all the pairs of decision-making methods (except with the VIKOR,
PROMETHEE, and GRA methods) is greater than 0.5. Spearman’s correlation values
for the VIKOR-SAW, VIKOR-COPRAS, and VIKOR-GRA pairs are negative, and
with other methods are less than 0.5. Similarly, the PROMETHEE method has
Spearman’s correlation values less than 0.5 with the SAW and COPRAS methods.
The GRA method has Spearman’s correlation values less than 0.5 with the WPM,
TOPSIS, and MTOPSIS methods. Hence, the ranks suggested by the VIKOR,
PROMETHEE, and GRA methods cannot be considered for calculating the average
ranks. Now, the corrected average ranks are calculated, excluding the ranks suggested
by the VIKOR, PROMETHEE, and GRA methods and presented in Table 7.30 as
corrected ranks.
The ERao-3 algorithm solution has the least average rank, which is 1.4. Thus, it can
be regarded as the best solution. The SAP-Rao, ERao-2, ERao-1, and AMTPG-Jaya
algorithm’s solutions have the next best average ranks, which are 2.4, 2.6, 4.8, and
6.2, respectively. The ERao-3 algorithm solution ranked one by three methods (SAW,
Table 7.20 Pareto-optimal solutions obtained by the Jaya algorithm for the solar-assisted Stirling heat engine case study-1
Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)
1 0.9 0.8 0.8 1800 1658.125 1000 400 58.9167 36.4942 102.5694
2 0.9 0.8 0.8 1800 1800 1000 508.8175 70.3137 31.6357 138.9775
3 0.899991 0.4 0.4 300 300 1000 400 8.1776 32.2937 21.9196
4 0.89996 0.8 0.8 1800 1800 1000 454.1751 67.6780 34.1603 124.2274
5 0.9 0.8 0.8 1800 1800 1000 482.4467 69.3978 32.8731 132.0330
6 0.899967 0.704188 0.8 1570.617 300 1000 440.3984 21.9561 33.7163 43.0688
7 0.9 0.8 0.8 1800 1285.157 1000 400 51.8134 36.4175 90.7505
8 0.89993 0.597117 0.698768 1744.565 300 1000 400 16.5111 34.6532 34.1212
9 0.9 0.77818 0.408339 1589.13 354.4436 1000 400.3019 12.0975 34.3312 28.2523
10 0.899946 0.792293 0.799236 1314.779 300 1000 400 18.4562 35.2864 35.3098
11 0.9 0.8 0.8 1800 1800 1000 503.9247 70.1866 31.8677 137.7023
12 0.899994 0.79626 0.799964 1271.048 768.6807 1000 400 36.7631 36.1514 65.7481
13 0.9 0.8 0.8 1183.395 1202.477 1000 400 47.2120 36.3557 83.0944
14 0.899995 0.8 0.8 657.2448 423.4564 1000 432.614 25.4262 34.4116 48.1493
15 0.9 0.8 0.8 1800 1800 1000 425.7529 64.9363 35.4189 115.5213
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System

16 0.899995 0.787161 0.8 1319.502 1292.852 1000 406.9967 50.8080 36.0785 89.8207
17 0.899954 0.8 0.8 1123.41 678.7381 1000 400 33.4460 36.0686 60.2063
18 0.899981 0.523338 0.46452 468.7195 300 1000 400 10.3932 33.3658 25.3967
19 0.899996 0.607953 0.790123 1498.395 435.6892 1000 411.2081 25.5686 34.9166 49.0071
20 0.899996 0.795085 0.692617 1275.527 392.8277 1000 400.2389 20.4383 35.4386 40.0208
21 0.9 0.8 0.8 1800 1800 1000 418.5825 64.0346 35.7285 113.1250
(continued)
181
Table 7.20 (continued)
182

Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)

22 0.899981 0.556517 0.715937 1103.67 1381.123 1000 400 43.9929 35.7976 82.7098
23 0.899989 0.79149 0.798764 1080.825 540.97 1000 426.7059 31.5009 34.8672 58.4120
24 0.9 0.8 0.8 1433.868 510.697 1000 400.3588 28.1693 35.8777 51.4242
25 0.9 0.68388 0.67073 597.289 1800 1000 400 42.6730 36.0409 80.2348
7 Optimization of the Selected Solar-Assisted Energy Systems
Table 7.21 Pareto-optimal solutions obtained by the Rao-1 algorithm for the solar-assisted Stirling heat engine case study-1
Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)
1 0.9 0.7993391 0.4 300 300 1000 400 9.1893 33.6236 22.6777
2 0.9 0.7998926 0.8 1800 1800 1000 400 61.1940 36.5149 106.3603
3 0.9 0.7991472 0.8 1800 1800 1000 510 70.3278 31.5780 139.2756
4 0.9 0.7904755 0.8 1800 1800 1000 510 70.1851 31.5629 139.1747
5 0.8999992 0.7990978 0.4 300 300 1000 449.84906 11.1113 32.2664 28.5796
6 0.8999948 0.7977162 0.8 1800 723.36467 1000 445.46871 42.3253 34.2993 78.4398
7 0.9 0.7999288 0.8 1427.562 976.44811 1000 431.57393 47.4430 34.9905 86.0785
8 0.8999959 0.8 0.68864 497.71392 390.70304 1000 454.36008 21.9000 33.2993 45.3083
9 0.9 0.7954883 0.8 1800 1800 1000 479.9931 69.2135 32.9783 131.3343
10 0.8999999 0.7980945 0.8 1738.5758 1800 1000 427.71157 64.7946 35.3275 115.5526
11 0.8999987 0.7998035 0.7358987 300 300 1000 400 14.6285 34.8663 29.5143
12 0.8999959 0.7999697 0.5836188 497.71392 390.70304 1000 419.71686 17.6958 34.4209 37.5064
13 0.8999987 0.8 0.7613621 764.91225 649.6838 1000 400 29.8551 35.9538 54.9369
14 0.8999996 0.7999219 0.554752 1593.3967 628.1976 1000 400 25.1472 35.7367 50.4324
15 0.8999994 0.7999531 0.8 1014.3719 963.43731 1000 400 40.8272 36.2473 72.4717
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System

16 0.9 0.794358 0.8 1800 1800 1000 492.82592 69.7394 32.3798 134.7348
17 0.8999986 0.8 0.7611377 1740.7027 740.78232 1000 400 35.8488 36.1345 65.0491
18 0.8999923 0.8 0.7461217 1516.9655 1239.4437 1000 417.19205 50.5072 35.6554 91.9435
19 0.8999995 0.7999323 0.8 1167.7503 1800 1000 400 56.9574 36.4747 99.3107
20 0.8999996 0.7999779 0.7615532 1593.3967 628.1976 1000 400 31.7681 36.0187 58.1590
21 0.9 0.7998574 0.8 1511.6418 1491.5292 1000 431.57393 58.9219 35.1131 105.8793
(continued)
183
Table 7.21 (continued)
184

Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)

22 0.8999998 0.7988243 0.7990539 1689.386 1367.0839 1000 409.67036 54.6618 36.0305 96.4075
23 0.8999951 0.7976211 0.5115517 1060.6435 440.2292 1000 430.38431 19.7992 34.1461 43.3318
24 0.8999925 0.7999699 0.7504792 1717.1978 1800 1000 417.19205 61.6812 35.7683 111.1097
25 0.8999991 0.7999573 0.7983882 784.03596 920.36155 1000 400 37.9666 36.1871 67.7496
7 Optimization of the Selected Solar-Assisted Energy Systems
Table 7.22 Pareto-optimal solutions obtained by the Rao-2 algorithm for the solar-assisted Stirling heat engine case study-1
Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)
1 0.9 0.8 0.4 300 300 1000 400 9.1902 33.6254 22.6773
2 0.9 0.8 0.8 1800 1800 1000 510 70.3417 31.5795 139.2852
3 0.9 0.8 0.8 1800 1800 1000 400 61.1952 36.5151 106.3605
4 0.89849 0.8 0.8 1800 300 1000 476.264 24.3999 32.3919 48.8891
5 0.9 0.8 0.54796 300 839.736 996.853 400 23.1091 35.5890 46.9788
6 0.89997 0.8 0.78054 833.376 1032.08 999.54 426.698 43.3582 35.1350 79.3279
7 0.89749 0.8 0.77452 434.719 300 1000 424.315 17.6961 34.0878 35.3926
8 0.89995 0.8 0.8 1800 1800 1000 475.228 69.0345 33.2031 130.1067
9 0.9 0.8 0.78212 1587.17 519.566 999.645 400 28.2197 35.8892 51.8265
10 0.9 0.8 0.8 374.233 1800 994.786 400 40.3039 36.1748 71.9599
11 0.9 0.8 0.69471 724.233 763.005 999.751 402.18 31.1674 35.9004 58.6238
12 0.8996 0.8 0.8 1800 675.108 1000 400 34.8627 36.0806 62.6978
13 0.9 0.8 0.55558 300 300 1000 400 11.9058 34.3738 26.4405
14 0.9 0.8 0.8 1800 1800 1000 447.668 67.1569 34.4544 122.3130
15 0.9 0.8 0.8 1800 733.798 999.725 400 36.9327 36.1602 66.0082
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System

16 0.89986 0.8 0.8 1690.33 1800 1000 439.919 65.7667 34.7841 118.8811
17 0.9 0.8 0.76482 614.81 1800 999.994 400 47.5994 36.3597 84.7584
18 0.9 0.8 0.44915 702.996 437.89 999.811 400 15.0281 34.8825 33.2781
19 0.9 0.8 0.8 1800 1800 1000 462.356 68.2694 33.7938 126.5347
20 0.89934 0.8 0.611 300 322.502 1000 417.985 14.4749 34.0827 31.2396
21 0.9 0.8 0.8 1800 1800 999.874 505.109 70.2179 31.8092 138.0354
(continued)
185
Table 7.22 (continued)
186

Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)

22 0.89913 0.8 0.8 1800 1645.32 1000 422.99 62.0488 35.4543 110.8163
23 0.89998 0.8 0.8 1800 1800 999.892 428.497 65.2527 35.2963 116.4362
24 0.9 0.8 0.79721 1800 1676.84 999.639 427.318 63.1123 35.3308 112.7617
25 0.89903 0.8 0.77101 1089.26 1563.81 999.879 400 51.8502 36.3396 92.2784
7 Optimization of the Selected Solar-Assisted Energy Systems
Table 7.23 Pareto-optimal solutions obtained by the Rao-3 algorithm for the solar-assisted Stirling heat engine case study-1
Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)
1 0.9 0.8 0.8 1800 1800 1000 400 61.1952 36.5151 106.3605
2 0.9 0.8 0.8 1800 1800 1000 510 70.3417 31.5795 139.2852
3 0.9 0.759994 0.4 300 300 1000 400 9.1307 33.5151 22.6933
4 0.9 0.8 0.8 1800 1800 1000 480.7015 69.3158 32.9538 131.5648
5 0.9 0.8 0.8 1800 1800 999.8908 492.5472 69.8168 32.4009 134.7371
6 0.9 0.8 0.8 1029.872 1800 1000 400 55.4940 36.4596 96.8746
7 0.9 0.602877 0.739097 679.3916 830.7619 999.8761 420.6716 33.1765 34.7873 64.0671
8 0.9 0.8 0.764364 904.36 1800 1000 400 52.8852 36.4289 93.6753
9 0.9 0.78514 0.8 1800 788.0152 1000 400.8879 38.8023 36.1386 69.3042
10 0.9 0.737009 0.8 1800 940.6118 1000 430.7336 47.2919 34.9050 86.5230
11 0.9 0.8 0.8 1800 1800 999.8031 506.6204 70.2571 31.7361 138.4435
12 0.9 0.8 0.8 1800 1800 999.9968 418.4326 64.0147 35.7349 113.0741
13 0.9 0.796108 0.8 1800 1800 1000 449.0885 67.2234 34.3836 122.7064
14 0.9 0.8 0.542715 1800 1455.145 999.7456 400 44.5934 36.2996 86.0298
15 0.9 0.8 0.727704 1800 360.5263 1000 400 20.2042 35.4445 39.1078
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System

16 0.9 0.740574 0.566445 1423.274 300 999.9941 400 13.8188 34.5943 30.0496
17 0.9 0.8 0.8 1771.044 1800 999.8016 433.0936 65.5975 35.0948 117.6151
18 0.9 0.71598 0.8 1800 809.9594 999.6809 421.3975 42.0913 35.1851 77.0150
19 0.9 0.8 0.669274 300 640.3507 1000 400 22.2237 35.5801 43.4088
20 0.9 0.587261 0.795635 1800 498.2299 1000 400 27.3234 35.4105 51.6205
21 0.9 0.8 0.8 1800 1800 1000 460.7346 68.1595 33.8673 126.0785
(continued)
187
Table 7.23 (continued)
188

Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)

22 0.9 0.771132 0.8 932.3202 1085.606 999.9201 407.9979 43.6879 35.9002 78.1181
23 0.9 0.782422 0.451933 1157.008 1042.876 999.9834 400 30.4433 35.9178 62.3189
24 0.9 0.727855 0.695886 1800 484.1864 1000 400 24.5437 35.5681 47.5053
25 0.9 0.787554 0.4 300 1763.795 1000 400 28.1942 35.8388 59.0871
7 Optimization of the Selected Solar-Assisted Energy Systems
Table 7.24 Pareto-optimal solutions obtained by the AMTPG-Jaya algorithm for the solar-assisted Stirling heat engine case study-1
Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)
1 0.9 0.7997667 0.4 300 300 1000 400 9.1899 33.6248 22.6774
2 0.9 0.8 0.8 1800 1800 1000 400 61.1952 36.5151 106.3605
3 0.9 0.8 0.8 1800 1800 1000 510 70.3417 31.5795 139.2852
4 0.9 0.8 0.7280096 300 300 1000 488.89549 17.9903 31.5698 38.8740
5 0.9 0.795675 0.7651346 1800 886.78117 1000 497.17225 50.2091 32.0413 100.6726
6 0.9 0.7998091 0.8 300 683.48999 1000 401.42904 25.3323 35.7071 46.7413
7 0.9 0.7992839 0.8 1800 1170.7675 1000 400 49.2301 36.3828 86.4613
8 0.9 0.8 0.8 1433.027 1732.6344 1000 408.77653 59.4577 36.1191 104.3307
9 0.9 0.7993774 0.7347561 635.03515 300 1000 401.32919 16.4163 35.0392 32.6046
10 0.9 0.7960857 0.7780951 1519.4813 853.20034 1000 424.0343 42.7386 35.2415 78.1069
11 0.9 0.7994338 0.4835576 1800 300 1000 400 12.1547 34.4122 27.5551
12 0.9 0.79957 0.8 1462.5721 1032.6008 1000 400 44.7435 36.3165 78.9921
13 0.9 0.8 0.8 1800 1454.501 1000 458.55906 62.3999 33.9292 115.4491
14 0.9 0.7995988 0.7500945 631.66397 391.31351 1000 400 20.2652 35.4505 38.9164
15 0.9 0.7992906 0.8 300 997.75294 1000 400 29.8357 35.9545 54.1883
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System

16 0.9 0.8 0.8 1800 1800 1000 472.80059 68.9079 33.3176 129.4248
17 0.9 0.7998415 0.8 1505.7135 835.98003 1000 400 39.5294 36.2209 70.3132
18 0.9 0.8 0.8 1800 1800 1000 479.14725 69.2404 33.0256 131.1465
19 0.9 0.8 0.8 1800 1800 1000 453.86381 67.6589 34.1771 124.1197
20 0.9 0.7972386 0.7982886 707.17944 673.4566 1000 415.03194 32.6803 35.4109 59.7417
21 0.9 0.8 0.8 1800 1800 1000 496.58112 69.9606 32.2139 135.7796
(continued)
189
Table 7.24 (continued)
190

Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)

22 0.9 0.8 0.8 1648.3672 693.08294 1000 408.92034 36.6103 35.7783 65.9608
23 0.9 0.8 0.8 1293.1109 583.83794 1000 439.81491 35.1887 34.4156 65.4908
24 0.9 0.7991705 0.8 1800 1800 1000 441.50501 66.5952 34.7267 120.4671
25 0.9 0.790367 0.8 1800 1800 1000 427.31977 65.0008 35.3318 115.9869
7 Optimization of the Selected Solar-Assisted Energy Systems
Table 7.25 Pareto-optimal solutions obtained by the SAP-Rao algorithm for the solar-assisted Stirling heat engine case study-1
Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)
1 0.9 0.8 0.8 1800 1800 999.487 400 61.1910 36.5088 106.4076
2 0.9 0.8 0.4 300 300 999.693 400 9.1888 33.6223 22.6793
3 0.9 0.8 0.8 1800 1800 1000 510 70.3417 31.5795 139.2852
4 0.9 0.79144 0.8 1800 1800 1000 492.684 69.6880 32.3812 134.6675
5 0.9 0.8 0.8 1800 1800 999.884 501.222 70.1074 31.9931 137.0180
6 0.89977 0.79562 0.8 1588.27 326.146 999.883 404.088 20.4030 35.2752 38.6770
7 0.89988 0.8 0.8 1800 835.521 998.804 417.629 42.8593 35.5050 77.2393
8 0.9 0.79767 0.8 1800 895.226 999.682 419.679 44.9869 35.4664 80.9142
9 0.9 0.8 0.8 624.594 455.955 1000 400.36 23.6571 35.6631 43.9138
10 0.9 0.79767 0.79978 1469.99 864.425 999.761 400 40.2361 36.2282 71.5335
11 0.89856 0.79957 0.78421 953.614 407.592 999.874 433.086 25.6267 34.2974 49.2339
12 0.9 0.79641 0.8 1800 1800 999.911 439.244 66.3428 34.8200 119.7760
13 0.8992 0.79966 0.79938 1052.09 598.05 999.705 400 30.4068 35.9111 55.4246
14 0.89889 0.8 0.8 1560.48 1800 1000 415.379 62.0635 35.7697 110.1454
15 0.9 0.8 0.8 1800 1800 998.867 484.438 69.4788 32.7602 132.7598
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System

16 0.89993 0.79996 0.49596 414.555 300 999.94 400 11.3677 34.2410 25.9941
17 0.89983 0.79479 0.68601 1456.87 300 999.612 400 16.3297 35.0491 33.0279
18 0.89997 0.8 0.74426 373.746 1208.09 999.92 400 34.0686 36.0832 62.4218
19 0.89998 0.79626 0.8 1800 1800 999.874 420.826 64.2788 35.6213 113.8938
20 0.9 0.8 0.4934 752.86 359.949 999.748 400 13.8936 34.7220 30.6847
21 0.9 0.8 0.8 1800 1634.38 999.621 444.846 64.3546 34.5572 117.0447
(continued)
191
Table 7.25 (continued)
192

Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)

22 0.9 0.8 0.8 1160.23 300 999.678 400 18.3419 35.2928 35.0680
23 0.89994 0.79788 0.8 1375.47 1245.01 999.631 401.416 49.3976 36.3125 86.9306
24 0.9 0.8 0.8 1800 993.484 999.571 439.187 50.1814 34.6822 91.5984
25 0.89959 0.8 0.8 593.161 925.798 998.95 405.246 36.3941 35.8853 65.6120
7 Optimization of the Selected Solar-Assisted Energy Systems
Table 7.26 Pareto-optimal solutions obtained by the ERao-1 algorithm for the solar-assisted Stirling heat engine case study-1
Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)
1 0.9 0.8 0.8 1800 1800 999.9999999 510 70.3417 31.5795 139.2852
2 0.9 0.8 0.8 1800 1800 999.9999999 510 70.3417 31.5795 139.2852
3 0.9 0.8 0.8 1800 1384.223 999.9999942 400 53.8803 36.4418 94.1895
4 0.9 0.8 0.4 300 300 1000 400 9.1902 33.6254 22.6773
5 0.9 0.8 0.8 1800 1800 999.9999978 402.7928 61.6738 36.3984 107.4368
6 0.9 0.8 0.8 1800 1800 999.9999975 464.3999 68.4037 33.7010 127.1065
7 0.9 0.8 0.8 1800 1560.526 999.9999 404.1432 57.9419 36.3058 101.3368
8 0.9 0.8 0.758302 300 484.8919 1000 403.7585 20.6328 35.3282 39.5468
9 0.9 0.8 0.8 1800 1800 1000 477.3006 69.1478 33.1108 130.6479
10 0.9 0.8 0.799987 1780.1 1800 1000 501.2853 69.9676 31.9917 136.7422
11 0.9 0.8 0.8 1800 1137.797 999.9998096 414.3926 50.7820 35.7835 90.2424
12 0.9 0.8 0.490663 1658.091 398.9937 1000 400 15.7543 34.9841 34.1529
13 0.9 0.8 0.8 1800 824.046 1000 446.8504 45.9118 34.2949 84.8385
14 0.9 0.8 0.8 1800 348.3517 1000 410.0981 22.3250 35.1873 41.9989
15 0.9 0.8 0.656474 1520.034 1001.658 1000 400 39.1493 36.2055 73.1958
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System

16 0.9 0.8 0.8 1800 1800 1000 483.3937 69.4411 32.8292 132.2865
17 0.9 0.8 0.8 999.7795 693.4501 999.9999994 403.2892 33.8632 35.9459 61.0506
18 0.9 0.8 0.605649 300 300 1000 403.382 12.9052 34.4463 27.8750
19 0.9 0.8 0.671495 568.1433 574.2821 1000 400 24.1527 35.6942 46.7317
20 0.9 0.8 0.8 976.1455 1740.567 999.9999991 402.1646 54.3748 36.3542 95.2007
21 0.9 0.8 0.6999 300 959.0352 1000 406.9635 28.2064 35.5972 53.6635
(continued)
193
Table 7.26 (continued)
194

Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)

22 0.9 0.8 0.8 1800 1036.765 999.9998973 411.1758 47.7480 35.8833 84.8526
23 0.9 0.8 0.8 1656.925 454.2977 999.9999931 400 26.0326 35.7998 47.8548
24 0.9 0.8 0.452302 300 414.7224 999.9999939 400 13.0072 34.5703 29.4453
25 0.9 0.8 0.4 793.5581 300 1000 400 9.9568 33.8660 24.1422
7 Optimization of the Selected Solar-Assisted Energy Systems
Table 7.27 Pareto-optimal solutions obtained by the ERao-2 algorithm for the solar-assisted Stirling heat engine case study-1
Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)
1 0.89999 0.79967 0.4 300 302.674 1000 402.058 9.3562 33.6054 23.0647
2 0.9 0.79976 0.8 1800 1800 1000 400 61.1924 36.5144 106.3609
3 0.9 0.8 0.8 1800 1800 1000 509.812 70.3373 31.5884 139.2364
4 0.9 0.7984 0.78778 1800 1056.17 1000 413.622 48.1912 35.7799 86.2197
5 0.9 0.79728 0.8 313.946 949.652 1000 400 29.7390 35.9465 54.0450
6 0.9 0.8 0.7849 885.621 1708.63 1000 400 52.0730 36.4199 91.6585
7 0.9 0.8 0.78267 1098.45 729.57 1000 400 34.4642 36.0994 62.2540
8 0.9 0.8 0.4 549.054 444.381 1000 400 13.5309 34.6527 30.9718
9 0.9 0.8 0.74273 1056.76 300 1000 400 17.1622 35.1743 33.7460
10 0.9 0.79803 0.78403 1273.06 918.127 1000 418.425 43.1103 35.4948 78.0850
11 0.9 0.79939 0.8 1800 1800 1000 491.404 69.7651 32.4553 134.4097
12 0.9 0.7988 0.78215 1331.8 791.613 1000 400 37.1375 36.1646 66.7524
13 0.89999 0.79876 0.78903 628.371 1181.12 1000 400 40.7100 36.2416 72.5643
14 0.89999 0.7979 0.8 300 724.382 1000 400 25.9269 35.7898 47.6964
15 0.9 0.79796 0.46828 300 300 1000 410.919 10.9771 33.7392 25.8874
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System

16 0.9 0.79649 0.8 1800 1800 1000 435.596 65.9810 34.9821 118.6415
17 0.9 0.79572 0.8 1800 1800 1000 454.517 67.6495 34.1396 124.2791
18 0.89999 0.79887 0.8 1161.01 1800 1000 400 56.8764 36.4712 99.1957
19 0.9 0.79679 0.8 1800 1800 1000 502.268 70.0878 31.9403 137.2358
20 0.9 0.8 0.8 1800 1800 1000 475.106 69.0333 33.2117 130.0530
21 0.89999 0.79946 0.7783 1474.38 378.29 1000 400 21.9403 35.5702 41.3556
(continued)
195
Table 7.27 (continued)
196

Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)

22 0.89999 0.79874 0.8 1800 1800 1000 414.186 63.4169 35.9137 111.6009
23 0.89999 0.79959 0.8 1178.3 685.771 1000 470.573 40.7407 33.1555 77.9010
24 0.9 0.8 0.8 1800 1800 1000 460.278 68.1280 33.8879 125.9497
25 0.89999 0.79684 0.78858 668.962 324.351 1000 423.058 20.2857 34.5139 39.1573
7 Optimization of the Selected Solar-Assisted Energy Systems
Table 7.28 Pareto-optimal solutions obtained by the ERao-3 algorithm for the solar-assisted Stirling heat engine case study-1
Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)
1 0.9 0.7993 0.8 1800 1800 1000 510 70.3303 31.5782 139.2776
2 0.9 0.8 0.40638 300 300 1000 400 9.3092 33.6663 22.8556
3 0.9 0.8 0.8 1800 1800 1000 510 70.3417 31.5795 139.2852
4 0.9 0.8 0.8 1800 1800 1000 400 61.1952 36.5151 106.3605
5 0.9 0.8 0.79984 1089.25 1377.47 1000 400 49.8168 36.3920 87.4332
6 0.89999 0.8 0.8 300 300 1000 479.001 18.8275 32.0670 38.4535
7 0.89999 0.79664 0.79917 1015.18 1618.7 1000 422.925 55.4980 35.4562 99.1495
8 0.9 0.8 0.8 1792.16 1800 1000 491.39 69.7198 32.4568 134.3101
9 0.9 0.79968 0.79993 1800 1800 1000 485.228 69.5157 32.7435 132.7746
10 0.89999 0.8 0.8 577.98 677.395 1000 400 30.0039 35.9612 54.4655
11 0.9 0.8 0.8 1322.66 657.441 1000 400 33.3474 36.0693 60.0257
12 0.9 0.79981 0.65757 300 300 1000 400 13.4957 34.6832 28.3266
13 0.9 0.79966 0.8 1800 980.22 1000 400.909 44.5633 36.2746 78.7565
14 0.9 0.8 0.54313 300 300 1000 400 11.7022 34.3286 26.1813
15 0.9 0.79963 0.8 1800 1800 1000 438.524 66.3147 34.8590 119.5639
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System

16 0.89999 0.79731 0.79926 1728.64 300 1000 459.957 23.5262 33.1505 46.1309
17 0.9 0.8 0.8 1800 1800 1000 457.399 67.9233 34.0177 125.1345
18 0.89999 0.8 0.8 300 390.407 1000 400 18.4383 35.3045 35.2203
19 0.9 0.79961 0.8 321.168 1800 1000 400 37.4744 36.1750 66.8981
20 0.9 0.79824 0.79972 1800 641.843 1000 413.759 35.6716 35.5464 64.6867
21 0.89999 0.79717 0.79887 365.85 1800 1000 417.639 40.4816 35.4807 73.1211
(continued)
197
Table 7.28 (continued)
198

Solution εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)

22 0.9 0.79997 0.7049 300 364.575 1000 400 16.2424 35.0690 32.6000
23 0.9 0.79897 0.8 1549.66 1632.02 1000 403.93 57.8917 36.3122 101.2500
24 0.89999 0.79952 0.8 1800 1662.64 1000 433.037 63.6254 35.0847 114.1846
25 0.9 0.8 0.49372 300 300 1000 400 10.8711 34.1284 25.0840
7 Optimization of the Selected Solar-Assisted Energy Systems
Table 7.29 Best solutions obtained by various algorithms in MOO scenario of the solar-assisted Stirling heat engine case study-1
Algorithm εR εH εL CH CL Th Tc P (kW) ηm (%) σ (W/K)
TOPSIS NSGA-II 0.9 0.8 0.8 1424 1252 996.9 400.5 49.64 36.56 89.47
LINMAP 0.9 0.8 0.8 1413 823 996.7 400.5 38.86 36.36 69.50
Bellman–Zadeh 0.9 0.8 0.8 1410 774 996.7 400.5 37.37 36.33 70.00
Jaya 0.9 0.8 0.8 1183.39 1202.477 1000 400 47.2120 36.3557 83.0944
Rao-1 0.89 0.79 0.8 1014.37 963.437 1000 400 40.8272 36.2473 72.4717
Rao-2 0.9 0.8 0.76482 614.81 1800 999.994 400 47.5994 36.3597 84.7584
Rao-3 0.9 0.8 0.76436 904.36 1800 1000 400 52.8852 36.4289 93.6753
AMTPG-Jaya 0.9 0.79 0.8 1800 1170.767 1000 400 49.2301 36.3828 86.4613
SAP-Rao 0.9 0.8 0.8 1800 1800 999.4868 400 61.1910 36.5088 106.4076
ERao-1 0.9 0.8 0.8 1800 1384.223 999.99999 400 53.8803 36.4418 94.1895
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System

ERao-2 0.89 0.79 0.8 1161.010 1800 1000 400 56.8764 36.4712 99.1957
ERao-3 0.9 0.8 0.8 1800 1800 1000 400 61.1952 36.5151 106.3605
Source NSGA-II—Ahmadi et al. (2013a)
Result in boldface indicates better values
199
200 7 Optimization of the Selected Solar-Assisted Energy Systems

Fig. 7.2 Plot of Pareto-fronts of the proposed algorithms and solutions of the methods compared
in solar-assisted Stirling heat engine system case study-1

WPM, and COPRAS) and ranked two by two methods (TOPSIS and MTOPSIS).
Here an observation can be made that the ERao-3 algorithm solution has a higher
power output, relatively similar thermal efficiency, and higher entropy generation.
Even though the ERao-3 algorithm solution has higher entropy generation than all
other solutions, it is ranked one because the relative improvement in the power output
is higher than the relative deterioration in the entropy generation with the ERao-3
algorithm solution. Here, it can be noted that the MADM methods have ranked these
solutions based on the relative importance of the solutions. The power output of the
system for the ERao-3 algorithm solution is 23.4%, 57.5%, 63.8%, 29.6%, 49.9%,
28.6%, 15.7%, 24.3%, 13.6%, and 7.6% higher than that of the NSGA-II (TOPSIS,
LINMAP, and Bellman–Zadeh), Jaya, Rao-1, Rao-2, Rao-3, AMTPG-Jaya, ERao-
1, and ERao-2 algorithms solutions, respectively. Whereas, by the ERao-3 solu-
tion the entropy generation is deteriorated by 18.8%, 53%, 51.9%, 28%, 46.8%,
25.5%, 13.5%, 23%, 12.9%, and 7.2% when compared to that of the NSGA-II
(TOPSIS, LINMAP, and Bellman–Zadeh), Jaya, Rao-1, Rao-2, Rao-3, AMTPG-
Jaya, ERao-1, and ERao-2 algorithms solutions, respectively. Besides, the thermal
efficiency is increased by the ERao-3 solution about 0.43%, 0.51%, 0.44%, 0.74%,
0.43%, 0.24%, 0.36%, 0.20%, and 0.12% when compared to that of the NSGA-II
(LINMAP and Bellman–Zadeh), Jaya, Rao-1, Rao-2, Rao-3, AMTPG-Jaya, ERao-1,
and ERao-2 algorithms solutions, respectively. Hence, the ERao-3 algorithm solution
has achieved the least average rank.
Now, the performances of the Jaya algorithm, Rao algorithms, and their modified
versions in the MOO scenario are evaluated based on the hypervolume, coverage,
and spacing indicators. Table 7.32 presents the hypervolume and spacing values of
Table 7.30 Ranks suggested by the MADM methods for different algorithm solutions presented in Table 7.29
Algorithm SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA AR CR FR
TOPSIS 12 11 9 9 1 1.5 12 3 7.3125 10.60 11
LINMAP 3 10 11 11 10 7.5 3 4 7.4375 7.60 9
Bellman–Zadeh 10 12 12 12 11 11.5 10 6 10.5625 11.2 12
Jaya 9 6 7 7 7 10 8 11 8.125 7.40 7
Rao-1 5 8 10 10 12 11.5 5 9 8.8125 7.60 8
Rao-2 11 9 8 8 6 9 11 12 9.25 9.40 10
Rao-3 8 7 5 5 4 6 9 8 6.5 6.80 6
AMTPG-Jaya 7 5 6 6 2 7.5 7 10 6.3125 6.20 5
SAP-Rao 2 2 3 3 9 3 2 2 3.25 2.40 2
ERao-1 6 4 4 4 3 5 6 7 4.875 4.80 4
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System

ERao-2 4 3 1 1 5 4 4 5 3.375 2.60 3

ERao-3 1 1 2 2 8 1.5 1 1 2.1875 1.40 1
AR average rank, CR corrected rank, FR final rank
201
 202

Table 7.31 Spearman’s rank correlation coefficients between different pairs of MADM method’s rankings presented in Table 7.30
Method SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA
SAW 1 0.699 0.510 0.510 −0.392 0.344 0.993 0.545
WPM 0.699 1 0.923 0.923 0.133 0.524 0.706 0.266
TOPSIS 0.510 0.923 1 1 0.371 0.668 0.5 0.280
MTOPSIS 0.510 0.923 1 1 0.371 0.668 0.5 0.280
VIKOR −0.392 0.133 0.371 0.371 1 0.5 −0.413 −0.119
PROMETHEE 0.344 0.524 0.668 0.668 0.5 1 0.316 0.728
COPRAS 0.993 0.706 0.5 0.5 −0.413 0.316 1 0.524
GRA 0.545 0.266 0.280 0.280 −0.119 0.728 0.524 1
7 Optimization of the Selected Solar-Assisted Energy Systems
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System 203

Table 7.32 Hypervolume

Algorithm Hypervolume Spacing
and spacing values of the
Pareto-fronts obtained by the Jaya 73,710.7 0.080901
proposed algorithms in MOO Rao-1 73,044.9 0.090238
for the solar-assisted Stirling
Rao-2 73,628.1 0.110388
engine case study-1
Rao-3 73,055.7 0.086255
AMTPG-Jaya 74,402.4 0.146992
SAP-Rao 74,487.9 0.079901
ERao-1 73,906.6 0.080655
ERao-2 74,152.9 0.090039
ERao-3 74,388.8 0.087872
Results in the bold figure indicate better values

the Pareto-fronts obtained by different algorithms in the MOO scenario. The Pareto-
front obtained by the NSGA-II algorithm was not reported by Ahmadi et al. (2013a).
Thus, the performance metric values for the NSGA-II algorithm were not presented
here. Here, an observation can be made that the SAP-Rao algorithm Pareto-front
has higher hypervolume and the lower spacing value. The SAP-Rao algorithm has a
better spacing, which is 1.2%, 11.5%, 27.61%, 7.4%, 45.6%, 1%, 11.3%, and 9.1%
lesser value when compared to that of the Jaya, Rao-1, Rao-2, Rao-3, AMTPG-
Jaya, ERao-1, ERao-2, and ERao-3 algorithms, respectively. Also, the SAP-Rao
algorithm Pareto-front has a higher hypervolume value than the other algorithms.
The AMTPG-Jaya, ERao-3, ERao-2, and ERao-1 algorithms have achieved the next
better hypervolume values, respectively. The performances of the proposed modified
algorithms are better or competitive to that of the basic algorithms in terms of the
hypervolume and spacing values.
The coverage metric values of the Pareto-fronts obtained by the Jaya algorithm,
Rao algorithms, and their modified versions are presented in Table 7.33. The ERao-1

Table 7.33 Coverage (%) values of the Pareto-fronts obtained by the proposed algorithms in MOO
for the solar-assisted Stirling engine case study-1
Algorithm Jaya Rao-1 Rao-2 Rao-3 AMTPG-Jaya SAP-Rao ERao-1 ERao-2 ERao-3
Jaya – 28 8 16 16 8 8 4 4
Rao-1 4 – 8 20 8 16 4 16 4
Rao-2 0 16 – 28 20 16 12 4 8
Rao-3 0 16 16 – 16 8 0 8 12
AMTPG-Jaya 12 24 24 32 – 16 8 8 8
SAP-Rao 0 24 20 16 16 – 8 16 12
ERao-1 16 28 8 16 8 12 – 0 0
ERao-2 12 28 16 16 12 16 8 – 8
ERao-3 12 12 20 20 24 12 8 4 –
204 7 Optimization of the Selected Solar-Assisted Energy Systems

and ERao-3 algorithms have achieved better coverage values than the other algo-
rithms compared. From the coverage values, it can be observed that three solutions
from the Pareto-fronts of the ERao-1 and ERao-3 algorithms are dominated. Simi-
larly, four solutions are dominated from the Pareto-fronts achieved by the Jaya,
SAP-Rao, and ERao-2 algorithms; five solutions are dominated from the Pareto-
fronts achieved by the AMTPG-Jaya and Rao-2 algorithms; seven solutions are
dominated from the Pareto-fronts achieved by the Rao-1 algorithm; eight solutions
are dominated from the Pareto-fronts achieved by the Rao-3 algorithm. The perfor-
mance of the proposed modified algorithms is better or competitive to that of the
basic algorithms in terms of the coverage values.
From the computational results of the solar-assisted Stirling heat engine case
study-1, it can be observed that the performance of the considered system can be
improved with the solutions achieved by the proposed algorithms. The solution of the
ERao-3 algorithm has better compromise among the power output, thermal efficiency,
and entropy generation. By the ERao-3 algorithm’s solution, the power output is
increased by 63.8%, and the thermal efficiency is increased by 1% compared to that
of the NSGA-II (TOPSIS, LINMAP, and Bellman–Zadeh) algorithm solutions. The
SAP-Rao algorithm’s performance in terms of the hypervolume and spacing values
is much better than that achieved by other algorithms. The ERao-1 and ERao-3
algorithms have achieved better performance in terms of coverage values. In addition,
the performances of the modified versions in this case study are better or competitive
to those of the basic algorithms as well as the NSGA-II algorithm.

7.2.2 Case Study-2

The description of this case study is presented in Sect. 2.2.2.2. The detailed descrip-
tion and thermodynamic analysis of the solar-assisted Stirling heat engine system
considered in this case study were presented by Ahmadi et al. (2013b). The objective
functions of this case study are thermal efficiency of solar-assisted Stirling system
(ηth ), the thermo-economic objective function (F), and the dimensionless power
output (P), which are given in Eqs. 2.29–2.32. The decision variables of these objec-
tive functions are internal irreversibility parameter (φ), heat transfer area ratio (Ar ),
temperature ratio (χ ), the temperature of the heat source (T H ), the temperature of
the working fluid in the high-temperature isothermal process (T h ).
All three objective functions considered are maximization functions. To maintain
the uniformity and to facilitate a fair comparison of the computational results of the
proposed algorithms with previous works, characteristics of the solar-assisted Stirling
system taken are the same as given by the Ahmadi et al. (2013b). The characteristics
of the solar-assisted Stirling system are as follows:
h h = h c = 200WK−1 m−2 , f = 0.7, C = 1300, δ = 5.67 × 10−8 Wm−2 K −4
 ,
−1 −2 −2
TL = 320 K, T0 = 300 K, h = 20WK m , I = 1000Wm , M1 + M2 = 1 1

2 × 10−5 sK−1 ,R = 4.3 J mol−1 K−1 , n = 1mol, λ = 2, ε = 0.9, K 0 = 2.5WK−1 ,

7.2 Optimization of a Solar-Assisted Stirling Heat Engine System 205

Table 7.34 Results obtained by the proposed algorithms in single-objective optimization scenarios
of the solar-assisted Stirling heat engine case study-2
Objective φ χ Ar TH Th P ηth F
Maximize-P 1 0.4752 10 1400 850 0.615117604 0.358187 0.1163736
Maximize-η 1 0.45 10 1100 850 0.342704804 0.4292346 0.064836
Minimize-F 1 0.47809 1.63617 1400 989.9205124 0.352630625 0.3561972 0.2072814

η0 = 0.85, 0.45 ≤ χ ≤ 0.7, 1100 ≤ TH ≤ 1400 K, 850 ≤ Th ≤ 1000 K,

0.25 ≤ Ar ≤ 10, and φ ≥ 1.
Ahmadi et al. (2013b) solved this MOO problem using the NSGA-II algorithm and
reported three Pareto-optimal solutions. These three solutions are documented based
on the similarity to the ideal solutions calculated by employing TOPSIS, LINMAP,
and Fuzzy Bellman–Zadeh decision-making methods. The number of function eval-
uations taken by the NSGA-II algorithm was not specified. However, in this work,
the Jaya algorithm, Rao algorithms, and their modified versions are tested by taking
20,000 function evaluations as the termination criterion. The best Pareto-optimal
solutions are reported using the average rank based on multiple decision-making
methods. Similar to the previous case studies, in this case study also, first single-
objective optimization is performed independently for each objective function, and
then MOO is performed by the proposed algorithms. In all the computational exper-
iments, the population size is maintained as 25 for all the algorithms. The elite size
for the elitist Rao algorithms is taken as 20%.
The Jaya algorithm, Rao algorithms, and their modified versions have obtained
identical solutions in all the single-objective optimization scenarios of this case study,
and those solutions are presented in Table 7.34. The maximum non-dimensional
power output, thermo-economic function, and thermal efficiency obtained by the
proposed algorithms are 0.615117604, 0.2072814, and 0.4292346, respectively.
The Pareto-optimal solutions achieved by the Jaya algorithm, Rao algorithms, and
their modified versions in MOO are presented in Tables 7.35, 7.36, 7.37, 7.38, 7.39,
7.40, 7.41, 7.42 and 7.43. Here an observation can be made that the non-dimensional
power output is varied between 0.17014 and 0.61512, the thermal efficiency is varied
between 0.292 and 0.4293, and non-dimensional thermo-economic function is varied
from 0.0616 to 0.20725. The best solutions from the Pareto-fronts obtained by the
proposed algorithms are identified based on the average rank method. Solution 9 of
the Jaya algorithm, Solution 16 of the Rao-1 algorithm, Solution 9 of the Rao-2 algo-
rithm, Solution 12 of the Rao-3 algorithm, Solution 8 of the AMTPG-Jaya algorithm,
Solution 14 of the SAP-Rao algorithms, Solution 16 of the ERao-1 algorithm, Solu-
tion 9 of the ERao-2, and Solution 16 of the ERao-3 algorithm are identified as the
best solutions from the respective algorithm Pareto-front. Now, these best solutions
are compared with those of the NSGA-II algorithm in Table 7.44.
In Table 7.44, the solutions with the best average rank from the Pareto-optimal
solutions of the proposed algorithms are compared with those reported by the NSGA-
II algorithm. Figure 7.3 presents the Pareto-fronts achieved by the AMTPG-Jaya,
206 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.35 Pareto-optimal solutions obtained by the Jaya algorithm for the solar-assisted Stirling
heat engine case study-2
Solution φ χ Ar TH Th P ηth F
1 1 0.47243 1.66268 1400 989.501 0.3548535 0.3600603 0.2072045
2 1 0.47519 10 1400 850 0.6151176 0.3581945 0.1163736
3 1 0.45 9.8232 1100 850 0.3418216 0.4292344 0.0656095
4 1 0.57164 4.8373 1400 871.554 0.4805860 0.2923630 0.1563832
5 1 0.45 0.79922 1202.84 1000 0.1798537 0.4147325 0.1339669
6 1 0.53305 5.42186 1400 852.312 0.5195827 0.3187032 0.1563288
7 1 0.45 1.09691 1288.87 1000 0.2486357 0.3996464 0.1691279
8 1 0.5081 2.45764 1214.43 855.798 0.3167602 0.3692959 0.1542707
9 1 0.48666 2.51716 1400 901.694 0.4103274 0.3503541 0.1973881
10 1 0.45 2.2319 1100 857.86 0.2298330 0.4292051 0.1174698
11 1 0.51777 3.84643 1400 866.249 0.4739180 0.3291281 0.1789402
12 1 0.45 9.75773 1370.65 871.385 0.5850816 0.3823743 0.1129090
13 1 0.51106 6.36388 1400 860.526 0.5483205 0.3337116 0.1471062
14 1 0.45 1.10705 1349.82 989.018 0.2720724 0.3870409 0.1845249
15 1 0.50426 7.59815 1400 869.281 0.5714123 0.3383530 0.1342494
16 1 0.50777 5.69527 1362.65 863.752 0.5092648 0.3438474 0.1480064
17 1 0.45 7.62834 1238.33 856.033 0.4504540 0.4089154 0.1055103
18 1 0.45 10 1176.52 850 0.4214861 0.4188383 0.0797406
19 1 0.46591 1.28909 1400 1000 0.3170798 0.3645035 0.2042427
20 1 0.45 10 1138.71 850 0.3837377 0.4242221 0.0725990
21 1 0.45625 1.18381 1300.16 999.708 0.2633766 0.3929243 0.1747284
22 1 0.45 2.11063 1125.15 885.101 0.2392401 0.4260053 0.1256145
23 1 0.46839 5.08283 1291.51 863.271 0.4474201 0.3858162 0.1407708
24 1 0.48211 8.25916 1400 850 0.5893103 0.3534718 0.1298143
25 1 0.45 7.67138 1190.33 875.933 0.4028174 0.4167435 0.0939465

SAP-Rao, and elitist Rao algorithms, including the optimal solutions reported for
the NSGA-II, Jaya, and Rao algorithms. The conflicting nature of these objectives
can be observed in Fig. 7.3. Any change in the design variables, which will result
in the improvement of one objective, is also leading to the deterioration of the other
two objectives, which is not desirable.
Furthermore, the non-dimensional power output of the system for the Rao-2 algo-
rithm solution is highest, which is 38.1%, 32.6%, 34%, 46.7%, 47.3%, 26.5%, 21%,
33%, 40.5%, 37%, and 38.5% higher than that of the NSGA-II (TOPSIS, LINMAP,
and Fuzzy), Jaya, Rao-1, Rao-3, AMTPG-Jaya, SAP-Rao, ERao-1, ERao-2, and
ERao-3 algorithms solutions, respectively. Similarly, the thermo-economic function
of the system for the Rao-1 algorithm solution is highest, which is 7%, 12%, 26%,
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System 207

Table 7.36 Pareto-optimal solutions obtained by the Rao-1 algorithm for the solar-assisted Stirling
heat engine case study-2
Solution φ χ Ar TH Th P ηth F
1 1 0.47522 10 1400 850 0.6151176 0.3581761 0.1163736
2 1 0.48054 1.62447 1400 983.749 0.3515125 0.3545243 0.2072350
3 1 0.54076 4.32582 1400 850 0.4848592 0.3134404 0.1698922
4 1 0.45 10 1100 865.777 0.3298162 0.4292333 0.0623977
5 1 0.45917 1.26622 1100 899.987 0.1841288 0.4220305 0.1193575
6 1 0.48579 3.02882 1400 864.08 0.4285399 0.3509485 0.1864785
7 1 0.45 1.78525 1188.78 915.915 0.2567197 0.4169539 0.1454414
8 1 0.50046 5.86113 1400 861.614 0.5395252 0.3409488 0.1536272
9 1 0.46217 4.58789 1267.55 850 0.4133015 0.3947666 0.1393352
10 1 0.45092 9.1107 1162.74 868.512 0.3913541 0.4201571 0.0797935
11 1 0.45 2.77357 1347.49 996.766 0.3898552 0.3875787 0.1781242
12 1 0.45989 7.87454 1173.7 850.477 0.4014021 0.4117169 0.0917532
13 1 0.45189 2.14718 1100 919.42 0.2131402 0.4277327 0.1109979
14 1 0.45 1.82094 1276.04 951.23 0.3036086 0.4021049 0.1705281
15 1 0.45387 8.37204 1250.51 864.38 0.4698091 0.4039235 0.1023992
16 1 0.45 2.49222 1399.45 948.093 0.4088411 0.3755073 0.1976900
17 1 0.52486 4.71035 1400 850 0.5021918 0.3242925 0.1663590
18 1 0.53429 7.19616 1400 850 0.5551247 0.3178606 0.1359245
19 1 0.46721 10 1382.89 850 0.6014358 0.3676388 0.1137851
20 1 0.47044 2.26178 1399.45 941.75 0.3978422 0.3615525 0.2020185
21 1 0.46192 5.06017 1343.27 874.853 0.4792003 0.3800878 0.1512319
22 1 0.46381 2.30579 1347.94 1000 0.3674042 0.3777486 0.1847929
23 1 0.45 1.09509 1164.53 900.588 0.1930444 0.4205449 0.1313831
24 1 0.45655 3.36859 1382.78 921.189 0.4462364 0.3750051 0.1826083
25 1 0.45 8.36698 1100 850 0.3333641 0.4292328 0.0726941

0.2%, 74%, 21%, 17%, 8%, 0.6%, 4.8%, and 2% higher than that of the NSGA-
II (TOPSIS, LINMAP, and Fuzzy), Jaya, Rao-2, Rao-3, AMTPG-Jaya, SAP-Rao,
ERao-1, ERao-2, and ERao-3 algorithms solutions, respectively. The thermal effi-
ciency of the system for the Fuzzy solution is higher than that of the other algorithm’s
solutions. As these solutions are non-dominated, to identify the best solution among
the solutions of different algorithms, which has the best compromise among the
objectives, the MADM methods-based average ranks are calculated and presented
in Table 7.45.
The decision-making methods ranking of the Pareto-optimal solutions obtained
by different algorithms is shown in Table 7.45. Spearman’s correlation coefficients
for different pairs of rankings given by decision-making methods for different algo-
rithm’s solutions are shown in Table 7.46. The ranks given by the TOPSIS and
208 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.37 Pareto-optimal solutions obtained by the Rao-2 algorithm for the solar-assisted Stirling
heat engine case study-2
Solution φ χ Ar TH Th P ηth F
1 1 0.47535 10 1400 850 0.6151174 0.3580891 0.1163736
2 1 0.48467 1.67085 1400 983.613 0.3555712 0.3517084 0.2072000
3 1 0.55429 4.48677 1400 850 0.4845037 0.3042074 0.1657613
4 1 0.45 10 1176.72 850 0.4216784 0.4188087 0.0797770
5 1 0.45 0.73455 1190.15 963.299 0.1701430 0.4167082 0.1294055
6 1 0.45 3.71136 1100 850 0.2756510 0.4292195 0.1064050
7 1 0.45141 1.387 1400 1000 0.3252330 0.3743961 0.2039807
8 1 0.45 10 1337.86 850 0.5607651 0.3896826 0.1060907
9 1 0.45208 10 1391.63 850 0.6020914 0.3759945 0.1139092
10 1 0.45 6.56738 1131.2 850 0.3473813 0.4252275 0.0910665
11 1 0.45 10 1214.81 850 0.4575556 0.4128630 0.0865646
12 1 0.49313 6.3795 1399.22 906.063 0.5444270 0.3461292 0.1457999
13 1 0.457 1.15377 1240.73 982.986 0.2344893 0.4032685 0.1569041
14 1 0.45041 2.80281 1352.93 998.311 0.3944172 0.3860868 0.1791825
15 1 0.53276 5.68522 1400 850 0.5262827 0.3188994 0.1531439
16 1 0.50965 1.91942 1400 954.327 0.3734701 0.3346602 0.2049099
17 1 0.45 1.86284 1100 868.388 0.2141476 0.4291996 0.1190795
18 1 0.53412 6.63805 1400 850 0.5455282 0.3179733 0.1418844
19 1 0.47145 5.95551 1252.29 850 0.4404524 0.3906165 0.1239885
20 1 0.45 1.01229 1142.15 926.471 0.1805630 0.4236977 0.1259297
21 1 0.45931 1.50596 1259.99 941.329 0.2750647 0.3982035 0.1671706
22 1 0.55156 4.19592 1400 850 0.4763056 0.3060659 0.1702154
23 1 0.53359 2.8402 1389.5 850 0.4151693 0.3204855 0.1872470
24 1 0.54203 2.30053 1398.38 963.251 0.3857805 0.3129021 0.1942558
25 1 0.52513 2.4346 1400 850 0.3925763 0.3240960 0.1921192

MTOPSIS methods for each solution are identical. Furthermore, Spearman’s corre-
lation for all the pairs formed by the SAW, WPM, TOPSIS, MTOPSIS, and COPRAS
methods is greater than 0.5. Spearman’s correlation values for all the pairs formed by
the VIKOR, PROMETHEE, and GRA methods are either negative or very nearer to
zero. Hence, the ranks suggested by the VIKOR, PROMETHEE, and GRA methods
cannot be considered for calculating the average ranks. Now, the corrected average
ranks are calculated, excluding the ranks suggested by the VIKOR, PROMETHEE,
and GRA methods and presented in Table 7.45 as the corrected ranks.
The ERao-3 algorithm solution has the least average rank, which is 1.4. Thus, it can
be regarded as the best solution. The AMTPG-Jaya, ERao-1, SAP-Rao, and ERao-2
algorithms’ solutions have the next best average ranks, which are 2.6, 3.2, 3.8, and
4.4, respectively. The ERao-3 algorithm solution ranked 1 by three methods (SAW,
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System 209

Table 7.38 Pareto-optimal solutions obtained by the Rao-3 algorithm for the solar-assisted Stirling
heat engine case study-2
Solution φ χ Ar TH Th P ηth F
1 1 0.45 1.69862 1400 1000 0.3556207 0.3753673 0.2058012
2 1 0.45 10 1400 876.669 0.6116321 0.3753879 0.1157142
3 1 0.45 1.89605 1100 887.321 0.2151054 0.4292021 0.1186729
4 1 0.45 10 1100 850 0.3427048 0.4292346 0.0648360
5 1 0.45 4.60016 1325.77 944.944 0.4463408 0.3922457 0.1502074
6 1 0.45 2.45894 1380.49 966.273 0.3994052 0.3800659 0.1944683
7 1 0.45 4.37848 1222.09 1000 0.3075491 0.4116512 0.1069181
8 1 0.45 3.90207 1315.52 934.754 0.4225595 0.3943697 0.1581249
9 1 0.45 10 1387.47 850 0.5976537 0.3784222 0.1130696
10 1 0.45 2.55625 1342.04 946.331 0.3831792 0.3887658 0.1828552
11 1 0.45 1.89216 1138.01 901.326 0.2354851 0.4242865 0.1300357
12 1 0.45 4.45221 1366.51 925.322 0.4760553 0.3833191 0.1637003
13 1 0.45 8.35976 1188.52 885.494 0.4014555 0.4170229 0.0876014
14 1 0.45 7.04411 1396.61 914.073 0.5603855 0.3762152 0.1394374
15 1 0.45 8.76887 1178.92 911.711 0.3731480 0.4184761 0.0784240
16 1 0.45 6.11491 1241.44 997.089 0.3465512 0.4083705 0.0957145
17 1 0.45 2.05639 1275.46 989.029 0.3104563 0.4022167 0.1650214
18 1 0.45 5.92725 1289.27 909.757 0.4547534 0.3996025 0.1284524
19 1 0.45 4.04331 1256.3 976.134 0.3582437 0.4057442 0.1310881
20 1 0.45 9.38173 1155.99 906.705 0.3542621 0.4218193 0.0705597
21 1 0.45 1.86434 1298.75 918.927 0.3119856 0.3977249 0.1734213
22 1 0.45 5.97463 1348.65 980.338 0.4732221 0.3873317 0.1329069
23 1 0.45 6.48322 1400 850.588 0.5335651 0.3753805 0.1412100
24 1 0.45 1.22715 1361.97 958.164 0.2831489 0.3843213 0.1855593
25 1 0.45 4.39652 1360.78 906.79 0.4695661 0.3846194 0.1628050

WPM, and COPRAS) and ranked 2 by two methods (TOPSIS and MTOPSIS). Here
an observation can be made that the ERao-3 algorithm solution has a better fitness
value in none of the three objectives. However, it is ranked one because it has better
compromise among the objectives. Here, it can be noted that the MADM methods
have ranked these solutions based on the relative importance of the solutions. If the
NSGA-II (TOPSIS) and ERao-3 algorithms solutions are compared, the NSGA-II
(TOPSIS) solution has better power output and thermal efficiency, and the ERao-3
algorithm has a better thermo-economic function. However, the combined deterio-
ration in the power output (0.3%) and thermal efficiency (2.25%) is relatively much
lesser than the improvement in the thermo-economic function (4.9%) by the ERao-
3 solution. Similarly, the ERao-3 algorithm solution has better relative importance
when compared to the solutions of the other algorithms.
210 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.39 Pareto-optimal solutions obtained by the AMTPG-Jaya algorithm for the solar-assisted
Stirling heat engine case study-2
Solution φ χ Ar TH Th P ηth F
1 1 0.47523 10 1400 850 0.6151176 0.3581708 0.1163736
2 1 0.47429 1.70012 1400 984.776 0.3581183 0.3587884 0.2071696
3 1 0.54385 3.37925 1400 874.013 0.4479567 0.3113258 0.1829702
4 1 0.45 1.06142 1149.28 910.127 0.1866062 0.4227112 0.1282612
5 1 0.45 10 1100 870.232 0.3257729 0.4292329 0.0616327
6 1 0.4707 6.99366 1400 917.136 0.5592433 0.3612581 0.1399058
7 1 0.45 9.98302 1201.36 889.024 0.4233368 0.4150250 0.0802012
8 1 0.45377 4.54838 1393.86 923.845 0.4981830 0.3742949 0.1689154
9 1 0.53189 2.9567 1400 893.116 0.4322267 0.3194890 0.1906471
10 1 0.48782 5.0417 1400 891.645 0.5190372 0.3495748 0.1642143
11 1 0.51059 3.55161 1400 856.574 0.4595718 0.3340258 0.1822166
12 1 0.45 1.48514 1285.29 993.68 0.2814204 0.4003456 0.1719659
13 1 0.45473 10 1274.38 850 0.5112880 0.3989772 0.0967302
14 1 0.49373 3.94899 1400 933.421 0.4790397 0.3455400 0.1779214
15 1 0.49536 8.8077 1400 850 0.5960797 0.3444305 0.1248405
16 1 0.50232 6.37904 1400 879.452 0.5486068 0.3396782 0.1469270
17 1 0.45 1.35865 1244.47 1000 0.2478882 0.4078197 0.1566653
18 1 0.45 1.55317 1364.8 977.08 0.3245880 0.3836902 0.1948723
19 1 0.45 10 1157.6 917.761 0.3473153 0.4215906 0.0657083
20 1 0.52618 1.93654 1400 917.31 0.3705050 0.3233798 0.2024679
21 1 0.45 1.57438 1121.24 884.211 0.2100277 0.4265042 0.1254094
22 1 0.45848 6.52616 1195.05 850 0.4032398 0.4095936 0.1062017
23 1 0.47417 4.65899 1316.61 863.824 0.4530867 0.3768290 0.1511948
24 1 0.45 7.77134 1345.71 850 0.5306336 0.3879771 0.1225319
25 1 0.52424 5.6177 1400 850 0.5278638 0.3247167 0.1549084

In addition, the non-dimensional power output of the system for the ERao-3 algo-
rithm solution is 6%, 6.4%, and 1.5% higher than that of the Jaya, Rao-1, and ERao-1
solutions, respectively. Also, the non-dimensional thermo-economic function of the
system for the ERao-3 solution is 4.9%, 9.8%, 23.5%, 70.1%, 18.4%, 14.7%, 6%,
and 2.8% higher than that of the NSGA-II (TOPSIS, LINMAP, and Bellman–Zadeh),
Rao-2, Rao-3, AMTPG-Jaya, SAP-Rao, and ERao-2 algorithms solutions, respec-
tively. In addition, the ERao-3 algorithm solution’s thermal efficiency is 7.14%,
0.29%, and 1.62% higher when compared to that of the Jaya, AMTPG-Jaya, and
ERao-1 algorithms solutions, respectively. Hence, the ERao-3 algorithm solution
can be considered as the best solution.
Now, the performances of the Jaya algorithm, Rao algorithms, and their modified
versions in the MOO scenario are evaluated based on the hypervolume, coverage,
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System 211

Table 7.40 Pareto-optimal solutions obtained by the SAP-Rao algorithm for the solar-assisted
Stirling heat engine case study-2
Solution φ χ Ar TH Th P ηth F
1 1 0.45 1.71204 1400 1000 0.3567995 0.3753675 0.2057987
2 1 0.45 10 1400 876.612 0.6116322 0.3753879 0.1157142
3 1 0.45 1.24252 1145.2 876.651 0.1935361 0.4232790 0.1262873
4 1 0.45 10 1100 850 0.3427048 0.4292346 0.0648360
5 1 0.45 1.2553 1300.33 992.743 0.2696118 0.3974069 0.1753017
6 1 0.45 10 1171 850 0.4161101 0.4196557 0.0787235
7 1 0.45 1.92037 1285.48 894.594 0.3019813 0.4003063 0.1656492
8 1 0.45 2.23873 1100 862.539 0.2305106 0.4292059 0.1176402
9 1 0.45 7.01104 1375.63 850 0.5330005 0.3812121 0.1330927
10 1 0.45 5.61806 1383.96 875.37 0.5138091 0.3792526 0.1507770
11 1 0.45 10 1260.89 878.674 0.4898969 0.4049275 0.0926832
12 1 0.45 2.9261 1311.57 937.723 0.3837980 0.3951712 0.1702709
13 1 0.45 1.80745 1356.36 973.487 0.3426676 0.3856023 0.1930934
14 1 0.45 3.44796 1393.93 981.638 0.4531617 0.3768601 0.1828964
15 1 0.45 2.05026 1261.23 988.123 0.3006076 0.4048442 0.1600097
16 1 0.45 2.42015 1194.09 884.873 0.2909584 0.4161412 0.1428223
17 1 0.45 7.5917 1221.14 850 0.4364477 0.4118194 0.1026070
18 1 0.45 2.91216 1204.13 859.136 0.3117243 0.4145640 0.1386632
19 1 0.45 8.95597 1242.3 908.541 0.4456168 0.4082301 0.0921025
20 1 0.45 10 1364.2 893.804 0.5790010 0.3838529 0.1095407
21 1 0.45 10 1149.17 850.615 0.3940691 0.4227820 0.0745536
22 1 0.45 10 1120.52 850 0.3647684 0.4266386 0.0690102
23 1 0.45 10 1233.54 850 0.4744574 0.4097393 0.0897622
24 1 0.45 3.9718 1326.75 868.76 0.4216985 0.3920336 0.1560574
25 1 0.45 5.68606 1225.44 906.469 0.3968878 0.4111016 0.1154790

and spacing indicators. Table 7.47 presents the hypervolume and spacing values of
the Pareto-fronts obtained by different algorithms in the MOO scenario. The Pareto-
front obtained by the NSGA-II algorithm was not reported by Ahmadi et al. (2013b).
Thus, the performance metric values for the NSGA-II algorithm were not presented
here. Here, an observation can be made that the SAP-Rao algorithm Pareto-front
has the highest hypervolume value, and the AMTPG-Jaya algorithm Pareto-front
has the least spacing value. The SAP-Rao algorithm has 5.72%, 5.8%, 4.74%, 6.1%,
4.6%, 4.7%, 1.6%, and 4.7% higher hypervolume value when compared to that
of the Jaya, Rao-1, Rao-2, Rao-3, AMTPG-Jaya, ERao-1, ERao-2, and ERao-3
algorithms, respectively. The ERao-2, AMTPG-Jaya, ERao-3, and ERao-1 algo-
rithms have achieved the next better hypervolume values, respectively. Similarly, the
AMTPG-Jaya algorithm has 18%, 20.7%, 24.6%, 17.2%, 13.1%, 7%, 4.14%, and
212 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.41 Pareto-optimal solutions obtained by the ERao-1 algorithm for the solar-assisted
Stirling heat engine case study-2
Solution φ χ Ar TH Th P ηth F
1 1 0.47707 1.64488 1400 995.219 0.3533372 0.3568923 0.2072421
2 1 0.47437 10 1400 850 0.6151102 0.3587564 0.1163722
3 1 0.45 1.60377 1100 894.457 0.2007590 0.4291962 0.1189803
4 1 0.45 9.93364 1100 850 0.3423765 0.4292345 0.0651243
5 1 0.53359 3.73404 1400 893.423 0.4647414 0.3183302 0.1787259
6 1 0.52033 4.02608 1400 872.952 0.4807894 0.3273836 0.1764064
7 1 0.48953 5.69076 1399.8 860.51 0.5356768 0.3484499 0.1557699
8 1 0.45 1.58619 1219.23 895.985 0.2527946 0.4121014 0.1504912
9 1 0.45603 9.20774 1400 921.443 0.5910918 0.3712751 0.1195049
10 1 0.46265 3.05367 1321.07 887.254 0.3934555 0.3841716 0.1704217
11 1 0.4511 10 1211.41 850 0.4546602 0.4125888 0.0860168
12 1 0.4541 10 1283.77 850 0.5190538 0.3976800 0.0981994
13 1 0.45 1.39254 1175.48 906.515 0.2235552 0.4189527 0.1400017
14 1 0.45 4.34434 1128.28 883.008 0.3009289 0.4256097 0.1051514
15 1 0.48922 7.75176 1400 850 0.5802420 0.3486192 0.1342475
16 1 0.4588 2.75666 1400 969.261 0.4285094 0.3693724 0.1964355
17 1 0.45806 9.25184 1341.88 877.305 0.5555832 0.3831168 0.1118983
18 1 0.45226 1.60377 1297.87 894.457 0.2801840 0.3962575 0.1660518
19 1 0.45 9.25184 1149.88 877.305 0.3724772 0.4226819 0.0750195
20 1 0.45286 2.34635 1372.1 906.5 0.3738316 0.3800321 0.1863959
21 1 0.45 5.88401 1298.82 892.381 0.4653791 0.3977351 0.1321455
22 1 0.47194 1.45158 1400 985.773 0.3342937 0.3603890 0.2060865
23 1 0.46402 5.0195 1400 850 0.5002558 0.3658123 0.1587502
24 1 0.45 1.45158 1379.3 985.773 0.3215900 0.3803318 0.1982548
25 1 0.46983 4.34434 1156.37 883.008 0.3259623 0.4065528 0.1138987

38% less spacing value when compared to that of the Jaya, Rao-1, Rao-2, Rao-3,
SAP-Rao, ERao-1, ERao-2, and ERao-3 algorithms, respectively. The performances
of the proposed modified algorithms are better or competitive to that of the basic
algorithms in terms of the hypervolume and spacing values.
The coverage metric values of the Pareto-fronts obtained by the Jaya algorithm,
Rao algorithms, and their modified versions are presented in Table 7.48. The ERao-2
algorithm has achieved better coverage values than the other algorithms compared.
From the coverage values, it can be observed that two solutions from the Pareto-front
of the ERao-2 algorithm are dominated. Similarly, three solutions of the Pareto-fronts
achieved by the SAP-Rao and ERao-1 algorithms are dominated, and four solutions
of the Pareto-front achieved by the ERao-3 algorithm are dominated. Also, 7, 6, 8,
6, and 5 solutions from the Pareto-fronts achieved by the Jaya, Rao-1, Rao-2, Rao-3,
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System 213

Table 7.42 Pareto-optimal solutions obtained by the ERao-2 algorithm for the solar-assisted
Stirling heat engine case study-2
Solution φ χ Ar TH Th P ηth F
1 1 0.45 1.69724 1400 1000 0.3554984 0.3753673 0.2058011
2 1 0.45 10 1400 876.611 0.6116322 0.3753879 0.1157142
3 1 0.45 1.61005 1110.87 882.977 0.2071486 0.4278372 0.1225717
4 1 0.45 9.92067 1100 850 0.3423118 0.4292345 0.0651810
5 1 0.45 1.28658 1371.47 1000 0.3027167 0.3821551 0.1951261
6 1 0.45 10 1317.06 867.94 0.5439979 0.3940618 0.1029185
7 1 0.45 10 1161.31 850 0.4065615 0.4210646 0.0769170
8 1 0.45 3.19021 1355.88 945.794 0.4223584 0.3857184 0.1784187
9 1 0.45 3.1075 1400 1000 0.4396704 0.3753790 0.1885551
10 1 0.45 10 1365 867.672 0.5836720 0.3836699 0.1104244
11 1 0.45 4.65624 1390.89 1000 0.4778360 0.3775984 0.1595163
12 1 0.45 3.63427 1333.79 973.544 0.4172122 0.3905473 0.1631300
13 1 0.45 5.85474 1350.76 912.226 0.5018472 0.3868653 0.1430101
14 1 0.45 10 1124.11 850 0.3685645 0.4261692 0.0697284
15 1 0.45 2.84316 1262.88 850 0.3274019 0.4045388 0.1475783
16 1 0.45 1.00523 1229.68 997.307 0.2128587 0.4103494 0.1487679
17 1 0.45 10 1242.35 850 0.4822558 0.4082209 0.0912376
18 1 0.45 7.5421 1382 866.634 0.5565623 0.3797197 0.1315026
19 1 0.45 10 1171.6 850 0.4166949 0.4195676 0.0788342
20 1 0.45 2.28606 1291.95 982.339 0.3345231 0.3990693 0.1689734
21 1 0.45 1.65632 1205.04 1000 0.2366704 0.4144083 0.1384157
22 1 0.45 10 1279.03 850 0.5136592 0.4015648 0.0971788
23 1 0.45 3.37096 1303.44 850 0.3706979 0.3968012 0.1516335
24 1 0.45 1.61418 1243.68 907.907 0.2676593 0.4079549 0.1582108
25 1 0.45 3.55997 1100 850 0.2720516 0.4292185 0.1077132

and AMTPG-Jaya algorithms are dominated, respectively. The performance of the

proposed modified algorithms is better or competitive to that of the basic algorithms
in terms of the coverage values.
From the computational results of the solar-assisted Stirling heat engine case
study-2, it can be observed that the performance of the considered system can be
modified with the solutions achieved by the proposed algorithms. The solution of the
ERao-3 algorithm has better compromise among the non-dimensional power output,
thermal efficiency, and thermo-economic function. By the ERao-3 algorithm solu-
tion, the power output is increased by 6.4%, thermal efficiency is increased by 7.14%,
and the thermo-economic function is increased by 70% compared to those of the solu-
tions of the other algorithms compared. The SAP-Rao algorithm’s performance in
214 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.43 Pareto-optimal solutions obtained by the ERao-3 algorithm for the solar-assisted
Stirling heat engine case study-2
Solution φ χ Ar TH Th P ηth F
1 1 0.47451 1.62478 1400 1000 0.3514687 0.3586402 0.2071931
2 1 0.4751 10 1400 850 0.6151175 0.3582596 0.1163736
3 1 0.45 10 1100 850 0.3427045 0.4292343 0.0648360
4 1 0.45291 1.77006 1100 861.682 0.2102446 0.4269176 0.1195524
5 1 0.53759 3.96958 1400 850 0.4729601 0.3155985 0.1750894
6 1 0.47884 7.9201 1400 850.581 0.5831853 0.3556895 0.1327132
7 1 0.51889 3.46544 1400 850 0.4537630 0.3283602 0.1825868
8 1 0.46253 5.47929 1400 894.458 0.5302809 0.3668325 0.1583746
9 1 0.45208 10 1244.99 850 0.4851544 0.4062189 0.0917860
10 1 0.53545 4.92092 1399.5 850 0.5050461 0.3171695 0.1624482
11 1 0.45919 2.09707 1400 993.046 0.3878196 0.3690938 0.2042506
12 1 0.49898 3.64363 1400 850 0.4601919 0.3419509 0.1796533
13 1 0.50407 6.35264 1400 870.121 0.5490857 0.3384836 0.1475022
14 1 0.50809 7.22694 1400 859.147 0.5659126 0.3357389 0.1381198
15 1 0.45 10 1345.76 850 0.5668061 0.3879677 0.1072336
16 1 0.45 2.90187 1400 972.377 0.4348128 0.3753697 0.1937964
17 1 0.45 7.3832 1207.84 850 0.4228139 0.4139878 0.1015348
18 1 0.45 1.88165 1335.52 912.826 0.3245572 0.3901581 0.1796687
19 1 0.46553 3.45372 1224.33 867.184 0.3555492 0.3996534 0.1433570
20 1 0.4591 2.16799 1347.44 860.828 0.3275047 0.3811583 0.1697673
21 1 0.4521 0.98559 1393.84 1000 0.2715280 0.3754085 0.1908946
22 1 0.45413 1.63117 1134.45 879.804 0.2209580 0.4215629 0.1300461
23 1 0.45 1.41582 1196.85 903.324 0.2329955 0.4156938 0.1450078
24 1 0.46382 1.45207 1340.21 1000 0.3075412 0.3793704 0.1895693
25 1 0.45 1.50705 1278.04 1000 0.2777241 0.4017240 0.1687390

terms of the hypervolume is much better than that achieved by other algorithms. Simi-
larly, The AMTPG-Jaya algorithm’s performance in terms of the spacing is much
better than that achieved by other algorithms. The ERao-2 algorithm has achieved
better performance in terms of coverage values. In addition, the performances of the
modified versions are better or competitive to those of the basic algorithms as well
as the NSGA-II algorithm.
Table 7.44 Best solutions obtained by various algorithms in MOO scenario of the solar-assisted Stirling heat engine case study-2
Algorithm φ χ Ar TH Th P ηth F
TOPSIS NSGA-II 1.002271 0.45466 3.17578 1371.415 920.6022 0.436108 0.384027 0.184709
LINMAP 1.001516 0.45921 3.67126 1371.466 919.8346 0.454229 0.381995 0.176510
Fuzzy 1.000984 0.45146 4.34813 1329.38 913.8662 0.449328 0.396756 0.156916
Jaya 1 0.4867 2.5172 1400.0000 901.6943 0.410327 0.350354 0.197388
Rao-1 1 0.4500 2.4922 1399.4499 948.0930 0.408841 0.375507 0.197690
Rao-2 1 0.4521 10.0000 1391.6298 850.0000 0.602091 0.375995 0.113909
Rao-3 1 0.4500 4.4522 1366.5084 925.3221 0.476055 0.383319 0.163700
AMTPG-Jaya 1 0.4538 4.5484 1393.8580 923.8445 0.498183 0.374295 0.168915
SAP-Rao 1 0.4500 3.4480 1393.9301 981.6382 0.453162 0.376860 0.182896
ERao-1 1 0.4588 2.7567 1400.0000 969.2610 0.428509 0.369372 0.196436
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System

ERao-2 1 0.4500 3.1075 1400.0000 1000.0000 0.439670 0.375379 0.188555

ERao-3 1.000023922 0.4500 2.9019 1400.0000 972.3766 0.434813 0.375370 0.193796
Source NSGA-II—Ahmadi et al. (2013b)
Result in boldface indicates better values
215
216 7 Optimization of the Selected Solar-Assisted Energy Systems

Fig. 7.3 Plot of Pareto-fronts of the proposed algorithms and solutions of the methods compared
in solar-assisted Stirling heat engine system case study-2

7.2.3 Case Study-3

The description of this case study is presented in Sect. 2.2.2.3. The detailed descrip-
tion and thermodynamic analysis of the solar-assisted Stirling heat engine system
considered in this case study were presented by Dai et al. (2018). The objective func-
tions of this case study are power output (P), system efficiency (η), and ecological
coefficient of performance (ECOP), which are given in Eqs. 2.35–2.37, respectively.
These objective functions are maximization functions. Four decision variables are
considered to assess the performance of the considered system. These decision vari-
ables and their ranges are as follows: 1000 ≤ T H ≤ 1200, 280 ≤ T L ≤ 300, 600 ≤
T h ≤ 1000, and 300 ≤ T c ≤ 600. To maintain the uniformity and to facilitate a fair
comparison of the computational results of the proposed algorithms with previous
works, characteristics of the solar-assisted Stirling system taken are the same as given
by the Dai et al. (2018). The characteristic features of the system considered in this
book are as follows: Cv = 3.214Jg−1 k−1 , μ2−3 = 0.3, μ4−1 = 0.2, T0 = 300K,
m = 4g, λ = 2, k1 = k2 = 5000 Ks , αleak = 12 K.s W
, αc = αh = 1000K.s
W
, n = 1mol
−1 −1
and R = 8.314Jmol K .
Dai et al. (2018) solved this MOO problem using the multi-objective PSO with
crowding distance (MOPSOCD) algorithm and reported a Pareto-optimal solution
using the TOPSIS decision-making method. The number of function evaluations
taken by the MOPSOCD algorithm was not specified. However, in this book, the Jaya
algorithm, Rao algorithms, and their modified versions are tested by taking 10,000
function evaluations as the termination criterion. The best Pareto-optimal solutions
Table 7.45 Ranks suggested by the MADM methods for different algorithm solutions presented in Table 7.44
Algorithm SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA AR CR FR
TOPSIS 5 6 7 7 6 2 6 6 5.625 6.2 6
LINMAP 8 7 8 8 4 2 8 9 6.75 7.8 8
Bellman–Zadeh 11 11 12 12 3 5 12 1 8.375 11.6 12
Jaya 10 10 9 9 12 12 11 10 10.37 9.80 10
Rao-1 6 8 6 6 11 8 7 3 6.875 6.60 7
Rao-2 12 12 11 11 10 6 10 2 9.25 11.20 11
Rao-3 9 9 10 10 2 2 9 11 7.75 9.40 9
AMTPG-Jaya 7 2 1 1 1 9 2 12 4.375 2.60 2
SAP-Rao 4 3 4 4 5 4 4 8 4.5 3.80 4
ERao-1 2 5 3 3 9 11 3 5 5.125 3.20 3
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System

ERao-2 3 4 5 5 7 7 5 7 5.375 4.40 5

ERao-3 1 1 2 2 8 10 1 4 3.625 1.40 1
AR average rank, CR corrected rank, FR final rank
217
 218

Table 7.46 Spearman’s rank correlation coefficients between different pairs of MADM method’s rankings presented in Table 7.45
Method SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA
SAW 1 0.853 0.825 0.825 −0.098 −0.254 0.867 −0.014
WPM 0.853 1 0.937 0.937 0.238 −0.204 0.951 −0.315
TOPSIS 0.825 0.937 1 1 0.042 −0.437 0.965 −0.273
MTOPSIS 0.825 0.937 1 1 0.042 −0.437 0.965 −0.273
VIKOR −0.098 0.238 0.042 0.042 1 0.570 0.112 −0.413
PROMETHEE −0.254 −0.204 −0.437 −0.437 0.570 1 −0.289 −0.070
COPRAS 0.867 0.951 0.965 0.965 0.112 −0.289 1 −0.189
GRA −0.014 −0.315 −0.273 −0.273 −0.413 −0.070 −0.189 1
7 Optimization of the Selected Solar-Assisted Energy Systems
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System 219

Table 7.47 Hypervolume

Algorithm Hypervolume Spacing
and spacing values of the
Pareto-fronts obtained by the Jaya 0.00329951 0.08681974
proposed algorithms in MOO Rao-1 0.00329721 0.08979312
for the solar-assisted Stirling
Rao-2 0.00333040 0.09439404
engine case study-2
Rao-3 0.00328939 0.08597348
AMTPG-Jaya 0.00333557 0.07115295
SAP-Rao 0.00348837 0.08190799
ERao-1 0.00333131 0.07647491
ERao-2 0.00343446 0.07422897
ERao-3 0.00333158 0.11455875
Results in the bold figure indicate better values

Table 7.48 Coverage (%) values of the Pareto-fronts obtained by the proposed algorithms in MOO
for the solar-assisted Stirling engine case study-2
Algorithm Jaya Rao-1 Rao-2 Rao-3 AMTPG-Jaya SAP-Rao ERao-1 ERao-2 ERao-3
Jaya – 20 16 12 12 4 4 0 8
Rao-1 16 – 28 4 12 4 4 4 16
Rao-2 8 8 – 4 8 0 12 0 4
Rao-3 8 24 24 – 16 12 8 4 16
AMTPG-Jaya 16 8 24 4 – 0 4 0 16
SAP-Rao 24 24 8 24 12 – 4 8 16
ERao-1 16 12 28 4 12 0 – 4 16
ERao-2 12 16 16 4 20 0 8 – 8
ERao-3 28 20 32 4 16 0 0 0 –

are reported using the average rank based on multiple decision-making methods.
Similar to the previous case studies, in this case study also, first single-objective
optimization is performed independently for each objective function, and then MOO
is performed by the proposed algorithms. In all the computational experiments, the
population size is maintained as 25 for all the algorithms. The elite size for the elitist
Rao algorithms is taken as 20%.
The Jaya algorithm, Rao algorithms, and their modified versions have obtained
identical solutions in all the single-objective optimization scenarios of this case study,
and those solutions are presented in Table 7.49. The maximum power output, ECOP,
and efficiency obtained by the proposed algorithms are 14.6026 kW, 1.0935, and
33.498%, respectively.
The Pareto-optimal solutions achieved by the Jaya algorithm, Rao algorithms,
and their modified versions in MOO are presented in Tables 7.50, 7.51, 7.52, 7.53,
7.54, 7.55, 7.56, 7.57 and 7.58. Here, the power output is varied from 10.31 to
14.6 kW, the system efficiency is varied between 28.36% and 33.5%, and ECOP is
varied from 0.63 to 1.0935. The best solutions from the Pareto-fronts obtained by
 220

Table 7.49 Results obtained by the proposed algorithms in single-objective optimization scenarios of the solar-assisted Stirling heat engine case study-3
Objective Algorithm T h (K) T c (K) T H (K) T L (K) P (kW) η (%) ECOP
Maximize-P MOPSOCD 874.3 455.4 1200 280 14.6 28.47 0.64
Proposed algorithms 874.29534 455.35696 1200 280 14.6026 28.472 0.6351
Maximize-η MOPSOCD 985.1 325.2 1119 280 12.23 33.49 0.92
Proposed algorithms 999.9 325.92772 1135.9649 280 12.3397 33.498 0.9094
Maximize-ECOP MOPSOCD 896.8 340.2 1000 300 10.86 32.74 1.09
Proposed algorithms 896.8394 340.20277 1000 300 10.8632 32.7473 1.0935
Source MOPSOCD—Dai et al. (2018)
7 Optimization of the Selected Solar-Assisted Energy Systems
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System 221

Table 7.50 Pareto-optimal solutions obtained by the Jaya algorithm for the solar-assisted Stirling
heat engine case study-3
Solution T h (K) T c (K) T H (K) T L (K) P (kW) η (%) ECOP
1 896.73 340.01 1000.00 300.00 10.8569 32.7473 1.0935
2 998.78 325.39 1136.32 280.00 12.3384 33.4976 0.9091
3 873.90 458.01 1200.00 280.00 14.6023 28.3561 0.6304
4 906.01 325.34 1000.00 291.04 10.5596 32.9467 1.0445
5 892.14 350.76 1005.14 290.95 11.7093 32.7967 1.0280
6 920.62 318.00 1014.57 280.00 10.9331 33.2978 0.9811
7 888.26 441.19 1200.00 280.00 14.5898 29.3014 0.6698
8 911.00 372.38 1200.00 285.74 14.0696 31.8430 0.8144
9 918.48 417.02 1200.00 280.00 14.4967 30.6522 0.7303
10 892.17 430.07 1200.00 283.90 14.5228 29.7895 0.7064
11 875.28 447.22 1200.00 280.00 14.5999 28.8169 0.6493
12 922.76 365.89 1121.67 298.60 13.0707 32.4468 0.9585
13 909.11 437.63 1200.00 291.84 14.4325 29.8281 0.7403
14 895.33 390.34 1093.86 299.43 13.2807 31.6452 0.9311
15 922.68 386.65 1117.96 293.58 13.4568 31.9927 0.9046
16 941.61 409.96 1200.00 280.00 14.4151 31.1895 0.7557
17 922.15 344.47 1074.99 281.98 12.6559 33.1404 0.9372
18 933.76 361.44 1175.55 280.00 13.8748 32.5168 0.8335
19 961.21 430.77 1200.00 289.45 14.3071 30.8157 0.7785
20 826.24 356.80 1000.00 300.00 12.0443 31.9825 1.0356
21 954.26 429.15 1181.42 285.47 14.2218 30.8405 0.7699
22 913.38 324.33 1000.00 284.94 10.7078 33.0993 1.0134
23 906.37 394.08 1155.77 300.00 13.7689 31.4039 0.8813
24 891.11 373.56 1077.55 300.00 12.9027 32.1111 0.9765
25 999.90 351.31 1200.00 280.00 13.6502 33.1160 0.8547

the proposed algorithms are identified based on the average rank method. Solution
12 of the Jaya algorithm, Solution 18 of the Rao-1 algorithm, Solution 7 of the
Rao-2 algorithm, Solution 12 of the Rao-3 algorithm, Solution 16 of the AMTPG-
Jaya algorithm, Solution 16 of the SAP-Rao algorithms, Solution 21 of the ERao-1
algorithm, Solution 13 of the ERao-2, and Solution 14 of the ERao-3 algorithm are
identified as the best solutions from the respective algorithm Pareto-front. Now, these
best solutions are compared with that of the MOPSOCD algorithm in Table 7.59.
In Table 7.59, the solutions with the best average rank from the Pareto-optimal
solutions of the various algorithms are compared with the MOPSOCD algorithm
solution. Figure 7.4 presents the Pareto-fronts achieved by the AMTPG-Jaya, SAP-
Rao, and elitist Rao algorithms, including the optimal solutions reported for the
MOPSOCD, Jaya, and Rao algorithms. The conflicting nature of these objectives
222 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.51 Pareto-optimal solutions obtained by the Rao-1 algorithm for the solar-assisted Stirling
heat engine case study-3
Solution T h (K) T c (K) T H (K) T L (K) P (kW) η (%) ECOP
1 999.90 326.48 1141.40 280.00 12.4337 33.4962 0.9062
2 896.46 339.90 1000.00 300.00 10.8567 32.7472 1.0935
3 875.69 454.65 1200.00 280.00 14.6025 28.5298 0.6375
4 912.92 324.46 1000.00 289.29 10.4907 32.9644 1.0337
5 940.91 448.72 1200.00 298.11 14.3239 29.9717 0.7736
6 904.42 406.85 1200.00 280.43 14.4694 30.7733 0.7377
7 884.43 442.85 1200.00 280.00 14.5937 29.1664 0.6641
8 940.69 331.84 1080.84 280.00 12.3334 33.4059 0.9381
9 873.50 418.15 1200.00 281.32 14.5231 29.8806 0.7003
10 949.30 443.37 1199.90 287.30 14.3984 30.2463 0.7415
11 870.48 425.34 1200.00 280.00 14.5632 29.5606 0.6810
12 931.41 359.99 1200.00 280.00 14.0067 32.3895 0.8159
13 899.58 360.06 1000.00 297.70 11.4304 32.5316 1.0605
14 915.50 390.47 1125.50 300.00 13.4711 31.7630 0.9208
15 930.82 411.20 1172.30 300.00 13.9735 31.1555 0.8582
16 875.88 452.47 1199.92 280.00 14.6018 28.6213 0.6412
17 901.77 340.53 1000.00 287.15 11.3312 32.9934 1.0210
18 931.33 358.47 1125.00 300.00 12.8447 32.6183 0.9755
19 911.47 350.45 1011.73 299.94 11.2107 32.7245 1.0788
20 921.50 413.28 1136.75 298.78 13.7508 31.1166 0.8691
21 977.12 415.15 1200.00 294.86 14.1132 31.4702 0.8374
22 892.22 377.86 1131.16 280.70 13.8560 31.8428 0.8222
23 903.53 380.08 1040.42 294.73 12.5247 32.1561 0.9768
24 877.64 366.77 1067.56 297.27 12.8263 32.1765 0.9720
25 920.47 331.76 1191.76 280.00 13.3151 32.7384 0.8380

can be observed in Fig. 7.4. Furthermore, the power output of the system for the Jaya
algorithm solution is highest, which is 1.1%, 2%, 10%, 15%, 0.6%, 4.7%, 5.4%,
0.5%, and 5.7% higher than that of the MOPSOCD, Rao-1, Rao-2, Rao-3, AMTPG-
Jaya, SAP-Rao, ERao-1, ERao-2, and ERao-3 algorithms solutions, respectively.
Similarly, the system efficiency for the Rao-2 algorithm solution is highest, which
is 1.32%, 1.87%, 1.33%, 1.34%, 0.9%, 1.15%, 1.23%, 1.01%, and 1.12% higher
than that of the MOPSOCD, Jaya, Rao-1, Rao-3, AMTPG-Jaya, SAP-Rao, ERao-
1, ERao-2, and ERao-3 algorithms solutions, respectively. The ECOP of the Rao-3
algorithm solution is 6.6%, 13%, 11%, 7.2%, 12.4%, 7%, 6.2%, 11.6%, and 6.1%
higher than that of the MOPSOCD, Jaya, Rao-1, Rao-2, AMTPG-Jaya, SAP-Rao,
ERao-1, ERao-2, and ERao-3 algorithms solutions, respectively.
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System 223

Table 7.52 Pareto-optimal solutions obtained by the Rao-2 algorithm for the solar-assisted Stirling
heat engine case study-3
Solution T h (K) T c (K) T H (K) T L (K) P (kW) η (%) ECOP
1 874.37 455.11 1200.00 280.00 14.6026 28.4835 0.6356
2 999.90 326.31 1135.66 280.00 12.3475 33.4979 0.9095
3 896.15 340.08 1000.00 300.00 10.8687 32.7472 1.0935
4 903.19 330.32 1000.00 296.64 10.5035 32.7928 1.0727
5 908.63 448.91 1200.00 280.00 14.5748 29.3839 0.6734
6 875.54 449.78 1200.00 280.00 14.6013 28.7210 0.6453
7 905.18 340.41 1036.45 289.90 11.8887 33.0520 1.0106
8 916.19 423.74 1200.00 280.00 14.5218 30.3955 0.7184
9 914.91 432.42 1200.00 288.58 14.4484 30.0983 0.7397
10 902.43 435.79 1200.00 280.00 14.5690 29.7538 0.6895
11 900.12 334.79 1000.00 290.71 11.0660 32.9967 1.0459
12 929.73 385.09 1200.00 297.87 13.9398 31.7167 0.8655
13 913.67 413.31 1200.00 287.32 14.3932 30.7105 0.7641
14 934.14 332.20 1083.46 288.86 12.0415 33.1894 0.9759
15 922.04 398.86 1142.21 300.00 13.6710 31.5369 0.8969
16 901.38 353.86 1000.00 298.57 11.2506 32.6403 1.0749
17 882.85 354.85 1141.51 291.22 13.2916 32.1649 0.8883
18 907.21 347.88 1111.45 292.44 12.8179 32.7016 0.9462
19 900.87 431.10 1200.00 300.00 14.3127 29.9265 0.7794
20 929.52 413.94 1200.00 300.00 14.1814 30.9245 0.8318
21 878.57 322.45 1006.92 281.00 11.5040 33.2636 0.9920
22 922.13 377.10 1067.81 300.00 12.6051 32.3407 0.9997
23 894.25 395.61 1200.00 291.47 14.2189 30.9657 0.7951
24 895.32 374.20 1000.00 300.00 11.6529 32.1732 1.0497
25 878.52 354.93 1118.39 291.93 13.1354 32.2356 0.9100

The decision-making methods ranking of the Pareto-optimal solutions obtained

by different algorithms is shown in Table 7.60. Spearman’s correlation coefficients
for different pairs of rankings given by decision-making methods for different algo-
rithm’s solutions are shown in Table 7.61. The ranks given by the TOPSIS and
MTOPSIS methods for each solution are identical. Furthermore, Spearman’s corre-
lation for all the pairs formed by all the MADM (except the pairs with VIKOR,
PROMETHEE, and GRA) methods is greater than 0.5. Spearman’s correlation values
for all the pairs formed by the GRA methods are either negative or less than 0.5.
Spearman’s correlation values for the VIKOR-SAW, VIKOR-WPM, VIKOR-GRA,
and VIKOR-COPRAS pairs are less than 0.5. Similarly, the PROMETHEE method
has Spearman’s correlation values less than 0.5 with the TOPSIS, MTOPSIS, and
GRA methods. Hence, the ranks suggested by the VIKOR, PROMETHEE, and
224 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.53 Pareto-optimal solutions obtained by the Rao-3 algorithm for the solar-assisted Stirling
heat engine case study-3
Solution T h (K) T c (K) T H (K) T L (K) P (kW) η (%) ECOP
1 897.74 340.47 1000.00 300.00 10.8600 32.7471 1.0935
2 996.99 327.09 1137.87 280.00 12.4336 33.4954 0.9081
3 874.02 454.09 1200.00 280.00 14.6025 28.5171 0.6369
4 907.51 312.27 1000.00 281.54 10.4627 33.1356 0.9929
5 906.57 391.25 1200.00 280.00 14.3785 31.2861 0.7604
6 911.57 414.80 1200.00 280.00 14.5003 30.6219 0.7288
7 887.85 428.94 1200.00 280.00 14.5685 29.7464 0.6892
8 958.43 388.14 1200.00 280.00 14.2409 31.9992 0.7958
9 899.96 395.18 1085.72 299.56 13.2166 31.5862 0.9338
10 837.93 341.55 1045.15 291.63 12.2937 32.3326 0.9656
11 897.97 453.01 1200.00 280.00 14.5901 29.0359 0.6585
12 891.97 354.23 1000.00 300.00 11.3788 32.6134 1.0831
13 837.89 328.64 1000.00 287.81 11.5739 32.7545 1.0085
14 999.90 327.61 1200.00 280.00 13.0735 33.3221 0.8661
15 897.59 420.11 1200.00 280.00 14.5362 30.2245 0.7106
16 844.15 361.69 1000.00 297.81 12.1487 32.1282 1.0314
17 901.84 340.84 1000.00 291.45 11.1833 32.9260 1.0459
18 868.04 368.09 1000.00 300.00 11.9835 32.1909 1.0510
19 998.52 382.99 1188.89 298.69 13.5721 32.4829 0.9202
20 996.01 355.50 1105.38 298.38 11.9584 32.9589 1.0026
21 996.00 359.99 1200.00 280.00 13.7998 32.9460 0.8454
22 947.83 396.11 1151.18 296.49 13.6732 31.8743 0.8934
23 983.90 377.16 1200.00 280.00 14.0477 32.5017 0.8218
24 999.90 382.26 1120.89 289.69 12.7322 32.5567 0.9156
25 999.90 350.84 1157.52 280.00 13.1917 33.2442 0.8827

GRA methods cannot be considered for calculating the average ranks. Now, the
corrected average ranks are calculated, excluding the ranks suggested by the VIKOR,
PROMETHEE, and GRA methods and presented in Table 7.60 as the corrected ranks.
The SAP-Rao algorithm solution has the least average rank of 1.8. Thus, it can be
regarded as the best solution. The ERao-1, ERao-2, and ERao-3 algorithm’s solutions
have the next best average ranks, which are 2.2, 2.8, and 4, respectively. Here an
observation can be made that the SAP-Rao algorithm solution has a better fitness
value in none of the three objectives. However, it is ranked 1 because it has a better
compromise among the objectives. Here, it can be noted that the MADM methods
have ranked these solutions based on the relative importance of the solutions. If
the MOPSOCD and SAP-Rao algorithms solutions are compared, the MOPSOCD
solution has better ECOP, and the SAP-Rao algorithm solution has better power
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System 225

Table 7.54 Pareto-optimal solutions obtained by the AMTPG-Jaya algorithm for the solar-assisted
Stirling heat engine case study-3
Solution T h (K) T c (K) T H (K) T L (K) P (kW) η (%) ECOP
1 873.65 455.35 1200.00 280.00 14.6026 28.4589 0.6345
2 896.94 340.25 1000.00 300.00 10.8633 32.7473 1.0935
3 999.90 326.28 1133.23 280.00 12.3110 33.4974 0.9109
4 896.16 314.26 1000.00 287.17 10.3098 32.9054 1.0149
5 889.45 446.63 1200.00 280.00 14.5943 29.1180 0.6620
6 890.50 433.74 1200.00 285.69 14.5120 29.6306 0.7063
7 927.36 373.50 1115.72 300.00 13.0939 32.3501 0.9644
8 892.14 375.51 1131.74 295.38 13.5013 31.8524 0.8976
9 958.36 413.90 1187.21 286.29 14.1962 31.3381 0.7963
10 901.97 423.86 1200.00 285.99 14.4698 30.1769 0.7328
11 975.49 340.10 1104.42 280.00 12.4927 33.3731 0.9208
12 911.14 384.66 1161.13 296.17 13.7677 31.7080 0.8766
13 914.02 413.91 1171.49 295.73 14.0958 30.8412 0.8209
14 966.79 411.06 1200.00 282.21 14.3020 31.4567 0.7786
15 923.98 428.87 1200.00 289.09 14.4157 30.3606 0.7545
16 964.68 359.77 1154.30 299.80 12.9979 32.7558 0.9638
17 895.24 324.42 1000.00 280.20 11.2797 33.2961 0.9950
18 892.16 402.93 1162.68 297.67 13.9670 30.9021 0.8377
19 898.96 411.84 1200.00 280.00 14.5039 30.5263 0.7244
20 870.21 371.56 1000.00 300.00 12.0120 32.1160 1.0455
21 917.99 322.01 1000.00 284.31 10.5438 33.0680 1.0069
22 857.33 342.53 1000.00 300.00 11.3787 32.5431 1.0777
23 881.56 361.70 1000.00 300.00 11.6965 32.4340 1.0694
24 874.12 350.96 1000.00 291.04 11.8643 32.7101 1.0273
25 940.99 385.29 1179.50 292.41 13.9064 31.9686 0.8629

output and system efficiency. In addition, combined improvement in the power output
(1.1%) and system efficiency (0.17%) is relatively more significant than the descent
in the ECOP (0.3%) by the SAP-Rao solution. Similarly, the SAP-Rao algorithm
solution has better relative importance when compared to the solutions of the other
algorithms.
The system’s power output for the SAP-Rao algorithm solution is 1.1%, 5%,
9.7, 0.7, and 1% higher than that of the MOPSOCD, Rao-2, Rao-3, ERao-1, and
ERao-3 solutions, respectively. Also, ECOP of the system for the SAP-Rao solution
is 5.6%, 3.8%, 0.2%, 5.1%, and 4.3% higher than that of the Jaya, Rao-1, Rao-
2, AMTPG-Jaya, and ERao-2 algorithms solutions, respectively. In addition, the
thermal efficiency of the SAP-Rao algorithm solution is 0.17%, 0.71%, 0.18%, and
0.19% higher when compared to that of the MOPSOCD, Jaya, Rao-1, and Rao-3
226 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.55 Pareto-optimal solutions obtained by the SAP-Rao algorithm for the solar-assisted
Stirling heat engine case study-3
Solution T h (K) T c (K) T H (K) T L (K) P (kW) η (%) ECOP
1 872.34 453.89 1200.00 280.00 14.6024 28.4900 0.6358
2 894.63 339.21 1000.00 300.00 10.8546 32.7457 1.0934
3 999.90 327.25 1141.72 280.00 12.4626 33.4953 0.9059
4 904.96 325.56 1000.00 294.76 10.3366 32.7880 1.0589
5 870.66 430.05 1200.00 280.00 14.5751 29.3879 0.6735
6 921.98 429.36 1200.00 280.00 14.5244 30.2944 0.7138
7 936.73 411.43 1200.00 293.59 14.2496 31.0967 0.8115
8 982.07 394.08 1176.26 292.45 13.7699 32.1577 0.8755
9 942.92 406.17 1200.00 280.00 14.3949 31.3194 0.7620
10 907.01 434.95 1200.00 280.00 14.5616 29.8613 0.6943
11 913.37 434.02 1200.00 297.75 14.3428 30.0340 0.7752
12 938.61 344.19 1037.21 294.24 11.3494 32.9614 1.0323
13 926.89 393.48 1115.41 298.82 13.3700 31.8488 0.9261
14 912.42 321.83 1000.00 284.08 10.6758 33.1246 1.0093
15 925.73 340.10 1042.67 284.95 11.9135 33.1872 0.9826
16 930.04 363.18 1080.46 300.00 12.4802 32.6758 1.0126
17 998.81 327.40 1135.20 284.59 12.1824 33.3915 0.9292
18 916.04 398.40 1200.00 293.41 14.1929 31.1971 0.8160
19 896.95 367.93 1070.23 300.00 12.6930 32.3321 0.9972
20 934.76 392.86 1193.86 295.91 14.0278 31.6220 0.8536
21 933.75 394.35 1171.12 294.27 13.9217 31.6842 0.8604
22 974.55 374.10 1167.07 292.72 13.5197 32.5908 0.9071
23 959.58 352.95 1120.68 290.17 12.8400 33.0333 0.9480
24 920.58 356.43 1066.89 287.84 12.6091 32.8680 0.9609
25 938.61 388.64 1185.93 289.89 14.0403 31.8295 0.8397

algorithms solutions, respectively. Hence, the SAP-Rao algorithm solution can be

considered as the best solution.
Now, the performances of the Jaya algorithm, Rao algorithms, and their modified
versions in the MOO scenario are evaluated based on the hypervolume, coverage,
and spacing indicators. Table 7.62 presents the hypervolume and spacing values of
the Pareto-fronts obtained by different algorithms in the MOO scenario. The Pareto-
front obtained by the MOPSOCD algorithm was not reported. Thus, the performance
metric values for the MOPSOCD algorithm were not presented here. Here, an obser-
vation can be made that the SAP-Rao algorithm Pareto-front has the highest hyper-
volume value, and the ERao-2 algorithm Pareto-front has the least spacing value. The
SAP-Rao algorithm has 1.74%, 1.71%, 1.26%, 1.64%, 1%, 0.4%, and 0.39% higher
hypervolume value when compared to that of the Jaya, Rao-1, Rao-2, Rao-3, ERao-1,
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System 227

Table 7.56 Pareto-optimal solutions obtained by the ERao-1 algorithm for the solar-assisted
Stirling heat engine case study-3
Solution T h (K) T c (K) T H (K) T L (K) P (kW) η (%) ECOP
1 872.98 455.81 1200.00 280.00 14.6025 28.4259 0.6332
2 897.15 340.42 1000.00 300.00 10.8671 32.7473 1.0935
3 999.90 326.14 1134.35 280.00 12.3230 33.4978 0.9103
4 907.67 331.15 1000.00 296.60 10.4718 32.7900 1.0722
5 903.46 448.26 1200.00 280.00 14.5817 29.3161 0.6705
6 934.71 402.33 1189.11 295.62 14.0879 31.3869 0.8416
7 948.34 442.07 1200.00 295.86 14.3117 30.2980 0.7804
8 910.96 409.30 1200.00 280.01 14.4782 30.7915 0.7368
9 876.13 390.30 1165.64 280.00 14.2159 31.0540 0.7628
10 872.05 451.19 1200.00 280.00 14.6018 28.5929 0.6400
11 901.65 431.59 1200.00 281.34 14.5474 29.8935 0.7010
12 874.25 386.59 1016.89 300.00 12.4283 31.7038 0.9994
13 915.33 364.61 1200.00 295.81 13.6529 31.9894 0.8708
14 919.37 431.87 1200.00 280.00 14.5349 30.1688 0.7081
15 970.07 327.84 1063.36 286.68 11.1782 33.1839 0.9771
16 954.37 369.74 1183.52 285.98 13.8437 32.4707 0.8569
17 968.89 323.15 1083.72 280.00 11.7757 33.4592 0.9395
18 999.90 324.05 1200.00 286.45 12.5709 33.1041 0.8869
19 938.72 373.91 1053.93 300.00 12.0953 32.4755 1.0208
20 920.15 368.58 1152.13 295.45 13.4618 32.2461 0.9095
21 918.01 361.00 1070.01 300.00 12.3983 32.6516 1.0195
22 999.90 339.04 1113.62 289.37 11.9788 33.2449 0.9614
23 888.07 351.11 1093.53 294.56 12.7391 32.5124 0.9587
24 916.59 366.76 1127.50 300.00 13.1151 32.3266 0.9549
25 988.64 358.56 1179.83 294.46 13.2580 32.8981 0.9278

ERao-2, and ERao-3 algorithms, respectively. The AMTPG-Jaya, ERao-3, ERao-2,

and ERao-1 algorithms have achieved the next better hypervolume values, respec-
tively. Similarly, the ERao-2 algorithm has 20.5%, 30.8%, 16.7%, 25.1%, 0.5%,
34.8%, 27.3%, and 10.6% less spacing value when compared to that of the Jaya,
Rao-1, Rao-2, Rao-3, AMTPG-Jaya, SAP-Rao, ERao-1, and ERao-3 algorithms,
respectively. The performances of the proposed modified algorithms are better or
competitive to that of the basic algorithms in terms of the hypervolume and spacing
values.
The coverage values of the Pareto-fronts obtained by the Jaya algorithm, Rao
algorithms, and their modified versions are presented in Table 7.63. The SAP-Rao
algorithm has achieved better coverage values than the other algorithms compared.
Only three solutions of the SAP-Rao algorithm are dominated by the solutions of
228 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.57 Pareto-optimal solutions obtained by the ERao-2 algorithm for the solar-assisted
Stirling heat engine case study-3
Solution T h (K) T c (K) T H (K) T L (K) P (kW) η (%) ECOP
1 872.34 453.70 1200.00 280.00 14.6024 28.4977 0.6361
2 895.53 340.07 1000.00 300.00 10.8774 32.7470 1.0935
3 999.90 327.08 1136.86 280.00 12.3894 33.4973 0.9088
4 907.56 327.65 1003.43 296.46 10.3683 32.7603 1.0654
5 892.16 446.99 1200.00 280.00 14.5926 29.1553 0.6636
6 896.19 431.67 1200.00 280.00 14.5681 29.7943 0.6913
7 903.60 419.42 1200.00 280.00 14.5275 30.3450 0.7161
8 868.75 326.91 1000.00 289.98 11.1651 32.9411 1.0367
9 945.03 422.94 1200.00 280.00 14.4479 30.8312 0.7386
10 958.86 335.74 1100.22 293.76 11.9432 33.1306 0.9897
11 953.26 434.23 1200.00 297.68 14.2529 30.6226 0.8052
12 979.27 406.14 1194.13 280.00 14.1979 31.7411 0.7852
13 970.96 367.76 1142.30 300.00 13.0042 32.7205 0.9704
14 850.71 344.46 1000.00 289.78 11.9807 32.6685 1.0158
15 849.30 333.04 1000.00 300.00 10.9081 32.4208 1.0683
16 999.90 356.80 1167.84 290.08 13.1562 33.0611 0.9206
17 880.24 369.01 1030.42 300.00 12.3428 32.2456 1.0254
18 953.82 407.18 1200.00 280.13 14.3626 31.4198 0.7675
19 999.90 401.11 1200.00 300.00 13.8263 32.0488 0.8952
20 997.52 328.45 1098.89 286.45 11.5115 33.2549 0.9547
21 851.17 378.83 1109.87 300.00 13.3367 31.2502 0.8993
22 999.90 343.44 1175.34 280.00 13.2704 33.3106 0.8774
23 933.72 378.50 1185.67 290.01 13.9388 32.0303 0.8515
24 979.30 353.30 1078.49 300.00 11.5592 32.8751 1.0282
25 976.08 370.16 1200.00 280.00 14.0150 32.5957 0.8267

the other compared methods. From the coverage values, it can be observed that four
solutions from the Pareto-fronts of the ERao-1, ERao-2, and Rao-2 algorithms are
dominated solutions. Similarly, five solutions from the Pareto-fronts achieved by
the Rao-1, Rao-3, AMTPG-Jaya, and ERao-3 algorithms are dominated, and six
solutions of the Pareto-front achieved by the Jaya algorithm are dominated. The
performance of the proposed modified algorithms is better or competitive to that of
the basic algorithms in terms of the coverage values.
From the computational results of the solar-assisted Stirling heat engine case
study-3, it can be observed that the performance of the considered system can be
modified with the solutions achieved by the proposed algorithms. The solution of
the SAP-Rao algorithm has better compromise among the power output, system
efficiency, and ECOP. The SAP-Rao algorithm’s solution has 9.7% higher power
7.2 Optimization of a Solar-Assisted Stirling Heat Engine System 229

Table 7.58 Pareto-optimal solutions obtained by the ERao-3 algorithm for the solar-assisted
Stirling heat engine case study-3
Solution T h (K) T c (K) T H (K) T L (K) P (kW) η (%) ECOP
1 897.69 341.15 1000.00 300.00 10.8873 32.7465 1.0934
2 999.90 328.74 1131.84 280.00 12.3646 33.4921 0.9114
3 873.55 456.99 1200.00 280.00 14.6025 28.3900 0.6317
4 898.56 325.36 1000.00 295.07 10.3949 32.7799 1.0605
5 917.21 389.35 1181.48 287.20 14.1168 31.5753 0.8154
6 949.41 328.16 1049.81 280.11 11.5437 33.3600 0.9580
7 954.39 395.17 1200.00 280.00 14.3008 31.7675 0.7841
8 849.67 363.52 1000.00 300.00 12.0732 32.1257 1.0462
9 970.06 382.34 1074.66 300.00 12.0694 32.3624 0.9957
10 931.80 415.02 1200.00 280.00 14.4609 30.9046 0.7421
11 855.32 403.64 1026.87 300.00 12.8718 30.8588 0.9339
12 914.18 314.74 1005.91 280.00 10.7075 33.2464 0.9850
13 999.90 369.97 1200.00 280.00 13.8938 32.7744 0.8362
14 922.96 361.01 1070.02 299.81 12.3618 32.6868 1.0207
15 882.08 407.89 1053.29 299.35 13.0467 31.0280 0.9207
16 885.93 375.54 1126.71 292.31 13.5579 31.8110 0.8819
17 987.55 338.70 1088.53 296.14 11.3769 33.0356 1.0068
18 864.82 441.46 1200.00 280.00 14.5932 28.8305 0.6499
19 908.28 420.45 1200.00 280.00 14.5247 30.3838 0.7178
20 999.90 332.15 1200.00 280.00 13.2206 33.3141 0.8656
21 910.89 442.44 1200.00 280.00 14.5660 29.6588 0.6853
22 894.08 420.81 1200.00 280.00 14.5416 30.1422 0.7068
23 999.90 350.87 1125.14 280.00 12.6923 33.2346 0.9000
24 999.90 378.06 1170.13 280.00 13.6431 32.6782 0.8448
25 968.24 342.54 1115.77 290.93 12.4269 33.1873 0.9655

output and 5.6% higher ECOP when compared to those of the solutions achieved by
the other algorithms compared. The SAP-Rao algorithm’s performance in terms of
coverage and hypervolume is much better than those achieved by the other algorithms.
Similarly, the ERao-2 algorithm’s performance in terms of the spacing is much better
than that achieved by other algorithms. In addition, the performances of the modified
versions are better or competitive to those of the basic algorithms as well as the
MOPSOCD algorithm. Here it can be observed that, in the three solar-assisted Stirling
heat engine case studies, the performances of the proposed modified algorithms are
superior or competitive to those of the basic Jaya and Rao algorithms. The next
section presents the application of the Jaya and Rao algorithms along with their
modified versions, to the solar-assisted Carnot-like heat engine case study.
230 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.59 Best solutions obtained by various algorithms in MOO scenario of the solar-assisted
Stirling heat engine case study-3
Algorithm T h (K) T c (K) T H (K) T L (K) P (kW) η (%) ECOP
MOPSOCD 917 362.3 1070.4 300 12.44 32.62 1.016
Jaya 922.755 365.895 1121.67 298.6 13.07068 32.446789 0.95852
Rao-1 931.330 358.470 1125.00 300 12.84471 32.618340 0.97547
Rao-2 905.184 340.412 1036.45 289.9 11.88868 33.051957 1.01062
Rao-3 891.967 354.235 1000.00 300 11.37881 32.613382 1.08310
AMTPG-Jaya 964.678 359.767 1154.30 299.8 12.99792 32.755804 0.96379
SAP-Rao 930.039 363.177 1080.46 300 12.48016 32.675822 1.01262
ERao-1 918.009 361.004 1070.01 300 12.39834 32.651557 1.01949
ERao-2 970.964 367.758 1142.30 300 13.00417 32.720541 0.97044
ERao-3 922.959 361.006 1070.02 299.81 12.36179 32.686821 1.02074
Source MOPSOCD—Dai et al. (2018)
Bold figure indicates a better value

Fig. 7.4 Plot of Pareto-fronts of the proposed algorithms and solutions of the methods compared
in solar-assisted Stirling heat engine system case study-3

7.3 Optimization of a Solar-Assisted Carnot-like Heat

Engine System

The description of the selected solar-assisted Carnot-like heat engine system is

presented in Sect. 2.2.3. The detailed description and thermodynamic analysis of the
solar-assisted Carnot-like heat engine system considered in this book were presented
Table 7.60 Ranks suggested by the MADM methods for different algorithm solutions presented in Table 7.59
Algorithm SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA AR CR FR
MOPSOCD 5 5 3 3 5 7 5 10 5.4 4.2 5
Jaya 8 8 8 8 10 10 8 5 8.1 8 8
Rao-1 7 7 7 7 8 8 7 6 7.1 7 7
Rao-2 10 10 10 10 1 5.5 10 1 7.2 10 10
Rao-3 9 9 9 9 9 9 9 4 8.4 9 9
AMTPG-Jaya 6 6 6 6 7 2.5 6 2 5.2 6 6
SAP-Rao 2 1 2 2 1 4 2 7 2.6 1.8 1
ERao-1 3 3 1 1 1 5.5 3 9 3.3 2.2 2
ERao-2 1 2 5 5 6 1 1 3 3.0 2.8 3
ERao-3 4 4 4 4 1 2.5 4 8 3.9 4 4
7.3 Optimization of a Solar-Assisted Carnot-like Heat Engine System

AR average rank, CR corrected rank, FR final rank

231
 232

Table 7.61 Spearman’s rank correlation coefficients between different pairs of MADM method’s rankings presented in Table 7.60
Method SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA
SAW 1 0.988 0.855 0.855 0.438 0.671 1 −0.442
WPM 0.988 1 0.891 0.891 0.481 0.634 0.988 −0.491
TOPSIS 0.855 0.891 1 1 0.550 0.415 0.855 −0.758
MTOPSIS 0.855 0.891 1 1 0.550 0.415 0.855 −0.758
VIKOR 0.438 0.481 0.550 0.550 1 0.563 0.438 −0.300
PROMETHEE 0.671 0.634 0.415 0.415 0.563 1 0.671 0.152
COPRAS 1 0.988 0.855 0.855 0.438 0.671 1 −0.442
GRA −0.442 −0.491 −0.758 −0.758 −0.300 0.152 −0.442 1
7 Optimization of the Selected Solar-Assisted Energy Systems
7.3 Optimization of a Solar-Assisted Carnot-like Heat Engine System 233

Table 7.62 Hypervolume

Algorithm Hypervolume Spacing
and spacing values of the
Pareto-fronts obtained by the Jaya 71.14207 0.058204
proposed algorithms in MOO Rao-1 71.16683 0.066851
for the solar-assisted Stirling
Rao-2 71.47948 0.055525
engine case study-3
Rao-3 71.21845 0.061742
AMTPG-Jaya 72.32720 0.046464
SAP-Rao 72.38306 0.070969
ERao-1 71.66467 0.063585
ERao-2 72.09578 0.046245
ERao-3 72.09954 0.051722
Results in the bold figure indicate better values

Table 7.63 Coverage (%) values of the Pareto-fronts obtained by the proposed algorithms in MOO
for the solar-assisted Stirling engine case study-3
Algorithm Jaya Rao-1 Rao-2 Rao-3 AMTPG-Jaya SAP-Rao ERao-1 ERao-2 ERao-3
Jaya – 20 0 4 12 4 4 4 12
Rao-1 4 – 0 12 12 12 4 16 12
Rao-2 12 16 – 20 16 12 16 8 20
Rao-3 16 16 16 – 16 8 12 16 20
AMTPG-Jaya 12 16 8 16 – 4 16 4 20
SAP-Rao 16 16 16 12 20 – 16 4 8
ERao-1 12 8 8 12 12 8 – 8 20
ERao-2 12 20 8 16 12 8 12 – 12
ERao-3 24 16 4 4 8 8 8 12 –

by Sayyaadi et al. (2015). The three objective functions considered in this case study
are non-dimensional power output (W ), non-dimensional ecological function (E),
and thermal efficiency (η). These three objective functions are maximization func-
tions and are presented in Eqs. 2.40–2.42. The decision variables are the temperature
ratio (γ ), the convective coefficient ratio (φ), the operating temperature ratio (τ ),
and the heat source to the heat sink allocation parameter (β), and their ranges are as
follows:0.1 ≤ φ ≤ 0.6,0.1 ≤ β ≤ 0.6, 0.2 ≤ τ ≤ 0.4, and 0.7 ≤ γ ≤ 0.8.
Sayyaadi et al. (2015) employed NSGA-II to this MOO case study for finding
the optimal decision variables and documented three non-dominated solutions using
Bellman–Zadeh, TOPSIS, and LINMAP decision-making methods based on the
similarity to the ideal solutions. The number of function evaluations taken by NSGA-
II to achieve convergence was not specified. However, in this work, the proposed
algorithms are executed using only 10,000 function evaluations as the termination
criterion and reported the Pareto-optimal solutions using the average rank based
on the multiple decision-making methods. Firstly, single-objective optimization is
234 7 Optimization of the Selected Solar-Assisted Energy Systems

performed for each objective function independently, and then multi-objective opti-
mization is performed in this work. In all the computational experiments, the popu-
lation size is maintained as 25 for all the algorithms. The elite size for the elitist Rao
algorithms is taken as 20%.
Table 7.64 presents the single-objective optimization results of this case study.
The Jaya algorithm, Rao algorithms, and their modified versions have obtained
identical solutions in maximizing thermal efficiency. However, the proposed algo-
rithms have obtained different combinations of design variables, which have resulted
in the same objective values in the remaining two single-objective optimization
scenarios. The maximum non-dimensional power output, ecological function, and
efficiency obtained by the proposed algorithms are 0.2, 0.120645329, and 0.7225921,
respectively.

Table 7.64 Results obtained by the proposed algorithms in single-objective optimization scenarios
of the solar-assisted Carnot-like heat engine case study
Objective Algorithm φ β τ γ η W E
Maximize-η NSGA-II 0.1023 0.103 0.2 0.8 0.7216 0.0587 0.0523
Proposed 0.1 0.1 0.2 0.8 0.722592 0.05711 0.050995
algorithms
Maximize-W NSGA-II 0.3638 0.5552 0.2 0.8 0.4993 0.2 0.0795
Jaya 0.5338884 0.496649 0.2 0.8 0.500009 0.2 0.0800006
Rao-1 0.392703 0.544478 0.2 0.8 0.499998 0.2 0.0799992
Rao-2 0.6 0.474255 0.2 0.8 0.499999 0.2 0.0799995
Rao-3 0.2459854 0.594178 0.2 0.8 0.499999 0.2 0.0799999
AMTPG-Jaya 0.2288 0.6 0.2 0.8 0.499999 0.2 0.0799999
SAP-Rao 0.2403719 0.596079 0.2 0.8 0.500002 0.2 0.0800001
ERao-1 0.2287994 0.6 0.2 0.8 0.500001 0.2 0.0800001
ERao-2 0.4991504 0.508417 0.2 0.8 0.500001 0.2 0.0800001
ERao-3 0.2862074 0.580554 0.2 0.8 0.499999 0.2 0.0799994
Maximize-E NSGA-II 0.2328 0.4008 0.2 0.8 0.6157 0.1721 0.1206
Jaya 0.4574605 0.325392 0.2 0.8 0.612700 0.1737 0.1206453
Rao-1 0.6 0.277104 0.2 0.8 0.612701 0.1737 0.1206453
Rao-2 0.54593 0.2954205 0.2 0.8 0.612701 0.1737 0.1206453
Rao-3 0.6 0.2771036 0.2 0.8 0.612701 0.1737 0.1206453
AMTPG-Jaya 0.3340455 0.3671949 0.2 0.8 0.612702 0.1737 0.1206453
SAP-Rao 0.3754643 0.3531646 0.2 0.8 0.612702 0.1737 0.1206453
ERao-1 0.1400834 0.4329009 0.2 0.8 0.612702 0.1737 0.1206453
ERao-2 0.5715787 0.2867318 0.2 0.8 0.612701 0.1737 0.1206453
ERao-3 0.4772283 0.3186931 0.2 0.8 0.612701 0.1737 0.1206453
Source NSGA-II—Sayyaadi et al. (2015)
7.3 Optimization of a Solar-Assisted Carnot-like Heat Engine System 235

The Pareto-optimal solutions achieved by the Jaya algorithm, Rao algorithms,

and their modified versions in MOO are presented in Tables 7.65, 7.66, 7.67, 7.68,
7.69, 7.70, 7.71, 7.72 and 7.73.
Here, the non-dimensional power output is varied from 0.057 to 0.2, the thermal
efficiency is varied between 0.49996 and 0.7226, and ecological function is varied
from 0.05099 to 0.12065. The best solutions from the Pareto-fronts obtained by the
proposed algorithms are identified based on the average rank method. Solution 7
of the Jaya algorithm, Solution 25 of the Rao-1 algorithm, Solution 1 of the Rao-2
algorithm, Solution 17 of the Rao-3 algorithm, Solution 21 of the AMTPG-Jaya algo-
rithm, Solution 10 of the SAP-Rao algorithms, Solution 5 of the ERao-1 algorithm,

Table 7.65 Pareto-optimal solutions obtained by the Jaya algorithm for the solar-assisted Carnot-
like heat engine case study
Solution φ β τ γ η W E
1 0.3344 0.3671 0.2 0.8 0.6127 0.1738 0.1206
2 0.3800 0.5488 0.2 0.8 0.5000 0.2000 0.0800
3 0.1000 0.1016 0.2 0.8 0.7222 0.0578 0.0515
4 0.3166 0.1000 0.2 0.8 0.7049 0.0862 0.0746
5 0.1475 0.1051 0.2 0.8 0.7177 0.0657 0.0582
6 0.1000 0.3144 0.2 0.8 0.6635 0.1364 0.1084
7 0.2483 0.4250 0.2 0.8 0.5995 0.1802 0.1200
8 0.4796 0.1000 0.2 0.8 0.6899 0.1069 0.0899
9 0.4308 0.1417 0.2 0.8 0.6826 0.1159 0.0960
10 0.1182 0.4289 0.2 0.8 0.6177 0.1710 0.1205
11 0.1000 0.3543 0.2 0.8 0.6496 0.1489 0.1144
12 0.1206 0.1000 0.2 0.8 0.7210 0.0600 0.0534
13 0.3207 0.3161 0.2 0.8 0.6358 0.1595 0.1183
14 0.2718 0.4955 0.2 0.8 0.5586 0.1938 0.1100
15 0.1781 0.3734 0.2 0.8 0.6323 0.1619 0.1190
16 0.2103 0.1000 0.2 0.8 0.7138 0.0722 0.0635
17 0.1955 0.4792 0.2 0.8 0.5816 0.1873 0.1169
18 0.1000 0.1543 0.2 0.8 0.7097 0.0788 0.0688
19 0.2714 0.2914 0.2 0.8 0.6514 0.1474 0.1138
20 0.4170 0.1000 0.2 0.8 0.6958 0.0991 0.0843
21 0.3563 0.1000 0.2 0.8 0.7014 0.0914 0.0785
22 0.3762 0.4403 0.2 0.8 0.5697 0.1910 0.1138
23 0.2705 0.2262 0.2 0.8 0.6734 0.1264 0.1026
24 0.2181 0.1000 0.2 0.8 0.7132 0.0732 0.0643
25 0.2210 0.2560 0.2 0.8 0.6692 0.1307 0.1052
236 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.66 Pareto-optimal solutions obtained by the Rao-1 algorithm for the solar-assisted Carnot-
like heat engine case study
Solution φ β τ γ η W E
1 0.5466 0.4923 0.2 0.8 0.5000 0.2000 0.0800
2 0.5261 0.3021 0.2 0.8 0.6127 0.1738 0.1206
3 0.1004 0.1000 0.2 0.8 0.7226 0.0572 0.0510
4 0.2143 0.2977 0.2 0.8 0.6560 0.1434 0.1119
5 0.1000 0.1236 0.2 0.8 0.7171 0.0667 0.0590
6 0.2363 0.1181 0.2 0.8 0.7072 0.0827 0.0719
7 0.5494 0.1198 0.2 0.8 0.6771 0.1223 0.1001
8 0.3214 0.1136 0.2 0.8 0.7009 0.0921 0.0790
9 0.2318 0.1034 0.2 0.8 0.7112 0.0764 0.0669
10 0.2140 0.1000 0.2 0.8 0.7135 0.0727 0.0639
11 0.3385 0.3338 0.2 0.8 0.6263 0.1658 0.1199
12 0.1000 0.3946 0.2 0.8 0.6344 0.1605 0.1186
13 0.2764 0.1389 0.2 0.8 0.6982 0.0959 0.0819
14 0.6000 0.4443 0.2 0.8 0.5212 0.1993 0.0927
15 0.3442 0.2901 0.2 0.8 0.6428 0.1543 0.1166
16 0.4894 0.4387 0.2 0.8 0.5486 0.1958 0.1061
17 0.3999 0.4557 0.2 0.8 0.5565 0.1942 0.1093
18 0.2168 0.1730 0.2 0.8 0.6944 0.1010 0.0857
19 0.4068 0.1174 0.2 0.8 0.6920 0.1042 0.0880
20 0.5813 0.4422 0.2 0.8 0.5268 0.1988 0.0957
21 0.6000 0.3572 0.2 0.8 0.5736 0.1898 0.1149
22 0.1000 0.2380 0.2 0.8 0.6873 0.1103 0.0922
23 0.4614 0.3902 0.2 0.8 0.5810 0.1875 0.1168
24 0.6000 0.4189 0.2 0.8 0.5378 0.1975 0.1012
25 0.2039 0.4481 0.2 0.8 0.5957 0.1819 0.1195

Solution 11 of the ERao-2, and Solution 9 of the ERao-3 algorithm are identified as
the best solutions from the respective algorithm Pareto-front.
In Table 7.74, the solutions with the best average rank from the Pareto-optimal
solutions of the various algorithms are compared with the NSGA-II (TOPSIS,
LINMAP, and Fuzzy) algorithm solutions. Figure 7.5 presents the Pareto-fronts
achieved by the AMTPG-Jaya, SAP-Rao, and elitist Rao algorithms, including
the optimal solutions reported for the NSGA-II, Jaya, and Rao algorithms. The
conflicting nature of these objectives can be observed in Fig. 7.5. Furthermore, the
ecological function of the system for the Rao-2 algorithm solution is highest, which is
2.7%, 2.6%, 3.9%, 0.6%, 1%, 1.3%, 0.5%, 0.4%, 0.4%, 0.3%, and 0.3% higher than
that of the NSGA-II (TOPSIS, LINMAP, and Fuzzy), Jaya, Rao-1, Rao-3, AMTPG-
Jaya, SAP-Rao, ERao-1, ERao-2, and ERao-3 algorithms solutions, respectively.
7.3 Optimization of a Solar-Assisted Carnot-like Heat Engine System 237

Table 7.67 Pareto-optimal solutions obtained by the Rao-2 algorithm for the solar-assisted Carnot-
like heat engine case study
Solution φ β τ γ η W E
1 0.1777 0.4201 0.2 0.8 0.6127 0.1738 0.1206
2 0.2288 0.6000 0.2 0.8 0.5000 0.2000 0.0800
3 0.1030 0.1000 0.2 0.8 0.7224 0.0575 0.0513
4 0.1000 0.3238 0.2 0.8 0.6604 0.1394 0.1099
5 0.1000 0.2395 0.2 0.8 0.6868 0.1108 0.0926
6 0.1079 0.3854 0.2 0.8 0.6369 0.1587 0.1181
7 0.1269 0.5599 0.2 0.8 0.5496 0.1956 0.1065
8 0.2780 0.1000 0.2 0.8 0.7082 0.0812 0.0707
9 0.1000 0.2209 0.2 0.8 0.6921 0.1041 0.0879
10 0.2523 0.4385 0.2 0.8 0.5924 0.1833 0.1190
11 0.1130 0.1698 0.2 0.8 0.7046 0.0866 0.0749
12 0.1642 0.4969 0.2 0.8 0.5779 0.1885 0.1161
13 0.1000 0.2910 0.2 0.8 0.6712 0.1287 0.1040
14 0.1000 0.1411 0.2 0.8 0.7129 0.0736 0.0646
15 0.1574 0.5209 0.2 0.8 0.5662 0.1919 0.1127
16 0.1929 0.1175 0.2 0.8 0.7110 0.0767 0.0671
17 0.1000 0.2565 0.2 0.8 0.6818 0.1169 0.0966
18 0.3980 0.1000 0.2 0.8 0.6976 0.0967 0.0825
19 0.2412 0.5087 0.2 0.8 0.5569 0.1941 0.1094
20 0.1987 0.6000 0.2 0.8 0.5074 0.1999 0.0846
21 0.1000 0.4163 0.2 0.8 0.6256 0.1663 0.1199
22 0.1990 0.5616 0.2 0.8 0.5334 0.1981 0.0991
23 0.1643 0.6000 0.2 0.8 0.5156 0.1996 0.0895
24 0.1983 0.3087 0.2 0.8 0.6540 0.1451 0.1127
25 0.1815 0.1000 0.2 0.8 0.7162 0.0683 0.0603

However, these solutions are non-dominated. Hence, to identify the best solution
among the solutions of different algorithms, which has the best compromise among
the objectives, the MADM methods based average ranks are calculated and presented
in Table 7.75.
The decision-making methods ranking of the Pareto-optimal solutions obtained
by different algorithms is shown in Table 7.75. Spearman’s correlation coefficients
for different pairs of rankings given by decision-making methods for different algo-
rithm’s solutions are shown in Table 7.76. The ranks given by the SAW and COPRAS
methods for each solution are identical. Similarly, the ranks given by the VIKOR,
PROMETHEE, and GRA methods are equal. The ranks given by the TOPSIS and
MTOPSIS methods for each solution are identical. Furthermore, Spearman’s corre-
lation for all the pairs formed by all the MADM (except the pairs with TOPSIS and
238 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.68 Pareto-optimal solutions obtained by the Rao-3 algorithm for the solar-assisted Carnot-
like heat engine case study
Solution φ β τ γ η W E
1 0.2290 0.6000 0.2 0.8 0.5000 0.2000 0.0800
2 0.1000 0.4464 0.2 0.8 0.6127 0.1738 0.1206
3 0.1000 0.1000 0.2 0.8 0.7226 0.0571 0.0510
4 0.1886 0.2961 0.2 0.8 0.6595 0.1402 0.1103
5 0.1866 0.2409 0.2 0.8 0.6777 0.1217 0.0997
6 0.2764 0.3532 0.2 0.8 0.6270 0.1654 0.1198
7 0.1094 0.3804 0.2 0.8 0.6387 0.1574 0.1177
8 0.3035 0.1000 0.2 0.8 0.7060 0.0845 0.0733
9 0.1270 0.2234 0.2 0.8 0.6888 0.1083 0.0908
10 0.1165 0.1000 0.2 0.8 0.7213 0.0594 0.0529
11 0.2725 0.1270 0.2 0.8 0.7017 0.0909 0.0781
12 0.1000 0.3043 0.2 0.8 0.6669 0.1331 0.1066
13 0.2274 0.1630 0.2 0.8 0.6962 0.0986 0.0839
14 0.1000 0.5318 0.2 0.8 0.5708 0.1907 0.1141
15 0.1669 0.1000 0.2 0.8 0.7173 0.0663 0.0587
16 0.1269 0.2114 0.2 0.8 0.6922 0.1040 0.0878
17 0.4685 0.3646 0.2 0.8 0.5927 0.1831 0.1191
18 0.1699 0.4763 0.2 0.8 0.5875 0.1852 0.1182
19 0.1000 0.5928 0.2 0.8 0.5349 0.1979 0.0998
20 0.1412 0.5475 0.2 0.8 0.5541 0.1948 0.1083
21 0.1182 0.6000 0.2 0.8 0.5262 0.1988 0.0954
22 0.1000 0.2879 0.2 0.8 0.6722 0.1277 0.1034
23 0.1729 0.6000 0.2 0.8 0.5136 0.1997 0.0883
24 0.1307 0.5093 0.2 0.8 0.5773 0.1887 0.1159
25 0.1000 0.1416 0.2 0.8 0.7128 0.0739 0.0648

MTOPSIS) methods is greater than 0.5. The TOPSIS and MTOPSIS methods have
Spearman’s correlation values less than 0.5 with the VIKOR, PROMETHEE, and
GRA methods. Hence, the ranks suggested by the TOPSIS and MTOPSIS methods
cannot be considered for calculating the average ranks. Now, the corrected average
ranks are calculated, excluding the ranks suggested by the TOPSIS and MTOPSIS
methods and presented in Table 7.75 as the corrected ranks.
The AMTPG-Jaya algorithm solution has the least average rank, which is 3.5.
Thus, it can be regarded as the best solution. The SAP-Rao, ERao-1, and ERao-
3 algorithm’s solutions have the next best and the same average rank, which is 4.
The SAP-Rao and ERao-3 algorithms solutions are ranked three (least among the
ranks suggested to these three solutions) by three methods each, whereas the ERao-1
algorithm solution ranked four by all the methods. Hence, the ERao-1 solution is
7.3 Optimization of a Solar-Assisted Carnot-like Heat Engine System 239

Table 7.69 Pareto-optimal solutions obtained by the AMTPG-Jaya algorithm for the solar-assisted
Carnot-like heat engine case study
Solution φ β τ γ η W E
1 0.2288 0.6000 0.2 0.8 0.5000 0.2000 0.0800
2 0.2731 0.3878 0.2 0.8 0.6127 0.1738 0.1206
3 0.1017 0.1000 0.2 0.8 0.7225 0.0573 0.0512
4 0.1000 0.2429 0.2 0.8 0.6858 0.1121 0.0934
5 0.2844 0.2553 0.2 0.8 0.6624 0.1375 0.1090
6 0.1836 0.5552 0.2 0.8 0.5407 0.1971 0.1026
7 0.2301 0.2368 0.2 0.8 0.6743 0.1253 0.1020
8 0.1274 0.1698 0.2 0.8 0.7034 0.0884 0.0763
9 0.1080 0.5362 0.2 0.8 0.5670 0.1917 0.1129
10 0.1624 0.1850 0.2 0.8 0.6962 0.0986 0.0839
11 0.1005 0.2241 0.2 0.8 0.6912 0.1053 0.0888
12 0.2294 0.3279 0.2 0.8 0.6432 0.1540 0.1165
13 0.2123 0.1000 0.2 0.8 0.7137 0.0724 0.0637
14 0.2458 0.5074 0.2 0.8 0.5568 0.1942 0.1094
15 0.1578 0.6000 0.2 0.8 0.5171 0.1995 0.0904
16 0.2379 0.1000 0.2 0.8 0.7116 0.0759 0.0664
17 0.1000 0.1578 0.2 0.8 0.7088 0.0802 0.0699
18 0.2245 0.2392 0.2 0.8 0.6742 0.1255 0.1021
19 0.2073 0.4684 0.2 0.8 0.5851 0.1861 0.1177
20 0.2121 0.4724 0.2 0.8 0.5822 0.1871 0.1171
21 0.1013 0.4741 0.2 0.8 0.5999 0.1801 0.1200
22 0.2931 0.1000 0.2 0.8 0.7069 0.0832 0.0722
23 0.4032 0.4980 0.2 0.8 0.5298 0.1985 0.0973
24 0.3183 0.1374 0.2 0.8 0.6948 0.1006 0.0854
25 0.1196 0.3905 0.2 0.8 0.6334 0.1612 0.1188

assigned with a final rank of four in Table 7.75. The SAP-Rao and ERao-3 solutions
are ranked 3 and 5 by three different methods each. Thus, the values of the objectives
are compared for these two solutions. The ERao-3 solution has better values in two
objectives (η and E), and the SAP-Rao solution has better value in one objective
(W ). Hence, the ERao-3 and SAP-Rao solutions are assigned with final ranks of 2
and 3, respectively.
Here an observation can be made that the AMTPG-Jaya algorithm solution has a
better compromise among the three objectives. The non-dimensional power output
for the AMTPG-Jaya algorithm solution is 17.6%, 3.62%, 0.47%, 0.49%, 1.09%, and
0.73% higher than that of the NSGA-II (Fuzzy), Rao-2, SAP-Rao, ERao-1, ERao-2,
and ERao-3 solutions, respectively. Also, thermal efficiency for the AMTPG-Jaya
solution is 2.7%, 2.6%, 0.1%, 0.7%, and 1.2% higher than that of the NSGA-II
240 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.70 Pareto-optimal solutions obtained by the SAP-Rao algorithm for the solar-assisted
Carnot-like heat engine case study
Solution φ β τ γ η W E
1 0.6000 0.2770 0.2 0.8 0.6127 0.1737 0.1206
2 0.6000 0.4743 0.2 0.8 0.5000 0.2000 0.0800
3 0.1000 0.1000 0.2 0.8 0.7226 0.0571 0.0510
4 0.1000 0.2131 0.2 0.8 0.6943 0.1012 0.0858
5 0.1000 0.3861 0.2 0.8 0.6377 0.1581 0.1179
6 0.1555 0.2192 0.2 0.8 0.6873 0.1103 0.0922
7 0.1782 0.1212 0.2 0.8 0.7113 0.0762 0.0667
8 0.4453 0.4081 0.2 0.8 0.5745 0.1896 0.1151
9 0.6000 0.4314 0.2 0.8 0.5298 0.1985 0.0972
10 0.4829 0.3409 0.2 0.8 0.6017 0.1792 0.1202
11 0.5922 0.4537 0.2 0.8 0.5166 0.1995 0.0901
12 0.1941 0.1371 0.2 0.8 0.7060 0.0846 0.0733
13 0.1661 0.3253 0.2 0.8 0.6520 0.1469 0.1135
14 0.1192 0.1831 0.2 0.8 0.7006 0.0924 0.0793
15 0.4147 0.4699 0.2 0.8 0.5450 0.1964 0.1045
16 0.3740 0.1590 0.2 0.8 0.6832 0.1153 0.0955
17 0.1000 0.2944 0.2 0.8 0.6701 0.1299 0.1047
18 0.1346 0.3916 0.2 0.8 0.6309 0.1629 0.1192
19 0.1000 0.2769 0.2 0.8 0.6756 0.1239 0.1011
20 0.5982 0.1618 0.2 0.8 0.6580 0.1416 0.1110
21 0.1000 0.1342 0.2 0.8 0.7146 0.0709 0.0624
22 0.5381 0.3933 0.2 0.8 0.5654 0.1921 0.1124
23 0.3376 0.4163 0.2 0.8 0.5891 0.1846 0.1185
24 0.4408 0.4473 0.2 0.8 0.5533 0.1949 0.1080
25 0.1938 0.2272 0.2 0.8 0.6811 0.1177 0.0972

(TOPSIS and LINMAP), Jaya, Rao-1, and Rao-3 algorithms solutions, respectively.
In addition, the ecological function of the AMTPG-Jaya algorithm solution is 2.2%,
2%, 3.3%, 0.4%, and 0.8% higher when compared to that of the NSGA-II (TOPSIS,
LINMAP, and Fuzzy), Rao-1, and Rao-3 algorithms solutions, respectively. Hence,
the AMTPG-Jaya algorithm solution can be considered as the best solution.
Now, the performances of the Jaya algorithm, Rao algorithms, and their modified
versions in the MOO scenario are evaluated based on the hypervolume, coverage,
and spacing indicators. Table 7.77 presents the hypervolume and spacing values of
the Pareto-fronts obtained by different algorithms in the MOO scenario. Here, an
observation can be made that the ERao-1 algorithm’s Pareto-front has the highest
hypervolume value, and the Rao-2 algorithm’s Pareto-front has the least spacing
value. The ERao-1 algorithm has 3.79%, 0.12%, 0.13%, 0.1%, and 0.13% higher
7.3 Optimization of a Solar-Assisted Carnot-like Heat Engine System 241

Table 7.71 Pareto-optimal solutions obtained by the ERao-1 algorithm for the solar-assisted
Carnot-like heat engine case study
Solution φ β τ γ η W E
1 0.1000 0.4463 0.2 0.8 0.6128 0.1737 0.1206
2 0.2288 0.6000 0.2 0.8 0.5000 0.2000 0.0800
3 0.1000 0.1000 0.2 0.8 0.7226 0.0571 0.0510
4 0.1000 0.1747 0.2 0.8 0.7045 0.0868 0.0750
5 0.3586 0.3828 0.2 0.8 0.6018 0.1792 0.1202
6 0.1014 0.1523 0.2 0.8 0.7101 0.0783 0.0683
7 0.1465 0.1208 0.2 0.8 0.7140 0.0718 0.0632
8 0.1905 0.3015 0.2 0.8 0.6575 0.1421 0.1113
9 0.1003 0.2850 0.2 0.8 0.6730 0.1268 0.1028
10 0.1000 0.6000 0.2 0.8 0.5303 0.1984 0.0975
11 0.1130 0.1000 0.2 0.8 0.7216 0.0589 0.0525
12 0.1257 0.5306 0.2 0.8 0.5667 0.1918 0.1128
13 0.1000 0.2406 0.2 0.8 0.6865 0.1113 0.0929
14 0.1614 0.1000 0.2 0.8 0.7178 0.0655 0.0580
15 0.2520 0.3577 0.2 0.8 0.6286 0.1644 0.1196
16 0.1000 0.1996 0.2 0.8 0.6979 0.0962 0.0822
17 0.3128 0.2365 0.2 0.8 0.6655 0.1345 0.1073
18 0.2119 0.3173 0.2 0.8 0.6493 0.1492 0.1145
19 0.2863 0.5131 0.2 0.8 0.5452 0.1964 0.1046
20 0.1000 0.1480 0.2 0.8 0.7113 0.0764 0.0668
21 0.5550 0.3558 0.2 0.8 0.5824 0.1870 0.1171
22 0.3733 0.5284 0.2 0.8 0.5162 0.1996 0.0898
23 0.2037 0.2812 0.2 0.8 0.6629 0.1370 0.1087
24 0.4408 0.4449 0.2 0.8 0.5548 0.1946 0.1086
25 0.3082 0.4503 0.2 0.8 0.5767 0.1889 0.1158

hypervolume value when compared to that of the Jaya, Rao-1, Rao-2, Rao-3, and
AMTPG-Jaya algorithms, respectively. The ERao-3, ERao-2, SAP-Rao, and Rao-3
algorithms have achieved the next better hypervolume values, respectively. Similarly,
the Rao-2 algorithm has 71%, 30.5%, 2.5%, 46.5%, 26.4%, 41%, 21%, and 29% less
spacing value when compared to that of the Jaya, Rao-1, Rao-3, AMTPG-Jaya, SAP-
Rao, ERao-1, ERao-2, and ERao-3 algorithms, respectively. The performances of the
proposed modified algorithms are better or competitive to that of the basic algorithms
in terms of the hypervolume and spacing values.
The coverage values of the Pareto-fronts obtained by the Jaya algorithm, Rao algo-
rithms, and their modified versions are presented in Table 7.78. The AMTPG-Jaya,
Rao-1, and ERao-2 algorithms have achieved better coverage values than the other
algorithms compared. Not a single solution of these algorithms is dominated by the
242 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.72 Pareto-optimal solutions obtained by the ERao-2 algorithm for the solar-assisted
Carnot-like heat engine case study
Solution φ β τ γ η W E
1 0.2287 0.6000 0.2 0.8 0.5000 0.2000 0.0800
2 0.3869 0.3493 0.2 0.8 0.6127 0.1738 0.1206
3 0.1027 0.1000 0.2 0.8 0.7224 0.0575 0.0513
4 0.5153 0.1000 0.2 0.8 0.6865 0.1113 0.0929
5 0.1456 0.1000 0.2 0.8 0.7190 0.0634 0.0563
6 0.4480 0.1000 0.2 0.8 0.6930 0.1030 0.0871
7 0.2696 0.2556 0.2 0.8 0.6640 0.1360 0.1081
8 0.6000 0.1819 0.2 0.8 0.6507 0.1480 0.1140
9 0.6000 0.2338 0.2 0.8 0.6310 0.1628 0.1192
10 0.2019 0.1000 0.2 0.8 0.7145 0.0710 0.0625
11 0.3902 0.3673 0.2 0.8 0.6041 0.1781 0.1203
12 0.2940 0.1000 0.2 0.8 0.7068 0.0833 0.0723
13 0.1794 0.5347 0.2 0.8 0.5540 0.1948 0.1083
14 0.6000 0.4520 0.2 0.8 0.5159 0.1996 0.0897
15 0.6000 0.4421 0.2 0.8 0.5226 0.1991 0.0935
16 0.4196 0.4027 0.2 0.8 0.5819 0.1872 0.1170
17 0.3379 0.1000 0.2 0.8 0.7030 0.0890 0.0767
18 0.2696 0.5276 0.2 0.8 0.5398 0.1972 0.1022
19 0.6000 0.4245 0.2 0.8 0.5342 0.1980 0.0995
20 0.2216 0.3414 0.2 0.8 0.6391 0.1571 0.1176
21 0.6000 0.1000 0.2 0.8 0.6779 0.1214 0.0995
22 0.6000 0.1038 0.2 0.8 0.6767 0.1227 0.1004
23 0.3539 0.4550 0.2 0.8 0.5658 0.1920 0.1125
24 0.3881 0.1068 0.2 0.8 0.6967 0.0980 0.0835
25 0.2353 0.4524 0.2 0.8 0.5883 0.1848 0.1183

Pareto-fronts of the other algorithms. In addition, one solution from the Pareto-fronts
of the Jaya, Rao-2, SAP-Rao, ERao-1, and ERao-3 algorithms are dominated. Simi-
larly, two solutions from the Pareto-front achieved by the Rao-3 algorithm are domi-
nated. The performance of the proposed modified algorithms is better or competitive
to that of the basic algorithms in terms of the coverage values.
From the computational results of the solar-assisted Carnot-like heat engine case
study, it can be observed that the performance of the considered system can be
improved with the solutions achieved by the proposed algorithms. The solution of the
AMTPG-Jaya algorithm has better compromise among the non-dimensional power
output, efficiency, and ecological function. The AMTPG-Jaya algorithm’s solution
has 17% higher power output, 3.3% higher ecological function, and 2.7% higher
7.3 Optimization of a Solar-Assisted Carnot-like Heat Engine System 243

Table 7.73 Pareto-optimal solutions obtained by the ERao-3 algorithm for the solar-assisted
Carnot-like heat engine case study
Solution φ β τ γ η W E
1 0.2288 0.6000 0.2 0.8 0.5000 0.2000 0.0800
2 0.2245 0.4042 0.2 0.8 0.6127 0.1737 0.1206
3 0.1000 0.1000 0.2 0.8 0.7226 0.0571 0.0510
4 0.1000 0.1607 0.2 0.8 0.7081 0.0814 0.0708
5 0.1065 0.1000 0.2 0.8 0.7221 0.0580 0.0517
6 0.1915 0.1000 0.2 0.8 0.7154 0.0696 0.0614
7 0.1000 0.3289 0.2 0.8 0.6586 0.1410 0.1108
8 0.2270 0.1000 0.2 0.8 0.7125 0.0744 0.0653
9 0.2398 0.4211 0.2 0.8 0.6027 0.1787 0.1202
10 0.4158 0.1000 0.2 0.8 0.6960 0.0990 0.0842
11 0.2133 0.2251 0.2 0.8 0.6797 0.1193 0.0982
12 0.2541 0.5299 0.2 0.8 0.5416 0.1970 0.1030
13 0.1838 0.5257 0.2 0.8 0.5584 0.1938 0.1100
14 0.2167 0.5709 0.2 0.8 0.5233 0.1991 0.0938
15 0.1634 0.1559 0.2 0.8 0.7038 0.0878 0.0758
16 0.1973 0.5607 0.2 0.8 0.5344 0.1980 0.0996
17 0.3347 0.3528 0.2 0.8 0.6189 0.1703 0.1205
18 0.1000 0.2919 0.2 0.8 0.6709 0.1290 0.1042
19 0.2045 0.5966 0.2 0.8 0.5084 0.1999 0.0852
20 0.2121 0.3277 0.2 0.8 0.6454 0.1523 0.1158
21 0.2354 0.3606 0.2 0.8 0.6297 0.1637 0.1194
22 0.3118 0.1605 0.2 0.8 0.6889 0.1082 0.0908
23 0.1494 0.4904 0.2 0.8 0.5839 0.1865 0.1174
24 0.3490 0.1000 0.2 0.8 0.7020 0.0904 0.0778
25 0.3667 0.4445 0.2 0.8 0.5692 0.1911 0.1136

thermal efficiency when compared to those of the solutions achieved by other algo-
rithms compared. In addition, the performances of the modified versions are better
or competitive to those of the basic algorithms as well as the NSGA-II algorithm.
The next section presents the Jaya and Rao algorithms’ application to the selected
bio-energy systems, along with their modified versions.
244 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.74 Best solutions obtained by various algorithms in MOO scenario of the solar-assisted
Carnot-like heat engine case study
Algorithm φ β τ γ η W E
TOPSIS NSGA-II 0.4447 0.3903 0.2 0.8 0.583877 0.186475 0.117451
LINMAP 0.3684 0.4149 0.2 0.8 0.584513 0.186248 0.117585
Fuzzy 0.4401 0.2533 0.2 0.8 0.644401 0.153089 0.116124
Jaya 0.248 0.425 0.20 0.80 0.599529 0.180211 0.119952
Rao-1 0.204 0.448 0.20 0.8 0.595680 0.181886 0.119499
Rao-2 0.178 0.420 0.20 0.80 0.612702 0.173763 0.120645
Rao-3 0.468 0.365 0.20 0.8 0.592698 0.183122 0.119073
AMTPG-Jaya 0.101 0.474 0.20 0.80 0.599867 0.180060 0.119987
SAP-Rao 0.483 0.341 0.20 0.799999937 0.601716 0.179219 0.120160
ERao-1 0.359 0.383 0.20 0.8 0.601801 0.179179 0.120168
ERao-2 0.390 0.367 0.20 0.8 0.604056 0.178123 0.120343
ERao-3 0.240 0.421 0.20 0.80 0.602729 0.178749 0.120245
Source NSGA-II—Sayyaadi et al. (2015)
Bold figure indicates a better value

Fig. 7.5 Plot of Pareto-fronts of the proposed algorithms and solutions of the methods compared
in solar-assisted Carnot-like heat engine case study
Table 7.75 Ranks suggested by the MADM methods for different algorithm solutions presented in Table 7.74
Algorithm SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA AR CR FR
TOPSIS 11 11 10 10 11 11 11 11 10.8 11 11
LINMAP 10 10 9 9 10 10 10 10 9.8 10 10
Fuzzy 12 12 12 12 12 12 12 12 12.0 12 12
Jaya 2 2 3 3 7 7 2 7 4.1 4.5 6
Rao-1 6 7 1 1 8 8 6 8 5.6 7.2 8
Rao-2 9 9 11 11 1 1 9 1 6.5 5 7
Rao-3 8 8 2 2 9 9 8 9 6.9 8.5 9
AMTPG-Jaya 1 1 4 4 6 6 1 6 3.6 3.5 1
SAP-Rao 3 3 5 5 5 5 3 5 4.3 4 3
ERao-1 4 4 6 6 4 4 4 4 4.5 4 4
ERao-2 7 6 8 8 2 2 7 2 5.3 4.3 5
7.3 Optimization of a Solar-Assisted Carnot-like Heat Engine System

ERao-3 5 5 7 7 3 3 5 3 4.8 4 2
AR average rank, CR corrected rank, FR final rank
245
246 7 Optimization of the Selected Solar-Assisted Energy Systems

Table 7.76 Spearman’s rank correlation coefficients between different pairs of MADM method’s
rankings presented in Table 7.75
Method SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA
SAW 1 0.9 0.7 0.7 0.5 0.5 1 0.5
WPM 0.9 1 0.6 0.6 0.5 0.5 0.9 0.5
TOPSIS 0.7 0.6 1 1 0.0 0.0 0.7 0.0
MTOPSIS 0.7 0.6 1 1 0.0 0.0 0.7 0.0
VIKOR 0.5 0.5 0.0 0.0 1 1 0.5 1
PROMETHEE 0.5 0.5 0.0 0.0 1 1 0.5 1
COPRAS 1 0.9 0.7 0.7 0.5 0.5 1 0.5
GRA 0.5 0.5 0.0 0.0 1 1 0.5 1

Table 7.77 Hypervolume

Algorithm Hypervolume Spacing
and spacing values of the
Pareto-fronts obtained by the Jaya 4.39053E-03 0.139464
proposed algorithms in MOO Rao-1 4.55133E-03 0.057768
for the solar-assisted
Rao-2 4.55104E-03 0.040155
Carnot-like engine case study
Rao-3 4.55232E-03 0.041179
AMTPG-Jaya 4.55097E-03 0.074990
SAP-Rao 4.55484E-03 0.054526
ERao-1 4.55675E-03 0.068134
ERao-2 4.55517E-03 0.050868
ERao-3 4.55632E-03 0.056517
Results in the bold figure indicate better values

Table 7.78 Coverage (%) values of the Pareto-fronts obtained by the proposed algorithms in MOO
for the solar-assisted Carnot-like heat engine case study
Algorithm Jaya Rao-1 Rao-2 Rao-3 AMTPG-Jaya SAP-Rao ERao-1 ERao-2 ERao-3
Jaya – 0 0 4 0 4 0 0 0
Rao-1 0 – 0 4 0 4 0 0 0
Rao-2 4 0 – 4 0 4 0 0 0
Rao-3 0 0 0 – 0 0 0 0 0
AMTPG-Jaya 4 0 4 4 – 4 0 0 0
SAP-Rao 0 0 0 4 0 – 0 0 0
ERao-1 0 0 0 8 0 4 – 0 4
ERao-2 0 0 0 4 0 0 0 – 0
ERao-3 0 0 0 8 0 4 4 0 –
References 247

References

Ahmadi, M. H., Mohammadi, A. H., Dehghani, S., & Barranco-Jiménez, M. A. (2013a). Multi-
objective thermodynamic-based optimization of output power of solar dish-Stirling engine by
implementing an evolutionary algorithm. Energy Conversion and Management, 75, 438–445.
https://doi.org/10.1016/j.enconman.2013.06.030
Ahmadi, M. H., Sayyaadi, H., Mohammadi, A. H., & Barranco-Jimenez, M. A. (2013b). Thermo-
economic multi-objective optimization of solar dish-Stirling engine by implementing evolu-
tionary algorithm. Energy Conversion and Management, 73, 370–380. https://doi.org/10.1016/j.
enconman.2013.05.031
Dai, D., Yuan, F., Long, R., Liu, Z., & Liu, W. (2018). Performance analysis and multi-objective opti-
mization of a Stirling engine based on MOPSOCD. International Journal of Thermal Sciences,
124, 399–406. https://doi.org/10.1016/j.ijthermalsci.2017.10.030
Li, Y., Liao, S., & Liu, G. (2015). Thermo-economic multi-objective optimization for a solar-dish
Brayton system using NSGA-II and decision making. International Journal of Electrical Power &
Energy Systems, 64, 167–175. https://doi.org/10.1016/j.ijepes.2014.07.027
Sayyaadi, H., Ahmadi, M. H., & Dehghani, S. (2015). Optimal design of a solar-driven heat engine
based on thermal and ecological criteria. Journal of Energy Engineering, 141(3), 4014012. https://
doi.org/10.1061/(asce)ey.1943-7897.0000191
Chapter 8
Optimization of the Selected Bio-Energy
Systems

Abstract This chapter presents the applications of different versions of Jaya and
Rao algorithms to the optimization problems of the biodiesel engine system and
microalgae-based biomass cultivation process. Three multiobjective optimization
case studies of biodiesel engine design and a multiobjective optimization case study
of microalgae cultivation process optimization are considered for optimization to
see if there can be any improvement in the performances of the selected systems.
The optimization is carried out using the Jaya algorithm, Rao algorithms, and their
modified versions. Furthermore, to identify the best solutions from the Pareto-fronts,
the average rank method described in Chap. 5 is employed. Computational results
revealed that the performances of the modified Jaya and Rao algorithms are superior
to those of the other algorithms. Also, the performances of the selected systems are
improved by the solutions of the proposed algorithms.

8.1 Design Optimization of the Single-Cylinder

Direct-Injection Diesel Engine

The description of the selected single-cylinder direct-injection diesel engine system

is presented in Sect. 2.3.1. The detailed description and specifications of this system
were presented by Dhingra et al. (2014). This is a case study of a single-cylinder
direct-injection compression ignition engine that runs using Jatropha biodiesel
blends. The seven objectives of this case study are: minimizing the combustion
parameters such as brake specific fuel consumption (BSFC—kg/kWh) and peak
cylinder pressure (Pmax —bar); maximizing the performance in terms of brake-
thermal efficiency (BTE—N-m); and minimizing the emissions such as carbon
mono oxide emission (CO—%), nitrogen oxides emission (NOx —ppm), hydro-
carbon emission (HC—ppm), and smoke emission opacity (S m ). The design vari-
ables are biodiesel blending ratio (X 1 ), load torque (X 2 ), and compression ratio (X 3 ),
and their ranges are 11.25 ≤ X 1 ≤ 33.75 (% V/V ), 7.5 ≤ X 2 ≤ 12.5 (N-m), and
13.5 ≤ X 3 ≤ 16.5 (V/V ). The regression models of this case study’s objectives
proposed by Dhingra et al. (2014) are as given below:

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 249
V. R. Ravipudi et al., Design Optimization of Renewable Energy Systems Using Advanced
Optimization Algorithms, Green Energy and Technology,
https://doi.org/10.1007/978-3-030-95589-2_8
250 8 Optimization of the Selected Bio-Energy Systems

BSFC = − 46.68493 + 0.13685X 1 + 1.32378X 2 + 5.31712X 3

− 3.21186 × 10−3 X 12 − 0.056480X 22 − 0.16911X 32
+ 2.33817 × 10−3 X 1 X 2 − 1.13137 × 10−3 X 1 X 3
− 0.022062X 2 X 3 (8.1)

BTE = − 2400.88522 + 10.28829X 1 + 71.43483X 2

+ 259.72937X 3 − 0.18203X 12 − 4.07476X 22
− 9.07056X 32 − 0.19143X 1 X 2 − 4.19026 × 10−3 X 1 X 3
+ 1.26091X 2 X 3 (8.2)

Pmax = − 3669.50268 + 10.41183X 1 + 129.7155X 2

+ 396.37136X 3 − 0.28479X 12 − 5.68691X 22
− 12.90809X 32 − 0.037712X 1 X 2 + 0.18856X 1 X 3
− 1.03709X 2 X 3 (8.3)

Sqrt(CO) = − 102.47076 + 0.41566X 1 + 1.86746X 2

+ 11.94069X 3 − 7.55818 × 10−3 X 12 − 0.11396X 22
− 0.40724X 32 − 3.75222 × 10−3 X 1 X 2
− 2.57927 × 10−3 X 1 X 3 + 0.035673X 2 X 3 (8.4)

Sqrt(NOx ) = − 765.06345 + 3.33115X 1 + 12.59961X 2

+ 91.17741X 3 − 0.062945X 12 − 0.70774X 22
− 3.08084X 32 − 0.026616X 1 X 2 − 0.01625X 1 X 3
+ 0.1666X 2 X 3 (8.5)

log10 (HC) = − 182.12527 + 0.57186X 1 + 3.99162X 2

+ 21.03397X 3 − 0.012787X 12 − 0.19348X 22
− 0.70153X 32 + 8.11844 × 10−5 X 1 X 2
+ 2.71724 × 10−4 X 1 X 3 + 1.69324 × 10−3 X 2 X 3 (8.6)

Sqrt(Sm ) =133.78384 − 0.22401X 1 − 3.48682X 2

− 15.04873X 3 + 0.010848X 12 + 0.18082X 22
+ 0.49577X 32 + 3.59009 × 10−3 X 1 X 2
− 5.06832 × 10−3 X 1 X 3 + 0.041469X 2 X 3 (8.7)
8.1 Design Optimization of the Single-Cylinder … 251

Dhingra et al. (2014) presented the optimum design parameters using the NSGA-
II algorithm and reported 30 Pareto-optimal solutions after 50,000 function evalua-
tions in the simultaneous optimization of these seven objectives. Now, the proposed
algorithms, along with basic Jaya and Rao algorithms, are used to find the optimal
parameters. For a fair comparison of the proposed algorithms with the NSGA-II, the
proposed and other considered algorithms are executed for 50,000 function evalua-
tions. In multi-objective optimization using the Jaya algorithm, Rao algorithms, and
their modified versions, a priori approach is followed. In multi-objective optimiza-
tion through a priori approach, the combined objective function of this case study is
given by the following equation:

BSFC BTE Pmax

Z combined = w1 − w2 + w3
BSFCmin BTEmax Pmaxmin
CO NOx HC Sm
+ w4 + w5 + w6 + w7 (8.8)
COmin NOxmin HCmin Smmin

Here, the terms BSFCmin , Pmax min , COmin , NOx min , HCmin, and Sm min represents
the objective function values achieved in single-objective optimization (minimiza-
tion) of BSFC, Pmax , CO, NOx , HC, and Sm , respectively. Similarly, the term BTEmax
represents the maximum BTE value achieved in single-objective optimization (maxi-
mization) of BTE. The weights of the objectives are varied between zero and one
randomly such that w1 + w2 + w3 + w4 + w5 + w6 + w7 = 1.
Dhingra et al. (2014) presented 30 Pareto-optimal solutions in this case study.
Hence, for a fair comparison of spacing values, in this work also 30 Pareto-optimal
solutions are found. By using different combinations of the weights, the algorithms
are executed 30 times. For each combination of the weights, a single Pareto-optimal
solution is found by executing the algorithm. The population size for all the algo-
rithms is taken as 50 in each execution. The elite population size is taken as 20%
of the population for the elitist Rao algorithms. Furthermore, the solutions obtained
by various algorithms are non-dominated. Hence, the average rank method was used
to identify the best solution from the Pareto-front of the respective algorithm. The
best solution from the Pareto-front of the NSGA-II algorithm was not identified by
Dhingra et al. (2014). Hence, the best solution from the NSGA-II Pareto-front is also
selected using the average rank method and reported in this work.
The Pareto-optimal solutions achieved by the Jaya algorithm, Rao algorithms, and
their modified versions in this case study are presented in Tables 8.1, 8.2, 8.3, 8.4,
8.5, 8.6, 8.7, 8.8 and 8.9. Similar to the previous case studies, the best solutions from
these Pareto-fronts are identified. Solution 29 of the NSGA-II algorithm, Solution
15 of the Jaya algorithm, Solution 11 of the AMTPG-Jaya algorithm, Solution 30 of
the Rao-1 algorithm, Solution 18 of the Rao-2 algorithm, Solution 22 of the Rao-3
algorithm, Solution 25 of the ERao-1 algorithm, Solution 2 of the ERao-2 algorithm,
Solution 23 of the ERao-3 algorithm, and Solution 12 of the SAP-Rao algorithm are
identified as the best solutions from the respective algorithm’s Pareto-front. Now,
these best solutions are compared with those of the NSGA-II algorithm in Table
8.10.
Table 8.1 Pareto-optimal solutions obtained by the Jaya algorithm for the single-cylinder direct-injection diesel engine case study
252

Solution X1 X2 X3 BSFC BTE Pmax CO (%) NOx HC Sm

1 16.3402 11.8641 16.2418 0.1 23.0257 38.6459 0.15229 174.967 8.14667 116.8354
2 15.6022 11.7793 16.2331 0.1 22.2531 37.9039 0.12906 166.392 7.05306 112.5002
3 15.647 11.782 16.2357 0.1 22.2842 37.9481 0.12981 166.699 7.08652 112.7368
4 17.3386 11.8013 16.2317 0.16337 25.537 43.8038 0.24389 197.422 12.8797 116.6346
5 16.2086 11.8476 16.2419 0.1 22.886 38.5201 0.14778 173.338 7.92446 116.0407
6 16.7768 11.8776 16.2369 0.11508 23.8532 40.0643 0.18045 182.992 9.52364 118.0359
7 16.2128 11.8471 16.2427 0.1 22.8829 38.5229 0.14763 173.294 7.91757 116.0551
8 15.7734 11.7953 16.2387 0.1 22.4136 38.0801 0.13344 168.072 7.25007 113.4601
9 15.761 11.795 16.2376 0.1 22.4088 38.0688 0.13336 168.03 7.24631 113.3987
10 15.7364 11.7925 16.237 0.1 22.3846 38.0435 0.13268 167.773 7.21554 113.259
11 17.208 11.8194 16.2336 0.15196 25.1413 42.9257 0.22805 193.969 11.9926 116.9767
12 15.372 11.7557 16.2265 0.1 22.0135 37.6556 0.1226 163.917 6.77064 111.198
13 17.1168 11.8318 16.2347 0.14407 24.8671 42.3166 0.21741 191.598 11.4156 117.2086
14 16.8395 11.8693 16.2367 0.12039 24.0395 40.478 0.18696 184.555 9.84479 117.887
15 15.372 11.755 16.227 0.1 22.006 37.654 0.1224 163.83 6.7597 111.187
16 15.8525 11.8039 16.2402 0.1 22.4957 38.1617 0.1358 168.956 7.3581 113.9172
17 16.8714 11.865 16.2366 0.12309 24.1343 40.6886 0.19033 185.353 10.0126 117.8107
18 15.3736 11.7551 16.2271 0.1 22.0092 37.656 0.12245 163.865 6.76381 111.1998
19 16.5586 11.8952 16.2382 0.1 23.2753 38.8494 0.16088 177.997 8.58057 118.2026
20 16.1849 11.8441 16.2423 0.1 22.8566 38.4964 0.14681 172.994 7.87793 115.8926
21 16.5661 11.8965 16.2379 0.1 23.2847 38.8562 0.16122 178.115 8.59798 118.2512
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.1 (continued)
Solution X1 X2 X3 BSFC BTE Pmax CO (%) NOx HC Sm
22 16.9288 11.8573 16.2363 0.12797 24.3052 41.0682 0.19647 186.798 10.3227 117.6724
23 15.6679 11.7846 16.2359 0.1 22.3089 37.9708 0.13052 166.964 7.11823 112.8598
24 17.0952 11.8348 16.2349 0.14221 24.8023 42.1726 0.21494 191.041 11.2836 117.2628
25 16.7624 11.8795 16.237 0.11386 23.8103 39.969 0.17897 182.633 9.45124 118.0701
26 15.5027 11.7686 16.2307 0.1 22.1468 37.7969 0.12614 165.282 6.92413 111.9316
27 16.5914 11.9003 16.2373 0.1 23.314 38.8791 0.16226 178.48 8.65208 118.4122
28 15.9692 11.8171 16.2418 0.1 22.6186 38.2807 0.13943 170.303 7.5264 114.5996
29 16.4252 11.8757 16.2409 0.1 23.1207 38.7261 0.15549 176.103 8.30633 117.3608
30 16.7865 11.8763 16.2369 0.1159 23.882 40.1282 0.18145 183.233 9.57255 118.013
8.1 Design Optimization of the Single-Cylinder …
253
Table 8.2 Pareto-optimal solutions obtained by the AMTPG-Jaya algorithm for the single-cylinder direct-injection diesel engine case study
254

Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm

1 16.589 11.8999 16.2374 0.1 23.3112 38.877 0.16216 178.445 8.64685 118.397
2 15.7072 11.7897 16.2361 0.1 22.3569 38.0136 0.13191 167.481 7.18097 113.0941
3 16.2279 11.8496 16.2423 0.1 22.9036 38.5383 0.14832 173.537 7.95092 116.1524
4 16.2905 11.8573 16.2424 0.1 22.9691 38.5982 0.15042 174.298 8.05413 116.5288
5 15.441 11.7631 16.2284 0.10001 22.0884 37.7318 0.12461 164.686 6.85748 111.5893
6 16.5741 11.8976 16.2378 0.1 23.2936 38.8634 0.16153 178.225 8.61424 118.3015
7 15.7837 11.7984 16.2374 0.1 22.4392 38.0938 0.13428 168.366 7.28817 113.5386
8 15.9675 11.8171 16.2416 0.1 22.6182 38.2792 0.13943 170.3 7.52632 114.5911
9 16.6281 11.8973 16.2371 0.10253 23.411 39.083 0.16548 179.315 8.80476 118.3853
10 16.4734 11.8826 16.2401 0.1 23.1759 38.771 0.15738 176.772 8.40211 117.6627
11 15.263 11.746 16.222 0.1 21.908 37.538 0.1199 162.85 6.6552 110.6
12 16.4961 11.8858 16.2397 0.1 23.2012 38.7919 0.15826 177.081 8.44667 117.8042
13 16.0203 11.8245 16.2413 0.1 22.6824 38.334 0.14142 171.024 7.61991 114.9138
14 15.7273 11.7903 16.2378 0.1 22.3655 38.032 0.13207 167.557 7.18806 113.1943
15 16.1045 11.8348 16.2417 0.1 22.774 38.4181 0.14425 172.057 7.75425 115.4162
16 16.5164 11.8887 16.2393 0.1 23.2248 38.8106 0.15908 177.37 8.48864 117.9325
17 16.5006 11.8862 16.2398 0.1 23.2049 38.7959 0.15838 177.124 8.45274 117.8302
18 16.1819 11.8433 16.2426 0.1 22.8501 38.4929 0.14659 172.917 7.86711 115.87
19 17.1317 11.8298 16.2345 0.14536 24.9117 42.4157 0.21912 191.983 11.5075 117.1711
20 15.9913 11.8215 16.2407 0.1 22.6546 38.3053 0.1406 170.717 7.581 114.7465
21 17.2407 11.8149 16.2332 0.1548 25.24 43.1447 0.23194 194.826 12.2076 116.8922
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.2 (continued)
Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm
22 15.9711 11.8215 16.2387 0.1 22.6505 38.2881 0.14059 170.694 7.58052 114.6506
23 17.2735 11.8103 16.2327 0.15767 25.3395 43.3655 0.23591 195.693 12.4285 116.8065
24 16.6809 11.8903 16.2371 0.10698 23.568 39.4312 0.17072 180.615 9.05316 118.2619
25 15.6502 11.7842 16.2344 0.1 22.3017 37.9546 0.1304 166.902 7.11284 112.7725
26 17.0948 11.8348 16.2349 0.14218 24.8013 42.1704 0.2149 191.032 11.2816 117.2637
27 17.2638 11.8117 16.2328 0.15682 25.3099 43.3 0.23473 195.436 12.3625 116.832
28 15.8955 11.8122 16.2382 0.1 22.5664 38.2107 0.13805 169.76 7.46188 114.2011
29 16.4863 11.8844 16.2399 0.1 23.1903 38.7829 0.15788 176.947 8.42728 117.7433
30 15.8122 11.799 16.2398 0.1 22.4504 38.1196 0.13447 168.463 7.29703 113.6792
8.1 Design Optimization of the Single-Cylinder …
255
Table 8.3 Pareto-optimal solutions obtained by the Rao-1 algorithm for the single-cylinder direct-injection diesel engine case study
256

Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm

1 15.5953 11.7824 16.2302 0.1 22.2741 37.9027 0.12984 166.649 7.0881 112.496
2 16.5377 11.8918 16.2389 0.1 23.249 38.8301 0.15993 177.669 8.53216 118.0669
3 15.8479 11.8042 16.2395 0.1 22.4972 38.1582 0.13589 168.98 7.362 113.8986
4 17.3107 11.8052 16.2321 0.16092 25.4522 43.6157 0.24045 196.679 12.684 116.7087
5 16.7671 11.8789 16.237 0.11426 23.8242 40 0.17945 182.75 9.47478 118.059
6 17.1826 11.8228 16.2339 0.14976 25.0648 42.7558 0.22505 193.306 11.8286 117.0418
7 17.3206 11.8038 16.232 0.16179 25.4822 43.6824 0.24166 196.942 12.753 116.6824
8 16.195 11.8442 16.2432 0.1 22.859 38.5051 0.14683 173.011 7.87913 115.9408
9 17.1256 11.8306 16.2346 0.14483 24.8935 42.3753 0.21842 191.826 11.4699 117.1864
10 16.4931 11.8861 16.2392 0.1 23.2028 38.7898 0.15834 177.105 8.45066 117.7928
11 17.0519 11.8407 16.2353 0.13849 24.6726 41.8845 0.21003 189.928 11.0242 117.3708
12 15.4602 11.7625 16.2308 0.1 22.0894 37.7483 0.1245 164.672 6.85244 111.6749
13 16.9412 11.8556 16.2362 0.12903 24.3423 41.1508 0.19782 187.113 10.3915 117.6421
14 15.7474 11.7975 16.2345 0.1 22.4236 38.0607 0.13399 168.227 7.27489 113.3572
15 16.0083 11.8201 16.2433 0.1 22.6484 38.318 0.14024 170.618 7.56486 114.814
16 16.9958 11.8483 16.2358 0.1337 24.5052 41.5127 0.20379 188.498 10.6986 117.5089
17 16.2472 11.8522 16.2421 0.1 22.9254 38.5572 0.14902 173.791 7.98546 116.2701
18 16.5337 11.8917 16.2386 0.1 23.2476 38.8268 0.15991 177.656 8.53059 118.0461
19 16.6524 11.8941 16.2371 0.10457 23.4833 39.2432 0.16788 179.913 8.9182 118.3286
20 15.9283 11.814 16.2401 0.10001 22.5868 38.2417 0.13856 169.966 7.4856 114.3738
21 16.1931 11.8474 16.2406 0.1 22.8814 38.507 0.14771 173.299 7.92086 115.964
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.3 (continued)
Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm
22 16.8582 11.8667 16.2367 0.12198 24.0952 40.6018 0.18894 185.024 9.94312 117.8422
23 16.7039 11.8873 16.2371 0.10892 23.6364 39.5831 0.17302 181.184 9.16379 118.2079
24 15.6407 11.7842 16.2335 0.1 22.2989 37.9461 0.13037 166.882 7.11163 112.7269
25 16.6823 11.8902 16.2371 0.1071 23.5722 39.4405 0.17086 180.65 9.0599 118.2586
26 16.7538 11.8807 16.237 0.11314 23.7848 39.9124 0.17809 182.42 9.40854 118.0903
27 16.5473 11.8931 16.2388 0.1 23.2594 38.8388 0.1603 177.797 8.55083 118.1265
28 15.8361 11.8006 16.2409 0.1 22.4681 38.1428 0.13493 168.644 7.31807 113.8086
29 17.0702 11.8382 16.2352 0.14007 24.7275 42.0064 0.2121 190.398 11.1332 117.3252
30 15.357 11.75 16.229 0.1 21.963 37.631 0.121 163.36 6.702 111.071
8.1 Design Optimization of the Single-Cylinder …
257
Table 8.4 Pareto-optimal solutions obtained by the Rao-2 algorithm for the single-cylinder direct-injection diesel engine case study
258

Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm

1 17.2824 11.8091 16.2326 0.15844 25.3663 43.4251 0.23699 195.928 12.4889 116.7832
2 17.2802 11.8094 16.2326 0.15825 25.3596 43.4103 0.23672 195.869 12.4739 116.789
3 16.5939 11.9007 16.2372 0.1 23.3171 38.8814 0.16237 178.519 8.65789 118.4283
4 16.5666 11.8964 16.238 0.1 23.2846 38.8567 0.16121 178.113 8.59767 118.2537
5 17.3282 11.8027 16.2318 0.16246 25.5053 43.7335 0.2426 197.144 12.8062 116.6623
6 17.1426 11.8283 16.2344 0.1463 24.9446 42.4887 0.22039 192.266 11.5757 117.1435
7 15.3801 11.7569 16.2265 0.10001 22.025 37.6655 0.12293 164.037 6.7847 111.2464
8 16.997 11.8481 16.2358 0.1338 24.5088 41.5206 0.20393 188.529 10.7055 117.506
9 16.5034 11.8869 16.2395 0.1 23.2104 38.7987 0.15858 177.195 8.46316 117.8515
10 16.4495 11.8782 16.2412 0.1 23.1422 38.748 0.15618 176.356 8.34172 117.5038
11 17.2064 11.8196 16.2336 0.15182 25.1366 42.9151 0.22786 193.928 11.9824 116.9807
12 16.6793 11.8905 16.2371 0.10685 23.5634 39.421 0.17056 180.577 9.04583 118.2655
13 17.0555 11.8402 16.2353 0.1388 24.6834 41.9085 0.21044 190.021 11.0456 117.3618
14 16.7719 11.8783 16.237 0.11467 23.8386 40.032 0.17995 182.87 9.49908 118.0475
15 17.1421 11.8284 16.2344 0.14626 24.943 42.4852 0.22033 192.253 11.5724 117.1448
16 15.499 11.7653 16.2328 0.1 22.1206 37.7879 0.12528 164.981 6.88591 111.8833
17 15.908 11.8079 16.2426 0.1 22.5382 38.2156 0.13693 169.399 7.41049 114.22
18 15.283 11.751 16.221 0.1 21.949 37.564 0.1211 163.3 6.708 110.732
19 17.0718 11.838 16.2352 0.14021 24.7324 42.0174 0.21229 190.441 11.143 117.3211
20 15.8931 11.8135 16.2371 0.1 22.5748 38.2104 0.13838 169.867 7.47695 114.2018
21 16.4232 11.8757 16.2407 0.1 23.1204 38.7245 0.15549 176.102 8.30639 117.3513
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.4 (continued)
Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm
22 17.225 11.817 16.2334 0.15344 25.1926 43.0395 0.23007 194.414 12.1039 116.9328
23 16.4392 11.8782 16.2403 0.1 23.1401 38.7397 0.15617 176.34 8.34054 117.4534
24 16.9269 11.8575 16.2363 0.12782 24.2997 41.0561 0.19627 186.752 10.3126 117.6768
25 16.2259 11.8492 16.2424 0.1 22.9004 38.5362 0.14821 173.499 7.94565 116.1391
26 15.4719 11.763 16.2316 0.1 22.0963 37.7596 0.12465 164.737 6.85885 111.7344
27 16.9159 11.859 16.2364 0.12688 24.267 40.9833 0.19509 186.474 10.2525 117.7034
28 16.6361 11.8963 16.2371 0.1032 23.4349 39.1359 0.16627 179.513 8.84205 118.3666
29 16.6239 11.8979 16.2371 0.10218 23.3988 39.0558 0.16508 179.214 8.78565 118.3949
30 16.9652 11.8524 16.236 0.13108 24.4137 41.3094 0.20042 187.719 10.5249 117.5839
8.1 Design Optimization of the Single-Cylinder …
259
Table 8.5 Pareto-optimal solutions obtained by the Rao-3 algorithm for the single-cylinder direct-injection diesel engine case study
260

Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm

1 17.053 11.8405 16.2353 0.13859 24.6761 41.8922 0.21016 189.958 11.031 117.3679
2 15.8465 11.7995 16.2428 0.1 22.4624 38.1502 0.13465 168.561 7.30527 113.8472
3 16.4707 11.8819 16.2403 0.1 23.171 38.7682 0.1572 176.71 8.39302 117.6429
4 16.8741 11.8646 16.2366 0.12333 24.1425 40.7069 0.19062 185.423 10.0274 117.804
5 15.9876 11.8202 16.2413 0.1 22.6445 38.3004 0.14025 170.598 7.56492 114.7163
6 15.5124 11.7719 16.2294 0.10001 22.1739 37.8115 0.12699 165.583 6.96179 112.0065
7 15.98 11.8155 16.244 0.1 22.6094 38.2878 0.13901 170.175 7.50722 114.6355
8 15.5056 11.7666 16.2325 0.10001 22.1322 37.7965 0.12562 165.106 6.90103 111.9254
9 16.3735 11.8685 16.2416 0.1 23.0622 38.6774 0.15351 175.4 8.20716 117.0397
10 16.2724 11.854 16.2431 0.1 22.9428 38.5798 0.14952 173.982 8.01015 116.4091
11 17.1303 11.83 16.2346 0.14524 24.9078 42.4069 0.21897 191.949 11.4994 117.1744
12 17.2495 11.8136 16.2331 0.15557 25.2667 43.2041 0.23301 195.059 12.2666 116.8692
13 16.5825 11.899 16.2375 0.1 23.304 38.8711 0.16191 178.356 8.63365 118.356
14 16.6335 11.8966 16.2371 0.10298 23.427 39.1185 0.16601 179.448 8.82976 118.3727
15 15.4233 11.7616 16.2276 0.1 22.0723 37.7129 0.1242 164.525 6.83972 111.4929
16 15.4359 11.7625 16.2283 0.10001 22.0824 37.7263 0.12445 164.624 6.85038 111.5598
17 16.1982 11.845 16.2429 0.1 22.8652 38.5084 0.14705 173.086 7.88948 115.9642
18 16.3509 11.8659 16.2415 0.1 23.0402 38.6565 0.15279 175.142 8.17136 116.905
19 17.2983 11.8069 16.2323 0.15984 25.4146 43.5322 0.23893 196.35 12.5982 116.7414
20 17.1689 11.8247 16.2341 0.14858 25.0238 42.6646 0.22345 192.95 11.7416 117.0766
21 16.0688 11.8282 16.2432 0.1 22.7195 38.3797 0.14245 171.421 7.66906 115.1814
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.5 (continued)
Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm
22 15.356 11.75 16.229 0.1 21.968 37.632 0.1212 163.42 6.7102 111.075
23 16.1202 11.8356 16.2425 0.1 22.7834 38.4322 0.14449 172.154 7.76604 115.4995
24 16.9902 11.849 16.2359 0.13321 24.4883 41.475 0.20317 188.354 10.6662 117.5228
25 17.1271 11.8304 16.2346 0.14496 24.898 42.3851 0.21859 191.864 11.479 117.1827
26 16.6385 11.8959 16.2371 0.10341 23.4422 39.1521 0.16651 179.573 8.85351 118.3608
27 16.715 11.8858 16.2371 0.10986 23.6694 39.6562 0.17414 181.458 9.21758 118.1818
28 15.7669 11.7967 16.237 0.1 22.4226 38.0767 0.13381 168.188 7.26672 113.4425
29 17.0368 11.8427 16.2355 0.1372 24.6276 41.7845 0.20834 189.543 10.9356 117.4081
30 15.5978 11.7809 16.2315 0.1 22.2639 37.9028 0.12946 166.525 7.07136 112.4941
8.1 Design Optimization of the Single-Cylinder …
261
Table 8.6 Pareto-optimal solutions obtained by the ERao-1 algorithm for the single-cylinder direct-injection diesel engine case study
262

Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm

1 17.2796 11.8095 16.2326 0.1582 25.3578 43.4063 0.23664 195.854 12.4698 116.7906
2 15.8354 11.8013 16.2403 0.1 22.473 38.1432 0.13511 168.705 7.32639 113.812
3 16.2465 11.8517 16.2424 0.1 22.922 38.556 0.1489 173.75 7.97946 116.2627
4 16.3932 11.8719 16.2408 0.1 23.0893 38.6967 0.15445 175.73 8.25406 117.1691
5 16.1699 11.8433 16.2415 0.1 22.8479 38.483 0.14659 172.904 7.86679 115.8126
6 16.5794 11.8984 16.2377 0.1 23.2996 38.8683 0.16174 178.3 8.62535 118.3352
7 16.0447 11.8271 16.2417 0.1 22.7066 38.358 0.14215 171.293 7.65426 115.0557
8 16.4941 11.8857 16.2396 0.1 23.2007 38.7903 0.15825 177.077 8.44619 117.7944
9 15.7818 11.7961 16.2389 0.1 22.4221 38.0887 0.13368 168.162 7.26093 113.5076
10 16.0572 11.827 16.2429 0.1 22.7089 38.3685 0.14214 171.305 7.65431 115.1148
11 17.0543 11.8403 16.2353 0.13871 24.68 41.901 0.21031 189.992 11.0389 117.3646
12 16.1414 11.838 16.2428 0.1 22.8043 38.4528 0.14513 172.389 7.79679 115.624
13 16.4357 11.8773 16.2406 0.1 23.1337 38.7361 0.15594 176.261 8.32896 117.4279
14 16.7449 11.8818 16.237 0.11238 23.7583 39.8537 0.17718 182.199 9.36438 118.1114
15 15.827 11.8024 16.2388 0.1 22.4788 38.1376 0.13538 168.785 7.33854 113.7815
16 17.3075 11.8056 16.2322 0.16064 25.4425 43.5943 0.24005 196.594 12.6619 116.7171
17 17.0476 11.8412 16.2354 0.13813 24.66 41.8565 0.20956 189.82 10.9993 117.3812
18 16.7757 11.8778 16.237 0.11499 23.8499 40.0569 0.18033 182.964 9.51803 118.0386
19 16.0517 11.827 16.2424 0.1 22.707 38.3637 0.14211 171.289 7.65274 115.0879
20 15.8426 11.803 16.2398 0.1 22.4872 38.1519 0.13557 168.866 7.34722 113.8622
21 16.828 11.8708 16.2368 0.11942 24.0054 40.4023 0.18576 184.268 9.7852 117.9143
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.6 (continued)
Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm
22 15.7222 11.7891 16.2381 0.1 22.3557 38.0257 0.13177 167.447 7.17419 113.1589
23 16.6102 11.8997 16.2371 0.10102 23.3578 38.9649 0.16373 178.876 8.72216 118.427
24 16.6613 11.8929 16.2371 0.10533 23.5099 39.3022 0.16877 180.133 8.96035 118.3077
25 15.287 11.748 16.223 0.1 21.934 37.564 0.1206 163.12 6.6848 110.732
26 15.5268 11.7704 16.2319 0.1 22.1665 37.8216 0.12663 165.478 6.94567 112.0614
27 16.5992 11.9011 16.2371 0.10009 23.3251 38.8924 0.16265 178.606 8.67181 118.4526
28 15.697 11.7878 16.2365 0.1 22.34 38.0015 0.1314 167.295 7.15767 113.0278
29 17.282 11.8091 16.2326 0.15841 25.365 43.4223 0.23693 195.917 12.486 116.7844
30 17.2285 11.8165 16.2333 0.15374 25.2032 43.063 0.23049 194.506 12.127 116.9238
8.1 Design Optimization of the Single-Cylinder …
263
Table 8.7 Pareto-optimal solutions obtained by the ERao-2 algorithm for the single-cylinder direct-injection diesel engine case study
264

Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm

1 16.5122 11.8883 16.2393 0.1 23.221 38.8069 0.15896 177.325 8.48212 117.9075
2 16.014 11.634 16.363 0.1035 21.34 38.243 0.0977 155.61 5.7295 112.813
3 15.5787 11.7759 16.2333 0.1 22.2209 37.8772 0.12812 166.045 7.01131 112.3569
4 15.5859 11.7776 16.2327 0.1 22.2362 37.8866 0.12859 166.215 7.0324 112.4077
5 16.7848 11.8765 16.2369 0.11576 23.8769 40.1169 0.18127 183.19 9.5639 118.017
6 16.9125 11.8595 16.2364 0.12658 24.2566 40.9603 0.19471 186.387 10.2335 117.7118
7 15.4728 11.7652 16.2301 0.1 22.1128 37.764 0.1252 164.927 6.88316 111.7589
8 16.3177 11.8607 16.2423 0.1 22.9979 38.624 0.15136 174.635 8.10047 116.6935
9 16.0103 11.8226 16.2417 0.1 22.6666 38.3229 0.14091 170.842 7.59576 114.8476
10 15.3888 11.756 16.228 0.1 22.0205 37.6716 0.12271 163.974 6.77508 111.2798
11 16.5203 11.8896 16.239 0.1 23.2315 38.8145 0.15933 177.456 8.5013 117.9607
12 16.9924 11.8487 16.2358 0.1334 24.4948 41.4895 0.20341 188.41 10.6787 117.5175
13 16.3439 11.865 16.2415 0.1 23.0327 38.6499 0.15255 175.053 8.15906 116.8624
14 15.9647 11.8173 16.2412 0.1 22.6185 38.277 0.13946 170.308 7.52773 114.5793
15 17.1252 11.8307 16.2346 0.1448 24.8922 42.3724 0.21837 191.815 11.4673 117.1875
16 17.323 11.8034 16.2319 0.162 25.4895 43.6985 0.24196 197.006 12.7698 116.6761
17 16.2843 11.8567 16.2422 0.1 22.9642 38.5925 0.15027 174.242 8.04672 116.4936
18 16.8876 11.8628 16.2365 0.12447 24.1826 40.7959 0.19205 185.761 10.0992 117.7717
19 16.8552 11.8672 16.2367 0.12172 24.0861 40.5815 0.18861 184.947 9.92691 117.8495
20 17.3186 11.8041 16.232 0.16161 25.4762 43.6689 0.24142 196.889 12.739 116.6878
21 15.9223 11.8116 16.2413 0.1 22.5681 38.2329 0.13792 169.746 7.45611 114.3231
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.7 (continued)
Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm
22 16.086 11.8323 16.2417 0.1 22.7527 38.3995 0.14358 171.814 7.72225 115.3038
23 15.6742 11.7852 16.2361 0.1 22.3149 37.9773 0.13068 167.026 7.12548 112.8954
24 15.9376 11.8138 16.2412 0.1 22.5873 38.249 0.13851 169.96 7.4834 114.4168
25 15.6208 11.7786 16.2356 0.1 22.252 37.9192 0.12888 166.354 7.04509 112.5807
26 15.6233 11.7797 16.235 0.1 22.2609 37.9232 0.12917 166.455 7.05805 112.6029
27 16.9596 11.8531 16.2361 0.1306 24.3972 41.2727 0.19982 187.579 10.4939 117.5974
28 17.1294 11.8301 16.2346 0.14516 24.905 42.4008 0.21886 191.925 11.4936 117.1768
29 16.1184 11.8352 16.2427 0.1 22.7797 38.4302 0.14436 172.11 7.76002 115.4867
30 16.915 11.8591 16.2364 0.1268 24.2642 40.9771 0.19499 186.451 10.2474 117.7057
8.1 Design Optimization of the Single-Cylinder …
265
Table 8.8 Pareto-optimal solutions obtained by the ERao-3 algorithm for the single-cylinder direct-injection diesel engine case study
266

Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm

1 16.914 11.8593 16.2364 0.12672 24.2613 40.9708 0.19488 186.427 10.2422 117.708
2 16.6787 11.8906 16.2371 0.10679 23.5615 39.4167 0.1705 180.561 9.04272 118.267
3 16.9641 11.8525 16.236 0.13098 24.4104 41.302 0.2003 187.691 10.5187 117.5866
4 16.5952 11.9008 16.2372 0.1 23.3185 38.8826 0.16242 178.537 8.66048 118.4367
5 17.0863 11.836 16.235 0.14145 24.7758 42.1137 0.21393 190.813 11.2301 117.285
6 16.3435 11.8645 16.2418 0.1 23.0292 38.649 0.15241 175.008 8.1524 116.8553
7 17.3256 11.8031 16.2319 0.16223 25.4974 43.7161 0.24228 197.076 12.7881 116.6691
8 17.175 11.8239 16.234 0.1491 25.0419 42.7049 0.22416 193.108 11.78 117.0612
9 15.5565 11.7743 16.2321 0.1 22.2036 37.8548 0.12769 165.872 6.99224 112.2379
10 16.5397 11.8924 16.2387 0.1 23.253 38.8321 0.16009 177.721 8.53997 118.0819
11 15.3693 11.7551 16.2267 0.1 22.0075 37.6519 0.12243 163.852 6.7629 111.1792
12 16.0728 11.8303 16.242 0.1 22.7352 38.3857 0.14301 171.612 7.6955 115.2205
13 17.3102 11.8052 16.2321 0.16087 25.4505 43.6119 0.24038 196.664 12.6801 116.7101
14 16.9818 11.8502 16.2359 0.13249 24.4632 41.4192 0.20224 188.14 10.6184 117.5434
15 16.9207 11.8584 16.2363 0.12729 24.2813 41.0151 0.1956 186.596 10.2787 117.6918
16 16.5182 11.8895 16.2389 0.1 23.2299 38.8126 0.15928 177.436 8.49857 117.9483
17 16.5341 11.8916 16.2387 0.1 23.2469 38.8271 0.15987 177.645 8.52892 118.0472
18 15.7344 11.7906 16.2382 0.1 22.37 38.0388 0.13218 167.601 7.19284 113.2313
19 17.2308 11.8162 16.2333 0.15394 25.2102 43.0785 0.23076 194.567 12.1422 116.9178
20 16.6908 11.889 16.2371 0.10781 23.5974 39.4966 0.17171 180.86 9.10064 118.2386
21 16.297 11.8583 16.2422 0.1 22.9777 38.6047 0.15071 174.4 8.06837 116.5705
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.8 (continued)
Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm
22 17.3485 11.7999 16.2315 0.16424 25.5671 43.8707 0.24512 197.687 12.9501 116.6081
23 15.727 11.719 16.284 0.102 21.907 38.052 0.1166 162.08 6.5024 112.385
24 17.0656 11.8388 16.2352 0.13967 24.7136 41.9755 0.21157 190.279 11.1054 117.3368
25 16.6427 11.8954 16.2371 0.10376 23.4545 39.1794 0.16692 179.675 8.87281 118.3512
26 15.7438 11.7924 16.2379 0.1 22.3851 38.0497 0.13265 167.77 7.214 113.292
27 15.7236 11.79 16.2376 0.1 22.3624 38.0282 0.13199 167.526 7.18438 113.1738
28 16.1841 11.8425 16.2434 0.1 22.845 38.4938 0.14637 172.848 7.85704 115.873
29 17.3512 11.7995 16.2315 0.16447 25.5752 43.8886 0.24545 197.758 12.9691 116.601
30 15.9353 11.8134 16.2413 0.1 22.5838 38.2465 0.1384 169.92 7.47821 114.4019
8.1 Design Optimization of the Single-Cylinder …
267
Table 8.9 Pareto-optimal solutions obtained by the SAP-Rao algorithm for the single-cylinder direct-injection diesel engine case study
268

Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm

1 16.6691 11.8919 16.2371 0.10599 23.5331 39.3538 0.16954 180.326 8.99731 118.289
2 15.9034 11.8114 16.2397 0.1 22.5618 38.2162 0.13783 169.693 7.45157 114.23
3 17.3444 11.8005 16.2316 0.16388 25.5546 43.8429 0.24461 197.577 12.9208 116.619
4 17.3289 11.8026 16.2318 0.16251 25.5073 43.7379 0.24268 197.162 12.8109 116.661
5 15.8558 11.804 16.2404 0.1 22.4975 38.1649 0.13584 168.974 7.35994 113.934
6 16.9089 11.86 16.2364 0.12628 24.246 40.9367 0.19433 186.297 10.2141 117.72
7 16.3171 11.8614 16.2417 0.1 23.0027 38.6243 0.15155 174.698 8.10981 116.697
8 15.3124 11.7547 16.2209 0.1 21.9889 37.5992 0.12222 163.716 6.75548 110.909
9 16.5316 11.8917 16.2385 0.1 23.2471 38.8252 0.1599 177.651 8.53 118.036
10 17.1936 11.8213 16.2338 0.15072 25.0981 42.8297 0.22635 193.594 11.8997 117.013
11 16.9405 11.8557 16.2362 0.12898 24.3403 41.1462 0.19774 187.096 10.3876 117.644
12 15.247 11.747 16.22 0.1 21.909 37.525 0.1201 162.9 6.6633 110.53
13 17.0192 11.8451 16.2356 0.13569 24.5749 41.6675 0.20638 189.093 10.8329 117.452
14 16.173 11.8426 16.2423 0.1 22.8436 38.4848 0.1464 172.846 7.85808 115.821
15 16.0149 11.8243 16.2409 0.1 22.6804 38.3293 0.14138 171.006 7.6181 114.887
16 17.1224 11.8311 16.2346 0.14455 24.8838 42.3538 0.21805 191.742 11.45 117.195
17 16.548 11.8936 16.2385 0.1 23.2628 38.8397 0.16043 177.842 8.55778 118.135
18 16.2295 11.849 16.2428 0.1 22.8997 38.539 0.14816 173.486 7.94326 116.154
19 15.4404 11.7654 16.2266 0.10001 22.1056 37.7353 0.12519 164.887 6.88345 111.608
20 17.05 11.8409 16.2354 0.13833 24.6671 41.8723 0.20983 189.881 11.0133 117.375
21 15.5717 11.777 16.2317 0.1 22.2277 37.873 0.1284 166.134 7.02411 112.334
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.9 (continued)
Solution X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm
22 16.8775 11.8642 16.2366 0.12361 24.1524 40.7288 0.19097 185.506 10.045 117.796
23 16.4984 11.8862 16.2396 0.1 23.2046 38.7941 0.15838 177.124 8.45288 117.82
24 16.0253 11.8267 16.2401 0.1 22.6996 38.3412 0.14203 171.231 7.64823 114.959
25 17.2144 11.8185 16.2335 0.15252 25.1608 42.9689 0.22881 194.138 12.0347 116.96
26 15.589 11.7807 16.2308 0.1 22.2596 37.8944 0.12938 166.487 7.0677 112.45
27 16.3666 11.8678 16.2415 0.1 23.0558 38.671 0.1533 175.325 8.19681 116.999
28 16.9537 11.8539 16.2361 0.1301 24.3796 41.2335 0.19918 187.429 10.4609 117.612
29 15.9446 11.8146 16.2412 0.1 22.5951 38.2565 0.13874 170.046 7.49442 114.458
30 15.4099 11.7598 16.2275 0.1 22.0548 37.6976 0.12371 164.341 6.81843 111.413
8.1 Design Optimization of the Single-Cylinder …
269
 270

Table 8.10 Best solutions obtained by various algorithms in MOO scenario of the single-cylinder direct-injection diesel engine case study
Algorithm X1 X2 X3 BSFC BTE Pmax CO NOx HC Sm Spacing
NSGA-II 33.75 7.5 13.6 0.112 21.37 14.73 0.75 11.56 48.97 154.5 0.4003
Jaya 15.3716 11.7547 16.2272 0.1000002 22.01 37.65 0.1224 163.8 6.7597 111.2 0.1857
Rao1 15.3568 11.7496 16.2293 0.1000019 21.96 37.63 0.1210 163.4 6.7020 111.1 0.1264
Rao2 15.2827 11.7505 16.2207 0.1000030 21.95 37.56 0.1211 163.3 6.7080 110.7 0.1277
Rao3 15.3563 11.7503 16.2287 0.1000022 21.97 37.63 0.1212 163.4 6.7102 111.1 0.1581
AMTPG-Jaya 15.2634 11.7459 16.2221 0.1000009 21.91 37.54 0.1199 162.9 6.6552 110.6 0.0911
SAP-Rao 15.2472 11.7467 16.2198 0.1000040 21.91 37.53 0.1201 162.9 6.6633 110.5 0.1057
ERao-1 15.2868 11.7484 16.2227 0.1000002 21.93 37.56 0.1206 163.1 6.6848 110.7 0.1174
ERao-2 16.0140 11.6339 16.3633 0.1035358 21.34 38.24 0.0977 155.6 5.7295 112.8 0.0905
ERao-3 15.7274 11.7189 16.2840 0.1020057 21.91 38.05 0.1166 162.1 6.5024 112.4 0.1079
Source NSGA-II:—Dhingra et al. (2014)
Result in boldface indicates better values
8 Optimization of the Selected Bio-Energy Systems
8.1 Design Optimization of the Single-Cylinder … 271

Now, to identify the best solution among the solutions of different algorithms
given in Table 8.10, which has the best compromise among the objectives, the
MADM methods based average ranks are calculated and presented in Table 8.11. The
Spearman’s correlation coefficients for different pairs of rankings given by decision-
making methods for different algorithm’s solutions are shown in Table 8.12. The
ranks given by the SAW, WPM, TOPSIS, and MTOPSIS methods for each solution
are identical. Furthermore, the Spearman’s correlation for all the pairs formed by all
the MADM (except the pairs with GRA) methods is greater than 0.5. The Spearman’s
correlation values for the pairs formed by the GRA method are less than 0.5. Hence,
the ranks suggested by the GRA method cannot be considered for calculating the
average ranks. Therefore, the corrected average ranks are calculated, excluding the
ranks suggested by the GRA method, and presented in Table 8.11 as the corrected
ranks.
The AMTPG-Jaya algorithm’s solution has achieved the least average rank of
2.29. Thus, it can be regarded as the best solution. The AMTPG-Jaya algorithm’s
solution ranked 1 by two methods (PROMETHEE and VIKOR), ranked two by
COPRAS, and ranked three by remaining methods. Hence, it has achieved the least
average rank. The ERao-3, SAP-Rao, ERao-2, and ERao-1 algorithm’s solutions
have achieved the next best average rank values, respectively.
Here, it can be noted that the MADM methods have ranked these solutions based
on the relative importance of the solutions and the AMTPG-Jaya solution has the
best compromise among the compared solutions. The brake-thermal efficiency for
the AMTPG-Jaya algorithm solution is 2.5 and 2.7% higher than that of the NSGA-
II and ERao-2 solutions, respectively. Also, the BSFC value for the AMTPG-Jaya
algorithm’s solution is 10.7, 3.4 and 2% lesser than that of the NSGA-II, ERao-2
and ERao-3 solutions, respectively. By the AMTPG-Jaya solution, CO emissions
are 84, 2, 0.9, 1, 1.1, 0.1 and 0.6% lesser when compared to that of the NSGA-II,
Jaya, Rao-1, Rao-2, Rao-3, SAP-Rao, and ERao-1 solutions, respectively. Similarly,
for the AMTPG-Jaya solution, NOx emissions are lesser compared to those of the
Jaya, Rao-1, Rao-2, Rao-3, SAP-Rao, and ERao-1 solutions. By the AMTPG-Jaya
solution, HC emissions are reduced about 86.4, 1.5, 0.7, 0.8, 0.8, 0.1 and 0.4% when
compared to that of the NSGA-II, Jaya, Rao-1, Rao-2, Rao-3, SAP-Rao, and ERao-
1 solutions, respectively. Similarly, by the AMTPG-Jaya solution, peak cylinder
pressure is reduced by 1.8, 1.4, 0.3, 0.2, 0.1, 0.2 and 0.1%, when compared to that of
the ERao-2, ERao-3, Jaya, Rao-1, Rao-2, Rao-3, and ERao-1 solutions, respectively.
Hence, the AMTPG-Jaya algorithm solution achieved the least average rank.
Furthermore, the spacing values of the Pareto-fronts obtained by the algorithms
compared are also presented in Table 8.10. The ERao-2 algorithm has the least
spacing value, which is 77.4, 51.3, 28.4, 29.2, 42.8, 0.6, 14.4, 22.9 and 16.1% lesser
value when compared to that of the NSGA-II, Jaya, Rao-1, Rao-2, Rao-3, AMTPG-
Jaya, SAP-Rao, ERao-1, and ERao-3 algorithms, respectively. The performances
of the proposed modified algorithms are better or competitive to that of the basic
algorithms in terms of the spacing metric. The coverage metric values of the Pareto-
fronts obtained by the Jaya algorithm, Rao algorithms, and their modified versions
 272

Table 8.11 Ranks suggested by the MADM methods for different algorithm solutions presented in Table 8.10
Algorithm SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA AR CR FR
NSGA-II 10 10 10 10 10 10 10 10 10.0 10.00 10
Jaya 9 9 9 9 8 9 9 1 7.88 8.86 9
Rao-1 7 7 7 7 6 4 7 3 6.00 6.43 7
Rao-2 6 6 6 6 5 6 6 4 5.63 5.86 6
Rao-3 8 8 8 8 7 8 8 2 7.13 7.86 8
AMTPG-Jaya 3 3 3 3 1 1 2 7 2.88 2.29 1
SAP-Rao 4 4 4 4 1 2 3 6 3.50 3.14 3
ERao-1 5 5 5 5 1 3 4 5 4.13 4.00 5
ERao-2 1 1 1 1 9 7 1 9 3.75 3.00 4
ERao-3 2 2 2 2 1 5 5 8 3.38 2.71 2
AR-Average Rank; CR-Corrected Rank; FR-Final Rank
8 Optimization of the Selected Bio-Energy Systems
8.1 Design Optimization of the Single-Cylinder … 273

Table 8.12 Spearman’s rank correlation coefficients between different pairs of MADM method’s
rankings presented in Table 8.11
Method SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA
SAW 1 1 1 1 0.52 0.60 0.93 −0.45
WPM 1 1 1 1 0.52 0.60 0.93 −0.45
TOPSIS 1 1 1 1 0.52 0.60 0.93 −0.45
MTOPSIS 1 1 1 1 0.52 0.60 0.93 −0.45
VIKOR 0.52 0.52 0.52 0.52 1 0.88 0.52 0.04
PROMETHEE 0.60 0.60 0.60 0.60 0.88 1 0.71 −0.05
COPRAS 0.93 0.93 0.93 0.93 0.52 0.71 1 −0.38
GRA −0.45 −0.45 −0.45 −0.45 0.04 −0.05 −0.38 1

are presented in Table 8.13. All the algorithms have performed well in terms of
coverage.
From the computational results of the single-cylinder direct-injection diesel
engine case study, it can be observed that the performance of the considered system
can be improved with the solutions achieved by the proposed algorithms. The solution
of the AMTPG-Jaya algorithm has better compromise among the seven objectives.
The performance objectives BTE, Pmax , CO, NOx , HC, BSFC, and smoke values
obtained by the ERao-3 solution are 21.908 N-m, 37.538 bar, 0.1199%, 162.85 ppm,
6.6552 ppm, 0.10000012 kg/kWh, and 110.6, respectively. The ERao-2 algorithm has
achieved the least spacing value (0.09049), which is much better than that achieved by
other algorithms compared, including the NSGA-II. All the algorithms have achieved
better performance in terms of coverage values. In addition, the performances of the
modified versions in this case study are better or competitive to those of the basic
algorithms as well as the NSGA-II algorithm.

8.2 Design Optimization of a Turbocharged DI Diesel

Engine

The description of the selected turbocharged DI diesel engine system is presented

in Sect. 2.3.2. In this case study, a turbocharged DI diesel engine using biodiesel
and diesel blends was considered. The waste vegetable cooking oil is considered as
the source of biodiesel. The specific fuel properties and engine specifications were
presented in Shirneshan et al. (2016).
The six objectives of this case study are: maximizing the brake power (P) and brake
torque (T ); minimizing the brake-specific fuel consumption (BSFC) and emissions
such as carbon monooxide emission (CO), nitrogen oxides emission (NOx ), and
hydrocarbon emission (HC). The design variables were the percentage of biodiesel
in fuel (X 1 ), engine speed (X 2 ), and engine load (X 3 ), and their ranges were 0 ≤
 274

Table 8.13 Coverage (%) values of the Pareto-fronts obtained by the proposed algorithms in MOO for the single-cylinder direct-injection diesel engine case
study
Algorithm Jaya Rao-1 Rao-2 Rao-3 AMTPG-Jaya SAP-Rao ERao-1 ERao-2 ERao-3 NSGA-II
Jaya – 0 0 0 0 0 0 0 0 0
AMTPG-Jaya 0 – 0 0 0 0 0 0 0 0
Rao1 0 0 – 0 0 0 0 0 0 0
Rao2 0 0 0 – 0 0 0 0 0 0
Rao3 0 0 0 0 – 0 0 0 0 0
SAP-Rao 0 0 0 0 0 – 0 0 0 3.33
ERao1 0 0 0 0 0 0 – 0 0 0
ERao2 0 0 0 0 0 0 0 – 0 0
ERao3 0 0 0 0 0 0 0 0 – 0
NSGA-II 0 0 0 0 0 0 0 0 0 –
8 Optimization of the Selected Bio-Energy Systems
8.2 Design Optimization of a Turbocharged DI Diesel Engine 275

X 1 ≤ 100 (%), 1000 ≤ X 2 ≤ 2800(rpm), and 25 ≤ X 3 ≤ 100 (%). The regression

models of this case study’s objectives are as follows:

P = − 47.32 − 0.08X 1 + 0.056X 2 + 0.205X 3 + 0.0002X 12

− 1.4 × 10−5 X 22 − 5.499 × 10−4 X 32 + 0.0004X 1 X 3
+ 0.0002X 2 X 3 (8.9)

T = − 299.277 − 0.524X 1 + 0.302X 2 + 4.654X 3

+ 0.00362X 12 − 7.401 × 10−5 X 22 − 0.00637X 32
− 3.771 × 10−4 X 2 X 3 (8.10)

BSFC =298.74 + 0.5X 1 − 0.088X 2 − 0.236X 3

+ 0.0014X 12 + 2.67 × 10−5 X 22 − 0.00018X 1 X 2
− 0.00336X 1 X 3 (8.11)

CO =0.109 − 3.43 × 10−4 X 1 − 3.96 × 10−5 X 2

− 7.53 × 10−4 X 3 + 6.48 × 10−7 X 12 + 6.32 × 10−9 X 22
+ 2.93 × 10−6 X 32 + 1.69 × 10−6 X 1 X 3 + 7.33 × 10−8 X 2 X 3 (8.12)

NOx =216.71 − 0.264X 1 + 0.158X 2 + 0.755X 3

+ 0.0114X 12 − 5.37 × 10−5 X 22 + 0.0558X 32
− 0.00188X 2 X 3 (8.13)

HC =56.38 − 0.028X 1 + 0.019X 2 − 0.554X 3

− 5.19 × 10−4 X 12 − 2.1 × 10−6 X 22 + 0.0026X 32
+ 3.14 × 10−5 X 1 X 2 − 0.0011X 1 X 3 + 8.38 × 10−5 X 2 X 3 (8.14)

Shirneshan et al. (2016) presented the optimum design parameters using the ABC
algorithm and reported a Pareto-optimal solution through a priori articulation of
preferences method. Now, the proposed algorithms, along with basic Jaya and Rao
algorithms, are used to find the optimal parameters. The function evaluations taken by
the ABC algorithm were not specified. However, the proposed and other considered
algorithms are executed for 50,000 function evaluations. In multi-objective optimiza-
tion using the Jaya algorithm, Rao algorithms, and their modified versions, a priori
approach is followed. In multi-objective optimization through a priori approach, the
combined objective function of this case study is given by the following equation:
276 8 Optimization of the Selected Bio-Energy Systems

P T BSFC CO
Z combined = − w1 − w2 + w3 + w4
Pmax Tmax BSFCmin COmin
NOx HC
+ w5 + w6 (8.15)
NOx min HCmin

Here, the terms BSFCmin , COmin , NOx min , and HCmin represent the objective
function values achieved in single-objective optimization (minimization) of BSFC,
CO, NOx , and HC, respectively. Similarly, the terms Pmax and Tmax represent the
maximum power and torque value achieved in single-objective optimization (maxi-
mization) of P and T, respectively. The weights of the objectives are varied between
zero and one randomly such that w1 + w2 + w3 + w4 + w5 + w6 = 1.
Shirneshan et al. (2016) presented a single Pareto-optimal solution in this case
study using a priori approach. Similar to the previous case study, for 30 combinations
of the weights, 30 Pareto-optimal solutions are identified. By using different combi-
nations of the weights, the algorithms are executed 30 times. For each combination of
the weights, a single Pareto-optimal solution is found by executing the algorithm. The
population size for all the algorithms is taken as 50 in each execution. The elite popu-
lation size is taken as 20% of the population for the elitist Rao algorithms. Further-
more, the solutions obtained by various algorithms are non-dominated. Hence, the
average rank method was used to identify the best solution from the Pareto-front of
the respective algorithm. The Pareto-optimal solutions achieved by the Jaya algo-
rithm, Rao algorithms, and their modified versions in this case study are presented
in Tables 8.14, 8.15, 8.16, 8.17, 8.18, 8.19, 8.20, 8.21 and 8.22
Similar to the previous case studies, the best solutions from these Pareto-fronts
are identified. Solution 26 of the Jaya algorithm, Solution 24 of the AMTPG-Jaya
algorithm, Solution 18 of the Rao-1 algorithm, Solution 10 of the Rao-2 algorithm,
Solution 7 of the Rao-3 algorithm, Solution 8 of the ERao-1 algorithm, Solution
16 of the ERao-2 algorithm, Solution 28 of the ERao-3 algorithm, and Solution 16
of the SAP-Rao algorithm are identified as the best solutions from the respective
algorithm Pareto-front. Now, these best solutions are compared with those of the
NSGA-II algorithm in Table 8.23.
In Table 8.23, the solutions with the best average rank from the Pareto-optimal
solutions of the various algorithms are compared with that reported by the ABC
algorithm. The ERao-3 algorithm has achieved higher brake power and least CO
and NOx emission values. The brake power for the ERao-3 algorithm solution is
3.4, 3.6, 1.9, 3.4, 3.8, 2.7, 2.0, 1.9 and 2.7% higher than that of the ABC, Jaya,
Rao-1, Rao-2, Rao-3, AMTPG-Jaya, SAP-Rao, ERao-1, and ERao-2 algorithms
solutions, respectively. The CO emissions are 13.8, 2.9, 1.2, 2.7, 3.1, 1.9, 1.3, 1.2
and 1.9% lesser than that of the ABC, Jaya, Rao-1, Rao-2, Rao-3, AMTPG-Jaya,
SAP-Rao, ERao-1, and ERao-2 algorithms solutions, respectively. Similarly, the
NOx emissions are 6.6, 6.2, 3.7, 6, 6.5, 4.9, 3.8, 3.6 and 4.9% lesser than that of
the ABC, Jaya, Rao-1, Rao-2, Rao-3, AMTPG-Jaya, SAP-Rao, ERao-1, and ERao-2
algorithms solutions, respectively. The Rao-3 algorithm has the least HC emissions
value, which is 0.9, 0.3, 3.3, 0.6, 1.9, 3.2, 3.4, 1.9 and 7.1% lesser than that of the
ABC, Jaya, Rao-1, Rao-2, AMTPG-Jaya, SAP-Rao, ERao-1, ERao-2 and ERao-3
8.2 Design Optimization of a Turbocharged DI Diesel Engine 277

Table 8.14 Pareto-optimal solutions obtained by the Jaya algorithm for the turbocharged DI diesel
engine case study
SN X 1 (%) X 2 (rpm) X 3 (%) P (kW) T (N.m) BSFC CO(%) NOx (ppm) HC
(g/kWh) (ppm)
1 63.211 2800.000 100.000 68.992 243.538 222.170 0.01378 374.064 114.869
2 100.000 1902.421 93.284 56.109 302.987 204.358 0.01274 632.901 73.666
3 66.829 2800.000 100.000 68.941 243.346 221.599 0.01346 378.473 114.443
4 48.408 2800.000 100.000 69.253 245.314 224.890 0.01529 359.136 116.468
5 100.000 1000.000 97.606 26.876 269.277 227.610 0.02597 830.404 40.769
6 66.169 2800.000 100.000 68.950 243.374 221.700 0.01352 377.646 114.522
7 52.973 2800.000 100.000 69.163 244.597 223.986 0.01479 363.207 115.999
8 100.000 1094.200 94.685 30.182 270.848 224.562 0.02398 789.873 44.251
9 97.438 2555.298 97.538 66.880 271.835 210.454 0.01057 488.258 99.936
10 50.579 2800.000 100.000 69.209 244.954 224.453 0.01505 361.013 116.247
11 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
12 89.241 2665.030 100.000 68.813 263.184 213.227 0.01152 456.090 105.883
13 100.000 1000.000 96.500 26.502 265.915 228.242 0.02590 819.671 40.853
14 90.845 2644.588 99.058 68.150 263.768 212.870 0.01123 458.908 104.719
15 51.683 2800.000 100.000 69.188 244.784 224.235 0.01493 362.009 116.133
16 60.828 2800.000 100.000 69.028 243.717 222.567 0.01401 371.324 115.141
17 100.000 2014.258 93.599 58.498 301.261 204.018 0.01188 609.767 77.933
18 80.039 2800.000 100.000 68.801 243.447 219.823 0.01242 397.103 112.775
19 53.744 2800.000 100.000 69.149 244.491 223.839 0.01471 363.941 115.917
20 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
21 100.000 1000.000 99.450 27.497 274.850 226.555 0.02609 848.609 40.644
22 100.000 1000.000 95.463 26.150 262.748 228.835 0.02585 809.730 40.937
23 69.737 2800.000 100.000 68.904 243.259 221.166 0.01321 382.232 114.092
24 100.000 1776.095 93.095 53.090 303.144 205.449 0.01393 658.575 68.911
25 59.571 2800.000 100.000 69.048 243.828 222.782 0.01413 369.930 115.283
26 100.000 2257.301 94.806 62.839 292.388 205.284 0.01065 558.130 87.419
27 56.640 2800.000 100.000 69.097 244.130 223.302 0.01442 366.822 115.606
28 100.000 1000.000 96.626 26.544 266.298 228.170 0.02591 820.884 40.843
29 54.189 2800.000 100.000 69.141 244.431 223.755 0.01467 364.372 115.870
30 100.000 2487.519 100.000 68.104 285.694 207.076 0.01091 530.902 96.903

algorithms solutions, respectively. The ABC algorithm’s solution has higher brake
torque and least BSFC values. Hence, to identify the best solution among the solutions
of different algorithms, which has the best compromise among the objectives, the
MADM methods based average ranks are calculated and presented in Table 8.24.
Table 8.24 presents the decision making methods ranking of the Pareto-optimal
solutions obtained by different algorithms. The Spearman’s correlation coefficients
278 8 Optimization of the Selected Bio-Energy Systems

Table 8.15 Pareto-optimal solutions obtained by the AMTPG-Jaya algorithm for the turbocharged
DI diesel engine case study
SN X 1 (%) X 2 (rpm) X 3 (%) P (kW) T (N.m) BSFC CO (%) NOx (ppm) HC
(g/kWh) (ppm)
1 100.000 1000.000 96.027 26.341 264.473 228.512 0.02588 815.124 40.890
2 100.000 1117.024 94.570 31.144 272.843 223.556 0.02355 785.648 45.052
3 99.568 2528.323 97.115 66.487 274.191 209.866 0.01040 497.620 98.473
4 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
5 50.495 2800.000 100.000 69.211 244.967 224.469 0.01506 360.939 116.256
6 49.828 2800.000 100.000 69.224 245.075 224.603 0.01513 360.351 116.324
7 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
8 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
9 100.000 1210.407 94.140 34.932 280.341 219.706 0.02185 768.256 48.347
10 99.905 2524.135 97.051 66.426 274.554 209.781 0.01037 499.096 98.245
11 100.000 2533.346 97.140 66.526 273.791 209.997 0.01038 497.175 98.611
12 85.095 2734.527 100.000 68.863 253.334 216.363 0.01190 426.709 109.360
13 58.162 2800.000 100.000 69.071 243.966 223.029 0.01426 368.411 115.440
14 100.000 1000.000 99.252 27.430 274.253 226.668 0.02608 846.634 40.656
15 100.000 1000.000 98.492 27.175 271.960 227.103 0.02602 839.103 40.707
16 100.000 1000.000 96.667 26.558 266.425 228.146 0.02591 821.287 40.839
17 100.000 1481.879 93.282 44.637 296.105 210.936 0.01753 716.620 58.079
18 100.000 2294.013 95.055 63.401 290.438 205.712 0.01054 550.164 88.880
19 57.246 2800.000 100.000 69.087 244.063 223.192 0.01436 367.448 115.540
20 100.000 1429.958 93.399 42.928 293.782 212.336 0.01828 726.629 56.199
21 100.000 1000.000 95.626 26.205 263.247 228.742 0.02586 811.285 40.923
22 48.795 2800.000 100.000 69.245 245.247 224.811 0.01525 359.463 116.429
23 100.000 1000.000 99.608 27.550 275.326 226.464 0.02610 850.187 40.634
24 100.000 2446.946 96.292 65.500 280.589 208.152 0.01033 516.514 95.054
25 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
26 53.242 2800.000 100.000 69.158 244.559 223.934 0.01477 363.461 115.970
27 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
28 100.000 1000.000 99.771 27.604 275.815 226.371 0.02611 851.816 40.624
29 86.791 2705.215 100.000 68.858 257.555 214.994 0.01173 439.026 107.903
30 100.000 1949.309 93.399 57.142 302.444 204.144 0.01236 623.249 75.448

for different pairs of rankings for different algorithm’s solutions are shown in Table
8.25. The ranks given by the pairs SAW-WPM and TOPSIS-MTOPSIS methods
for each solution are identical. Furthermore, the Spearman’s correlation for all
the pairs of decision-making methods (except with the PROMETHEE and GRA
methods) is greater than 0.5. Spearman’s correlation values for the pairs consisting of
PROMETHEE or GRA methods are either negative or nearer to zero. Hence, the ranks
8.2 Design Optimization of a Turbocharged DI Diesel Engine 279

Table 8.16 Pareto-optimal solutions obtained by the Rao-1 algorithm for the turbocharged DI
diesel engine case study
SN X 1 (%) X 2 (rpm) X 3 (%) P (kW) T (N.m) BSFC CO (%) NOx (ppm) HC(ppm)
(g/kWh)
1 68.315 2800.000 100.000 68.922 243.294 221.375 0.01333 380.370 114.265
2 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
3 49.884 2800.000 100.000 69.223 245.065 224.591 0.01513 360.400 116.318
4 62.218 2800.000 100.000 69.007 243.608 222.333 0.01388 372.906 114.983
5 95.832 2576.231 97.878 67.182 269.984 210.957 0.01071 481.160 101.065
6 100.000 1000.000 96.363 26.455 265.497 228.320 0.02589 818.351 40.863
7 57.669 2800.000 100.000 69.079 244.017 223.117 0.01431 367.890 115.494
8 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
9 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
10 100.000 2200.541 94.457 61.924 295.088 204.745 0.01086 570.359 85.175
11 55.833 2800.000 100.000 69.111 244.225 223.449 0.01450 365.999 115.694
12 49.093 2800.000 100.000 69.239 245.196 224.750 0.01521 359.717 116.399
13 54.832 2800.000 100.000 69.129 244.348 223.634 0.01460 365.001 115.802
14 100.000 2250.038 94.759 62.725 292.755 205.207 0.01067 559.700 87.131
15 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
16 73.391 2800.000 100.000 68.863 243.238 220.656 0.01291 387.229 113.637
17 79.438 2800.000 100.000 68.806 243.415 219.894 0.01246 396.169 112.855
18 100.000 2325.908 95.286 63.870 288.614 206.133 0.01047 543.208 90.155
19 100.000 1000.000 98.851 27.295 273.044 226.897 0.02605 842.654 40.683
20 100.000 2318.004 95.227 63.756 289.077 206.024 0.01048 544.935 89.839
21 99.596 2527.976 97.109 66.482 274.221 209.859 0.01039 497.742 98.454
22 59.817 2800.000 100.000 69.044 243.805 222.740 0.01410 370.200 115.255
23 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
24 64.026 2800.000 100.000 68.980 243.487 222.038 0.01371 375.031 114.774
25 59.643 2800.000 100.000 69.047 243.821 222.770 0.01412 370.009 115.275
26 100.000 2181.335 94.348 61.601 295.915 204.596 0.01094 574.474 84.420
27 68.206 2800.000 100.000 68.923 243.297 221.391 0.01334 380.229 114.278
28 100.000 1000.000 97.883 26.969 270.116 227.451 0.02598 833.112 40.749
29 56.199 2800.000 100.000 69.105 244.181 223.382 0.01446 366.370 115.654
30 100.000 1460.115 93.328 43.929 295.171 211.507 0.01784 720.824 57.290

suggested by the PROMETHEE and GRA methods cannot be considered for calcu-
lating the average ranks. Now, the corrected average ranks are calculated, excluding
the ranks suggested by the PROMETHEE and GRA methods, and presented in Table
8.24 as the corrected ranks.
The ERao-1 algorithm’s solution has the best compromise among the objectives
and has achieved the least average rank of 1.33. Hence, the ERao-1 algorithm’s
280 8 Optimization of the Selected Bio-Energy Systems

Table 8.17 Pareto-optimal solutions obtained by the Rao-2 algorithm for the turbocharged DI
diesel engine case study
SN X 1 (%) X 2 (rpm) X 3 (%) P (kW) T (N.m) BSFC CO (%) NOx (ppm) HC
(g/kWh) (ppm)
1 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
2 100.000 1000.000 98.123 27.050 270.845 227.313 0.02600 835.475 40.732
3 56.458 2800.000 100.000 69.100 244.151 223.335 0.01443 366.635 115.626
4 100.000 1682.108 93.063 50.612 302.025 206.752 0.01495 677.374 65.416
5 100.000 1000.000 96.612 26.540 266.257 228.178 0.02591 820.756 40.844
6 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
7 100.000 1131.410 94.500 31.743 274.068 222.935 0.02328 782.980 45.558
8 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
9 100.000 2476.862 96.573 65.868 278.336 208.753 0.01033 509.845 96.280
10 100.000 2264.319 94.852 62.948 292.028 205.361 0.01063 556.610 87.697
11 48.682 2800.000 100.000 69.248 245.266 224.834 0.01526 359.367 116.440
12 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
13 79.390 2800.000 100.000 68.806 243.412 219.899 0.01246 396.095 112.861
14 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
15 48.160 2800.000 100.000 69.258 245.357 224.941 0.01532 358.929 116.492
16 100.000 1491.586 93.263 44.950 296.504 210.688 0.01739 714.742 58.431
17 100.000 2407.655 95.942 64.997 283.385 207.424 0.01034 525.231 93.453
18 51.066 2800.000 100.000 69.200 244.878 224.356 0.01500 361.449 116.197
19 100.000 1000.000 95.560 26.183 263.043 228.780 0.02585 810.650 40.929
20 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
21 100.000 1549.446 93.166 46.762 298.642 209.308 0.01660 703.494 60.539
22 77.682 2800.000 100.000 68.821 243.336 220.104 0.01259 393.487 113.086
23 100.000 2302.161 95.113 63.523 289.983 205.815 0.01052 548.390 89.205
24 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
25 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
26 63.187 2800.000 100.000 68.992 243.540 222.174 0.01379 374.036 114.871
27 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
28 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
29 73.569 2800.000 100.000 68.861 243.239 220.632 0.01290 387.482 113.615
30 100.000 1000.000 98.668 27.234 272.491 227.002 0.02603 840.838 40.695

solution can be considered as the best solution. The brake power for the ERao-1
algorithm’s solution is 1.4, 1.7, 1.5, 1.8, 0.8, 0.1 and 0.8% higher than that of the ABC,
Jaya, Rao-2, Rao-3, AMTPG-Jaya, SAP-Rao, and ERao-2 solutions, respectively.
The CO emissions for the ERao-1 algorithm solution is 12.8, 1.7, 1.5, 1.9, 0.7, 0.1
and 0.7% lesser than that of the ABC, Jaya, Rao-2, Rao-3, AMTPG-Jaya, SAP-Rao
and ERao-2 solutions, respectively. The ERao-1 algorithm solution’s NOx emissions
8.2 Design Optimization of a Turbocharged DI Diesel Engine 281

Table 8.18 Pareto-optimal solutions obtained by the Rao-3 algorithm for the turbocharged DI
diesel engine case study
SN X 1 (%) X 2 (rpm) X 3 (%) P (kW) T (N.m) BSFC CO(%) NOx (ppm) HC
(g/kWh) (ppm)
1 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
2 78.555 2800.000 100.000 68.813 243.372 219.998 0.01252 394.812 112.971
3 56.931 2800.000 100.000 69.092 244.098 223.249 0.01439 367.121 115.575
4 73.003 2800.000 100.000 68.867 243.236 220.708 0.01294 386.685 113.686
5 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
6 72.965 2800.000 100.000 68.867 243.235 220.713 0.01295 386.631 113.691
7 100.000 2250.989 94.765 62.740 292.707 205.217 0.01067 559.495 87.168
8 55.058 2800.000 100.000 69.125 244.320 223.592 0.01458 365.225 115.777
9 48.748 2800.000 100.000 69.246 245.255 224.821 0.01525 359.423 116.433
10 100.000 1311.368 93.751 38.765 287.250 216.025 0.02013 749.248 51.939
11 100.000 1000.000 97.654 26.892 269.423 227.582 0.02597 830.874 40.766
12 50.323 2800.000 100.000 69.215 244.995 224.503 0.01508 360.786 116.273
13 100.000 1000.000 96.407 26.470 265.632 228.295 0.02590 818.777 40.860
14 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
15 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
16 100.000 1000.000 96.939 26.650 267.251 227.991 0.02593 823.914 40.819
17 100.000 2228.404 94.623 62.380 293.811 204.991 0.01075 564.369 86.274
18 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
19 57.056 2800.000 100.000 69.090 244.084 223.226 0.01437 367.251 115.561
20 66.034 2800.000 100.000 68.952 243.380 221.721 0.01353 377.478 114.538
21 50.655 2800.000 100.000 69.208 244.942 224.438 0.01504 361.081 116.239
22 100.000 1000.000 98.552 27.195 272.141 227.068 0.02603 839.696 40.703
23 100.000 1000.000 98.105 27.044 270.789 227.324 0.02600 835.291 40.733
24 68.638 2800.000 100.000 68.918 243.285 221.327 0.01330 380.790 114.225
25 100.000 1216.670 94.114 35.178 280.806 219.463 0.02174 767.083 48.569
26 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
27 100.000 2344.910 95.430 64.142 287.469 206.406 0.01043 539.049 90.918
28 60.715 2800.000 100.000 69.030 243.726 222.586 0.01402 371.197 115.154
29 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
30 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610

are 3, 2.7, 2.4, 3, 1.3, 0.2 and 1.3% lesser than those of the ABC, Jaya, Rao-2, Rao-3,
AMTPG-Jaya, SAP-Rao and ERao-2 solutions, respectively.
The spacing values of the Pareto-fronts obtained by different algorithms in the
MOO scenario are also presented in Table 8.23. Here, an observation can be made
that the SAP-Rao algorithm’s Pareto-front has the least spacing value. The SAP-Rao
algorithm has a better spacing, which is 34.2, 67.4, 29.3, 23.3, 49.6, 64.3, 52.2 and
282 8 Optimization of the Selected Bio-Energy Systems

Table 8.19 Pareto-optimal solutions obtained by the ERao-1 algorithm for the turbocharged DI
diesel engine case study
SN X 1 (%) X 2 (rpm) X 3 (%) P (kW) T (N.m) BSFC CO(%) NOx (ppm) HC
(g/kWh) (ppm)
1 84.151 2751.443 100.000 68.855 250.853 217.182 0.01200 419.645 110.195
2 78.260 2800.000 100.000 68.816 243.360 220.034 0.01254 394.362 113.010
3 71.624 2800.000 100.000 68.882 243.236 220.898 0.01305 384.775 113.859
4 100.000 1000.000 95.448 26.145 262.702 228.844 0.02585 809.588 40.938
5 67.683 2800.000 100.000 68.930 243.314 221.469 0.01338 379.556 114.341
6 71.908 2800.000 100.000 68.879 243.235 220.858 0.01303 385.165 113.823
7 51.511 2800.000 100.000 69.191 244.810 224.269 0.01495 361.852 116.151
8 100.000 2326.932 95.293 63.885 288.553 206.147 0.01046 542.984 90.196
9 100.000 2100.578 93.943 60.168 298.913 204.155 0.01134 591.644 81.267
10 50.185 2800.000 100.000 69.217 245.017 224.531 0.01509 360.664 116.288
11 100.000 1000.000 96.405 26.470 265.626 228.296 0.02590 818.758 40.860
12 70.205 2800.000 100.000 68.899 243.251 221.099 0.01317 382.855 114.034
13 68.016 2800.000 100.000 68.926 243.303 221.419 0.01336 379.984 114.301
14 100.000 1966.759 93.448 57.515 302.176 204.091 0.01223 619.640 76.114
15 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
16 100.000 1314.952 93.738 38.896 287.472 215.904 0.02007 748.569 52.067
17 100.000 1000.000 95.531 26.173 262.956 228.796 0.02585 810.378 40.931
18 60.018 2800.000 100.000 69.041 243.787 222.705 0.01408 370.422 115.233
19 100.000 2110.106 93.987 60.344 298.600 204.191 0.01129 589.629 81.637
20 84.684 2741.845 100.000 68.860 252.265 216.714 0.01194 423.650 109.721
21 100.000 1000.000 97.470 26.830 268.864 227.687 0.02596 829.075 40 .779
22 50.062 2800.000 100.000 69.220 245.037 224.556 0.01511 360.556 116.300
23 51.924 2800.000 100.000 69.183 244.748 224.188 0.01491 362.230 116.108
24 52.637 2800.000 100.000 69.170 244.645 224.050 0.01483 362.891 116.034
25 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
26 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
27 55.694 2800.000 100.000 69.114 244.242 223.475 0.01451 365.859 115.709
28 100.000 2231.141 94.640 62.425 293.680 205.017 0.01074 563.779 86.383
29 100.000 2446.645 96.289 65.497 280.611 208.146 0.01033 516.581 95.042
30 48.251 2800.000 100.000 69.257 245.341 224.922 0.01531 359.004 116.483

16.4% lesser when compared to that of the Jaya, Rao-1, Rao-2, Rao-3, AMTPG-Jaya,
ERao-1, ERao-2 and ERao-3 algorithms, respectively.
The coverage metric values of the Pareto-fronts obtained by the Jaya algorithm,
Rao algorithms, and their modified versions are presented in Table 8.26. The Jaya
algorithm has achieved better coverage values than the other algorithms compared.
8.2 Design Optimization of a Turbocharged DI Diesel Engine 283

Table 8.20 Pareto-optimal solutions obtained by the ERao-2 algorithm for the turbocharged DI
diesel engine case study
SN X 1 (%) X 2 (rpm) X 3 (%) P (kW) T (N.m) BSFC CO (%) NOx (ppm) HC
(g/kWh) (ppm)
1 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
2 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
3 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
4 100.000 1917.874 93.319 56.455 302.837 204.276 0.01261 629.727 74.252
5 100.000 1650.765 93.072 49.740 301.416 207.280 0.01532 683.588 64.258
6 100.000 1000.000 95.559 26.182 263.040 228.781 0.02585 810.641 40.929
7 100.000 1000.000 97.324 26.781 268.422 227.771 0.02595 827.658 40.790
8 50.235 2800.000 100.000 69.216 245.009 224.521 0.01509 360.709 116.282
9 70.206 2800.000 100.000 68.899 243.251 221.098 0.01317 382.857 114.034
10 50.491 2800.000 100.000 69.211 244.968 224.470 0.01506 360.935 116.256
11 100.000 1690.678 93.063 50.846 302.171 206.615 0.01485 675.670 65.733
12 100.000 1735.952 93.071 52.056 302.795 205.954 0.01435 666.635 67.414
13 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
14 96.310 2569.952 97.775 67.092 270.541 210.802 0.01067 483.274 100.727
15 100.000 2307.103 95.148 63.596 289.704 205.879 0.01051 547.313 89.402
16 100.000 2293.782 95.053 63.398 290.451 205.709 0.01054 550.214 88.870
17 100.000 2518.855 96.990 66.362 274.993 209.665 0.01036 500.435 98.011
18 60.213 2800.000 100.000 69.038 243.770 222.671 0.01407 370.637 115.211
19 52.679 2800.000 100.000 69.169 244.639 224.042 0.01482 362.931 116.030
20 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
21 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
22 59.462 2800.000 100.000 69.050 243.838 222.801 0.01414 369.812 115.295
23 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
24 100.000 1000.000 95.470 26.152 262.768 228.831 0.02585 809.792 40.936
25 53.670 2800.000 100.000 69.150 244.501 223.853 0.01472 363.871 115.925
26 51.685 2800.000 100.000 69.188 244.784 224.235 0.01493 362.010 116.133
27 88.363 2680.645 99.722 68.650 260.370 214.040 0.01155 447.690 106.632
28 72.044 2800.000 100.000 68.877 243.235 220.840 0.01302 385.352 113.806
29 100.000 1214.641 94.123 35.098 280.656 219.542 0.02177 767.463 48.497
30 100.000 1000.000 95.219 26.067 261.999 228.975 0.02584 807.405 40.957

From the coverage values, it can be observed that two solutions from the Pareto-
front of the Jaya, and three solutions from the Pareto-fronts of the SAP-Rao and
ERao-1 algorithm are dominated. Similarly, five solutions are dominated from the
Pareto-fronts achieved by the AMTPG-Jaya, Rao-1, and ERao-3 algorithms; six
solutions are dominated from the Pareto-front achieved by the ERao-2 algorithms;
eight solutions are dominated from the Pareto-front achieved by the Rao-3 algorithm;
284 8 Optimization of the Selected Bio-Energy Systems

Table 8.21 Pareto-optimal solutions obtained by the ERao-3 algorithm for the turbocharged DI
diesel engine case study
SN X 1 (%) X 2 (rpm) X 3 (%) P (kW) T (N.m) BSFC CO (%) NOx (ppm) HC
(g/kWh) (ppm)
1 100.000 1990.415 93.520 58.010 301.754 204.041 0.01205 614.732 77.019
2 100.000 1000.000 97.588 26.870 269.224 227.619 0.02597 830.234 40.770
3 100.000 1000.000 97.497 26.839 268.946 227.672 0.02596 829.340 40.777
4 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
5 100.000 1046.716 94.936 28.135 266.492 226.738 0.02490 798.631 42.588
6 100.000 1668.053 93.066 50.224 301.766 206.983 0.01512 680.164 64.896
7 100.000 1936.894 93.366 56.873 302.614 204.191 0.01246 625.811 74.975
8 54.062 2800.000 100.000 69.143 244.448 223.779 0.01468 364.248 115.884
9 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
10 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
11 100.000 2508.413 96.884 66.241 275.844 209.431 0.01035 502.780 97.579
12 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
13 85.361 2729.840 100.000 68.863 254.016 216.140 0.01187 428.672 109.128
14 100.000 1344.416 93.641 39.962 289.241 214.928 0.01960 742.977 53.122
15 100.000 1277.442 93.873 37.507 285.067 217.207 0.02069 755.660 50.728
16 100.000 2493.833 96.739 66.071 277.011 209.112 0.01034 506.048 96.978
17 48.919 2800.000 100.000 69.243 245.226 224.786 0.01523 359.568 116.416
18 53.264 2800.000 100.000 69.158 244.556 223.930 0.01476 363.483 115.968
19 82.958 2773.446 100.000 68.833 247.575 218.279 0.01214 410.500 111.275
20 65.597 2800.000 100.000 68.958 243.401 221.789 0.01357 376.938 114.590
21 100.000 2507.748 96.877 66.234 275.898 209.416 0.01035 502.929 97.552
22 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
23 100.000 1000.000 95.664 26.218 263.363 228.720 0.02586 811.649 40.920
24 63.347 2800.000 100.000 68.990 243.529 222.148 0.01377 374.225 114.853
25 100.000 2129.819 94.080 60.701 297.917 204.280 0.01119 585.451 82.405
26 100.000 1579.868 93.128 47.682 299.605 208.648 0.01620 697.546 61.652
27 58.890 2800.000 100.000 69.059 243.893 222.901 0.01419 369.190 115.359
28 100.000 2416.454 96.019 65.112 282.775 207.581 0.01034 523.283 93.811
29 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
30 80.621 2800.000 100.000 68.796 243.480 219.757 0.01238 398.015 112.697

10 solutions are dominated from the Pareto-front achieved by the Rao-2 algorithm.
The performances of the proposed modified algorithms are better or competitive to
that of the basic algorithms in terms of coverage and spacing values.
From the computational results of the turbocharged DI diesel engine case study,
it can be observed that the solutions obtained by the proposed algorithms are better
than those reported in the literature. The solution of the ERao-1 algorithm has better
8.2 Design Optimization of a Turbocharged DI Diesel Engine 285

Table 8.22 Pareto-optimal solutions obtained by the SAP-Rao algorithm for the turbocharged DI
diesel engine case study
SN X 1 (%) X 2 (rpm) X 3 (%) P (kW) T (N.m) BSFC CO (%) NOx (ppm) HC
(g/kWh) (ppm)
1 100.000 1570.341 93.135 47.395 299.305 208.852 0.01632 699.381 61.303
2 53.929 2800.000 100.000 69.146 244.466 223.804 0.01469 364.120 115.898
3 97.232 2557.269 97.574 66.912 271.663 210.498 0.01058 487.510 100.052
4 50.815 2800.000 100.000 69.205 244.917 224.406 0.01502 361.224 116.223
5 100.000 1986.985 93.508 57.939 301.816 204.048 0.01208 615.436 76.887
6 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
7 79.240 2800.000 100.000 68.807 243.405 219.917 0.01247 395.863 112.881
8 100.000 2511.125 96.917 66.276 275.637 209.488 0.01035 502.207 97.692
9 69.134 2800.000 100.000 68.912 243.272 221.254 0.01326 381.437 114.165
10 100.000 1661.957 93.070 50.055 301.653 207.085 0.01519 681.389 64.671
11 100.000 2110.318 93.978 60.342 298.566 204.198 0.01129 589.511 81.646
12 100.000 1000.000 95.332 26.105 262.347 228.910 0.02584 808.483 40.948
13 100.000 1025.188 95.055 27.186 264.426 227.761 0.02533 802.587 41.837
14 100.000 1000.000 98.383 27.138 271.631 227.165 0.02602 838.030 40.714
15 49.783 2800.000 100.000 69.225 245.082 224.612 0.01514 360.312 116.329
16 100.000 2323.141 95.267 63.831 288.780 206.094 0.01047 543.823 90.044
17 49.242 2800.000 100.000 69.236 245.171 224.720 0.01520 359.845 116.383
18 100.000 2440.745 96.235 65.422 281.042 208.033 0.01033 517.892 94.801
19 48.294 2800.000 100.000 69.256 245.334 224.913 0.01530 359.041 116.479
20 100.000 1000.000 96.583 26.530 266.168 228.195 0.02591 820.471 40.846
21 80.379 2800.000 100.000 68.798 243.466 219.784 0.01239 397.636 112.729
22 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
23 100.000 1043.465 94.956 27.993 266.191 226.889 0.02497 799.251 42.475
24 100.000 1000.000 100.000 27.681 276.503 226.240 0.02613 854.110 40.610
25 48.297 2800.000 100.000 69.256 245.333 224.913 0.01530 359.044 116.479
26 100.000 2193.723 94.420 61.811 295.391 204.689 0.01089 571.833 84.907
27 97.741 2550.169 97.486 66.825 272.340 210.322 0.01054 490.029 99.676
28 100.000 2214.027 94.555 62.158 294.531 204.849 0.01081 567.598 85.707
29 100.000 1000.000 97.540 26.854 269.078 227.647 0.02596 829.764 40.774
30 100.000 1000.000 97.895 26.973 270.154 227.444 0.02598 833.235 40.748

compromise among the six objectives. The performance objectives P, T, BSFC, CO,
NOx , and HC values obtained by the ERao-1 solution are 63.89 kW, 288.55 N m,
206.15 g/kWh, 0.01046%, 542.98 and 90.20 ppm, respectively. The SAP-Rao algo-
rithm has achieved the least spacing value (0.10705), which is much better than that
achieved by other algorithms compared. The ERao-1, SAP-Rao, and Jaya algorithms
 286

Table 8.23 Best solutions obtained by various algorithms in MOO scenario of the turbocharged DI diesel engine case study
Algorithm X 1 (%) X 2 (rpm) X 3 (%) P (kW) T (N.m) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Spacing
ABC 85.63 2208 97 63 298 202.85 0.012 560 88 –
Jaya 100 2257.30 94.806172 62.84 292.39 205.28 0.01065 558.13 87.42 0.16279
Rao-1 100 2325.90 95.28581 63.87 288.61 206.13 0.01047 543.21 90.16 0.32805
Rao-2 100 2264.31 94.85232 62.95 292.03 205.36 0.01063 556.61 87.70 0.15134
Rao-3 100 2250.98 94.76524 62.74 292.71 205.22 0.01067 559.49 87.17 0.13963
AMTPG-Jaya 100 2294.01 95.05487 63.40 290.44 205.71 0.01054 550.16 88.88 0.21219
SAP-Rao 100 2323.14 95.26661 63.83 288.78 206.09 0.01047 543.82 90.04 0.10705
ERao-1 100 2326.93 95.29346 63.89 288.55 206.15 0.01046 542.98 90.20 0.29966
ERao-2 100 2293.78 95.05325 63.40 290.45 205.71 0.01054 550.21 88.87 0.22398
ERao-3 100 2416.45 96.01875 65.11 282.78 207.58 0.01034 523.28 93.81 0.12800
Source ABC—Shirneshan et al. (2016)
Result in boldface indicates better values
8 Optimization of the Selected Bio-Energy Systems
Table 8.24 Ranks suggested by the MADM methods for different algorithm solutions presented in Table 8.23
Algorithm SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA AR CR FR
ABC 10 10 10 10 9 6 10 2 8.31 9.83 10
Jaya 8 8 8 8 7 6 8 4 7.06 7.83 8
Rao-1 4 4 3 3 1 9 3 10 4.63 3.00 3
Rao-2 7 7 6 6 6 6 7 5 6.19 6.50 7
Rao-3 9 9 9 9 10 6 9 3 7.94 9.17 9
AMTPG-Jaya 6 6 4 4 5 10 6 7 6.00 5.17 6
SAP-Rao 3 3 2 2 1 2 2 9 3.00 2.17 2
ERao-1 2 2 1 1 1 6 1 8 2.69 1.33 1
8.2 Design Optimization of a Turbocharged DI Diesel Engine

ERao-2 5 5 5 5 1 1 5 6 4.13 4.33 4

ERao-3 1 1 7 7 8 6 4 1 4.31 4.67 5
AR—Average Rank; CR—Corrected Rank; FR—Final Rank
287
288 8 Optimization of the Selected Bio-Energy Systems

Table 8.25 Spearman’s rank correlation coefficients between different pairs of MADM method’s
rankings presented in Table 8.24
Method SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA
SAW 1 1 0.73 0.73 0.61 0.11 0.93 −0.39
WPM 1 1 0.73 0.73 0.61 0.11 0.93 −0.39
TOPSIS 0.73 0.73 1 1 0.91 −0.01 0.92 −0.88
MTOPSIS 0.73 0.73 1 1 0.91 −0.01 0.92 −0.88
VIKOR 0.61 0.61 0.91 0.91 1 0.16 0.81 −0.89
PROMETHEE 0.11 0.11 −0.01 −0.01 0.16 1 0.11 0.11
COPRAS 0.93 0.93 0.92 0.92 0.81 0.11 1 −0.68
GRA −0.39 −0.39 −0.88 −0.88 −0.89 0.11 −0.68 1

Table 8.26 Coverage (%) values of the Pareto-fronts obtained by the proposed algorithms in MOO
for turbocharged DI diesel engine case study
Algorithm Jaya Rao-1 Rao-2 Rao-3 AMTPG-Jaya SAP-Rao ERao-1 ERao-2 ERao-3
Jaya – 16.67 33.33 26.67 16.67 10 10 13.33 10
Rao-1 6.67 – 33.33 26.67 16.67 10 10 13.33 10
Rao-2 6.67 16.67 – 26.67 16.67 10 10 13.33 10
Rao-3 6.67 16.67 33.33 – 16.67 10 10 13.33 10
AMTPG-Jaya 6.67 16.67 33.33 26.67 – 10 10 13.33 10
SAP-Rao 6.67 16.67 33.33 26.67 16.67 – 10 13.33 10
ERao-1 6.67 16.67 33.33 26.67 16.67 10 – 13.33 10
ERao-2 6.67 16.67 33.33 26.67 16.67 10 10 – 16.67
ERao-3 6.67 16.67 33.33 26.67 16.67 10 10 20 –

have achieved better performance in terms of coverage values. In addition, the perfor-
mances of the modified versions in this case study are better or competitive to those
of the basic algorithms as well as the ABC algorithm.

8.3 Design Optimization of a Compression Ignition

Biodiesel Engine with an EGR System

The description of the selected compression ignition biodiesel engine with an EGR
system is presented in Sect. 2.3.3. In this case study, a compression ignition biodiesel
engine with an exhaust gas recirculation system is considered for multi-objective
optimization. This case study was presented by Jaliliantabar et al. (2019).
The design variables of this case study are the exhaust gas recirculation rate (ER),
engine load percentage (EL), engine speed (ES) in rpm, and biodiesel percentage
(BP). The ranges of design variables are as follows:0 ≤ ER ≤ 30, 25 ≤ EL ≤ 75,
8.3 Design Optimization of a Compression … 289

1800 ≤ ES ≤ 2400, and 0 ≤ BP ≤ 15. The objective functions of this case

study are maximization of power output (P), and minimization of the brake specific
fuel consumption (BSFC) and emissions such as carbon monooxide emission (CO),
nitrogen oxides emission (NOx ), hydrocarbon emission (HC), and smoke opacity
(Sm ). The regression models of the objectives of this case study are presented in
Sect. 2.3.3.
Jaliliantabar et al. (2019) reported a single Pareto-optimal solution for this case
study by implementing the NSGA-II algorithm. Now, the proposed algorithms, along
with basic Jaya and Rao algorithms, are used to find the optimal parameters. The
function evaluations taken by the NSGA-II algorithm was not specified. However,
the proposed and other considered algorithms are executed for 10,000 function eval-
uations. The population sizes of all the algorithms are taken as 30. In the case of the
SAP-Rao algorithm, 30 best-ranked solutions at termination are selected as the final
population. The elite population size is taken as 20% of the population for the elitist
Rao algorithms.
Furthermore, the solutions obtained by various algorithms are non-dominated.
Hence, the average rank method was used to identify the best solution from the Pareto-
front of the respective algorithm. Furthermore, the performances of the proposed
algorithms are compared with the performances of the basic Jaya and Rao algorithms
in terms of spacing and coverage performance metrics. The spacing performance
metric is calculated for ten independent runs, and the statistical results are presented
in terms of best, worst, mean, and standard deviation (SD).
Table 8.27 presents the optimal design parameters achieved by the SAP-Rao algo-
rithm in single-objective optimization scenarios of each objective. In all the single
objective optimization scenarios, the solutions obtained by the Jaya algorithm, Rao
algorithms, and their modified versions are identical. Hence, the solutions achieved
by the SAP-Rao algorithm in the respective scenarios are reported in Table 8.27.
The Pareto-optimal solutions achieved by the Jaya algorithm, Rao algorithms, and
their modified versions in this case study are presented in Tables 8.28, 8.29, 8.30,
8.31, 8.32, 8.33, 8.34, 8.35 and 8.36. The power is varied between 0.86 and 3.53 kW,
the BSFC is varied from 218 to 577 g/kWh, the CO emission is varied from 0.00044
to 2.16%, the NOx emission is varied from 86 to 387 ppm, HC emission is varied
from 16.01 to 250 ppm, and smoke emission is varied from 2.3 to 17.4 m−1 .

Table 8.27 Optimal solutions obtained by the SAP-Rao algorithm in single-objective optimization
scenarios of the biodiesel engine with an EGR system case study
Objective ER (%) EL (%) ES (rpm) BP (%) Objective value
P (kW) 30 75 2400 0 3.52314
CO (%) 30 25 1890.68 15 1.42286E-08
NOx (ppm) 28.12839 25 2162.72 15 86.83693804
HC (ppm) 0 41.95999 1800 15 16.01092
BSFC (g/kWh) 5.18747 62.24392 1800 8.4683 218.6337404
Sm (1/m) 0 30.82292 1800 11.0903 2.301684075
Table 8.28 Pareto-optimal solutions obtained by the Jaya algorithm for the biodiesel engine with an EGR system case study
290

Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
1 0.000 42.020 1800.000 15.000 1.890 316.461 0.072 290.955 16.011 2.738
2 0.000 30.984 1800.000 10.894 1.361 409.994 0.117 218.026 22.028 2.302
3 30.000 25.000 2161.243 15.000 1.087 564.309 0.090 86.943 46.484 3.741
4 6.012 61.202 1800.000 8.078 2.548 218.904 0.399 357.422 54.371 6.732
5 29.378 25.988 1849.770 10.288 1.204 494.819 0.000 146.656 32.025 2.800
6 0.000 25.000 1992.674 2.002 0.831 548.314 0.162 130.140 45.304 2.773
7 0.000 25.000 2097.867 1.224 0.832 570.621 0.172 122.347 50.791 2.948
8 30.000 75.000 2400.000 0.000 3.523 424.208 2.160 256.076 249.948 17.390
9 0.000 71.693 1800.000 3.967 2.853 244.308 0.866 387.349 94.022 10.637
10 30.000 75.000 2400.000 1.848 3.503 411.455 2.077 251.066 239.231 16.854
11 30.000 75.000 2377.888 11.667 3.357 386.224 1.654 214.164 189.290 14.419
12 30.000 58.249 1948.955 4.356 2.445 292.100 0.484 278.509 89.589 7.341
13 19.402 72.140 1856.021 6.844 2.973 267.791 0.811 312.059 109.727 10.782
14 11.970 53.542 1800.000 2.226 2.170 247.489 0.358 330.792 60.766 5.974
15 29.323 75.000 2384.198 5.827 3.425 392.300 1.874 238.051 215.640 15.686
16 11.435 75.000 2053.031 12.919 3.009 343.749 1.011 271.798 126.986 11.725
17 30.000 75.000 2325.718 14.665 3.271 392.088 1.448 197.393 170.654 13.517
18 30.000 68.909 2396.850 3.834 3.227 367.370 1.643 259.566 195.277 13.767
19 0.441 25.000 1800.000 6.546 1.019 484.007 0.180 166.953 30.970 2.469
20 30.000 75.000 2400.000 6.429 3.449 390.737 1.888 236.584 215.312 15.676
21 20.082 75.000 1907.773 5.199 3.044 304.076 1.056 295.028 137.804 12.521
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.28 (continued)
Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
22 30.000 65.135 1995.096 8.373 2.757 298.936 0.637 260.534 102.257 8.822
23 0.000 27.056 2002.436 3.695 0.942 516.925 0.144 147.556 42.281 2.682
24 0.000 66.932 1800.000 1.124 2.631 243.933 0.788 384.518 89.300 9.623
25 30.000 42.333 2264.820 8.645 1.903 382.201 0.340 207.263 72.227 5.151
26 5.550 64.712 1800.000 3.088 2.598 230.785 0.643 367.217 80.551 8.640
27 0.000 27.946 1800.000 15.000 1.309 464.611 0.161 188.335 25.830 2.458
28 8.981 42.783 2145.635 5.634 1.700 377.501 0.293 231.746 64.452 4.534
29 28.210 70.095 1981.642 11.391 2.960 308.402 0.716 251.823 107.070 9.862
30 10.246 25.748 2175.782 13.267 0.979 550.222 0.161 106.736 49.393 3.328
8.3 Design Optimization of a Compression …
291
Table 8.29 Pareto-optimal solutions obtained by the AMTPG-Jaya algorithm for the biodiesel engine with an EGR system case study
292

Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
1 0.000 42.077 1800.000 15.000 1.893 316.017 0.072 291.288 16.012 2.742
2 0.000 31.343 1800.000 11.381 1.385 406.998 0.114 220.556 21.560 2.303
3 30.000 25.000 2161.219 15.000 1.087 564.308 0.090 86.943 46.483 3.741
4 25.712 46.073 1878.288 13.441 2.070 312.306 0.000 245.005 32.131 3.689
5 0.000 71.687 1800.000 6.809 2.896 237.333 0.773 386.442 83.199 10.025
6 2.850 63.752 1800.000 8.443 2.638 219.206 0.466 370.217 56.936 7.287
7 0.775 25.000 2069.627 0.000 0.818 572.216 0.180 122.608 52.868 3.067
8 30.000 75.000 2400.000 0.733 3.515 418.849 2.126 254.146 245.625 17.173
9 30.000 75.000 2352.340 12.434 3.319 387.064 1.573 208.803 182.539 14.081
10 30.000 75.000 2396.374 7.234 3.434 388.721 1.850 233.315 210.956 15.463
11 30.000 75.000 2400.000 5.906 3.456 392.315 1.908 238.384 217.849 15.800
12 0.000 25.227 1800.000 0.000 0.892 501.229 0.221 167.032 39.512 2.836
13 26.003 33.233 2400.000 15.000 1.538 458.455 0.327 160.011 60.135 4.676
14 30.000 75.000 2400.000 0.878 3.513 417.832 2.120 253.754 244.778 17.131
15 30.000 25.000 1938.071 15.000 1.170 543.703 0.006 106.402 35.074 3.128
16 23.486 66.170 2389.103 9.420 2.976 342.463 1.296 255.044 155.411 11.487
17 30.000 75.000 2319.562 15.000 3.262 393.063 1.425 195.585 168.616 13.420
18 11.127 59.763 2028.955 12.736 2.433 299.121 0.442 278.581 72.853 6.803
19 18.198 59.960 2400.000 10.249 2.675 331.936 1.024 265.692 126.695 9.321
20 22.859 74.583 1800.000 1.062 3.023 281.142 1.079 328.817 139.279 12.824
21 6.456 65.674 2306.992 9.682 2.719 346.879 1.090 289.051 130.349 10.155
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.29 (continued)
Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
22 0.000 25.000 2054.034 5.037 0.856 550.044 0.147 124.568 43.617 2.661
23 0.000 60.729 1800.000 13.008 2.585 227.793 0.304 368.882 36.798 5.833
24 19.874 69.307 1824.075 11.590 2.959 247.671 0.521 312.469 77.326 8.795
25 30.000 43.537 2400.000 7.386 2.107 370.547 0.566 230.943 88.534 6.166
26 18.525 66.823 1914.548 7.469 2.747 270.389 0.656 300.149 97.052 9.140
27 30.000 75.000 2158.985 9.830 3.182 368.471 1.267 219.017 162.497 13.097
28 17.200 43.910 2156.799 11.233 1.815 367.350 0.239 214.474 59.897 4.534
29 0.000 28.852 1800.000 14.854 1.345 452.015 0.148 196.317 24.531 2.420
30 18.661 75.000 1873.335 14.021 3.150 294.184 0.717 287.019 97.452 10.624
8.3 Design Optimization of a Compression …
293
Table 8.30 Pareto-optimal solutions obtained by the Rao-1 algorithm for the biodiesel engine with an EGR system case study
294

Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
1 0.000 30.567 1800.000 10.864 1.343 414.947 0.121 214.639 22.402 2.302
2 0.000 41.993 1800.000 15.000 1.889 316.677 0.072 290.793 16.011 2.735
3 23.749 28.556 1800.000 13.944 1.409 454.528 0.001 168.852 26.127 2.731
4 29.586 25.000 2154.583 15.000 1.084 563.987 0.088 86.926 46.412 3.721
5 6.484 62.534 1800.000 8.510 2.608 218.703 0.422 357.893 56.775 7.036
6 0.450 69.368 1800.000 4.932 2.784 233.771 0.747 385.098 82.687 9.613
7 30.000 75.000 2400.000 0.000 3.523 424.208 2.160 256.076 249.948 17.390
8 0.522 25.000 2263.460 0.000 0.931 591.255 0.230 127.383 59.038 3.359
9 1.341 25.000 2018.976 3.297 0.847 547.345 0.152 126.502 45.627 2.786
10 29.975 74.559 2400.000 13.267 3.346 385.895 1.631 210.914 184.387 14.148
11 30.000 66.744 2301.641 15.000 2.925 354.537 1.010 212.792 129.877 10.348
12 30.000 75.000 2400.000 4.225 3.475 398.763 1.976 243.916 226.344 16.215
13 30.000 69.881 2228.089 7.384 3.034 353.317 1.231 236.237 159.731 12.114
14 26.151 72.618 2391.127 14.595 3.201 377.472 1.473 218.468 168.634 13.095
15 26.862 75.000 2152.880 5.462 3.152 371.760 1.420 239.525 180.473 14.022
16 5.801 60.774 1896.858 7.080 2.448 256.468 0.480 330.285 69.241 7.110
17 0.000 71.894 1818.750 14.382 2.991 256.837 0.597 372.005 65.132 8.952
18 25.365 64.342 2400.000 3.408 3.001 352.834 1.421 272.931 173.811 12.125
19 30.000 75.000 2400.000 0.782 3.515 418.503 2.124 254.014 245.338 17.159
20 29.643 44.639 2370.336 15.000 2.055 377.221 0.446 200.928 72.700 5.664
21 30.000 75.000 2400.000 3.468 3.484 402.353 2.007 246.280 230.340 16.412
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.30 (continued)
Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
22 0.000 46.229 1862.286 9.992 1.911 292.669 0.142 301.795 28.103 3.485
23 30.000 55.591 2302.769 10.798 2.498 334.089 0.694 232.463 102.550 7.682
24 28.036 56.853 2293.258 15.000 2.498 342.603 0.641 218.799 95.060 7.528
25 8.646 65.839 2066.941 5.881 2.623 312.512 0.852 297.845 114.695 9.680
26 30.000 25.000 2286.419 9.064 1.130 546.286 0.182 114.576 56.307 4.076
27 30.000 60.002 2309.842 11.373 2.684 333.450 0.845 232.827 116.221 8.812
28 30.000 30.523 2400.000 5.795 1.523 476.051 0.352 177.085 70.267 4.836
29 6.032 58.228 1800.000 15.000 2.546 237.983 0.201 340.836 33.380 5.274
30 23.802 52.854 2155.279 12.720 2.242 332.286 0.381 228.110 74.524 6.029
8.3 Design Optimization of a Compression …
295
Table 8.31 Pareto-optimal solutions obtained by the Rao-2 algorithm for the biodiesel engine with an EGR system case study
296

Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
1 27.643 25.000 2169.690 15.000 1.068 563.530 0.106 86.863 48.052 3.768
2 4.999 61.961 1800.000 7.719 2.567 218.853 0.430 361.847 56.424 6.964
3 0.000 30.090 1800.000 11.065 1.326 421.273 0.125 210.554 22.778 2.304
4 30.000 25.000 1800.000 12.931 1.268 504.354 0.000 144.332 27.355 2.768
5 0.000 41.933 1800.000 15.000 1.887 317.144 0.072 290.444 16.011 2.731
6 0.000 25.000 2126.198 11.206 0.889 560.184 0.161 116.718 42.786 2.692
7 0.000 68.145 1800.000 4.758 2.734 231.291 0.710 385.811 78.768 9.217
8 30.000 75.000 2400.000 1.249 3.509 415.308 2.103 252.740 242.634 17.024
9 0.000 25.000 2333.125 15.000 0.924 580.589 0.275 121.555 49.643 3.215
10 30.000 75.000 2400.000 5.315 3.463 394.344 1.932 240.374 220.778 15.943
11 23.923 65.938 2400.000 0.000 3.089 378.219 1.622 281.555 197.310 13.504
12 11.060 34.513 2219.750 5.979 1.385 447.748 0.231 183.601 60.093 3.883
13 16.417 75.000 2365.919 6.766 3.262 389.166 1.768 261.837 200.992 15.063
14 19.874 75.000 1969.442 0.000 2.991 349.807 1.358 289.249 174.867 14.266
15 28.062 25.000 1800.000 15.000 1.300 511.886 0.015 137.623 27.951 2.886
16 30.000 63.485 2400.000 3.418 3.011 351.522 1.391 268.179 171.632 11.893
17 28.680 58.028 2400.000 0.000 2.798 365.038 1.257 279.040 163.704 10.858
18 0.000 73.337 1909.708 0.000 2.819 317.761 1.185 354.583 135.921 12.739
19 28.124 75.000 2163.837 0.000 3.172 402.749 1.676 249.998 212.051 15.574
20 7.138 64.799 1800.000 2.794 2.603 232.272 0.654 363.503 83.556 8.773
21 0.482 65.205 2125.464 10.792 2.568 326.257 0.773 308.574 96.607 8.596
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.31 (continued)
Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
22 0.000 68.258 1966.150 2.629 2.646 304.661 0.929 341.816 112.222 10.379
23 20.335 75.000 1813.400 1.057 3.018 286.152 1.117 328.547 142.179 13.079
24 14.712 59.237 1971.165 4.373 2.382 286.628 0.561 295.289 90.587 7.678
25 28.004 36.330 2332.027 6.442 1.687 426.060 0.349 195.064 72.706 4.923
26 19.540 62.520 2266.864 7.361 2.676 332.972 0.971 260.842 130.847 9.750
27 0.000 25.000 1800.000 6.506 1.017 484.278 0.182 167.441 30.684 2.451
28 8.723 36.454 2391.112 15.000 1.530 442.369 0.367 195.832 59.718 4.247
29 0.000 71.302 1941.942 1.358 2.753 312.919 1.076 347.664 126.050 11.715
30 0.000 45.481 1800.000 0.000 1.761 296.953 0.290 315.886 44.361 4.455
8.3 Design Optimization of a Compression …
297
Table 8.32 Pareto-optimal solutions obtained by the Rao-3 algorithm for the biodiesel engine with an EGR system case study
298

Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
1 27.776 25.000 2160.506 15.000 1.069 563.399 0.100 86.843 47.590 3.741
2 5.221 62.031 1800.000 7.501 2.567 218.985 0.437 361.555 57.440 7.023
3 0.000 41.676 1800.000 15.000 1.877 319.184 0.072 288.920 16.015 2.711
4 30.000 25.000 1800.000 12.941 1.268 504.391 0.000 144.304 27.350 2.768
5 0.000 31.282 1800.000 11.376 1.383 407.715 0.115 220.063 21.614 2.302
6 0.000 25.000 2185.362 0.000 0.867 587.425 0.201 122.356 56.129 3.183
7 0.000 25.000 2058.976 0.000 0.813 571.259 0.179 123.675 51.654 3.007
8 30.000 74.976 2400.000 2.719 3.492 406.157 2.037 248.612 234.251 16.602
9 0.000 75.000 1800.000 6.769 3.016 251.354 0.902 385.441 95.371 11.262
10 27.008 32.422 2309.250 6.215 1.476 463.028 0.276 172.120 67.050 4.491
11 13.295 61.429 1800.000 15.000 2.702 236.356 0.233 325.905 42.575 6.098
12 30.000 56.070 2326.973 2.688 2.592 350.552 0.964 261.082 135.923 9.231
13 23.049 60.578 2016.259 4.514 2.492 301.027 0.638 275.878 104.078 8.333
14 28.420 30.193 2400.000 0.000 1.530 505.750 0.432 188.810 85.740 5.364
15 30.000 70.863 2358.541 15.000 3.136 371.147 1.307 208.894 155.082 12.168
16 0.000 75.000 1800.000 13.092 3.105 256.721 0.718 379.217 75.251 10.125
17 26.971 25.000 1961.108 15.000 1.133 545.020 0.031 102.578 38.167 3.217
18 0.000 25.000 1806.297 9.024 1.060 486.971 0.177 165.133 29.570 2.421
19 24.002 75.000 1800.000 7.229 3.148 262.848 0.838 314.591 111.122 11.424
20 30.000 75.000 2246.424 15.000 3.211 388.704 1.272 194.456 157.312 12.860
21 21.497 52.685 1920.748 3.904 2.158 289.192 0.344 283.065 72.207 6.014
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.32 (continued)
Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
22 30.000 38.197 2400.000 2.081 1.906 422.617 0.544 226.404 93.029 5.997
23 18.810 67.825 2362.261 15.000 2.904 358.201 1.178 241.484 140.267 10.991
24 11.084 46.727 1800.000 10.269 2.038 266.491 0.087 298.123 28.654 3.706
25 27.910 62.539 2400.000 10.446 2.874 333.629 1.134 247.722 139.246 10.282
26 17.697 75.000 2003.670 12.217 3.061 330.005 0.949 262.775 124.803 11.646
27 17.489 53.688 1924.644 15.000 2.306 292.288 0.155 265.757 45.362 4.968
28 30.000 25.000 2382.543 14.613 1.184 549.795 0.261 107.233 55.159 4.518
29 30.000 59.205 2195.955 15.000 2.567 338.083 0.556 215.769 90.315 7.505
30 0.000 37.189 1800.000 9.568 1.597 340.655 0.095 265.523 19.811 2.520
8.3 Design Optimization of a Compression …
299
Table 8.33 Pareto-optimal solutions obtained by the ERao-1 algorithm for the biodiesel engine with an EGR system case study
300

Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
1 4.905 62.083 1800.000 8.889 2.591 218.709 0.405 361.159 53.314 6.818
2 30.000 25.000 1896.402 14.469 1.197 533.404 0.000 116.133 32.759 3.011
3 0.000 30.978 1800.000 11.099 1.365 410.586 0.117 217.822 21.961 2.302
4 28.350 25.000 2162.771 15.000 1.074 563.647 0.099 86.838 47.415 3.748
5 0.000 42.340 1800.000 15.000 1.903 313.967 0.073 292.829 16.018 2.763
6 0.000 70.533 1800.000 8.807 2.884 232.185 0.673 384.988 72.680 9.249
7 0.000 25.000 2146.485 15.000 0.898 576.037 0.193 110.696 44.714 2.880
8 30.000 75.000 2400.000 1.151 3.510 415.967 2.107 253.011 243.200 17.052
9 0.000 25.000 2098.864 6.808 0.868 553.977 0.148 121.431 43.597 2.657
10 30.000 75.000 2400.000 7.151 3.441 388.887 1.861 234.032 211.884 15.511
11 30.000 75.000 2340.388 12.402 3.307 386.760 1.548 208.125 180.848 13.992
12 30.000 75.000 2276.294 2.108 3.312 404.552 1.800 240.785 218.778 15.803
13 30.000 70.560 2172.665 9.690 3.017 347.898 1.086 228.807 145.007 11.532
14 30.000 75.000 2400.000 2.202 3.499 409.299 2.061 250.050 237.246 16.755
15 19.693 35.408 2353.998 10.646 1.573 432.536 0.322 183.774 64.227 4.508
16 30.000 75.000 2195.368 7.327 3.209 375.820 1.427 225.596 179.353 13.888
17 20.770 35.408 2228.764 11.353 1.494 435.696 0.193 169.828 56.185 4.017
18 30.000 75.000 1953.362 13.199 3.183 328.546 0.811 238.373 115.058 11.093
19 27.755 70.646 2259.762 8.277 3.066 356.821 1.296 235.811 163.673 12.444
20 30.000 75.000 1883.538 13.312 3.221 307.324 0.708 256.341 102.423 10.636
21 0.133 54.783 1800.000 6.122 2.247 231.881 0.311 357.363 40.846 5.282
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.33 (continued)
Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
22 0.000 45.509 1800.000 3.165 1.822 281.782 0.224 317.438 35.168 3.970
23 30.000 66.860 2175.816 10.769 2.871 335.732 0.899 230.254 126.177 10.095
24 19.801 65.965 1837.314 5.184 2.727 246.939 0.620 322.913 91.777 8.940
25 20.530 72.514 1817.279 13.278 3.112 260.956 0.573 305.449 81.686 9.557
26 24.569 66.921 2104.642 6.115 2.790 324.935 0.940 259.042 133.209 10.507
27 30.000 66.860 1925.840 11.852 2.875 286.844 0.506 262.805 85.217 8.399
28 6.151 25.000 1815.507 11.873 1.121 493.562 0.151 152.878 32.564 2.682
29 0.000 26.813 1800.000 10.080 1.167 461.400 0.158 182.901 26.604 2.359
30 30.000 75.000 2234.350 13.444 3.212 382.650 1.294 201.402 160.676 13.005
8.3 Design Optimization of a Compression …
301
Table 8.34 Pareto-optimal solutions obtained by the ERao-2 algorithm for the biodiesel engine with an EGR system case study
302

Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
1 19.981 43.830 1800.000 11.854 1.992 292.973 0.000 268.552 25.473 3.315
2 27.626 25.000 2159.930 15.000 1.068 563.336 0.101 86.848 47.635 3.739
3 5.719 62.420 1800.000 8.241 2.596 218.669 0.428 360.111 56.652 7.027
4 0.000 42.023 1800.000 15.000 1.891 316.436 0.072 290.974 16.011 2.738
5 0.000 31.547 1800.000 11.482 1.396 404.875 0.113 222.114 21.359 2.304
6 0.000 25.000 2306.596 15.000 0.912 581.378 0.260 118.372 49.061 3.159
7 0.000 25.000 1955.282 1.917 0.838 540.271 0.165 134.890 43.821 2.748
8 5.493 69.067 1800.000 7.101 2.826 228.375 0.663 368.019 79.204 9.233
9 30.000 75.000 2400.000 1.179 3.510 415.782 2.106 252.935 243.041 17.044
10 18.931 63.018 2400.000 4.302 2.876 346.906 1.312 280.894 160.207 11.306
11 30.000 75.000 2369.484 15.000 3.308 393.961 1.535 198.727 176.159 13.814
12 30.000 75.000 2400.000 6.334 3.451 391.007 1.892 236.912 215.766 15.699
13 30.000 32.660 2400.000 9.460 1.596 450.843 0.341 177.085 65.335 4.768
14 14.168 75.000 1840.842 3.970 3.016 278.815 1.031 330.477 128.194 12.426
15 30.000 75.000 2400.000 11.933 3.381 386.382 1.696 215.295 191.565 14.550
16 30.000 75.000 2400.000 2.385 3.497 408.223 2.053 249.520 236.231 16.704
17 30.000 29.396 2400.000 9.681 1.442 484.241 0.307 157.702 61.944 4.592
18 24.177 25.000 1800.000 15.000 1.278 508.136 0.045 138.157 30.167 2.945
19 15.486 47.382 2343.443 8.475 2.073 356.789 0.559 244.543 86.678 6.089
20 10.053 75.000 1800.000 7.319 3.068 251.504 0.870 351.286 103.425 11.401
21 30.000 36.093 2400.000 15.000 1.706 432.594 0.349 172.468 61.777 4.903
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.34 (continued)
Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
22 30.000 66.023 1990.820 4.076 2.755 308.824 0.812 272.711 122.955 9.934
23 11.008 50.835 1800.000 10.932 2.214 246.659 0.124 313.702 31.758 4.269
24 19.628 73.073 2076.026 11.368 2.994 337.462 0.997 251.235 132.170 11.519
25 29.673 70.826 2388.052 13.896 3.174 366.970 1.397 217.392 162.462 12.527
26 5.529 49.194 2206.387 6.728 1.982 351.969 0.462 262.620 76.093 5.591
27 30.000 75.000 2255.346 11.547 3.237 380.905 1.396 208.966 170.978 13.485
28 28.511 25.000 2017.495 4.913 0.996 539.701 0.076 123.885 51.112 3.386
29 27.720 56.235 2143.901 5.786 2.400 325.688 0.609 250.299 102.727 7.613
30 30.000 75.000 1935.987 8.379 3.145 317.987 0.953 259.010 131.075 11.881
8.3 Design Optimization of a Compression …
303
Table 8.35 Pareto-optimal solutions obtained by the ERao-3 algorithm for the biodiesel engine with an EGR system case study
304

Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
1 29.867 25.000 1800.000 15.000 1.311 513.937 0.000 137.682 26.716 2.848
2 0.000 30.989 1800.000 10.911 1.361 409.979 0.117 218.052 22.017 2.302
3 0.000 41.895 1800.000 15.000 1.885 317.444 0.072 290.219 16.011 2.728
4 27.357 25.000 2171.558 15.000 1.066 563.473 0.108 86.885 48.262 3.773
5 1.928 63.037 1800.000 8.534 2.609 219.055 0.446 371.823 53.969 7.043
6 30.000 75.000 2400.000 0.000 3.523 424.208 2.160 256.076 249.948 17.390
7 0.000 25.000 2139.850 15.000 0.900 575.441 0.191 110.805 44.503 2.871
8 0.000 25.000 2129.855 8.324 0.877 557.117 0.154 119.545 43.672 2.675
9 0.000 69.731 1800.000 7.005 2.828 230.796 0.697 385.914 75.945 9.305
10 30.000 71.998 2400.000 8.737 3.302 368.329 1.639 236.538 188.997 13.939
11 30.000 75.000 2102.253 8.411 3.156 359.287 1.216 229.231 159.267 12.985
12 30.000 66.596 2400.000 7.440 3.097 346.685 1.406 251.774 168.223 12.158
13 26.378 72.771 2400.000 3.961 3.351 385.223 1.847 256.839 213.765 15.279
14 9.552 75.000 2035.345 15.000 3.005 346.382 0.933 275.843 117.006 11.280
15 30.000 75.000 2400.000 3.924 3.479 400.140 1.988 244.866 227.921 16.293
16 30.000 75.000 2400.000 10.086 3.405 385.343 1.756 222.911 198.925 14.893
17 30.000 37.059 2353.974 15.000 1.710 429.216 0.302 170.096 60.094 4.749
18 20.508 69.508 2086.062 6.608 2.854 326.345 1.007 265.275 137.292 11.174
19 28.840 56.259 1940.448 0.000 2.301 312.255 0.566 287.243 101.783 7.701
20 22.916 75.000 2037.558 9.720 3.086 338.798 1.072 249.749 141.514 12.324
21 30.000 34.510 2253.641 15.000 1.537 456.784 0.167 148.476 51.689 4.130
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.35 (continued)
Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
22 30.000 75.000 2400.000 2.772 3.492 406.027 2.037 248.382 234.103 16.599
23 0.000 25.000 1800.000 13.410 1.153 497.183 0.195 163.656 29.460 2.522
24 0.000 55.588 1800.000 4.632 2.252 232.954 0.361 360.646 46.429 5.684
25 0.500 67.582 2098.699 15.000 2.658 337.176 0.743 304.802 91.731 8.726
26 29.881 73.546 1800.000 15.000 3.260 276.806 0.496 280.378 76.713 9.386
27 5.703 60.666 1854.740 7.149 2.469 241.250 0.445 341.493 63.029 6.914
28 18.037 50.083 1800.000 3.693 2.085 258.114 0.236 310.285 52.267 5.078
29 16.649 25.000 2323.298 15.000 1.028 561.634 0.244 100.656 55.869 4.041
30 10.647 65.579 2111.080 9.714 2.643 317.097 0.795 281.075 108.616 9.223
8.3 Design Optimization of a Compression …
305
Table 8.36 Pareto-optimal solutions obtained by the SAP-Rao algorithm for the biodiesel engine with an EGR system case study
306

Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
1 0.000 41.608 1800.000 15.000 1.874 319.726 0.072 288.515 16.017 2.706
2 2.573 62.466 1800.000 9.000 2.598 218.952 0.418 368.509 51.897 6.838
3 30.000 25.000 2164.146 15.000 1.087 564.355 0.092 86.943 46.614 3.750
4 0.000 30.081 1800.000 10.765 1.320 420.639 0.125 210.699 22.879 2.304
5 0.039 71.261 1800.000 7.199 2.887 235.347 0.745 386.066 80.418 9.799
6 30.000 25.000 1895.402 15.000 1.207 535.929 0.000 114.501 32.575 3.033
7 0.000 25.000 2156.082 15.000 0.896 576.845 0.195 110.599 45.015 2.894
8 0.000 25.000 2006.552 5.339 0.866 540.802 0.144 128.730 41.406 2.600
9 30.000 75.000 2400.000 0.000 3.523 424.208 2.160 256.076 249.948 17.390
10 21.019 74.220 2400.000 8.614 3.298 380.535 1.746 250.382 197.299 14.726
11 30.000 74.574 2400.000 15.000 3.324 390.991 1.582 202.916 178.579 13.891
12 16.915 75.000 2400.000 5.040 3.336 395.523 1.903 268.393 214.100 15.729
13 30.000 75.000 2400.000 1.607 3.505 412.975 2.087 251.746 240.594 16.922
14 29.825 75.000 2400.000 4.381 3.472 398.013 1.969 243.709 225.472 16.174
15 19.350 65.053 2229.668 10.244 2.735 332.716 0.936 252.536 125.951 9.861
16 30.000 67.251 2349.973 12.677 3.008 349.942 1.183 225.179 145.346 11.154
17 29.077 51.512 2136.747 7.645 2.215 332.489 0.414 236.483 82.577 6.201
18 10.248 68.575 1800.000 9.327 2.865 227.390 0.570 351.145 73.983 8.769
19 30.000 75.000 2255.077 15.000 3.216 389.408 1.290 194.374 158.665 12.925
20 0.000 57.276 1800.000 9.685 2.402 225.008 0.290 363.642 36.694 5.371
21 2.573 48.956 1800.000 10.333 2.096 254.435 0.141 325.837 25.783 3.819
(continued)
8 Optimization of the Selected Bio-Energy Systems
Table 8.36 (continued)
Solution ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
22 0.000 26.266 1800.000 8.369 1.110 466.624 0.165 178.753 27.945 2.380
23 23.501 67.410 1899.793 7.779 2.812 270.468 0.640 292.914 97.236 9.191
24 30.000 75.000 1893.304 4.722 3.113 311.412 1.039 280.125 140.763 12.457
25 0.000 52.570 1800.000 12.297 2.262 243.823 0.170 345.434 24.795 4.182
26 29.825 53.749 2305.836 6.440 2.445 339.513 0.740 245.849 110.963 7.828
27 18.792 32.541 2150.875 9.740 1.324 457.749 0.130 157.341 51.722 3.591
28 26.329 46.882 2168.578 13.360 2.017 358.695 0.255 207.096 62.543 5.007
29 14.377 40.087 2400.000 13.640 1.756 402.843 0.428 210.913 67.840 4.899
30 20.360 74.794 2400.000 4.152 3.371 396.828 1.937 263.050 220.051 15.967
8.3 Design Optimization of a Compression …
307
308 8 Optimization of the Selected Bio-Energy Systems

Similar to the previous case studies, the best solutions from these Pareto-fronts
are identified. Solution 2 of the Jaya algorithm, Solution 4 of the AMTPG-Jaya
algorithm, Solution 1 of the Rao-1 algorithm, Solution 3 of the Rao-2 algorithm,
Solution 24 of the Rao-3 algorithm, Solution 2 of the ERao-1 algorithm, Solution 1
of the ERao-2 algorithm, Solution 3 of the ERao-3 algorithm, and Solution 1 of the
SAP-Rao algorithm are identified as the best solutions from the respective algorithm
Pareto-front. Now, these best solutions are compared with those of the NSGA-II
algorithm in Table 8.37.
In Table 8.37, the solutions with the best average rank from the Pareto-optimal
solutions of the various algorithms are compared with that reported for the NSGA-II
algorithm. The AMPG-Jaya algorithm solution achieved a higher power output, the
Rao-3 solution has the least BSFC, the ERao-3 solution has the least HC emission,
the Rao-3 solution has the least smoke opacity, and the ERao-1 solution has the
least CO and NOx emission values. Here, an observation can be made that these
solutions are non-dominated. Hence, to identify the best solution among the solutions
of different algorithms, which has the best compromise among the objectives, the
MADM methods based average ranks are calculated and presented in Table 8.38.
The Spearman’s correlation coefficients for different pairs of MADM methods
rankings for different algorithm’s solutions are shown in Table 8.39. The ranks given
by the TOPSIS and MTOPSIS methods for each solution are identical. Furthermore,
the Spearman’s correlation for all the pairs of decision-making methods (except with
the VIKOR method) is greater than 0.5. Spearman’s correlation values for the pairs
consisting of the VIKOR method are either negative or nearer to zero. Hence, the
ranks suggested by the VIKOR method cannot be considered for calculating the
average ranks. Now, the corrected average ranks are calculated, excluding the ranks
suggested by the VIKOR method, and presented in Table 8.38 as the corrected ranks.
The ERao-3 algorithm solution has the least average rank, which is 1.6. All the
MADM methods, except the WPM and SAW, have ranked the ERao-2 solution as
one. The WPM method ranked this solution as two, and the SAW method ranked it
four. Hence, the ERao-2 solution has achieved the least rank.
The ERao-2 solution has the best compromise among the six objectives. Thus, it
can be considered as the best solution among the compared solutions. The ERao-2
solution has better values in five objectives (power output, BSFC, CO, HC, and Sm )
compared to those of the NSGA-II solution. Similarly, the ERao-2 solution has better
values in four objectives (NOx , CO, HC, and Sm ) compared to those of the Rao-3
solution. It has better values in four objectives (power output, BSFC, CO, and NOx )
compared to those of the SAP-Rao and ERao-3 solutions. It has better values for the
objectives BSFC, CO, HC, and Sm , when compared to those of the AMTPG-Jaya
algorithm solution. Similarly, it has better values in three objectives (Power output,
BSFC, and CO) compared to those of the Jaya, Rao-1, and Rao-2 solutions. Also, it
has better values for the objectives power output, BSFC, and HC, when compared to
those of the ERao-1 solution. Hence, the ERao-2 solution can be considered as the
best solution.
The spacing values of the Pareto-fronts obtained by different algorithms in the
MOO scenario are presented in Table 8.40. The ERao-2 algorithm has achieved better
Table 8.37 Best solutions obtained by various algorithms in MOO scenario of the biodiesel engine with an EGR system case study
Algorithm ER (%) EL (%) ES (rpm) BP (%) P (kW) BSFC (g/kWh) CO (%) NOx (ppm) HC (ppm) Sm (1/m)
NSGA-II 6.6 40 2125 10.9 0.32 391.83 0.19 215.19 49.01 3.67
Jaya 0 30.9842 1800 10.893 1.3609 409.99 0.11727 218.026 22.02787 2.30211
Rao-1 0 30.5670 1800 10.864 1.3427 414.95 0.12068 214.639 22.40188 2.30200
Rao-2 0 30.0902 1800 11.064 1.3262 421.27 0.12475 210.554 22.77768 2.30385
8.3 Design Optimization of a Compression …

Rao-3 11.084 46.7273 1800 10.269 2.0385 266.49 0.08665 298.123 28.65417 3.70621
AMTPG-Jaya 25.712 46.0734 1878.2 13.441 2.0698 312.31 0.00040 245.005 32.13142 3.68890
SAP-Rao 0 41.6085 1800 15 1.8739 319.73 0.07211 288.515 16.01710 2.70602
ERao-1 30 25.0000 1896.4 14.468 1.1969 533.40 0.00004 116.133 32.75884 3.01143
ERao-2 19.9815 43.8303 1800 11.853 1.9924 292.97 0.00018 268.552 25.47298 3.31472
ERao-3 0 41.8954 1800 15 1.8854 317.44 0.07232 290.219 16.01113 2.72791
Source NSGA-II:—Jaliliantabar et al. (2019)
Result in boldface indicates better values
309
 310

Table 8.38 Ranks suggested by the MADM methods for different algorithm solutions presented in Table 8.37
Algorithm SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA AR CR FR
NSGA-II 10 10 10 10 10 10 10 10 10 10.0 10
Jaya 5 7 7 7 1 4.5 7 7 5.69 6.4 7
Rao-1 8 8 8 8 1 6 8 8 6.88 7.7 8
Rao-2 9 9 9 9 1 8 9 9 7.88 8.9 9
Rao-3 6 6 6 6 9 8 6 6 6.63 6.3 6
AMTPG-Jaya 7 3 2 2 7 4.5 2 4 3.94 3.5 3
SAP-Rao 3 4 4 4 5 2.5 3 3 3.56 3.4 2
ERao-1 1 1 3 3 8 8 4 5 4.13 3.6 4
ERao-2 4 2 1 1 1 1 1 1 1.5 1.6 1
ERao-3 2 5 5 5 6 2.5 5 2 4.06 3.8 5
AR—Average Rank; CR—Corrected Rank; FR—Final Rank
8 Optimization of the Selected Bio-Energy Systems
8.3 Design Optimization of a Compression … 311

Table 8.39 Spearman’s rank correlation coefficients between different pairs of MADM method’s
rankings presented in Table 8.38
Method SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA
SAW 1 0.794 0.685 0.685 0.006 0.543 0.661 0.770
WPM 0.794 1 0.964 0.964 −0.056 0.531 0.927 0.830
TOPSIS 0.685 0.964 1 1 0.025 0.661 0.988 0.891
MTOPSIS 0.685 0.964 1 1 0.025 0.661 0.988 0.891
VIKOR 0.006 −0.056 0.025 0.025 1 0.526 0.063 0.113
PROMETHEE 0.543 0.531 0.661 0.661 0.526 1 0.729 0.846
COPRAS 0.661 0.927 0.988 0.988 0.063 0.729 1 0.915
GRA 0.770 0.830 0.891 0.891 0.113 0.846 0.915 1

spacing values in terms of mean value, and the Rao-3 algorithm has achieved better
spacing values in terms of best, and the ERao-3 algorithm has the least standard
deviation. The mean of the spacing values achieved by the ERao-2 algorithm is
11.8, 2.8, 3.7, 12.9, 6, 0.7, 1.7 and 4.7% lesser when compared to that of the Jaya,
Rao-1, Rao-2, Rao-3, AMTPG-Jaya, SAP-Rao, ERao-1, and ERao-3 algorithms,
respectively.
The coverage metric values of the Pareto-fronts obtained by the Jaya algorithm,
Rao algorithms, and their modified versions are presented in Table 8.41. The ERao-1
algorithm has achieved better coverage values than the other algorithms compared.
Only two solutions of the ERao-1 algorithm are dominated by other algorithm Pareto-
fronts. From the coverage values, it can be observed that four solutions from the
Pareto-front of the SAP-Rao are dominated, and six solutions from the Pareto-fronts
of the Jaya, Rao-1, Rao-3, AMTPG-Jaya, and ERao-3 algorithms are dominated.
Similarly, seven solutions of the Pareto-front achieved by the ERao-2 algorithm are
dominated, and 13 solutions are dominated from the Pareto-front achieved by the
Rao-2 algorithm. The performances of the proposed modified algorithms are better
or competitive to that of the basic algorithms in terms of coverage and spacing values.
From the computational results of the biodiesel engine with an EGR system case
study, it can be observed that the solutions obtained by the proposed algorithms are
better than those reported in the literature. The solution of the ERao-2 algorithm has a
better compromise among the six objectives. By the solution of the ERao-1 solution,
the output power is increased by 522.6% than that of the NSGA-II algorithm solution.
In addition, the objectives BSFC, CO, HC, and Sm values are reduced by 25.2, 99, 48
and 9.7%, respectively, when compared to those of the NSGA-II solution. The ERao-
2 algorithm has achieved the best mean spacing value (0.13532), which is much better
than that achieved by other algorithms compared. The ERao-1 algorithm has achieved
better performance in terms of coverage values. In addition, the performances of the
modified versions in this case study are better or competitive to those of the basic
algorithms as well as the NSGA-II algorithm.
 312

Table 8.40 Spacing values of the Pareto-fronts obtained by the proposed algorithms in MOO for the biodiesel engine with an EGR system case study
Algorithm Jaya Rao-1 Rao-2 Rao-3 AMTPG-Jaya SAP-Rao ERao-1 ERao-2 ERao-3
Best 0.11925 0.11453 0.09774 0.08456 0.10115 0.09628 0.09932 0.10129 0.11349
Worst 0.19209 0.18317 0.20105 0.21666 0.16852 0.17344 0.18537 0.17749 0.17338
Mean 0.15338 0.13915 0.14050 0.15537 0.14391 0.13626 0.13773 0.13532 0.14202
SD 0.02864 0.02076 0.03055 0.03471 0.02102 0.02328 0.02893 0.02066 0.01989
Result in boldface indicates better values
8 Optimization of the Selected Bio-Energy Systems
8.4 Process Optimization of a Microalgae-Based … 313

Table 8.41 Coverage (%) values of the Pareto-fronts obtained by the proposed algorithms in MOO
for the biodiesel engine with an EGR system case study
Algorithm Jaya Rao-1 Rao-2 Rao-3 AMTPG-Jaya SAP-Rao ERao-1 ERao-2 ERao-3
Jaya – 13.33 26.67 6.67 13.33 10.00 6.67 6.67 13.33
Rao-1 6.67 – 20.00 10.00 3.33 13.33 3.33 3.33 6.67
Rao-2 3.33 3.33 – 6.67 3.33 0.00 0.00 0.00 0.00
Rao-3 10.00 16.67 36.67 – 6.67 3.33 3.33 3.33 10.00
AMTPG-Jaya 20.00 13.33 26.67 6.67 – 6.67 3.33 6.67 3.33
SAP-Rao 16.67 20.00 33.33 6.67 6.67 – 3.33 10.00 6.67
ERao-1 16.67 16.67 43.33 20.00 20.00 10.00 – 23.33 20.00
ERao-2 6.67 10.00 26.67 3.33 3.33 6.67 3.33 – 3.33
ERao-3 10.00 13.33 26.67 16.67 16.67 13.33 3.33 16.67 –

8.4 Process Optimization of a Microalgae-Based Biomass

Cultivation Process

The description of the selected microalgae-based biomass cultivation process is

presented in Sect. 2.3.4. In this case study, a multi-objective optimization case study
of fertilizer-assisted cultivation of the Nannochloropsis species biomass production
for biofuel feedstock is considered. This case study was presented by Banerjee et al.
(2016).
The objective functions of this case study include the maximization of biomass
production (BMP), eicosapentaenoic acid (EPA), and lipid productions (TLP). The
design variables of this case study are cultivation light intensity (X 1 in µmol/m2 /s),
temperature (X 2 in °C), and concentrations of NaCl (X 3 in M), NaHCO3 (X 4 in g/L)
and NPK-10:26:26 fertilizers (X 5 in g/L). The ranges of design variables are as
follows: 25 ≤ X 1 ≤ 125, 17 ≤ X 2 ≤ 29, 0.25 ≤ X 3 ≤ 1.25, 0.05 ≤ X 4 ≤ 1.85
and 0.15 ≤ X 5 ≤ 1.1. The regression models of the objectives of this case study are
presented in Sect. 2.3.4.
Banerjee et al. (2016) presented the optimum design parameters for this case
study using the NSGA-II algorithm. Now, the proposed algorithms, along with basic
Jaya and Rao algorithms, are used to find the optimal parameters through multi-
objective optimization. Furthermore, the Pareto-optimal solutions obtained by the
NSGA-II algorithm were not reported in the literature. However, the performances
of the proposed algorithms are compared with those achieved by the basic Jaya and
Rao algorithms. Furthermore, the solutions obtained by various algorithms are non-
dominated. Hence, the average rank method was used to identify the best solution
from the Pareto-front of the respective algorithm. Furthermore, the performances
of the proposed algorithms are compared with the performances of the basic Jaya
and Rao algorithms in terms of hypervolume, spacing andcoverage performance
metrics. The hypervolume and spacing performance metrics are calculated for ten
independent runs and the statistical results are presented in terms of best, worst, mean
and standard deviation (SD).
314 8 Optimization of the Selected Bio-Energy Systems

The Pareto-optimal solutions achieved by the Jaya algorithm, Rao algorithms

and their modified versions in this case study are presented in Tables 8.42, 8.43,
8.44, 8.45, 8.46, 8.47, 8.48, 8.49 and 8.50 Here, the biomass productivity is varied
between 0.731 and 0.7471 g/L, the lipid generation is varied from 45.11 to 47.12%,
and the EPA generation is varied from 21.3 to 23.51%. Similar to the previous case
studies, the best solutions from these Pareto-fronts are identified. Solution 5 of the

Table 8.42 Pareto-optimal solutions obtained by the Jaya algorithm for the case study of a
microalgae-based biomass cultivation process
Solution X1 X2 X3 X4 X5 BMP (g/L) TLP (%) EPA (%)
1 125.0000 21.2426 0.2904 1.3984 0.6496 0.7410 47.1162 21.9670
2 125.0000 21.1474 0.3466 0.9651 0.8641 0.7471 46.2141 21.5826
3 101.4912 17.0000 0.2500 1.1607 0.6562 0.7320 45.1821 23.5014
4 125.0000 22.1296 0.3653 0.9700 0.8582 0.7471 46.2149 21.3031
5 125.0000 17.0000 0.2865 1.1115 0.7952 0.7452 46.5051 22.9416
6 123.6404 17.0000 0.2654 1.1757 0.7053 0.7429 46.6797 23.1403
7 125.0000 22.6981 0.3541 1.1111 0.7400 0.7455 46.8668 21.4290
8 125.0000 18.3874 0.2871 1.4147 0.6093 0.7392 46.9514 22.6363
9 124.9035 18.0441 0.2500 1.3972 0.5942 0.7384 46.8511 22.8469
10 125.0000 21.4339 0.3403 1.1318 0.7927 0.7464 46.7376 21.7001
11 125.0000 20.6925 0.2887 1.3413 0.6516 0.7417 47.1111 22.1250
12 125.0000 20.5394 0.3199 1.1934 0.8588 0.7462 46.4346 21.8569
13 105.2143 17.0000 0.2500 1.1571 0.6512 0.7336 45.4627 23.4904
14 123.9731 19.4937 0.3351 0.8201 0.8394 0.7460 46.0840 22.0355
15 112.7859 17.0000 0.2500 1.1481 0.6548 0.7371 45.9946 23.4198
16 110.9045 17.0000 0.2500 1.1481 0.6310 0.7356 45.8319 23.4397
17 125.0000 17.0000 0.3250 0.9773 0.8776 0.7457 45.9523 22.6162
18 125.0000 20.3025 0.2645 1.2128 0.6599 0.7429 47.0498 22.3302
19 122.2723 19.9065 0.2677 1.1183 0.7326 0.7440 46.7749 22.4524
20 125.0000 21.2644 0.3610 1.2117 0.6966 0.7441 47.0236 21.7617
21 104.3744 17.0000 0.2500 1.0788 0.6564 0.7336 45.3130 23.4909
22 121.2119 17.0000 0.2500 1.3724 0.7201 0.7404 46.6147 23.2149
23 125.0000 20.3670 0.3263 1.3712 0.7941 0.7444 46.7952 21.9599
24 109.5265 17.0000 0.2500 1.0581 0.7827 0.7381 45.5618 23.3661
25 125.0000 19.6649 0.2905 1.1817 0.8169 0.7459 46.6561 22.2529
26 117.8767 17.0000 0.2500 1.2384 0.6885 0.7396 46.4017 23.3266
27 122.4247 18.6161 0.3032 1.0617 0.8615 0.7452 46.1623 22.4289
28 108.5403 17.0000 0.2500 1.1589 0.7177 0.7365 45.7237 23.4485
29 107.1290 17.0000 0.2500 1.1390 0.6514 0.7346 45.5864 23.4786
30 122.4454 17.9353 0.2773 1.1639 0.7455 0.7437 46.6609 22.8750
8.4 Process Optimization of a Microalgae-Based … 315

Table 8.43 Pareto-optimal solutions obtained by the AMTPG-Jaya algorithm for the case study of
a microalgae-based biomass cultivation process
Solution X1 X2 X3 X4 X5 BMP (g/L) TLP (%) EPA (%)
1 125.0000 20.9305 0.3461 0.9710 0.8606 0.7471 46.2436 21.6447
2 100.8574 17.0000 0.2500 1.1464 0.6540 0.7317 45.1126 23.5018
3 125.0000 21.6497 0.3065 1.3463 0.6535 0.7417 47.1185 21.8378
4 125.0000 22.7495 0.3339 1.0552 0.8188 0.7467 46.5341 21.3666
5 101.8510 17.0000 0.2500 1.1703 0.6298 0.7314 45.1770 23.4970
6 125.0000 20.4115 0.2616 1.1333 0.7823 0.7457 46.7872 22.2285
7 125.0000 18.4588 0.3364 1.1909 0.7254 0.7446 46.8865 22.4897
8 125.0000 18.9523 0.2500 1.2596 0.6403 0.7416 46.9716 22.6900
9 125.0000 21.7584 0.3514 0.9477 0.8351 0.7471 46.3570 21.4834
10 118.5807 17.0000 0.3122 1.2156 0.6259 0.7387 46.4317 23.0909
11 125.0000 21.3996 0.2723 1.1752 0.7002 0.7443 47.0370 22.0373
12 122.3157 17.3352 0.2746 1.1137 0.7762 0.7439 46.4697 22.9974
13 125.0000 22.3862 0.3158 1.2864 0.6755 0.7429 47.0906 21.6501
14 125.0000 19.3623 0.2550 1.2769 0.6481 0.7419 47.0188 22.5729
15 106.4583 17.0000 0.2517 1.1979 0.6561 0.7341 45.6046 23.4758
16 125.0000 18.1483 0.2500 1.2573 0.6094 0.7404 46.8531 22.8704
17 108.9231 17.0000 0.2500 1.1077 0.6429 0.7352 45.6681 23.4619
18 125.0000 20.3645 0.2911 1.0541 0.7973 0.7465 46.6750 22.1108
19 118.4000 17.0000 0.2521 1.0093 0.6940 0.7408 46.1989 23.2958
20 122.2519 17.0000 0.2500 1.1927 0.5830 0.7384 46.4557 23.2056
21 125.0000 17.5127 0.2803 1.1610 0.7740 0.7450 46.6681 22.8705
22 113.5406 17.0000 0.2500 1.0879 0.6365 0.7371 45.9504 23.4043
23 125.0000 17.1629 0.2736 1.2161 0.6848 0.7429 46.8135 23.0409
24 125.0000 20.3038 0.2875 1.1791 0.7476 0.7453 46.9330 22.2065
25 105.1946 17.0000 0.2500 1.1291 0.6427 0.7335 45.4177 23.4895
26 118.5759 17.0000 0.2500 1.0554 0.6631 0.7400 46.2575 23.3150
27 110.1718 17.0000 0.2500 1.1488 0.6639 0.7362 45.8274 23.4514
28 104.1054 17.0000 0.2500 1.0773 0.6453 0.7332 45.2738 23.4914
29 109.7695 17.0000 0.2500 1.1035 0.6581 0.7360 45.7436 23.4543
30 114.5881 17.0000 0.2500 1.0809 0.6642 0.7384 46.0452 23.3896

Jaya algorithm, Solution 26 of the AMTPG-Jaya algorithm, Solution 19 of the Rao-1

algorithm, Solution 6 of the Rao-2 algorithm, Solution 19 of the Rao-3 algorithm,
Solution 17 of the ERao-1 algorithm, Solution 13 of the ERao-2 algorithm, Solution
9 of the ERao-3 algorithm, and Solution 22 of the SAP-Rao algorithm are identified
as the best solutions from the respective algorithm’s Pareto-front. Now, these best
solutions are compared in Table 8.51.
316 8 Optimization of the Selected Bio-Energy Systems

Table 8.44 Pareto-optimal solutions obtained by the Rao-1 algorithm for the case study of a
microalgae-based biomass cultivation process
Solution X1 X2 X3 X4 X5 BMP (g/L) TLP (%) EPA (%)
1 125.0000 21.3965 0.3447 0.9615 0.8585 0.7471 46.2462 21.5433
2 101.3886 17.0000 0.2500 1.1460 0.6485 0.7318 45.1473 23.5013
3 101.5146 17.0000 0.2500 1.0443 0.6691 0.7327 45.0537 23.4922
4 125.0000 21.7300 0.2819 1.3317 0.6505 0.7417 47.1169 21.9068
5 125.0000 23.7469 0.3355 1.0469 0.7852 0.7462 46.6322 21.1890
6 125.0000 19.5936 0.2757 1.2321 0.6863 0.7435 47.0285 22.4510
7 125.0000 18.3735 0.2873 1.0538 0.7738 0.7457 46.6537 22.6301
8 125.0000 22.3511 0.3263 0.9864 0.8407 0.7470 46.3769 21.4306
9 125.0000 19.9073 0.2993 0.9427 0.7063 0.7451 46.7661 22.2696
10 125.0000 22.3531 0.3048 1.1725 0.6914 0.7442 47.0431 21.7111
11 104.4553 17.0000 0.2500 1.0903 0.6587 0.7337 45.3368 23.4915
12 123.0035 17.7472 0.3020 1.0385 0.7592 0.7446 46.5297 22.7932
13 125.0000 21.2037 0.2983 1.0498 0.6925 0.7448 46.9529 22.0000
14 106.2541 17.0000 0.2500 1.1160 0.6620 0.7345 45.5081 23.4837
15 122.0862 17.2891 0.2812 1.3828 0.6521 0.7395 46.7486 23.0240
16 117.0355 17.6248 0.2535 1.0969 0.7834 0.7416 46.1553 23.1015
17 125.0000 20.5471 0.2901 1.3631 0.6643 0.7419 47.1092 22.1461
18 110.5626 17.1974 0.2500 1.2364 0.6519 0.7356 45.9469 23.3933
19 123.0035 17.7207 0.2811 1.3101 0.7150 0.7423 46.8012 22.9055
20 119.0762 17.2408 0.2513 1.1893 0.7403 0.7415 46.4141 23.2144
21 115.9004 17.0000 0.2500 1.1715 0.6617 0.7385 46.2226 23.3715
22 108.7157 17.0387 0.2500 1.1121 0.6472 0.7353 45.6708 23.4551
23 125.0000 20.9354 0.3379 1.1936 0.8046 0.7461 46.7194 21.8043
24 114.9694 17.0489 0.2674 1.3713 0.6746 0.7371 46.3161 23.2786
25 125.0000 21.7902 0.2907 1.2692 0.6622 0.7427 47.1078 21.8814
26 112.4077 17.0000 0.2500 1.0611 0.6725 0.7377 45.8838 23.4177
27 123.0035 17.9714 0.2853 1.3207 0.6403 0.7404 46.8440 22.8395
28 110.4306 17.0000 0.2500 1.0534 0.6572 0.7364 45.7250 23.4421
29 108.6494 17.0000 0.2500 1.3592 0.6174 0.7326 45.7997 23.4232
30 116.6414 17.0000 0.2500 1.1144 0.7404 0.7407 46.1766 23.3225

In Table 8.51, the solutions with the best average rank from the Pareto-optimal
solutions of the various algorithms are presented. Figure 8.1 presents the Pareto-fronts
achieved by the AMTPG-Jaya, SAP-Rao, and elitist Rao algorithms, including the
optimal solutions reported for the Jaya, and Rao algorithms. The Jaya algorithm
solution has achieved higher biomass productivity, the SAP-Rao algorithm solution
8.4 Process Optimization of a Microalgae-Based … 317

Table 8.45 Pareto-optimal solutions obtained by the Rao-2 algorithm for the case study of a
microalgae-based biomass cultivation process
Solution X1 X2 X3 X4 X5 BMP (g/L) TLP (%) EPA (%)
1 100.9575 17.0000 0.2500 1.1470 0.6611 0.7319 45.1302 23.5016
2 125.0000 21.1658 0.2500 1.3561 0.6599 0.7414 47.0986 22.1371
3 125.0000 21.3381 0.2500 0.9529 0.8439 0.7463 46.3402 21.9128
4 125.0000 22.5894 0.2500 1.0018 0.8117 0.7461 46.5431 21.6920
5 99.9722 17.0000 0.2500 1.0560 0.6961 0.7325 44.9585 23.4852
6 117.5120 17.0000 0.2500 0.9924 0.8103 0.7419 45.8898 23.1973
7 125.0000 17.4875 0.2500 1.1197 0.7359 0.7443 46.6991 23.0205
8 125.0000 20.6525 0.2500 1.2662 0.6756 0.7428 47.0700 22.2835
9 125.0000 19.5495 0.2500 1.0000 0.7955 0.7459 46.5736 22.4390
10 125.0000 18.7056 0.2500 1.0121 0.7483 0.7452 46.6751 22.7033
11 125.0000 20.5581 0.2500 0.9452 0.8728 0.7462 46.1487 22.0303
12 125.0000 17.7822 0.2500 1.1885 0.7088 0.7436 46.8240 22.9679
13 105.9998 17.0000 0.2500 1.0690 0.7069 0.7355 45.4465 23.4671
14 125.0000 17.7965 0.2500 1.5457 0.6838 0.7391 46.8894 22.8596
15 110.3063 17.0000 0.2500 0.9452 0.6817 0.7370 45.5722 23.4151
16 124.3719 18.5339 0.2500 1.4333 0.6774 0.7405 46.9533 22.7538
17 125.0000 22.6996 0.2500 1.1838 0.6874 0.7435 47.0301 21.8120
18 112.3285 17.0000 0.2500 1.1401 0.7182 0.7383 45.9618 23.4069
19 124.6895 19.5487 0.2500 1.3191 0.6400 0.7411 47.0176 22.5414
20 123.6743 19.3461 0.2500 1.2427 0.7091 0.7431 46.9152 22.6208
21 115.3399 17.0000 0.2500 1.0645 0.7396 0.7402 46.0437 23.3396
22 102.1732 17.0000 0.2500 1.1860 0.6673 0.7324 45.2740 23.4991
23 101.2771 17.0000 0.2500 1.0668 0.7004 0.7332 45.0796 23.4851
24 120.0439 17.0000 0.2500 1.1604 0.6470 0.7399 46.4430 23.2899
25 107.3820 17.0000 0.2500 1.2320 0.6664 0.7345 45.7070 23.4718
26 103.8756 17.0000 0.2500 1.0130 0.6754 0.7340 45.2006 23.4801
27 107.5547 17.0000 0.2500 1.3472 0.6633 0.7335 45.7836 23.4473
28 107.5884 17.0000 0.2500 1.1456 0.7252 0.7363 45.6361 23.4510
29 112.0074 17.0000 0.2500 1.0836 0.7513 0.7388 45.8333 23.3794
30 116.3680 17.0000 0.2500 1.1667 0.7057 0.7397 46.2504 23.3525

has a higher lipid production percentage, and the ERao-3 solution has a higher EPA
generation percentage.
These solutions are non-dominated, and to identify the best solution among the
solutions of different algorithms, which has the best compromise among the objec-
tives, the MADM methods based average ranks are calculated and presented in
Table 8.52. The Spearman’s correlation coefficients for different pairs of rankings
318 8 Optimization of the Selected Bio-Energy Systems

Table 8.46 Pareto-optimal solutions obtained by the Rao-3 algorithm for the case study of a
microalgae-based biomass cultivation process
Solution X1 X2 X3 X4 X5 BMP (g/L) TLP (%) EPA (%)
1 125.0000 21.6905 0.3045 1.3420 0.6383 0.7413 47.1150 21.8360
2 125.0000 22.0396 0.3182 0.9025 0.8602 0.7470 46.1803 21.4638
3 103.0106 17.0000 0.2500 1.0242 0.6506 0.7329 45.1210 23.4878
4 102.6524 17.0000 0.2501 1.1259 0.6522 0.7326 45.2311 23.4990
5 125.0000 18.8661 0.2523 1.3691 0.7036 0.7423 46.9894 22.6657
6 125.0000 19.3395 0.3043 0.9607 0.7962 0.7464 46.5362 22.2890
7 125.0000 18.4287 0.2934 0.8792 0.8130 0.7460 46.2879 22.4858
8 125.0000 17.0000 0.2935 1.5403 0.7020 0.7398 46.8312 22.9011
9 105.4449 17.0000 0.2500 1.1974 0.6482 0.7334 45.5157 23.4871
10 125.0000 20.7472 0.3478 1.1550 0.6864 0.7443 47.0135 21.9386
11 125.0000 17.8788 0.2822 1.2620 0.6408 0.7417 46.9069 22.8334
12 106.6410 17.0000 0.2500 1.0147 0.7429 0.7364 45.3872 23.4264
13 123.9464 19.3048 0.2879 0.7791 0.7126 0.7440 46.3887 22.4068
14 125.0000 22.4777 0.3026 1.1030 0.7472 0.7457 46.8873 21.6475
15 109.4060 17.0785 0.2500 1.1384 0.7705 0.7377 45.6840 23.3725
16 118.0435 17.0000 0.2500 1.5760 0.6919 0.7357 46.4390 23.2000
17 125.0000 21.3115 0.3503 0.8689 0.8250 0.7469 46.3148 21.5851
18 112.8090 17.0000 0.2673 1.1828 0.7581 0.7391 45.9900 23.3062
19 123.8676 17.0000 0.2810 1.1183 0.6252 0.7411 46.6109 23.0823
20 122.0134 20.0628 0.3297 1.0675 0.7788 0.7451 46.6034 22.1496
21 112.4611 17.0000 0.2574 1.2345 0.6728 0.7370 46.0735 23.3923
22 123.3290 17.8696 0.2500 1.1353 0.6400 0.7414 46.6968 23.0034
23 114.3495 17.0000 0.2614 1.2214 0.6931 0.7384 46.1919 23.3479
24 107.1340 17.0000 0.2500 1.2454 0.6851 0.7347 45.7072 23.4688
25 125.0000 21.4680 0.2512 1.3048 0.6902 0.7427 47.0803 22.0712
26 114.8987 17.0012 0.2772 0.9941 0.7247 0.7402 45.9779 23.2480
27 125.0000 22.1379 0.2973 1.1090 0.7155 0.7451 46.9736 21.7744
28 109.3780 17.0000 0.2529 1.3061 0.6512 0.7345 45.8867 23.4313
29 112.6610 17.0000 0.2500 1.1346 0.6030 0.7355 45.8713 23.4054
30 125.0000 17.0000 0.2603 1.2146 0.6180 0.7407 46.7216 23.1188

for different algorithm’s solutions are shown in Table 8.53. The ranks given by the
TOPSIS and MTOPSIS methods and WPM and COPRAS methods for each solution
are identical. Furthermore, the Spearman’s correlation for all the pairs of decision-
making methods formed by the SAW, WPM, TOPSIS, MTOPSIS, and COPRAS
methods is greater than 0.5. However, the Spearman’s correlation values for the
VIKOR-TOPSIS, VIKOR—MTOPSIS, and VIKOR-GRA pairs are less than 0.5.
8.4 Process Optimization of a Microalgae-Based … 319

Table 8.47 Pareto-optimal solutions obtained by the ERao-1 algorithm for the case study of a
microalgae-based biomass cultivation process
Solution X1 X2 X3 X4 X5 BMP (g/L) TLP (%) EPA (%)
1 101.0886 17.0000 0.2500 1.1427 0.6609 0.7320 45.1358 23.5015
2 125.0000 21.6117 0.3026 1.3323 0.6486 0.7417 47.1185 21.8648
3 125.0000 21.8584 0.3390 0.9676 0.8721 0.7471 46.1630 21.4243
4 100.7535 17.0000 0.2500 1.0690 0.6613 0.7321 45.0177 23.4969
5 103.6297 17.0000 0.2500 1.2669 0.6208 0.7313 45.3838 23.4789
6 120.5450 17.0400 0.2500 1.2386 0.6670 0.7402 46.5551 23.2646
7 123.7574 18.2917 0.2879 1.0709 0.7452 0.7448 46.6815 22.7148
8 123.7711 17.5470 0.2664 1.4219 0.6242 0.7388 46.8230 22.9481
9 109.8401 17.0000 0.2500 1.0267 0.7326 0.7377 45.6455 23.4116
10 113.3839 17.0245 0.2500 1.1501 0.6640 0.7376 46.0478 23.4054
11 125.0000 21.5395 0.3339 1.0043 0.8367 0.7470 46.4278 21.6053
12 111.9448 17.0276 0.2500 1.1746 0.6514 0.7365 45.9646 23.4235
13 125.0000 17.9650 0.2500 1.1241 0.6490 0.7424 46.7913 22.9379
14 123.7574 18.6865 0.2902 1.0049 0.8489 0.7458 46.2428 22.4412
15 123.5544 20.8428 0.2748 1.0131 0.8328 0.7459 46.4100 22.0196
16 110.6783 17.0000 0.2500 1.1905 0.6346 0.7354 45.8649 23.4417
17 125.0000 17.4446 0.2613 1.1088 0.6946 0.7436 46.7426 23.0149
18 125.0000 22.3978 0.3058 1.1217 0.7313 0.7454 46.9423 21.6714
19 124.4933 19.4536 0.3118 1.0306 0.8536 0.7465 46.3169 22.1582
20 107.4926 17.0000 0.2500 1.2361 0.6553 0.7343 45.7079 23.4705
21 103.7171 17.0000 0.2500 1.1019 0.6422 0.7329 45.2701 23.4945
22 124.8521 20.5644 0.2609 1.2806 0.7142 0.7436 47.0336 22.2512
23 108.2576 17.0000 0.2500 1.2000 0.6812 0.7355 45.7513 23.4655
24 104.5102 17.0000 0.2500 1.1006 0.6422 0.7333 45.3300 23.4913
25 123.7710 17.0000 0.2500 1.3310 0.6698 0.7408 46.7662 23.1783
26 125.0000 21.0258 0.2500 1.3227 0.6405 0.7412 47.0886 22.1813
27 124.5728 18.7620 0.2883 1.3876 0.6310 0.7402 46.9914 22.5751
28 123.1681 18.2812 0.3253 0.9763 0.8137 0.7455 46.3222 22.4865
29 125.0000 22.1679 0.3024 1.2163 0.6975 0.7441 47.0588 21.7525
30 123.7574 18.9931 0.2641 1.0384 0.7248 0.7445 46.7260 22.6445

Similarly, the PROMETHEE method has the Spearman’s correlation values less than
0.5 with the SAW, WPM, TOPSIS, MTOPSIS, and COPRAS methods. The GRA
method has the Spearman’s correlation values less than 0.5 with the SAW, WPM,
TOPSIS, MTOPSIS, VIKOR, and COPRAS methods. Hence, the ranks suggested by
the VIKOR, PROMETHEE, and GRA methods cannot be considered for calculating
the average ranks. Now, the corrected average ranks are calculated, excluding the
320 8 Optimization of the Selected Bio-Energy Systems

Table 8.48 Pareto-optimal solutions obtained by the ERao-2 algorithm for the case study of a
microalgae-based biomass cultivation process
Solution X1 X2 X3 X4 X5 BMP (g/L) TLP (%) EPA (%)
1 99.8541 17.0000 0.2500 1.1676 0.6668 0.7314 45.0679 23.5004
2 124.9695 21.1798 0.2925 1.3500 0.6577 0.7418 47.1179 21.9947
3 125.0000 21.8736 0.3401 0.9688 0.8579 0.7471 46.2554 21.4513
4 104.2653 17.0000 0.2500 1.1582 0.6817 0.7339 45.4181 23.4915
5 124.9639 19.5125 0.2534 1.3921 0.6572 0.7409 47.0494 22.5078
6 102.8818 17.0000 0.2500 1.0630 0.6607 0.7331 45.1803 23.4937
7 107.8746 17.0000 0.2500 1.1495 0.6679 0.7353 45.6691 23.4725
8 124.9974 18.4808 0.2541 1.1409 0.7702 0.7451 46.7301 22.7333
9 105.8389 17.0000 0.2593 1.1679 0.7481 0.7358 45.5131 23.4107
10 116.3560 17.0000 0.2500 1.2342 0.6079 0.7367 46.2159 23.3434
11 112.6548 17.0000 0.2510 1.0348 0.6975 0.7384 45.8727 23.3990
12 125.0000 22.4003 0.3084 1.2076 0.7335 0.7449 46.9780 21.6524
13 118.5760 17.1384 0.2500 1.2851 0.7153 0.7401 46.4671 23.2612
14 125.0000 18.5914 0.3554 1.4037 0.7159 0.7425 46.9500 22.3493
15 124.5756 20.1091 0.3304 1.1487 0.7706 0.7458 46.8114 22.0886
16 124.5918 20.1243 0.3389 1.0686 0.8096 0.7465 46.5918 21.9917
17 124.8914 17.0000 0.2640 1.1845 0.7537 0.7441 46.6696 23.0776
18 125.0000 20.5879 0.2611 1.4345 0.6234 0.7395 47.0728 22.1987
19 105.0548 17.0000 0.2500 1.2009 0.6342 0.7328 45.4665 23.4864
20 125.0000 17.9122 0.2500 1.0790 0.8148 0.7454 46.4608 22.8158
21 125.0000 21.3154 0.3281 1.1786 0.7929 0.7461 46.7719 21.7686
22 123.8124 20.7197 0.2762 1.2090 0.7436 0.7444 46.9106 22.1779
23 124.9894 21.4731 0.2500 1.1219 0.6670 0.7435 47.0087 22.1114
24 113.1203 17.0000 0.2500 1.1514 0.7283 0.7388 46.0129 23.3897
25 107.2724 17.0370 0.2548 1.1295 0.6768 0.7354 45.6236 23.4494
26 118.8557 17.0000 0.2783 1.0955 0.6839 0.7409 46.3656 23.2104
27 123.6308 17.0000 0.3196 1.3684 0.7027 0.7416 46.7941 22.9161
28 125.0000 21.4281 0.2693 1.0915 0.8163 0.7463 46.6253 21.9042
29 114.0206 17.0000 0.2500 1.1708 0.6503 0.7374 46.0922 23.4014
30 121.0026 17.1954 0.2556 1.3597 0.6159 0.7379 46.6027 23.1575

ranks suggested by the VIKOR, PROMETHEE, and GRA methods, and presented
in Table 8.52 as the corrected ranks.
The SAP-Rao algorithm solution has the least average rank, which is 1.4. The
SAP-Rao solution has the best compromise among the objectives. Thus, it can be
considered as the best solution among the compared solutions. The ERao-2 algorithm
8.4 Process Optimization of a Microalgae-Based … 321

Table 8.49 Pareto-optimal solutions obtained by the ERao-3 algorithm for the case study of a
microalgae-based biomass cultivation process
Solution X1 X2 X3 X4 X5 BMP (g/L) TLP (%) EPA (%)
1 100.5422 17.0000 0.2500 1.1464 0.6593 0.7317 45.0935 23.5017
2 125.0000 21.4992 0.3484 0.9525 0.8585 0.7471 46.2312 21.5037
3 125.0000 21.0129 0.3014 1.3673 0.6577 0.7417 47.1177 21.9966
4 100.7928 17.0000 0.2500 1.1453 0.6483 0.7315 45.0978 23.5015
5 110.5065 17.0000 0.2500 1.0698 0.6888 0.7372 45.7721 23.4375
6 105.0944 17.0000 0.2545 1.1319 0.6760 0.7343 45.4585 23.4731
7 107.4323 17.0000 0.2500 1.0831 0.6975 0.7359 45.5716 23.4640
8 103.5104 17.0000 0.2500 1.1444 0.6445 0.7327 45.3079 23.4968
9 116.5563 17.0000 0.2515 1.2811 0.7046 0.7390 46.3437 23.3340
10 113.3456 17.0000 0.2500 1.0952 0.6866 0.7383 45.9927 23.4052
11 101.3553 17.0000 0.2500 1.1320 0.7092 0.7331 45.1627 23.4860
12 124.9361 22.4904 0.2999 1.1160 0.7281 0.7453 46.9427 21.6748
13 124.9433 20.3368 0.3118 1.3365 0.7036 0.7433 47.0676 22.1202
14 110.0204 17.0000 0.2500 1.0135 0.7322 0.7378 45.6402 23.4073
15 125.0000 18.9546 0.3040 1.2732 0.6432 0.7420 47.0138 22.5004
16 125.0000 17.0000 0.2887 0.8857 0.7230 0.7442 46.3833 22.9537
17 118.7036 17.0000 0.2500 1.2418 0.7949 0.7414 46.2480 23.2216
18 125.0000 20.2792 0.3097 0.9884 0.8715 0.7469 46.2117 21.9056
19 125.0000 17.0000 0.2684 1.0401 0.8520 0.7453 46.1656 22.8961
20 124.9057 22.1211 0.3478 1.2476 0.7321 0.7446 46.9666 21.5794
21 119.4545 17.0000 0.2830 1.0168 0.8108 0.7431 46.0612 23.0467
22 125.0000 20.6079 0.2684 1.1476 0.7680 0.7456 46.8545 22.1777
23 122.8455 19.3137 0.2500 1.2251 0.6843 0.7423 46.8850 22.6624
24 120.4315 17.0000 0.2500 1.2350 0.6606 0.7400 46.5377 23.2773
25 124.8803 18.7456 0.3020 0.9784 0.8129 0.7463 46.4468 22.4181
26 122.4654 18.1019 0.2779 1.3559 0.7848 0.7427 46.6378 22.7615
27 124.9495 19.6096 0.3064 1.0638 0.7701 0.7461 46.7541 22.2774
28 124.8273 19.1598 0.2986 0.9547 0.8051 0.7463 46.4746 22.3399
29 119.2542 17.0000 0.2613 1.0370 0.8003 0.7427 46.0960 23.1496
30 125.0000 18.8517 0.2577 1.2427 0.7789 0.7447 46.7888 22.6154

has achieved the next least average rank of 1.8. Furthermore, the ERao-1 and ERao-
3 algorithms have attained the same average rank of 3.6. However, the ERao-1 has
achieved better ranks by three methods (SAW, WPM, and COPRAS), and the ERao-
3 algorithm has achieved better ranks by two methods (TOPSIS and MTOPSIS).
Hence, the ERao-1 and ERao-3 algorithms are given ranks 3 and 4, respectively.
322 8 Optimization of the Selected Bio-Energy Systems

Table 8.50 Pareto-optimal solutions obtained by the SAP-Rao algorithm for the case study of a
microalgae-based biomass cultivation process
Solution X1 X2 X3 X4 X5 BMP (g/L) TLP (%) EPA (%)
1 125.0000 21.4132 0.3410 0.9599 0.8519 0.7471 46.2889 21.5671
2 125.0000 21.6982 0.2970 1.3766 0.6539 0.7414 47.1181 21.8468
3 99.8128 17.0000 0.2500 1.1397 0.6521 0.7312 45.0167 23.5012
4 100.2454 17.0000 0.2500 1.1734 0.6400 0.7309 45.0696 23.4998
5 125.0000 21.3542 0.2617 1.2456 0.7253 0.7442 47.0193 22.0563
6 112.2826 17.0000 0.2500 1.3431 0.7161 0.7368 46.0928 23.3855
7 111.0456 17.0000 0.2515 1.2092 0.7270 0.7375 45.9301 23.4105
8 125.0000 19.3442 0.2511 1.4296 0.6446 0.7401 47.0324 22.5381
9 125.0000 20.4652 0.3414 1.0417 0.8720 0.7469 46.2398 21.7596
10 104.4634 17.0000 0.2500 1.0954 0.6701 0.7340 45.3554 23.4907
11 123.7532 17.9644 0.2500 1.4901 0.6447 0.7385 46.8613 22.8797
12 125.0000 17.7572 0.2912 1.0573 0.7333 0.7449 46.7009 22.8059
13 116.5568 17.0000 0.2621 1.3677 0.7438 0.7390 46.3308 23.2519
14 106.7963 17.0000 0.2500 1.2368 0.6857 0.7347 45.6766 23.4720
15 120.8521 17.0000 0.2500 1.4818 0.6834 0.7381 46.6260 23.1932
16 125.0000 20.6454 0.2570 1.0827 0.6620 0.7436 46.9606 22.2819
17 125.0000 18.0298 0.2988 1.1255 0.7737 0.7455 46.6948 22.6792
18 125.0000 21.3728 0.3405 1.1213 0.8313 0.7467 46.5427 21.6454
19 125.0000 18.4498 0.2500 1.4425 0.6459 0.7398 46.9709 22.7520
20 125.0000 17.9062 0.2833 1.0539 0.8696 0.7459 46.1718 22.5878
21 125.0000 19.9202 0.3592 1.0741 0.8246 0.7467 46.5149 21.9303
22 125.0000 17.0000 0.2594 1.3109 0.6330 0.7405 46.8018 23.1113
23 125.0000 19.4712 0.2965 1.3944 0.6894 0.7422 47.0593 22.3636
24 108.3264 17.0000 0.2500 1.2407 0.6820 0.7353 45.7883 23.4608
25 125.0000 17.4676 0.2558 1.3759 0.7341 0.7424 46.8336 22.9746
26 108.3264 17.0000 0.2573 1.1870 0.7082 0.7362 45.7527 23.4296
27 122.5491 19.6445 0.2987 1.2075 0.6967 0.7431 46.9021 22.4206
28 125.0000 18.0456 0.3256 1.1110 0.6972 0.7442 46.8305 22.6427
29 118.1135 17.0000 0.2500 1.0320 0.6233 0.7387 46.1400 23.3151
30 101.6586 17.0000 0.2500 1.2087 0.6687 0.7321 45.2551 23.4980

The hypervolume and spacing values of the Pareto-fronts obtained by different

algorithms in the MOO scenario are presented in Table 8.54. The Rao-1 algorithm
has achieved better spacing values in terms of mean value, the SAP-Rao algorithm
has achieved better spacing values in terms of best, and the Rao-3 algorithm has
the least standard deviation. Similarly, the SAP-Rao algorithm has achieved better
hypervolume values in terms of mean and best values, and the ERao-3 algorithm has
8.4 Process Optimization of a Microalgae-Based … 323

Table 8.51 Best solutions obtained by various algorithms in MOO scenario of the case study of a
microalgae-based biomass cultivation process
Algorithm X1 X2 X3 X4 X5 BMP TLP (%) EPA (%)
(g/L)
Jaya 125 17 0.2865 1.1115 0.7952 0.7452 46.5051 22.9416
Rao-1 123.003 17.7207 0.2811 1.3101 0.7150 0.7423 46.8012 22.9055
Rao-2 117.512 17 0.2500 0.9924 0.8103 0.7419 45.8898 23.1973
Rao-3 123.868 17 0.2810 1.1183 0.6252 0.7411 46.6109 23.0823
AMTPG-Jaya 118.576 17 0.2500 1.0554 0.6631 0.7400 46.2575 23.3150
SAP-Rao 125 17 0.2594 1.3109 0.6330 0.7405 46.8018 23.1113
ERao-1 125 17.4446 0.2613 1.1088 0.6946 0.7436 46.7426 23.0149
ERao-2 118.576 17.1384 0.2500 1.2851 0.7153 0.7401 46.4671 23.2612
ERao-3 116.556 17 0.2515 1.2811 0.7046 0.7390 46.3437 23.3340
Result in boldface indicates better values

Fig. 8.1 Plot of Pareto-fronts of the proposed algorithms and solutions of the methods compared
in the case study of a microalgae-based biomass cultivation process

the least standard deviation. The mean of the hypervolume values achieved by the
SAP-Rao algorithm is 0.3, 0.08, 0.34, 0.6, and 0.4% higher when compared to that
of the Jaya, AMTPG-Jaya, Rao-2, and Rao-3 algorithms, respectively.
The coverage metric values of the Pareto-fronts obtained by the Jaya algorithm,
Rao algorithms, and their modified versions are presented in Table 8.55. The ERao-3
algorithm has achieved better coverage values than the other algorithms compared.
Only four solutions of the ERao-3 algorithm are dominated by other algorithm Pareto-
fronts. From the coverage values, it can be observed that five solutions from the
Pareto-fronts of the SAP-Rao and ERao-1 are dominated, and six solutions from the
 324

Table 8.52 Ranks suggested by the MADM methods for different algorithm solutions presented in Table 8.51
Algorithm SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA AR CR FR
Jaya 8 8 8 8 5 3.5 8 1 6.187 8 8
Rao-1 7 7 7 7 7 3.5 7 4 6.187 7 7
Rao-2 9 9 9 9 9 7.5 9 9 8.812 9 9
Rao-3 6 6 6 6 2 5 6 8 5.625 6 6
AMTPG-Jaya 5 5 4 4 6 9 5 6 5.5 4.6 5
SAP-Rao 1 1 2 2 3 1.5 1 3 1.812 1.4 1
ERao-1 2 3 5 5 1 1.5 3 2 2.812 3.6 3
ERao-2 3 2 1 1 4 6 2 7 3.25 1.8 2
ERao-3 4 4 3 3 8 7.5 4 5 4.812 3.6 4
AR—Average Rank; CR—Corrected Rank; FR—Final Rank
8 Optimization of the Selected Bio-Energy Systems
Table 8.53 Spearman’s rank correlation coefficients between different pairs of MADM method’s rankings presented in Table 8.52
Method SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA
SAW 1 0.9833 0.8667 0.8667 0.6000 0.3798 0.9833 0.3000
WPM 0.9833 1 0.9333 0.9333 0.5500 0.3038 1 0.2167
TOPSIS 0.8667 0.9333 1 1 0.3333 0 0.9333 0.0333
MTOPSIS 0.8667 0.9333 1 1 0.3333 0 0.9333 0.0333
8.4 Process Optimization of a Microalgae-Based …

VIKOR 0.6000 0.5500 0.3333 0.3333 1 0.6498 0.5500 0.3333

PROMETHEE 0.3798 0.3038 0 0 0.6498 1 0.3038 0.6920
COPRAS 0.9833 1 0.9333 0.9333 0.5500 0.3038 1 0.2167
GRA 0.3000 0.2167 0.0333 0.0333 0.3333 0.6920 0.2167 1
325
 326

Table 8.54 Hypervolume and spacing values of the Pareto-fronts obtained by the proposed algorithms in MOO of a microalgae-based biomass cultivation
process
Algorithm Jaya Rao1 Rao2 Rao3 AMTPG-Jaya SAP-Rao ERao-1 ERao-2 ERao-3
Spacing B 0.0557 0.0565 0.0589 0.0655 0.0577 0.0549 0.0554 0.0560 0.0612
W 0.1022 0.0968 0.1176 0.1019 0.1078 0.0977 0.1042 0.0982 0.1106
M 0.0750 0.0728 0.0788 0.0806 0.0747 0.0750 0.0755 0.0811 0.0832
SD 0.0146 0.0157 0.0215 0.0096 0.0163 0.0132 0.0140 0.0136 0.0153
Hyper volume B 687.29 687.38 687.23 687.45 687.09 687.67 687.23 687.10 687.29
W 669.85 669.42 669.11 670.37 685.23 686.58 686.35 685.57 686.75
M 685.25 684.82 683.19 684.58 686.54 687.11 686.91 686.77 686.93
SD 5.41 5.46 6.99 5.06 0.71 0.33 0.24 0.45 0.15
Result in boldface indicates better values
8 Optimization of the Selected Bio-Energy Systems
8.4 Process Optimization of a Microalgae-Based … 327

Table 8.55 Coverage (%) values of the Pareto-fronts obtained by the proposed algorithms in MOO
of a microalgae-based biomass cultivation process
Algorithm Jaya Rao-1 Rao-2 Rao-3 AMTPG-Jaya SAP-Rao ERao-1 ERao-2 ERao-3
Jaya – 20 13.3 30 13.3 13.3 6.7 10 10
Rao-1 20 – 16.7 33.3 3.3 13.3 13.3 20 10
Rao-2 16.7 10 – 30 13.3 16.7 10 16.7 10
Rao-3 16.7 3.3 13.3 – 6.7 0 13.3 3.3 6.7
AMTPG-Jaya 30 23.3 26.7 36.7 – 6.7 10 20 13.3
SAP-Rao 20 10 16.7 23.3 6.7 – 13.3 10 6.7
ERao-1 6.7 20 13.3 33.3 13.3 13.3 – 20 10
ERao-2 26.7 10 13.3 23.3 20 3.3 16.7 – 13.3
ERao-3 30 13.3 13.3 23.3 10 16.7 6.7 6.7 –

Pareto-fronts of the AMTPG-Jaya, and ERao-2 algorithms are dominated. Similarly,

seven solutions of the Pareto-front achieved by the Rao-1 algorithm are dominated;
eight solutions are dominated from the Pareto-front achieved by the Rao-2 algorithm;
nine solutions are dominated from the Pareto-front achieved by the Jaya algorithm;
11 solutions are dominated from the Pareto-front achieved by the Rao-3 algorithm.
The performances of the proposed modified algorithms are better or competitive to
that of the basic algorithms in terms of the coverage and spacing values.
From the computational results of the case study of a microalgae-based biomass
cultivation process, it can be observed that the SAP-Rao algorithm has achieved
the best solution. The Rao-1 algorithm has achieved the best mean spacing value
(0.0728). The SAP-Rao algorithm has achieved the best mean hypervolume value
(687.11). The ERao-3 algorithm has achieved better performance in terms of
coverage values. In addition, the performances of the improved versions in this
case study are better or competitive to those of the basic algorithms. The next
chapter presents the application of the Jaya algorithm and Rao algorithms along
with their modified versions to the hydropower generation system and geothermal
energy system case studies.

References

Banerjee, A., Guria, C., & Maiti, S. K. (2016). Fertilizer assisted optimal cultivation of microalgae
using response surface method and genetic algorithm for biofuel feedstock. Energy, 115, 1272–
1290. https://doi.org/10.1016/j.energy.2016.09.066
Dhingra, S., Bhushan, G., & Dubey, K. K. (2014). Multi-objective optimization of combustion,
performance and emission parameters in a jatropha biodiesel engine using non-dominated sorting
genetic algorithm-II. Frontiers of Mechanical Engineering, 9(1), 81–94. https://doi.org/10.1007/
s11465-014-0287-9
Jaliliantabar, F., Ghobadian, B., Najafi, G., Mamat, R., & Carlucci, A. P. (2019). Multi-objective
NSGA-II optimization of a compression ignition engine parameters using biodiesel fuel and
exhaust gas recirculation. Energy, 187, 115970. https://doi.org/10.1016/j.energy.2019.115970
328 8 Optimization of the Selected Bio-Energy Systems

Shirneshan, A., Samani, B. H., & Ghobadian, B. (2016). Optimization of biodiesel percentage
in fuel mixture and engine operating conditions for diesel engine performance and emission
characteristics by artificial bees colony algorithm. Fuel, 184, 518–526. https://doi.org/10.1016/j.
fuel.2016.06.117
Chapter 9
Optimization of Hydroenergy
and Geothermal Energy Systems

Abstract This chapter presents the applications of optimization algorithms to the

problems of hydro- and geothermal energy systems. A single-objective optimization
case study of the Nigerian Jebba hydropower plant performance and a multi-objective
optimization case study of a ground source heat pump-radiant ceiling air conditioning
system are considered for optimization to see if there can be any improvement in the
performances of the selected systems. The optimization is carried out using the Jaya
algorithm, Rao algorithms, and their modified versions. Furthermore, to identify the
best solutions from the Pareto-fronts, the average rank method described in Chap. 5 is
used. Computational results revealed that the performances of the modified Jaya and
Rao algorithms are superior to those of the other algorithms. Also, the performances
of the selected systems are improved by the solutions of the proposed algorithms.

9.1 Optimization of a Hydropower Generation System

Optimization of the Nigerian Jebba hydropower plant performance characteristics

to enhance its electricity generation case study is considered to demonstrate the
application of advanced optimization algorithms. This case study was presented by
Onokwai et. al. (2020) and a brief description of this case study is presented in
Sect. 2.4.1. This is a single-objective optimization case study, and the objective is to
improve the estimated power generation (EPw ) of the Jebba hydropower plant which
is given in Eq. 2.65. The decision variables considered are the discharge (D), pressure
drop between the head and tail-water (Pd ), the stator temperature (Tst ), the water
pressure (Pwater ) for cooling the generator, and the oil pressure (Poil ) for cooling and
lubrication of the bearings. The lower and upper boundaries of the decision variables
are as follows:

7139.25m3 /s ≤ D ≤ 8783.3667m3 /s (9.1)

26.1257m ≤ Pd ≤ 28.5867m (9.2)

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 329
V. R. Ravipudi et al., Design Optimization of Renewable Energy Systems Using Advanced
Optimization Algorithms, Green Energy and Technology,
https://doi.org/10.1007/978-3-030-95589-2_9
330 9 Optimization of Hydroenergy and Geothermal Energy Systems

53◦ C ≤ Tst ≤ 58◦ C (9.3)

29N/m2 ≤ Pwater ≤ 33N/m2 (9.4)

31N/m2 ≤ Poil ≤ 37N/m2 (9.5)

Onokwai et al. (2020) had reported an optimal solution using the response surface
methodology (RSM) for this case study. Now, the Jaya algorithm, Rao algorithms,
and their improved versions are employed to find the optimal solution for this case
study. The proposed algorithms are executed using only 10,000 function evaluations
as the termination criterion. In all the computational experiments, the population size
is maintained as 25 for all the algorithms. The elite size for the elitist Rao algorithms
is taken as 20%. Table 9.1 presents the single-objective optimization results of this
case study. The Jaya algorithm, Rao algorithms, and their improved versions have
obtained identical solution in this case study. Hence, only one solution is presented
here. Also, the solutions obtained by the proposed algorithms are superior to those
obtained by the RSM. The maximum estimated power generation obtained by the
proposed algorithms are 3166.874 MW. The proposed algorithms solutions have
resulted 18.5% increment in the total estimated power generation.
Here, it can be observed that the values achieved by the proposed algorithms
for the decision variables discharge, stator temperature, water, and oil pressure are
at the boundaries of the variables. To investigate why this has happened, decision
variables versus estimated power generation plots are drawn and presented in Fig. 9.1.
From the plots, it can be observed that the increase in discharge, water pressure for
cooling the generator, and stator temperature will result in increase of estimated
power generation. Similarly, as the pressure drop and oil pressure for lubrication
increases, the net power generation is decreased.
Furthermore, the larger the discharge, the more energy will be made available for
conversion to electric power as shown in Fig. 9.1. Also, from the hydrological data
obtained from the Jebba hydropower plant, lower turbine discharge values rendered
the turbine less efficient such that at very low reservoir levels. This is as a result
of the fact that Jebba hydropower plant operates via fixed turbine blades which
cannot be adjusted to run at low reservoir level (Onokwai et al., 2020). In addition,
thermal instability of generator windings occurs when the temperature fluctuated.

Table 9.1 Optimum solutions obtained using the RSM and proposed algorithms in the optimization
of the Jebba hydropower plant system case study
     
Method D m3 /s Pd (m) Tst (◦ C) Pwater N/m2 Poil N/m2 E Pw (M W )
RSM 8783.37 28.59 58 31 31 2671.4
Proposed 8783.367 26.2361 58 33 31 3166.874
algorithms
Result in boldface indicates a better performing algorithm
9.1 Optimization of a Hydropower Generation System 331

Fig. 9.1 Plot of decision variables versus estimated power generation

Maintaining the windings temperature at 58 °C or lower would enable the generator

run at its maximum capacity without risk of insulation failure. The hotter the windings
temperature, the less load the generator would sustain (Onokwai et al., 2020). The
following section presents the application of the Jaya and Rao algorithms along with
their modified versions to a ground source heat pump system case study.

9.2 Optimization of a Ground Source Heat Pump-Radiant

Ceiling Air Conditioning System

The description of the ground source heat pump-radiant ceiling (GSHP-RC) air
conditioning system is presented in Sect. 2.4.2. The detailed system description and
technical specifications of the GSHP-RC system considered in this case study were
presented by Xie et al. (2020). This case study consists of three objectives, which are
seasonal performance factor (SPF), thermal comfort predicted mean value (PMV),
and the operating cost (OC). In the optimization of this system, SPF is a maximization
function and the absolute value of PMV and operating costs is minimization func-
tions. The decision variables of these functions are the water supply temperature of
radiant ceiling (a), the indoor set temperature (b), and the water supply temperature
of the heat pump (c). The lower and upper boundaries of these variables are 16 < a
< 20, 24 < b < 28, and 7 < c < 11.
332 9 Optimization of Hydroenergy and Geothermal Energy Systems

Xie et al. (2020) optimized this multi-objective optimization case study using the
NSGA-II algorithm and reported two Pareto-optimal solutions. These two solutions
were selected from a Pareto-front based on the proximity to a reference solution using
utility method for two different weight combinations of the objectives. The number
of function evaluations taken by the NSGA-II was not specified. However, the Jaya
algorithm, Rao algorithms, and their modified versions are tested by taking 10,000
function evaluations as the termination criterion. The best Pareto-optimal solutions
are reported using the average rank based on multiple decision-making methods. In
all the computational experiments, the population size is maintained as 20 for all the
algorithms. The elite size for the elitist Rao algorithms is taken as 20%.
The Pareto-optimal solutions achieved by the Jaya algorithm, Rao algorithms, and
their improved versions in this case study are presented in Tables 9.2, 9.3, 9.4, 9.5, 9.6,
9.7, 9.8, 9.9 and 9.10 Here, an observation can be made that the seasonal performance
factor is varied between 3.65 and 3.75, the thermal comfort predicted mean value
is varied between 1e-5 and 0.75, and the operating cost is varied from 2200 to
2500. The best solutions from the Pareto-fronts obtained by the proposed algorithms
are identified based on the average rank method. Solution 4 of the Jaya algorithm,

Table 9.2 Pareto-optimal solutions obtained by the Jaya algorithm for the GSHP-RC system case
study
Solution a b c SPF PMV OC
1 16.1691 25.9539 9.337048 3.718262 3.57E-06 24,945.33
2 16 28 9.746768 3.762294 0.71256 23,098.13
3 16 25.94701 9.194334 3.719283 0.007977 24,963.48
4 16.56782 25.9273 11 3.695793 0.00543 23,835.93
5 16.24631 28 11 3.740916 0.713688 22,087.41
6 16.38885 28 10.77912 3.744283 0.714341 22,324.12
7 16 27.67985 9.843536 3.755992 0.591089 23,197.08
8 16.31228 27.9283 10.47108 3.75012 0.686818 22,623.24
9 16 26.80147 9.260826 3.736557 0.275096 24,173.19
10 16.00461 27.63492 10.70988 3.744806 0.574356 22,513.75
11 16 27.39271 9.285821 3.748246 0.485004 23,741.47
12 16.48853 26.21918 11 3.702682 0.094432 23,500.48
13 16.40366 25.74062 10.095 3.710895 0.057692 24,748.86
14 16.04723 26.39836 9.646177 3.730488 0.139874 24,274.95
15 16 26.98305 10.25315 3.73975 0.338343 23,332.9
16 16.09935 26.65432 11 3.717215 0.227029 22,952.95
17 16.529 27.35783 11 3.724056 0.479647 22,550.85
18 16.40899 27.12582 11 3.721459 0.395881 22,684.19
19 16.34138 26.4258 10.61972 3.717977 0.157036 23,597.33
20 16.55944 26.76018 9.849759 3.729401 0.273479 24,010.42
9.2 Optimization of a Ground Source Heat Pump … 333

Table 9.3 Pareto-optimal solutions obtained by the Rao-1 algorithm for the GSHP-RC system case
study
Solution a b c SPF PMV OC
1 16.04694 25.96695 10.01825 3.721415 4.62E-05 24,406.63
2 16 25.97329 9.496396 3.722447 0.000372 24,750.09
3 16 28 9.743059 3.762294 0.71256 23,100.75
4 16.1572 25.96175 11 3.702918 0.002064 23,607.25
5 16.12582 28 10.99627 3.742915 0.713136 22,069.82
6 16.17336 28 10.34625 3.755532 0.713354 22,665.24
7 16.08983 27.25775 11 3.729029 0.437454 22,496.87
8 16 26.86286 9.586978 3.740045 0.296358 23,917.96
9 16.05193 27.77799 9.699609 3.757165 0.628371 23,255.22
10 16.0604 27.67006 10.23338 3.752299 0.587993 22,915.4
11 16.02723 26.20045 10.94375 3.711107 0.074317 23,376.86
12 16.10831 26.44729 10.70825 3.720252 0.157783 23,407.14
13 16.12806 27.54888 10.6124 3.743183 0.54379 22,679.87
14 16.16195 26.59041 9.99303 3.731804 0.206993 23,901.2
15 16.00082 26.66473 9.864731 3.736305 0.228203 23,878.66
16 16.2986 26.75558 11 3.716039 0.266037 22,939.54
17 16.03245 27.33034 10.16909 3.746825 0.462784 23,169.09
18 16.1184 26.83472 9.969582 3.737336 0.289137 23,704.57
19 16.07005 27.72215 10.65147 3.746676 0.607547 22,530.4
20 16.0433 27.49273 10.28793 3.748514 0.522165 22,972.61

Table 9.4 Pareto-optimal solutions obtained by the Rao-2 algorithm for the GSHP-RC system case
study
Solution a b c SPF PMV OC
1 16 28 9.742384 3.762294 0.71256 23,101.22
2 16 25.97139 10.43529 3.717768 0.000232 24,044.45
3 16 25.97362 11 3.70562 0.000476 23,526.32
4 16 27.94971 11 3.743818 0.693258 22,070.29
5 16 28 9.939083 3.761866 0.71256 22,958.33
6 16 25.96738 9.692646 3.722976 0.00151 24,623.5
7 16 26.1877 10.90721 3.712246 0.069336 23,412.64
8 16 27.65644 10.45286 3.750067 0.582338 22,727.75
9 16 27.5293 10.74163 3.742145 0.53513 22,545.8
10 16 26.32809 9.428732 3.728885 0.115307 24,463.06
(continued)
334 9 Optimization of Hydroenergy and Geothermal Energy Systems

Table 9.4 (continued)

Solution a b c SPF PMV OC
11 16 27.76579 10.86784 3.743749 0.623365 22,292.95
12 16 26.48194 10.78965 3.720812 0.166433 23,262.13
13 16 27.39644 10.2889 3.747323 0.486364 23,021.67
14 16 27.08576 10.60852 3.73632 0.374593 22,956.1
15 16 27.71986 10.19788 3.754584 0.606085 22,903.68
16 16 27.22655 10.60736 3.739064 0.424851 22,859.97
17 16 26.65326 10.17318 3.734208 0.224277 23,651.8
18 16 26.92921 10.50032 3.735239 0.319476 23,164.81
19 16 26.7355 9.655706 3.737768 0.252388 23,970.08
20 16 26.54602 10.07694 3.732951 0.187957 23,815.94

Table 9.5 Pareto-optimal solutions obtained by the Rao-3 algorithm for the GSHP-RC system case
study
Solution a b c SPF PMV OC
1 16.47459 25.91993 11 3.697117 8.03E-05 23,798.19
2 16.00771 25.9718 9.256978 3.720357 0.000161 24,905.23
3 16 28 9.750632 3.762293 0.71256 23,095.4
4 16.21532 28 10.39278 3.754226 0.713546 22,633.62
5 16.0022 28 11 3.744755 0.71257 22,045.85
6 16 27.91716 9.363954 3.759106 0.680806 23,394.16
7 16.0202 26.76323 9.572211 3.737748 0.262371 24,011.46
8 16.16614 26.70101 11 3.717067 0.244459 22,937.87
9 16.00432 27.11443 9.823609 3.745035 0.38485 23,570.24
10 16.33389 27.64992 11 3.732771 0.583127 22,304.72
11 16 27.38294 9.857441 3.75022 0.481441 23,365.09
12 16.30292 27.67971 9.762669 3.751333 0.59383 23,320.72
13 16.003 27.4504 10.02407 3.750745 0.506137 23,198.84
14 16.00398 26.92806 9.704893 3.741495 0.319155 23,789.27
15 16.00888 26.48933 9.395937 3.731623 0.169143 24,346.81
16 16.51486 25.76827 10.48706 3.70492 0.045152 24,452.34
17 16.06765 26.62883 10.00623 3.733955 0.217622 23,825.32
18 16.66508 25.89608 10.17595 3.709084 0.000771 24,648.12
19 16.13211 27.39453 11 3.731008 0.48743 22,417.02
20 16.03355 26.90615 9.608356 3.740422 0.312108 23,881.15
9.2 Optimization of a Ground Source Heat Pump … 335

Table 9.6 Pareto-optimal solutions obtained by the AMTPG-Jaya algorithm for the GSHP-RC
system case study
Solution a b c SPF PMV OC
1 16.05922 25.96562 11 3.704534 5.02E-05 23,559.94
2 16.03247 25.97029 9.584395 3.722272 0.00052 24,708.77
3 16 28 9.743278 3.762294 0.71256 23,100.59
4 16.00539 28 11 3.744705 0.712585 22,046.36
5 16.01162 28 10.75841 3.750687 0.712613 22,276.3
6 16.00659 27.1389 11 3.728041 0.393598 22,555.87
7 16.0174 27.85123 11 3.741641 0.655812 22,123.73
8 16.0054 27.71231 9.744761 3.756648 0.603297 23,250.75
9 16.06314 25.88255 10.61108 3.712019 0.026135 24,006.94
10 16.00972 26.06049 9.76879 3.724643 0.028555 24,484.16
11 16.03848 27.31381 10.42345 3.743292 0.456885 22,971.94
12 16 27.49515 9.592036 3.752284 0.522539 23,481.56
13 16.0341 26.5311 9.578951 3.733067 0.183815 24,201.24
14 16.03099 26.65468 9.611074 3.73561 0.225505 24,075.89
15 16.01505 27.63702 9.755525 3.755039 0.575239 23,287.78
16 16.01168 26.29423 10.71704 3.718625 0.104503 23,497.79
17 16.00976 27.24628 9.657993 3.747492 0.432099 23,598.15
18 16.01796 26.78955 9.938515 3.738188 0.27138 23,730.55
19 16.03975 27.02236 10.99397 3.725434 0.35292 22,654.08
20 16.05525 26.4448 9.926435 3.730988 0.155517 24,036.94

Table 9.7 Pareto-optimal solutions obtained by the SAP-Rao algorithm for the GSHP-RC system
case study
Solution a b c SPF PMV OC
1 16 28 9.743338 3.762294 0.71256 23,100.55
2 16.4169 25.92638 11 3.698149 8E-05 23,763.66
3 16 25.99068 9.078556 3.718573 0.005908 24,987.65
4 16.00436 28 11 3.744721 0.71258 22,046.19
5 16.1388 28 10.24448 3.757321 0.713196 22,743.03
6 16.0043 27.82129 10.57186 3.751158 0.64437 22,534.4
7 16.26081 26.32153 11 3.708243 0.120683 23,305.49
8 16.26984 26.02953 11 3.702456 0.027255 23,590.21
9 16 26.83961 9.46639 3.739019 0.28829 24,015.28
10 16.003 27.37273 11 3.732618 0.477765 22,399.04
(continued)
336 9 Optimization of Hydroenergy and Geothermal Energy Systems

Table 9.7 (continued)

Solution a b c SPF PMV OC
11 16.0023 26.21195 9.534165 3.727214 0.0773 24,500.35
12 16.24963 26.61405 10.53811 3.72457 0.217108 23,467.77
13 16 26.1147 9.618033 3.725679 0.045686 24,533.56
14 16 26.56128 9.641764 3.734371 0.193102 24,121.19
15 16.38782 27.69624 10.15899 3.748402 0.600709 23,032.4
16 16.18683 26.42839 10.55398 3.721687 0.153659 23,592.02
17 16.00294 26.4988 10.15166 3.731376 0.172158 23,798.14
18 16.00408 27.6691 11 3.738329 0.587105 22,220.65
19 16 26.23018 10.92118 3.712705 0.083177 23,360.3
20 16 27.49377 11 3.735004 0.52203 22,322.87

Table 9.8 Pareto-optimal solutions obtained by the ERao-1 algorithm for the GSHP-RC system
case study
Solution a b c SPF PMV OC
1 16.30606 25.93907 11 3.700138 4.66E-05 23,698.24
2 16 28 9.743159 3.762294 0.71256 23,100.68
3 16.03424 25.96858 10.58699 3.714594 3.69E-05 23,930.14
4 16 27.88057 11 3.742481 0.66685 22,105.38
5 16.00457 28 9.800415 3.762185 0.712581 23,060.65
6 16.04393 25.95323 9.52005 3.721491 0.0045 24,773.37
7 16.03211 26.45176 9.404069 3.730593 0.157211 24,382.42
8 16.00562 27.63591 9.658408 3.755089 0.574733 23,354.24
9 16 26.63605 9.80372 3.735887 0.218423 23,945.92
10 16.01316 27.78872 10.63283 3.749229 0.63212 22,499.29
11 16.00737 26.9612 10.49741 3.73579 0.330818 23,145.84
12 16.09543 26.08195 11 3.706213 0.038223 23,462.42
13 16 27.53361 10.62601 3.744638 0.536721 22,647.98
14 16 27.87609 10.67989 3.750172 0.665147 22,408.38
15 16 26.22183 10.1042 3.726474 0.080451 24,078.67
16 16.02753 27.43658 10.60461 3.742743 0.501388 22,732.83
17 16.02981 26.35499 10.48652 3.7239 0.125037 23,655.46
18 16.04679 26.85149 9.516769 3.738793 0.293404 23,988.1
19 16.01395 27.11582 11 3.727479 0.385512 22,574.03
20 16.03057 27.18223 9.350473 3.744304 0.409465 23,848.6
9.2 Optimization of a Ground Source Heat Pump … 337

Table 9.9 Pareto-optimal solutions obtained by the ERao-2 algorithm for the GSHP-RC system
case study
Solution a b c SPF PMV OC
1 16 25.9721 10.59807 3.714992 8.06E-06 23,902.01
2 16 25.97106 9.194449 3.719749 0.000338 24,939.89
3 16 28 9.738953 3.762294 0.71256 23,103.64
4 16.06596 25.96574 11 3.70443 0.000217 23,562.78
5 16.04473 28 10.46249 3.755852 0.712765 22,545.13
6 16.03005 28 11 3.744317 0.712698 22,050.3
7 16.04612 26.22184 9.747491 3.727196 0.081855 24,362.88
8 16 26.75796 9.558205 3.737909 0.260104 24,018.37
9 16 27.1879 9.664041 3.746528 0.410992 23,630.89
10 16.02597 26.93439 11 3.723783 0.321806 22,708.94
11 16 26.42954 10.86852 3.717902 0.148933 23,233.19
12 16.06419 27.89502 11 3.741751 0.672747 22,109.23
13 16.16997 27.37748 10.92519 3.732104 0.481764 22,510.1
14 16.02844 26.09683 11 3.707554 0.040837 23,419.93
15 16.05495 26.22524 10.92511 3.711642 0.083229 23,383.06
16 16.00725 26.34244 9.751825 3.730138 0.12025 24,235.5
17 16 27.5705 11 3.736488 0.55037 22,276.86
18 16 26.14559 10.02962 3.725536 0.055673 24,206.93
19 16 27.81759 9.614432 3.758586 0.642939 23,282.64
20 16 27.52896 10.19998 3.750872 0.535005 23,012.24

Solution 4 of the Rao-1 algorithm, Solution 3 of the Rao-2 algorithm, Solution 1 of

the Rao-3 algorithm, Solution 1 of the AMTPG-Jaya algorithm, Solution 2 of the
SAP-Rao algorithms, Solution 1 of the ERao-1 algorithm, Solution 4 of the ERao-2,
and Solution 2 of the ERao-3 algorithm are identified as the best solutions from the
respective algorithm Pareto-front. Now, these best solutions are compared with those
of the NSGA-II algorithm in Table 9.11.
In Table 9.11, the solutions with the best average rank from the Pareto-optimal
solutions of the proposed algorithms are compared with those reported by the NSGA-
II algorithm. Figure 9.2 presents the Pareto-fronts achieved by the AMTPG-Jaya,
SAP-Rao, and elitist Rao algorithms, including the optimal solutions reported for
the NSGA-II, Jaya, and Rao algorithms. The conflicting nature of these objectives
can be observed in Fig. 9.2. Any change in the design variables that will result in
the increment of SPF also leads to the increment of the PMV and OC, which is not
desirable. Also, the variation of the SPF is relatively less when compared to the other
objectives. It can also be observed from the Pareto-optimal solutions obtained by the
algorithms. It indicates that for a small change in SPF caused by the design variables,
there is a relatively considerable change in the PMV and operating costs.
338 9 Optimization of Hydroenergy and Geothermal Energy Systems

Table 9.10 Pareto-optimal solutions obtained by the ERao-3 algorithm for the GSHP-RC system
case study
Solution a b c SPF PMV OC
1 16 28 9.743323 3.762294 0.71256 23,100.56
2 16.48512 25.91899 10.97858 3.697524 5.4E-06 23,825.1
3 16.11637 28 10.98185 3.743461 0.713093 22,082.21
4 16.25939 28 9.913288 3.757891 0.713748 23,021.83
5 16.17793 25.93108 9.313292 3.717463 0.006879 24,986.94
6 16.35046 26.40046 9.705634 3.725847 0.148988 24,351.24
7 16 26.54618 9.816065 3.734132 0.188011 24,011.28
8 16.42168 26.27675 9.52748 3.721838 0.110881 24,618.27
9 16.39246 26.79812 11 3.715384 0.282573 22,937.78
10 16.05327 26.90949 9.99245 3.739685 0.313677 23,608.91
11 16.20253 26.28523 11 3.708458 0.107164 23,315.25
12 16.09541 27.40063 10.99107 3.731951 0.48916 22,412.56
13 16.55551 26.63564 10.55846 3.719813 0.231811 23,548.06
14 16.03401 27.75923 10.17975 3.754989 0.621166 22,903.1
15 16.06842 26.06045 11 3.706222 0.030461 23,471.58
16 16.17544 26.48085 9.229082 3.72725 0.170733 24,518.54
17 16.02089 26.04747 10.83608 3.710985 0.024746 23,620.61
18 16.08193 25.99947 11 3.704831 0.011461 23,536.64
19 16.05195 27.30341 10.19902 3.745705 0.453324 23,167.82
20 16 27.21867 9.891739 3.746944 0.422022 23,447.59

The decision-making methods ranking of the Pareto-optimal solutions obtained

by different algorithms is shown in Table 9.12. The Spearman’s correlation coeffi-
cients for different pairs of rankings given by decision-making methods for different
algorithm’s solutions are shown in Table 9.13. The ranks given by the TOPSIS and
MTOPSIS methods for each solution are identical. Furthermore, the Spearman’s
correlation for all the pairs of decision-making methods (except with the VIKOR,
PROMETHEE, and GRA methods) is greater than 0.5. The Spearman’s correlation
value for the GRA-WPM pair is negative, and for the GRA method pairs with SAW,
TOPSIS, MTOPSIS, and COPRAS, it is less than 0.5. Similarly, the PROMETHEE
method has the Spearman’s correlation values less than 0.5 with the SAW, WPM, and
COPRAS methods. The VIKOR method has the Spearman’s correlation values less
than 0.5 with the SAW, WPM, and COPRAS methods. Hence, the ranks suggested by
the VIKOR, PROMETHEE, and GRA methods cannot be considered for calculating
the average ranks. Now, the corrected average ranks are calculated, excluding the
ranks suggested by the VIKOR, PROMETHEE, and GRA methods and presented in
Table 9.12 as corrected ranks.
Table 9.11 Best solutions and performance metrics obtained by various algorithms in MOO scenario of the GSHP-RC system case study
Algorithm a b c SPF PMV OC Hypervolume Spacing
NSGA-II solution 1 16.1 26.6 10.4 3.741 0.225 23,525 – –
NSGA-II solution 2 16.2 26 9.7 3.734 0.0203 24,613 – –
Jaya 16.56782 25.9273 11 3.695793 0.00543 23,835.93 18,072.62 0.1154082
Rao-1 16.1572 25.96175 11 3.702918 0.002064 23,607.25 18,303.26 0.0997092
Rao-2 16 25.97362 11 3.70562 0.000476 23,526.32 18,066.16 0.0837749
Rao-3 16.47459 25.91993 11 3.697117 8.03E-05 23,798.19 17,803.86 0.1268176
9.2 Optimization of a Ground Source Heat Pump …

AMTPG-Jaya 16.05922 25.96562 11 3.704534 5.02E-05 23,559.94 18,028.67 0.0942512

SAP-Rao 16.4169 25.92638 11 3.698149 8E-05 23,763.66 17,998.07 0.0699101
ERao-1 16.30606 25.93907 11 3.700138 4.66E-05 23,698.24 18,098.15 0.1310586
ERao-2 16.06596 25.96574 11 3.70443 0.000217 23,562.78 18,773.68 0.1191293
ERao-3 16.48512 25.91899 10.97858 3.697524 5.4E-06 23,825.1 18,360.59 0.1408059
Source NSGA-II: Xie et al. (2020)
Result in boldface indicates better values
339
340 9 Optimization of Hydroenergy and Geothermal Energy Systems

Fig. 9.2 Plot of Pareto-fronts of the proposed algorithms and solutions of the methods compared
in the GSHP-RC system case study

The AMTPG-Jaya algorithm solution has the least average rank, which is 1.8.
Thus, it can be regarded as the best solution. The ERao-1, ERao-2, ERao-3, and SAP-
Rao algorithms‘ solutions have the next best average ranks, which are 2.8, 3.8, 4, and
4.6, respectively. The AMTPG-Jaya algorithm solution ranked one by three methods
(TOPSIS, MTOPSIS, and COPRAS) and ranked three by two methods (SAW and
WPM). Here, an observation can be made that the AMTPG-Jaya algorithm solu-
tion has a better value in any objective. The NSGA-II solution 1 has better seasonal
performance factor and operating costs, and the ERao-3 algorithm solution has better
predicted mean value. However, the AMTPG-Jaya algorithm’s solution is ranked
one because it has best compromise among the three objectives. If we compare the
AMTPG-Jaya algorithm solution with NSGA-II solution 1, the AMTPG-Jaya algo-
rithm’s solution has 0.97% deterioration in SPF and 0.15% deterioration in operating
cost, Whereas the PMV has improved 99% compared to NSAGA-II solution 1. This
indicates that the relative improvement in the PMV is dominating the relative deteri-
oration in the SPF and OC. Also, the solutions of the Jaya, Rao-1, Rao-3, SAP-Rao,
and ERao-2 algorithms are dominated by the solution of the AMTPG-Jaya algorithm.
Hence, the ERao-3 algorithm solution has achieved the least average rank and can
be considered as the best solution among the compared.
Now, the performances of the Jaya algorithm, Rao algorithms, and their modified
versions in the MOO scenario are evaluated based on the hypervolume and spacing
indicators. These performance indicators are presented in Table 9.11. Here, an obser-
vation can be made that the ERao-2 algorithm Pareto-front has higher hypervolume
and the SAP-Rao algorithm has lower spacing value. The SAP-Rao algorithm has
a better spacing, which is 39, 29, 16, 44, 25, 46, 41, and 50% lesser value when
compared to that of the Jaya, Rao-1, Rao-2, Rao-3, AMTPG-Jaya, ERao-1, ERao-
2, and ERao-3 algorithms, respectively. The ERao-2 algorithm Pareto-front has a
Table 9.12 Ranks suggested by the MADM methods for different algorithm solutions presented in Table 9.11
Algorithm SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA AR CR FR
NSGA-II solution 1 8 11 11 11 6 3 11 2 7.875 10.4 11
NSGA-II solution 2 11 10 10 10 10 10 10 10 10.125 10.2 10
Jaya 10 9 9 9 11 11 9 11 9.875 9.2 9
Rao-1 9 8 8 8 4 8 8 5 7.25 8.2 8
Rao-2 7 7 3 3 1 2 7 1 3.875 5.4 6
Rao-3 5 5 6 6 9 9 6 8 6.75 5.6 7
9.2 Optimization of a Ground Source Heat Pump …

AMTPG-Jaya 3 3 1 1 2 1 1 3 1.875 1.8 1

SAP-Rao 4 4 5 5 7 6.5 5 7 5.4375 4.6 5
ERao-1 2 2 4 4 5 4 2 6 3.625 2.8 2
ERao-2 6 6 2 2 3 5 3 4 3.875 3.8 3
ERao-3 1 1 7 7 8 6.5 4 9 5.4375 4 4
AR—Average Rank; CR—Corrected Rank; FR—Final Rank
341
 342

Table 9.13 Spearman’s rank correlation coefficients between different pairs of MADM method’s rankings presented in Table 9.12
Method SAW WPM TOPSIS MTOPSIS VIKOR PROMETHEE COPRAS GRA
SAW 1 0.945 0.591 0.591 0.273 0.478 0.836 0.127
WPM 0.945 1 0.645 0.645 0.209 0.296 0.891 −0.055
TOPSIS 0.591 0.645 1 1 0.700 0.592 0.864 0.445
MTOPSIS 0.591 0.645 1 1 0.700 0.592 0.864 0.445
VIKOR 0.273 0.209 0.700 0.700 1 0.843 0.455 0.900
PROMETHEE 0.478 0.296 0.592 0.592 0.843 1 0.460 0.875
COPRAS 0.836 0.891 0.864 0.864 0.455 0.460 1 0.164
GRA 0.127 −0.055 0.445 0.445 0.900 0.875 0.164 1
9 Optimization of Hydroenergy and Geothermal Energy Systems
9.2 Optimization of a Ground Source Heat Pump … 343

higher hypervolume value which is 3.9, 2.6, 3.9, 5.4, 4.1, 4.3, 3.7, and 2.2% higher
than Jaya, Rao-1, Rao-2, Rao-3, AMTPG-Jaya, SAP-Rao, ERao-1, and ERao-3 algo-
rithms, respectively. The ERao-3, Rao-1, and ERao-1 algorithms have achieved the
next better hypervolume values, respectively. The performances of the improved
algorithms are better or competitive to that of the basic algorithms in terms of the
hypervolume and spacing values.
From the computational results of the GSHP-RC system case study, it can be
observed that the performance of the considered system can be improved with the
solutions achieved by the proposed algorithms. The solution of the AMTPG-Jaya
algorithm has better compromise among the SPF, PMV, and OC. The SAP-Rao
algorithm’s performance in terms of spacing and the ERao-2 algorithm’s performance
in terms of hypervolume are much better than that achieved by other algorithms. In
addition, the performances of the improved versions in this case study are better or
competitive to those of the basic algorithms as well as the NSGA-II algorithm. The
next chapter presents the conclusions.

References

Onokwai, A. O., Owamah, H. I., Ibiwoye, M. O., Ayuba, G. C., & Olayemi, O. A. (2020). Application
of response surface methodology (RSM) for the optimization of energy generation from Jebba
hydro-power plant, Nigeria. ISH Journal of Hydraulic Engineering. https://doi.org/10.1080/097
15010.2020.1806120
Xie, Y., Hu, P., Zhu, N., Lei, F., Xing, L., & Xu, L. (2020). Collaborative optimization of ground
source heat pump-radiant ceiling air conditioning system based on response surface method and
NSGA-II. Renewable Energy, 147, 249–264. https://doi.org/10.1016/j.renene.2019.08.109
Appendices

Appendix A. Single-objective Optimization of Standard

Benchmark Problems

Appendix A.1. Unconstrained Standard Benchmark Problems

Problem Function Formulation D Interval C

D 2
F1 Sphere F1 (y) = i=1 yi 30 [−100, US
100]
D
F2 SumSquares F2 (y) = i=1 i yi2 30 [−10, 10] US
F3 Beale F3 (y) = 2 [−4.5, UN
D    
2 2 + 2.625 − y + y y 3 2 4.5]
i=1 (1.5 − y1 + y1 y2 ) + 2.25 − y1 + y1 y2
2
1 1 2

F4 Easom F4 (y) = − cos(y1 ) cos(y2 ) exp(−(y1 − π )2 − (y2 − π )2 ) 2 [−100, UN

100]
F5 Matyas F5 (y) = 0.26(y12 + y22 ) − 0.48y1 y2 2 [−10, 10] UN
F6 (y) = 100(y12 − y2 )2 + (y1 − 1)2 + (y3 − 1)2 − 90(y32 − y4 )
F6 Colville   4 [−10, 10] UN
+ 10.1 (y2 − 1)2 + (y4 − 1)2 + 19.8(y2 − 1)(y4 − 1)
D  D
F7 Trid 6 F7 (y) = i=1 (yi − 1)2 − i=2 yi yi−1 6 [−D2 , UN
D2 ]
D D
F8 Trid 10 F8 (y) = i=1 (yi − 1)2 − i=2 yi yi−1 10 [−D2 , UN
D2 ]
D  2  4
D D
F9 Zakharov F9 (y) = 2
i=1 yi + i=1 0.5i yi + i=1 0.5i yi 10 [−5, 10] UN
 D i 2
F10 Schwefel 1.2 F10 (y) = i=1 j=1 y 2j 30 [−100, UN
100]
D   2 
F11 Rosenbrock F11 (y) = 100 yi2 − yi+1 + (1 − yi )2
i=1 30 [−30, 30] UN
D   2
F12 Dixon-Price F12 (y) = (yi − 1)2 + i=2 i 2yi2 − yi−1 30 [−10, 10] UN
25
F13 Shekel’s F23 (y) = 1
+ j=1 j+2
1
2 [−65.536, MS
i=1 ( yi −ai j )
500 6
Foxholes 65.536]
 2  
F14 Branin F13 (y) = y2 − 5.1
4π 2
y12 + 5
π y1 − 6 + 10 1 − 1
8π cos y1 + 10 2 [−5, 10] MS
[0, 15]
(continued)

© The Editor(s) (if applicable) and The Author(s), under exclusive license 345
to Springer Nature Switzerland AG 2022
V. R. Ravipudi et al., Design Optimization of Renewable Energy Systems Using Advanced
Optimization Algorithms, Green Energy and Technology,
https://doi.org/10.1007/978-3-030-95589-2
346 Appendices

(continued)
Problem Function Formulation D Interval C
F15 Bohachevsky 1 F14 (y) = y12 + 2y22 − 0.3 cos(3π y1 ) − 0.4 cos(4π y2 ) + 0.7 2 [−100, MS
100]
F16 Booth F17 (y) = (y1 + 2y2 − 7)2 + (2y1 + y2 − 5)2 2 [−10, 10] MS
D   20
F17 Michalewicz 2 F18 (y) = − i=1 sin y1 sin i yi2 /π 2 [0, π] MS
D   2 20
F18 Michalewicz 5 F19 (y) = − i=1 sin y1 sin i yi /π 5 [0, π] MS
F19 Bohachevsky 2 F15 (y) = y12 + 2y22 − 0.3 cos(3π y1 )(4π y2 ) + 0.3 2 [−100, MN
100]
F20 Bohachevsky 3 F16 (y) = y12 + 2y22 − 0.3 cos(3π y1 + 4π y2 ) + 0.3 2 [−100, MN
100]
 
F20 (y) = 1 + (y1 + y2 + 1)2 19 − 14y1 + 3y12 − 14y2 + 6y1 y2 + 3y22
F21 GoldStein-Price   2 [−2, 2] MN
30 + (2y1 − 3y2 ) 18 − 32y1 + 12y1 + 48y2 − 36y1 y2 + 27y22
2 2


D D  k    2
F22 Perm F21 (y) = k=1 i=1 i + β (yi /i) − 1
k
4 [−D, D] MN
4    2 
F23 Hartmann 3 F24 (y) = − i=1 ci exp − 3j=1 ai j y j − pi j 3 [0, 1] MN
     
D D
F24 Ackley F22 (y) = −20 exp −0.2 D1 2 − exp 1
i=1 yi D i=1 cos 2π yi + 20 + e 30 [−32, 32] MN

   
 (yi − 1)2 1 + sin2 (3π yi+1 )
D−1
F25 (y) = 0.1 sin2 (π y1 ) +  
i=1 +(y D − 1) + 1 + sin (2π y D )
2 2

⎧ ⎫
⎪
⎪ k(yi − a)m yi  a, ⎪ ⎪
F25 Penalized 2 ⎪
⎪ ⎪
⎪ 30 [−50, 50] MN
D ⎨ ⎬
+ u(yi , 5, 100, 4), u(yi , a, k, m) = 0 − a ≤ yi ≤ a,
⎪
⎪ ⎪
⎪
i=1 ⎪
⎪ ⎪
⎪
⎩ ⎭
k(−yi − a)m yi ≺ −a

F26 Langerman 2 F26 (y) = 2 [0, 10] MN

D   D  2    D  2 
− i=1 ci exp − π1 j=1 x j − ai j cos π j=1 x j − ai j

F27 Langerman 5 F27 (y) = 5 [0, 10] MN

D    2    D  2 
1 D
− i=1 ci exp − π j=1 x j − ai j cos π j=1 x j − ai j

F28 Langerman 10 F28 (y) = 10 [0, 10] MN

D    2    D  2 
1 D
− i=1 ci exp − π j=1 x j − ai j cos π j=1 x j − ai j
D
F29 Fletcher Powell F29 (y) = i=1 (Ai − Bi )2 5 [−π, π] MN
5 D     
Ai = j=1 ai j Sin α j + bi j Cos α j , Bi =
D     
j=1 ai j Sin x j + bi j Cos x j
D
F30 Fletcher Powell F30 (y) = i=1 (Ai − Bi )2 10 [−π, π] MN
1 D     
Ai = j=1 ai j Sin α j + bi j Cos α j , Bi =
D     
j=1 ai j Sin x j + bi j Cos x j

D—Dimension, C—Characteristics, U—Unimodal, M—Multimodal, S—Separable, N—Non-separable

Appendix A.2. Unimodal and Multimodal Benchmark

Problems Set-1
Problem Formulation Interval
n
P1 G 1 (x) = i=1 xi2 [−100, 100]
n
Appendices

P2 n |x |
G 2 (x) = i=1 |xi | + i=1 i [−10, 10]
n  i 2
P3 G 3 (x) = i=1 j=1 x j [−100, 100]

P4 G 4 (x) = maxi {|xi |, 1 ≤ i ≤ n} [−100, 100]

n−1   2 
P5 G 5 (x) = i=1 100 xi+1 − xi2 + (xi − 1)2 [−30, 30]
n
P6 G 6 (x) = i=1 (|xi + 0.5|)2 [−100, 100]
n
P7 G 7 (x) = i=1 i xi4 + random[0, 1) [−1.28, 1.28]
n √ 
P8 G 8 (x) = i=1 −xi sin |xi | [−500, 500]
n
P9 G 9 (x) = i=1 xi2 − 10 cos(2π xi ) + 10 [−5.12, 5.12]

 
n   n 
P10 G 10 (x) = −20 exp −0.2 n1 i=1 xi2 − exp n1 i=1 cos(2π xi ) + 20 + e [−32, 32]
n  
1 2 xi
√
P11 G 11 (x) = 4000 i=1 x i
n cos
− i=1 +1 [−600, 600]
i

(continued)
347
(continued)
348

Problem Formulation Interval

 n−1

π   
G 12 (x) = 10 sin2 (π.y) + (yi − 1)2 1 + 10 sin2 (π.yi+1 ) + (yn − 1)2
n
i=1
P12 [−50, 50]
n

+ u(xi , 10, 100, 4)
i=1
 
xi +1
yi = 1 + 4
⎧ ⎫
⎪ m ⎪
⎪
⎨ k(xi − a) xi > a ⎪
⎬
u(xi , a.k, m) = 0 −a < xi < a
⎪
⎪ ⎪
⎪
⎭
i i
⎩ k(−x − a)m x < −a

⎧  ⎫
⎪
⎨ ⎬
n−1 (x i − 1)2 1 + sin2 (3π x i+1 ) ⎪

G 13 (x) = 0.1 sin2 (3π x1 ) +  
⎪
⎩ 2 2 ⎪
i=1 +(x n − 1) 1 + sin (2π x n ) ⎭
P13 [−50, 50]
n

+ u(xi , 5, 100, 4)
i=1

−1
1 25 1
P14 G 14 (x) = 500 + j=1 j+2 6 [−65, 65]
i=1 ( x i −ai j )

  2
11 x 1 bi2 +bi x 2
P15 G 15 (x) = i=1 ai − bi2 +bi x 3 +x 4
[−5, 5]

(continued)
Appendices
(continued)
Problem Formulation Interval
P16 [−5, 5]
Appendices

G 16 (x) = 4x12 + 2.1x14 + 13 x16 + x1 x2 − 4x22 + 4x24

 2  
5.1 2 5 1
P17 G 17 (x) = x2 − 4π 2 x1 + π x1 − 6 + 10 1 − 8π cos x1 + 10 [−5, 5]
  
G 18 (x) = 1 + (x1 + x2 + 1)2 19 − 14x1 + 3x12 − 14x2 + 6x1 x2 + 3x22
P18    [−2, 2]
× 30 + (2x1 − 3x2 )2 × 18 − 32x1 + 12x12 + 48x2 − 36x1 x2 + 27x22
4    2 
P19 G 19 (x) = − i=1 Ci exp − 3j=1 ai j x j − Pi j [0, 1]
4    2 
P20 G 20 (x) = − i=1 Ci exp − 6j=1 ai j x j − Pi j [0, 1]
5   2 =1
4
P21 G 21 (x) = − i=1 j=1 x j − ai j + ci [0, 10]
7   2 =1
4
P22 G 22 (x) = − i=1 j=1 x j − ai j + ci [0, 10]
10   2 =1
4
P23 G 23 (x) = − i=1 j=1 x j − ai j + ci [0, 10]
349
350 Appendices

Appendix A.3. Unimodal and Multimodal Benchmark

Problems Set-2

Problem Test function Interval

d
P1 TF1 (y) = i=1 yi2 [−100, 100]d
d d
P2 TF2 (y) = i=1 abs|yi | + i=1 abs|yi | [−10, 10]d
d m 2
P3 TF3 (y) = m=1 n=1 yn [−100, 100]d
P4 TF4 (y) = max{|yi |, 1 ≤ i ≤ d} [−100, 100]d
  
P5 TF5 (y) = dm=1 100 ym+1 − ym2 + (yi − 1)2 [−30, 30]d
d
P6 TF6 (y) = i=1 yi + 0.52 [−100, 100]d
d
P7 TF7 (y) = i=1 i yi4 + rand[0, 1] [−1.28, 1.28]d
 √ 
P8 TF8 (y) = dm=1 −ym sin |ym | [−500, 500]d
d
P9 TF9 (y) = m=1 ym2 − 10 cos(2π ym ) + 10 [−5.12, 5.12]d

P10 TF10 (y) = [−32, 32]d

     
d
20 −0.2 1
d
2
m=1 ym − exp d1 dm=1 cos(2π ym ) + 20 + e

Appendix B. Multi-objective Optimization Standard

Benchmark Problems

Multi-objective Optimization ZDT Test Problems

ZDT-1 problem:
Minimize: f 1 (x) = x1
Minimize: f 2 (x) = g(x) × h( f 1 (x), g(x))

N
where G(x) = 1 + 9
N −1
xi
i=2


f 1 (x)
h( f 1 (x), g(x)) = 1 −
g(x)

0 ≤ xi ≤ 1, 1 ≤ i ≤ 30
Appendices 351

ZDT-2 problem:
Minimize: f 1 (x) = x1
Minimize: f 2 (x) = g(x) × h( f 1 (x), g(x))

N
where G(x) = 1 + 9
N −1
xi
i=2

2
f 1 (x)
h( f 1 (x), g(x)) = 1 −
g(x)

0 ≤ xi ≤ 1, 1 ≤ i ≤ 30

ZDT-3 problem:
Minimize: f 1 (x) = x1
Minimize: f 2 (x) = g(x) × h( f 1 (x), g(x))

N
where G(x) = 1 + 9
29
xi
i=2

 
f 1 (x) f 1 (x)
h( f 1 (x), g(x)) = 1 − − sin(10π f 1 (x))
g(x) g(x)

0 ≤ xi ≤ 1, 1 ≤ i ≤ 30

ZDT-1 with linear Pareto-front (ZDT-1L) problem:

Minimize: f 1 (x) = x1
Minimize: f 2 (x) = g(x) × h( f 1 (x), g(x))

N
where G(x) = 1 + 9
N −1
xi
i=2

f 1 (x)
h( f 1 (x), g(x)) = 1 −
g(x)

0 ≤ xi ≤ 1, 1 ≤ i ≤ 30

ZDT-2 with three-objective (ZDT2-3O) problem:

Minimize: f 1 (x) = x1
Minimize: f 2 (x) = x2
352 Appendices

Minimize: f 3 (x) = g(x) × h( f 1 (x), g(x)) × h( f 2 (x), g(x))


N
where G(x) = 1 + 9
N −1
xi
i=3

2
f 1 (x)
h( f 1 (x), g(x)) = 1 −
g(x)
2
f 2 (x)
h( f 2 (x), g(x)) = 1 −
g(x)

0 ≤ xi ≤ 1, 1 ≤ i ≤ 30

Appendix C. Codes for Jaya and Rao Algorithms (and their

improved versions) for Unconstrained Optimization
Problems

The codes for unconstrained benchmark problems presented in Appendix A.1 for
sample single-objective optimization functions are given below. The user has to create
separate MATLAB files, but the files are to be saved in a single folder. These codes
may be used for reference, and the user may define the objective function(s), design
variables, and their ranges as per his or her own requirements. The algorithm codes
are made as function file which needs to be called in the ‘All_Algorithms_Main.m’
file.

C.1 All_Algorithms_Main.m: Main Program for Executing

All Algorithms

This program is to initialize the optimization problem parameters and algorithm

parameters. The variable TME defines the algorithm to be executed. The TME values
for different algorithms are as follows: TME = −2 executes the AMTPG-Jaya algo-
rithm, TME = −1 executes the MTPG-Jaya algorithm, TME = 0 executes the Jaya
algorithm, TME = 1 executes the Rao-1 algorithm, TME = 2 executes the Rao-2
algorithm, TME = 3 executes the Rao-3 algorithm, TME = 4 executes the Rao-4
algorithm, TME = 5 executes the Elitist_Rao-1 algorithm, TME = 6 executes the
Elitist_Rao-2 algorithm, TME = 7 executes the Elitist_Rao-3 algorithm, TME = 8
executes the Elitist_Rao-4 algorithm, and TME = 9 executes the SAP_Rao algorithm.
The unconstrained benchmark problems presented in Appendix A.1 are coded in the
‘Unconstrained_benchmark_func.m’ file. This program is only for demonstration
Appendices 353

purpose. The numbers assigned to the population size, generations, design variables,
and maximum function evaluations in this program need not be taken as default
values
%%%%Code Begins %%%%
%%%% All_Algorithms_Main.m%%%%
clear
clear global
clear global variables
clc
warning off
rehash
global func_num
global L
global U
global P
global D
global MFEs
global Fes
global Ter_Error
global TME
global convergence
global ES
global Max_N_Teams
global Min_N_Teams
convergence =[];
format short g
MFEs=100000; %% maximum number of function evaluations
for P=20 %% population size
runs=30; %% number of runs
%% TME is to select the algorithm that we want to run
%%% TME=-2 Executes the AMTPG-Jaya algorithm
%%% TME=-1 Executes the AMTPG-Jaya algorithm
%%% TME=0 Executes the Jaya algorithm
%%% TME=1 Executes the Rao-1 algorithm
%%% TME=2 Executes the Rao-2 algorithm
%%% TME=3 Executes the Rao-3 algorithm
%%% TME=4 Executes the Rao-4 algorithm
%%% TME=5 Executes the Elitist_Rao-1 algorithm
%%% TME=6 Executes the Elitist_Rao-2 algorithm
%%% TME=7 Executes the Elitist_Rao-3 algorithm
%%% TME=8 Executes the Elitist_Rao-4 algorithm
%%% TME=9 Executes the SAP_Rao algorithm
for TME=[-2,-1,0,1,2,3,4,5,6,7,8,9] %% Algorithm number
%%% select the unconstrained benchmark function number given
% in the function file named "Unconstrained_benchmark_func.m"
for func_num=1:30
% Problem boundaries & number of variables are presented in "Dand-
Bounds.m".
[D,lb1,ub1,Ter_Error]=DandBounds(func_num);
L=lb1*ones(1,D); %% Lower boundaries of variables
U=ub1*ones(1,D); %% Upper boundaries of variables
ObjectiveFunction=@Unconstrained_benchmark_func; %% Func-
tions file
354 Appendices

GO=zeros(runs,1); %% Global Optimum

GV=zeros(runs,D); %% Global Optimum variables
GNFE=zeros(runs,1); %% number of function evalua-
tions to reach optimum
% Avg_eva=nan((MFEs/P),runs);
Avg_obj=nan(round(MFEs/P),runs);
for run=1:runs
rng(’shuffle’)
if TME<=4 && TME>=0 %% Jaya and Rao algorithms
[xf]=single_obj(ObjectiveFunction);
elseif TME==-1 %% MTPG-Jaya algorithm
[xf]=MTPG_single_obj(ObjectiveFunction);
elseif TME==-2 %% AMTPG-Jaya algorithm
Min_N_Teams=2;
Max_N_Teams=6;
[xf]=AMTPG_single_obj(ObjectiveFunction);
elseif TME>=5 && TME<=8%% ElitistRao algorithms
ES=round(0.1*P); %% Elite size
[xf]=E_Rao_single_obj(ObjectiveFunction);
elseif TME==9 %% SAP-Rao algorithm
[xf]=SAP_Rao_single_obj(ObjectiveFunction);
end
[val,ind]=min(xf(:,D+1));
A=xf(ind,1:D);
B=val;
GO(run,1)= B;
GV(run,:)= A;
GNFE(run,1)= Fes;
[rows,~]=size(convergence);
% Avg_eva(1:rows,run)=convergence(:,1);
Avg_obj(1:rows,run)=convergence(:,2);
convergence =[];
end
% AE=mean(Avg_eva(:,:),2);
AO=mean(Avg_obj(:,:),2);
bbest=min(GO);
wbest=max(GO);
mbest=mean(GO);
stdbest=std(GO);
mFes=mean(GNFE);
stdFes=std(GNFE);
disp([’--------[ Algorithm’,’ TME No.= ’,num2str(TME), ’ ]-----
-----’])
disp([’--------[ ’,’Fun No.= ’,num2str(func_num), ’ ]---------
------’])
fprintf(’ best=%f worst=%f \n mean=%f std=%f \n mean Fes=
%f \n std Fes=%f \n’,bbest,wbest,mbest,stdbest,mFes,stdFes);
fprintf(’\n ’)
beep;
end
end
end
%%%%Code Ends %%%%
Appendices 355

C.2 single_obj.m: Jaya and Rao Algorithm Programs

This function file needs to be called in the ‘All_Algorithms_Main.m’ program to

execute the algorithms. As the flow of the Jaya and Rao algorithms is similar and
they differ only in movement equation, a common function file is created for these
algorithms with sub-function files for each algorithm. The sub-functions for the Jaya
and Rao algorithms are as follows: ‘updatepop’ for the Jaya algorithm, ‘Rao1’ for the
Rao-1 algorithm, ‘Rao2’ for the Rao-2 algorithm, ‘Rao3’ for the Rao-3 algorithm,
and ‘Rao4’ for the Rao-4 algorithm. The ‘Trimr’ function is to check and update the
boundary constraints of the variables.
%%%%Code Begins %%%%
%%%% single_obj.m%%%%
function [xf]=single_obj(ObjectiveFunction)
global L
global U
global P
global D
global MFEs
global TME
global Fes
global initial_flag
global Ter_Error
global convergence
x=repmat(L,P,1)+rand(P,D).*repmat((U-L),P,1);
initial_flag=0;
f=ObjectiveFunction(x);
xf=[x,f];
NFE=0;
Gfold(1,1)=min(f);
counter =1;
while NFE<MFEs
nx=[];
if TME==0
[nx]=updatepop(xf); %% Jaya algorithm
elseif TME==1
[nx]=Rao1(xf); %% Rao-1 algorithm
elseif TME==2
[nx]=Rao2(xf); %% Rao-2 algorithm
elseif TME==3
[nx]=Rao3(xf); %% Rao-3 algorithm
elseif TME==4
[nx]=Rao4(xf); %% Rao-4 algorithm
end
[nx,~]=trim(nx,L,U);
initial_flag=0;
newf = ObjectiveFunction(nx);
NFE = NFE + P;
for i=1:P
if(newf(i)<=f(i))
x(i,:)=nx(i,:);
f(i)=newf(i);
356 Appendices

end
end
xf=[x,f];
Gfnew(1,1)=min(f);
if Gfnew<Gfold
Fes=NFE;
Gfold=Gfnew;
end
if Gfold<=Ter_Error
Fes=NFE;
break
end
convergence(counter,1)=NFE;
convergence(counter,2)=Gfold;
counter =counter+1;
end
end

function [nx]=Rao1(f)
%% Rao-1 algorithm
[row,col]=size(f);
F=f(:,col);
[~,indx]=sort(F);
x=f(:,1:col-1);
xb=x(indx(1,1),:);
xw=x(indx(row,1),:);
bx=repmat(xb,row,1);
wx=repmat(xw,row,1);
r=rand(row,col-1);
nx=x+(r.*(bx-wx));
end

function [nx]=Rao2(f)
%% Rao-2 algorithm
[row,col]=size(f);
x=f(:,1:col-1);
F=f(:,col);
[~,indx]=sort(F);
Best=x(indx(1,1),:);
worst=x(indx(row,1),:);
xnew=zeros(row,col-1);
for i=1:row
k=unidrnd(row);
while i==k
k=randi(row);
end
if F(i)<F(k)%% minimization objective
xnew(i,:)=x(i,:)+rand(1,(col-1)).*(Best-worst)+rand(1,(col-
1)).*(abs(x(i,:))-abs(x(k,:)));
else
xnew(i,:)=x(i,:)+rand(1,(col-1)).*(Best-worst)+rand(1,(col-
1)).*(abs(x(k,:))-abs(x(i,:)));
end
end
nx=xnew;
Appendices 357

end

function [nx]=Rao3(f)
%% Rao-3 algorithm
[row,col]=size(f);
x=f(:,1:col-1);
F=f(:,col);
[~,indx]=sort(F);
Best=x(indx(1,1),:);
worst=x(indx(row,1),:);
xnew=zeros(row,col-1);
for i=1:row
k=unidrnd(row);
while i==k
k=randi(row);
end
if F(i)<F(k)%% minimization objective
xnew(i,:)=x(i,:)+rand(1,(col-1)).*(Best-
abs(worst))+rand(1,(col-1)).*(abs(x(i,:))-x(k,:));
else
xnew(i,:)=x(i,:)+rand(1,(col-1)).*(Best-
abs(worst))+rand(1,(col-1)).*(abs(x(k,:))-x(i,:));
end
end
nx=xnew;
end

function [nx]=Rao4(f)
%% Rao-4 algorithm
[row,col]=size(f);
F=f(:,col);
[~,indx]=sort(F);
x=f(:,1:col-1);
xb=x(indx(1,1),:);
xw=x(indx(row,1),:);
bx=repmat(xb,row,1);
wx=repmat(xw,row,1);
r=rand(row,col-1);
r1=rand(row,col-1);
r2=rand(row,col-1);
r3=rand(row,col-1);
nx=x+(r.*(bx-wx))+(r1.*(wx-x)+r2.*(bx-x))*0.5-r3.*(bx-(x));
end

function [nx]=updatepop(f)
%% Jaya algorithm
[pop,c]=size(f);
F=f(:,c);
[~,indx]=sort(F);
x=f(:,1:c-1);
xb=x(indx(1,1),:);
xw=x(indx(pop,1),:);
bx=repmat(xb,pop,1);
wx=repmat(xw,pop,1);
r=rand(pop,c-1);
358 Appendices

r1=rand(pop,c-1);
nx=x+r.*(bx-abs(x))-r1.*(wx-abs(x));
end

function [nx,violations]=trim(nx,L,U)
%% Boundary constraints violation check
[~,D]=size(nx);
violations=0;
for i = 1:D
indexes = (nx(:,i) < L(i));
violations = violations + abs(sum(L(i)-nx(indexes,i)));
if isempty(indexes)==0
nx(indexes,i) = L(i);
end
indexes = (nx(:,i) > U(i));
violations = violations + abs(sum(nx(indexes,i)-U(i)));
if isempty(indexes)==0
nx(indexes,i) = U(i);
end
end
end
%%%%Code Ends %%%%

C.3 E_Rao_single_obj.m: Elitist Rao Algorithm Programs

This function file needs to be called in the ‘All_Algorithms_Main.m’ program to

execute the algorithms. As the flow of the elitist Rao algorithms is similar and they
differ only in movement equation, a common function file is created for these algo-
rithms with sub-function files for movement equation of each algorithm. The sub-
functions for the elitist Rao algorithms are as follows: ‘Rao1’ for the elitist Rao-1
algorithm, ‘Rao2’ for the elitist Rao-2 algorithm, ‘Rao3’ for the elitist Rao-3 algo-
rithm, and ‘Rao4’ for the elitist Rao-4 algorithm. The ‘Trimr’ function is to check
and update the boundary constraints of the variables.
%%%%Code Begins %%%%
%%%% E_Rao_single_obj.m%%%%
function [xf]=E_Rao_single_obj(ObjectiveFunction)
global L
global U
global P
global D
global MFEs
global TME
global Fes
global initial_flag
global Ter_Error
global convergence
global ES
initial_flag=0;
Appendices 359

x=repmat(L,P,1)+rand(P,D).*repmat((U-L),P,1);
f=nan(P,1);
for i =1:P
f(i,1)=ObjectiveFunction(x(i,:));
end
xf=[x,f];
NFE=0;
Gfold(1,1)=min(f);
counter =1;
while NFE<MFEs
xf=unique(xf, ’rows’);
xf=sortrows(xf,D+1); % sort the solutions in ascending order
[r,~]= size(xf);
if r<P
for i=r+1:P
for j = 1 : D
xf(i,j) = L(j) + (U(j) - L(j))*rand(1);
end
end
for i = r+1:P
xf(i,D+1) = ObjectiveFunction(xf(i,1:D));
end
else
xf(:,:)=xf(:,:);
end
%% Replace the worst solution with elite solutions
if mod(counter,round(P))==0
xf(P-ES+1:P,:)= xf(1:ES,:);
end
%%
nx=[];
if TME==5 %% Elitist Rao-1 algorithm
[nx]=Rao1(xf);
elseif TME==6 %% Elitist Rao-2 algorithm
[nx]=Rao2(xf);
elseif TME==7 %% Elitist Rao-3 algorithm
[nx]=Rao3(xf);
elseif TME==8 %% Elitist Rao-4 algorithm
[nx]=Rao4(xf);
end
[nx,~]=trim(nx,L,U);
newf=nan(P,1);
for i =1:P
newf(i,1)=ObjectiveFunction(nx(i,:));
end
NFE = NFE + P;
for i=1:P
if(newf(i)<=f(i))
x(i,:)=nx(i,:);
f(i)=newf(i);
end
end
xf=[x,f];
Gfnew(1,1)=min(f);
360 Appendices

if Gfnew<Gfold
Fes=NFE;
Gfold=Gfnew;
end
if Gfold<=Ter_Error
Fes=NFE;
break
end
convergence(counter,1)=NFE;
convergence(counter,2)=Gfold;
counter =counter+1;
end
end

function [nx]=Rao1(f)
%% Elitist Rao-1 algorithm
[row,col]=size(f);
F=f(:,col);
[~,indx]=sort(F);
x=f(:,1:col-1);
xb=x(indx(1,1),:);
xw=x(indx(row,1),:);
bx=repmat(xb,row,1);
wx=repmat(xw,row,1);
r=rand(row,col-1);
nx=x+(r.*(bx-wx));
end

function [nx]=Rao2(f)
%% Elitist Rao-2 algorithm
[row,col]=size(f);
x=f(:,1:col-1);
F=f(:,col);
[~,indx]=sort(F);
Best=x(indx(1,1),:);
worst=x(indx(row,1),:);
xnew=zeros(row,col-1);
for i=1:row
k=unidrnd(row);
while i==k
k=randi(row);
end
if F(i)<F(k)%% minimization objective
xnew(i,:)=x(i,:)+rand(1,(col-1)).*(Best-worst)+rand(1,(col-
1)).*(abs(x(i,:))-abs(x(k,:)));
else
xnew(i,:)=x(i,:)+rand(1,(col-1)).*(Best-worst)+rand(1,(col-
1)).*(abs(x(k,:))-abs(x(i,:)));
end
end
nx=xnew;
end

function [nx]=Rao3(f)
%% Elitist Rao-3 algorithm
Appendices 361

[row,col]=size(f);
x=f(:,1:col-1);
F=f(:,col);
[~,indx]=sort(F);
Best=x(indx(1,1),:);
worst=x(indx(row,1),:);
xnew=zeros(row,col-1);
for i=1:row
k=unidrnd(row);
while i==k
k=randi(row);
end
if F(i)<F(k)%% minimization objective
xnew(i,:)=x(i,:)+rand(1,(col-1)).*(Best-
abs(worst))+rand(1,(col-1)).*(abs(x(i,:))-x(k,:));
else
xnew(i,:)=x(i,:)+rand(1,(col-1)).*(Best-
abs(worst))+rand(1,(col-1)).*(abs(x(k,:))-x(i,:));
end
end
nx=xnew;
end
function [nx]=Rao4(f)
%% Elitist Rao-4 algorithm
[row,col]=size(f);
F=f(:,col);
[~,indx]=sort(F);
x=f(:,1:col-1);
xb=x(indx(1,1),:);
xw=x(indx(row,1),:);
bx=repmat(xb,row,1);
wx=repmat(xw,row,1);
r=rand(row,col-1);
r1=rand(row,col-1);
r2=rand(row,col-1);
r3=rand(row,col-1);
nx=x+(r.*(bx-wx))+(r1.*(wx-x)+r2.*(bx-x))*0.5-r3.*(bx-(x));
end
function [nx,violations]=trim(nx,L,U)
%% Boundary constraints violation check
[~,D]=size(nx);
violations=0;
for i = 1:D
indexes = (nx(:,i) < L(i));
violations = violations + abs(sum(L(i)-nx(indexes,i)));
if isempty(indexes)==0
nx(indexes,i) = L(i);
end
indexes = (nx(:,i) > U(i));
violations = violations + abs(sum(nx(indexes,i)-U(i)));
if isempty(indexes)==0
nx(indexes,i) = U(i);
end
end
end
%%%%Code Ends %%%%
362 Appendices

C.4 SAP_Rao_single_obj.m: Self-Adaptive Population Rao

Algorithm Program

This function file needs to be called in the ‘All_Algorithms_Main.m’ program to

execute the SAP-Rao algorithm. This program is only for demonstration purpose.
%%%%Code Begins %%%%
%%%% SAP_Rao_single_obj.m%%%%
function [xf]=SAP_Rao_single_obj(ObjectiveFunction)
global L
global U
global P
global D
global MFEs
global Fes
global initial_flag
global Ter_Error
global convergence
%%
initial_flag=0;
x=repmat(L,P,1)+rand(P,D).*repmat((U-L),P,1);
f=nan(P,1);
for i =1:P
f(i,1)=ObjectiveFunction(x(i,:));
end
xf=[x,f];
NFE=0;
Gfold(1,1)=min(f);
GOFV=min(f);
counter =1;
while NFE<MFEs
nx=nan(P,D);
sub_pop_size=floor(P/4);
IND=randperm(P);
from=1;
for k=1:4
if(k~=4)
to=from+sub_pop_size-1;
ind=IND(from:to);
from=from+sub_pop_size;
else
ind=IND(from:end);
end
tempxf=xf(ind,:);
tme=k;%% this is a local variable
if tme==4
[tempxnew]=Rao1(tempxf);
elseif tme==1
[tempxnew]=Rao2(tempxf);
elseif tme==2
[tempxnew]=Rao3(tempxf);
elseif tme==3
[tempxnew]=Rao4(tempxf);
Appendices 363

end
nx(ind,:)=tempxnew;
end
[nx,~]=trim(nx,L,U);
newf=nan(P,1);
for i =1:P
newf(i,1)=ObjectiveFunction(nx(i,:));
end
NFE = NFE + P;
for i=1:P
if(newf(i)<f(i))
x(i,:)=nx(i,:);
f(i)=newf(i);
end
end
xf=[x,f];
Gfnew(1,1)=min(f);
%% %%%%%%%%%%%%%%%%%
xf=sortrows(xf,D+1);
%%%%%%%%%%%%%%%%%%%%
%%
% disp(mod(Gfold,Gfnew))
if abs(mod(Gfold,Gfnew))<1e-1
if P>20
P=P-round(0.05*P);
x=xf(1:P,1:D);
f=xf(1:P,D+1);
xf=xf(1:P,:);
end
else
if P<1000
P_old=P;
P=P+round(0.05*P);
P_diff=P-P_old;
xfadded=xf(1:P_diff,:);
x=[x;xfadded(:,1:D)];
f=[f;xfadded(:,D+1)];
xf=[x,f];
end
end
if Gfnew<GOFV
Fes=NFE;
GOFV=Gfnew;
end
if GOFV<=Ter_Error
Fes=NFE;
break
end
Gfold=Gfnew;
convergence(counter,1)=NFE;
convergence(counter,2)=GOFV;
convergence(counter,3)=P;
counter =counter+1;
end
364 Appendices

xf=[x,f];
end
function [nx]=Rao1(f)
[row,col]=size(f);
F=f(:,col);
[~,indx]=sort(F);
x=f(:,1:col-1);
xb=x(indx(1,1),:);
xw=x(indx(row,1),:);
bx=repmat(xb,row,1);
wx=repmat(xw,row,1);
r=rand(row,col-1);
nx=x+(r.*(bx-wx));
end
function [nx]=Rao2(f)
[row,col]=size(f);
x=f(:,1:col-1);
F=f(:,col);
[~,indx]=sort(F);
Best=x(indx(1,1),:);
worst=x(indx(row,1),:);
xnew=zeros(row,col-1);
for i=1:row
k=unidrnd(row);
while i==k
k=randi(row);
end
if F(i)<F(k)%% minimization objective
xnew(i,:)=x(i,:)+rand(1,(col-1)).*(Best-worst)+rand(1,(col-
1)).*(abs(x(i,:))-abs(x(k,:)));
else
xnew(i,:)=x(i,:)+rand(1,(col-1)).*(Best-worst)+rand(1,(col-
1)).*(abs(x(k,:))-abs(x(i,:)));
end
end
nx=xnew;
end
function [nx]=Rao3(f)
[row,col]=size(f);
x=f(:,1:col-1);
F=f(:,col);
[~,indx]=sort(F);
Best=x(indx(1,1),:);
worst=x(indx(row,1),:);
xnew=zeros(row,col-1);
for i=1:row
k=unidrnd(row);
while i==k
k=randi(row);
end
if F(i)<F(k)%% minimization objective
xnew(i,:)=x(i,:)+rand(1,(col-1)).*(Best-
abs(worst))+rand(1,(col-1)).*(abs(x(i,:))-x(k,:));
else
Appendices 365

xnew(i,:)=x(i,:)+rand(1,(col-1)).*(Best-
abs(worst))+rand(1,(col-1)).*(abs(x(k,:))-x(i,:));
end
end
nx=xnew;
end
function [nx]=Rao4(f)
[row,col]=size(f);
F=f(:,col);
[~,indx]=sort(F);
x=f(:,1:col-1);
xb=x(indx(1,1),:);
xw=x(indx(row,1),:);
bx=repmat(xb,row,1);
wx=repmat(xw,row,1);
r=rand(row,col-1);
r1=rand(row,col-1);
r2=rand(row,col-1);
r3=rand(row,col-1);
nx=x+(r.*(bx-wx))+(r1.*(wx-x)+r2.*(bx-x))*0.5-r3.*(bx-(x));
end
function [nx,violations]=trim(nx,L,U)
[~,D]=size(nx);
violations=0;
for i = 1:D
indexes = (nx(:,i) < L(i));
violations = violations + abs(sum(L(i)-nx(indexes,i)));
if isempty(indexes)==0
nx(indexes,i) = L(i);
end
indexes = (nx(:,i) > U(i));
violations = violations + abs(sum(nx(indexes,i)-U(i)));
if isempty(indexes)==0
nx(indexes,i) = U(i);
end
end
end
%%%%Code Ends %%%%

C.5 MTPG_ single_obj.m: MTPG-Jaya Algorithm Program

This function file needs to be called in the ‘All_Algorithms_Main.m’ program to

execute the MTPG-Jaya algorithm. This program is only for demonstration purpose.
The numbers assigned to the population size, generations, design variables, and
maximum function evaluations in this program need not be taken as default values.
%%%%Code Begins %%%%
%%%% MTPG_single_obj.m%%%%
function [xf]=MTPG_single_obj(ObjectiveFunction)
global L
366 Appendices

global U
global P
global D
global MFEs
global Fes
global initial_flag
global Ter_Error
global convergence
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
N_Teams = 4; %4
NE=5; %5
Pacc=0.35; %0.1
MaxStagnation = 5; % The maximum iterations without improve-
ment to consider not stagnated
Stagnation=0;
Command_Center_Iter = 10;
x=repmat(L,P,1)+rand(P,D).*repmat((U-L),P,1);
initial_flag=0;
f=ObjectiveFunction(x);
NFE = 0;
sol_current = x ;
FV_current= f ;

%% Update Global Best

[FV_Global, index] = min(FV_current);
sol_Global = sol_current (index,:); % Global Best
Fes = NFE;
FV_Global_Previous=inf;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% MAIN LOOP
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Teams_sols = cell(1, N_Teams); % matrix of solutions
Teams_FV = Inf * ones(P, N_Teams); % matrix of objective func-
tion values
Teams_Rank = Teams_FV; % matrix of objective function values
Teams_viols = zeros(1, N_Teams);
FirmWare=randi(NE,[1,N_Teams]);
% FirmWare=1;
r1=rand;
r2=rand;
itr=0;
counter =1;
while NFE<MFEs
itr=itr+1;
sol_iteration = sol_current; % Local Best
FV_iteration = FV_current;
%% Find New solutions
violations = zeros(1, N_Teams);
for team=1:N_Teams
f=[sol_iteration,FV_iteration];
if FirmWare(team)==1
[nx]=updatepop(f);
elseif FirmWare(team)==2
[nx,r1]=cjupdate(f,r1);
Appendices 367

elseif FirmWare(team)==3
[nx]=randupdate(sol_iteration);
elseif FirmWare(team)==4
[nx,r2]=chaosupdate(sol_iteration,r2);
elseif FirmWare(team)==5
nx=quasi(L,U,sol_iteration);
elseif FirmWare(team)==6
[nx]=updatepopjh(f);
end
nx,violations(team)]=boundcorrection(nx,L,U,violations(team));
Teams_sols{team} = nx;
initial_flag=0;
Teams_FV(:,team) = ObjectiveFunction(Teams_sols{team});
NFE = NFE + P;
end
%% %% COMMAND CENTER
for i = 1:P
[sorted, indices] = sort(Teams_FV(i, :)); % Sort the row posi-
tion idx
FV_iteration(i) = sorted(1); % Get the minimum value
[~,c]=size(Teams_sols{indices(1)}(i, :));
if c<D
disp(’error’)
end
sol_iteration (i,:) = Teams_sols{indices(1)}(i, :);
Teams_Rank(i, :) = indices;
end
Teams_viols = Teams_viols + mean(Teams_Rank, 1) + viola-
tions; % average rank of each team
improved = FV_iteration < FV_current;
%% ACCEPT WORSE SOLUTIONS IF SEARCH STAGNATED ? Pacc% OF CHANCE
if (Stagnation > MaxStagnation-1)
% fprintf(’ Stagnated!!! Selecting non-improved solu-
tions.\n’);
indexes = find( (improved + (rand(P, 1) < Pacc)) > 0);
Stagnation = 0;
else
indexes = find( improved );
end;
%% Update Current Solution with iteration solution
FV_current(indexes) = FV_iteration(indexes);
sol_current(indexes,:) = sol_iteration (indexes,:);
%% Update Global Best
[FV_Global_New,index] = min(FV_current);
if FV_Global_New < FV_Global
FV_Global=FV_Global_New;
sol_Global=sol_current (index,:);
Fes=NFE;% NFE_FinalBest = NFE;
end
%% % TIME TO UPDATE THE FIRMWARE
if (rem (itr, Command_Center_Iter) == 0)
[~, index]=sort(Teams_viols);
worst_FW_Num=index(N_Teams);
worst_FW=FirmWare(worst_FW_Num);
368 Appendices

% best_FW_Num=index(1); % best_FW=FirmWare(best_FW_Num); % h=1;

h=randi(NE);
while h==worst_FW
h=randi(NE);
end
FirmWare(index(N_Teams))=h; % FirmWare(index(N_Teams))=best_FW;
Teams_viols = zeros(1, N_Teams);
end %% Command_Center_Iters
%% Stagnation Check
if ( abs(FV_Global - FV_Global_Previous) / abs(FV_Global_Previous)
< 1e-6 ) %% Needs better improvement
Stagnation = Stagnation + 1;
else
Stagnation = 0;
end
%% update previous global by current global solution
FV_Global_Previous=FV_Global;
xf=[sol_Global,FV_Global];
if FV_Global<=Ter_Error
Fes=NFE;
break
end
convergence(counter,1)=NFE;
convergence(counter,2)=FV_Global;
counter =counter+1;
end
end
function [nx,violations]=boundcorrection(nx,LB,UB,violations)
[~,D]=size(nx);
for i = 1:D
indexes = find(nx(:,i) < LB(i));
violations = violations + abs(sum(LB(i)-nx(indexes,i)));
if isempty(indexes)==0
nx(indexes,i) = LB(i);
end
indexes = find(nx(:,i) > UB(i));
violations = violations + abs(sum(nx(indexes,i)-UB(i)));
if isempty(indexes)==0
nx(indexes,i) = UB(i);
end
end
end
function [nx]=updatepopjh(f)
[pop,NT]=size(f);
x=f(:,1:NT-1);
[sorted, ~]=sortrows(f,NT);
b1=sorted(1,1:NT-1);
b2=sorted(2,1:NT-1);
b3=sorted(3,1:NT-1);
b=[b1; b2; b3];
w1=sorted(pop,1:NT-1);
w=repmat(w1,pop,1);
nx=zeros(pop,NT-1);
for j=1:NT-1
Appendices 369

rb=randi(3,[pop,1]);
r=rand(pop,1);
nx(:,j)=x(:,j)+r.*(b(rb,j)-abs(x(:,j)))-r.*(w(:,j)-
abs(x(:,j)));
end
end
function [nx]=randupdate(f)
[pop,NT]=size(f);
x=f(:,1:NT);
nx=zeros(pop,NT);
r1=rand;
for i=1:pop
for j=1:NT
r1=4*r1*(1-r1);
nx(i,j)=x(i,j)+rand*(2*r1-1);
end
end
end
function xn=quasi(umin,umax,x)
[pop,var]=size(x);
a1=(umin+umax)/2;
b1=repmat(umin+umax,pop,1);
a=repmat(a1,pop,1);
b=b1-x;
c=rand(pop,var);
xn= a+(b-a).*c;
end
function [nx,r1]=cjupdate(f,r1)
[pop,NT]=size(f);
x=f(:,1:NT-1);
[~, ind]=min(f(:,NT));
xb=x(ind,:);
bx=repmat(xb,pop,1);
[~, ind]=max(f(:,NT));
xw=x(ind,:);
wx=repmat(xw,pop,1);
r=zeros(pop,NT-1);
for i=1:pop
for j=1:NT-1
r(i,j)=4*r1*(1-r1);
r1=r(i,j);
end
end
nx=x+r.*(bx-abs(x))-r.*(wx-abs(x));
end
function [nx,r1]=chaosupdate(f,r1)
[pop,NT]=size(f);
x=f(:,1:NT);
nx=zeros(pop,NT);
for i=1:pop
for j=1:NT
r1=4*r1*(1-r1);
nx(i,j)=x(i,j)+rand*(2*r1-1);
end
370 Appendices

end
end
function [nx]=updatepop(f)
[pop,NT]=size(f);
x=f(:,1:NT-1);
[~, ind]=min(f(:,NT));
xb=x(ind,:);
bx=repmat(xb,pop,1);
[~, ind]=max(f(:,NT));
xw=x(ind,:);
wx=repmat(xw,pop,1);
nx=x+rand(pop,NT-1).*(bx-abs(x))-rand(pop,NT-1).*(wx-abs(x));
end
%%%%Code Ends %%%%

C.6 AMTPG_single_obj.m: MTPG-Jaya Algorithm Program

This function file needs to be called in the ‘All_Algorithms_Main.m’ program

to execute the AMTPG-Jaya algorithm. This program is only for demonstration
purpose. The numbers assigned to the population size, generations, design variables,
and maximum function evaluations in this program need not be taken as default
values.
%%%%Code Begins %%%%
%%%% AMTPG_single_obj.m%%%%
function [xf]=AMTPG_single_obj(ObjectiveFunction)
global L
global U
global P
global D
global MFEs
global Fes
global initial_flag
global Ter_Error
global convergence
global FirmWare
global NE
global N_Teams
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
N_Teams = 4; %4
NE=5; %5
Pacc=0.35; %0.1
MaxStagnation = 5; % The maximum iterations without improve-
ment to consider not stagnated
Stagnation=0;
Command_Center_Iter = 10;
x=repmat(L,P,1)+rand(P,D).*repmat((U-L),P,1);
initial_flag=0;
f=ObjectiveFunction(x);
Appendices 371

NFE = 0;
sol_current = x;
FV_current= f;

%% Update Global Best

[FV_Global, index] = min(FV_current);
sol_Global = sol_current (index,:); % Global Best
Fes = NFE;
FV_Global_Previous=inf;
FV_Global_Old=FV_Global;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% MAIN LOOP
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Teams_sols = cell(1, N_Teams); % matrix of solutions
Teams_FV = Inf * ones(P, N_Teams); % matrix of objective func-
tion values
Teams_Rank = Teams_FV; % matrix of objective function values
Teams_viols = zeros(1, N_Teams);
FirmWare=randi(NE,[1,N_Teams]);
% FirmWare=1;
r1=rand;
r2=rand;
itr=0;
counter =1;
while NFE<MFEs
itr=itr+1;
sol_iteration = sol_current; % Local Best
FV_iteration = FV_current;
%% Find New solutions
violations = zeros(1, N_Teams);
for team=1:N_Teams
f=[sol_iteration,FV_iteration];
if FirmWare(team)==1
[nx]=updatepop(f);
elseif FirmWare(team)==2
[nx,r1]=cjupdate(f,r1);
elseif FirmWare(team)==3
[nx]=randupdate(sol_iteration);
elseif FirmWare(team)==4
[nx,r2]=chaosupdate(sol_iteration,r2);
elseif FirmWare(team)==5
nx=quasi(L,U,sol_iteration);
elseif FirmWare(team)==6
[nx]=updatepopjh(f);
end
[nx,violations(team)]=boundcorrection(nx,L,U,violations(team));
Teams_sols{team} = nx;
initial_flag=0;
Teams_FV(:,team) = ObjectiveFunction(Teams_sols{team});
NFE = NFE + P;
end
%% %% COMMAND CENTER
for i = 1:P
[sorted, indices] = sort(Teams_FV(i, :)); % Sort the row posi-
tion idx
372 Appendices

FV_iteration(i) = sorted(1); % Get the minimum value

sol_iteration (i,:) = Teams_sols{indices(1)}(i, :);
Teams_Rank(i, :) = indices;
end
Teams_viols = Teams_viols + mean(Teams_Rank, 1) + viola-
tions; % average rank of each team
improved = FV_iteration < FV_current;
%% ACCEPT WORSE SOLUTIONS IF SEARCH STAGNATED ? Pacc% OF CHANCE
if (Stagnation > MaxStagnation-1)
% fprintf(’ Stagnated!!! Selecting non-improved solu-
tions.\n’);
indexes = find( (improved + (rand(P, 1) < Pacc)) > 0);
Stagnation = 0;
else
indexes = find( improved );
end;
%% Update Current Solution with iteration solution
FV_current(indexes) = FV_iteration(indexes);
sol_current(indexes,:) = sol_iteration (indexes,:);
%% Update Global Best
[FV_Global_New,index] = min(FV_current);
if FV_Global_New < FV_Global
FV_Global=FV_Global_New;
sol_Global=sol_current (index,:);
Fes=NFE;% NFE_FinalBest = NFE;
end
%% % TIME TO UPDATE THE FIRMWARE
if (rem (itr, Command_Center_Iter) == 0)
[Teams_viols,Teams_sols,Teams_FV,Teams_Rank ]=command-
centre(Teams_viols,FV_Global_Old,FV_Global);
FV_Global_Old=FV_Global;
end %% Command_Center_Iters
%% Stagnation Check
if ( abs(FV_Global - FV_Global_Previous) / abs(FV_Global_Previous)
< 1e-6 ) %% Needs better improvement
Stagnation = Stagnation + 1;
else
Stagnation = 0;
end
%% update previous global by current global solution
FV_Global_Previous=FV_Global;
xf=[sol_Global,FV_Global];
if FV_Global<=Ter_Error
Fes=NFE;
break
end
convergence(counter,1)=NFE;
convergence(counter,2)=FV_Global;
counter =counter+1;
end
end
function [Teams_viols,Teams_sols,Teams_FV,Teams_Rank ]=command-
centre(Teams_viols,FV_Global_Old,FV_Global_New)
%N_Teams
Appendices 373

%FirmWare
%Teams_viols,FV_Global_Old,CViol_Global_Old,
FV_Global_New,CViol_Global_New
global P
global FirmWare
global NE
global N_Teams
global Max_N_Teams
global Min_N_Teams
% global NFE
Old_N_Teams=N_Teams;
[~, index]=sort(Teams_viols);
worst_FW_Num=index(N_Teams);
worst_FW=FirmWare(worst_FW_Num);
best_FW_Num=index(1);
best_FW=FirmWare(best_FW_Num);
if NE==1
h=1;
else
h=randi(NE);
while h==worst_FW
h=randi(NE);
end
end
FirmWare(index(N_Teams))=h;% number of teams not changed
% if NFE>4e5
% N_Teams=2;
% else
if FV_Global_New >= FV_Global_Old
if N_Teams<Max_N_Teams
N_Teams=N_Teams+1;
end
else
if N_Teams>Min_N_Teams
N_Teams=N_Teams-1;
end
end
% end
if Old_N_Teams<N_Teams %% number of teams increased
FirmWare(N_Teams)=best_FW;
elseif Old_N_Teams>N_Teams %% number of teams decreased
FirmWare(worst_FW_Num)=[];
end
Teams_viols = zeros(1, N_Teams);
Teams_sols = cell(1, N_Teams); % matrix of solutions
Teams_FV = Inf * ones(P, N_Teams); % matrix of objective func-
tion values
Teams_Rank = Teams_FV; % matrix of objective function values
end
function [nx,violations]=boundcorrection(nx,LB,UB,violations)
[~,D]=size(nx);
for i = 1:D
indexes = find(nx(:,i) < LB(i));
violations = violations + abs(sum(LB(i)-nx(indexes,i)));
374 Appendices

if isempty(indexes)==0
nx(indexes,i) = LB(i);
end
indexes = find(nx(:,i) > UB(i));
violations = violations + abs(sum(nx(indexes,i)-UB(i)));
if isempty(indexes)==0
nx(indexes,i) = UB(i);
end
end
end
function [nx]=updatepopjh(f)
[pop,NT]=size(f);
x=f(:,1:NT-1);
[sorted, ~]=sortrows(f,NT);
b1=sorted(1,1:NT-1);
b2=sorted(2,1:NT-1);
b3=sorted(3,1:NT-1);
b=[b1; b2; b3];
w1=sorted(pop,1:NT-1);
w=repmat(w1,pop,1);
nx=zeros(pop,NT-1);
for j=1:NT-1
rb=randi(3,[pop,1]);
r=rand(pop,1);
nx(:,j)=x(:,j)+r.*(b(rb,j)-abs(x(:,j)))-r.*(w(:,j)-
abs(x(:,j)));
end
end
function [nx]=randupdate(f)
[pop,NT]=size(f);
x=f(:,1:NT);
nx=zeros(pop,NT);
r1=rand;
for i=1:pop
for j=1:NT
r1=4*r1*(1-r1);
nx(i,j)=x(i,j)+rand*(2*r1-1);
end
end
end
function xn=quasi(umin,umax,x)
[pop,var]=size(x);
a1=(umin+umax)/2;
b1=repmat(umin+umax,pop,1);
a=repmat(a1,pop,1);
b=b1-x;
c=rand(pop,var);
xn= a+(b-a).*c;
end
function [nx,r1]=cjupdate(f,r1)
[pop,NT]=size(f);
x=f(:,1:NT-1);
[~, ind]=min(f(:,NT));
xb=x(ind,:);
Appendices 375

bx=repmat(xb,pop,1);
[~, ind]=max(f(:,NT));
xw=x(ind,:);
wx=repmat(xw,pop,1);
r=zeros(pop,NT-1);
for i=1:pop
for j=1:NT-1
r(i,j)=4*r1*(1-r1);
r1=r(i,j);
end
end
nx=x+r.*(bx-abs(x))-r.*(wx-abs(x));
end
function [nx,r1]=chaosupdate(f,r1)
[pop,NT]=size(f);
x=f(:,1:NT);
nx=zeros(pop,NT);
for i=1:pop
for j=1:NT
r1=4*r1*(1-r1);
nx(i,j)=x(i,j)+rand*(2*r1-1);
end
end
end
function [nx]=updatepop(f)
[pop,NT]=size(f);
x=f(:,1:NT-1);
[~, ind]=min(f(:,NT));
xb=x(ind,:);
bx=repmat(xb,pop,1);
[~, ind]=max(f(:,NT));
xw=x(ind,:);
wx=repmat(xw,pop,1);
nx=x+rand(pop,NT-1).*(bx-abs(x))-rand(pop,NT-1).*(wx-abs(x));
% nx=round(nx);
end
%%%%Code Ends %%%%

C.7 Unconstrained_benchmark_func.m: 30 Unconstrained

Benchmark Function Program

This function file needs to be called in the ‘All_Algorithms_Main.m’ program to

execute any algorithm. This program is only for demonstration purpose. The numbers
assigned in this program need not be taken as default values.
%%%%Code Begins %%%%
%%%% Unconstrained_benchmark_func.m%%%%
function f=Unconstrained_benchmark_func(x)
global func_num
376 Appendices

global initial_flag
persistent fhd
% benchmark_func.m is the main function for 25 test func-
tions, all minimize
% problems
% e.g. f=benchmark_func(x,func_num)
% x is the variable, f is the function value
% func_num is the function num,

% 30 TEST FUCNTIONS
% Problem No. Problem Name Boundaries Dimension
% 1. sphere Bounds[-100,100] D=30
% 2. SumSquares (sum2) Bounds[-10,10] D=30
% 3. Beale Bounds[-4.5,4.5] D=2,5
% 4. Easom Bounds[-100,100] D=2
% 5. Matyas Bounds[-10,10] D=2
% 6. Colville Bounds[-10,10] D=4
% 7. Trid6 Bounds[-6,6] D=6
% 8. Trid10 Bounds[-10,10] D=10
% 9. Zakharov (zakh) Bounds[-5,10] D=10
% 10. Schwefel 1.2(schw12) Bounds[-100,100] D=30
% 11. Rosenbrock (rosen) Bounds[-30,30] D=30
% 12. Dixon-Price (dp) Bounds[-10,10] D=30
% 13. Foxholes Bounds[-65.536,65.536] D=2
% 14. Branin Bounds[0,15] D=2
% 15. Bohachevsky 1(bh1) Bounds[-100,100] D=2
% 16. Booth Bounds[-10,10] D=2
% 17. Michalewicz 2(mich2)Bounds[0,pi] D=2
% 18. Michalewicz 5(mich5)Bounds[0,pi] D=5
% 19. Bohachevsky 2(bh2) Bounds[-100,100] D=2
% 20. Bohachevsky 3(bh3) Bounds[-100,100] D=2
% 21. GoldStein-Price(gold)Bounds[-2,2] D=2
% 22. Perm Bounds[-4,4] D=4
% 23. Hartman 3 Bounds[0,1] D=3
% 24. Ackley Bounds[-32,32] D=30
% 25. Penalized 2 Bounds[-50,50] D=30
% 26. Langarman 2 Bounds[0,10] D=2
% 27. Langarman 5 Bounds[0,10] D=5
% 28. Langarman 10 Bounds[0,10] D=10
% 29. Fletcher Powell 5 Bounds[-pi,pi] D=5
% 30. Fletcher Powell 1 Bounds[-pi,pi] D=10

if initial_flag==0
if func_num==1; fhd=str2func(’sphere’);
elseif func_num==2; fhd=str2func(’sum2’);
elseif func_num==3; fhd=str2func(’beale’);
elseif func_num==4; fhd=str2func(’easom’);
elseif func_num==5; fhd=str2func(’matyas’);
elseif func_num==6; fhd=str2func(’colville’);
elseif func_num==7; fhd=str2func(’trid6’);
elseif func_num==8; fhd=str2func(’trid10’);
elseif func_num==9; fhd=str2func(’zakh’);
elseif func_num==10; fhd=str2func(’schw12’);
elseif func_num==11; fhd=str2func(’rosen’);
elseif func_num==12; fhd=str2func(’dp’);
Appendices 377

elseif func_num==13; fhd=str2func(’Foxholes’);

elseif func_num==14; fhd=str2func(’branin’);
elseif func_num==15; fhd=str2func(’bh1’);
elseif func_num==16; fhd=str2func(’booth’);
elseif func_num==17; fhd=str2func(’mich2’);
elseif func_num==18; fhd=str2func(’mich5’);
elseif func_num==19; fhd=str2func(’bh2’);
elseif func_num==20; fhd=str2func(’bh3’);
elseif func_num==21; fhd=str2func(’gold’);
elseif func_num==22; fhd=str2func(’perm’);
elseif func_num==23; fhd=str2func(’hartman3’);
elseif func_num==24; fhd=str2func(’ackley’);
elseif func_num==25; fhd=str2func(’Penalized2’);
elseif func_num==26; fhd=str2func(’Langarman2’);
elseif func_num==27; fhd=str2func(’Langarman5’);
elseif func_num==28; fhd=str2func(’Langarman10’);
elseif func_num==29; fhd=str2func(’FletcherPowell5’);
elseif func_num==30; fhd=str2func(’FletcherPowell10’);
end
end
[row,~]=size(x);
f=nan(row,1);
for i=1:row
f(i)=feval(fhd,x(i,:));
end
function [f]=FletcherPowell10(x)
% Bounds[-pi,pi] D=10
% FletcherPowell-10 function.
%
a=[-79 56 -62 -9 92 48 -22 -34 -39 -40;91 -9 -18 -59 99 -45 88 -14
-29 26;-38 8 -12 -73 40 26 -64 29 -82 -32;-78 -18 -49 65 66 -40 88
-95 -57 10;-1 -43 93 -18 -76 -68 -42 22 46 -14;34 -96 26 -56 -36 -85
-62 13 93 78;52 -46 -69 99 -47 -72 -11 55 -55 91;81 47 35 55 67 -13 33
14 83 -42;-50 66 -47 -75 89 -16 82 6 -85 -62];
b=[-65 -11 76 78 30 93 -86 -99 -37 52;59 67 49 -45 52 -33 -34 29
-39 -80;21 -23 -80 86 86 -30 39 -73 -91 5;-91 -75 20 -64 -15 17 -89 36
-49 -2;-79 99 -31 -8 -67 -72 -43 -55 76 -57;-89 -35 -55 75 15 -6 -53
-56 -96 87;-76 45 74 12 -12 -69 2 71 75 -60;-50 -88 93 68 10 -13 84
-21 65 14;-23 -95 99 62 -37 96 27 69 -64 -92;-5 -57 -30 -6 -96 75 25
-6 96 77];
alpha=[-2.7910 2.5623 -1.0429 0.5097 -2.8096 1.1883 2.0771 -2.9926
0.0715 0.4142];
xx=x;
s1=0;
for i=1:2
sA=0;
sB=0;
for j=1:2
sA=sA+(a(i,j)*sin(alpha(j)))+(b(i,j)*cos(alpha(j)));
sB=sB+(a(i,j)*sin(xx(j)))+(b(i,j)*cos(xx(j)));
end
A(i)=sA;
B(i)=sB;
s1=s1+(A(i)-B(i))ˆ2;
378 Appendices

end
f=s1;

function [f]=FletcherPowell5(x)
% Bounds[-pi,pi] D=5
% FletcherPowell-5 function.
%
a=[-79 56 -62 -9 92;91 -9 -18 -59 99;-38 8 -12 -73 40;-78 -18 -49
65 66;-1 -43 93 -18 -76];
b=[-65 -11 76 78 30;59 67 49 -45 52;21 -23 -80 86 86;-91 -75 20 -64
-15;-79 99 -31 -8 -67];
alpha=[-2.7910 2.5623 -1.0429 0.5097 -2.8096];
xx=x;
s1=0;
for i=1:2
sA=0;
sB=0;
for j=1:2
sA=sA+(a(i,j)*sin(alpha(j)))+(b(i,j)*cos(alpha(j)));
sB=sB+(a(i,j)*sin(xx(j)))+(b(i,j)*cos(xx(j)));
end
A(i)=sA;
B(i)=sB;
s1=s1+(A(i)-B(i))ˆ2;
end
f=s1;

function [f]=Langarman10(x)
% Bounds[0,10] D=10
% Langarman10 function.
%
d = size(x,2);
m=10;
a=[9.681 0.667 4.783 9.095 3.517 9.325 6.544 0.211 5.122 2.020;9.4
2.041 3.788 7.931 2.882 2.672 3.568 1.284 7.033 7.374;8.025 9.152
5.114 7.621 4.564 4.711 2.996 6.126 0.734 4.982;2.196 0.415 5.649
6.979 9.510
9.166 6.304 6.054 9.377 1.426;8.074 8.777 3.467 1.863 6.708 6.349
4.534
0.276 7.633 1.567;7.650 5.658 0.720 2.764 3.278 5.283 7.474 6.274
1.409
8.208;1.256 3.605 8.623 6.905 0.584 8.133 6.071 6.888 4.187
5.448;8.314
2.261 4.224 1.781 4.124 0.932 8.129 8.658 1.208 5.762;0.226 8.858
1.420
0.945 1.622 4.698 6.228 9.096 0.972 7.637;7.305 2.228 1.242 5.928
9.133
1.826 4.060 5.204 8.713 8.247];
C=[0.806 0.517 1.5 0.908 0.965 0.669 0.524 0.902 0.531 0.876];
xx=x;
s2=0;
for i=1:m
s1=0;
for j=1:d
s1=s1+(xx(j)-a(i,j))ˆ2;
Appendices 379

end
s2=s2+C(i)*(exp((-1/pi)*s1))*(cos(pi*s1));
end
f=-s2;

function [f]=Langarman5(x)
% Bounds[0,10] D=5
% Langarman5 function.
%
d = size(x,2);
m=5;
a=[9.681 0.667 4.783 9.095 3.517 9.325 6.544 0.211 5.122 2.020;9.4
2.041 3.788 7.931 2.882 2.672 3.568 1.284 7.033 7.374;8.025 9.152
5.114
7.621 4.564 4.711 2.996 6.126 0.734 4.982;2.196 0.415 5.649 6.979
9.510
9.166 6.304 6.054 9.377 1.426;8.074 8.777 3.467 1.863 6.708 6.349
4.534
0.276 7.633 1.567;7.650 5.658 0.720 2.764 3.278 5.283 7.474 6.274
1.409
8.208;1.256 3.605 8.623 6.905 0.584 8.133 6.071 6.888 4.187
5.448;8.314
2.261 4.224 1.781 4.124 0.932 8.129 8.658 1.208 5.762;0.226 8.858
1.420
0.945 1.622 4.698 6.228 9.096 0.972 7.637;7.305 2.228 1.242 5.928
9.133
1.826 4.060 5.204 8.713 8.247];
C=[0.806 0.517 1.5 0.908 0.965 0.669 0.524 0.902 0.531 0.876];
xx=x;
s2=0;
for i=1:m
s1=0;
for j=1:d
s1=s1+(xx(j)-a(i,j))ˆ2;
end
s2=s2+C(i)*(exp((-1/pi)*s1))*(cos(pi*s1));
end
f=-s2;

function [f]=Langarman2(x)
% Bounds[0,10] D=2
% Langarman2 function.
%
d = size(x,2);
m=2;
a=[9.681 0.667 4.783 9.095 3.517 9.325 6.544 0.211 5.122 2.020;9.4
2.041 3.788 7.931 2.882 2.672 3.568 1.284 7.033 7.374;8.025 9.152
5.114
7.621 4.564 4.711 2.996 6.126 0.734 4.982;2.196 0.415 5.649 6.979
9.510
9.166 6.304 6.054 9.377 1.426;8.074 8.777 3.467 1.863 6.708 6.349
4.534
0.276 7.633 1.567;7.650 5.658 0.720 2.764 3.278 5.283 7.474 6.274
1.409
380 Appendices

8.208;1.256 3.605 8.623 6.905 0.584 8.133 6.071 6.888 4.187

5.448;8.314
2.261 4.224 1.781 4.124 0.932 8.129 8.658 1.208 5.762;0.226 8.858
1.420
0.945 1.622 4.698 6.228 9.096 0.972 7.637;7.305 2.228 1.242 5.928
9.133
1.826 4.060 5.204 8.713 8.247];
C=[0.806 0.517 1.5 0.908 0.965 0.669 0.524 0.902 0.531 0.876];
xx=x;
s2=0;
for i=1:m
s1=0;
for j=1:d
s1=s1+(xx(j)-a(i,j))ˆ2;
end
s2=s2+C(i)*(exp((-1/pi)*s1))*(cos(pi*s1));
end
f=-s2;

function [f]=Penalized2(x)
% 30 [-50, 50]
% Penalized2 function.
%
c = size(x,2);
a=5;k=100;m=4;
u=[];
xx=x;
for i=1:c
if(xx(i)>a)
u(i)=k*((xx(i)-a)ˆm);
end
if(xx(i)>=-a && xx(i)<=a)
u(i)=0;
end
if(xx(i)<-a)
u(i)=k*(-xx(i)-a)ˆm;
end
end
s1=0;
for i=1:c-1
s1=s1+((xx(i)-1)ˆ2)*(1+(sin(3*pi*xx(i+1)))ˆ2)+((xx(c)-
1)ˆ2)*(1+(sin(2*pi*xx(c)))ˆ2);
end
f=0.1*(((sin(pi*xx(1)))ˆ2)+s1)+sum(u);

function y = ackley(x)
% 30 [-32, 32]
% Ackley function.
%
n = size(x,2);
a = 20; b = 0.2; c = 2*pi;
s1 = 0; s2 = 0;
for i=1:n;
s1 = s1+x(i)ˆ2;
s2 = s2+cos(c*x(i));
Appendices 381

end
y = -a*exp(-b*sqrt((1/n)*s1))-exp((1/n)*s2)+a+exp(1);

function [f]=hartman3(x)
% 3 [0, 1]
% hartman3 function
c = size(x,2);
C=[1.0 1.2 3.0 3.2];
P=(10ˆ-4)*[3689 1170 2673;4699 4387 7470;1091 8732 5547;381 5743
8828];
a=[3 10 30;0.1 10 35;3.0 10 30;0.1 10 35];
s1=0;
for i=1:4
s2=0;
for j=1:c
s2=s2+a(i,j)*(x(j)-P(i,j))ˆ2;
end
s1=s1+C(i)*exp(-s2);
end
f=-s1;

function y = perm(x)
% 4 [-D, D]
% Perm function
%
n = 2;
b = 10;
s_out = 0;
for k = 1:n;
s_in = 0;
for j = 1:n
s_in = s_in+(j+b)*(x(j)ˆk-(1/j)ˆk);
end
s_out = s_out+s_inˆ2;
end
y = s_out;

function y = gold(x)
% 2 [-2, 2]
% Goldstein and Price function
%
a = 1+ ((x(1)+x(2)+1)ˆ2) * (19-14*x(1)+3*x(1)ˆ2 - 14*x(2) +
6*x(1)*x(2) + 3*x(2)ˆ2);
b = 30+ (2*x(1)-3*x(2))ˆ2 * (18 - 32*x(1) + 12*x(1)ˆ2 + 48 * x(2) -
36*x(1)*x(2) + 27*x(2)ˆ2);
y = a*b;

function y = bh3(x)
% 2 [-100 100]
% Bohachecsky function 3
%
y = x(1)ˆ2+2*x(2)ˆ2-0.3*cos(3*pi*x(1)+4*pi*x(2))+0.3;

function y = bh2(x)
% 2 [-100 100]
382 Appendices

% Bohachecsky function 2
%
% z(i)=(x1ˆ2)+(2*x2ˆ2)-0.3*cos((3*pi*x1)*(4*pi*x2))+0.3;
y = x(1)ˆ2+2*x(2)ˆ2-0.3*cos(3*pi*x(1))*cos(4*pi*x(2))+0.3;

% function y = mich10(x)
% % 10 [0, pi]
% % Michalewicz function
% %
% n = size(x,2);
% m = 10;
% s = 0;
% for i = 1:n;
% s = s+sin(x(i))*(sin(i*x(i)ˆ2/pi))ˆ(2*m);
% end
% y = -s;

function y = mich5(x)
% 5 [0, pi]
% Michalewicz function
%
n = size(x,2);
% m = 10;
s = 0;
for i = 1:n;
% s = s+sin(x(i))*(sin(i*x(i)ˆ2/pi))ˆ(2*m);
s = s+sin(x(i))*(sin(i*x(i)ˆ2/pi))ˆ20;
end
y = -s;

function y = mich2(x)
% 2 [0, pi]
% Michalewicz function
%
n = size(x,2);
m = 10;
s = 0;
for i = 1:n;
s = s+sin(x(i))*(sin(i*x(i)ˆ2/pi))ˆ(2*m);
end
y = -s;

function y = booth(x)
% 2 [-10, 10]
% Booth function
%
y = (x(1)+2*x(2)-7)ˆ2+(2*x(1)+x(2)-5)ˆ2;

function y=bh1(x)
% 2 [-100 100]
% Bohachecsky function 1
%
y=x(1)ˆ2+2*x(2)ˆ2-0.3*cos(3*pi*x(1))-0.4*cos(4*pi*x(2))+0.7;

function y = branin(x)
Appendices 383

% 2 [0, 15]
% Branin function
%
% x1=x(1);
% x2=x(2);
% z=(x2-(5.1/(4*piˆ2))*(x1ˆ2)+(5/pi)*x1-6)ˆ2+10*(1-
(1/(8*pi)))*cos(x1)+10;
y = (x(2)-(5.1/(4*piˆ2))*x(1)ˆ2+5*x(1)/pi-6)ˆ2+10*(1-
1/(8*pi))*cos(x(1))+10;

function [f]=Foxholes(x)
% 2 [-65.536, 65.536]
% Foxholes function.
%
a=[-32 -32;-16 -32;0 -32;16 -32;32 -32;-32 -16;-16 -16;0 -16;16
-16;32 -16;-32 0;-16 0;0 0;16 0;32 0;-32 16;-16 16;0 16;16 16;32
16;-32 32;-16 32;0 32;16 32;32 32];
s2=0;
for j=1:25
s1=0;
for i=1:2
s1=s1+(x(i)-a(j,i))ˆ6;
end
s2=s2+(1/(j+s1));
end
f=((1/500)+s2)ˆ-1;

function y = dp(x)
% 30 [-10, 10]
% Dixon and Price function.
%
n =size(x,2);
s1 = 0;
for j = 2:n;
% s1 = s1+j*( (2 * ( x(j)ˆ2 ) - x(j-1) )ˆ2 );
s1=s1+j*(2*x(j)ˆ2-x(j-1))ˆ2;

end
y = s1+((x(1)-1)ˆ2);

function y = rosen(x)
% 30 [-30, 30] UN
% Rosenbrock function
%
n =size(x,2);
sum = 0;
for j = 1:n-1;
sum = sum+100 * (( x(j)ˆ2 -x(j+1) )ˆ2+( x(j)-1 )ˆ2);
end
y = sum;

function y = schw12(x)
% 30 [-100, 100]
% Schwefel function
%
384 Appendices

n =size(x,2);
sum=0;
for i=1:n
sum2=0;
for j=1:i
sum2=sum2+(x(1,j))ˆ2;
end
sum=sum+(sum2ˆ2);
end
y =sum;

function y = zakh(x)
% 10 [-5, 10]
% Zakharov function
%
n = size(x,2);
s1 = 0;
s2 = 0;
for j = 1:n;
s1 = s1+x(j)ˆ2;
s2 = s2+0.5*j*x(j);
end
y = s1+s2ˆ2+s2ˆ4;

function y = trid10(x)
% 6 [-D2, D2] 10 [-D2, D2]
% Trid function
%
n = size(x,2);
s1 = 0;
s2 = 0;
for j = 1:n;
s1 = s1+(x(j)-1)ˆ2;
end
for j = 2:n;
s2 = s2+x(j)*x(j-1);
end
y = s1-s2;

function y = trid6(x)
% 6 [-D2, D2] 10 [-D2, D2]
% Trid function
%
n = size(x,2);
s1 = 0;
s2 = 0;
for j = 1:n;
s1 = s1+(x(j)-1)ˆ2;
end
for j = 2:n;
s2 = s2+x(j)*x(j-1);
end
y = s1-s2;

function y = colville(x)
Appendices 385

% 4 [-10, 10]
% Colville function
%
x1=x(1);
x2=x(2);
x3=x(3);
x4=x(4);
y=100*(((x1ˆ2)-x2)ˆ2)+((x1-1)ˆ2)+((x3-1)ˆ2)+90*(((x3ˆ2)
-x4)ˆ2)+10.1*(((x2-1)ˆ2)+((x4-1)ˆ2))+19.8*(x2-1)*(x4-1);

function y = matyas(x)
% 2 [-10, 10]
% Matyas function
%
y = 0.26*(x(1)ˆ2+x(2)ˆ2)-0.48*x(1)*x(2);

function y = easom(x)
% 2 [-100, 100]
% Easom function
%
y = -cos(x(1))*cos(x(2))*exp(-(x(1)-pi)ˆ2-(x(2)-pi)ˆ2);

function y = beale(x)
% 2, 5 [-4.5, 4.5]
% Beale function.
%
y = (1.5-x(1)*(1-x(2)))ˆ2+(2.25-x(1)*(1-x(2)ˆ2))ˆ2+(2.625-
x(1)*(1-x(2)ˆ3))ˆ2;

function y = sum2(x)
% 30 [-10, 10]
% Sum Squares function
%
[~, n]=size(x);
s = 0;
for j = 1:n
s=s+j*x(j)ˆ2;
end
y = s;

function y = sphere(x)
% 30 [-100, 100]
% Sphere function
%
[~, n]=size(x);
s = 0;
for j = 1:n
s = s+x(j)ˆ2;
end
y = s;
%%%%Code Ends %%%%
386 Appendices

C.8 DandBounds.m: 30 Unconstrained Benchmark Function

Variable Dimensions and Their Boundary Programs

This function file needs to be called in the ‘All_Algorithms_Main.m’ program to

execute any algorithm. This program is only for demonstration purpose. The numbers
assigned in this program need not be taken as default values.
%%%%Code Begins %%%%
%%%% DandBounds.m%%%%
function [D,L,U,Opt_Val]=DandBounds(func_num)
if func_num==1
% 1. sphere
L=-100;U=100;D=30;Opt_Val=0;
elseif func_num==2
% 2. SumSquares (sum2)
L=-10;U=10;D=30;Opt_Val=0;
elseif func_num==3
% 3. Beale
L=-4.5;U=4.5;D=2;Opt_Val=0;
elseif func_num==4
% 4. Easom
L=-100;U=100;D=2;Opt_Val=-1;
elseif func_num==5
% 5. Matyas
L=-10;U=10;D=2;Opt_Val=0;

elseif func_num==6
% 6. Colville
L=-10;U=10;D=4;Opt_Val=0;

elseif func_num==7
% 7. Trid6
L=-36;U=36;D=6;Opt_Val=-49.9999999999984;

elseif func_num==8
% 8. Trid10
L=-100;U=100;D=10 ;Opt_Val=-210;

elseif func_num==9
% 9. Zakharov (zakh)
L=-5;U=10;D=10;Opt_Val=0;

elseif func_num==10
% 10. Schwefel 1.2(schw12)
L=-100;U=100;D=30;Opt_Val=0;

elseif func_num==11
% 11. Rosenbrock (rosen)
L=-30;U=30;D=30;Opt_Val=0;

elseif func_num==12
% 12. Dixon-Price (dp)
L=-10;U=10;D=30;Opt_Val=0;
Appendices 387

elseif func_num==13
% 13. Foxholes
L=-65.536;U=65.536;D=2;Opt_Val=0.99800383;

elseif func_num==14
% 14. Branin
L=0;U=15;D=2;Opt_Val=0.397887;

elseif func_num==15
% 15. Bohachevsky 1(bh1)
L=-100;U=100;D=2;Opt_Val=0;

elseif func_num==16
% 16. Booth
L=-10;U=10;D=2;Opt_Val=0;

elseif func_num==17
% 17. Michalewicz 2(mich2)
L=0;U=pi;D=2;Opt_Val=-1.801303;

elseif func_num==18
% 18. Michalewicz 5(mich5)
L=0;U=pi;D=5;Opt_Val=-4.6877;

elseif func_num==19
% 19. Bohachevsky 2(bh2)
L=-100;U=100;D=2;Opt_Val=0;

elseif func_num==20
% 20. Bohachevsky 3(bh3)
L=-100;U=100;D=2;Opt_Val=0;

elseif func_num==21
% 21. GoldStein-Price(gold)
L=-2;U=2;D=2;Opt_Val=3;

elseif func_num==22
% 22. Perm
L=-4;U=4;D=4;Opt_Val=0;

elseif func_num==23
% 23. Hartman 3
L=0;U=1;D=3;Opt_Val=-3.862779;

elseif func_num==24
% 24. Ackley
L=-32;U=32;D=30;Opt_Val=0;

elseif func_num==25
% 25. Penalized 2
L=-50;U=50;D=30;Opt_Val=1e-16;

elseif func_num==26
% 26. Langarman 2
L=0;U=10;D=2;Opt_Val=-1.08094;

elseif func_num==27
388 Appendices

% 27. Langarman 5
L=0;U=10;D=5;Opt_Val=-1.49;

elseif func_num==28
% 28. Langarman 10
L=0;U=10;D=10;Opt_Val=-1.0528;

elseif func_num==29
% 29. Fletcher Powell 5
L=-pi;U=pi;D=5;Opt_Val=0;

elseif func_num==30
% 30. Fletcher Powell 1
L=-pi;U=pi;D=2;Opt_Val=0;
end
%%%%Code Ends %%%%
Index

A E
Adaptive multi-team perturbation guiding Elitist Rao algorithm, 71–75, 82, 91, 93,
Jaya, 61 103, 106, 114, 122, 142, 152, 153,
156, 160, 166, 179, 180, 205, 206,
219, 221, 234, 236, 251, 276, 289,
B 316, 330, 332, 337, 358
Benchmark functions, 353, 375, 386
Bioenergy system, 5, 23, 37, 243
G
Geothermal energy system, 4, 5, 29, 41, 43,
C 327, 329
Codes, 352, 353, 355, 358, 362, 365, 370, Gray relational analysis (GRA), 39, 137,
375, 386 138, 163, 164, 166, 168, 169, 178,
Complex proportional assessment 180, 201, 202, 208, 217, 218, 223,
(COPRAS), 136, 138, 163, 164, 224, 231, 232, 237, 238, 245, 246,
166, 168, 169, 177, 178, 180, 271–273, 278, 279, 287, 288, 310,
200–202, 208, 209, 217, 218, 223, 311, 319, 320, 324, 325, 338, 341,
231, 232, 237, 245, 246, 271–273, 342
287, 288, 310, 311, 318, 319, 321, Ground source heat pump-radiant ceiling
324, 325, 338, 340–342 air conditioning system, 11, 29, 329,
Compression ignition biodiesel engine, 26, 331
288
Compromise ranking method (VIKOR),
135, 138, 163, 164, 166, 169, 177,
178, 180, 201, 202, 208, 217, 218, H
223, 224, 231, 232, 237, 245, 246, Hydro energy system, 3
271–273, 287, 288, 308, 310, 311, Hydro power generation, 29, 39, 41, 327,
319, 320, 324, 325, 338, 341, 342 329
Cost model, 13 Hypervolume, 53, 81, 170, 171, 174, 178,
Coverage, 53, 80, 170, 172, 174, 178, 200, 200, 203, 204, 210–212, 214, 219,
203, 204, 210, 212–214, 219, 226, 227, 229, 233, 240, 241, 246,
226–229, 233, 240–242, 246, 271, 313, 322, 326, 327, 339, 340, 343
273, 274, 282, 284, 288, 289, 311,
313, 323, 327
Crowding distance, 66, 68, 70, 71, 79, 80, I
216 Inverted generational distance, 53, 81
© The Editor(s) (if applicable) and The Author(s), under exclusive license 389
to Springer Nature Switzerland AG 2022
V. R. Ravipudi et al., Design Optimization of Renewable Energy Systems Using Advanced
Optimization Algorithms, Green Energy and Technology,
https://doi.org/10.1007/978-3-030-95589-2
390 Index

J R
Jaya algorithm, 40–42, 45, 53–55, 57–59, Rao algorithm, 45, 53, 69–72, 74, 76, 77,
61–63, 66–68, 82, 83, 90, 91, 93, 98, 79, 82, 83, 90, 91, 93, 98, 103, 104,
103, 104, 106, 110, 115, 122, 106, 110, 114, 115, 123, 139,
141–146, 149–151, 159, 160, 166, 141–147, 149–151, 154, 155, 159,
167, 170–172, 179–181, 189, 200, 160, 162, 164, 166, 169–174, 179,
203, 205, 206, 210–214, 216, 219, 180, 191, 200, 203–206, 210–213,
221, 222, 225–228, 234, 235, 216, 219, 221, 224–229, 234–236,
238–242, 249, 251, 252, 254, 271, 240, 241, 243, 249, 251, 268, 271,
273, 275–278, 282, 285, 289, 290, 275, 276, 281, 282, 285, 289, 306,
292, 308, 311, 314–316, 323, 327, 308, 311, 313–316, 320, 322, 323,
329, 330, 332, 335, 337, 340, 343, 327, 329–332, 335, 337, 340, 343,
352, 353, 355, 365, 370 352, 353, 355, 358, 362

M S
Microalgae-based biomass cultivation Self-adaptive population algorithm, 53, 74,
process, 28, 249, 313–323, 326, 327 362
Multi-attribute decision-making methods, Simple additive weighing (SAW), 132, 138,
123, 131 163, 164, 166, 168, 169, 177, 178,
Multi-objective AMTPG-Jaya algorithm, 180, 201, 202, 208, 217, 218, 223,
66 231, 232, 237, 245, 246, 271–273,
Multi-objective elitist Rao algorithm, 74 287, 288, 308, 310, 311, 318, 319,
Multi-objective Jaya algorithm, 66 321, 324, 325, 338, 340–342
Multi-objective optimization, 14, 21, 23, Single-cylinder direct-injection diesel
26, 33, 35–37, 39, 42–44, 53, 66–72, engine, 24, 39, 249, 252, 254, 256,
74, 75, 79–82, 110, 114, 116–120, 258, 260, 262, 264, 266, 268, 270,
122, 123, 131, 138, 159, 160, 179, 273, 274
234, 249, 251, 275, 288, 313, 329, Single-objective optimization, 53, 54, 63,
332, 350 66, 71, 73, 76, 82, 110, 160, 161,
Multi-objective Rao algorithm, 70, 74 179, 205, 219, 220, 233, 234, 251,
Multi-objective SAP-Rao algorithm, 79 276, 289, 329, 330, 345, 352
Multi-team perturbation guiding Jaya Solar-assisted Brayton engine system, 161,
algorithm, 57, 59 162, 167–176
Solar-assisted Carnot-like heat engine
system, 11, 22, 23, 159, 230
N Solar-assisted Stirling engine system, 11,
Nuclear energy system, 6, 7 17–19, 21, 35, 159, 174, 176, 179,
181, 183, 185, 187, 189, 191, 193,
195, 197, 199, 200, 203–216,
O 219–230, 233
Ocean thermal energy system, 4 Solar energy system, 1, 2
Spacing, 53, 80, 81, 114–117, 119, 120,
170, 171, 174, 178, 200, 203, 204,
P 211, 212, 214, 219, 226, 227, 229,
Power generation model, 13 233, 240, 241, 246, 251, 270, 271,
Preference ranking organization method for 273, 281, 284–286, 289, 308,
enrichment evaluations 311–313, 322, 326, 327, 339, 340,
(PROMETHEE), 132, 133, 138, 343
163, 164, 166, 168, 169, 177, 178,
180, 201, 202, 208, 217, 218, 223,
224, 231, 232, 237, 238, 245, 246, T
271–273, 278, 279, 287, 288, 310, Technique for order preference by
311, 319, 324, 325, 338, 341, 342 similarity to ideal solution
Index 391

(TOPSIS), 43, 133, 134, 138, 160, Unimodal and multimodal benchmark
163, 164, 166–170, 174, 175, problems, 99, 104, 106, 346–350
177–180, 199–202, 204–207, 209,
210, 215–218, 223, 231–233,
236–238, 240, 245, 271–273, 287, W
288, 308, 310, 311, 318, 321, 324, Wake model, 11, 35
325, 338, 340–342 Weighted product method (WPM), 132,
Turbocharged DI diesel engine, 25, 273, 138, 163, 164, 166, 168, 169, 177,
277–286, 288 180, 200–202, 208, 209, 217, 218,
223, 231, 232, 245, 246, 271–273,
287, 288, 308, 310, 311, 318, 319,
321, 324, 325, 338, 340, 342
U Wind energy system, 2, 3
Unconstrained benchmark problems, 82, Wind farm layout, 11, 14, 33, 34, 139, 141,
83, 90, 91, 352 149, 151–156