Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Reliability Analysis of Modern Power Systems
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Desineni Subbaram Naidu

Thomas Robertazzi
Diomidis Spinellis
Tony Q. S. Quek
Behzad Razavi
Sarah Spurgeon, Editor-in-Chief
IEEE Press Editorial Board
Piscataway, NJ 08854
445 Hoes Lane
IEEE Press

Ekram Hossain
Brian Johnson

Joydeep Mitra
James Lyke
Hai Li
Jón Atli Benediktsson
Moeness Amin

James Duncan
Adam Drobot
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Reliability Analysis of Modern Power Systems

Indian Institute of Technology (Banaras Hindu University)

Department of Electrical Engineering IT and Cybernetics

Department of Electrical Engineering

University of South-Eastern Norway

P. Sanjeevikumar
Varanasi (UP)
R. K. Saket
Edited by

Porsgrunn
Norway
India
 k

Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Copyright © 2024 by The Institute of Electrical and Electronics Engineers, Inc.
All rights reserved.

Published by John Wiley & Sons, Inc., Hoboken, New Jersey.

Published simultaneously in Canada.

No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any
means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section
107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or
authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222
Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978) 750-4470, or on the web at www.copyright.com.
Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons,
Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008, or online at http//www.wiley.com/
go/permission.

Trademarks Wiley and the Wiley logo are trademarks or registered trademarks of John Wiley & Sons, Inc. and/or its
affiliates in the United States and other countries and may not be used without written permission. All other
trademarks are the property of their respective owners. John Wiley & Sons, Inc. is not associated with any product
or vendor mentioned in this book.

Limit of Liability/Disclaimer of Warranty While the publisher and author have used their best efforts in preparing
this book, they make no representations or warranties with respect to the accuracy or completeness of the contents
of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose.
No warranty may be created or extended by sales representatives or written sales materials. The advice and
strategies contained herein may not be suitable for your situation. You should consult with a professional where
appropriate. Neither the publisher nor author shall be liable for any loss of profit or any other commercial damages,
including but not limited to special, incidental, consequential, or other damages. Further, readers should be aware
that websites listed in this work may have changed or disappeared between when this work was written and when it
k is read. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, k
including but not limited to special, incidental, consequential, or other damages.

For general information on our other products and services or for technical support, please contact our Customer
Care Department within the United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax (317)
572-4002.

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be
available in electronic formats. For more information about Wiley products, visit our web site at www.wiley.com.

Library of Congress Cataloging-in-Publication Data

Names: Saket, R. K., author. | P. Sanjeevikumar, author.

Title: Reliability analysis of modern power systems / R. K. Saket, P. Sanjeevikumar.
Description: [Hoboken] : Wiley, [2024] | Includes index.
Identifiers: LCCN 2024007086 (print) | LCCN 2024007087 (ebook) | ISBN
9781394226740 (hardback) | ISBN 9781394226764 (adobe pdf) | ISBN
9781394226757 (epub)
Subjects: LCSH: Reliability Engineering | Electric Power Systems | Renewable Energy Systems.
Classification: LCC TA169 .S234 2024 (print) | LCC TA169 (ebook) | DDC
620/.00452--dc23/eng/20240315
LC record available at https://lccn.loc.gov/2024007086
LC ebook record available at https://lccn.loc.gov/2024007087

Cover Design: Wiley

Cover Image: © TebNad/iStockGetty Images

Set in 9.5/12.5pt STIXTwoText by Straive, Chennai, India

k
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Dedicated to the Blessed One, the Worthy One, the fully Enlightened One
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
vii

Contents

About the Authors xix

List of Contributors xxi
Foreword xxvii
Preface xxix
Acknowledgments xxxiii

Section 1 Reliability Principles and Applications 1

1 Basic Principles and Scientific Importance of Reliability Theory 3

Aanchal Verma, Akanksha Singh S. Vardhan, Vanitha Bagana, R. K. Saket,
and P. Sanjeevikumar
1.1 Introduction 3
1.2 Basic Concept of Reliability Engineering 4
1.2.1 The Role of Reliability Engineering 4
1.2.2 Key Features of Reliability Engineering 5
1.3 Scientific Importance of Reliability in Modern Technology 6
1.4 Basic Concept of Probability Theory 7
1.5 Basic Concepts of System Reliability 9
1.5.1 Bathtub Curve 9
1.5.1.1 Initial Period (debugging/Burn-In/Infant-Monolith Period) 11
1.5.1.2 Constant Hazard Model 11
1.5.1.3 Wear-Out Region 11
1.5.1.4 Burn-In Screening 11
1.5.2 Reliability Functions 12
1.5.3 Mean Time to Failure of Component 14
1.5.4 Additional Examples 14
1.6 Conclusion 17
References 18

2 Bayesian Approach for Reliability Evaluation and Remaining Useful Life

Prediction 19
Debasis Jana, Suprakash Gupta, and Deepak Kumar
2.1 Introduction 19
2.2 Bayesian Network 20
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
viii Contents

2.3 Bayesian Reliability 22

2.4 Application of BN in Reliability and Remaining Useful Life 23
2.4.1 BN Structure Modeling 23
2.4.2 BN Parameter Learning 25
2.4.3 BN Inference 25
2.4.4 Validation and Verification 26
2.5 Dynamic Bayesian Networks 26
2.6 Advantages and Limitations of BN and DBN 27
2.7 Conclusion 28
References 29

3 Evaluation of Basic Reliability Indices Using State Enumeration Method 31

Rajesh Arya, Chandrima Roy, Atul Koshti, Ramesh C. Bansal, and Liladhar Arya
3.1 Introduction 31
3.2 Markov Process 31
3.3 Solution of State Equations 34
3.3.1 Steady State Solution: Continuous Time Case 34
3.3.2 Complete Solution: Continuous Time Case 35
3.3.2.1 Laplace Transform Method 35
3.3.2.2 Solution by Computing eAT 35
3.3.2.3 Discretization 35
3.3.2.4 Steady State Solution of Discrete Markov Process 36
3.3.2.5 Complete Solution of Discrete Markov Chain 36
3.4 Functions of a Single Component’s Availability and Unavailability 37
3.5 Two-Component State Model and State Probabilities 38
3.6 Three-Component State Transition Diagram 40
3.7 Concept of Frequency and Mean Duration 41
3.8 Frequency of Combined Events 42
3.9 State Enumeration Technique for Obtaining Frequency-Duration (FD) 44
3.10 Conclusion 49
References 49

4 Methodologies for Reliability Evaluation of Network 51

Rajesh Arya, Atul Koshti, Aanchal Verma, Baseem Khan, and Liladhar Arya
4.1 Introduction 51
4.2 Series Network 51
4.3 Parallel Network 53
4.4 Partially Redundant System 56
4.5 Reliability Evaluation of Complex Networks 57
4.5.1 Event Space Method 57
4.5.2 Decomposition Method 59
4.5.3 Tie-Set Method 61
4.5.4 Cut-Set Method 62
4.6 Determination of Tie-Sets 63
4.6.1 Connection Matrix Method Using Node Elimination 64
4.7 Method of Obtaining Cut-Set 65
4.8 Multistate Model 66
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Contents ix

4.8.1 Event-Space Method 66

4.8.2 Decomposition Approach 67
4.9 Illustrative Examples 68
4.10 Conclusions 72
References 73

5 Probabilistic Approach for Standby and Load-Sharing System Reliability

Evaluation 75
Rajesh Arya, R. K. Saket, Atul Koshti, Saad Mekhilef, and Pradeep Purey
5.1 Introduction 75
5.2 Reliability Evaluation Under Ideal Condition 75
5.3 Standby System Reliability Evaluation Under Nonideal Condition 78
5.3.1 Switching is Imperfect (Endrenyi 1978) 78
5.3.2 Switching is Imperfect and Switch has a Failure Rate 𝜆s (Imperfect Switch) 79
5.3.3 Consider the Standby Components Might Fail While they are in Idle Mode in
Addition 80
5.4 Reliability Evaluation of Load-Sharing System (Endrenyi 1978) 81
5.5 Illustrative Examples 83
5.6 Conclusion 88
References 89

Section 2 Reliability-Based Systems Design 91

6 Physical Reliability Methods and Design for System Reliability 93

Smriti Singh, Jyoti Maurya, Eram Taslima, Bharat B. Sagar, and R. K. Saket
6.1 Introduction 93
6.2 Reliability Methods 94
6.2.1 Block Diagram Analysis 94
6.2.1.1 Cut Sets and Tie Sets 94
6.2.1.2 Common Mode Failures 95
6.2.1.3 Enabling Events 95
6.2.2 Fault Tree Analysis 96
6.2.2.1 Steps of FTA 98
6.2.2.2 Applications of FTA 98
6.2.3 State Space Analysis (Markov Analysis) 98
6.2.3.1 Advantages, Limitations, and Application of Markov Analysis 99
6.2.4 Petri Nets 99
6.2.4.1 Fault Tree and Petri Net Transformation 99
6.2.5 Failure Mode Effects and Criticality Analysis (FMECA) 100
6.2.5.1 Steps in Performing an FMECA 103
6.2.5.2 Advantages of FMEA 104
6.2.6 Accelerated Life Testing (ALT) 104
6.2.7 Reliability Apportionment 104
6.3 Design Analysis and Process 105
6.3.1 Design for Reliability 105
6.3.1.1 Design for Reliability Tools 106
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
x Contents

6.3.1.2 Design for Reliability Metrics 107

6.3.2 Reliability Verification 108
6.3.3 Analytical Physics 108
6.3.4 Reliability Life Cycle 108
6.4 Conclusions 110
Bibliography 110

7 Design for Maintainability and Availability Analysis for System Design 113
Jyoti Maurya, Om P. Bharti, K. S. Anand Kumar, and R. K. Saket
7.1 Introduction 113
7.2 Elements of Maintainability 114
7.2.1 Maintainability and System Engineering 115
7.2.2 Maintainability Analysis Process 118
7.2.3 Maintainability Analysis Mathematics 119
7.3 Availability of the Systems 120
7.3.1 Availability of Repairable Systems 121
7.3.2 Availability of Nonrepairable Systems 121
Problems 122
7.4 Conclusion 123
Bibliography 123

8 Genetic Algorithm and Artificial Neural Networks in Reliability-Based Design

Optimization 125
Heeralal Gargama, Sanjay Kumar Chaturvedi, and Rajiv Nandan Rai
8.1 Introduction 125
8.2 Reliability-based Design 127
8.2.1 Performance-based Reliability 127
8.2.2 RBDO Problem 128
8.2.3 Probability Sufficiency Factor 129
8.2.4 MCS Approach 130
8.2.5 Latin Hyper Cube Design 131
8.2.6 Artificial Neural Networks 131
8.2.7 Real-coded GA 132
8.3 RBDO Methodology Using PSF and ANNs 134
8.3.1 Case Study: EMI Shielding Design for the Required Load of 80 dB 135
8.4 Conclusion 137
8.A Evaluation of Electromagnetic Shielding Effectiveness 138
References 139

9 Parametric Estimation Models for Minimal and Imperfect Maintenance 143

Rajiv Nandan Rai, Sanjay Kumar Chaturvedi, and Heeralal Gargama
9.1 Introduction 143
9.2 Maintenance Actions on Maintained Systems 145
9.3 Classifications of Imperfect Maintenance Categories 146
9.3.1 Perfect, Imperfect, and Minimal Repairs 146
9.4 Parametric Reliability Estimation Models for Maintained Systems 149
9.4.1 Definitions 149
9.4.2 Parametric Analysis Approaches 151
9.4.2.1 Renewal Process 151
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Contents xi

9.4.2.2 Nonhomogeneous Poisson Process 151

9.5 NHPP: Illustrative Example 153
9.6 Generalized Renewal Process 156
9.6.1 Arithmetic Reduction of Age Models 156
9.6.1.1 Kijima - I Model 157
9.6.1.2 Kijima - II Model 159
9.6.1.3 Virtual Age based Reliability Metrics 161
9.7 GRP: Illustrative Examples 161
9.8 Conclusion 164
Practice Problems 165
References 165

Section 3 Reliability Analysis of Transmission Systems 167

10 Transmission System Reliability Evaluation Including Security 169

Pushpendra Singh, Rajesh Arya, Lakhan Singh Titare, Mohd. Tauseef Khan,
and Sharat Chandra Choube
10.1 Introduction 169
10.2 Problem Formulation 171
10.3 Monte Carlo Simulation for Evaluation of the Security Index: With and Without
Considering the Absence of Transmission Lines 172
10.4 Evaluation of the Load Flow’s Minimal Eigenvalue Jacobian 174
10.5 Evaluation of Schur’s Inequality 175
10.6 Evaluation of the PSI and the Cut-set Approach 175
10.7 Recurrent Neural Network (RNN) Assessment of Probabilistic Insecurity 177
10.8 Results and Discussions 178
10.8.1 IEEE Six-Bus System 178
10.8.2 IEEE 14-Bus System 179
10.8.3 IEEE 25-Bus System 186
10.9 Conclusions 190
10.A.1 Data for IEEE six-bus, seven-line test system (100 MVA Base) 191
10.A.2 Data for IEEE 14-bus, 20-line system (100 MVA Base) 192
10.A.3 Data for IEEE 25-bus, 35 line system (100 MVA Base) 194
References 196

11 Probabilistic Voltage Security Assessment and Enhancement Using

Rescheduling of Reactive Power Control Variables 199
Lakhan Singh Titare, Aanchal Singh S. Vardhan, Liladhar Arya, and Devkaran Sakravdia
11.1 Introduction 199
11.2 Computation of Probabilistic Insecurity Index (PII) Using Cut-set Technique 201
11.3 Computation of Probabilistic Insecurity Index (PII) Sensitivity using ANN 202
11.4 Voltage Security Enhancement using a Monovariable Approach 205
11.5 Results and Discussion 206
11.5.1 14-bus System 206
11.5.2 25-bus System 210
11.6 Conclusions 214
References 215
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xii Contents

Section 4 Reliability Analysis of Distribution Systems 217

12 Modern Aspects of Probabilistic Distributions for Reliability Evaluation of

Engineering Systems 219
Aanchal Singh S. Vardhan, Aanchal Verma, Jyotsna Ogale, R. K. Saket, and Stuart Galloway
12.1 Introduction 219
12.2 Life Distribution of Power Components: An Overview 220
12.2.1 Binomial Distribution 220
12.2.2 Exponential Distribution 221
12.2.3 Poisson Distribution 221
12.2.4 Geometric Distribution 223
12.2.5 Weibull Distribution 224
12.2.5.1 Weibull Distribution Mean and Variance 224
12.2.6 Normal Distribution 225
12.2.7 Gamma Distribution 227
12.3 Failure Distribution Functions for Reliability Evaluation 227
12.3.1 Evaluation of Reliability Based on Exponential Distribution 228
12.3.2 Reliability Evaluation Based on Weibull Distribution 230
12.3.3 Normal Distribution-based Reliability Evaluation 232
12.4 Use of Exponential Model to Evaluate Reliability and MTBF 232
12.4.1 Components Connected in Serial 232
12.4.2 Parallel System 233
12.5 Probabilistic Methods For Reliability Evaluation 233
12.5.1 Gaussian Distribution Approach (GDA) 233
12.5.2 Safety Factor Concept (SFC) and Peak Load Consideration (PLC)-based Evaluation of
Reliability 235
12.5.3 LOLP Evaluation using Simpson’s 1/3rd rule 236
12.6 Additional Solved Examples 242
12.7 Conclusion 244
References 244

13 Reliability Enhancement of Electrical Distribution Systems Considering

Active Distributed Generations 247
Kalpesh B. Kela, Bhavik N. Suthar, Smriti Singh, Rajesh Arya, and Liladhar Arya
13.1 Introduction 247
13.2 Electrical Distribution Reliability Indices: Customer and Energy Based 249
13.3 Defining the Problem 250
13.3.1 DG Location Determination 250
13.3.2 Connecting DGs in the System as Standby Units 250
13.3.3 Analysis of Costs and Benefits 252
13.4 The Flower Pollination Algorithm Overview 253
13.5 Solution Approach 254
13.5.1 Evaluation of the DG Locations (Steps) 255
13.5.2 Evaluating the Optimized Solution with the Help of Flower Pollination Method
(Steps) 255
13.5.3 Analyzing the Costs and Benefits 256
13.6 Discussions and Outcomes 258
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Contents xiii

13.6.1 Roy Billinton Test System (RBTS-2) 258

13.6.2 Comparison Research 261
13.7 Conclusion 261
References 263

14 Reliability Enhancement Strategy for Electrical Distribution Systems

Considering Reward and Penalty 267
Kalpesh B. Kela, Bhavik N. Suthar, Liladhar Arya, and Rajesh Arya
14.1 Introduction 267
14.2 Reward and Penalty System (RPS) 269
14.3 Problem Identification 271
14.4 Rao Algorithms: An Overview 273
14.5 Steps to Solve the Problem 274
14.6 A Discussion of the Findings 274
14.6.1 Test System-Bus 2 by Roy Billinton (RBTS-2) 274
14.7 Conclusion 281
References 281

15 Reliability Analysis of Composite Distribution System Using Frequency

Duration Concept 285
Atul Koshti, Eram Taslima, Pradeep Purey, Liladhar Arya, and Sharat C. Choube
15.1 Introduction 285
15.2 Components Modeling in Composite Distribution System (CDS) 286
15.2.1 Capacity Modeling 286
15.2.2 Load Modeling 286
15.3 Frequency-Duration Concept for Reliability Indices Evaluation 286
15.4 MCS-Based Reliability Indices Evaluation of CDS 288
15.5 Result and Discussion 289
15.6 Illustrative Examples 290
15.7 Conclusions 298
References 298

Section 5 Reliability Analysis of Distribution Systems Integrated With

Renewable Energy Systems 301

16 Reliability Assessment of Distribution Systems Integrated with Renewable

Energy Systems 303
Sachin Kumar, Sandeep Kumar, Aanchal Singh S. Vardhan, R. K. Saket, and P. Sanjeevikumar
16.1 Introduction 303
16.2 Reliability Functions 305
16.3 Renewable Energy Sources 307
16.3.1 Solar Energy 307
16.3.2 Wind Energy 309
16.3.3 Vehicle Charging-Discharging 310
16.3.4 Other Renewable Energy Sources 313
16.4 Optimization and Control 313
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xiv Contents

16.5 Case Study 315

16.5.1 IEEE Reliability Test System-79 315
16.5.2 Roy Billinton Test System 318
16.5.3 Bahir Dar’s Power Distribution Test System 319
16.6 Challenges and Future Directions 320
16.6.1 Challenges 320
16.6.2 Future Directions 320
Problems 321
Solution 322
16.7 Conclusion 323
References 323

17 Reliability Evaluation and Performance of Hybrid Photovoltaic Energy

Systems for Rural Electrification Using Markov Process 325
Santosh S. Raghuwanshi, Smriti Singh, Akanksha Singh S. Vardhan, Rajesh Arya,
and R. K. Saket
17.1 Introduction 325
17.2 Reliability Indices 326
17.3 Markov Process 327
17.4 Reliability of the System 329
17.4.1 System with Series Components 329
17.4.2 System with Parallel Components 330
17.4.2.1 Complex (Series–Parallel) System 330
17.4.3 Renewable Hybrid PV System 331
17.4.3.1 PV Arrays 332
17.4.3.2 Converter 332
17.4.3.3 Inverter 332
17.4.3.4 Charge Controller and Solar Batteries 332
17.4.3.5 DG Set 333
17.5 Conclusion 338
References 338

18 Probabilistic Distribution and Monte Carlo Approach for Reliability

Evaluation of SEIG-Based Micro Hydro Power Generation System 341
Lokesh Varshney, Kanhaiya Kumar, Gautam Singh Dohare, Udaya M. Bhaskara Rao,
and Jitendra Singh Shakya
18.1 Introduction 341
18.2 Residual Magnetism in SEIG: Restoration and Loss 342
18.2.1 Residual Magnetism Losses 342
18.2.2 Residual Magnetism Restoration 343
18.3 Problems with SEIG Excitation Failure in RE Systems 343
18.4 SEIG Tests with Lowest Capacitive Excitation 343
18.5 Rotor Core Magnetization of SEIG Reliability Assessment Using Least Capacitor
Score 344
18.5.1 Assessment of Reliability-Related Functions 345
18.5.2 Probabilities of Failure and Success 345
18.6 Discussion and Outcomes 349
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Contents xv

18.7 Conclusion 350

References 350

19 Reliability and Mean Life Assessment of Solar Panel by Cooling 353

Rahul Agrawal, Jyotsna Ogale, Nga T. T. Nguyen, R. K. Saket, and Joydeep Mitra
19.1 Introduction 353
19.2 Methodology 355
19.2.1 Cell Temperature 356
19.2.2 Degradation Equations 357
19.2.3 Mean Life Without Cooling 358
19.2.4 Mean Life After Cooling 359
19.2.4.1 Mean Life with Cooling up to 25 ∘ C 359
19.2.4.2 Mean Life with Cooling up to 35 ∘ C 360
19.2.4.3 Mean Life with Cooling up to 45 ∘ C 361
19.3 Reliability Assessment 365
19.3.1 Cooling up to 25 ∘ C 366
19.3.2 Cooling up to 35 ∘ C 367
19.3.3 Cooling up to 45 ∘ C 368
19.4 Probability Density Function 369
19.4.1 Cooling up to 25 ∘ C 369
19.4.2 Cooling up to 35 ∘ C 371
19.4.3 Cooling up to 45 ∘ C 371
19.5 Cumulative Distribution Function 371
19.5.1 Cooling up to 25 ∘ C 374
19.5.2 Cooling up to 35 ∘ C 374
19.5.3 Cooling up to 45∘ C 377
19.6 Results 378
19.7 Conclusion 378
References 379

20 Reliability Assessment of Different Topologies in Photovoltaic System 381

Laxman Chaudhary, Aanchal Verma, Ramesh C. Bansal, and R. K. Saket
20.1 Introduction 381
20.1.1 Basic Reliability Estimation Concepts 382
20.1.2 Hazard Rate Function 382
20.1.3 Reliability Block Diagram 384
20.2 Reliability Modeling of PV Topology 385
20.2.1 Reliability Modeling of Central PV Topology 385
20.2.2 Reliability Modeling of String PV Topology 385
20.2.3 Reliability Modeling of Micro PV Topology 386
20.3 Estimation of Failure Rate 387
20.3.1 DC-Link Capacitor 387
20.3.2 Diodes 387
20.3.3 Switch 388
20.3.4 Filter 388
20.4 Reliability Estimation Using RBD 388
20.4.1 Central Topology 389
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xvi Contents

20.4.2 String Topology 391

20.4.3 Micro Topology 396
20.5 Results 400
20.6 Conclusions 405
References 405

Section 6 Reliability Analysis of Power Electronics Components and

Systems for Modern Power System Applications 409

21 Reliability Evaluation of Power Electronics Converters for Modern Power

System Applications 411
Amit Kumar, Sachin Kumar, Sunil K. Singh, R. K. Saket, and P. Sanjeevikumar
21.1 Introduction 411
21.2 Failures in Power Electronics Converters 412
21.2.1 Catastrophic Failure 412
21.2.2 Wear-Out Failure 413
21.3 Estimation and Monitoring of Junction Temperature 414
21.3.1 Direct Method 414
21.3.2 Indirect Method 414
21.3.2.1 Estimation by Vceon(Ihigh) 415
21.3.2.2 Estimation by Threshold Voltage (Vthr ) 415
21.3.2.3 Estimation by Short Circuit Current (SCC) 416
21.3.2.4 Estimation by Turn-On/Off Delay Time 417
21.3.2.5 Estimation by Rgint with the Help of Igpeak 418
21.4 Reliability of a Modern Power System 420
21.4.1 Power Converter Availability Model 422
21.4.2 HVDC Availability Model 422
21.4.3 Reliability Model of Modern Power System 423
21.5 Challenges and Future Directions 424
References 424

22 Reliability Assessment of Sub-components of Electric Vehicle for

Performance Enhancement Grid Integrated Power System 427
Saumya Singh, Dhawal Dwivedi, Sandeep K. Soni, R. K. Saket, and Dwarkadas P. Kothari
22.1 Introduction 427
22.2 Electric Vehicles and Grid Integration 428
22.2.1 Charging Infrastructure 429
22.2.2 Load Management and Grid Stability 429
22.2.3 Benefits and Challenges of Grid Integration 430
22.3 Sub-components of EVs 431
22.3.1 Reliability Study of Sub-components in EVs 433
22.3.1.1 Methodologies of Reliability Studies 434
22.4 Reliability Assessment Techniques in EVs 435
22.4.1 Markov Model 435
22.4.2 Monte Carlo Simulation (MCS) 436
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Contents xvii

22.4.3 Contingency Enumeration Method 437

22.4.4 State Space Method 438
22.4.5 Electric Vehicles Trend 439
22.4.6 Fundamental Considerations in Assessing the Reliability of EVs 439
22.4.7 EVs Battery Management Systems (BMS) 442
22.5 Evaluation of Distribution Systems Reliability with Integrated EVs 443
22.5.1 Distribution Network (DN) Reliability Assessment 444
22.5.2 Reliability Evaluation Using a V2G Approach and High EV Penetration 444
22.5.3 Reliability Parameters 446
22.5.3.1 Reliability Parameters of EVs 446
22.5.3.2 Grid System Reliability Parameters 447
22.6 Conclusion 448
References 448

23 Reliability Assessment of Multilevel Inverter for Modern Power System

Applications 451
Saumya Singh, Dhawal Dwivedi, Kumari Sarita, R. K. Saket, and P. Sanjeevikumar
23.1 Introduction 451
23.2 Reliability Assessment Techniques 453
23.3 Types of Multilevel Inverters (MLIs) 456
23.3.1 Cascaded H-Bridge Multilevel Inverters 456
23.3.2 Neutral-Point Clamped Three-Level Inverter (NPC) 458
23.3.3 Flying Capacitors Three-Level Inverter 458
23.3.4 Three-Level T-Type Inverter 460
23.4 Comparative Reliability Assessment of MLIs 463
23.5 Conclusion 464
References 464

24 Reliability Aspects in Snubber Circuit for Industrial Power Applications 467

Dhawal Dwivedi, Saumya Singh, Kumari Sarita, R. K. Saket, and P. Sanjeevikumar
24.1 Introduction 467
24.2 Passive Snubber Circuit 468
24.2.1 Surge Across Switch 468
24.3 Selection of Turn-OFF Snubber 469
24.3.1 Snubbers Across Each Switching Device 469
24.3.1.1 RC Snubber Circuit 469
24.3.1.2 RCD Charge-Discharge Snubber 469
24.3.1.3 Discharge-Suppressing RCD Snubber 470
24.3.2 Lumped Snubbers Between Power Buses 470
24.3.2.1 C Snubber 470
24.3.2.2 RCD Snubber 471
24.4 Design of a Discharge-Suppressing RCD Snubber 471
24.5 Simulation Results of RCD Snubber 472
24.6 Reliability Aspects in Snubber Design for Industrial Power Applications 476
24.7 Conclusion 478
References 478
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xviii Contents

25 Reliability Assessment of Power Electronics Devices and Systems for Modern

Power Applications 481
Jyoti Maurya, Saumya Singh, Sachin Kumar, P. Sanjeevikumar, and R. K. Saket
25.1 Introduction 481
25.2 Concept of PEDS Reliability in Modern Power System 483
25.3 V-Shape Model-Based Reliability Assessment in PEDS 486
25.3.1 Hierarchical Reliability Modeling and Assessment 486
25.3.2 Model-Based Reliability Enhancement 488
25.4 Converter Reliability Modeling 489
25.5 Conclusion and Future Challenges 492
References 492

26 Reliability Aspects in the Design and Development of Microgrids 493

Amit Kumar, Sachin Kumar, Almoataz Y. Abdelaziz, R. K. Saket, and D. P. Kothari
26.1 Introduction 493
26.2 Architecture and Operation of Microgrid 494
26.2.1 Types of Microgrid 494
26.2.2 Microgrid Technology 496
26.3 Microgrid Control Strategies 496
26.3.1 Advantages and Disadvantages of Microgrid 497
26.4 Reliability Aspects in Microgrid Planning and Design 499
26.4.1 Reliability Evaluation 500
26.4.2 Energy Scheduling, Forecasting, and Optimization Techniques 502
26.4.3 Reliability Aspects of Power Electronics in Microgrids 502
26.4.4 Component to System: Layer-Wise Reliability 503
26.5 Conclusion and Future Challenges 504
References 505

Abbreviations 507
Notations 513
Index 525
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xix

About the Authors

Professor (Dr.) R. K. Saket (Member’12–Senior Member’18, IEEE)

is a professor with the Department of Electrical Engineering,
Indian Institute of Technology (Banaras Hindu University),
Varanasi, Uttar Pradesh, India. Previously, he was a faculty
member at Government Engineering College, Rewa (Madhya
Pradesh), India; Birla Institute of Technology and Science, Pilani
(Rajasthan), India; Sam Higginbottom University of Agriculture,
Technology and Sciences, Allahabad (Uttar Pradesh), India;
and University Institute of Technology, Rajiv Gandhi University
of Technology, Bhopal (Madhya Pradesh), India. He has pro-
vided his industrial services to Electrical Safety Division of the
Department of Energy, Government of Madhya Pradesh, Ujjain
Division (Madhya Pradesh), India, as an electrical safety engi-
neer. He has more than 25 years of academic, industrial, and research experience. He is the
author/co-author of six national/international patents, approximately 200 scientific articles, book
chapters, and research papers in indexed international journals and prestigious conference pro-
ceedings. He received the Best Paper Award(s) at Thammasat University, Bangkok, Thailand on
23 January 2017 and The Maldives National University, Republic of Maldives on March 11, 2023.
He is the author/co-author of the prestigious textbook Modern Power System Analysis, 5th edition,
2022, published by McGraw Hill, New Delhi, India. He has supervised 21 PhD research scholars
and 55 MTech students. He has delivered many technical talks and is honored as a resource person
in power system reliability engineering. He has delivered his academic and research lecture(s) at
World Academy of Science, Engineering & Technology, Paris (France) on 21–23 September 2014;
International Theravada University, Yangon (Myanmar) on 2–4 December 2014; Thammasat Uni-
versity, Bangkok (Thailand) on 21–23 January 2017; Tokyo City University, Tokyo (Japan) on 9–11
March 2018; Brunel University, London (UK) on 20–23 February 2023; and The Maldives National
University, Male City (Maldives) on 11–12 March 2023. His research interests include reliability
engineering, electrical machines and drives, modern power system reliability, reliability enhance-
ment of industrial components and systems, and reliability aspects in renewable energy systems.
Professor Saket is a fellow of the Institution of Engineers (India), fellow of the Institution of Elec-
tronics and Telecommunication Engineers (India), senior member of IEEE (USA), member of IET
(UK), and a life member of the Indian Society for Technical Education, New Delhi (India). He is an
associate editor of the IET Renewable Power Generation (UK), IET Electrical Systems in Transporta-
tion (USA), and IEEE Access (USA). He is the Managing Guest Editor of the IEEE Journal of the Elec-
tron Devices Society (USA), Computers & Electrical Engineering, Elsevier, and Electrical Engineering,
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xx About the Authors

Springer Nature. He is an Editorial Board Member of the Journal of Electrical Systems (Algeria),
and Engineering, Technology and Applied Science Research (Greece). He has received many awards,
honors, and recognitions for his excellent academic and research contributions including the pres-
tigious Gandhian Young Technological Innovation Award (2018) appreciated by the hon’ble presi-
dent of India at Rashtrapati Bhavan, New Delhi, India; Design Impact Award (2018) appreciated by
Padma Vibhushan Ratan Tata at Mumbai, India; Nehru Encouragement Award (1988 and 1990)
awarded by the hon’ble chief minister of Madhya Pradesh state government, Bhopal, India; and
IEEE IAS Global Distinguished Educator Award 2023 (USA) at the Maldives National University,
Republic of Maldives.

Professor (Dr.) P. Sanjeevikumar (Member’12–Senior Mem-

ber’15, IEEE) received the PhD degree in electrical engineering
from the University of Bologna, Bologna, Italy, in 2012.
He is a full professor in electrical power engineering at
the Department of Electrical Engineering, Information
Technology, and Cybernetics, University of South-Eastern
Norway, Norway. He has authored over 750 scientific
papers and received the Best Paper cum Most Excellence
Research Paper Awards from IET-SEISCON’13, IET-CEAT’16,
IEEE-EECSI’19, IEEE-CENCON’19, and five best paper awards
from ETAEERE’16-sponsored Lecture Notes in Electrical
Engineering, Springer book. He is a fellow of the Institution of
Engineers (India), the Institution of Electronics and Telecom-
munication Engineers (India), and the Institution of Engineering and Technology (UK). He
received a lifetime achievement award from Marquis Who’s Who in USA in 2017 for contributing
to power electronics and renewable energy research. He has been listed among the world’s top 2%
of scientists since 2019 by Stanford University, USA.
He is an editor/associate editor/editorial board for refereed journals, in particular the IEEE Sys-
tems Journal (USA), IEEE Transaction on Industry Applications (USA), IEEE Access (USA), IET
Power Electronics (UK), IET Electronics Letters (UK), and International Transactions on Electrical
Energy Systems, Wiley; subject editorial board member—Energy Sources—Energies Journal, MDPI;
and the subject editor for the IET Renewable Power Generation (UK), IET Generation, Transmission
and Distribution (UK), and FACETS Journal (Canada).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xxi

List of Contributors

Almoataz Y. Abdelaziz Ramesh C. Bansal

Faculty of Engineering & Technology Department of Electrical Engineering
Future University in Egypt University of Sharjah
Cairo Sharjah
Egypt United Arab Emirates

Rahul Agrawal Om P. Bharti

Department of Electrical Engineering Department of Electrical Engineering
Indian Institute of Technology (BHU) Government Polytechnic
Varanasi, Uttar Pradesh Ghazipur, Uttar Pradesh
India India

K. S. Anand Kumar Udaya M. Bhaskara Rao

Electrical Division Engineering Division
CSIR National Aerospace Laboratory CSIR National Metallurgical Laboratory
Bangalore, Karnataka Jamshedpur, Jharkhand
India India
Liladhar Arya
Sanjay Kumar Chaturvedi
Department of Electrical Engineering
Subir Chowdhury School of Quality and
Shri G.S. Institute of Technology & Science
Reliability
Indore, Madhya Pradesh
Indian Institute of Technology Kharagpur
India
Kharagpur, West Bengal
Rajesh Arya India
Department of Information Technology
Acropolis Institute of Technology and Research Laxman Chaudhary
Indore, Madhya Pradesh Department of Electrical Engineering
India Indian Institute of Technology (BHU)
Varanasi, Uttar Pradesh
Vanitha Bagana India
Department of Electronics and
Communication Engineering
National Institute of Technology
Jamshedpur, Jharkhand
India
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xxii List of Contributors

Sharat Chandra Choube Baseem Khan

Department of Electrical & Electronics Department of Electrical and Computer
Engineering Engineering
University Institute of Technology, RGPV Hawassa University
Bhopal, Madhya Pradesh Hawassa
India Ethiopia

Gautam Singh Dohare Mohd. Tauseef Khan

Department of Electrical, Electronics & Department of Electrical Engineering
Communication Engineering Rajkiya Engineering College
Galgotias University Banda, Uttar Pradesh
Greater Noida, Uttar Pradesh India
India
Atul Koshti
Dhawal Dwivedi
Department of Electrical Engineering
Department of Electrical Engineering
GES’S R.H. Sapat College of Engineering
Indian Institute of Technology (BHU)
Management Studies & Research
Varanasi, India
Nashik, Maharashtra
Stuart Galloway India
University of Strathclyde
Glasgow Dwarkadas P. Kothari
UK Department of Electrical Engineering
Indian Institute of Technology
Heeralal Gargama Delhi
Subir Chowdhury School of Quality and India
Reliability
Indian Institute of Technology Kharagpur Amit Kumar
Kharagpur, West Bengal Flameproof and Equipment Safety Department
India CSIR-Central Institute of Mining and Fuel
Research
Suprakash Gupta Dhanbad, Jharkhand
Department of Mining Engineering India
Indian Institute of Technology (BHU)
Varanasi (UP) Deepak Kumar
India Department of Mining Engineering
Indian Institute of Technology (BHU)
Debasis Jana
Varanasi (UP)
Department of Mining Engineering
India
Indian Institute of Technology (BHU)
Varanasi (UP)
Sachin Kumar
India
Department of Electrical Engineering
Kalpesh B. Kela G.B. Pant Institute of Engineering and
Department of Electrical Engineering Technology
L.D. College of Engineering Pauri-Garhwal, Uttarakhand
Ahmedabad, Gujarat India
India
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
List of Contributors xxiii

Sandeep Kumar Pradeep Purey

Department of Electronics and Department of Electronics Engineering
Communication Engineering Maharaja Ranjit Singh College of Professional
G.B. Pant Institute of Engineering and Sciences
Technology Indore, Madhya Pradesh
Pauri-Garhwal, Uttarakhand India
India
Santosh S. Raghuwanshi
Kanhaiya Kumar Department of Electrical Engineering
Electrical & Electronics Engineering Medi-Caps University
Department Indore, Madhya Pradesh
Galgotias College of Engineering & Technology India
Greater Noida, Uttar Pradesh
India Rajiv Nandan Rai
Subir Chowdhury School of Quality and
Jyoti Maurya Reliability
Department of Electrical Engineering Indian Institute of Technology Kharagpur
Indian Institute of Technology (BHU) Kharagpur, West Bengal
Varanasi, Uttar Pradesh India
India
Chandrima Roy
Saad Mekhilef Department of Electronics & Communication
Department of Electrical Engineering Engineering
Swinburne University of Technology Heritage Institute of Technology
Melbourne Kolkata, West Bengal
Australia India

Joydeep Mitra Bharat B. Sagar

Department of Electrical Engineering Department of Computer Science and
Michigan State University Engineering
Michigan School of Engineering, Harcourt Butler
USA Technical University
Kanpur, Uttar Pradesh
Nga T. T. Nguyen India
Department of Electrical Engineering and
Computer Science R. K. Saket
University of Wyoming Department of Electrical Engineering
Wyoming Indian Institute of Technology
USA (Banaras Hindu University)
Varanasi (UP)
Jyotsna Ogale India
Department of Electronics Engineering
Samrat Ashok Technological Institute
Vidisha, Madhya Pradesh
India
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xxiv List of Contributors

Devkaran Sakravdia Sunil K. Singh

Department of Electrical Engineering Department of Electrical Engineering
Government Engineering College Shri Ramswaroop Memorial University
Ujjain, Madhya Pradesh Lucknow, Uttar Pradesh
India India

P. Sanjeevikumar Sandeep K. Soni

Department of Electrical Engineering Department of Electrical Engineering
IT and Cybernetics Shiv Nadar Institution of Eminence
University of South-Eastern Norway Delhi, NCR
Porsgrunn, Norway India

Kumari Sarita Bhavik N. Suthar

Department of Electrical Engineering Department of Electrical Engineering
Government Engineering College Government Engineering College
Aurangabad, Aurangabad, Bihar Bhuj, Gujarat
India India

Jitendra Singh Shakya Eram Taslima

Department of Electrical Engineering Department of Electrical Engineering
Samrat Ashok Technical Institute Indian Institute of Technology (BHU)
Vidisha, Madhya Pradesh Varanasi, Uttar Pradesh
India India

Pushpendra Singh Lakhan Singh Titare

Department of Electrical Engineering Department of Electrical Engineering
Rajkiya Engineering College Government Engineering College
Banda, Uttar Pradesh Ujjain, Madhya Pradesh
India India

Saumya Singh Aanchal Singh S. Vardhan

Department of Electrical Engineering Department of Electrical Engineering
Indian Institute of Technology (BHU) National Institute of Technology
Varanasi, Uttar Pradesh Jamshedpur, Jharkhand
India India

Smriti Singh Akanksha Singh S. Vardhan

Department of Electrical Engineering Department of Electrical Engineering
Indian Institute of Technology (BHU) Shri GS Institute of Technology & Science
Varanasi, Uttar Pradesh Indore, Madhya Pradesh
India India
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xxv
List of Contributors

Indian Institute of Technology (BHU)

Department of Electrical Engineering

Varanasi, Uttar Pradesh

Aanchal Verma

India
Department of Electrical, Electronics &
Communication Engineering

Greater Noida, Uttar Pradesh

Galgotias University
Lokesh Varshney

India
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xxvii

Foreword

Writing a foreword is both a pleasure and a challenge. Still, I always enjoy writing the foreword
to recommend bookwork in my fields of power, energy, drives, reliability, renewables, and power
systems technologies. This application benefits society and solves power demand, reliability,
power quality, generation, transmission, and distribution crises through modern power systems.
I read this book titled Reliability Analysis of Modern Power Systems, Wiley-IEEE Publications,
authored/edited by learned colleagues Prof. R. K. Saket, Department of Electrical Engineering,
Indian Institute of Technology (Banaras Hindu University), Varanasi, Uttar Pradesh, India, and
Prof. P. Sanjeevikumar, Department of Electrical Engineering, IT and Cybernetics, University of
South-Eastern Norway, Porsgrunn, Norway, which has proven exciting and innovative research
findings for the readers. In particular, power generation, transmission, and distribution reliability
are continuously gaining attention as crucial, empowering technology with a smart control
approach to enhance the system’s reliability and efficiency. These challenges are solvable through
modern power electronics, with higher flexibility, adjustable electronics loads, energy storage
systems, and renewable energy systems, including windmills, solar PV, batteries, supercapacitors,
hydropower, and tidal power.
On reading, the topics that draw strong reader attention are the basic principles and scientific
importance of reliability theory, the Bayesian approach and useful life prediction, genetic algo-
rithms, artificial neural networks in reliability-based optimization, and reactive power control vari-
ables rescheduling; and also reliability evaluation of power electronics converters and devices in
modern power system applications, reliability assessment of multilevel inverters, design and devel-
opment of microgrids, and reliability aspects in snubber circuits for industrial power applications.
I hope the readers will enjoy reading this book and be able to grasp the research findings as a text-
book for bachelor’s, master’s, and PhD programs in universities and further future enhancement
in modern science, engineering, technology, and skills!
Finally, I congratulate the editors, authors, Wiley-IEEE Publishers, reviewers and editorial assis-
tants, and the IEEE and Wiley press production team; without their support, it would not be a
successful outcome. Credit goes to their effort in making flyers and editing, proof correction, etc.

With Best Compliments

Prof. Frede Blaabjerg, Fellow IEEE

Villum Investigator Professor,
Center of Reliable Power Electronics (CoRPE),
Department of Energy Engineering,
Aalborg University, Aalborg, Denmark
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xxviii Foreword

It is a pleasure to write the foreword to this important book, Reliability Analysis of Modern Power
Systems, edited by Dr. R. K. Saket and Dr. P. Sanjeevikumar. The book has been contributed by many
authors who are university professors specialized in respective topics and are editors. The book is
hybrid, heterogeneous, and interdisciplinary in nature. All the contributions appear authoritative
in nature. The book contains 26 chapters.
Reliability is an extremely important topic in modern power systems. It is particularly important
for tomorrow’s large and complex smart grid power system, where many generators (fossil, nuclear,
and renewables with storage elements) are interconnected through long transmission, distribution,
and utilization systems to supply reliable electricity to consumers at economical prices. Smart grid
uses state-of-the-art technologies in power systems, power electronics, computers, communica-
tions, information, cyber, and AI to optimize system availability, reliability, power quality, security,
and energy efficiency with optimum resource utilization.
The book deals with reliability theories and their applications in power systems very comprehen-
sively. The topics include reliability studies in the distribution system, photovoltaic energy system,
hydro-power system, power electronic converters (including devices and snubbers), electric vehi-
cles, and microgrids. It also covers studies on genetic algorithms and artificial neural networks
in reliability-based optimization, parametric estimation models, security of transmission systems,
probabilistic voltage security assessment, reactive power control rescheduling, etc.
The book can be considered an important reference book for engineers involved in designing
reliable power systems. It can also be considered a graduate-level text. Selected topics from the
book can also be taught in undergraduate courses.
It should be mentioned here that Dr. R. K. Saket and Dr. P. Sanjeevikumar are well known
internationally for their research contributions. While Dr. R. K. Saket specializes in reliability the-
ory and applications, Dr. P. Sanjeevikumar is more renowned in the area of power electronics.
The publication of this book is very timely and will be accepted with enthusiasm.

Best Compliments

Prof. Bimal K. Bose, PhD, DSc (Honoris Causa), IEEE Life Fellow
Emeritus Chair Professor of Electrical Engineering
(Formerly Condra Chair of Excellence in Power Electronics),
Department of Electrical Engineering and Computer Science,
The University of Tennessee, Knoxville, USA
http://en.wikipedia.org/wiki/Bimal_Kumar_Bose
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xxix

Preface

Reliability is an old concept and a new discipline in engineering and technology. Reliability Analysis
of Modern Power Systems (RAMPS) has evolved from our deep interest in academics and innovative
research and our long-standing involvement in quantitative reliability evaluation and application
of probabilistic techniques to modern power system problems. Modern power systems have grown
larger and spread over larger geographical area with many interconnections between neighbor-
ing systems and integration of renewable energy sources. The applications benefit society and
solve power demand, power quality, reliability, generation, transmission, and distribution crises
through modern power systems. The systems that involve high-risk and high-cost factors need to
be reliable enough. This book is basically intended to present important aspects of reliability and
its analysis in the field of modern power systems. The primary objective of RAMPS is to present
the historical importance of reliability engineering, dependability assessment of power compo-
nents, design enhancement for reliability, impact of renewable energy systems, and innovative
research findings for engineering students, PhD research scholars, practicing power engineers, and
technical managers. The emphasis throughout is on practical applications, and the mathematical
concepts described are accordingly suitable for solving the types of reliability problems covered
in RAMPS. The references given in the bibliographies are limited to those considered to provide a
direct continuation of the chapter material, with an emphasis on experimental research and practi-
cal applications. Tables, figures, flowcharts, and graphs are provided to complement the analytical
methods described, and numerous worked examples are included. Power generation, transmis-
sion, and distribution are continuously gaining attention as crucial, empowering technology with
a smart control approach to enhance the system’s reliability and efficiency. The challenges are solv-
able through modern power electronics with high flexibility, adjustable electronic loads, energy
storage systems, and renewable energy systems including wind energy conversion systems, solar
photovoltaic, electric vehicles, hydropower, and tidal power.
The text of RAMPS is divided into six sections as shown in Figure P.1. The first section describes
the historical background, reliability principles, and comprehensive importance of reliability engi-
neering; the second section addresses the design for reliability and design enhancement for depend-
ability of modern power components and systems; the third section provides information about
reliability assessment aspects for transmission system of modern power systems; the fourth section
describes the reliability issues associated with conventional power distribution system; the fifth
section of RAMPS illustrates the reliability analysis methodologies of modern distribution systems
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xxx Preface

SECTION 1
Reliability
Principles and
Applications
SECTION 6
Reliability
Analysis of Power SECTION 2
Electronics Reliability-Based
Components and Systems Design
System for Modern
Power Application Reliability
Analysis of
Modern Power
Systems
SECTION 5 (RAMPS) SECTION 3
Reliability Analysis
Reliability
of Distribution
Analysis of
System Integrated
Transmission
with Renewable
System
Energy Systems
SECTION 4
Reliability Analysis
of Distribution
System

Figure P.1 Organization of Reliability Analysis of Modern Power Systems.

integrated with renewable energy systems; and the reliability analysis of power electronics compo-
nents, integrated circuits and systems for modern power system applications are described in the
sixth section.
The sections of the RAMPS textbook are subdivided into 26 innovative chapters as shown in
Figure P.2. Section 1: Reliability Principles and Applications provides an overview of reliability
principles and comprehensive applications in the fields of modern engineering and technology,
where the basic concepts of reliability, life distribution of power components, reliability analysis
of composite power system structure, failure distribution functions, and useful life predictions
are described in detail. In Chapter 1, the basic reliability principles, applicability of bathtub
curve, and scientific importance of reliability theory with numerical illustrations are described
in detail. The Bayesian approach for reliability evaluation and remaining useful life prediction is
presented in Chapter 2. The evaluation of basic reliability indices using state enumeration method
with industrial numerical problems is described in Chapter 3. In Chapter 4, the methodologies
for reliability evaluation of network with industrial problems are discussed. The probabilistic
approach for standby and load-sharing system reliability evaluation is illustrated in Chapter 5.
Section 2: Reliability-Based Power Systems Design contains four innovative chapters. The physical
reliability methods and design for system reliability including experimental questions and indus-
trial real problems are presented in Chapter 6. Design for maintainability and availability analysis
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Preface xxxi

Basic principle, scientific importance of reliability,

Reliability Principles and bayesian approach, reliability evaluation of network,
Section 1 Applications life prediction, reliability indices and evaluation methods

Reliability Based Systems Physical reliability methods, maintainability and

Section 2 availability design for reliability, genetic algorithm and
Design
artificial neural network product design and life estimation

Reliability Reliability Analysis of Transmission Transmission system reliability, security

Analysis Section 3
System aspects, probabilistic voltage security,
of Modern assessment, distribution system reliability
Power Modern aspects of probabilistic
Reliability Analysis of Distribution
Systems Section 4 distribution, reliability, aspects based on
System active distributed generations, frequency-
(RAMPS) duration concept, reward and penalty
Reliability Analysis of Distribution Distribution system integrated with renewable energy
Section 5 System Integrated with Renewable systems, mean life assessment of solar panel by
Energy Systems cooling, Markov process, monte carlo approach

Reliability Analysis of Power Power electronics converters for modem power system
Section 6 Electronics Components and Systems applications, electric vehicles, reliability of power
for Modern Power System Application electronic devices multilevel inverters, design and
development of micro grids

Figure P.2 Section classification of RAMPS and organization of the chapters.

for system design are described in Chapter 7. Chapter 8 presents the genetic algorithm and arti-
ficial neural networks in reliability-based design optimization. The parametric estimation models
for minimal and imperfect maintenance are illustrated in Chapter 9.
Section 3: Reliability Analysis of Transmission Systems contains only two innovative chapters. The
transmission system reliability evaluation including security aspects and dependability illustra-
tions is illustrated in Chapter 10. In Chapter 11, the probabilistic voltage security assessment and
stability enhancement using rescheduling of reactive power control variables of modern power
systems are presented.
Section 4: Reliability Analysis of Distribution Systems is incorporated with four comprehensive
chapters. Chapter 12 presents the modern aspects of probabilistic distributions with industrial
numerical problems for reliability evaluation of engineering systems. Reliability enhancement
of electrical distribution systems considering active distributed generations is discussed in
Chapter 13. In Chapter 14, the reliability enhancement strategy for electrical distribution system
considering reward and penalty is presented and supported by industrial data. The reliability
analysis of composite distribution system using frequency–duration concept is incorporated in
Chapter 15.
Section 5: Reliability Analysis of Distribution Systems Integrated with Renewable Energy Systems
is decorated with five innovative chapters. The reliability assessment of distribution systems inte-
grated with renewable energy systems is presented in Chapter 16. In Chapter 17, the reliability
evaluation and performance analysis of hybrid photovoltaic energy systems for rural electrifica-
tion using Markov process are described successfully. The probabilistic distributions and Monte
Carlo approach for reliability evaluation of SEIG-based modern power systems are presented in
Chapter 18. In Chapter 19, the reliability and mean life assessment of solar panels by cooling are
discussed. The reliability assessment of different topologies in photovoltaic systems is incorporated
into Chapter 20.
Section 6: Reliability Analysis of Power Electronics Components and Systems for Modern Power
System Applications is incorporated with six innovative chapters. The reliability evaluation of
power electronics converters for modern power system applications is covered in Chapter 21.
In Chapter 22, the reliability assessment of sub-components of electric vehicles for performance
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xxxii Preface

enhancement of grid-integrated power systems is incorporated excellently. The reliability assess-

ment of multilevel inverters for modern power system applications is illustrated successfully
in Chapter 23. In Chapter 24, the reliability aspects of snubber circuits for industrial power
applications are discussed. The reliability assessment of power electronics devices and systems for
modern power applications is incorporated into Chapter 25. Finally, the reliability aspects of the
design and development of microgrids are illustrated in Chapter 26.
The requirements of the course syllabi of the All India Council for Technical Education (India),
University Grants Commission (India), Indian Institute of Technology (IITs), American Society for
Quality Control (USA), and the Institute of Quality Assurance (UK) in reliability engineering are
covered, so the book will be suitable for use in courses leading to these organizations. The empha-
sis on practical approaches and experimental methodologies in modern power system engineering,
comprehensive coverage of reliability standards and specifications, and the overall layout of the
book should make it equally suitable as a general, up-to-date reference for use in academia, indus-
try, and government agencies. We hope that the readers will enjoy reading this book of reasonable
length and comprehensive coverage. This book is timely and addresses the needs of engineers, pro-
fessionals, and researchers. Readers will be able to grasp the research findings as a textbook for
graduation, postgraduation, and PhD programs in universities and will refer it for future enhance-
ment in modern science, engineering, technology, and skills!
Finally, we congratulate the authors, co-authors, expert reviewers, editorial assistants,
Wiley-IEEE Publishers, and the IEEE and Wiley press production team; without their valuable
support, it would not be a successful outcome.

18 January 2024 R. K. Saket

Varanasi (UP) India P. Sanjeevikumar
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xxxiii

Acknowledgments

While writing this textbook, we have had the benefit of valuable advice and suggestions from many
academicians, senior faculty colleagues, practicing engineers, and other innovators who used the
earlier editions of the book entitled Modern Power System Analysis, authored by D.P. Kothari,
I.J. Nagrath, R. K. Saket, and our previous research publications. I want to thank the expert
reviewers of Reliability Analysis of Modern Power Systems (RAMPS) for their many helpful
suggestions. To the extent that I have been able to respond to their criticism, the text has been
improved immeasurably. All these individuals have influenced this edition. We express our thanks
and appreciation to them. The publishers and authors gratefully acknowledge the suggestions
and also express their appreciation for all those reviewers who took the time to review the book.
We hope this support and response will continue in future.
We acknowledge our sincere thanks and regards to Prof. L.D. Arya, Professor of Power System
Reliability Engineering, Shri G.S. Institute of Technology and Science, Indore, Madhya Pradesh,
India, for his wholehearted encouragement, excellent contributions, and cooperation to complete
this prestigious book. I owe this moment of satisfaction with deep gratitude for his valuable time,
exhorting guidance, and valuable suggestions during the various stages of this book. His lifelong
experience in the field of power system reliability engineering and awareness of recent develop-
ments in these areas have been very useful to us during the writing of this book. No amount of
words can really suffice to express our sincere thanks to him for discussing the problems critically
and providing concrete suggestions.
We are heartily thankful to Prof. D.P. Kothari, Former Director and Professor of Power System
Engineering, Indian Institute of Technology, New Delhi, India, for his valuable guidance, dedicated
efforts, keen interest in developing the writing skills, mental satisfaction, encouragement, and help-
ful criticism at various stages of the book chapter preparation. His positive attitude, unassailable
optimism, constant encouragement, and invaluable assistance made it possible to bring out this
book in its present shape.
Prof. R. K. Saket would like to acknowledge sincere thanks and kind regards to his great Guru,
Prof. I.J. Nagrath, Former Deputy Director and Professor of Power System Engineering, Birla Insti-
tute of Technology and Science, Pilani (Rajasthan), India, who provided the historical opportunity
to join the PhD program under his valuable supervision and helped directly or indirectly wherever
needed during his research work and BITSian life.
Prof. R. K. Saket would like to express his heartily regards and dedication to his great father
Parinivutta Ram Das Saket Vardhan and beloved mother Parinivutta Terasita Devi who gave him
inspiration and perennial encouragement adventurously during his engineering education and
academic career. He will be failing his basic duties if he does not mention the golden soil of his
small village Dhaurahara, located in the Vindhya–Rewanchal region of Madhya Pradesh, India,
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xxxiv Acknowledgments

which was far away from the city and was lacking the basic amenities such as electricity, quality
education, drinking water, and road during his engineering education, even then he could rise to
this level, could publish his innovative research work in SCIE journals of all the global publishers,
and could write many books including prestigious RAMPS, Wiley-IEEE (USA).
Prof. P. Sanjeevikumar would like to express his heartily regards and dedication to his parents,
the late Mr. Padmanaban Sanjeevi and the late Mrs. Kamala Padmanaban, for their inspiration
and encouragement adventurously during his engineering education, academic engagements, and
research activities.
The editors would also like to thank the higher authorities, learned faculty colleagues, and
research scholars of the Indian Institute of Technology (Banaras Hindu University), Varanasi,
Uttar Pradesh, India, and the University of South - Eastern Norway, Norway, who have been very
generous and helpful toward us for maintaining a suitable atmosphere, providing manuscript
preparation support, and extending all facilities required for the completion of this prestigious
book. The editors, and authors acknowledge the financial support extended by Science and
Engineering Research Board (SERB); a statutory body of the Department of Science & Technology
(DST), Government of India; New Delhi, India under the sponsored research project sanction order
No. EEQ/2021/000177, and project code No.: R&D/IIT(BHU) Varanasi (SERB)/EE/21-22/09/348/
02.05.2022.
The editors are thankful to Ms. Victoria Bradshaw and Ms. Storry Aileen, Senior Editorial
Assistant for Electrical Engineering, Technology & Engineering Careers; Wiley – IEEE publishers
(USA) for handling peer-review process of the RAMPS proposal and for providing appreciable
improvement suggestions to enhance the book proposal contents. We would also like to acknowl-
edge Ms. Kavipriya Ramachandran, Managing Editor; Mr. Deenadayalu Govindanagaraj, Content
Refinement Specialist; Mr. Vasanth Koilraj, Support Service Administrator; and production staff of
Wiley – IEEE publishers (USA) for completing publication requirements and copyright agreement
formalities for contributors.
The editors are thankful to Prof. Bimal Kumar Bose, Department of Electrical Engineering and
Computer Science, the University of Tennessee, Knoxville, USA, and Prof. Frede Blaabjerg, Depart-
ment of Energy Engineering, Aalborg University, Aalborg, Denmark, for writing the foreword notes
on this book.
We want to express our affection to the future talents of next-generation Aanchal, Aakanksha,
Siddharth, Siddhant, Kishankumar, and Yomeka for providing emotional support and natural
happiness directly or indirectly throughout the publication process of this textbook.

18 January 2024 R. K. Saket

P. Sanjeevikumar
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1

Reliability Principles and Applications

Section 1
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3

Basic Principles and Scientific Importance of Reliability Theory

Aanchal Verma 1 , Akanksha Singh S. Vardhan 2* , Vanitha Bagana 3 , R. K. Saket 1 , and
P. Sanjeevikumar 4
1
Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
2
Department of Electrical Engineering, Shri GS Institute of Technology & Science, Indore, Madhya Pradesh, India
3
Department of Electronics and Communication Engineering, National Institute of Technology, Jamshedpur, Jharkhand, India
4
Department of Electrical Engineering, IT and Cybernetic, University of South-Eastern Norway, Porsgrunn, Norway

1.1 Introduction

Reliability, an age-old concept, has evolved into a contemporary discipline within modern engi-
neering and technology. Originating as an extension of probability theory, reliability theory found
its early applications in industries such as nuclear, electronics, and aerospace after World War II.
As engineering systems become more complex, the demand for elevated reliability has emerged
as an essential requirement. Reliability as a concept was primarily formulated for nonrepairable
technical systems due to the critical importance of initial failure in such contexts.
Over time, reliability studies expanded to encompass diverse engineering fields, both for systems
that are repairable and those that are not. The evolution of reliability theory traces back to World
War II, with formal evaluations emerging as a response to issues with German missiles. However,
the challenges of data scarcity, computational limitations, and a reluctance to employ probabilistic
techniques have progressively diminished.
In this chapter, our objective is to outline contemporary reliability evaluation techniques suit-
able for modern power systems while elucidating the significance of various reliability indices. As
reliability often intertwines with safety concerns, particularly in scenarios involving human lives,
it becomes integral to broader safety considerations (Saket et al. 2022).
Reliability is a pivotal factor in engineering design, influencing system effectiveness and account-
ing for constraints such as cost. The analysis also weighs factors like operating environment, oper-
ational time, and service quality. To quantitatively assess system utility, reliability indices rooted in
probability theory are essential. Qualitative judgments are insufficient for engineers planning and
designing systems.
Operating environment plays a substantial role in reliability modeling, affecting failure and repair
rates. Consideration of time in operation is imperative, as the probability of successful system oper-
ation diminishes over prolonged periods. Equally crucial is the quality of service, as the system’s
availability does not necessarily guarantee consumer satisfaction.
In essence, reliability can be described as the likelihood of a component or system fulfilling its
intended function adequately throughout the designated period of operation, contingent upon the
given operating environment.
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4 1 Basic Principles and Scientific Importance of Reliability Theory

1.2 Basic Concept of Reliability Engineering

Reliability is described as the probability that a component or system effectively carries out its
intended function within the designated timeframe and under the specified operating conditions.
The classical definition of reliability underscores four fundamental attributes: the likelihood of
success, adequate functionality, duration of operation, and prevailing operating conditions.
Reliability engineering delves into the exploration of equipment lifespan and failure. By applying
scientific principles and mathematical methodologies, it investigates the aging and failure patterns
of devices. The goal is to gain insights into failure mechanisms, leading to enhanced product designs
that promote longevity and mitigate the consequences of failure. The emphasis here is on design
improvement.
A pivotal aspect is that new system and product designs must demonstrate safety and reliabil-
ity before being manufactured and utilized industrially. This practice is widespread across various
engineering domains, encompassing both nonrepairable and repairable systems.
Electric power systems serve as a prime example of environments where an exceedingly high
level of reliability is anticipated (Bansal et al. 2002; Arya et al. 2000, 2001). The reliability concept
is often divided into two facets: security and adequacy. Adequacy concerns the presence of ample
facilities within the system to meet customer load demands. This encompasses energy generation
and the requisite transmission and distribution infrastructure. Probabilistic models are commonly
employed for adequacy assessment across different segments of the power system, like generation,
and load. In contrast, security pertains to the system’s capacity to respond effectively to internal
disruptions.
In engineering applications, concepts necessitate quantification, translating into procedures
that establish correlations between various attributes and numerical scales for measurement. This
necessitates the conversion of reliability into measurable quantities through suitable definitions.
Reliability is rooted in the mathematical realm of probability, and its practical implications can
vary significantly across different contexts.
Electric power systems exemplify environments where exceptional reliability is imperative. For
numerous power systems, customers typically experience an average of 2–3 hours of interruptions
per year, with even shorter durations in certain regions (Bharti et al. 2019; Kumar et al. 2020; Saket
et al. 2006, 2007, 2009). Industries with stringent requirements also demand a high degree of reli-
ability. Consequently, reliability has consistently been a pivotal factor in the design, arrangement,
maintenance, and operation of electrical power systems. Reliability theory’s formal concepts and
methodologies have been extensively employed to evaluate power system reliability from various
angles.

1.2.1 The Role of Reliability Engineering

The chapter also delves into the primary challenges and potential solutions presented in this book.
Additionally, it explores the contemporary context, including the implications of digital transfor-
mation and the influence of the Industrial Internet of Things (IIoT) on the functions of reliability
engineers. The central role of the power reliability engineer (PRE) is to elevate the reliability of
essential assets, which encompass fixed power plants, earth-moving equipment, power-related
components, and the associated utilities and facilities. Reliability is established through the inte-
gration of reliability engineering principles into the asset management (AM) system, alongside
a rigorous and organized assessment of physical failures, asset conditions, and component relia-
bility data. Reliability engineering is a cornerstone discipline that ensures products, systems, and
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1.2 Basic Concept of Reliability Engineering 5

services operate as intended for their intended lifespan. This comprehensive field plays a pivotal
role in numerous industries, contributing to better safety, cost-effectiveness, enhanced productiv-
ity, customer satisfaction, and adherence to regulations. The importance of reliability engineering
is described as follows:

1. Improving Safety:
Reliability engineering serves as a safeguard against catastrophic failures. In industries where
human lives are at stake, like modern power systems, aviation, and medical devices, reliable
products are essential. Consider an aircraft: A reliable one is less likely to crash, instilling confi-
dence in passengers and airline companies alike. Similarly, medical devices, such as life-support
systems, must function without fail to prevent harm to patients.
2. Reducing Costs:
Reliable products and systems require fewer maintenance and repair interventions, translating
into substantial cost savings for businesses. Think about a reliable car; it demands fewer repairs
compared to an unreliable counterpart. The same principle applies to industries like energy,
where a reliable power plant requires less maintenance than an unreliable one, minimizing
downtime and costs.
3. Increasing Productivity:
Operational efficiency is a hallmark of reliability engineering. When products and systems oper-
ate reliably, they contribute to enhanced productivity. A reliable factory machine, for instance,
can churn out more products per hour than an unreliable one. This efficiency boost directly
impacts the bottom line and overall business success.
4. Enhancing Customer Satisfaction:
Customers are naturally drawn to products and services they can rely on. Reliability fosters
customer satisfaction, which, in turn, leads to increased sales and higher profits. Imagine a cus-
tomer purchasing a new car and experiencing minimal issues over time; they are more likely to
become brand advocates and recommend the product to others.
5. Regulatory Compliance:
Numerous industries are subject to stringent regulations that demand adherence to specific reli-
ability standards. For instance, the aviation industry operates under strict regulations governing
aircraft reliability. Compliance is crucial not only for ensuring safety but also for avoiding legal
repercussions and maintaining the integrity of the industry.

1.2.2 Key Features of Reliability Engineering

1. Failure Analysis:
Failure analysis is at the core of reliability engineering. By identifying the root causes of failures,
engineers can develop strategies to prevent similar issues in the future. This proactive approach
not only enhances product performance but also contributes to long-term cost savings.
2. Risk Assessment:
Reliability engineering involves quantifying risks associated with potential failures. This infor-
mation helps prioritize preventive measures, ensuring that resources are directed where they are
most needed. Effective risk assessment minimizes surprises and contributes to a well-structured
risk management strategy.
3. Probabilistic Modeling:
Probabilistic modeling leverages mathematical models to predict the likelihood of failure.
These predictions inform decision-making processes regarding design, maintenance, and
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6 1 Basic Principles and Scientific Importance of Reliability Theory

testing strategies. By quantifying potential failures, businesses can allocate resources more
effectively and make informed choices.
4. Life Cycle Management (LCM):
Reliability engineering extends throughout the entire life cycle of a system or product. From
design to disposal, engineers engage in activities such as reliability testing, maintenance plan-
ning, and failure data analysis. This comprehensive approach ensures optimal performance and
longevity.

1.3 Scientific Importance of Reliability in Modern Technology

The electric power supply system’s reliability can be described as its ability to consistently offer
users uninterrupted and satisfactory service. This criterion of quality encompasses the mainte-
nance of power supply frequency and voltage within defined tolerances. Certainly, the level of
reliability customers encounter will naturally vary depending on their particular geographic loca-
tions. Furthermore, different segments of the power network, including distribution, generation
and transmission systems, will demonstrate varying levels of reliability.
It is obvious that in order to meet the expected customer-level reliability, each of these individual
components and systems must maintain an elevated level of reliability. The concept of adequacy
revolves around ensuring that the system possesses ample facilities to meet consumer demands.
This encompasses the generation of adequate energy and the interconnected transmission and dis-
tribution networks responsible for efficiently conveying this energy to end consumers. Adequacy
predominantly pertains to static conditions and does not encompass system disturbances.
In contrast, security concentrates on the system’s capacity to effectively respond to disturbances
originating within itself. Security is intricately linked to the system’s ability to counteract diverse
disruptions it may encounter, spanning from localized to widespread effects and even encompass-
ing the loss of significant generation and transmission facilities.
Dividing the system into its primary functional zones provides a pragmatic approach. These
zones encompass the composite generation, generation systems and transmission (bulk power)
systems, and distribution systems. Each major functional zone can be further subdivided, facil-
itating the focused analysis of specific subsets of the issue. Notable subzones include individual
generating stations, substations, and protection systems.
Establishing the required level of system-generating capacity to guarantee a reliable power supply
is a crucial component of power system planning and operation (Huang et al. 2019; Gandini 1990;
Freixas and Pons 2008). This intricate issue can be divided into two separate domains: static and
operational capacity demands. In power system planning, the static capacity domain focuses on the
long-term assessment of the system’s overarching need, while the operating capacity domain deals
with the short-term evaluation of the actual capacity required to meet a specific load demand. Both
of these considerations must be taken into account during the planning phase, especially when
exploring alternative facilities. After decisions are made, the short-term requirement becomes a
matter of operational concern.
In the context of static capacity evaluation, a fundamental parameter for generating units is the
probability of encountering a forced outage at a future point in time. This probability, referred to as
unit unavailability in engineering systems and historically referred as the unit-forced outage rate
in power system applications, serves as a fundamental element for evaluating the system’s static
capacity.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1.4 Basic Concept of Probability Theory 7

1.4 Basic Concept of Probability Theory

The Bayesian approach has an important aspect related to conditional probability for assessing
the reliability of power system components. The Bayes theorem is introduced, along with a brief
explanation. A sample space denoted as S, let Ai be a series of exhaustive and mutually exclusive
events that P(Ai ) > 0, ∀i . Within “S” sample space, consider an event “B” where the probability of
event “B” occurring, P(B) > 0. Then,
P(A )P(B∕Ai )
P(Ai ∕B) = ∑ i (1.1)
i P(Ai )P(B∕Ai )

Proof: Consider Ai as an event within the set S, and B is event where P(B) > 0. Subsequently, in
accordance with the definition of conditional probability,
P(Ai ∩ B)
P(Ai ∕B) =
P(B)
and
P(B)P(Ai ∕B) = P(Ai ∩ B) (1.2)

P(Ai ∩ B)
P(B∕Ai ) =
P(Ai )
P(Ai )P(B∕Ai ) = P(Ai ∩ B) (1.3)
From (1.1),
P(Ai ∕B)P(B) = P(Ai )P(B∕Ai )

P(Ai )P(B∕Ai )
P(Ai ∕B) =
P(B)
∑
P(B) = P(Ai )P(B∕Ai ) (1.4)
i

We now assert that, given the event (Ai ) being mutually exclusive and collectively exhaustive within
the set S, it can be concluded that ∪i Ai = S and that the individual Ai events are nonoverlapping or
disjoint.
B = B ∩ S = B(∪i Ai ) = ∪i (B ∩ Ai )
Thus,
∑
P(B) = P[∪i (Bi ∩ Ai )] = P(B ∩ Ai )
i

Thus, by axiom of probability

∑
P(B) = P(Ai )P(B∕Ai ) (1.5)
i

Exercise 1.1: Among a collection of circuit breakers, Factory F1 produces 1000 of them, with 20
being faulty. Factory F2 manufactures 4000 circuit breakers, and 40 of them are defective. Factory
F3 contributes 5000 circuit breakers, with 50 being faulty. All these circuit breakers are combined
into a single stockpile. What is the probability that a defective circuit breaker selected at random
being produced in Factory F1 ?
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8 1 Basic Principles and Scientific Importance of Reliability Theory

Solution:
Consider A1 , A2 , and A3 as events representing the selection of circuit breakers produced by F1 , F2 ,
and F3 , respectively. The overall number of products is given by:

N(S) = 5000 + 4000 + 1000 = 10,000

1000 1
p(A1 ) = =
10,000 10,
4000 2
P(A2 ) = =
10,000 5
5000 1
p(A3 ) = =
10,000 2
Let B represent the event where the selected circuit breakers are found to be faulty.
Then
20 1
p(B∕A1 ) = =
1000 50
40 1
p(B∕A2 ) = =
4000 100
50 1
p(B∕A3 ) = =
5000 100
To calculate P(A1 ∕B) using Bayes’ theorem:
P(A )P(B∕A1 )
P(A1 ∕B) = ∑3 1
i=1 P(Ai )P(B∕Ai )
( )
1 1
10 50 2
= ( ) ( ) ( ) =
1 1 2 1 1 1 11
+ +
10 50 5 100 2 100

Exercise 1.2: For a power transformer, rejection can occur due to either a flawed design or other
reasons. The chance of having a flawed design is 0.1, and the probability of rejection specifically
due to this design issue is 0.95, while it is 0.45 for rejections arising from different causes. In the
event of a power transformer’s rejection, what is the probability that this rejection is specifically
due to a design issue?
Solution:
Consider events F1 and F2 to represent whether the design is flawed or not, and let event A represent
the rejection of the power transformer.
Then, P(F1 ) = 0.1, P(F2 ) = 0.9 and P(A∕F1 ) = 0.95, P(A∕F2 ) = 0.45,
P(F1 ∕A) = P[rejection due to faulty design]
( ) ( )
P F1 P A∕F1 0.1 × 0.95
( ) ( ) ( ) ( ) = = 0.19
P F1 P A∕F1 + P F2 P A∕F2 0.1 × 0.95 + 0.9 × 0.45

Exercise 1.3: Three machines are involved in manufacturing wattmeters for power system
applications, with 10,000, 20,000, and 30,000 watts, respectively. Defectives are produced by these
machines at rates of 1%, 2%, and 1%, respectively. What is the probability of a randomly chosen
defective wattmeter is manufactured by the third machine?
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1.5 Basic Concepts of System Reliability 9

Solution:
Assume that A1 represents a wattmeter manufactured by the first machine, A2 by the second
machine, A3 by the third machine, and B denotes the case where the selected wattmeter is
identified to be faulty.
Hence n(S) = 30,000 + 10,000 + 20,000 = 60,000
( ) 10,000 1
P A1 = =
60,000 6
( ) 20,000 1
P A2 = = ;
60,000 3
( ) 30,000 1
P A3 = =
60,000 2
( ) ( ) ( )
B 1 B 2 B 1
P = ;P = ;P =
A1 100 A2 100 A3 100
( )
A
To find the P B3 by Bayes theorem:
( ) (B)
( ) P A3 P A
A3
( ) (B)
3
P = ∑
B 5
i=1 P Ai P A
3
( )
1 1
×
= ( 2 100
) ( ) ( )
1 1 1 2 1 1
× + × + ×
6 100 3 100 2 100
1 1 600 3
= ( ) ( ) ( ) = × =
1 2 1 200 (1 + 4 + 2) 8
+ +
600 300 200

1.5 Basic Concepts of System Reliability

Based on the fundamental concept of reliability, the key characteristics encompass probability,
proper functioning, duration, and operational context of the system. The reliability of both com-
ponents and systems frequently hinges on the duration they have spent in service. Hence, in the
context of reliability research, an essential element to consider is the failure-time distribution,
which portrays how the time until component failure is distributed under defined environmen-
tal circumstances. Components within a power system might be categorized as nonrepairable or
repairable. A component earns the designation of “repairable” if it can be restored to its initial state
post-failure without disrupting the system’s operation.
Key concepts in power system reliability include availability, maintainability, mean up time
(MUT), loss of load probability (LOLP), mean time to repair (MTTR), mean down time (MDT),
mean time to failure (MTTF), and mean time between failures (MTBF). A practical approach to
describing this distribution is through its corresponding instantaneous failure rate.

1.5.1 Bathtub Curve

The bathtub curve is a graphical representation in Figure 1.1 that illustrates how the rate of failure
of an item changes over time. This curve demonstrates the initial period of an item’s life, known as
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10 1 Basic Principles and Scientific Importance of Reliability Theory

the “infant-mortality period,” during which the failure rate goes down. Afterward, it reaches a sta-
ble intrinsic failure rate level, maintaining this rate during its useful life. As the item continues to
age and surpasses its designed lifespan, the curve then shows an increase in the failure rate, known
as wear-out failures. This curve is constructed by mapping three distinct phases: (i) The initial
infant-mortality failure rate upon introduction, (ii) the ongoing random failure rate during its oper-
ational lifespan, and (iii) the wear-out failure rate as the product exceeds its designated lifespan.
The hazard function represents the pace at which units still in operation experience failure. Reli-
ability analysts frequently employ this function to elucidate the failure patterns of devices and sys-
tems. Instantaneous rate of failure i.e. hazard function, may be decreasing, increasing, or constant.
Usually in the life of a component/device/system all three types of failure rates are predominant
in three stages of the system. These three failure rate are described in three different regions called
infant-mortality period, random failures, and wear-out failures, as shown in Figure 1.1. This curve
also shows the combination of three regions known as 𝜆-curve/bathtub curve. Curves (a), (b), and
(c) represent endogenous failure rates, exogenous failure rate, and wear-out failure rates, respec-
tively. Curve (d) represents total failure rate. Even in the early stages, wear-out failure could be
minimal, but it emerges as the dominant factor in curve (c) as it enters region III. Similarly, even
in debugging period random failures may be present. The incorporation of the hazard function is
derived from the observation that devices collectively display a hazard pattern that mirrors a bath-
tub over the course of their lives. The shape of the curve is meant to demonstrate the concept that
the aging process in a population of devices goes throughout distinct phases. During initial phases
of the devices’ lifespans, failures happen at a relatively elevated rate. This initial phase of high fail-
ure rates is often attributed to weaknesses in some devices within the population, resulting from
manufacturing defects, material flaws, or other physical irregularities. Following the initial phase,
referred to as the infant-mortality period, the device population transitions into a stage known as
the functional life period, characterized by a relatively low and stable hazard function. Ultimately,
when individuals in the population approach the later stages of their lives, survivors encounter a
rising failure rate as a result of accumulated wear and tear. The hazard model, or the bathtub curve,
can be dissected into three distinct sections:
(i) Initial period,
(ii) Constant hazard model, and
(iii) Wear-out region.

Region I Region II Region III

Infant mortality Normal life Wear out
Decreasing failure rate

Increasing failure rate

Instantaneous failure rate

γ<1 Constant failure rate γ>1

γ=1
0 Time t

Figure 1.1 Bathtub curve.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1.5 Basic Concepts of System Reliability 11

1.5.1.1 Initial Period (debugging/Burn-In/Infant-Monolith Period)

Initial period where the failure rate is decreasing is known as debugging period or infant-mortality
period. Failure rate is decreasing in this period because as time passes it is design/manufacturing
defects are removed. Failure in this period is due to endogenous reasons. That is internal regions
are mainly responsible for the system’s failure rate. During the burn-in phase, a diminishing hazard
function implies a decreased likelihood of an impending failure as time progresses. In general, this
is best suited model for the initial phase of a component’s operation, characterized by a hazard
function that consistently diminishes over its entire lifespan.

1.5.1.2 Constant Hazard Model

The second stage represents the device’s useful period, during which the failure rate remains rela-
tively stable. Failures in this region are also known as random failures. Here in this useful period,
failures are by chance and reasons are exogenous. Reasons for failures are external, e.g. overstress-
ing, accident, etc. The constant hazard model is relevant to components in which the probability of
failure within a given interval remains consistent throughout their entire lifespan. During the mid-
dle phase of their life, there is no anticipation of unusual clusters of failures, and thus, the so-called
chance failures with relatively low and unchanging hazard rates take precedence.

1.5.1.3 Wear-Out Region

In the third region which is known as wear-out region, aging effect dominates, and increasing fail-
ure rate is observed due to wear-out. The rising hazard function basically demonstrates that these
components are gradually becoming more susceptible to failure as they age, particularly within a
group of the same size that has endured for the same duration, is expressed as:
h(t) = f (t)∕R(t)
where R(t) = reliability function;
f (t) = probability density function;
h(t) = hazard function of the system or device.

1.5.1.4 Burn-In Screening

This is a design and test procedure to filter out the defective, weak or low quality product from the
population, so that they do not turn-out at a later stage during application as shown in Figure 1.2.
This reduces the risk at higher level and minimizes the possibility of rework. Ideally, only those

Stage of Start-Up /
Normal operation End of life
service life commissioning
Failure rate Decreased Quasi-constant Increasing
characteristics failure rates failure rates failure rates

Root causes “Infant mortality” Random failures Wear-out

Overall failure rate (d)

Failure rate

(a)
(c)
(b)

Time

Figure 1.2 Representation of hazard function.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12 1 Basic Principles and Scientific Importance of Reliability Theory

components must be used in the assembly of the system, which have undergone burn-in screening
test already, because it considerably enhances the system reliability.
Burn-in is an essential step if the output of the production is not matching the reliability specifi-
cations. It must be ensured that the burn-in stress or temperature is not so high that the units under
test are damaged. Burn-in screening is an expensive process and thus the time period for which the
screening is done is limited. How long a burn-in screening is run depends on the desired level of
reliability. Some of the related references are (Saket 2011) and (Saket 2013).

1.5.2 Reliability Functions

To delve into this concept further, a component taken into consideration that commences its oper-
ation at a specific time, for example, t = 0. Here f (t) represents the probability density associated
with the time until the component fails. This probability density (t) indicates the likelihood of the
component failing in the range of time t and t + Δt, denoted by the function f (t)Δt.
Consequently, the probability of the component failing within the range from 0 to time t can be
t
calculated using the cumulative distribution function t, F(t) = ∫0 f (x)dx. The reliability function,
which denotes the probability of the component enduring until time t, is articulated as R(t) = 1 −
F(t). This function complements the cumulative distribution by indicating the likelihood of survival
up to a certain time t.
The probability of a successful outcome, which is equivalent to the probability of the component
failing in the range of t and t + Δt, is given by F(t + Δt) − F(t). The probability of failure in this
interval, considering the component’s survival until time t is articulated as:
F(t + Δt) − F(t)
(1.6)
R(t)
Dividing by Δt, we determine the mean failure rate within the time interval from t to t + Δt, given
that the component has survived up to time t:
F(t + Δt) − F(t) 1
• (1.7)
Δt R(t)
By approaching the limit with Δt converging to zero, we obtain the instantaneous failure rate,
commonly well known as the failure rate:
F ′ (t) 1 dF
Z(t) = = (t) (1.8)
R(t) R(t) dt
dF(t)
Finally, observing that the PDF =f (t) = dt
,
f (t)
we get the relation Z(t) = (1.9)
1 − F(t)
Now, let us derive a pivotal relationship that conveys the density of failure times using the fail-
ure rate function. By utilizing the relationship between the reliability function R(t) = 1 − F(t) and
consequently, F ′ (t) = −R′ (t), the expression for Z(t) is formulated as follows:
R′ (t)
Z(t) = − (1.10)
R(t)
After successfully solving this differential equation for R(t), we acquire
t
log R(t) = − Z(x)ax
∫0
t
R(t) = e∫0 Z(x)ax (1.11)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1.5 Basic Concepts of System Reliability 13

Table 1.1 Relation between probabilistic functions.

— F(t) f (t) R(t) h(t)

t
l
F(t) — f (t)dt 1 − R(t) 1 − e− ∫0 h(t)dt
∫0
dF(t) dR(t) l
f (t) — − h(t)e− ∫0 h(t)dt
dt dt
∞
t
R(t) 1 − F(t) f (t)dt — e− ∫0 h(t)dt
∫0

dF(t) f (t) d
h(t) ∕[1 − F(t)] ∞ − log R(t) —
dt dt
f (t)dt
∫0

[ ]
1
Furthermore, by utilizing the relationship f (t) = Z(t)R(t), we obtain f (t) = Z(t) exp − ∫0 Z(x)ax .
This formula is referred to as the general equation for the distribution of failure times. Table 1.1
displays the relationships between reliability functions.
For constant failure rate, the following relations are given for reliability assessment of the power
components.

(1) Reliability function

t
R(t) = e− ∫o h(t)dt
put h(t) = 𝜆 (constant failure rate of the components)
∞
R(t) = e− ∫o 𝜆dt

⇒ R(t) = e−𝜆t
R(t) = (1 − Pf )

The failure probability of power component

Pf = [1 − R(t)]
⇒ Pf = (1 − e−𝜆t )

(2) The probability density function (PDF) of power component

−dR(t)
f (t) =
dt
⇒ f (t) = 𝜆e−𝜆t

(3)
∞
MTTF = R(t)dt
∫o
∞
MTTF = e−𝜆t dt
∫o
1
MTTF =
𝜆
⇒ R(t) = e−𝜆t
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
14 1 Basic Principles and Scientific Importance of Reliability Theory

1.5.3 Mean Time to Failure of Component

Our focus frequently lies in determining the average time until a component fails (MTTF), rather
than focusing on the specifics of each failure. This parameter is assumed to remain consistent for
all components that share identical designs and function under same operating conditions. With
life test data available for a group of N items and their corresponding failure times as t1 , t2 , … tn ,
the MTTF is articulated as follows:
1∑
n
MTTF = t (1.12)
N i=1 i
Although when a component’s characteristics are captured through its reliability function, the
MTTF is calculated as the arithmetic mean of the random variable T, which signifies the duration
until the component experiences a failure.
Henceforth,
∞
MTTF = E[T] = tf (t) dt (1.13)
∫0
dF(t)
But f (t) = dt
= − dR(t)
dt
.
Hence,
∞ ∞
MTTF = − t dR(t) dt = −[tR(t)]0 ∞ + R(t) dt
∫0 ∫0
∞
= R(t) dt (∵ R(∞) = 0)
∫0
∞
MTTF = R(t) dt (1.14)
∫0
The MTTF is determined by applying the Laplace transform to R(t),
∞ ∞
MTTF = R(t) dt Lt R(t) dt (1.15)
∫0 t→∞ ∫0
t
However, Lt ∫0 R(x) dx = Lt R(s),
t→∞ s→0
t
Thus, MTTF = Lt R(s) ∫0 t2 f (t) dt − (MTTF)2
s→0
[ ]
Remark: Var(T) = E[T − E(T)]2 = E T 2 − [E(T)]2 .

1.5.4 Additional Examples

Exercise 1.4: In a composite power system, where R(t) represents the reliability function:
√
R(t) = 1∕e 0.001t
,t ≥ 0

a) Calculate the reliability for a mission lasting 50 hours.

b) Demonstrate that the hazard rate exhibits a decreasing trend.
c) Determine the reliability for a 50-hour mission, considering a wear-in period of 10 hours.
d) Find the design life required to achieve a reliability of 0.95, considering 10 hours of wear-in
period.
Solution: √
The system’s reliability function is: R(t) = 1∕e 0.001t , t ≥ 0
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1.5 Basic Concepts of System Reliability 15

a)
√
0.001 × 50
R(50) = 1∕e = 0.9512
b)
( √ )
√
−R′ (t) 1 0.001
𝜆(t) = ⇒ −e 0.001t
× √ × −
R(t) e 0.001t t
√
0.001
⇒ , which is a decreasing function of t.
t
c)
( )
( ) R T0 + t
R t∕T0 = ( )
R T0
√
R(60) e 0.001×10
R(50∕10) = = √ = 0.8651
R(10) e 0.001 × 60
d)
( )
R tD + 10
= 0.95
R(10)
( ) √
R tD + 10 = R(10) × 0.95 = 1∕e 0.001×10 × 0.95
√( )
⇒ tD + 10 × 0.001 = 0.15129
tD = 12.89 hours.

Exercise 1.5: In the context of a protective relay circuit with an MTBF of 0.8 years and under the
assumption of random failures, address the following inquiries:
i) Determine the likelihood that the circuit will remain operational without experiencing failure
over the course of 1 year.
ii) Calculate the probability of encountering over two failures within the initial year.
iii) Ascertain the anticipated count of failures anticipated annually.
Solution:
Given the random occurrence of failures, the count of failures within a time interval of “t” is mod-
−𝜆t
eled by a Poisson distribution, given by P[N(t) = n] = e n!(𝜆t)n , n ≥ 0, where 𝜆 signifies the rate of
failure. Consequently, the time gaps between failures are distributed exponentially with an average
of 1∕𝜆.
Now MTTF = 1∕𝜆 = 0.8 year
∴ 𝜆 = 1.25 per year
i)
e−𝜆 𝜆̂
0
P[N(1) = 0] = − e−125 = 0.2865
0!
ii)
[ ]
𝜆̂ 𝜆̂ 𝜆̂
0 1 2
P[N(1) > 2] = 1 − e −𝜆
+ +
0! 1! 2!
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
16 1 Basic Principles and Scientific Importance of Reliability Theory

[ ]
(1.25)2
⇒ 1 − 0.2865 1 + 1.25 + = 0.1315.
2
iii)

E[N(t)] = 𝜆t

E[Annual failure count] = 𝜆 = 1.25.

Exercise 1.6: In a redundant power system with “n” independent, identical components, each
having a constant failure rate 𝜆, the formula to compute the MTTF is expressed as: MTTF =
1 ∑ n (−1)i−1
n

𝜆 C1 i
. What is the minimum system reliability when the failure rate is 𝜆 = 0.02 per hour?
i=1

Solution:
If Rs (t) is the reliability of system expressed as Rs (t) = 1 − (1 − e−𝜆t )n where the reliability of indi-
∑n
n ∑n
vidual components is represented as e−𝜆t = 1 − (−1)i Ci e−i𝜆t = − (−1)i−1 n Ci e−i𝜆t
i=0 i=0
∞
∴ MTTF = Rs (t) dt
∫0
∑n ∞
1 ∑ n (−1)i−1
n
= (−1)i−1 n C1 e−i𝜆t dt =
i=1
∫0 𝜆 i=1 Ci i

∑
2 i−1
[ ]
When n = 2, MTTF = 200 × 2 Ci (−1)i = 200 2 − 12 = 300
i=1

∑3
3 i−1
( )
3 1 11
When n = 3, MTTF = 200 × Ci (−1)i = 200 3 − 2
+ 3
= 200 × 6
= 366
i=1
Hence the required minimum value of n = 4.
Exercise 1.7: The power transmission system comprises five consecutive segments within its pro-
tection scheme. Each individual segment has a consistent failure rate of 0.23 failures per year, with
the failures of segments being independent from each other. Compute the system’s MTTF and its
reliability function.
Solution:
The failure rate remains constant, and the failure times are distributed exponentially with a param-
eter value of 𝜆 = 0.23.
For a single segment, the reliability function is represented as R1 (t) = e−(0.23)t .
The system comprises five consecutive segments. As these segments experience independent fail-
ures, The system’s overall reliability is represented as follows:
R(t) = R1 (t)R2 (t)R3 (t)R4 (t)R5 (t)
= e−(0.23)t .e−(0.23)t .e−(0.23)t .e−(0.23)t .e−(0.23)t = e−(1.15)t
∞ ∞ [ −(1.15)t ]
−(1.15)t e
MTTF = R(t)dt = e dt = = 0.8696
∫0 ∫0 −1.15

Exercise 1.8: A newly acquired distribution transformer exhibits a constant daily failure rate of
0.03 during continuous operation, along with a fixed daily repair rate of 0.2.

i) Calculate the interval availability for the initial 30 days and determine the steady-state avail-
ability.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1.6 Conclusion 17

ii) Establish the steady-state availability when introducing a stand-by unit, assuming zero failures
during stand-by mode.
iii) In the scenario where both units are operational, ascertain the steady-state availability.
Solution:
𝜇 = 0.2 per day, 𝜆 = 0.03 per day
a)
𝜇 𝜇
A1 (T) + + [1 − e−(𝜆+𝜇)T ]
𝜆 + 𝜇 (𝜆 + 𝜇)2 T
0.2 0.03
A1 (30) + + [1 − e−(0.03+0.2)30 ]
0.03 + 0.2 (0.03 + 0.2)2 X30
= 0.8696 + 0.0031 + 0.8727
𝜇
A(∞) = = 0.8696
𝜆+𝜇
b) In the stand-by redundant system,
𝜆𝜇 + 𝜇 2
Ai (∞) =
𝜆2 + 𝜆𝜇 + 𝜇 2
0.03 × 0.2 + 0.22
=
0.2 + 0.3 × 0.2 + 0.032
2

0.046
= = 0.9808
0.0469
c) In the setup of an active redundant system
Ai (∞) = 1 − [1 − A(∞)]2
⇒ 1 − [1 − 0.8696]2
⇒ 0.9830

1.6 Conclusion
The analysis provided in this chapter creates a framework for correlating the reliability of power
systems to the reliability of their individual components across various equipment designs. The uti-
lization of system configurations based on binary component states and independent components
allows for an approach to reliability analysis that simplifies the complex system. It is imperative
to focus on component-level reliability due to the clear and well-defined impact of component’s
reliability on the reliability of overall system. Particularly in the context of composite power system
designs, the ability to independently assess the reliability of individual components is crucial for
achieving elevated levels of overall reliability.
The chapter offers practical examples to showcase the application of these methods, illustrat-
ing both the advantages and limitations of various techniques. These practical cases are supported
by real-life examples taken from the author’s substantial experience in academia, research, and
consulting. The chapter provides a thorough exploration of fundamental concepts in probability
theory, the use of redundant systems to enhance reliability, evaluations of power component struc-
tures with a focus on reliability, application of the Hazard model for failure analysis of components,
and the presentation of various probability distributions related to reliability. Solved and unsolved
numerical examples are included in each section and subsection, offering a comprehensive and
illustrative view of the concepts discussed in the chapter.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
18 1 Basic Principles and Scientific Importance of Reliability Theory

References

Arya, L.D., Choube, S.C., and Saket, R.K. (2000). Composite System Reliability Evaluation Based on
Static Voltage Stability Limit, 133–139. Journal-Institution of Engineers India Part EL Electrical
Engineering Division.
Arya, L.D., Choube, S.C., and Saket, R.K. (2001). Generation System Adequacy Evaluation Using
Probability Theory, 170–174. Journal-Institution of Engineers India Part EL Electrical Engineering
Division.
Bansal, R.C., Bhatti, T.S., and Kothari, D.P. (2002). Discussion and closure of “bibliography on the
application of probability methods in power system reliability evaluation”. IEEE Transactions on
Power Systems 17 (3): 924.
Bharti, O.P., Saket, R.K., and Nagar, S.K. (2019). Reliability assessment and performance analysis of
DFIG-based WT for wind energy conversion system. International Journal of Reliability and Safety 13
(4): 235–266.
Freixas, J. and Pons, M. (2008). Identifying optimal components in a reliability system. IEEE
Transactions on Reliability 57 (1): 163–170.
Gandini, A. (1990). Importance and sensitivity analysis in assessing system reliability. IEEE
Transactions on Reliability 39 (1): 61–70.
Huang, X., Coolen, F.P.A., Coolen-Maturi, T., and Zhang, Y. (2019). A new study on reliability
importance analysis of phased mission systems. IEEE Transactions on Reliability 69 (2): 522–532.
Kumar, S., Saket, R.K., Dheer, D.K. et al. (2020). Reliability enhancement of electrical power system
including impacts of renewable energy sources: a comprehensive review. IET Generation,
Transmission & Distribution 14 (10): 1799–1815.
Saket, R.K. (2011). Reliability evaluation of defence support systems. In: Innovations in Defence Support
Systems-2: Socio-Technical Systems, vol. 338 (ed. L.C. Jain, E.V. Aidman, and C. Abeynayake),
241–286. Springer-Verlag.
Saket, R.K. (2013). Design aspects and probabilistic approach for generation reliability evaluation of
MWW based micro-hydro power plant. Renewable and Sustainable Energy Reviews 28: 917–929.
Saket, R.K., Bansal, R.C., and Singh, G. (2006). Generation capacity adequacy evaluation based on peak
load consideration. The South Pacific Journal of Natural and Applied Sciences 24 (1): 38–44.
Saket, R.K., Bansal, R.C., and Singh, G. (2007). Reliability evaluation of power system considering
voltage stability and continuation power flow. Journal of Electrical Systems 3 (2): 48–60.
Saket, R.K., Bansal, R.C., and Singh, G. (2009). Power system components modelling and reliability
evaluation of generation capacity. International Journal of Reliability and Safety 3 (4): 427–441.
Saket, R.K., Kothari, D.P., and Nagrath, I.J. (2022). Modern Power System Analysis. Tata McGraw-Hill
Publishing Company.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19

Bayesian Approach for Reliability Evaluation and Remaining Useful

Life Prediction
Debasis Jana, Suprakash Gupta, and Deepak Kumar
Department of Mining Engineering, Indian Institute of Technology (BHU) Varanasi (UP), India

2.1 Introduction
The quality of trustworthiness of any system, product, or service over a specified period and
conditions is termed reliability. This quality of any system, product, or service is inherently
uncertain and stochastic. Engineers and researchers are facing difficult challenges with accessing
the evaluation of the reliability of any complex system. There are relevant statistical inference
methods are available for reliability analysis. For example, the evaluation of hardware reliability
can be approached through the reliability block diagrams (RBD), fault tree analysis (FTA), Markov
and semi-Markov models (MKV), and Petri Nets models. The models like Gaussian processes,
vector regression, and the Markov-modulated Poisson process are used in the software reliability
study. ATHEANA, CREAM, and SPAR-H are utilized for the human reliability analysis. Every
approach to assessing reliability has a unique set of benefits and drawbacks that are inherently
associated with it. Due to the dependencies between interconnected components or factors
and environmental effects, evaluation of reliability using RBD, and FTA is difficult because of
the restriction of binary variables. Markov and Petri net models have problems like state space
explosion.
Power system reliability is the ability of an electrical power system to deliver electricity consis-
tently and without interruptions to its customers. It is a measure of how dependable and resilient
the power system is in providing electricity to meet the demands of consumers. Reliability in power
systems is crucial because interruptions in the power supply can have significant consequences for
various sectors, including residential, commercial, industrial, and critical infrastructure. Power
outages can result in economic losses, inconvenience, compromised safety, and disruptions in
essential services. Ensuring high reliability in the power sector is very important. For ensuring
power system reliability, several key factors like adequate generation capacity, robust transmission,
and distribution infrastructure, effective system planning, fault detection and rapid response,
redundancy and backup systems, maintenance and asset management are considered. Power
system reliability is typically measured using metrics such as the Customer Average Interruption
Duration Index (CAIDI), System Average Interruption Duration Index (SAIDI), and System
Average Interruption Frequency Index (SAIFI). These metrics quantify the customers’ average
duration and frequency of power outages.
The reliability estimation of a complex system is challenging and often requires the incorporation
of multiple techniques and multiple sources of data for building a credible statistical model. Reli-
ability evaluation using a probabilistic graphical model like Bayesian network (BN) is a research
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20 2 Bayesian Approach for Reliability Evaluation and Remaining Useful Life Prediction

interest in recent days that is attracting considerable attention from researchers and reliability engi-
neers. BN is an acyclic-directed graphical model that is capable of characterizing and analyzing
uncertainty effectively, making it a significant problem-solving model in real-world encountered
problems. Within the literature, the application of BNs extends to many fields including risk anal-
ysis, fault diagnosis, safety engineering, and uncertainty prediction.

2.2 Bayesian Network

BNs are graphical models to describe probabilistic relationships within a group of random variables.
BNs are a directed acyclic graph, consisting of points (represent variables) and directed edges con-
necting these points. In BN, the arc-starting nodes are called the parent or cause nodes, and the
arc-ending node is the child or effect node. For example, if an arc starts from node A and ends in
node B, then node A is the parent of node B and B is the child of A. A root node is a node without
a parent, and a node without any child node is considered a leaf node. A typical BN is shown in
Figure 2.1. In Figure 2.1, the “parent or cause nodes” is the node at the tail (nodes E1, E2, E3, and
E4), and the “child or effect node” is the node at the head (node IE1 and IE2).
The conditional probability is the main-stay of a BN and is defined as the likelihood of occurrence
of an event depending on a previous event occurrence. Each node in a BN has a CPT that illustrates
how the state of the node is characterized by every state of its parents’ conditional probability.
The joint probability distribution is also calculated from the CPT. The conditional probability is
mathematically expressed as:
P(A).P(B∣A) P(A, B)
P(A∣B) = = (2.1)
P(B) P(B)
In the equation, the hypothetical event is denoted by A and B is the observed evidence; PA prior
probability before observing B; P(B∣A) is the conditional probability that B is present in every state
of A; P(B) is the marginal probability; P(A∣B) is the posterior probability after observing B, and P(A,
B) is the joint probability.
The joint probability distribution function is estimated using the conditional independence
and the probability theory chain rule. The joint probability distribution of a set of variables U =
{X 1 , … , X n } is expressed as follows:
∏
n
P(U) = P(X1 , … , Xn ) = P(Xi ∣Pa (Xi )) (2.2)
i=1

where P(X i ∣Pa (X i )) denotes the conditional probability of X i , and Pa (X i ) is the parent node of X i .

TE

IE1 IE2

E1 E2 E3 E4

Figure 2.1 Typical BN with two parent nodes and one child node.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2.2 Bayesian Network 21

Through the marginalization of the joint probability distribution, system reliability can be
obtained. During the BN reliability analysis, each event’s prior probability is updated by new
information or posterior probability. The posterior probability will be updated by conditional
probability function or Bayesian inference as soon as a new observation or evidence E is given, as
follows:
P(U, E) P(E∣U) ⋅ P(U) P(E∣U) ⋅ P(U)
P(U∣E) = = = ∑ (2.3)
P(E) P(E) P(E∣U) ⋅ P(U)
i

Example 2.1 Let us attempt to comprehend the workings of BNs with a simple illustration.
For this illustration, let us pretend that we are modeling a student’s marks for an examination.
The marks depend on two additional factors: examination difficulty (ed) and intelligence level
(IQ) of the student.
Exam difficulty (ed) is a discrete variable and measures the difficulty of the exam with two pos-
sible values, 0 for easy and 1 for difficult. The easier the exams, the lower the value.
Intelligence Level (iq) represents the student’s IQ level. It is also discrete with two values 0 for
low and 1 for high. It is a measure of the student’s overall intelligence.
Additionally, the student’s intelligence level leads us to another component, which is the stu-
dents’ aptitude score (as), which can be further broken down. With the marks that the student has
earned up to this point, he is now able to get admission to a specific university.
Therefore, we have five variables: student marks (m), exam difficulty (ed), intelligence (iq),
aptitude score (as), and university admission (ad). The probability distribution values of the given
five variables in the form of CPT and the BN are presented in Figure 2.2.
The joint probability distribution is given by
P(m, ed, iq, as, ad) = P(ed).P(iq).P(m∣iq, ed).P(ad∣m).P(as∣iq)
where the probability of exam difficulty is denoted by P(ed), the probability of intelligence is
denoted by P(iq), P(m|iq, ed) is the conditional probability of marks obtained given his/her

iq0 iq1 Exam

Intelligence ed0 ed1
difficulty
0.8 0.2 level 0.7 0.3

Marks

as0 as1
Score
iq0 0.75 0.25
m0 m1
Admission iq1 0.4 0.6
iq0,ed0 0.6 0.4
iq0,ed1 0.9 0.1
iq1,ed0 0.5 0.5 ad0 ad1
iq1,ed1 0.8 0.2 m0 0.6 0.4
m1 0.9 0.1

Figure 2.2 BN example.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
22 2 Bayesian Approach for Reliability Evaluation and Remaining Useful Life Prediction

intelligence level and exam difficulty, P(ad|m) is the conditional probability of getting admission
given students’ marks and P(as|iq) is the conditional probability of students’ aptitude score given
his her intelligence.
Now, find out what is the probability that a student with a low aptitude score and a low intelli-
gence level will still be able to pass the exam and get into the university even if the difficulty of the
exam is hard.
For the given scenario, the joint probability distribution can be written as

P(m = 1; iq = 0; as = 0; ad = 1) =?

From the CPT and BN presented in Figure 2.2, the probability for the given conditions is
calculated as

P(m = 1; iq = 0; as = 0; ad = 1) = 0.1 × 0.1 × 0.75 × 0.3 × 0.8 = 0.0018

Practice 2.1 Determine the likelihood that a student will have a high IQ and a high aptitude score,
the test will be easy, and yet the student will still not pass the test and will not be admitted to the
institution.

2.3 Bayesian Reliability

The Bayesian statistics make it possible to conduct simultaneous evaluations of the reliability of
the complex system in addition to its subsystems and its individual components’ reliability. The
purpose of the Bayesian reliability evaluation is to get a posterior distribution that is as accurate
as feasible, and then to utilize this distribution to compute failure rate estimates with confidence
intervals, which are referred to as credibility intervals. A credibility interval is used to represent
an unknown parameter. Several statistical distributions and parameters that define the models are
utilized for reliability estimates. The exponential and the Weibull are frequently used distributions
to estimate the mean of failures data. On the other hand, the binomial is frequently used distri-
bution to predict the chance of failure for pass/fail systems. To be more precise, we cannot make
probability claims about the real parameter. The Bayesian approach deals with the model parame-
ters even though they are random values. There is no presumption that the model parameters are
random; rather, we utilize the tools of probability to convey our uncertainty about the values of
the parameters. Mean time between failures (MTBF) is a good example of such parameters. Cre-
ating a prior distribution for the model parameters is the first step, and then we use that revised
prior distribution to help us derive the updated assessment or posterior distribution for the model
parameters. The concept of Bayesian reliability can be expressed using the following formula:

fpos (R∣Data) ∝ Lik (Data∣R)fpor (R) (2.4)

The posterior distribution, denoted by the f pos with the reliability R, given data is equal to the
product of the prior knowledge, denoted by f por with likelihood function (Lik). Because the proba-
bility function of the data is the only thing that is considered in a standard statistical analysis, the
basic shift in mathematics that occurs in a Bayesian analysis is the addition of the prior information.
To perform a Bayesian reliability evaluation, the following steps are typically necessary:

1. Specify Prior Distribution: Define a prior distribution that represents the initial knowledge
or beliefs about the reliability parameters. This distribution can be based on expert opinions,
historical data, or other relevant information.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2.4 Application of BN in Reliability and Remaining Useful Life 23

2. Collect Data: Gather data on system performance, failures, or any other relevant information
related to reliability.
3. Likelihood Function: Specify a likelihood function that represents the statistical relationship
between the data and the reliability parameters. This function describes how likely the observed
data is under different reliability scenarios.
4. Bayesian Inference: Combine the prior distribution and the likelihood function to obtain the
posterior distribution using Bayes’ theorem. The posterior distribution represents the updated
belief about the reliability parameters given the prior knowledge and the observed data.
5. Posterior Analysis: Perform various analyses on the posterior distribution to obtain reliability
metrics of interest, such as failure rates, MTBF, or system availability. Uncertainty intervals can
also be calculated to quantify the uncertainty associated with these metrics.
6. Sensitivity Analysis: Assess the sensitivity of the results to different assumptions or inputs.
This helps identify influential factors and provides insights into the reliability behavior of the
system.

2.4 Application of BN in Reliability and Remaining Useful Life

BN provides an effective framework in the context of probabilistic events and is recognized as a
robust formalism to represent random variables with complex dependencies between them. The
main utility of a BN is to estimate the posterior probabilities of random variables set by combining
the prior knowledge with the real-time observation of some other variables. BN framework provides
a compact representation of the event-dependent failure behaviors and common-cause failures in
today’s complex systems. In addition, BN is also a solution to the state space explosion problem
associated with Markov chain-based modeling approaches (Liu and Singh 2010).
The BN-based dependability evaluation primarily consists of four processes, which include struc-
ture modeling, parameter modeling, inference, and verification and validation. For evaluating the
reliability of any field of system, like hardware, structures, software, and human reliability, the
aforesaid four steps are necessary to perform. Figure 2.3 provides a flowchart that outlines this
process in detail.
A typical hardware system comprised of two subsystems is presented in Figure 2.4. The paral-
lel subsystem comprises three components connected in parallel, whereas the series subsystem
consists of three components connected in series. In BN, the status of each component should be
represented by the root node, and the reliability of the entire system should be represented by the
final leaf node.

2.4.1 BN Structure Modeling

The first step of the BN model deals with qualitative aspects corresponding to the directed acyclic
graph. Determining nodes and identifying connection arcs are the two main steps involved in this
step of modeling. The techniques utilized to accomplish BN structure modeling include knowledge
representation for reliability estimation of structures, hardware, and humans; mapping technique
for reliability analysis of hardware, and humans; and structure learning.
The structure of the BN is developed in such a way that it considers the impact linkages between
the system’s components responsible for monitoring, data analysis, and the distribution of infor-
mation. The minimal cut-set matrix approach is used to extract the structure of the BN. The BN
structure is developed by making use of the logical relationships existing between the components.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
24 2 Bayesian Approach for Reliability Evaluation and Remaining Useful Life Prediction

Start

Step 1: Structure modelling

• Knowledge representation
• Mapping
• Structure representation
Revise Revise

Step 2: Parameter modelling

• Expert elicitation
• Mapping
• Parameter learning

Revise Revise

Step 3: BN Inference
• Exact inference
• Approximate inference

Network Network
Step 4: Validation verification
• Sensitivity analysis
• Well known scenario
• Real data
• Simulator data
• Constructive modeling

Figure 2.3 BN method flowchart.

Component 1

Parallel
Component 2 subsystem R
Subsystem R
Series
Component 3 subsystem R

Component 1 Component 2 Component 3

Figure 2.4 Typical BN of a series-parallel system for reliability study.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2.4 Application of BN in Reliability and Remaining Useful Life 25

These methods of representing information are extremely subjective and have the potential to pro-
duce dependability evaluation models. The developed network is often too large in techniques like
mapping and knowledge-based representation and takes considerable computational time. This
makes the structure learning technique preferable during BN structure modeling.
The algorithm for structure learning is typically the machine learning algorithm. The component
failure data along with the complete system failure data, are used to learn the structure of BN.
The reliability evaluation using this method is more accurate than those obtained using methods
that involve knowledge representation and mapping. This is one of the advantages of using this
structured learning method. The apparent downside of using structure learning is that it can be
difficult or even impossible to obtain adequate data for exact structure learning.
The BN reliability evaluation method has already been incorporated into power system reliability
modeling. For example, Eliassi et al. (2015) used BN for the evaluation of composite power systems
by giving importance to components. In their work, the training data has been generated by utiliz-
ing the Monte Carlo simulation with state sampling. They established the BN structure with the
data by using common structure learning algorithms.

2.4.2 BN Parameter Learning

The process of BN parameters learning involves assigning a prior probability as well as a condi-
tional probability. When evidence is gathered, the posterior probability of variables is referred to
as the conditional probability of those variables. Typically, a CPT is utilized in the case of discrete
variables, while with continuous variables, a conditional probability distribution is utilized. The
techniques that are used to accomplish BN parameter modeling include expert opinion and param-
eter learning for hardware, structures, and software reliability analysis and mapping, particularly
for hardware reliability analysis.
BNs’ capacity to learn from data is a particularly useful trait. Only the CPT parameters are learned
from historical data during parameter learning, even though the structure of the BN may be eas-
ily perceptible. The BN’s structure must be learned, nevertheless, when using structure learning.
This implies that historical data must be used to realize the interdependence between the domain
variables.
Every unknown parameter in a CPT must be learned from historical data. A beta distribution
defined in Eq. (2.5) is used for estimating the parameter of CPT:
¬(𝛼 + 𝛽) 𝛼−1
P(𝜃) = Beta(𝜃∣𝛼, 𝛽) = .𝜃 .(1 − 𝜃)𝛽−1 (2.5)
¬(𝛼).¬(𝛽)
According to the researchers, The Beta distribution is effective for parameter estimation of a
CPT. This is because it has a convenient property that, over the interval [0, 1] they are nonzero
and integrate to 1. The distribution parameters α and β are the number of events + 1. Also, the
𝛼
distribution mean value is 𝛼+𝛽 . This implies that a proportionally larger α value gives 𝜃 gets closer
to unity, signifying a larger data set gives more certainty about 𝜃.

2.4.3 BN Inference
Interface is the most important characteristic of a BN. In this stage of network evaluation, the
evaluated probability of networks is updated when new observations are given. For hardware and
software reliability evaluation, exact inference, and mapping inference are used methods for per-
forming BN inference.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
26 2 Bayesian Approach for Reliability Evaluation and Remaining Useful Life Prediction

A variable that was observed and instantiated in the BN serves as the evidence. By giving some
evidence or observations, the probability of one or more variables can be estimated in the BN infer-
ence. This indicates that BN generates the posterior probability for a target node of interest given
a collection of input values, or evidence. The probability distribution in the BN is calculated by
combining the data using various inference techniques.
There are both precise and approximate methods for carrying out Bayesian inferences. Variable
elimination algorithms and junction tree algorithms are two commonly used precise inference
techniques. In the variable elimination methodology, the variables are eliminated in a particular
order. In addition to improving computing efficiency, this provides a better approach to rewrite the
inference computations. On the other hand, the junction tree methods transform the multiple con-
nected BNs into junction trees. Then, inferences are carried out using the Lauritzen-Spiegelhalter
algorithm (Lauritzen and Spiegelhalter 1988) or variable elimination (Shafer-Shenoy algorithm).
When the model is large, with hundreds of variables, the precise inference approach is slow.
According to the literature, approximation techniques frequently perform well in such situations.
Variational approaches (mean-field approximation), parametric approximation, sampling tech-
niques, loopy belief propagation (Boyen-Koller algorithm, Boyen and Koller 1998), etc., are a few
examples of approximation techniques.

2.4.4 Validation and Verification

BN model validation is an extremely important facet of the reliability evaluation process. This is
because these aspects instill a level of acceptable confidence in the calculated results. How accu-
rately the BN model reflects the reality of the evaluated reliability is the validation. In BN anal-
ysis, for verification and validation, sensitivity analysis is the most used technique for hardware,
structures, software, and human reliability. This can be done by utilizing contrastive modeling,
simulated data, and real data.

2.5 Dynamic Bayesian Networks

The dynamic modeling of system behavior is the focus of the most recent work in the field of
real-time reliability evaluation. DBN is used to represent temporal aspects of a system’s behavior,
which makes it easier to evaluate reliability (Murphy 2002). By including pertinent temporal depen-
dencies inside the random variables, which will represent the variables’ dynamic behavior over
time, the Bayesian network model is expanded to a DBN model. To formalize complicated dynamic
system models using prior knowledge of the system’s functional and malfunctioning states, Weber
and Jouffe (2006) introduced an approach based on Dynamic Object-Oriented Bayesian Networks
(DOOBNs).
DBNs are a visual model for performing causal analysis and time-varying probabilistic inference
in the presence of uncertainty in the system. Contemporaneous dependencies and noncontempo-
raneous dependencies are the two categories of dependencies found in a DBN. Contemporaneous
dependencies are said to be when the arcs or connections between nodes represent variables within
the same period. Arcs between nodes representing variables at various times are referred to as non-
contemporaneous dependencies.
A typical DBN with two time slices t and t + 1 is presented in Figure 2.5. The two-slice temporal
Bayesian network (2TBN) that defines the transition model P(X t ∣ X t − 1 ) can be defined as given in
Eq. (2.6):
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2.6 Advantages and Limitations of BN and DBN 27

X 1t X2t X1t+1 X2t+1

Xnt Xnt+1

Time slice t Time slice t+1

Figure 2.5 DBN consisting of two-time slices.

∏
N
( ( ))
P(Xt ∣ Xt−1 ) = P Xti ∣ Pa Xti (2.6)
i=1
( )
where Pa Xti are the parents of Xti and Xti is the ith node at time t.
In a 2TBN, the node in the second time slice has its associated CPT or conditional probability
( ( ))
data (CPD) depending on the variable type defined by P Xti ∣ Pa Xti for all t > 1. For a sequence
of length T, the joint probability distribution of the 2TBN network is given in Eq. (2.7):
∏ ∏ (
T N
( ))
P(X1∶T ) = P Xti ∣ Pa Xti (2.7)
t=1 i=1

2.6 Advantages and Limitations of BN and DBN

The Bayesian approach for reliability evaluation is a powerful tool that integrates prior knowl-
edge, data, and statistical models to obtain reliable and up-to-date assessments. It provides a robust
framework for decision-making and can be applied in various industries, including engineering,
manufacturing, and infrastructure systems.
In the context of reliability evaluation, the Bayesian approach offers several advantages:
1. Incorporation of Prior Information: Bayesian analysis allows the integration of prior knowl-
edge or beliefs about system reliability into the evaluation process. This prior information can
be based on expert opinions, historical data, or previous studies. By combining prior knowledge
with new data, the Bayesian approach provides a more comprehensive and informed analysis.
2. Flexibility in Modeling: Bayesian methods allow for flexible modeling of the underlying reli-
ability distributions. Different distribution models can be specified based on the specific char-
acteristics of the system under evaluation. This flexibility enables the incorporation of complex
and realistic models that better represent the reliability behavior of the system.
3. Sequential Updating: One of the key features of Bayesian analysis is the ability to update esti-
mates as new data becomes available. This is particularly useful in reliability evaluation, where
data on failures or system performance may accumulate over time. The Bayesian approach
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
28 2 Bayesian Approach for Reliability Evaluation and Remaining Useful Life Prediction

allows for sequential updating of reliability estimates, resulting in more accurate and up-to-date
assessments.
4. Uncertainty Quantification: Bayesian inference provides a natural framework for quantify-
ing uncertainty in reliability evaluations. Through the posterior distribution, which represents
the updated belief about the reliability parameters, uncertainty intervals or credibility intervals
can be calculated. These intervals provide a measure of the uncertainty associated with the esti-
mated reliability metrics.
5. Integration of Heterogeneous Data: Bayesian analysis allows for the integration of different
types of data from multiple sources. This is particularly relevant in reliability evaluation, where
data can come from various sources such as field observations, laboratory testing, and histori-
cal records. Bayesian methods provide a unified framework for combining heterogeneous data,
resulting in a more robust and comprehensive reliability assessment.

Despite the many benefits of BNs, there are two significant application-related limitations. First
off, a lack of sufficient failure data frequently makes it impossible to estimate the CPT. The CPT
can be counted using expert judgment as a solution to this issue. The ability of BN to update the
conditional probabilities regularly based on new data or evidence helps handle the problem to a
considerable extent, even though the effectiveness of this solution is highly dependent on expert
capabilities. The second issue of BN that has been extensively studied is its inability to accurately
model continuous variables because of the limits of the inference method. All continuous nodes
must have some predefined range and intervals defined to achieve uniform discretization. The
number of set intervals heavily influences accuracy, but it comes at a high cost due to computational
complexity.
Another limitation of BN and DBN is that each node can only represent the state associated with a
single observed variable. Therefore, to represent a component state with two or more observed vari-
ables, more child nodes need to be added to the network structure, which increases the complexity
of the BN. The challenges with the use of BNs are:

● Discretization of continuous Variables.

● Collecting and structuring expert knowledge.
● No support for Feedback loops.
● The quality of the prior beliefs or model is directly correlated to the quality of the outputs pro-
duced by the network. If you feel that the system is dependent on a certain variable, then you
can only include that variable as a component of a BN.

2.7 Conclusion

In this chapter, the use of BNs in reliability analysis has been demonstrated. Since BNs were first
introduced in the early 1980s by Pearl, their use in reliability studies has been the subject of exten-
sive research and has seen successful results. This chapter presents an overview of the reliability
evaluation approach using BNs from the standpoint of the objects of evaluation. Because of its
user-friendly interaction with domain experts, the modeling framework that BNs form is a prac-
tical, useful tool and utilized effectively. In addition, because BNs are found in probability theory,
many of the fundamental debates that stymie the development of other modeling frameworks may
be avoided. The robust mathematical formulation was applied to build learning approaches in BNs
which are effective and efficient.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 29

References

Boyen, X. and Koller, D. (1998). Tractable inference for complex stochastic processes. UAI.
Eliassi, M., Khoshkholgh Dashtaki, A., Seifi, H. et al. (2015). Application of Bayesian networks in
composite power system reliability assessment and reliability-based analysis. IET Generation,
Transmission and Distribution 9 (13): 1755–1764. https://doi.org/10.1049/iet-gtd.2014.0660.
Lauritzen, S.L. and Spiegelhalter, D.J. (1988). Local computations with probabilities on graphical
structures and their application to expert systems. Journal of the Royal Statistical Society: Series B:
Methodological 50 (2): 157–194. https://doi.org/10.1111/j.2517-6161.1988.tb01721.x.
Liu, Y. and Singh, C. (2010). Reliability evaluation of composite power systems using Markov cut-set
method. IEEE Transactions on Power Systems 25 (2): 777–785. https://doi.org/10.1109/tpwrs.2009
.2033802.
Murphy, K. (2002). Dynamic Bayesian networks: representation, inference, and learning. Thesis of the
University of California, Berkley.
Weber, P. and Jouffe, L. (2006). Complex system reliability modeling with dynamic object oriented
Bayesian networks (DOOBN). Reliability Engineering and System Safety 91 (2): 149–162. https://doi
.org/10.1016/j.ress.2005.03.006.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
31

Evaluation of Basic Reliability Indices Using State Enumeration

Method
Rajesh Arya 1 , Chandrima Roy 2 , Atul Koshti 3 , Ramesh C. Bansal 4 , and Liladhar Arya 5
1
Department of Information Technology, Acropolis Institute of Technology and Research, Indore, Madhya Pradesh, India
2
Department of Electronics & Communication Engineering, Heritage Institute of Technology, Kolkata, West Bengal, India
3 Department of Electrical Engineering, GES’S R.H. Sapat College of Engineering, Management Studies & Research, Nashik,

Maharashtra, India
4
Department of Electrical Engineering, University of Sharjah, Sharjah, United Arab Emirates
5
Department of Electrical Engineering, Shri G.S. Institute of Technology & Science, Indore, Madhya Pradesh, India

3.1 Introduction
The concept of reliability function is of great significance for nonrepairable or mission-oriented
systems. A repairable system once fails to be repaired and then again is ready for service. In such
a situation one is interested in knowing the percentage of system in the time is upstate and the
failure frequency of the systems. Usually, a constant failure rate is assumed in such analysis. Since
the repair time of the system is also a random variable. Repair rate of the system is also treated as
a constant i.e. repair density function is also assumed as exponential. Usually, it is expected that a
large number of failures in a system occurs, which requires small repair time, whereas few failure
will occur, which requires a long repair time. Hence, the plot of the estimate of the repair density
function will look as shown in Figure 3.1.
To determine availability function Markov process is utilized. Precisely function of a repair den-
sity is defined as
P[tr ≤ Tr ≤ tr + Δtr ]
fr (tr ) = lim ≈
Δtr →0 Δtr
nr (tr , Δtr ): Number of repairs completed between tr to tr + Δtr
N r : Total number of repairs carried over in a specified period of study.

3.2 Markov Process

It is a stochastic process that explains a process changing from one state to another. Assume that a
process is described by “n” states, which are mutually exclusive with associated probabilities at “t”
as P1 (t), P2 (t), . …, Pn (t). The Markov process (Billinton & Allan 1983; Ebeling 2000) is defined by
following conditional probability equation

Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
32 3 Evaluation of Basic Reliability Indices Using State Enumeration Method

fr(tr)

tr

Figure 3.1 Repair density function plot.

[ ]
j
P X(t + Δt) = = i, X(t + Δt) = k, X(t − 2Δt) = 1 ,
X(t)
[ ]
j
= P X(t + Δt) = = i = pij (t, Δt) (3.1)
X(t)
It is obvious that in a Markov process, a state “j” at (t + Δt) depend on the occupied state by
the process at t i.e. no on earlier states. In this sense, a Markov process is memoryless process.
The conditional probability in Eq. (3.1) i.e. pij (t, Δt) are known as transitional probabilities.
Further, if transitional probabilities are independent on “t,” then the Markov process is called
homogeneous/stationary process. In view of the above Eq. (4.1) is written as
[ ]
j
P X(t + Δt) = = i = qij (t, Δt) (3.2)
X(t)
qij may be defined state as transitions rate i to j. In terms of reliability as shall be made clear, this
may be either a failure or repair rate, provided both have exponential probability density functions.
Consider the general state transition diagram shown below.
Let us assume P1 (t), P2 (t), . . …, Pn (t) are state probabilities at time “t.” General expression for
pi (t + Δt) system of probability in ith state at [(t + Δt)] is written below as (Figure 3.2)

⎛ ⎞
∑
n ⎜ ∑n ⎟
pi (t + Δt) = qki Δt pk (t) + ⎜1 − qki Δt ⎟ pi (t)
⎜ ⎟
k=1 ⎜ k=1 ⎟
k≠i ⎝ k≠i ⎠
The above equation is rewritten below as
∑
n
pi (t + Δt) = qki Δt pk (t) + pi (t) (3.3)
k=1

Further, the above can be written as

pi (t + Δt) − pi (t) ∑
n
= qki pk (t)
Δt k=1
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3.2 Markov Process 33

1–ΣqijΔt
qi1Δt
1

qi3Δt

qi2Δt

qi3Δt 2

qijΔt
3

1–Σq3jΔt 1–ΣqjiΔt

Figure 3.2 General state transition diagram.

Taking the limit as Δt → 0

∑
n
ṗ i = qki pk (t) (3.4)
k=1

The following matrix can be used to express Eq. (3.4):

ṗ T (t) = pT A (3.5)

Matrix “A” follows from (3.4) as follows:

⎡ a11 a12 a13 .. a1n ⎤
⎢ ⎥
⎢ a21 a22 a23 .. a2n ⎥
⎢ ∶ ∶ ∶ ∶ ⎥
A=⎢
⎢ ak1 ak2 ak3 .. akn ⎥⎥
⎢ ∶ ∶ ∶ ∶ ⎥
⎢a an2 an3 … ann ⎥⎦
⎣ n1
Elements of “A” matrix are identified as follows:
aij = rate of departure from ith state to jth state = qij
∑
n
aii = −(sum of total transition rate from ith state) = − qij
j=1
j≠i
Matrix “A” is known as the transition intensity matrix.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
34 3 Evaluation of Basic Reliability Indices Using State Enumeration Method

If one puts t = rΔt in Eq. (3.3) to define pki = qki Δt state k to state i transition probability in an
interval, then Eq. (3.3) gets modified as
∑
pi (r + 1) = pki pk (r) + pi (r)
This represents a discrete Markov process or chain (Billinton & Allan 1996; Lewis 1987) and
represented in matrix form as
pT (r + 1) = pT (r) P (3.6)

⎡p11 p12 p13 .. p1n ⎤

⎢ ⎥
p p22 p23 .. p2n ⎥
P = ⎢ 21
⎢∶ ∶ ∶ ∶ ∶ ⎥
⎢p pn2 pn3 … pnn ⎥⎦
⎣ n1
P is known as the transition probability matrix.
pii = 1−(sum of the state i transition probabilities)
pij = probabilities of transition from state i to j

3.3 Solution of State Equations

When one is interested in long-term solutions to reliability problems, then a steady-state solution
of Eqs. (3.5) and (3.6) is required.

3.3.1 Steady State Solution: Continuous Time Case

Long run behavior of Eq. (3.5) is obtained by setting ṗ T = 0 i.e. by solving the algebraic matrix eqn.
pT A = 0 (3.7)
Since matrix A is singular (3.7), it will have an infinite solution. Hence, (n − 1) equations are
taken from (3.7), and one more equation as follows
∑
pk = 1 (3.8)

Example 3.1
Consider following two two-state continuous Markov process models. Obtain steady-state
probabilities.
q12 = 0.2
q21 = 2
The matrix equation is written as follows:
[ ]
−0.2 0.2
[ṗ 1 ṗ 2 ] = [p1 p2 ] =0
2 −2
To get a steady-state solution
[ ]
−0.2 0.2
[p1 p2 ] = [0 0]
2 −2
− 0.2 p1 + 2 p2 = 0, 0.2 p1 − 2 p2 = 0
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3.3 Solution of State Equations 35

These two equations are not independent. Hence, one of them is considered.
0.2 p1 − 2 p2 = 0, 2 p1 + p2 = 1
Solve the equation
p1 = 0.9090
p2 = 0.0909

3.3.2 Complete Solution: Continuous Time Case

The complete solution of the matrix equation is required
ṗ T (t) = pT A, pT (0) = [1 0 ...0]

3.3.2.1 Laplace Transform Method

s pT (s) − pT (0) = pT (s) A, pT (s)[s I − A] = pT (0)

pT (s) = pT (0)[s I − A]−1 , pT (t) = L−1 {pT (0)[s I − A]−1 }

3.3.2.2 Solution by Computing eAT

The solution of state space equation is written as
pT (t) = pT (0)eAT
Matrix exponential can be expressed as
[ ]
(At)2 (At)m
eAT = 1 + At + +···+ +···
2! m!
For a specific time “t,” “m” terms may be considered such that the probabilities do not change
significantly by considering (m + 1) terms.

3.3.2.3 Discretization
One of the most powerful methods for solution of state equation is by discretization. The matrix
differential equation is discretized as
pT (t + Δt) − pT (t)
= pT (t)A
Δt
pT (t + Δt) − pT (t) AΔt + pT (t)

pT (t + Δt) − pT (t) [I + AΔt]

pT (t + Δt) − pT (t)A′ , A′ = I + A Δt
Putting t = rΔt and dropping Δt from argument of pT (.), above is explained as
pT (r + 1) = pT (r) A′ , Put r = 0
pT (1) = pT (0) A′ , Put r = 1
2
pT (2) = pT (1) A′ = pT (0) A′
The general solution is written as
r
pT (r) = pT (0) A′
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
36 3 Evaluation of Basic Reliability Indices Using State Enumeration Method

3.3.2.4 Steady State Solution of Discrete Markov Process

pT (r + 1) = pT (r) P

Long run solution of the above is obtained by letting

pT (r + 1) = pT (r) P = pT
∑
pT = pT P, pT [I − P] = 0 and pi = 1

3.3.2.5 Complete Solution of Discrete Markov Chain

The following general solution follows from Eq. (3.6)

pT (r) = pT (0) Pr

Example 3.2
Consider a two-state continuous Markov process with 𝜆 = q12 = 0.1 and 𝜇 = q21 = 10. Obtain prob-
abilities function for Figure 3.3.
Solution:
The matrix equation is written as follows
[ ]
−0.1 0.1
[ṗ 1 ṗ 2 ] = [p1 p2 ]
10 − 10
[ ] [ ]
s+𝜆 −𝜆 s + 0.1 − 0.1
[s I − A] = =
−𝜇 s+𝜇 −10 s + 10
[ ]−1
s+𝜆 −𝜆
[s I − A]−1 =
−𝜇 s+𝜇
[ ]
1 s+𝜇 𝜆
[p1 (s) p2 (s)] = [1 0]
s(s + 𝜇 + 𝜆) 𝜇 s+𝜆
s+𝜇 𝜆
p1 (s) = , p (s) =
s(s + 𝜇 + 𝜆) 2 s(s + 𝜇 + 𝜆)
Taking Inverse laplace transform given below for correctness
𝜇 𝜆 −(𝜆+𝜇)t 𝜆 𝜆 −(𝜆+𝜇)t
p1 (t) = + e , p2 (t) = − e
𝜆+𝜇 𝜆+𝜇 𝜆+𝜇 𝜆+𝜇
Taking inverse Laplace transform gives
10 0.1 −10.1t 0.1 0.1 −10.1t
p1 (t) = + e , p2 (t) = − e
10.1 10.1 10.1 10.1

0.2 Figure 3.3 Two-state continuous Markov process model.

1 2

2
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3.4 Functions of a Single Component’s Availability and Unavailability 37

3.4 Functions of a Single Component’s Availability

and Unavailability
Consider single components with rates of failure and repair of 𝜆 and 𝜇, respectively. Figure 3.4
displays the system’s state transition diagram (Arya & Koshti 2008; Arya et al. 2013; Halve et al.
2023).
The states 1 and 2 stand for up and down, respectively. The following matrix form represents the
state space equations.
[ ]
−𝜆 𝜆
[ṗ 1 ṗ 2 ] = [p1 p2 ]
𝜇 −𝜇
p1 (0) = 1, p2 (0) = 0
[ ] [ ]
s 0 −𝜆 𝜆
[s I − A] = −
0 s 𝜇 −𝜇
[ ]
s+𝜆 −𝜆
=
−𝜇 s+𝜇
[ ]−1 [ ]
s+𝜆 −𝜆 1 s+𝜇 𝜆
[s I − A]−1 = =
−𝜇 s+𝜇 s(s + 𝜇 + 𝜆) 𝜇 s+𝜆
[ ]
1 s+𝜇 𝜆
[p1 (s) p2 (s)] = [1 0]
s(s + 𝜇 + 𝜆) 𝜇 s+𝜆
s+𝜇 𝜆
p1 (s) = , p (s) =
s(s + 𝜇 + 𝜆) 2 s(s + 𝜇 + 𝜆)
Taking inverse Laplace transform gives
𝜇 𝜆 −(𝜆+𝜇)t
P1 (t) = + e
𝜆+𝜇 𝜆+𝜇
𝜆 𝜆 −(𝜆+𝜇)t
P2 (t) = − e
𝜆+𝜇 𝜆+𝜇
The state probability p1 (t) is known as availability and p2 (t) is known as unavailability. Figure 3.5
shows the plot of these functions. It is observed that both availability and unavailability approaches
to steady state values as given
𝜇 𝜆
P1ss = , P2ss = ,
𝜆+𝜇 𝜆+𝜇
The same figure shows the plot of reliability, which is exponentially decreasing and does not
consider repair. No path of return from absorbing states.

Figure 3.4 A system with a single component’s state λ

transition diagram.

1 2

μ
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
38 3 Evaluation of Basic Reliability Indices Using State Enumeration Method

p1(t)
p1(t)

p2(t)
p2(t)

R(t)

Figure 3.5 Plot of availability and unavailability.

3.5 Two-Component State Model and State Probabilities

Assume that two components “a” and “b” have failure and repair rates as 𝜆a , 𝜆b and 𝜇 a , 𝜇 b . The
system will have four states, as shown in the diagram (Figure 3.6).

1 2

μα λa
–
a b
a b

μb μb

λb λb

– – –

a b a b
μα λa

3 4

Figure 3.6 Two-component state model.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3.5 Two-Component State Model and State Probabilities 39

State space equations are written as follows

⎡ −(𝜆a + 𝜆b ) 𝜆a 𝜆b 0 ⎤
⎢ ⎥
𝜇a −(𝜇a + 𝜆b ) 0 𝜆b
[ṗ 1 ṗ 2 ṗ 3 ṗ 4 ] = [p1 p2 p3 p4 ] ⎢ ⎥
⎢ μb 0 −(𝜆a + 𝜇b ) 𝜆a ⎥
⎢ 0 𝜇b 𝜇a −(𝜇a + 𝜇b ) ⎥⎦
⎣
State probabilities may be easily evaluated by using the method of discretization. It requires a
suitable choice of “Δt.” If one does not have experience with the system initially, any value of Δt
may be selected, and state probabilities trajectories are obtained. Then, a smaller value of Δt is
selected, and then again, state probabilities are obtained. The difference in two calculated state
probabilities should be less than the tolerance value specified. Otherwise, further reduce Δt and
repeat the calculation of state probabilities till convergence.
State probabilities in the long run are evaluated by solving the following algebraic equations:

⎡ −(𝜆a + 𝜆b ) 𝜆a 𝜆b 0 ⎤
⎢ ⎥
𝜇a −(𝜇a + 𝜆b ) 0 𝜆b
[p1 p2 p3 p4 ] ⎢ ⎥ = [0 0 0 0]
⎢ 𝜇b 0 −(𝜆a + 𝜇b ) 𝜆a ⎥
⎢ 0 𝜇b 𝜇a −(𝜇a + 𝜇b ) ⎥⎦
⎣
p1 + p2 + p3 + p4 = 1

The state probabilities are obtained as

𝜇a 𝜇 b 𝜆 𝜇 𝜇 a 𝜆b 𝜆a 𝜆 b
p1 = , p2 = a b , p3 = , p4 =
D D D D
D = (𝜆a + 𝜇a )(𝜆b + 𝜇b )

If the two components are identical i.e.

𝜆a = 𝜆b = 𝜆, 𝜇 a = 𝜇b = 𝜇

The diagram of state transition will reduce Figure 3.7, which is shown below.

State 1 Both units are in up state

State 2 One unit is up and other is down
State 3 Both units are down

2μ λ

1 2 3

2μ

Figure 3.7 Diagram of the reduced form state transition for a two-component system.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
40 3 Evaluation of Basic Reliability Indices Using State Enumeration Method

State equations are written by inspection

⎡ −2𝜆 2𝜆 0 ⎤
[ṗ 1 ṗ 2 ṗ 3 ] = [p1 p2 p3 ] ⎢ 𝜇 −(𝜆 + 𝜇) 𝜆 ⎥
⎢ ⎥
⎣ 0 2𝜇 −2𝜇 ⎦
Steady-state probabilities are evaluated as follows

⎡ −2𝜆 2𝜆 0 ⎤
[p1 p2 p3 ] ⎢ 𝜇 −(𝜆 + 𝜇) 𝜆 ⎥ = [0 0 0], p1 + p2 + p3 = 1
⎢ ⎥
⎣ 0 2𝜇 −2𝜇 ⎦
Solution gives
𝜇2 2𝜆𝜇 2 𝜆2
p1 = , p2 = , p3 =
(𝜆 + 𝜇)2 (𝜆 + 𝜇) 2 (𝜆 + 𝜇)2
System availability is evaluated by applying system adequacy criterion. For example, at least one
of the units is required for the success system, then the availability of a system is given by
𝜇 2 + 2𝜆𝜇
Asys = p1 + p2 =
(𝜆 + 𝜇)2
Unavailability is given as
𝜆2
Asys =
(𝜆 + 𝜇)2

3.6 Three-Component State Transition Diagram

Three components of state transition Figure 3.8 are shown below.

If the units are identical, then the above state transition diagram reduces, as shown in Figure 3.9.

μa ab c μc
1
μb

λa λb λc
– – –
ab c ab c ab c
2 3 4
μb μc μa μc μb
μa
λb λc λa λc λb
–– – – – –
ab c λa ab c ab c
5 6 7
μc μb μa

λb
–––
λc ab c
λa
8

Figure 3.8 Three-component state transition diagram.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3.7 Concept of Frequency and Mean Duration 41

3μ 2μ λ

1 2 3 4

μ
3μ
2μ

Figure 3.9 Diagram of the reduced state transition for a three-component system.

For steady-state probabilities, state equations are written, and solutions for eight states (different
components case) are written (long run):
𝜇a 𝜇b 𝜇c 𝜆 𝜇𝜇 𝜇 𝜆𝜇 𝜇a 𝜇 b 𝜆c
p1 = , p2 = a b c , p3 = a b c , p4 =
D D D D
𝜆a 𝜆 b 𝜇 c 𝜆a 𝜇 b 𝜆c 𝜇 a 𝜆b 𝜆c 𝜆a 𝜆 b 𝜆 c
p5 = , p6 = , p7 = , p8 =
D D D D
D = (𝜆a + 𝜇a ) (𝜆b + 𝜇b ) (𝜆c + 𝜇c )
If the components are identical, then states are reduced from eight to four, as shown in Figure 3.9:
𝜇3 3𝜇 2 𝜆 3𝜆2 𝜇 𝜆3
p1 = , p2 = , p3 = , p4 =
(𝜆 + 𝜇)3 (𝜆 + 𝜇) 3 (𝜆 + 𝜇) 3 (𝜆 + 𝜇)3
If at least two units are required from an adequacy viewpoint, the system availability is given as
𝜇 3 + 3𝜇 2 𝜆 3𝜆2 𝜇 + 𝜆3
Asys = p1 + p2 = , Asys = p3 + p4 =
(𝜆 + 𝜇) 3 (𝜆 + 𝜇)3

3.7 Concept of Frequency and Mean Duration

Assume a component has failure and repair rates are, respectively, 𝜆 and 𝜇. MTTF and MTTR are
given as 1/𝜆 and 1/𝜇 respectively. Life length (T f ) and time to repair (T r ) are random variables, and
each has an exponential distribution function (Billinton & Allan 1996; Lewis 1987; Koshti et al.
2023). Then cycle time i.e. sums of “T f ” and “T r ” is also a random variable when the process of
failure and repair comes in stages. Moreover, there are two random variables that are independent.
Hence, as in the case of the standby system, the cycle time (T c = T f + T r ) has hypo-exponential
distribution function (Lewis 1987) and is given as
𝜆𝜇 −𝜆t
g(t) = [e − e−𝜇t ]
𝜇−𝜆
Average or mean value of cycle time (T c ) is given as
1 1
Tc = +
𝜆 𝜇
Thus, it can be seen as a statistical cycle, as shown in Figure 3.10.
Cycle time is also known as MTBF i.e. mean time between failures. Average failure frequency or
frequency is given as ( ) ( )
1 1 𝜇𝜆 𝜇 𝜆
f = = = = 𝜆= 𝜇
Tc 1 1
+ 𝜆+𝜇 𝜆+𝜇 𝜆+𝜇
𝜆 𝜇
= (Probability of being in that state) (rate of departure from that state)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
42 3 Evaluation of Basic Reliability Indices Using State Enumeration Method

1 1

λ μ

Tc
t

Figure 3.10 Component statistical cycle.

Now consider a general state transition diagram with state probabilities P1 (t), P2 (t). . , Pn (t). Fur-
ther assume t1 , t2 , … , tn are time spent in these states respectively.
Where,
T0 = t1 + t2 + t3 , … + tn
Further assume that q1 , q2 , q3 , … , qn are total rate of departure from these states, respectively,
i.e.
∑
qi = qij
j

Then, the frequency of ith state i.e. state “i” on one side and remaining states on the other side
is given as the number of transitions per unit time from ith state:
( )
qt ti
fi = i i = q i = pi q i
To To
= (probability of being in ith state) (total rate of departure from ith state).

3.8 Frequency of Combined Events

Individual frequency of a state is not of importance when large numbers of components are
involved in the systems. Transitions per unit of time (frequency) from a collection of events, such
as a group of states that correlate to system success, are important under such circumstances. This
frequency matches the frequency of failure. Figure 3.11 depicts a generic state transition diagram.
The total states are divided into two categories, “W” and “F.” Working (up) and failing (down)
states, respectively, are represented by the states in “W” and “F.” If the system is in a state of “W”
set the system is “up” otherwise if the system encounters a state in “F” the system is in “down”
state.
In such situations, transitions outside the group are considered. Transitions among states in a
group are not considered. Hence, the frequency of transitions from “W” to “F” is written as:
∑ ∑
f = pi qij (3.9)
i∈W j∈F
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3.8 Frequency of Combined Events 43

i qji

qij

Figure 3.11 General state transition diagram.

Frequency will be the same as calculated as transitions from “F” to “W” and given
∑ ∑
f = pj qji (3.10)
j∈F i∈W

Example 3.3
A system having two component consists of a 𝜇 1 = 9.6, 𝜇 2 = 9.4, 𝜆1 = 0.4 and 𝜆2 = 0.6. Obtain (1)
f 1 , f 2 , f 3 , f 4 (2) f 12 , f 34 (Figure 3.12).

Figure 3.12 General state transition diagram. μ1 λ1

–
1 2
1 2

μ2 μ2

λ2 λ2

– – –

1 2 1 2
μ1 λ1
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
44 3 Evaluation of Basic Reliability Indices Using State Enumeration Method

The probabilities in long run are given as

𝜇1 𝜇2 𝜆𝜇
p1 = = 0.94 × 0.96 = 0.9024, p2 = 1 2 = 0.04 × 0.94 = 0.0376
D D
D = (𝜆1 + 𝜇1 )(𝜆2 + 𝜇2 )
𝜇 1 𝜆2 𝜆𝜆
p3 = = 0.96 × 0.06 = 0.0576, p4 = 1 2 = 0.04 × 0.06 = 0.0024
D D
f1 = p1 (𝜆1 + 𝜆2 ) = (0.94 × 0.96)(0.4 + 0.6) = 0.94 × 0.96 = 0.9024
f2 = p2 (𝜇1 + 𝜆2 ) = 0.04 × 0.94 × (9.6 + 0.6) = 0.38352
f3 = p3 (𝜆1 + 𝜇2 ) = (0.96 × 0.06)(0.4 + 9.4) = 0.96 × 0.06 × 9.8 = 0.56448
f4 = p4 (𝜇1 + 𝜇2 ) = 0.04 × 0.06 × (9.4 + 9.6) = 0.04 × 0.06 × 19 = 0.0456
f12 = 𝜆2 (p1 + p2 ) = 0.6 × [0.94 × 0.96 + 0.04 × 0.94] = 0.564
f34 = 𝜇2 (p3 + p4 ) = 9.4 × [ 0.96 × 0.06 + 0.04 × 0.06] = 0.564

Note:

f12 = f34

This is required since frequency balance must hold.

3.9 State Enumeration Technique for Obtaining

Frequency-Duration (FD)

For determining failure frequency, mean up/time of a technique has been explained Figure 3.13
(Arya et al. 2013; Halve et al. 2023; Koshti et al. 2023; Endrenyi 1978).

Step 1: Failure and repair rate is assumed that each component present in the system is provided.
Draw a state transition diagram using these states.
Step 2: Solving for steady-state probabilities.
Solve steady-state probabilities
∑
pT A = 0, pi = 1

Step 3: Divide the state using the “system adequacy criterion” space into the “W” and “F” groups.
Get steady state availability and unavailability using the following method:
∑ ∑
pup = pi , pdn = pj
i∈W j∈F

Step 4: The following relations to determine failure frequency

∑ ∑
f = pi qij
i∈W j∈F

Step 5: The diagram’s representation of the general state transition consists of two states and one
component where 𝜆e and 𝜇 e are the rate of failure and repair being equal. The frequency is
expressed as follows:
f f
f = pup 𝜆e = pdn 𝜇e , 𝜆e = ,𝜇 =
pup e pdn
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3.9 State Enumeration Technique for Obtaining Frequency-Duration (FD) 45

μe λe

Pup Pdn

Figure 3.13 Two state one component equivalent state transition diagram.

As for a single component

1 pup
MUT = = = pup Tc
𝜆e f
1 p
MDT = = dn = pdn Tc
𝜇e f
Hence, pup , pdn , f , MUT, and MDT are calculated as reliability index.

Example 3.4
Obtain a system consisting of three identical components having failure and repair rate as 𝜆 and 𝜇,
respectively. System success requires at least two components in upstate.

(a) Frequency
(b) MUT
(c) MDT

Take 𝜇 = 9.8 per year and 𝜆 = 0.2 per year

Solution:
The system will have four states, as shown in Figure 3.9. State probabilities are given as follows
𝜇3
p1 = = 0.983 = 0.9411
(𝜆 + 𝜇)3
3𝜇 2 𝜆
p2 = = 3 × 0.982 × 0.02 = 0.057624
(𝜆 + 𝜇)3
3𝜆2 𝜇
p3 = = 3 × 0.98 × 0.022 = 0.001176
(𝜆 + 𝜇)3
𝜆3
p4 = = 0.023 = 0.000008
(𝜆 + 𝜇)3
Adequacy criterion is applied and pup and pdn are obtained as
𝜇3 3𝜇 2 𝜆
pup = p1 + p2 = + = 0.998724
(𝜆 + 𝜇) 3 (𝜆 + 𝜇)3
𝜇 2 (𝜇 + 3𝜆) 3𝜆2 𝜇 𝜆3
pup = , p dn = p 3 + p 4 = +
(𝜆 + 𝜇)3 (𝜆 + 𝜇)3 (𝜆 + 𝜇)3
(3𝜇 + 𝜆)𝜆2
pdn = = 0.001184
(𝜆 + 𝜇)3
Frequency f is calculated as follows
f = p1 × 0 + p2 × 2𝜆
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
46 3 Evaluation of Basic Reliability Indices Using State Enumeration Method

3𝜇 2 𝜆 6𝜇 2 𝜆2 6 × 9.82 × 0.22
f = × 2𝜆 = = = 0.0230496∕year
(𝜆 + 𝜇) 3 (𝜆 + 𝜇) 3 1000
Mean cycle time
(𝜆 + 𝜇)3 1000
Tc = 1∕f = = = 43.38 year
6𝜇 2 𝜆2 6 × 9.82 × 0.22
pup 𝜇 2 (𝜇 + 3𝜆)
MUT = =
f 6𝜇 2 𝜆2
9.82 (9.8 + 0.6)
= = 43.33
6 × 9.82 × 0.22
p 𝜆2 (3𝜇 + 𝜆) (𝜆 + 𝜇)3 𝜆2 (3𝜇 + 𝜆)
MDT = dn = × =
f (𝜆 + 𝜇)3 6𝜇 2 𝜆2 6𝜇 2 𝜆2
0.22 (29.4 + 0.2)
= = 0.05
6 × 9.82 × 0.22

Example 3.5
A generator has failure, repair, and re-installation rate as 𝜆, 𝜇, and 𝛼 , respectively. Obtain expres-
sions for
(a) Availability
(b) Frequency
(c) Mean up time (MUT)
(d) Mean down time (MDT)

Solution:
Figure 3.14 shows the state transition
State 1: Operable state
State 2: Failed state
State 3: Generator repaired but not re-installed for operation

1
α

λ μ

2 3

Figure 3.14 State transition diagram for Example 3.5.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3.9 State Enumeration Technique for Obtaining Frequency-Duration (FD) 47

Hence states 1 is up state and states 2 and 3 are down state.

⎡ −𝜆 𝜆 0 ⎤
[ṗ 1 ṗ 2 ṗ 3 ] = [p1 p2 p3 ] ⎢ 0 −𝜇 𝜇 ⎥
⎢ ⎥
⎣ 𝛼 0 −𝛼 ⎦
Long run probabilities are calculated by setting derivative terms zero and taking along two
equations from the above set of equation
−𝜆p1 + 𝛼p3 = 0, 𝜆p1 − 𝜇p2 = 0 and p1 + p2 + p3 = 1
Solving the above three equations
𝜆 𝜆 𝜆 𝜆
ṗ 3 = p , ṗ = p , p + p + p = 1
𝛼 1 2 𝜇 1 1 𝜇 1 𝛼 1
𝜇𝛼 𝛼𝜆 𝜇𝜆
p1 = ,p = ,p =
𝜇𝛼 + 𝛼𝜆 + 𝜇𝜆 2 𝜇𝛼 + 𝛼𝜆 + 𝜇𝜆 3 𝜇𝛼 + 𝛼𝜆 + 𝜇𝜆

(a) Availability is given as

𝜇𝛼
A = p1 =
𝜇𝛼 + 𝛼𝜆 + 𝜇𝜆
(b) Frequency is given by
𝜇𝛼𝜆
f = p1 𝜆 =
𝜇𝛼 + 𝛼𝜆 + 𝜇𝜆
(c) Mean up time
p1 1
MUT = =
f 𝜆
(d) Mean down time
p + p3 𝜇𝜆 + 𝛼𝜆
MDT = 2 =
f 𝜇𝛼𝜆
Take numerical values
𝜇 = 9.6∕year,
𝜆 = 0.4∕year
𝛼 = 20∕year
(a) Availability is given as
9.6 × 20 192
A = p1 = = = 0.9421
9.6 × 20 + 9.6 × 0.4 + 0.4 × 20 203.8
Unavailability
A = 1 − A = 0.0579
(b) Frequency is given by
9.6 × 20 × 0.4
f = = 0.3768∕year
203.8
(c) Mean up time
p 0.9421
MUT = 1 = = 2.50 year
f 0.3768
(d) Mean down time
p + p3 0.0579
MDT = 2 = = 0.15366 year
f 0.3768
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
48 3 Evaluation of Basic Reliability Indices Using State Enumeration Method

Example 3.6
A generator with a 100 MW rating has failure rate and repair rate of 0.2 and 9.8 per year. Obtain-

(a) Availability and unavailability

(b) Frequency
(c) MUT and MDT
(d) Mean capacity
(e) Expected power and energy are not supplied if the load on the generator is constant at 80 MW.
Solution:

(a) Availability is given as

9.8
A= = 0.98
9.8 + 0.2
0.2
A= = 0.02
9.8 + 0.2
(b) Frequency is given by

f = 0.98 × 0.2 = 0.196∕year

Note: f < 𝜆 [this will always be the case]

(c) Mean up time

A 0.98 1
MUT = = =5=
f 0.196 𝜆
Mean down time

A 0.02 1
MDT = = = 0.102 =
f 0.196 𝜇
(d) Mean capacity
∑
c= ci pi = c1 p1 + c2 p2 = 100 × 9.8 + 0 × 0.02 = 98 MW

Variance = 𝜎c2
∑
= (ci − c)2 pi = (c1 − c)2 p1 + (c2 − c)2 p2 = (100 − 98)2 × 0.98 + (0 − 98)2 × 0.02
= 196 MW2
√
𝜎c = 196 = 14 MW
∑
(e) Expected power not supplied (EPNS) = PNSi × pi

= PNS1 × p1 + PNS2 × p2

PNSi power not supplied in ith state

= 0 × 0.98 + 80 × 0.02 = 160 MW

Expected energy not supplied (EENS)=1.6 × 8760 = 14, 016 MWh/year

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 49

3.10 Conclusion

A methodology has been presented in this chapter for evaluating reliability indices of repairable
systems employing the Markov process. This methodology gives analysis for the evaluation of
frequency failure, MUT, and MDT, along with probabilities of success and probabilities of fail-
ure. The Markov process was utilized to evaluate state probabilities based on the state transition
diagram. Steady-state solution of state equations has been demonstrated to evaluate availability
and unavailability based on state transition diagram of single-component, two-component, and
three-component models. Further reduced state transition diagrams are used to evaluate the avail-
ability and unavailability of identical components, assuming equal failure rates of components and
equal repair rates of components. Illustrative examples are also presented to show the use of state
enumeration technique for reliability indices evaluation. Therefore, this chapter presents a method-
ology to evaluate frequency and duration indices, which will be helpful for power system planning
engineers to assess the reliability of the system in the perspective of reliability analysis.

References

Arya, L.D. and Koshti, A. (2008). Distributed generation capacity reliability evaluation using safety
index. Journal of Institution of Engineering (India) 89: 3–7.
Arya, L.D., Koshti, A., and Choube, S.C. (2013). Frequency-duration analysis of composite distribution
system using a non-sequential Monte Carlo simulation. Electrical Power and Energy Systems 46:
17–25.
Billinton, R. and Allan, R.N. (1983). Reliability Evaluation of Engineering Systems, 2e. Springer.
Billinton, R. and Allan, R.N. (1996). Reliability Evaluation of Power Systems, 2e. Plenum Press.
Ebeling, C.E. (2000). An Introduction to Reliability and Maintainability Engineering. McGraw Hill
Education (India) Edition.
Endrenyi, J. (1978). Reliability Modeling in Electric Power Systems. Wiley-Blackwell.
Halve, S.S., Koshti, A., and Arya, R. (2023). A sampling method based on system state transition for
distribution system adequacy assessment using distributed generation. Journal of Operation and
Automation in Power Engineering 11 (4): 249–257.
Koshti, A.S., Verma, A., Arya, R. et al. (2023). Tolerable random interruption duration based reliability
estimation of stand alone hybrid renewable energy system (SAHRES) by network reduction and
sequential Monte Carlo simulation. IET Generation Transmission & Distribution 1–12. (Early view)
https://doi.org/10.1049/gtd2.12910.
Lewis, E.E. (1987). Introduction to Reliability Engineering. John Wiley & Sons.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
51

Methodologies for Reliability Evaluation of Network

Rajesh Arya 1 , Atul Koshti 2 , Aanchal Verma 3 , Baseem Khan 4 , and Liladhar Arya 5
1 Department of Information Technology, Acropolis Institute of Technology and Research, Indore, Madhya Pradesh, India
2
Department of Electrical Engineering, GES’S R.H. Sapat College of Engineering, Management Studies & Research, Nashik,
Maharashtra, India
3
Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
4 Department of Electrical and Computer Engineering, Hawassa University, Hawassa, Ethiopia
5
Department of Electrical Engineering, SGSITS, Indore, Madhya Pradesh, India

4.1 Introduction
This chapter presents reliability evaluation methodologies which are represented by reliability net-
work or logic diagram. The reliability network may be a series or parallel or series-parallel or a
complex configuration. It is to be stressed here that the reliability networks in no way represent the
connection of components in a physical system. For example two transmission lines operating in
electrically parallel will be considered in series configuration for reliability calculation if successful
operation of both is required to meet the demand. The relation used is derived will be applicable for
reliability evaluation without time factor or with time dependents function under the assumption
that all the component, are put in service simultaneously (this will not be the case in a standby
system). Generally, it is assumed that components in system fail independently.

4.2 Series Network

A series network is simplest and block diagram representation of it is shown in Figure 4.1.
Where I = Input, O = Output denote input and output nodes.
In terms of its physical setup, a system configuration can be classified as a series reliability net-
work when it either experiences a failure when a single component fails or remains operational
when all its components are functioning properly. This is also known as nonredundant system
(Billinton and Allan 1992; Saket et al. 2022). Assuming time-independent reliabilities r1 , r2 , … , rn ,
the reliability of a series system can be expressed as follows:
∏
n
Rs = ri (4.1)
i=1

Sometimes this is also known as static reliability.

Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
52 4 Methodologies for Reliability Evaluation of Network

Figure 4.1 Series reliability network.

1 2 n

Now assume T1 , T2 , … , Tn are life lengths of these components and f1 (t), f2 (t), … , fn (t) are the
failure density function of respective components. Life length “T” of the series system will be min-
imum of individual life lengths, i.e.

T = min [T1 , T2 , … , Tn ] (4.2)

Failure density function of “T” is calculated using following relation

(T > t) = (T1 > t) ⋅ (T2 > t) ⋅ · · · ⋅ Tn > t) (4.3)

The above relation implies that if (T > t) then all T1 , T2 , … , Tn must be greater than “t” (if “T” is
minimum). Above relation can be written in terms of probability as follows

P(T > t) = P(T1 > t) ⋅ P(T2 > t) ⋅ · · · ⋅ (Tn > t) (4.4)

The relationship stated above is expressed using the reliability function as follows:
∏
n
Rs = ri (t) (4.5)
i=1

Failure density function of series system may also obtained using (4.4)
∏
n
( )
1 − F(t) = 1 − Fi (t)
i=1
∏
n
( )
F(t) = 1 − 1 − Fi (t)
i=1

Failure density function of series system

{ } { n }
d ∏n
( ) d ∏( )
f (t) = 1− 1 − Fi (t) =− 1 − Fi (t)
dt i=1
dt i=1

If

f1 (t) = f2 (t) = · · · = fn (t) = 𝜑(t)

d ∏ d
f (t) = − Ri (t) = − Rn (t)
dt dt
= nRn−1 (t) 𝜑(t)

Let us consider that each component maintains a consistent failure rate denoted by 𝜆1 , 𝜆2 , … , 𝜆n
∏
n
Rs (t) = e−𝜆i t
i=1
Rs (t) = e−𝜆s t
∑
𝜆s = 𝜆i

Failure density function of series system

f (t) = 𝜆s e−𝜆s t
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4.3 Parallel Network 53

Hence, the series system also has exponential failure law. Failure rate is sum of individual failure
Rates (Arya et al. 2013; Halve et al. 2023). Further mean time to failure (MTTF) of system
1 1
MTTF = = ∑
𝜆s 𝜆i
Reliability importance IRk of kth component is given as for a specified time
dRs ∏
n
IRk = = r
drk i=1i≠k i

It is obvious for series system, the reliability importance is highest of “weakest” component.
Hence, the component which has the lowest reliability may be improved in design/quality/
strengthening if at all overall system reliability need to be improved.

4.3 Parallel Network

A parallel network consisting of “n” component is shown in Figure 4.2.

The system can fulfill its function as long as a single component is operational. Because all com-
ponents are interconnected within the system, this configuration is also referred to as an active
parallel setup. Probability of failure of the system i.e. unreliability is given by (if there is reliability
or independent time factor).
∏
n
Qp = (1 − ri )
i=1
∏
n
Rp = 1 − (1 − ri )
i=1

Above relation is expanded and written as

∑ ∑ ∑
Rp = ri − ri rj + ri rj rk − · · · + (−1)n−1 r1 r2 ...rn
i

If component are identical then

Rp = 1 − (1 − r)n
Now assume each component has f1 (t), f2 (t), … , fn (t) as failure density function and correspond-
ing r1 (t), r2 (t), … rn (t) as reliability functions. Life length of system is given as
T = max [T1 , T2 , … , Tn ]

Figure 4.2 Parallel reliability network.

1

n
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
54 4 Methodologies for Reliability Evaluation of Network

Following is written

(T ≤ t) = (T1 ≤ t) ⋅ (T2 ≤ t) ⋅ · · · ⋅ (Tn ≤ t)

P(T ≤ t) = P(T1 ≤ t) ⋅ P(T2 ≤ t) ⋅ · · · ⋅ P(Tn ≤ t)
G(t) = F1 (t)F2 (t)...Fn (t)

Therefore Cumulative distribution function of parallel network

∏
G(t) = Fi (t)

Failure law of parallel network

d ∏
g(t) = F (t)
dt ∏ i
R(t) = 1 − Fi (t)
∏
R(t) = 1 − (1 − ri (t))

If each component has identical failure law

d n
g(t) = F (t)
dt
g(t) = nF n−1 (t) f (t)
R(t) = 1 − (1 − r(t))n

Let us assume each component has constant failure rate 𝜆1 , 𝜆2 , … , 𝜆n

ri (t) = e−𝜆i t
∏( )
Rp (t) = 1 − 1 − e−𝜆i t
i

Failure density function of parallel system

d ∏( )
g(t) = 1 − e−𝜆i t
dt
Hazard rate of the system (Ebeling 2019) is given as-
d ∏( )
1 − e−𝜆i t
h(t) = dt ∏ ( )
1− 1 − e−𝜆i t
It is noted that each component has constant failure rate, but the failure rate is time-dependent.
The plot of the hazard rate is given in Figure 4.3 as increasing.
{ }
𝜆k = min 𝜆1 , 𝜆2 , … , 𝜆n
MTTF of the parallel system is given as
[ ]
∞ ∏( )
MTTF = 1− 1 − e−𝜆i t dt
∫0 i
[ ]
∞ ∑ ∑ ∑ ∑
−𝜆i t −(𝜆i +𝜆j )t −(𝜆i +𝜆j +𝜆k )t n − 𝜆i t
MTTF = e − e + e + · · · + (−1) e dt
∫0 i j>i k>j>i
∑1 ∑ 1 ∑ 1 1
= − + + · · · + (−1)n ∑
i
𝜆 i j>i
(𝜆 i + 𝜆 j ) k>j>i i
(𝜆 + 𝜆 j + 𝜆 k ) 𝜆i
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4.3 Parallel Network 55

Figure 4.3 Plot of hazard rate versus time.

h (t) 2λ

For example n = 2
1 1 1
MTTF2 = + −
𝜆 1 𝜆 2 𝜆 1 + 𝜆2
For n = 3
1 1 1 1 1 1 1
MTTF3 = + + − − − +
𝜆 1 𝜆 2 𝜆 3 𝜆 1 + 𝜆2 𝜆 2 + 𝜆3 𝜆 3 + 𝜆1 𝜆 1 + 𝜆2 + 𝜆3
If n units are identical having each as constant failure rate 𝜆 than expression for MTTF is given as
∑n
n Ck (−1)
k+1
MTTF =
k=1
k𝜆
It is true that reliability of parallel system increases as more number of components are con-
nected. But common failure may not permit to achieve ideal results. The cause of common failure
includes common operational environment, external catastrophic events, operational errors, com-
mon external power source, etc.
The cause of common mode failure (CMF) effects (Allan 2013) is modeled by connecting a hypo-
thetical unit in series with a parallel combination representing the reliability owing common cause
failure as shown in Figure 4.4.
𝜆1 , 𝜆2 are failure rate of each unit independently and is failure rate due to common cause. The
reliability of the system is written as
[ ]
R = e−𝜆1 t + e−𝜆2 t + e−(𝜆1 +𝜆2 )t e−𝜆c t
It is necessary that in physical consideration should be small; otherwise, effect of redundancy
may be neglected. In other words total failure rate of each component may be written as
𝜆1T = 𝜆1 + 𝜆c 𝜆2T = 𝜆2 + 𝜆c
Putting in previous relation resulting into
[ ]
R = e−𝜆1T t + e−𝜆2T t + e−(𝜆1T +𝜆2T −𝜆c )t

Figure 4.4 Common mode failure.

1
CMF unit
λ1

λc
2

λ2
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
56 4 Methodologies for Reliability Evaluation of Network

If units are identical then

𝜆1 = 𝜆2 = 𝜆 𝜆1T = 𝜆2T = 𝜆T
Above relation will be written as
[ ]
R = 2e−𝜆T t − e−(2𝜆T −𝜆c )t
At times, a factor 𝛽 is defined as the ratio of common cause failure to total failure rate as
𝜆c
𝛽=
𝜆T
𝛽-factor may be substituted in above relation and reliability expression is written as follows
[ ]
R = 2e−𝜆T t − e−(2𝜆T −𝛽𝜆T )t
[ ]
R = 2e−𝜆T t − e−(2−𝛽)𝜆T t
[ ]
R = 2 − e−(1−𝛽)𝜆T t e−𝜆T t
At times, this technique is known 𝛽 factor technique. General expression for “n” component in
parallel accounting common cause failure is written as
[ ]
R(t) = 1 − (1 − e−𝜆t )n e−𝜆c t

4.4 Partially Redundant System

This arrangement is also known as k-out of n configuration. In this “n” units are operating in par-
allel, the success requires at least “k” unit in operation or in upstates, where k is given as
1 < k < n.
if k = 1 it is fully redundant system and k = n it is series system (nonredundant system). k∕n system
network is represented as shown in Figure 4.5.
Assuming each component is identical and has reliability “r” for a specified duration. Using the
binomial distribution probability of “m” success out of “n” unit is written as follows:
p [X = m] =n Cm r m (1 − r)n−m
Since at least “k” units are required for system success. Reliability expression is written as
∑
n
n
Rk∕n = Cm r m (1 − r)n−m
m=k

Figure 4.5 k∕n partially redundant system.

1

2
k/n
3

n
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4.5 Reliability Evaluation of Complex Networks 57

If each component has failure rate 𝜆, then the reliability is given as Rk∕n

∑
n
n
Rk∕n = Cm e−𝜆mt (1 − e−𝜆t )n−m
m=k

Using above failure density function, hazard rate and MTTF of partially redundant system may
be calculated. If the units are not identical than Binomial distribution is not applicable. In this situ-
ation, a product in the following manner is formed and states where at least “k” units are available
are selected. Their probabilities are calculated and added to give the required results.

(p1 + q1 )(p2 + q2 )...(pn + qn )

= p1 p2 ...pn + q1 p2 ...pn + p1 q2 ...pn + · · · + q1 q2 ...qn

In above total events will be 2n. Success events may be separated to calculate reliability of system.
Exercise 4.1: A system of three components, each having failure rate as 𝜆1 , 𝜆2 , 𝜆3 . System success
requires at least two components in operation. Obtain (1) reliability function and(2) MTTF.
Solution:

(1) From the product (r1 + q1 )(r2 + q2 )(r3 + q3 ) = r1 r2 r3 + q1 r2 r3 + r1 q2 r3 + r1 r2 q3 + q1 q2 r3 +

q 1 r2 q3 + r1 q2 q3 + q1 q2 q3
Selecting the events which forms system success

R2∕3 = r1 r2 r3 + q1 r2 r3 + r1 q2 r3 + r1 r2 q3
= e−(𝜆1 +𝜆2 )t + e−(𝜆2 +𝜆3 )t + e−(𝜆1 +𝜆3 )t − 2e−(𝜆1 +𝜆2 +𝜆3 )t

(2)
1 1 1 2
MTTF = + + −
𝜆 1 + 𝜆2 𝜆 2 + 𝜆3 𝜆 1 + 𝜆3 𝜆 1 + 𝜆 2 + 𝜆3

4.5 Reliability Evaluation of Complex Networks

This section deals with the reliability evaluation of a non-series-parallel network (Breneman et al.
2022; Endrenyi 1978). The following four methods are discussed

(1) Event space,

(2) Decomposition,
(3) Tie-set, and
(4) Cut-set.

Simulation-based techniques may also be employed for obtaining solution of the power system
reliability (Arya and Koshti 2008; Koshti et al. 2023).

4.5.1 Event Space Method

This is one of the simplest techniques for evaluating the reliability of a network. The method is
implemented in the following steps.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
58 4 Methodologies for Reliability Evaluation of Network

c Figure 4.6 Event space tree for three

abc components.
b
–
a C abc–
–
– c abc
b
– –
C abc–
–
abc
c
b
–
C – –
abc
–a –
– –
abc
b c
–
C –––
abc

Step 1: Based on the total number of components, enumerate all possible states of the network.
Assuming each component has two states (up and down), total no. of states will 2n. n
denotes total no. of components. These events may be identified using a event tree, as
shown in Figure 4.6 for three components, namely a, b, c. The event tree will grow in size
as the number of component increases.
Step 2: Looking at the network, partition the total number of states in two sets i.e. “W”—success
set of event and “F”—fail set of event.
Step 3: Evaluate the probabilities of these events and calculate the system’s reliability as outlined
below
∑
R= P(Ei )
i∈w
∑
R=1−R= P(Ej )
j∈f

Ei and Ej are events

P(Ei ) the is probability of occurrence of ith event

Note: Simple addition in above relation is due to the event enumerated being mutually exclusive.
Exercise 4.2: Evaluate the reliability of following network using event space method.
Assume ra , rb , rc , and rd are the reliabilities of each component as shown in Figure 4.7.
Solution:
The network will have 16 states written as follows
E1 = abcd E9 = abcd
E2 = abcd E10 = abcd
E3 = abcd E11 = abcd
E4 = abcd E12 = abcd
E5 = abcd E13 = abcd
E6 = abcd E14 = abcd
E7 = abcd E15 = abcd
E8 = abcd E16 = abcd
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4.5 Reliability Evaluation of Complex Networks 59

Figure 4.7 Reliability evaluation using event space method.

a b

c d

Matching the event with the network following events are in “w” sets
W ∶ [E1 , E2 , E3 , E4 , E5 , E7 , E9 , E13 ]
R = P(E1 ) + P(E2 ) + P(E3 ) + P(E4 ) + P(E5 ) + P(E7 ) + P(E9 ) + P(E13 )
R = P(abcd) + P(abcd) + P(abcd) + P(abcd) + P(abcd) + P(abcd) + P(abcd) + P(abcd)
R = ra rb rc rd + ra rb (1 − rc )rd + ra rb rc (1 − rd ) + ra rb (1 − rc )(1 − rd ) + (1 − ra )rb rc rd
+ (1 − ra )rb rc (1 − rd ) + ra (1 − rb )rc rd + (1 − ra )(1 − rb )rc rd
If ra = rb = rc = rd = r
the above relation is simplified as
R = r 4 + 4r 3 (1 − r) + 3r 2 (1 − r)2
R = 3r 2 − 2r 3
Further assume each component has constant failure rate 𝜆, i.e. r = e−𝜆t . The above reliability
expression reduces to
R(t) = 3e−2𝜆t − 2e−3𝜆t
Failure density function of the system is evaluated as follows
f (t) = 6𝜆e−2𝜆t − 6𝜆e−3𝜆t
And hazard rate is expressed as-
6𝜆e−2𝜆t − 6𝜆e−3𝜆t
h(t) =
3e−2𝜆t − 2e−3𝜆t
MTTF of the system is given as
3 2 5
MTTF = − =
2𝜆 3𝜆 6𝜆
The method is simple but as number of component increases states also increases and thus com-
putational burden increases. Total number of states are 2n, if n denotes total no. of component, but
total number of states can be reduced by considering the failure of a small number of components.
Since failure of large number of components is unlikely i.e. probability of such events/ states with
failure of more than a specific number of component is negligible.

4.5.2 Decomposition Method

In this technique, a key-stone component is selected as “X” and based on this component’s state
system success is written as follows
S=S⋅X +S⋅X
Probability of system success is given as follows
P(S) = P(S ⋅ X) + P(S ⋅ X)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
60 4 Methodologies for Reliability Evaluation of Network

Decomposition is based on that system success will occur either in intersection with “X” working
or with intersection “X” has failed. From the above relation, reliability is written as follows
P(S) = P(S∕X)P(X) + P(S∕X)P(X)
Where, P(X) = Reliability of “X” component
P(S∕X) = Probability of system success provided “X” is working.
P(S∕X) = Probability of system success given that “X” has failed (X)
The above decomposition is shown in Figure 4.8.
As shown in Figure 4.8, the network is decomposed into two networks “A” and “B.” In “A” sys-
tem, success is represented when “X” is working which is shown as “short-circuit,” whereas in “B”
system success is evaluated while “X” has failed which is shown as “open-circuit” condition. The
methodology is explained by evaluating the reliability of the following system shown in Figure 4.9.
Let us assume “X” as key-stone which is component “c”
P(C) = P(X) = rc
P(C) = P(X) = (1 − rc )
Decomposed network corresponding to P(S∕X) is shown in Figure 4.10.
It is easier to evaluate the P(S∕X) from above network. Close observation of above network
reveals that reliability of above network is simply “b” and “d” in parallel combination i.e.
P(S∕X) = rb + rd − rb rd
The original network for P(S∕X) is evaluated as follows

P(s) Network

x –x

–
P(s/x)
P(s/x)
A B

Figure 4.8 Decomposition method illustration.

Figure 4.9 Reliability evaluation using decomposition

a b method.

Input Output
c d
x

Figure 4.10 Decomposed network for X.

a b

d
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4.5 Reliability Evaluation of Complex Networks 61

Figure 4.11 Decomposed network for X. a b

From Figure 4.11 P(S∕X) is evaluated as follows:

P(S∕X) = ra rb
Hence over all reliability of system is written as
P(S) = (rb + rd − rb rd )rc + ra rb (1 − rc )
P(S) = rb rc + rc rd − rb rc rd + ra rb − ra rb rc
Note: If each component has reliability “r” then
P(S) = 3r 2 − 2r 3 as obtained using event space method.
Reliability importance of each component is given as follows
𝜕P(S)
= rb − rbrc = rb(1 − rc)
𝜕ra
𝜕P(S)
= rc (1 − rd ) + ra (1 − rc )
𝜕rb
𝜕P(S)
= rb + rd − rb rd − ra rb
𝜕rc
𝜕P(S)
= rc − rb rc = rc (1 − rb )
𝜕rd
If each component has same reliability then
𝜕P(S) 𝜕P(S)
= = r − r2
𝜕ra 𝜕rd
𝜕P(S) 𝜕P(S)
= = 2(r − r 2 )
𝜕rb 𝜕rc
Hence it is obvious component “b” and “c” has twice reliability importance as compared to “a”
and “d.” Hence if system reliability is to be improved then it should be improved by investing money
in components “b” and “c” both or “b” or “c.”

4.5.3 Tie-Set Method

This method is also known as path tracing method. A tie-set is a set of components which provide
path from input to output while all of the components in the set are in upstate. Minimal tie-set
is a set of minimum number of components which while all of them in upstate provides a path
(success). Even if a single component is removed from this “minimal tie-set,” the remaining set of
component no longer provides a path (success) even if all of them are in upstate. The methodology
is explained in the following steps
Step 1: Given a reliable network, obtain all possible minimal tie-sets as
[ ]
T1 , T2 , T3 , … , Tr
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
62 4 Methodologies for Reliability Evaluation of Network

Figure 4.12 Equivalent network using tie sets.

T1

T2

Tr

Step 2: For system success, at least one tie-set must be available. Thus equivalent network is drawn
in Figure 4.12.

Hence, the reliability expression is written as follows:

( r )
∑
R=P Ti
i=1
∑ ∑ ∑
R= P(Ti ) − P(Ti Tj ) + P(Ti Tj Tk ) + · · · + (−1)r−1 P(T1 T2 ......Tr )
i j>i k>j>i

Consider the complex network used in previous sub-section. Looking at the network Tie-sets are
obtained as follows T1 = a ⋅ b, T2 = c ⋅ d, T3 = b ⋅ c

R = P(T1 ) + P(T2 ) + P(T3 ) − P(T1 T2 ) − P(T1 T3 ) − P(T2 T3 ) + P(T1 T2 T3 )

R = ra rb + rc rd + rb rc − ra rb rc − rb rc rd

If ra = rb = rc = rd

R = 3r 2 − 2r 3

4.5.4 Cut-Set Method

A cut-set is defined as a group of components such that the system fails if any of them fails. A
minimal cut-set is a set of component which fail system fails; even if a single component is removed
from minimal cut-set the remaining components do not form a cut-set. The method is explained
in following steps.

Step 1: Identify all possible minimal cut-sets as C1 , C2 , … , CT

Step 2: For system failure at least one cut-set should occur i.e. system will fail even if all the
elements in any one of the minimal “cut-set” fails. The equivalent diagram is shown in
Figure 4.13.

In view of this unreliability, the system is written as follows:

Q = 1 − R = P(C1 + C2 + · · · · · · + Cr )
∑ ∑ ∑
Q= P(Ci ) − P(Ci Cj ) + P(Ci Cj Ck ) + · · · + (−1)r−1 P(C1 C2 ......Cr )
i j>i k>j>i

Let us consider the following example of Figure 4.14

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4.6 Determination of Tie-Sets 63

C1 C2 Cr

Figure 4.13 Equivalent network using cut-sets.

Figure 4.14 Reliability network.

a b

c d

Following cut-sets are identified

C1 = a c C2 = b d
C3 = b c C4 = c e b
Q = P(C1 ) + P(C2 ) + P(C3 ) + P(C4 ) − P(C1 C2 ) − P(C1 C3 ) − P(C1 C4 ) − P(C2 C3 )
− P(C2 C4 ) − P(C3 C4 ) + P(C1 C2 C3 ) + P(C1 C2 C4 ) + P(C2 C3 C4 ) + P(C3 C4 C1 )
− P(C1 C2 C3 C4 )
If each component has the same reliability r then resulting into
Q = 3(1 − r)2 − (1 − r)3 − 2(1 − r)4 + (1 − r)5
The cut-set method has the following advantages
(i) The method can be programmed on a digital computer. No doubt tie-set method also can be
programmed.
(ii) While one identifies cut-set, various failure modes are directly available. This enables one to
strengthen the weak components or devices.
(iii) In cut-set method approximate evaluation of reliability is possible. This may be achieved in
one or two following ways
(a) Limiting the order of cut-sets i.e. neglecting the higher-order cut-sets which contain more
than a specified number of elements. Since probabilities of occurrence of such “cuts” may
be negligible.
(b) By truncating the unreliability relation “Q” by retaining only first-order terms as follows.
∑
Q≅ P(Ci )
R≅1−Q

Failure probability thus obtained will be on higher side and reliability will be on lower side.
Usually in engineering application, each component is highly reliable and above approximations
are justified in practice.
No doubt decomposition method is very powerful but it cannot be programmed on computer.
Moreover, it may not be used when large numbers of components are present in the system.

4.6 Determination of Tie-Sets

Method for obtaining tie-sets which are required in Tie-Set method of reliability evaluation.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
64 4 Methodologies for Reliability Evaluation of Network

4.6.1 Connection Matrix Method Using Node Elimination

Assuming a reliability network is given, form a connection matrix N defined as follows

Nij = 1

Nij = Sum of the elements connected between and directed from ith to jth node
= 0 Otherwise
Eliminate one by one all intermediate nodes using following formula

N ′ ij = Nij + Nik Nkj

where “k” is the node being eliminated. Ultimately after eliminating all intermediate nodes a 2×2
matrix will result. All minimal tie-set can be obtained from this matrix from Ninput × output
element. In this process at times a nonminimal tie-set may result which should be ignored and
identified.
Exercise 4.3: Obtain tie-sets of the following network shown in Figure 4.15 using Node elimina-
tion method
Solution:
By inspection connection matrix is written as follows

⎡1 c a 0⎤
⎢ ⎥
0 1 1 d⎥
N=⎢
⎢0 0 1 b⎥
⎢0 0 0 1⎥⎦
⎣
node k = 2 to be eliminated

N ′ 11 = 1
N ′ 13 = N13 + N12 N23 = a + c ⋅ 1 = a + c
N ′ 14 = N14 + N12 N24 = 0 + c ⋅ d = cd
N ′ 31 = 0
N ′ 33 = 1
N ′ 34 = N34 + N32 N24 = b + 0 ⋅ d = b
N ′ 41 = 0
N ′ 43 = 0
N ′ 44 = 1

3 Figure 4.15 Tie-sets using node elimination method.

a b
4

1
c d
2
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4.7 Method of Obtaining Cut-Set 65

The resulting matrix is written

⎡1 a + c cd⎤
N=⎢0 1 b⎥
⎢ ⎥
⎣0 0 1⎦
k=3
N ′′ 11 = 1
N ′′ 14 = N ′ 14 + N ′ 13 N ′ 34
= cd + (a + c) ⋅ b
= cd + ab + bc
N ′′ 41 = 0
N ′′ 44 = 1
[ ]
1 ab + bc + cd
N ′′ =
0 1
Tie-sets are obtained as element of N ′ 14 i.e.
T1 = ab, T2 = bc, T3 = cd

4.7 Method of Obtaining Cut-Set

Once tie-set are obtained, a matrix in the following format is prepared Aik represent an element
present in ith path, and kth column is considered for this element. This means Aik = 1, if ith tie-set
contains kth element and 0 otherwise. Then, if a specific column contains all ones, then the corre-
sponding elements form a first-order cut. Now examine sum of two column’s combinations (exclud-
ing the first-order cut). Combinations giving one at every place will form second-order cut. Thus,
continuing the increased numbers of combinations higher-order cuts are obtained. It is stressed
here that while looking at combinations, the combinations of lower order which forms cut should
not be considered. The procedure is illustrated with the network shown in Figure 4.16.
The following three tie-sets are obtained
T1 = ab, T2 = cd, T3 = bce
Following matrix is prepared

Elements a b c d e
Tie sets

T1 1 1 0 0 0

T2 0 0 1 1 0

T3 0 1 1 0 1

It is obvious from the above table; there is no column which contains all ones. Hence there is no
first-order cut-set.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
66 4 Methodologies for Reliability Evaluation of Network

Figure 4.16 Bridge reliability network.

a b

c d

Taking combination of two at a time following second order cut-sets are observed

C1 = a c, C2 = b c, C3 = b d
Looking at three combinations following one minimal cut-set is observed

C4 = c e b

4.8 Multistate Model

It has been considered that the component has two states i.e. “up” and “down.” In numerous sce-
narios, components may exhibit more than two states. A generator may have “up,” “down,” and
a “derated” states. A diode may function properly; fail by open circuit or short circuit. Thus diode
has two failure modes and one state of normal operation. Reliability of such systems are evaluated
either by event-space or decomposition method.

4.8.1 Event-Space Method

Consider the example of two diodes operating in parallel for rectification purpose as shown in
Figure 4.17.
Reliability of above configuration is required for rectification purpose. Assume Pn1 , Po1 , Ps1 are
the probabilities of normal, open circuit, and short circuit for diode one and Pn2 , Po2 , Ps2 are the
same probabilities of diode number two. Events may be obtained by product rule or by tree.
( )( )
P01 + Ps1 + Pn1 P02 + Ps2 + Pn2
= P01 P02 + P01 Ps2 + P01 Pn2 + Ps1 P02 + Ps1 Ps2 + Ps1 Pn2 + Pn1 P02 + Pn1 Ps2 + Pn1 Pn2
In all there are nine states. States corresponding to system success are added and reliability is
evaluated as
Pn′ = Pn1 Pn2 + P01 Pn2 + P02 Pn1
Probability of configuration failing by short circuit is given by
Ps′ = Ps1 Ps2 + P01 Ps2 + P02 Ps1 + Pn1 Ps2 + Pn2 Ps1

1 Figure 4.17 Two diodes in parallel.

2
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4.8 Multistate Model 67

Probability of configuration failing by open circuit is given as

Po′ = Po1 Po2

In view of this, above configuration is replaced by a single diode shown in Figure 4.18
The equivalent diode has as probabilities of normal, open, and short cir-
cuit states. If the diodes are identical i.e.

Pn1 = Pn2 = Pn′

Figure 4.18
Po1 = Po2 = Po′
Equivalent diode of
Ps1 = Ps2 = Ps parallel combination.

Then reliability of the configuration is given as

Pn′ = Pn 2 + 2Pn P0

If configuration is in series as follows (Figure 4.19)

In this case total numbers of states are same i.e. nine as earlier but the equivalent three states are
identified as follows

Pn′′ = Pn1 Pn2 + Pn1 Ps2 + Pn2 Ps1

Po′′ = Po1 Po2 + Po1 Pn2 + Po2 Pn1 + Po1 Ps2 + Po2 Ps1
Ps′′ = Ps1 Ps2

The equivalent diode has above three state probabilities

Again if each diode has same probabilities then

Pn′′ = Pn 2 + 2Pn Ps
Po′′ = Po 2 + 2Po Pn + 2Po Ps
Ps′′ = Ps 2

It is obvious from expressions of and that for reliability enhancement two diodes should be con-
nected in series if probability of failure of diode by short circuit is more than that of probability of
failure by open circuit, otherwise they should be connected in parallel.

4.8.2 Decomposition Approach

Assume the two diodes have been connected in parallel as shown below
Assume diode # 2 as key-stone as X. Then system success “S” and reliability is written as follows

S = S ⋅ X0 + S ⋅ XN + S ⋅ XSC
P(S) = P(S∕X0 )P(X0 ) + P(S∕XN )P(XN ) + P(S∕XSC )P(XSC )
= pn1 p02 + (pn1 + p01 )pn2 + 0 ⋅ ps2
= Pn1 p02 + pn1 pn2 + p01 pn2

Figure 4.19 Two diode in series configuration. 1 2

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
68 4 Methodologies for Reliability Evaluation of Network

1 Figure 4.20 Diode in parallel with X as key stone.

Figure 4.21 Series parallel diode configuration.

1

Assume again following configuration Parallel combination of diode may be assumed as a single
component “X.” (Figures 4.20 and 4.21) Then reliability expression is written as
P(S) = P(S∕XN ) ⋅ P(XN ) + P(S∕XSC ) ⋅ P(XSC ) + P(S∕X0 ) ⋅ P(X0 )
= (Ps3 + Pn3 )(Pn1 ⋅ Pn2 + P01 ⋅ Pn2 + P02 ⋅ Pn1 ) + Pn3 (ps1 ⋅ ps2 + P01 ⋅ Ps2 + P02 ⋅
Ps1 + Pn1 ⋅ Ps2 + Pn2 ⋅ Ps1 ) + 0 ⋅ (P01 ⋅ P02 )
If all diodes are identical then above reduces to
P(S) = (Ps + Pn )(Pn 2 + 2Pn P0 ) + Pn (Ps 2 + 2P0 Ps + 2Pn Ps )

4.9 Illustrative Examples

Exercise 4.4: An electrical radial distributor has 10 distributer sections to meet farthest load point.
Each distributer section has reliability “r.” What should be minimum value of “r” such that farthest
load point reliability does not drop below 0.90.
Solution:
This is series system and overall reliability to farthest load point is
r 10 = 0.90
10 ln r = ln 0.90
ln r = 0.1 ln 0.90
Gives
r = 0.9898
Exercise 4.5: An important hospital load to be meet by parallel feeder in which each feeder has
reliability 0.9. Determine minimum number of parallel feeder such that reliability to meet these
load is 0.99.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4.9 Illustrative Examples 69

Solution:
Reliability expression for “n” component in parallel is

Rp = 1 − (1 − r)n = 0.99
= 1 − (1 − 0.9)n = 0.99
0.1n = 0.01

Taking natural logarithm

n ln 0.1 = ln 0.01
0.01
n=
ln 0.1
≅2

Exercise 4.6: An administrative block distributed load is supplied by three parallel feeder each
having capacity of 50 MW and load is 80 MW. Each feeder has availability of 0.95. Determine the
availability of system.
Solution:
To meet a demand of 80 MW of total capacity of three feeders 150 MW at least two feeders must be
operable. Hence it is a partially redundant system

n = 3, k = 2

System availability is written as (using binomial distribution)

R = r 3 + 3r 2 (1 − r) = 3r 2 − 2r 3
= 3 × 0.952 − 2 × 0.953
= 0.9025

Exercise 4.7: Obtain reliability expression for the following bridge network in Figure 4.22 by all
four methods discussed in the chapter. Assume reliability of each component as ra = rb = rc = rd =
re = r = 0.99
Solution:

(a) Event space method

There are five components and will have 32 states.

E1 = abcde E9 = abcde
E2 = abcde E10 = abcde
E3 = abcde E11 = abcde
E4 = abcde E12 = abcde

Figure 4.22 Bridge network.

a b

c d
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
70 4 Methodologies for Reliability Evaluation of Network

E5 = abcde E13 = abcde

E6 = abcde E14 = abcde
E7 = abcde E15 = abcde
E8 = abcde E16 = abcde
E17 = abcde E25 = abcde
E18 = abcde E26 = abcde
E19 = abcde E27 = abcde
E20 = abcde E28 = abcde
E21 = abcde E29 = abcde
E22 = abcde E30 = abcde
E23 = abcde E31 = abcde
E24 = abcde E32 = abcde
{ }
W: E1 , E2 , E3 , E4 , E5 , E7 , E9 , E10 , E13 , E17 , E18 , E19 , E20 , E21 , E25 , E29
R = P(E1 ) + P(E2 ) + P(E3 ) + P(E4 ) + P(E5 ) + P(E7 ) + P(E9 ) + P(E10 ) + P(E13 ) + P(E17 )
+ P(E18 ) + P(E19 ) + P(E20 ) + P(E21 ) + P(E25 ) + P(E29 )
= ra rb rc rd re + ra rb (1 − rc )rd re + ra rb rc (1 − rd )re + ra rb (1 − rc )(1 − rd )re
+ (1 − ra )rb rc rd re + (1 − ra )rb rc (1 − rd )re + ra (1 − rb )rc rd re
+ ra (1 − rb )(1 − rc )rd re + ra rb (1 − rc )rd (1 − re ) + (1 − ra )(1 − rb )rc rd re
+ ra rb (1 − rc )rd (1 − re ) + ra rb rc (1 − rd )(1 − re ) + ra rb (1 − rc )(1 − rd )(1 − re )
+ (1 − ra )rb rc rd (1 − re ) + (1 − ra )(1 − rb )rc rd (1 − re ) + ra (1 − rb )rc rd (1 − re )
if ra = rb = rc = rd = re = r = 0.99
R = 2r 5 − 5r 4 + 2r 3 + 2r 2

if r = 0.99 then

R = 0.99979805

(b) Decomposition method

The problem can be very easily solved by selecting key stone as component “e.”

P(X) = P(e) = re
P(X) = P(e) = (1 − re )

Then conditional probability P(S/X) is written as follows:

P(S∕X) = (ra + rc − ra rc ) + (rb + rd − rb rd ) − (ra + rc − ra rc )(rb + rd − rb rd )

P(S∕X) = (ra rb + rc rd − ra rb rc rd )
[ ] [ ]
R = (ra + rc − ra rc ) + (rb + rd − rb rd ) re + (ra rb + rc rd − ra rb rc rd ) (1 − re )
If ra = rb = rc = rd = re = r

Following reliability expression is obtained.

R = 2r 5 − 5r 4 + 2r 3 + 2r 2 if r = 0.99 then

R = 0.99979805
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4.9 Illustrative Examples 71

(c) Tie-set method

Following minimal tie-sets are obtained.
T1 = ab, T2 = cd, T3 = aed, T4 = ceb
[ ]
R = P T1 + T2 + T3 + T4
R = P(T1 ) + P(T2 ) + P(T3 ) + P(T4 ) − P(T1 T2 ) − P(T1 T3 ) − P(T1 T4 ) − P(T2 T3 ) − P(T2 T4 )
− P(T3 T4 ) + P(T1 T2 T3 ) + P(T1 T2 T4 ) + P(T2 T3 T4 ) + P(T3 T4 T1 ) + P(T1 T2 T3 T4 )

Putting
If ra = rb = rc = rd = re = r

R = 2r 5 − 5r 4 + 2r 3 + 2r 2

if r = 0.99 then

R = 0.99979805

(d) Cut-set method

Following minimal cut-sets are obtained.

C1 = a c C2 = b d
C3 = a e d C4 = b c e
Unreliability expression is written as
Q=1−R
Q = P(C1 ) + P(C2 ) + P(C3 ) + P(C4 ) − P(C1 C2 ) − P(C1 C3 ) − P(C1 C4 ) − P(C2 C3 )
− P(C2 C4 ) − P(C3 C4 ) + P(C1 C2 C3 ) + P(C1 C2 C4 ) + P(C2 C3 C4 ) + P(C3 C4 C1 )
− P(C1 C2 C3 C4 )
Q=1−R

For equal component reliability

R = 2r 5 − 5r 4 + 2r 3 + 2r 2

if r = 0.99 then

R = 0.99979805

Exercise 4.8: Consider the following reliability network, each component has same reliability
r = 0.99. Obtain system reliability (Figure 4.23).

a b c

d e f

Figure 4.23 Complex reliability network.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
72 4 Methodologies for Reliability Evaluation of Network

a c

d e f

Figure 4.24 Decomposed network with X is working.

a c

d e f

Figure 4.25 Decomposed network with X is failed.

Solution:
Careful examination of the network suggest that the problem can be easily solved by decomposition
method selecting component “b” as key stone (Figures 4.24 and 4.25)
P(X) = P(b) = r
P(X) = P(b) = (1 − r)
When X is working then P(S∕X)
P(S∕X) = (2r − r 2 )2
When X is not working then
P(S∕X)
P(S∕X) = r 3
R = (2r − r 2 )2 r + r 3 (1 − r)
R = r 5 − 5r 4 + 5r 3
if r = 0.99
R = 0.999505
Note: Reader is advised to solve the same problem by event space, tie-set and cut-set method.

4.10 Conclusions
This chapter has described four basic methodologies for reliability evaluation of systems. Every
methodology has been discussed with examples. It is stressed here that each method is applicable
in different situations and limitations in some cases. Event space method is not suitable when a
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 73

large number of components are present in the system. In advance application, state pruning may
adopted i.e. truncation of state space. State pruning will require lesser number of states to be enu-
merated. Decomposition is justified when number of components in the systems are small. More-
over further limitation of the method is that it cannot be programmed on digital computer. Tie-set
and cut-set methodologies are effective and both are programmable on digital computer. Usually
in practice cut-set method is quite efficient in practice due to the fact approximations are justified.
Additional advantage of cut-set method is that, while identifying cut-sets various failure modes
are also obtained. Hence cut-set method is a powerful and efficient technique especially for power
network problems.

References

Allan, R.N. (2013). Reliability Evaluation of Power Systems. Springer Science & Business Media.
Arya, L.D. and Koshti, A. (2008). Distributed generation capacity reliability evaluation using safety
index. Journal of the Institution of Engineers (India) 89: 3–7.
Arya, L.D., Koshti, A., and Choube, S.C. (2013). Frequency-duration analysis of composite distribution
system using a non-sequential Monte Carlo simulation. International Journal of Electrical Power &
Energy Systems 46: 17–25.
Billinton, R. and Allan, R.N. (1992). Reliability Evaluation of Engineering Systems, vol. 792. Springer.
Breneman, J.E., Sahay, C., and Lewis, E.E. (2022). Introduction to Reliability Engineering. John Wiley &
Sons.
Ebeling, C.E. (2019). An Introduction to Reliability and Maintainability Engineering. Waveland Press.
Endrenyi, J. (1978). Reliability Modeling in Electric Power Systems. Wiley-Blackwell.
Halve, S.S., Koshti, A., and Arya, R. (2023). A sampling method based on system state transition for
distribution system adequacy assessment using distributed generation. Journal of Operation and
Automation in Power Engineering 11 (4): 249–257.
Koshti, A.S., Verma, A., Arya, R. et al. (2023). Tolerable random interruption duration based reliability
estimation of stand alone hybrid renewable energy system by network reduction and sequential
Monte Carlo simulation. IET Generation, Transmission & Distribution https://doi.org/10.1049/gtd2
.12910.
Saket, R.K., Kothari, D.P., and Nagrath, I.J. (2022). Modern Power System Analysis. Tata McGraw-Hill
Publishing Company.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
75

Probabilistic Approach for Standby and Load-Sharing System

Reliability Evaluation
Rajesh Arya 1 , R. K. Saket 2 , Atul Koshti 3 , Saad Mekhilef 4 , and Pradeep Purey 5
1
Department of Information Technology, Acropolis Institute of Technology and Research, Indore, Madhya Pradesh, India
2
Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
3 Department of Electrical Engineering, GES’S R.H. Sapat College of Engineering, Management Studies & Research, Nashik,

Maharashtra, India
4
Department of Electrical Engineering, Swinburne University of Technology, Melbourne, Australia
5
Department of Electronics Engineering, Maharaja Ranjit Singh College of Professional Sciences, Indore, Madhya Pradesh, India

5.1 Introduction
Reliability evaluation of a standby system needs different considerations and requires a state tran-
sition diagram (Hegazy et al. 2003; Arya & Koshti 2008; Costa & Matos 2009; Xu & Singh 2012;
Arya & Koshti 2012). Accounting for preventive maintenance, the hazard rate of a component may
be treated as constant as an average value during a period. Still nonrepairable systems are being
considered. A standby system (Billinton & Li 1993; Wang et al. 2013; Ahmadi-Khatir et al. 2009;
Arya et al. 2013; Halve et al. 2023) is one where one of the components is in operation, and upon
failure of this component, one of the idle components is switched in for service. It is obvious as the
number of idle (standby) components increases, the reliability of the standby system increases. This
may not always be true. Under ideal conditions certainly. reliability will increase, but the magni-
tude of the increase in reliability may reduce owing to imperfect switch and switching (Koshti et al.
2023; Khezri & Mahmoudi 2020; Roy et al. 2022; Ranjbar et al. 2022; Yahaya et al. 2020; Hamzeh
& Vahidi 2020; Hassan et al. 2022). These aspects need consideration. Further, the standby compo-
nent may fail under idle mode. The “switch” which puts the idle component in operation may fail
on demand.

5.2 Reliability Evaluation Under Ideal Condition

This section discusses the reliability evaluation of a standby system (Ebeling 2000; Lewis 1987;
Endrenyi 1978) under the assumptions
(a) Failure rate of switch is zero
(b) Probability of failing the switch on demand is zero
(c) Failure rate of idle component is zero while in standby mode
A standby system with “r” standby components is shown in Figure 5.1

Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
76 5 Probabilistic Approach for Standby and Load-Sharing System Reliability Evaluation

Figure 5.1 Standby system representation.

λ1dt λ2dt λ3dt λndt

0 1 2 3 n

1–λ1dt 1–λ2dt 1–λ3dt

Figure 5.2 Standby system state transition diagram.

Assume that the failure rates of each component as 𝜆1 , 𝜆2 , 𝜆3 …𝜆r+1 as seen in the following
system’s state transition diagram (Figure 5.2).
State-0 Represents no failure i.e. main component is in operation
State-1 Main failed and “first” standby is in operation
State-2 Second standby is in operation [main and first standby has failed]

State-n nth standby is in operation [main and (n−1) standby component have failed]
Sate 0 equation is written as follows-

P0 (t + dt) = P0 (t) (1 − 𝜆1 dt)

(Probability of system being in “0th” state at t + dt) = (Probability of system in “0th ” state at “t”)
(probability of the components do not fail in interval t to (t + dt))

P0 (t) = −𝜆1 P0 (t) (5.1)

Similarly, equation for state 1 is written as

p1 (t + dt) = p0 (t) (𝜆1 dt) + p1 (t) (1 − 𝜆1 dt)

p1 (t + dt) = p0 (t) (𝜆1 dt) + p1 (t) (1 − 𝜆2 dt)
p1 (t − dt) − p1 (t)
= 𝜆1 p 0 − 𝜆 2 p 1
dt
p1 = 𝜆1 ṗ 0 − 𝜆2 p1 (5.2)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
5.2 Reliability Evaluation Under Ideal Condition 77

Equation for state 2

p2 (t + dt) = p1 (t)(𝜆2 dt) + p2 (t)(1 − 𝜆3 dt)

p2 (t − dt) − p2 (t)
= 𝜆1 p 0 − 𝜆 2 p 1
dt
ṗ 2 = 𝜆2 p1 − 𝜆3 p2 (5.3)

In general, the equation for nth state is written as follows

ṗ n = 𝜆n pn−1 − 𝜆n+1 pn (5.4)

The above equations are written in matrix form (Endrenyi 1978) as follows

ṗ T (t) = pT A
ṗ T = [ṗ 0 ṗ 1 ṗ 2 . …]
pT = [p0 p1 p2 . …] (5.5)

Matrix “A” follows

⎡−𝜆1 𝜆1 0 0 0 .. ⎤
⎢ ⎥
0 −𝜆2 𝜆2 0 0 .. ⎥
A=⎢
⎢ 0 0 −𝜆3 𝜆3 0 .. ⎥
⎢ ∶ ∶ ∶ ∶ ∶ ⋱⎥⎦
⎣
Matrix A may be formed as follows:
aii = − (Total rate of departure from ith state)
aij = Total rate of departure from ith state to jth state
In solving the above state equations, the initial condition vector is given as [1 0 0 . . . . 0]
In sequence, the above equation is solved, and if there are “r” standby, then the reliability of the
system is given below
∑
r
R(t) = pi (t), R(t) = 1 − R(t)
i=0

Example 5.1
Assume the system consists of one main component and one standby component having failure
rate as 𝜆1 and 𝜆2 respectively.
Solution:
The system’s reliability is described as

R(t) = p0 (t) + pi (t)

P0 = e−𝜆1 t
Ṗ 1 = 𝜆1 P0 − 𝜆2 P2
𝜆1
P1 (t) =
(s + 𝜆1 ) (s + 𝜆2 )
𝜆1 𝜆1
P1 (s) = −
(𝜆2 − 𝜆1 ) (s + 𝜆2 ) (𝜆2 − 𝜆1 ) (s + 𝜆2 )
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
78 5 Probabilistic Approach for Standby and Load-Sharing System Reliability Evaluation

Taking inverse Laplace transform gives

𝜆1 𝜆1
pi (t) = e−𝜆1 t − e−𝜆2 t
(𝜆2 − 𝜆1 ) (𝜆2 − 𝜆1 )
𝜆1 𝜆1
R(t) = e−𝜆1 t e−𝜆1 t − e−𝜆2 t
(𝜆2 − 𝜆1 ) (𝜆2 − 𝜆1 )
1 [ −𝜆 t ]
R(t) = 𝜆 e 1 − 𝜆1 e−𝜆2 t
(𝜆2 − 𝜆1 ) 2
Failure density function
dR 𝜆1 𝜆2 [ −𝜆 t ]
f (t) = − = e 1 − e−𝜆2 t
dt (𝜆2 − 𝜆1 )
Note: It is hypo-exponential failure law.
Further, if two failure rates are equal 𝜆1 = 𝜆2 = 𝜆, then R (t), f (t) expression reduces to

R(t) = (1 + 𝜆t)e−𝜆 t

f (t) = 𝜆2 t e−𝜆 t

It is an Erlang distribution function (Billinton & Allan 1983; Billinton & Allan 1996).
In general, for “r” standby system where each component has “𝜆” failure rate
(𝜆t)n −𝜆t
pn = e
n!
Given by a Poisson process, the reliability expression reduces to
∑
r
(𝜆t)n
R(t) = e−𝜆t
n=0
n!

5.3 Standby System Reliability Evaluation Under Nonideal

Condition

5.3.1 Switching is Imperfect (Endrenyi 1978)

Assume that the system consists of one main and one standby component, each having the same
failure rate. Further, assume that Ps is the probability of successful change-over. Figure 5.3 shows
the diagram of state transitions.

Ps λdt Figure 5.3 Standby system state transition

λdt diagram with perfect switching.

0 1 2

1–λdt qsλdt
1–λdt
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
5.3 Standby System Reliability Evaluation Under Nonideal Condition 79

State 0 Represents the main component of operation

State 1 Main component has failed change-over has been successful, and standby is operating.
State 2 an absorbing state and is either reached by failure of standby component or unsuccessful
change-over when the main component fails.
q s = 1 − ps

State space equation is written as

⎡−𝜆 ps 𝜆 q s 𝜆⎤
[ṗ 0 ṗ 1 ṗ 2 ] = [p0 p1 p2 ] ⎢0 −𝜆 𝜆 ⎥ lim
⎢ ⎥ 𝛿x→0
⎣0 0 0 ⎦
ṗ 0 = −𝜆p0 , p0 = e−𝜆t , ṗ 1 = ps 𝜆p0 − 𝜆p1
ps 𝜆 ps 𝜆
(s + 𝜆) p1 (s) = , p (s) =
(s + 𝜆) 1 (s + 𝜆)2
Taking inverse Laplace transform
p1 (t) = ps 𝜆 te−𝜆t
Reliability is described as
R(t) = p0 (t) + p1 (t), R(t) = (1 + ps 𝜆 t) e−𝜆t

1 + ps
MTTF =
𝜆
Hence MTTF will reduce depending on the value of ps and it provides a functional failure
density as
dR
f (t) = − , f (t) = ps 𝜆2 te−𝜆t + 𝜆e−𝜆t (1 − ps ), f (t) = 𝜆e−𝜆t [(1 − ps ) + ps 𝜆t]
dt

5.3.2 Switching is Imperfect and Switch has a Failure Rate 𝝀s (Imperfect Switch)
In addition to unsuccessful change-over, the switch while in connection has a failure rate is 𝜆s .
The diagram of the state is shown in Figure 5.4.
State 0 represents the main component is in operation, and standby is in standing mode
State 1 main component has failed, and standby is in operation (with successful) change-over
State 2 an absorbing state which is reached by
(a) by unsuccessful change-over
(b) by failure of switch while main component is in operation

Figure 5.4 Standby system state transition Psλ (λ + λs)

diagram with imperfect switching.

0 1 2

(qsλ + λs)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
80 5 Probabilistic Approach for Standby and Load-Sharing System Reliability Evaluation

(c) failure of switch while standby is in operation

(d) failure of standby component while in operation
State equation in matrix form is written as
⎡−(𝜆 + 𝜆s ) ps 𝜆 q s 𝜆 + 𝜆s ⎤
[ṗ 0 ṗ 1 ṗ 2 ] = [p0 p1 p2 ] ⎢0 −(𝜆 + 𝜆s ) (𝜆 + 𝜆s ) ⎥
⎢ ⎥
⎣0 0 0 ⎦
ṗ 0 = −(𝜆 + 𝜆s )p0 , p0 = e−(𝜆+𝜆s )t
ps 𝜆
ṗ 1 = ps 𝜆p0 − (𝜆 + 𝜆s )p1 (s + 𝜆 + 𝜆s )p1 =
(s + 𝜆 + 𝜆s )
ps 𝜆
p1 (s) = , p (t) = ps 𝜆te−(𝜆+𝜆s )t
(s + 𝜆 + 𝜆s )2 1
Reliability is given as
R (t) = p0 (t) + p1 (t), R(t) = (1 + ps 𝜆t)e−𝜆t e−𝜆s t
The result is expected. The previous result multiplied by the reliability of switch e−𝜆s t this is equiv-
alent to an imperfect switch in series.

5.3.3 Consider the Standby Components Might Fail While they are in Idle Mode
in Addition
𝜆i is failure rate of standby components in idle mode. State transition diagram is shown below
(Figure 5.5).
State space equations are written as
⎡−(𝜆 + 𝜆s + 𝜆i ) ps 𝜆 𝜆i q s 𝜆 + 𝜆s ⎤
⎢ ⎥
0 −(𝜆 + 𝜆s ) 0 −(𝜆 + 𝜆s )⎥
[ṗ 0 ṗ 1 ṗ 2 ṗ 3 ] = [p0 p1 p2 p3 ] ⎢
⎢0 0 −𝜆s 𝜆s ⎥
⎢0 0 0 0 ⎥
⎣ ⎦
The following states are described:
State 0 Represents the main component in operation, and standby is in idle mode
State 1 Main component has failed, successful change-over has taken place, and standby
component is in operation

Psλ λ + λs Figure 5.5 Standby system state transition

diagram with the assumption of component
failure during idle mode.

0 1 3

λs
qs λ + λs

2
λi
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
5.4 Reliability Evaluation of Load-Sharing System (Endrenyi 1978) 81

State 2 Main component is in operation, but the standby component has failed in idle mode
State 3 Absorbing state, which is reached via either unsuccessful change over, switch failure or by
standby component failure.
Reliability is given by
R(t) = P + P1 + P2
ṗ 0 = −(𝜆 + 𝜆s + 𝜆i )p0
p0 = e−(𝜆+𝜆s +𝜆i )t
ṗ 1 = ps 𝜆 p0 − (𝜆 + 𝜆s )p1
ps 𝜆
(s + 𝜆 + 𝜆s )p1 =
(s + 𝜆 + 𝜆s + 𝜆i )
ps 𝜆
p1 (s) =
(s + 𝜆 + 𝜆s )(s + 𝜆 + 𝜆s + 𝜆i )
p𝜆[ ]
p1 (t) = s e−(𝜆+𝜆s )t − e−(𝜆+𝜆s +𝜆i )t
𝜆i
ṗ 2 = 𝜆i p0 − 𝜆s p2
𝜆i
p2 (s) =
(s + 𝜆s ) (s + 𝜆 + 𝜆s + 𝜆i )
Therefore
𝜆i 𝜆i
P2 (s) = −
(𝜆 + 𝜆i ) (s + 𝜆s ) (𝜆 + 𝜆i ) (s + 𝜆 + 𝜆s + 𝜆i )
𝜆i [ −𝜆 t ]
p2 (t) = e s − e−(𝜆+𝜆s +𝜆i )t
(𝜆 + 𝜆i )
Reliability is given as
ps 𝜆 [ −(𝜆+𝜆 )t ]
p0 = e−(𝜆+𝜆s +𝜆i )t , p1 (t) = e s − e−(𝜆+𝜆s +𝜆i )t
𝜆i
𝜆i [ −𝜆 t ]
p2 (t) = e s − e−(𝜆+𝜆s +𝜆i )t
(𝜆 + 𝜆i )
Therefore reliability is given as
ps 𝜆 [ −(𝜆+𝜆 ) t ] 𝜆i [ −𝜆 t ]
R(t) = e−(𝜆+𝜆s +𝜆i ) t + e s − e−(𝜆+𝜆s +𝜆i ) t + e s − e−(𝜆+𝜆s +𝜆i ) t
𝜆i (𝜆 + 𝜆i )

5.4 Reliability Evaluation of Load-Sharing System (Endrenyi 1978)

Assume two systems are operating in active parallel and sharing the total load/demand. If one of
the system fails, the other can meet the demand but operate under stressed conditions, and hence,
its failure rate will increase. A load-sharing system’s reliability is obtained using state transition
diagram. In an earlier analysis of parallel system the two components were operating in parallel
and sharing the demand, but when one of them fails, the other meets the demand, but the failure
rate remains the same. The diagram of a load-sharing system’s state transition is given as follows.
If each failure rate is 𝜆1 , 𝜆2 , then the component while both are in operation. 𝜆+1 and 𝜆+2 are fail-
ure rate of these components while only one of them is working. The four-state model of the
load-sharing system is given in Figure 5.6.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
82 5 Probabilistic Approach for Standby and Load-Sharing System Reliability Evaluation

Figure 5.6 Four-state model of the load-sharing system.

λ1 State 1
State 0
-
1, 2
1, 2

λ2 λ2 +

State 2 State 3
- - - -
1, 2 λ1 + 1, 2

State 1 Both elements are operational.

State 2 Component 2 is working; however, component 1 failed.
State 3 Component 1 is functional, whereas component 2 is not.
State 4 both components 1 and 2 have failed.

State space equation of load-sharing system is given as follows:

⎡−(𝜆1 + 𝜆2 ) 𝜆1 𝜆2 0⎤
⎢ ⎥
0 −𝜆+2 0 𝜆+2 ⎥
[ṗ 0 ṗ 1 ṗ 2 ṗ 3 ] = [p0 p1 p2 p3 ] ⎢
⎢ 0 0 −𝜆+1 𝜆1 ⎥
+
⎢ 0 0 0 0 ⎥⎦
⎣
ṗ 0 = −(𝜆1 + 𝜆2 )p0 , p0 = e−(𝜆1 +𝜆2 )t , ṗ 1 = 𝜆1 p0 − 𝜆+2 p1
( ) 𝜆1 𝜆1
s + 𝜆+2 p1 = , p1 (s) = ( )
(s + 𝜆1 + 𝜆2 ) (s + 𝜆1 + 𝜆2 ) s + 𝜆+2
( ) ( )
𝜆1 ∕ 𝜆+2 − 𝜆1 − 𝜆2 𝜆1 ∕ 𝜆+2 − 𝜆1 − 𝜆2
p1 (s) = − ( )
(s + 𝜆1 + 𝜆2 ) s + 𝜆+2
𝜆1 [ + ]
p1 (t) = ( ) e−𝜆2 t
− e−(𝜆1 +𝜆2 )t
𝜆1 + 𝜆2 − 𝜆+2
ṗ 2 = 𝜆2 p0 − 𝜆+1 p2
λ2
p2 (s) = ( )
(s + 𝜆1 + 𝜆2 ) s + 𝜆+1
𝜆1 [ + ]
p2 (t) = ( ) e−𝜆1 t
− e−(𝜆1 +𝜆2 )t
𝜆1 + 𝜆2 − 𝜆+1
Reliability is expressed as follows:

R(t) = p0 (t) + p1 (t) + p2 (t)

𝜆1 [ + ]
R(t) = e−(𝜆1 +𝜆2 )t + ( ) e−𝜆2 t
− e−(𝜆1 +𝜆2 )t
𝜆1 + 𝜆2 − 𝜆+2
𝜆1 [ + ]
+( ) e−𝜆1 t
− e−(𝜆1 +𝜆2 )t
𝜆1 + 𝜆2 − 𝜆+2
If,

𝜆+1 = 𝜆1 , 𝜆+2 = 𝜆2
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
5.5 Illustrative Examples 83

Then, the above expression reduces to the expression of parallel system i.e.
R (t) = e−𝜆1 t + e−𝜆2 t − e−(𝜆1 +𝜆2 )t
Further, if the component are identically i.e.
𝜆1 = 𝜆2 = λ, 𝜆+1 = 𝜆+2 = 𝜆+
Then, the expression for the reliability of load sharing is given to the system by
2𝜆
R(t) = e−2𝜆t + [e−λ+t − e−2λt ]
(2𝜆 − 𝜆+ )
1 +
R(t) = [2𝜆e−𝜆 t − 𝜆+ e−2𝜆t ]
(2𝜆 − 𝜆 )
+

If 𝜆+ = 2𝜆, then the following relation is obtained

R(t) = (1 + 2 𝜆 t) e−2𝜆t

5.5 Illustrative Examples

This section describes various numerical problems to understand the mathematical interpretation.
The following examples are illustrated for the reliability assessment of industrial systems.

Example 5.2
Obtain the reliability of a standby system where main and standby components have failure rates
of 0.2 and 0.3 per year, respectively. Estimate the reliability value for 1 year.
Solution:
𝜆M = 0.2, 𝜆S = 0.3

1 [ −𝜆 t ]
R(t) = 𝜆S e M − 𝜆M e−𝜆S t
(𝜆S − 𝜆M )
R(t) = 10 × [0.3e−0.2t − 0.2e−0.3t ]
For one year
R(1) = 10 × [0.3e−0.2 − 0.2e−0.3 ]

R(1) = 0.97519

Example 5.3
A standby system has a failure rate of 0.1/year of main as well as standby component. Obtain relia-
bility for one year if failure rates of the transfer switch are 0.05/year and the probability of successful
change-over is 0.95.
Solution:
𝜆M = 𝜆S = 𝜆 = 0.1
𝜆sw = 0.05
ps = 0.95
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
84 5 Probabilistic Approach for Standby and Load-Sharing System Reliability Evaluation

Reliability expression for the system is written as

R(t) = (1 + ps 𝜆t)e−(𝜆+𝜆sw )t

R (t) = (1 + 0.5 × 0.1 × t) e−0.15t

For (t = 1 year)
R (t) = (1.095) e−0.15 , R(t) = 0.9424

Example 5.4
Obtain a nonredundant system’s reliability considering 50 components, and each component has
a failure rate of 0.002/year and also has one similar component in spare.
Solution:
Failure rate of the system is given as-
𝜆SYS = 50 × 0.02, 𝜆SYS = 1.0∕year
If one of the components fails, then it is replaced by the spare component, and again, the failure
rate will be 1.0. Hence, the system can be represented as standby system where “both” component
has a failure rate of 1.0/year. Then, reliability under ideal conditions is given as
R(t) = (1 + 𝜆sys t)e−𝜆sys t , R(t) = (1 + t)e−t
R(1) = 2e−1.0 , R(t) = 0.7357
Look, if you do not have spare components, then reliability is
R(t) = e−1.0 , R(t) = 0.3678

Example 5.5
Obtain reliability expression for the following system. Each component has a failure rate 𝜆 per year.
Assume ideal condition (Figure 5.7).
Solution:
This is a series system, where two standby sub-systems are in series. Each sub-system has reliability
R = (1 + 𝜆t)e−𝜆t
Hence, the overall system has reliability
RSYS = R2 , RSYS = [(1 + 𝜆t)e−𝜆t ]2

Example 5.6
Obtain reliability expression for the following system given in Figure 5.8. Each component has
failure rate 𝜆 per year.

Figure 5.7 Figure for Example 5.5.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
5.5 Illustrative Examples 85

Figure 5.8 Figure for Example 5.6.

Solution:
Hence two standby systems are connected in parallel. Each sub-system has reliability
R = (1 + 𝜆t)e−𝜆t
System reliability overall is described as (Figure 5.9)
RSYS = 2 R − R2
RSYS = 2 (1 + 𝜆t) e−𝜆t − [(1 + 𝜆t) e−𝜆t ]2
RSYS = 2 (1 + 𝜆t) e−𝜆t − (1 + 𝜆t)2 e−2𝜆t

Example 5.7
Two generators, each of 100 MW rating, are meeting demand of 150 MW, each having a failure rate
of 0.01/year. If one of the generators fails, the other can meet the demand but increase the failure
rate 𝜆+ = 0.04 per year. Compute the reliability for one year.
Solution:
Reliability expression is written as
1 +
R(t) = [𝜆+ e−2λt − 2𝜆 e−𝜆 t ]
(𝜆+
− 2𝜆)
𝜆+ = 0.04, 𝜆 = 0.01
100
R(t) = [0.04e−0.02t − 0.02e−0.04t ]
2
R(t) = 50[0.04e−0.02 − 0.02e−0.04 ]
R(t) = 50[0.03920 − 0.01921]
R(t) = 0.9995

Example 5.8
A load is supplied for three generators in parallel, having failure rate 𝜆 per year. If one of the gen-
erators fails, the remaining two generators can meet the demand, but failure rate of each becomes
double. If the next generator fails only one generator can meet the demand with failure rate six
times. Obtain reliability expression (Figure 5.9).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
86 5 Probabilistic Approach for Standby and Load-Sharing System Reliability Evaluation

3λ 4λ 6λ Figure 5.9 Figure for Example 5.8.

1 2 3 4

P1(t) P2(t) P3(t) P4(t)

Solution:
⎡−3𝜆 3𝜆 0 0 ⎤
⎢ ⎥
0 −4𝜆 4𝜆 0 ⎥
R(t) = p1 + p2 + p3 , [ṗ 1 ṗ 2 ṗ 3 ] = [p1 p2 p3 ] ⎢
⎢0 0 −6𝜆 6𝜆⎥
⎢0 0 0 0 ⎥⎦
⎣
ṗ 1 = 3𝜆p1 , sP1 (s) − 1 = 3𝜆P1 (s)
1
P1 (s) = , p (t) = e−3𝜆t , ṗ 2 = 3𝜆p1 − 4𝜆p2
s + 3𝜆 1
sP2 (s) = 3𝜆P1 (s) − 4𝜆P2 (s), (s + 4𝜆)P2 (s) = 3𝜆P1 (s)
3𝜆
P2 (s) = , p (t) = 3e−3𝜆t − 3e−4𝜆t
(s + 4𝜆) (s + 3𝜆) 2
4𝜆 × 3𝜆
ṗ 3 = 4𝜆p2 − 6𝜆p3 , sP3 (s) = 4𝜆P2 (s) − 6𝜆P3 (s), (s + 6𝜆)P3 (s) =
(s + 4𝜆) (s + 3𝜆)
12𝜆2 4 6 2
P3 (s) = , P (s) = − +
(s + 3𝜆) (s + 4𝜆) (s + 6𝜆) 3 (s + 3𝜆) (s + 4𝜆) (s + 6𝜆)
p3 (t) = 4e−3𝜆t − 6e−4𝜆t + 2e−6𝜆t
R(t) = p1 (t) + p2 (t) + p3 (t)
R(t) = e−3𝜆t + 3 e−3𝜆t − 3 e−4𝜆t + 4e−3𝜆t − 6 e−4𝜆t + 2 e−6𝜆t
R(t) = 8e−3𝜆t − 9e−4𝜆t + 2e−6𝜆t

Example 5.9
A generator meets a load of 200 MW via two parallel transmission lines with 150 MW capaci-
ties each. Each transmission line has reliability of 0.99 for 1 year. The generator failure rate is
0.1 per year.

(a) Obtain reliability of the system.

(b) Recalculate the reliability of another generator that is operating as a hot reserve and may take
a load in the event of failure of the previous generator. The failure rate of this generator is also
0.1 per year.
Solution:

(a) The logic diagram is shown as (Figure 5.10)

Figure 5.10 Figure for Example 5.9 (a).

G T1 T2
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
5.5 Illustrative Examples 87

Reliability of the system for one year is given as

RG = e−0.1t = e−0.1 = 0.9405, RT1 = RT2 = 0.99
R = 0.9405 × 0.99 × 0.99, R = 0.9217
(b) For this case reliability diagram is shown (Figure 5.11).

Figure 5.11 Figure for Example 5.9 (b).

G2

G1 T1 T2

Reliability of generator system as standby

RG = (1 + 𝜆t)e−𝜆t , RG = (1.1)e−0.1 , RG = 0.9953
System reliability is
R′ = RG RT1 RT2 , R′ = 0.9953 × 0.99 × 0.99, R′ = 0.97549
Note: R′ > R

Example 5.10
A system consists of two components. The main component is in the operation and another com-
ponent is in standby (spare) mode. Each component has failure, repair, and re-installation rate
as 𝜆, 𝜇, and 𝛼, respectively. Draw the state transition diagram and write down state probabilities
(Figure 5.12).

Figure 5.12 Figure for Example 5.10.

α α
4 1
w
λ λ
F
5 2 3
2μ μ

Solution:
Step 1. Main component up standby available
Step 2. Main component down standby available but not installed
Step 3. Main component down main as well as spare component is available but not reinstalled
Step 4. Standby is in operation spare component is down
Step 5. Both main and standby down
States 1 and 4 are up state
States 2, 3, and 5 are down state.
State probabilities in the long run are obtained by solving
∑ 2𝛼𝜇 2 (𝜆 + 𝜇) + 𝛼 2 𝜇 2
pT A = 0, pi = 1, p1 =
Δ
2𝜆𝛼𝜇 (𝜆 + 𝜇) 2𝜆𝜇 2 (𝜆 + 𝜇) 2𝜆𝛼 2 𝜇 𝜆2 𝛼 2
p2 = , p3 = , p4 = , p5 =
Δ Δ Δ Δ
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
88 5 Probabilistic Approach for Standby and Load-Sharing System Reliability Evaluation

Δ = 2𝜇[(𝛼 + 𝜇) (𝜆 + 𝛼) (𝜆 + 𝜇)] + 𝜆2 𝛼 2
System availability in long run is given as
ASYS = p1 + p4
And system unavailability is given as
ASYS = p2 + p3 + p5
Failure frequency
f = p1 𝜆 + p4 𝜆, f = (p1 + p4 )𝜆
Putting the p1 and p4
2𝛼𝜆𝜇[(𝜆 + 𝜇)(𝛼 + 𝜇)]
f =
Δ
Mean up time is
p + p4 1
MUT = 1 =
f 𝜆
Mean down time
p + p3 + p 5
MDT = 2
f
1 𝛼𝜆
MDT = +
𝛼 2𝜇(𝜆 + 𝜇)(𝜇 + 𝛼)
Take 𝜆 = 0.4, 𝜇 = 9.6 and 𝛼 = 100 per year.
System availability
ASYS = p1 + p4
p1 = 0.88084, p4 = 0.0378, ASYS = 0.9188
ASYS = 1 − ASYS = 1 − 0.9188
ASYS = 0.08132
f = (p1 + p4 )𝜆 = 0.9186 × 0.4 = 0.36744
A 0.9186
MUT = SYS =
f 0.36744
MUT = 2.5 years, MDT = 0.22 years

5.6 Conclusion
A methodology has been presented in this chapter for reliability evaluation of standby and
load-sharing system. The following cases have been illustrated with expressions for reliability
evaluation of a standby system under ideal conditions with (a) assumptions that the failure rate of
the switch is zero, (b) probability of failing the switch on demand is zero, and (c) failure rate of idle
component is zero in standby mode. Further, standby system reliability under nonideal conditions
is evaluated with conditions such as (a) imperfect switch, (b) imperfect switching, and (c) standby
components failure during idle mode. Moreover, at the end of the chapter, illustrative exercises
are presented to help readers understand the concept. This methodology will give an insight into
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 89

using the philosophy of reliability evaluation for standby and load-sharing systems, which will be
helpful for power system planning engineers to assess the reliability of such systems.

References

Ahmadi-Khatir, A., Fotuhi-Firuzabad, M., and Goel, L. (2009). Customer choice of reliability in
spinning reserve procurement and cost allocation using well-being analysis. Electric Power Systems
Research 79 (10): 1431–1440.
Arya, L.D. and Koshti, A. (2008). Distributed generation capacity reliability evaluation using safety
index. Journal of Institution of Engineers (India) 89: 3–7.
Arya, L.D. and Koshti, A. (2012). Probabilistic simulation approach for distributed generation (DG)
capacity evaluation using artificial neural network representation of load duration curve. Journal of
Institution of Engineers (India) 93 (1): 57–62.
Arya, L.D., Koshti, A., and Choube, S.C. (2013). Frequency-duration analysis of composite distribution
system using a non-sequential Monte Carlo simulation. International Journal of Electrical Power and
Energy Systems 46: 17–25.
Billinton, R. and Allan, R.N. (1983). Reliability Evaluation of Engineering Systems, 2e. Springer.
Billinton, R. and Allan, R.N. (1996). Reliability Evaluation of Power Systems, Springer International
Edition, 2e.
Billinton, R. and Li, W. (1993). A system state transition sampling method for composite system
reliability evaluation. IEEE Transactions on Power Systems 8 (3): 761–770.
Costa, P.M. and Matos, M.A. (2009). Assessing the contribution of microgrids to the reliability of
distribution networks. Electric Power Systems Research 79 (2): 382–389.
C.E. Ebeling, ‘An Introduction to Reliability and Maintainability Engineering’, McGraw Hill Education
(India) edition, 2000.
Endrenyi, J. (1978). Reliability Modeling in Electric Power Systems. Wiley-Blackwell.
Halve, S.S., Koshti, A., and Arya, R. (2023). A sampling method based on system state transition for
distribution system adequacy assessment using distributed generation. Journal of Operation and
Automation in Power Engineering 11 (4): 249–257.
Hamzeh, M. and Vahidi, B. (2020). Reliability evaluation of distribution transformers considering the
negative and positive effects of rooftop photovoltaics. IET Generation, Transmission & Distribution 14
(15): 3063–3069.
Hassan, R., Das, B.K., and Hasan, M. (2022). Integrated off-grid hybrid renewable energy system
optimization based on economic, environmental, and social indicators for sustainable development.
Energy 250: 123823.
Hegazy, Y.G., Salama, M.M.A., and Chikhani, A.Y. (2003). Adequacy assessment of distributed
generation systems using Monte Carlo simulation. IEEE Transactions on Power Systems 18 (1): 48–52.
Khezri, R. and Mahmoudi, A. (2020). Review on the state-of-the-art multi-objective optimisation of
hybrid standalone/grid-connected energy systems. IET Generation, Transmission & Distribution 14
(20): 4285–4300.
Koshti, A.S., Verma, A., Arya, R. et al. (2023). Tolerable random interruption duration based reliability
estimation of stand alone hybrid renewable energy system (SAHRES) by network reduction and
sequential Monte Carlo simulation. IET Generation, Transmission & Distribution 1–12. (Early view)
https://doi.org/10.1049/gtd2.12910.
Lewis, E.E. (1987). Introduction to Reliability Engineering. John Wiley & Sons.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
90 5 Probabilistic Approach for Standby and Load-Sharing System Reliability Evaluation

Ranjbar, H., Saber, H., and Sharifzadeh, M. (2022). Bi-level planning of distributed energy resources
into existing transmission grids: pathway to sustainable energy systems. IET Generation,
Transmission & Distribution 16 (24): 4963–4979.
Roy, P., He, J.B., Zhao, T., and Singh, Y.V. (2022). Recent advances of wind-solar hybrid renewable
energy systems for power generation: a review. IEEE Open Journal of the Industrial Electronics
Society 3: 81–104.
Wang, Y., Guo, C., and Wu, Q.H. (2013). A Cross-entropy-based three-stage sequential importance
sampling for composite power system short-term reliability evaluation. IEEE Transactions on Power
Systems 28 (4): 4254–4263.
Xu, Y. and Singh, C. (2012). Adequacy and economy analysis of distribution systems integrated with
electric energy storage and renewable energy resources. IEEE Transactions on Power Systems 27 (4):
2332–2341.
Yahaya, A., AlMuhaini, M., and Heydt, G.T. (2020). Optimal design of hybrid DG systems for microgrid
reliability enhancement. IET GTD 14 (5): 816–823.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
91

Reliability-Based Systems Design

Section 2
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
93

Physical Reliability Methods and Design for System Reliability

Smriti Singh 1 , Jyoti Maurya 1 , Eram Taslima 1 , Bharat B. Sagar 2 , and R. K. Saket 1
1
Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
2 Department of Computer Science and Engineering, School of Engineering, Harcourt Butler Technical University, Kanpur,
Uttar Pradesh, India

6.1 Introduction
The impetus behind conducting reliability and maintainability analysis stems from various factors,
including the heightened intricacy and advancement of systems, a growing societal emphasis on
product quality, the existence of rules and regulations governing product liability, and the neces-
sity to meet performance criteria outlined in government contracts to attain the desired level of
reliability (Ebeling 2019). Reliability and maintainability constitute a significant component of the
engineering design process (Wang et al. 2004) and assume a crucial role in life cycle assessment,
cost–benefit evaluation, studies of operational capability, allocation of resources for repair and facil-
ities, calculation of inventory, need for spare part, replacement decision-making, as well as the
formulation of proactive maintenance initiatives.
The prioritized goals of reliability engineering, as per O’Connor and Kleyner’s (2012) study are
as follows:
● Employing engineering expertise and specialized methodologies to avert or minimize the occur-
rence of failures.
● Recognizing and rectifying the root causes of failures that transpire despite prevention endeavors.
● Developing strategies to manage failures that arise when their underlying causes remain unad-
dressed.
● Utilizing techniques to predict the prospective reliability of novel designs and to scrutinize
reliability-related data.
In order to achieve these goals, certain reliability methods are used (Saket 2011). Physical reliabil-
ity methods refer to a set of techniques and approaches used to evaluate and ensure the reliability of
physical systems, components, or structures. These methods are employed to assess the capacity of
a system to carry out its intended operations without encountering errors and to quantify the prob-
ability or likelihood of failure under various operating conditions and stresses (Kothari et al. 2022).
Physical reliability methods typically involve the analysis of modes of failure, the identification of
potential failure mechanisms, and the establishment of strategies to mitigate or eliminate those
failures. These methods can be applied to a wide range of systems, including mechanical, electri-
cal, electronic, and structural systems. In Sections 6.2.1, 6.2.2, etc provides a detailed explanation
Sometimes this is also known as static reliability.
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
94 6 Physical Reliability Methods and Design for System Reliability

The chapter is structured in the following manner: Section 6.1 provides a concise Introduction,
Section 6.2 discusses the Reliability methods, Section 6.3 delves into the Design analysis and pro-
cess, and Section 6.4 serves as the chapter’s conclusion.

6.2 Reliability Methods

6.2.1 Block Diagram Analysis

Any reliability design starts with the building up of a model. Block Diagrams and Fault Trees are
used to provide graphical representation which helps in evaluation of the relationship between the
various components of the system.
Logical connections between system components and the concept of system failure logic can be
depicted using a reliability block diagram (RBD). If the connection is complex, then the construc-
tion of the RBD is difficult. After the construction of the block diagram, the process of analysis
involves simplifying the overall intricate system into a more basic one, which can then be exam-
ined through arrangements of series and parallel components. This method is also known as block
diagram decomposition. Some of the methods used to examine a complex system are described in
Sections 6.2.1.1, 6.2.1.2, etc.

6.2.1.1 Cut Sets and Tie Sets

Cut set and tie set methods are used to analyze complex reliability block diagrams. A cut set is
formed by drawing a line across the system’s blocks to emphasize the minimal number of malfunc-
tioning blocks that would lead to system failure. Tie sets (or path sets) are generated by drawing
lines through blocks that, if all of them operated correctly, would enable the system to operate.
The employment of cut set and tie set techniques is fit for computer applications. Their utilization
is well-suited for examining extensive systems featuring numerous configurations, like control sys-
tems in the aviation industry, electricity producing systems, or control and instrumentation systems
for large-scale industrial installations.
Figures 6.1 and 6.2 show how cut sets and tie sets are produced. It’s noticeable that there are cut
sets, which are three in number, and tie sets which are two in number.
The limitations on the reliability of the system, as determined through cut sets and tie sets, are
expressed as follows:
n
∑
N
∏j

Rs > 1 − (1 − Ri ) (6.1)
j i
T nj
∑ ∏
Rs < Ri (6.2)
j i

A B

Figure 6.1 Cut sets.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6.2 Reliability Methods 95

A B

Figure 6.2 Tie sets.

where
N = cut set quantity
T = tie set quantity
nj = quantity of blocks in the jth cut set or tie set

6.2.1.2 Common Mode Failures

A common mode failure is a shared source of failure that can lead to failed paths in redundant
systems. It is important to identify and evaluate a common mode failure since it increases the like-
lihood of malfunction. There are many cases when more than one unit fails in an industry due to
a single cause. The likelihood of failure in redundant systems, when considering only individual
failure paths, is notably elevated. Some of the common example of the sources of common mode
failure are:

● A sensor system which detects the common mode failure.

● An indicator system which notifies the path failure.
● A common power supply.
● An act of maintenance or operation that is frequently encountered on many different paths.
● Software that assists different paths.
● When one element fails, the remaining elements or redundant units in series are more likely to
fail.

Hence, the occurrence of common cause failures holds significant importance in reliability anal-
ysis and necessitates inclusion for precise reliability forecasting. Various methods have been devel-
oped that consider the common mode failure for reliability evaluation. An approach of this kind
operates under the premise that all common mode failures linked to the redundant system are
consolidated into a single hypothetical unit, which is then integrated in series with the redun-
dant system. Instances of such redundant systems encompass a parallel network, a k-out-of-n unit
system, and a bridge configuration.

6.2.1.3 Enabling Events

Enabling events are one which will cause a high level failed event. Quite similar to common failure
modes, it is difficult to predict an enabling event. Some of the common examples of enabling event
are:

● Incorrect control settings.

● Disabled warning systems, due to false warnings.
● Following wrong procedures by operating and maintenance personnel.
● Temporarily unavailable standby element.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
96 6 Physical Reliability Methods and Design for System Reliability

Dimly lit
chamber Top event

Combination event Combination event

No electricity Switch Both bulbs fuse
open

Basic event
AND gate OR gate

No Fuse Bulb1 Bulb2

power failure fuse fuse

Basic event Basic event Basic event Basic event

Figure 6.3 Fault tree.

6.2.2 Fault Tree Analysis

Fault tree analysis (FTA) is a systematic and graphical method used in reliability engineering, risk
assessment, and safety analysis to evaluate the possible system failure modes and spot out the
underlying causes that could lead to those failures. It is particularly useful for understanding and
mitigating complex systems’ reliability issues and safety concerns.
The primary purpose of FTA is to identify the chain of events or conditions that could lead to
an undesired event, often referred to as the “top event” or “undesired outcome.” It furnishes a
systematic method for examining the logical connections among diverse events and conditions
that contribute to the unfolding of the primary event. Figure 6.3 shows a FTA diagram. It roots out
the possible causes of failure through tree branching and focuses on prevention.
In 1962, H.A. Watson created FTA while working at Bell Laboratory, as part of a contract with
the U.S. Air Force Ballistic System Division. This contract aimed to assess the Minuteman I Inter-
national Ballistic Missile (ICBM) Launch Control System (Ericson and Ll 1999), (Vesely 2002).
Standard symbols are used to define the basic events corresponding to such logical connections
and for construction of a fault tree. Figure 6.4, shows the symbols used in the construction of
a fault tree. The AND and OR function gates are used to depict the symptoms of failure, down
to the real root cause. Finally, from all levels of failure, the top event probability is quantified
using FTA.
As shown in Figure 6.3, failure of a component is depicted using a circle, signifying a basic event.
The failure can happen due to one or more failure events (failure of sub-components). Components
can thus be arranged in series (OR Gate) or parallel (AND Gate) fashion. Components are linked in
series if the breakdown of either component results in the collapse of the whole system. Conversely,
if the failure of one component doesn’t lead to the failure of the entire system, then the blocks
are linked in a parallel arrangement. In cases where system blocks are arranged in a sequential
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6.2 Reliability Methods 97

Fault tree symbol Meaning Fault tree symbol Meaning

A basic fault tree Is used to include or
event which is exclude a part of
independent of Fault Tree, which
other event and may or may not be
does not require required.
BASIC EVENT SWITCH
further
development.

A basic event which

depends on lower Allows failure only
events developed as when all inputs fail.
BASIC EVENT separate fault tree.
AND GATE

A basic event which

depends on lower Allows failure in case
event, but not of any one of the
developed input fails.
BASIC EVENT OR GATE
downwards.

Defines a causal
An event which is relationship between
the result of one fault and
combination of basic another. The output
events through the event is produced
COMBINATION input logic gates. INHIBIT GATE
directly from the
EVENT input event if the
indicated condition
is satisfied.

A line from the apex of the triangle indicates a transfer in;

OUT A line from the side denotes a transfer out.

IN
TRANSFERRED
EVENT

Figure 6.4 Fault tree symbol.

configuration, and Ri be the reliability of component Ci , i = 1, 2, … n then the system reliability is

given by:
∏
n
Rsystem = R1 ∗ R2 ∗ · · · ∗ Rn = Ri (6.3)
i=1

Similarly, for system components connected in parallel, system reliability is as follows:

∏
n
Rsystem = 1 − (1 − Ri ) (6.4)
i=1

Different FTA is constructed for each defined top event. It is difficult to construct an FTA for a
large complex system. Drawing and evaluating them manually is an impractical task. In such cases,
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
98 6 Physical Reliability Methods and Design for System Reliability

computer programs are used to generate and evaluate FTAs. In this way, an effective, economic and
easy iterative analysis is provided by the assistance of computer programming. FTA is an essential
part of reliability analysis as it considers single as well as multiple failure event.

6.2.2.1 Steps of FTA

Following are the basic steps to be followed for a FTA:
1. Level 1 is defined by the top failure event.
2. Level 2 defines the input symptoms which are the contributing cause for the top event.
3. Levels 3, 4 or more are defined by subsequently breaking down the failures with additional gate
levels.
4. Finalize the diagram.

6.2.2.2 Applications of FTA

Given are the main applications of FTA:
● To breakdown failure into more branches.
● FTA is generally applied when a backward approach identifies the potential cause of failure.
● To define inter-relationship between the cause and failure.
Some of the recent applications of FTA for reliability evaluations are: Kang et al. (2019) where
FTA finds application in the reliability assessment of floating offshore wind turbines. In Badida
et al. (2019) risk assessment of oil and natural gas pipelines is done.
Fuzzy fault tree analysis (FFTA) (Kuzu et al. 2019; Singh et al. 2022) is an extension of traditional
FTA that incorporates the concepts of fuzzy logic and uncertainty into the analysis process. Fuzzy
logic enables the portrayal of uncertain or ambiguous data, a common occurrence in real-world
scenarios where exact probabilities or values may be difficult to determine. FFTA is particularly
useful when dealing with systems that involve subjective judgments, incomplete data, or qualitative
information.

6.2.3 State Space Analysis (Markov Analysis)

Every framework or constituent has the potential to exist in one of two conditions: active or inac-
tive. The active state denotes operational status, while the inactive state signifies a malfunction.
Employing state space analysis, it becomes possible to determine the probabilities linked to these
conditions. Among state space analysis techniques, Markov analysis is the most widely recognized
method, applicable under the following presumption:
● The transition process from one state to another should exhibit homogeneity, with a consistent
probability that remains unchanged.
● The future state of the system does not depend on the previous state except the immediate one
that precedes the future state.
This technique bears the name of a Russian mathematician, Andrei Andreyevich Markov
(1856–1922). Markov analysis is implemented for the analysis of complicated system such as
power generation and communication system. It is generally used for reliability analysis of systems
where block diagram analysis cannot be applied.
In recent times, this method has been used to model the volatile nature of renewable energy
systems (Theristis and Papazoglou 2013; Singh et al. 2023), so that it is possible to execute the
reliability analysis. In Peiravi et al. (2020), the authors have solved an optimization problem based
on Markov model using redundancy strategy. Markov model is used for the reliability analysis of
wind turbine in Li et al. (2019).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6.2 Reliability Methods 99

Denotes an event

Immediate transition which denotes

transfer of event with no delay time

Timed transition which denotes transfer of

event with a period of delay

Arch, drawn between event and transitions

as arrow

Token, depicted as a dot. Indicates system

state

Inhibitor arch, drawn between places and

transitions.

Figure 6.5 Petri nets basic symbols.

6.2.3.1 Advantages, Limitations, and Application of Markov Analysis

● The process of transitioning from one state to another must be homogeneous, and its probability
must remain constant.
● The future state of the system does not depend on the previous state except the immediate one
that precedes the future state.

6.2.4 Petri Nets

Petri nets is an extension of state space analysis technique. Carl Adam Petri introduced this tech-
nique in 1962. Petri nets describes the existing relation between condition and event related to
failure event using graphical and mathematical analysis. In late 1970s, stochastic Petri nets (SPN)
or time Petri nets came into the picture. SPN deals with the flaws of Markov chain analysis. It mod-
els the state of the component such that its state can be inferred from previous states, unlikely, in the
Markov chain process. In this way, Petri nets are often used as a pre-processor in modeling, which
is converted internally to Markov state space. The dependence of the Markov method on constant
rate of occurrence is a major drawback of the Markov method. To overcome this disadvantage, the
transition process is simulated using Monte Carlo method. In software design, original Petri nets
are employed, whereas in reliability engineering, timed Petri nets are utilized.
Following are the fundamental symbols in Petri nets:
Figure 6.6 lists the basic structures depicting the logical relations for Petri nets. Petri nets is an effi-
cient tool for system modeling, monitoring of failure, simulation and reliability analysis. Dynamic
behavior can also be observed form Petri nets.

6.2.4.1 Fault Tree and Petri Net Transformation

A fault tree can be correlated with Petri nets, as shown in Figure 6.7. This is known as conversion
from fault trees to Petri nets.
Figure 6.8 shows the Petri net conversion of a fault tree.
In recent literature, Petri nets find its application in reliability evaluation. The authors in Kabir
and Papadopoulos (2019) present a review of Bayesian Network and Petri nets in reliability assess-
ment, safety and risk analysis. Stochastic PetriNets are used to model the critical systems for safety
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
100 6 Physical Reliability Methods and Design for System Reliability

Petri nets Logic relation Boolean function

B
TRANSFER B = A, lf A then B
A

R
AND R = A.B, If A and B then R

A B

R
OR R = A+B, If A OR B then R

A B

B R
TRANSFER AND B = R = A, If A then B and R

B R

TRANSFER OR B+R = A, If A then B OR R

A

R
INHIBITION R = A.B', If A AND B' then R

A B

Figure 6.6 Logical relations with Petri Nets.

of Nuclear Power Plants to check for reliability qualification in Jyotish et al. (2023). In Kumar et al.
(2022), the authors have employed a Petri Nets-based method to forecast the reliability of software
in industrial contexts. In Kumar et al. (2020), performance analysis of milk processing systems is
done using PetriNets.

6.2.5 Failure Mode Effects and Criticality Analysis (FMECA)

Failure modes effects and criticality analysis (FMECA) also known as Failure modes and effects
analysis (FMEA) is a widely used reliability method in system design. FMECA examines the
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6.2 Reliability Methods 101

Fault tree Petri net Probability of

failure

1
1
AND-Model F1 = FP * FQ

P Q P Q

2 2
OR-Model F2 = 1-(1-FP) (1-FQ)

P Q P Q

Figure 6.7 Correlation between fault tree and Petri Net.

Figure 6.8 Petri net transformation of Fault Tree

of Figure 6.3. 0

1 E 2

A B C D

potential modes of failure within the system, classifies them by severity and also determines the
effect of these failures on the operation of the system as a whole.
FMEA was initially introduced for military purposes in the late 1940s by the US Armed Forces,
as noted by Dhillon (2004). Subsequently, its application extended to aerospace and rocket devel-
opment. Table 6.1 lists some of the prospective failure modes experienced by the industries where
the product is manufactured. The motive of FMEA is to reduce failures, beginning with the highest
priority failure. Few of the common causes that require special attention are depicted in Table 6.2.
Failure Modes are ways through which an error or defect can occur in the design or item that
affects the consumer. Effect Analysis refers to the assessment of the consequences of these failures.
Table 6.3 shows the possible effects of failure.
Prior to commencing the FMEA process, a worksheet is prepared that encompasses crucial sys-
tem specifications. This worksheet systematically outlines all components and their respective
functions. Figure 6.9 shows a FMEA Worksheet.
Variety of industries which find application of FMEA include semiconductor processing, soft-
ware, healthcare, and food services. FMEA combined with Advanced Product Quality Planning
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
102 6 Physical Reliability Methods and Design for System Reliability

Table 6.1 Failure modes.

Corrosion Broken Burrs Omitted Wrapped

Bent Crack Melted Sharp Edges Not fulfilled
Discolor Porous Shortness Discoloring Leaking
Defective Delay Bound Eccentricity Open Circuited
Out of Tolerance Damage Tightness Misassemble Brittle
Misalignment Blistered Roughness Grounding Deformity

Table 6.2 Causes of failure.

Inadequate venting Damaged part Missed operation

Handling damage Incorrect tooling Misalignment
Improper surface preparation Improper torque Improper tool setup
Out of tolerance Inadequate gating Clamping
Inadequate control system Inadequate control system Tool damage/blurs
Inadequate holding Material failure Overload capacity
Assembly error Incorrect speed/feeds Heat treatment shrinkage

Table 6.3 Effects of failure.

Brake failure Air/water leakage Excess interest

Oil carryover Wrong carryover Wrong material
Poor coordination Delay in delivery/payment Unable to assemble
Poor valve performance Customer dissatisfaction Oil carry over
Inadequate service life Abrupt use of material Contamination-valve system

S. Process Potential Potential SEVERITY Potential OCCURRENCE Current Detection RPN Action
No. Step/ Modes of Effects (1-10) Causes (1-10) Controls (1-10) Recommen
Input Failure of ded
Failure
1
2
3
4
5
6

Figure 6.9 A sample of FMEA worksheet.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6.2 Reliability Methods 103

(APQP) serves as a risk mitigation tool. To assess the risk, risk priority number (RPN) is used.
Following step are followed in determining RPN:

1. Rating of the effect of failure in the order of severity.

2. Determining the likelihood of failure occurrence.
3. Likelihood of pre-detection of failure.

RPN = Severity × Occurrence × Detection (6.5)

1. Severity: Severity determines all the failure modes. A malfunctioning mode within a single
component initiates a malfunctioning mode within another component. Failure effect results
from the failure mode. Every effect is allotted a severity number (S) which varies from 1 to 10.
1 represents no danger and 10 represents critical. With the help of these numbers the engineers
rank the failure modes and their impacts. If severity number is 9 or 10, the failure modes need to
be eliminated and the design has to be changed. Such a high severity number generally results
in an effect that is injurious to the consumer and may cause litigation.
2. Occurrence: This step takes into consideration the frequency with which a particular cause is
resulting in failure. It is essential to recognize and record all major factors contributing to failure
modes using technical terminology. The occurrence rank is given to the failure mode from 1 to
10. If the occurrence is high (greater than 4 or greater than 1 and the severity number is 9 or 10),
action must be taken.
3. Detection: Detection refers to the quantification of the probability that a malfunction can be
detected. This is done using testing and monitoring. Once the detection number (D) is known,
the ability to remove or detect the failure on time can be prioritized. D assesses the possibility
that the failure might go unnoticed. The higher the detection number, the higher the chance
that the detection of failure is escaped or the chance of detection is low.

RPN rating ranges from 1 to 10. Higher the risk or severity, higher will be the RPN rating. RPN can
thus be used to sort out the problems in the priority order; so that corrective actions can be made.
The Automotive Industry Action Group (AIAG) mandates the incorporation of FMEA within
the automotive APQP process. Additionally, AIAG provides a comprehensive manual outlining the
application of this method. Furthermore, the American Society for Quality (ASQ) offers support
for FMEA practices.
Criticality rating of component can also be given by the failure mode criticality number:

Cm = 𝛽𝛼𝜆p t (6.6)

Where
𝛽 = the likelihood of function or mission loss given specific conditions.
𝛼 = failure mode ratio (for an item, 𝛼 = 1).
𝜆p = item failure or hazard rate.
t = operating or at-risk time of item.

6.2.5.1 Steps in Performing an FMECA

The first and foremost step is to collect all the information regarding the design. These include
Computer-Aided Engineering (CAE) data, drawings, full specifications, stress analysis, test results,
etc. Reliability prediction information must be available for criticality analysis, if not available it
needs to be generated. For complete analysis functional block diagram and RBD needs to be pre-
pared. The FMECA process should commence once initial design details become accessible.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
104 6 Physical Reliability Methods and Design for System Reliability

The authors in Mzougui and El Felsoufi (2019) propose a modified FMEA to improve the reliabil-
ity of any item. FMEA is applied to the semiconductor industry in Cabanes et al. (2021). Healthcare
risk analysis is done using FMEA in Liu and Liu (2019). The authors in Balaraju et al. (2019) use a
Fuzzy-FMEA-based risk assessment method of a Load-Haul-Dumper machine.

6.2.5.2 Advantages of FMEA

The benefits of employing FMEA encompass:

● Enhancements are made to the product’s quality, safety, and reliability.

● The image of the organization manufacturing the product is improved.
● User is more satisfied.
● Time and cost for system development is minimized.
● Future failures can be reduced.
● Identifies and eliminates the failure modes early.
● Minimizes the possibility of same type of failure again and again.

6.2.6 Accelerated Life Testing (ALT)

Accelerated Life Testing (ALT) is a testing methodology used to assess the reliability, durability,
and lifespan of a product or component in a shorter time frame than its actual expected life under
normal operating conditions. This testing approach involves subjecting the product or component
to elevated levels of stress, such as increased temperature, humidity, voltage, mechanical load, or
other environmental factors, to accelerate the aging and wear-out processes. The purpose of ALT
is to predict and identify potential failures and weaknesses in the product more quickly, allowing
manufacturers to make design improvements and ensure that the product meets reliability and
safety standards before it reaches the market.
In Woo et al. (2020), the authors outline the reliability-focused design of a household compressor
that undergoes recurrent internal stress. The book Balakrishnan (2023) brings forward the recent
advances in reliability and life testing. In Diao et al. (2019), the authors have described the utiliza-
tion of ALT and modeling capacity degradation for a LiCoO2 -graphite cell.

6.2.7 Reliability Apportionment

Individual reliability of subsystems can be evaluated against overall system reliability using Relia-
bility Apportionment which is sometimes necessary. It holds a significant role during the designing
process of a system, especially when a desired system reliability is expected or when the system is
large enough. The subsystem specifications have to be decided based on the requirement of indi-
vidual reliability of the sub-subsystem. This activity is known as a reliability “budgeting.”
Reliability Apportionment starts with the drawing of the RBD for the system. Each block is
assigned the requisite reliability, taking into account, the complex nature, different operating con-
ditions, and risk associated with each block.
Forcina et al. (2020) present a comprehensive overview of the reliability allocation techniques
employed in recent times through a systematic literature review. Maheve and Peter (2021) put
light on its various applicability and limitations. In Choudhary and Monisha (2020), reliability
estimation of power supply modules of aerospace is done using ARINC Reliability Apportionment
Technique. The authors in Shen et al. (2022) provide a cost minimization model for reliability allo-
cation of mechanical system.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6.3 Design Analysis and Process 105

6.3 Design Analysis and Process

Reliability is an essential part of today’s marketplace. As a result, there is a need to establish a

team that supports the entire process of design and development. The three basic activities that the
reliability team will undertake are:

1. Design for Reliability

2. Reliability Verification
3. Analytical Physics

With these activities, a robust reliability program can be built up.

DfR takes precedence as the initial and most crucial step in product development. It makes use of
knowledge about the physics of failure (PoF) to design the prospective problems. The next activity
is Reliability Verification. In this, the objectives for customer reliability are ensured through ver-
ification studies and demonstration tests. Analytical Physics, which is the third activity, collects
information about the PoF of the product. In this way, comprehending how a product will fail and
the reason behind it is essential to the design of a product that meets the customer’s expectations.

6.3.1 Design for Reliability

DfR is a technique that begins at an early stage of the design process and is incorporated into every
subsequent stage. It sets reliability goals for the design team. It makes use of several tools and
practices to ensure reliability of product. The primary stages of the DfR Program comprise the
subsequent measures, as shown in Figure 6.10 (sequential phases represented in block format):

● Identify,
● Design,
● Analyze,
● Verify,
● Validate, and
● Monitor and Control.

Customer
Front end business process
metrics critical
to quality Post sales
identification business
Design process process
Phase 5
Phase 1
Phase 2 Phase 3 Phase 4 Implement Collection
Setting production
System Design and
appropriate Reliability and field
model product for analysis of
reliability verification reliability
development reliability field data
goals systems

Specify Design Implement

Figure 6.10 Design for reliability process.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
106 6 Physical Reliability Methods and Design for System Reliability

Phase 1, known as the identification phase, involves establishing reliability goals. Phases 2 and
3 encompass the design and analysis stages, focusing on modeling the system with reliability
considerations. Phase 4 involves verifying and quantifying the design. Phase 5, validation, entails
conducting functional and environmental tests on the system.
The reliability goals are defined quantitatively in the identify phase. Tools that help in the defini-
tion are: Goal Setting, Bench-marking, Gap Analysis, Reliability Program Plan Writing, and Quality
Function Deployment (QFD).
In the design phase stage, specific design activities are started. These activities include circuit
layout, mechanical drawing, selection of components/suppliers etc. Some of the important tools
used in this phase are: FMEA, Tolerance evaluation, Reliability prediction, Evaluating Tolerance and
trade off cost.
The third phase in the DfR is the analysis phase. In this phase, a rough estimate of the reliability
is done. Tools used in this phase are RBD, Finite Element Analysis (FEA), Physics of failure (PoF),
Reliability Prediction, FMEA, Stress-Strength Analysis, Design Review by Failure Mode (DRBFM), and
Warranty Analysis of the existing products.
The next phase is the Verify phase. In this, the hardware prototype is built. All the test results are
quantified. The prototype is ready for testing and thorough analysis. Tools used in the Verify phase
include: Accelerated Life Testing (ALT), Highly Accelerated Life Testing (HALT), Test to Failure (Life
Data Analysis), Degradation Analysis, Reliability Growth Process, and Design Review Based on Test
Results (DRBTR).
The functional and environmental testing of the system is done in the Validation phase. It may
happen that there is modification in the design in order to make the system more robust. Tools
used in the process include: Design Validation (Including ALT and Reliability Demonstration) and
Process Validation.
To keep the variation within limits, the Control and Monitor phase is used. Tools implemented
for this phase are: Control Chart and process capability Studies, Human Reliability, Continuous
Compliance, Ongoing Reliability Testing (ORT), Audits, and Field Return Analysis.

6.3.1.1 Design for Reliability Tools

DfR tools are used in the development of the product, such that the design is robust and matches
the customer’s requirement.
1. Failure Mode and Effect Analysis: Failure mode and effect analysis (FMEA) is a major tool
to integrate reliability in the design of products. FMEA can identify the reason for failure, the
gravity of the failure, and the likelihood of malfunction.
2. Competitive Benchmarking: Competitive benchmarking measures how and where the prod-
uct stands against its competitors. This is done by using a set of predetermined metrics. The
design process with such benchmarking makes sure that all the significant design aspects have
been included. It is always important to assess the competition. This guarantees a product’s
cutting-edge performance, optimal choice of materials, robust packaging, mechanical and elec-
trical soundness, and the identification of potential cost concerns.
3. Reliability Predictive Modeling: Reliability predictive modeling estimates the product mean
time between failure (MTBF) and mean time to failure (MTTF). These indices hold a significant
role in comprehending the viability of the design’s capability to achieve the reliability objectives
required to satisfy the needs of consumers. Also, the evaluation of these metrics directs and
assists in the trade-off design of the product so that the best design approach can be adopted.
Although, reliability predictive modeling provides an excellent comparative result; it is known
to have limited accuracy.
4. Physics of Failure: It is via the concept of PoF that we gain insight into the material’s phys-
ical attributes, processes, and technologies employed in the design, and the interplay of these
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6.3 Design Analysis and Process 107

attributes with the potentially detrimental conditions encountered by the design throughout the
product’s lifecycle.
5. Design Studies: Design studies find out the capabilities of platforms and products to meet life’s
requirements. The data provided by design studies is used to calculate the precise MTBF or
MTTF. The design maturity testing platform (with no failure) is designed to meet the customer’s
need.

6.3.1.2 Design for Reliability Metrics

1. Risk Assessment: Risk assessment is performed to manage and keep the level of DfR activity
at an appropriate level. Some inquiries that should be posed during the initial stages of program
development include:
● Will this product use any new technology with an unproven track record of reliability?

● Will the new design be different from the old one (approximately more than 30% new con-

tent)?
● Will there be a different environmental exposure to the product?

● Are there any new requirements for this product?

● Are there new applications for this product?

● Is there any new material used in the design process?

● Will there be any new suppliers for the products?

● Will the product be manufactured at a different location?

● Are there any other changes in the product design compared to the old one that affect relia-

bility? If the answer is yes, then the risk is high and more DfR tools will be required to design
the product development.
2. Assessing Program using IEEE 1624 Scoring System: Reliability assessment is a detailed
analysis of the organization techniques and practices used in the development of a reliable
product. It results in a reliability program plan (RPP) and captures the present situation of the
organization. The outcomes of the reliability assessment are mapped onto a reliability matu-
rity matrix. IEEE 1624 scoring system defines the key reliability practices through assessment
methodology. Some of these practices are:
● Reliability requirements and strategic planning

● Training and development

● Reliability assessment

● Testing for reliability

● Supply chain management

● Monitoring and examining field information

● Confirmation and authentication

● Enhancements in reliability

3. Assessing Program using AIAG Scoring System: AIAG stands for AIAG which also contains
assessment methodologies enumerated below:
● Planning for reliability

● Reliability-centric design

● Reliability modeling and prediction

● Reliability of mechanical components and systems

● Concepts of statistics

● Analyzing and reporting the failures

● Analysis of reliability Data

● Reliability testing

● Manufacturing reliability
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
108 6 Physical Reliability Methods and Design for System Reliability

6.3.2 Reliability Verification

The third building block activity in the design process is reliability verification. It involves ensuring
that a product or system functions as intended and remains operational over its expected lifespan,
even in the presence of various stresses, environmental conditions, and potential failures (Wasser-
man 2002). A structured method for conducting reliability testing begins by establishing a reliability
objective, followed by executing tests that are directly tied to performance in order to assess the
product’s reliability. In contemporary industries, it is imperative that a reliability verification test
explicitly establishes their connection to the overall reliability performance of the product, as well
as their influence on warranty expenses and customer contentment.
Reliability verification studies, also known as reliability testing, can be done effectively only when
a proper environmental tool is used to simulate the design weakness. The equipment used for test-
ing facilities includes thermal shock chambers, accelerated life test chambers, vibration systems,
and humidity chambers. Data collection methods play a significant role in grasping the intricacies
of the design during the testing phase of reliability verification. The authors in Wen et al. (2023)
introduce an accelerated reliability verification approach for testing the reliability of the motor
drive mechanism within the precision corn seed-metering device.

6.3.3 Analytical Physics

Physical analysis, or failure analysis, also called analytical diagnostics, plays a major role in
reliability engineering. It helps in understanding and determining the reason for failure and
finding corrective action for it. Repeated failures can be minimized by identifying the underlying
reasons for failure. Similarly, finding out the PoF gives a better understanding of the product and
its limitations. Customer satisfaction increases as problem resolution improves and abnormalities
are eliminated. Some of the major factors for finding the reason behind failure are: corrosion,
contamination, effects of stress strain, diffusion, etc.
Failure in a system may also be due to human errors. This cannot be ignored as an important
contributing factor. In Saket et al. (2011), the authors introduce a novel methodology that employs
a probabilistic approach to conduct biorhythmic analysis, aiming to mitigate aviation accidents.
The methodology is crafted using Gaussian distribution techniques to assess the reliability of the
aviation system while accounting for the influence of biorhythmic factors on the pilot. The theory of
biorhythms predominantly employs scientific techniques to map the patterns or cycles that impact
the internal processes of the body and human conduct. This theory is especially relevant to physical,
emotional, and intellectual capacities. According to the biorhythm theory, three statistical cycles
commence at birth and persist consistently throughout one’s lifetime.

6.3.4 Reliability Life Cycle

Figure 6.11 shows the phases of reliability life cycle. It comprises of several phases, each of which
plays a critical role in ensuring the reliability and longevity of a system or product.
These phases can be summarized as follows:
● Specification for Reliability: Specification for reliability: reliability specifications are crucial
in ensuring that products and systems meet the desired levels of performance and reliability.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6.3 Design Analysis and Process 109

Reliability life cycle

Specification Design for Reliability Reliability Reliability

for reliability reliability verification maintenance audit

Figure 6.11 Phases of reliability life cycle.

They provide a quantitative and qualitative framework for designing, testing, and maintaining
reliable solutions.
● Design for Reliability: In this the product is designed incorporating reliability specifications
into the design process. The risk of costly failures is reduced, customer satisfaction is improved.
● Reliability Verification: This confirms that a product or system meets its specified reliability
requirements and standards.
● Reliability Maintenance: Reliability maintenance is a proactive approach to manage and pre-
serve the reliability of system. Its primary goal is to prevent failures, reduce downtime, extend
the life of asset, and optimize the performance.
● Reliability Audit: Reliability audit evaluates whether the organization is effectively managing
reliability to meet its objective and compliance requirements.
The reliability life cycle consists of a series of interconnected process steps aimed at ensuring that
a product or system maintains its reliability and performance throughout its operational lifespan.
These steps help organizations proactively manage reliability considerations.
Figure 6.12 shows the various process steps followed by a product in its entire life cycle along
with the tools and methods corresponding to these steps.

Reliability Testing for

FMEA to Predicting reliability reliability.
goal setting identify through Weibull,
through problem area exponential, Monte Including
expectation and allocation of Carlo etc. and reliability in
of customer. reliability. design. service and
manufacturing.

Figure 6.12 Process Steps with related tools and methods in Reliability Life Cycle.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
110 6 Physical Reliability Methods and Design for System Reliability

6.4 Conclusions

The chapter offers an extensive overview of physical reliability methods, which are used to evaluate
system reliability. It delves into each method’s description, outlines their respective procedures,
and highlights their recent applications in scholarly literature. Additionally, it addresses both the
benefits and constraints associated with these methods.
Furthermore, the chapter explores the realm of design analysis and processes, encompassing
topics such as DfR, reliability verification, and analytical physics. It also provides a concise overview
of the tools and metrics utilized in DfR.
Ultimately, the chapter concludes by examining the phases of the reliability life cycle, delineating
the steps involved in this process, and presenting the relevant tools for each phase.

Bibliography

Badida, P., Balasubramaniam, Y., and Jayaprakash, J. (2019). Risk evaluation of oil and natural gas
pipelines due to natural hazards using fuzzy fault tree analysis. Journal of Natural Gas Science and
Engineering 66: 284–292.
Balakrishnan, N. (2023). Recent Advances in Life-Testing and Reliability. CRC Press.
Balaraju, J., Raj, M.G., and Murthy, C.S. (2019). Fuzzy-FMEA risk evaluation approach for LHD
machine–a case study. Journal of Sustainable Mining 18 (4): 257–268.
Cabanes, B., Hubac, S., Le Masson, P., and Weil, B. (2021). Improving reliability engineering in product
development based on design theory: the case of FMEA in the semiconductor industry. Research in
Engineering Design 32: 309–329.
Choudhary, K. and Monisha, S. (2020). Aerospace power supply modules reliability estimation using
ARINC reliability apportionment technique. In: Reliability, Safety and Hazard Assessment for
Risk-Based Technologies: Proceedings of ICRESH 2019, 825–832. Springer.
Dhillon, B.S. (2004). Reliability, Quality, and Safety for Engineers. CRC Press.
Diao, W., Saxena, S., and Pecht, M. (2019). Accelerated cycle life testing and capacity degradation
modeling of LiCoO2-graphite cells. Journal of Power Sources 435: 226830.
Ebeling, C.E. (2019). An Introduction to Reliability and Maintainability Engineering. Waveland Press.
Ericson, C.A. and Ll, C. (1999). Fault tree analysis. System Safety Conference, Volume 1, Orlando,
Florida, USA, 1–9.
Forcina, A., Silvestri, L., Di Bona, G., and Silvestri, A. (2020). Reliability allocation methods: a
systematic literature review. Quality and Reliability Engineering International 36 (6): 2085–2107.
Jyotish, N.K., Singh, L.K., Kumar, C., and Singh, P. (2023). Batch deterministic and stochastic Petri nets
modeling for reliability quantification for safety critical systems of nuclear power plants. Nuclear
Engineering and Design 404: 112191.
Kabir, S. and Papadopoulos, Y. (2019). Applications of Bayesian networks and Petri nets in safety,
reliability, and risk assessments: a review. Safety Science 115: 154–175.
Kang, J., Sun, L., and Soares, C.G. (2019). Fault tree analysis of floating offshore wind turbines.
Renewable Energy 133: 1455–1467.
Kothari, D.P., Nagrath, I.J., and Saket, R.K. (2022). Modern Power System Analysis. New Delhi: Tata
McGraw-Hill Publishing Company.
Kumar, N., Tewari, P.C., and Sachdeva, A. (2020). Performance assessment of milk processing system
using Petri nets. SN Applied Sciences 2: 1–12.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Bibliography 111

Kumar, K., Sumit, Kumar, S. et al. (2022). Predicting reliability of software in industrial systems using a
Petri net based approach: a case study on a safety system used in nuclear power plant. Information
and Software Technology 146: 106895.
Kuzu, A.C., Akyuz, E., and Arslan, O. (2019). Application of fuzzy fault tree analysis (FFTA) to
maritime industry: a risk analysing of ship mooring operation. Ocean Engineering 179: 128–134.
Li, J., Zhang, X., Zhou, X., and Lu, L. (2019). Reliability assessment of wind turbine bearing based on
the degradation-Hidden-Markov model. Renewable Energy 132: 1076–1087.
Liu, H.-C. and Liu, H.-C. (2019). FMEA for proactive healthcare risk analysis: a systematic literature
review. In: Improved FMEA Methods for Proactive Healthcare Risk Analysis, 15–45. Singapore:
Springer.
Maheve, M. and Peter, P. (2021). Applicability and limitations of reliability allocation methods. In: 2021
Annual Reliability and Maintainability Symposium (RAMS), 1–4. IEEE.
Mzougui, I. and El Felsoufi, Z. (2019). Proposition of a modified FMEA to improve reliability of
product. Procedia CIRP 84: 1003–1009.
O’Connor, P. and Kleyner, A. (2012). Practical Reliability Engineering. John Wiley & Sons.
Peiravi, A., Ardakan, M.A., and Zio, E. (2020). A new Markov-based model for reliability optimization
problems with mixed redundancy strategy. Reliability Engineering & System Safety 201: 106987.
Saket, R.K. (2011). Reliability evaluation of defence support systems. In: Innovations in Defence Support
Systems-2: Socio-Technical Systems, 241–286. Springer.
Saket, R.K., Kaushik, W.C.S.P., and Singh, C.G. (2011). Biorhythmic analysis to prevent aviation
accidents. In: Innovations in Defence Support Systems-2: Socio-Technical Systems, vol. 338
(ed. L.C. Jain, E.V. Aidman, and C. Abeynayake), 207–240. Berlin, Heidelberg: Springer-Verlag.
Shen, L., Zhang, Y., Zhao, Q. et al. (2022). A reliability allocation methodology for mechanical systems
with motion mechanisms. IEEE Systems Journal 16 (4): 5596–5607.
Singh, K., Kaushik, M., and Kumar, M. (2022). Integrating 𝛼-cut interval based fuzzy fault tree analysis
with bayesian network for criticality analysis of submarine pipeline leakage: a novel approach.
Process Safety and Environmental Protection 166: 189–201.
Singh, S., Saket, R.K., and Khan, B. (2023). A comprehensive review of reliability assessment
methodologies for grid-connected photovoltaic systems. IET Renewable Power Generation 17 (7):
1859–1880.
Theristis, M. and Papazoglou, I.A. (2013). Markovian reliability analysis of standalone photovoltaic
systems incorporating repairs. IEEE Journal of Photovoltaics 4 (1): 414–422.
Vesely, B. (2002). Fault Tree Analysis (FTA): Concepts and Applications. NASA HQ.
Wang, W., Loman, J., and Vassiliou, P. (2004). Reliability importance of components in a complex
system. In: Annual Symposium Reliability and Maintainability, 2004-RAMS, 6–11. IEEE.
Wasserman, G. (2002). Reliability Verification, Testing, and Analysis in Engineering Design, vol. 153.
CRC Press.
Wen, C., Zhang, J., Zheng, K. et al. (2023). Accelerated verification method for the reliability of the
motor drive mechanism of the corn precision seed-metering device. Computers and Electronics in
Agriculture 212: 108163.
Woo, S.-w., Pecht, M., and O’Neal, D.L. (2020). Reliability design and case study of the domestic
compressor subjected to repetitive internal stresses. Reliability Engineering & System Safety 193:
106604.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
113

Design for Maintainability and Availability Analysis for System Design

Jyoti Maurya 1 , Om P. Bharti 2 , K. S. Anand Kumar 3 , and R. K. Saket 1
1 Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
2
Department of Electrical Engineering, Government Polytechnic, Ghazipur, Uttar Pradesh, India
3
Electrical Division, CSIR National Aerospace Laboratory, Bangalore, Karnataka, India

7.1 Introduction
It is clearly essential that systems under maintenance are structured to facilitate the easy comple-
tion of tasks, ensuring that the expertise required for diagnosing issues, conducting repairs, and
performing routine maintenance is not excessively intricate. This consideration should encompass
the skill set and training of anticipated maintenance staff and users (Kothari et al. 2022).
The overarching objective for a completed system (be it a car, aircraft, avionics device, or com-
puter) should encompass its capability to function and fulfill its intended function for the entire
projected duration of its planned design lifespan (Diatte et al. 2022). Consequently, while develop-
ing a system, we must avoid the mindset of just delivering the system and disengaging. Instead,
a supplier’s system should offer continuous support throughout the entire operational lifespan
of the product. This involves several concepts illustrated in Figure 7.1. Achieving supportability
necessitates a skillful integration of reliability, maintainability, logistics, operations engineering,
and safety engineering. This integration ensures that the system remains accessible and functional
for its intended purpose throughout its designated operational lifespan.
Maintainability plays a crucial role in ensuring the system’s successful maintenance and con-
tinuous support, as well as its adaptability for modifications and improvements throughout its
operational lifespan (Sarita et al. 2023). Through system analysis and considerations of maintain-
ability, the alignment of project development with a broader systems perspective becomes evident.
The focus of maintainability procedures and documentation lies in guiding the process of upkeep
and highlighting the typical documentation generated for extensive software programs. Through
maintainability analysis, the process of making decisions among various alternative components
while considering trade-offs becomes apparent. When equipment or machinery malfunctions, an
eventuality given that no design can be entirely infallible, it becomes crucial to promptly address
the issue for its speedy restoration, allowing for its continued utilization. Indeed, users are often pri-
marily interested in its availability, denoting the portion of time the equipment remains functional
as required, akin to household television sets.

Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
114 7 Design for Maintainability and Availability Analysis for System Design

Reliability Mean time

engineering between
failure

Maintainability Mean time

engineering to repair
Design, Availability
development,
production, and
operation Logistics Time to
engineering aquare
spare

Operations Operational
engineering constraints No Yes

Requirement satisfied
and minimized
END

Figure 7.1 Requirement of system supportability.

Paying customers, such as those who cover their fares, constitute the primary income stream
for an airline. Consequently, the majority of commercial aircraft need to operate for extensive
hours daily to remain financially viable (Sayed et al. 2020). Irrespective of the intricacy involved,
all equipment must possess favorable maintainability, which gauges the speed at which any dip in
performance is recognized, the issue is pinpointed, repairs are executed, and a confirmation is con-
ducted to reinstate the equipment’s regular functionality. Integrating maintainability into the initial
design is crucial, as with other reliability factors. Trying to enhance it later by making adjustments
to already manufactured equipment won’t yield a satisfactory outcome.
Enhancements that boost maintainability can often be seamlessly integrated into the initial
design without incurring notable extra expenses. Nevertheless, the designer needs to exercise
caution in enhancing maintainability, as doing so might inadvertently compromise reliability.
For instance, replacing soldered connections with plugs and sockets to expedite the removal of
assemblies could potentially result in decreased dependability at times (O’Connor 2010).
After an introduction, this chapter is structured in the following manner: Section 7.2 includes
elements of maintainability, Section 7.3 delves availability of the systems, and Section 7.4 serves as
the conclusion of the chapter.

7.2 Elements of Maintainability

We need to factor in the necessary tasks for sustaining the system right from the initial stages of our
design. The system should either remain operational throughout the mission without any failures,
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
7.2 Elements of Maintainability 115

or the failed components must be substituted. If our system lacks impeccable dependability and is
not immune to wear and tear, we need to address the subsequent inquiries:
● Which components are susceptible to malfunction, and how will their faults be identified? For
instance, in the case of a cathode ray tube (CRT) screen that doesn’t display an image, is the issue
due to screen failure, power supply failure, or the computer ceasing to transmit screen data?
● Is it possible to easily diagnose a range of issues? How fast can the problem be pinpointed? Is
data from sporadic malfunctions retrievable for later analysis? In cases where a problem can’t be
isolated or the system lacks adequate diagnostic tools, the process of restoring the system might
become a lengthy endeavor.
● When can we expect the system to be restored? Is the system designed with modular compo-
nents for easy replacement? Are the components stacked with numerous attachment points?
Can software also assist in identifying issues?
● Will a defective unit be discarded or undergo repairs? If repairs are needed, where should the
process take place? Additionally, what specific tools and personnel are required to carry out the
repair task?
● Would specialized parts be accessible for the purpose of repairing the device? Will specialized
components like traveling wave tubes or low-noise amplifiers be produced continuously if they
are necessary for unit repairs? Will the original manufacturer handle the unit repairs? Will the
supplier sustain their operations if repair agreements are reached, considering potential logistical
challenges?
All these inquiries need to be tackled during the product development phase. While a few of
these matters intersect with logistics (which deals with providing and sustaining a system over its
complete lifespan), they all require attention. The system’s maintenance approach and the plans
for ensuring easy maintenance should both be contemplated in the initial stages of the product’s
design. To achieve this, we should start by examining certain definitions. Which components are
susceptible to malfunction, and how will their breakdowns be identified? As an illustration, in the
case of a CRT monitor failing to display an image, potential culprits could include a defective screen,
a malfunctioning power supply, or a cessation of screen data transmission from the computer.

7.2.1 Maintainability and System Engineering

Figure 7.2 illustrates a global representation of a continuous research endeavor with a long-term
scope, like the space program, and its integration of maintainability. The original purpose of the
horizon mission methodology (HMM) was to examine pioneering space technologies. HMMs are
imaginary space missions characterized by criteria for performance that exceed the capabilities of
current space technologies. These missions are designed to promote abstract thinking by steering
clear of uncomplicated forecasts and variations of existing expertise (Lalli and Packar 2018).
This framework has been created to offer a structured and analytical method for evaluating and
identifying the technological needs of breakthrough technologies (BTOs). Additionally, it aims to
evaluate the extent to which these technologies could bring about groundbreaking capabilities
for advanced space missions. Consequently, we can conceptualize space programs, or similarly
impactful scientific endeavors, as unified and comprehensive initiatives directed toward overarch-
ing objectives, such as achieving milestones like lunar landings, manned missions to Mars, or the
establishment of a sustained human presence on the moon.
The fundamental idea of the program assumes a singular unchanging objective. This involves
gathering well-tested tools to move closer to achieving the objective. Another element of the work
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
116 7 Design for Maintainability and Availability Analysis for System Design

Horizon mission (HMM/BTO)

Space program

Current Proving Ongoing

applied technology exploration

Project A: Space experiment phases

A B C D E

Logistics/maintainability

Assurance: Reliability

Safety and Quality

gem /
ana ring
ent
m m nginee

Manufacturing
pro tems e

Test and evaluation

gra
Sys

Research engineering and system analysis

Figure 7.2 System engineering and maintainability.

involves advancing technology and components, as well as continuously exploring the distant pos-
sibilities of what is known to lie ahead. At the level of a specific project, diverse fields of expertise
come together to plan, construct, roll out, and manage the designated project. As exemplified,
one of these fields is maintainability (Zhang et al. 2016). Extending the diverse efforts for main-
tainability across the different phases of a project leads to the visual representation displayed in
Figure 7.3. The National Aeronautics and Space Administration (NASA) employs five stages to
define a mission. Our aim is to integrate our maintainability program throughout all five stages.
The figure illustrates the descriptions of the tasks involved.
In Sections 7.2.2 and 7.2.3 delineate the various actions. The pivotal aspect is the early introduc-
tion of the project’s maintainability concept within the program. If this step is neglected, long-term
missions will entail higher costs and more periods of inactivity. Granted, the initial expenses dur-
ing early development may increase, but the overall expenditure will be reduced. Several projects,
in their pursuit of securing budgetary approval for a new system, have failed to consider main-
tainability and integrated diagnostics (Muralidharan and Syamsundar 2012). Nevertheless, this
consistently leads to higher overall costs.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
7.2 Elements of Maintainability 117

System engineering: Maintainability

Phase A Phase B Phase C Phase D Phase E Prod.

Pre. Analysis Define Design Develop/Test operation
maintenance

Maintainability program management

Maintainability analysis/trade OFF

Maintenance concepts/requirement/availability

Sparing concept

Level of maintenance

Personal and training

Diagnostic principles/concepts/bit

Schedule/preventive maintenance

Replacement policy

Support equipment

Maintainability design criteria

Supplier maintainability analysis

Maintainability demonstration

Figure 7.3 System engineering and maintainability.

The interconnected nature of the project’s diverse activities, as well as the flow of work and
information between the operations, reliability, and logistics units, can be understood by the
maintainability of the systems engineering process as illustrated in Figure 7.4. The concept of
maintainability evolves through an assessment of system operation and mission prerequisites. This
concept is also shaped by the reliability of components and the numerous evaluations of reliability
conducted. Subsequently, this maintenance analysis collaborates with design engineering to
formulate a design that is both reparable and maintainable (Leslie 1986).
Ultimately, data and requirements related to maintainability are transmitted to the logistics
sector, facilitating the creation of effective support resource initiatives. The outcome of the main-
tenance analysis also holds significant importance for logistical support analysis. The Logistics
Support Analysis Record (LSAR) and the development of Support Resources contribute to the
formulation of a comprehensive plan encompassing several elements:
● Designing suitable facilities for equipment housing and ground operations.
● Procuring ground support equipment.
● Devising a logistics plan and other associated activities.
● Generating technical publications containing equipment operation and maintenance
instructions.
● Identifying necessary personnel and training prerequisites for equipment maintenance, repair,
and support.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
118 7 Design for Maintainability and Availability Analysis for System Design

Operational requirement Maintainability

and system requirement program plan

Engineering Maintenance
Reliability
design data concept/
document plan
Specification Allocations life
layout/drawing Maintainability prediction
technical analysis failure data
reports design Maintenance failure analysis
changes analysis FMEA/CIL

Logistics support analysis

Support resources development/LSAR

Support Tech.
Facilities Logistics Personnel
equipment Pub.

Maintainability
demonstration Customer

Figure 7.4 Maintainability in the system engineering process.

Ultimately, a maintainability demonstration is conducted to assess the actual time required for
diagnosing and physically replacing a line replaceable unit (LRU) or an orbital replaceable unit
(ORU).

7.2.2 Maintainability Analysis Process

The assessment of maintainability (Figure 7.5) also considers three factors:

● Estimated time for corrective maintenance multiplied by the number of failures.

● Estimated time for preventive maintenance (PM) multiplied by the number of scheduled PMs.
● Estimated time for replacing limited-life items multiplied by the number of scheduled
change-outs. By incorporating these values, an estimation of the total maintenance time per
period is derived.

Moreover, the potential downtime due to maintenance allows us to predict the system’s avail-
ability. The overall availability, along with other indicators of maintainability, is adjusted as the
design progresses and as the Failure Mode and Effects Analysis/Critical Items List (FMEA/CIL)
and supplier maintainability program data become more refined. The insights garnered from the
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
7.2 Elements of Maintainability 119

Specifications FMEA/CIL
Maintenance
concept/
Supplier maintainability Maintainability
maintainability demonstration
program

Design Failure
Maintainability surveillance reporting
design
checklist

Maintainability
analysis M
Design REQR.
Maintainability
Design problem
trade OFFs define

Design reviews

Corrective
action systems

Maintenance
analysis (M
Task/support)

Figure 7.5 Maintainability analysis process.

maintainability analysis are employed by project management to evaluate the overall maturity of
the design and its capacity to fulfill program objectives.

7.2.3 Maintainability Analysis Mathematics

The availability needs can be achieved through the utilization of an extremely reliable system, a
system that is easily repairable and possesses an ample reserve of spare parts, or a fusion of both
approaches. The degree of system employment and the specific mission profile will similarly influ-
ence the stipulations for system availability. Several instances from NASA and other sources that
showcase continuous and intermittent mission requirements have been given as follows:

● Continuous operations
1. Spacecraft (LEO)
2. Space station
3. Air traffic control system
● Intermittent operation—on demand
1. Emergency vehicle
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
120 7 Design for Maintainability and Availability Analysis for System Design

2. Research fighter
3. Shipboard Gatling gun
● Intermittent operation—schedule
1. Space experiment
2. CAT scan I MRI equipment in hospital
3. Pace Shuffle main engines

7.3 Availability of the Systems

Availability is the likelihood of an item being accessible when required, represented as the propor-
tion of time the item is usable. This aspect is determined for a repairable item through its failure
rate, denoted as A, in conjunction with its rate of repair or replacement. The steady-state availabil-
ity is the proportion of the total period that the item is available. The steady-state availability of a
simple unit with a constant failure rate X and a constant mean repair rate, where MTTR is mean
time to repair
𝜇 MTBF
A= = (7.1)
𝜆+𝜇 MTBF + MTTR
The likelihood of the item being available at time t is equivalent to its immediate availability:
𝜇 𝜆
A= + ∗ exp[−(𝜆 + 𝜇)t] (7.2)
𝜆+𝜇 𝜆+𝜇
As the value of t increases, availability tends to approach a steady state. When assessing design
choices, it is usually more insightful to focus on system nonavailability. The steady-state unavail-
ability of system is illustrated by Eq. (7.3):
𝜇
=1−A= (7.3)
𝜆+𝜇
and instantaneous unavailability of system is illustrated by Eq. (7.4):
𝜇 𝜆
= − ∗ exp[−(𝜆 + 𝜇)t] (7.4)
𝜆+𝜇 𝜆+𝜇
If there is a need for planned maintenance that involves taking the system offline, this down-
time should be taken into account when calculating the system’s availability. The availability of
extra units for replacement repairs is often a relevant aspect to think about, which depends on past
usage of spare units and the rate at which replacement units can be repaired. In intricate systems
like power plants, satellites, chemical plants, and radar stations, the emphasis is on availability.
While strong reliability is necessary, it’s not sufficient to guarantee the system’s availability when
required. Equally important is the ability to swiftly perform repairs and essential routine mainte-
nance tasks, ideally without causing a system shutdown. Consequently, ensuring maintainability
is a vital aspect of designing for optimal availability, often necessitating a balance between reliabil-
ity and maintainability attributes. Several electronic systems, such as those found in built-in test
equipment (BITE), exemplify this approach.
The heightened intricacy may diminish dependability and result in erroneous failure alerts.
Nevertheless, BITE can substantially curtail maintenance durations and consequently enhance
accessibility by promptly pinpointing defect locations (This is not the sole rationale for BITE
adoption. It can also obviate the necessity for external testing apparatus and training for diagnostic
procedures, among other benefits.), redundancy also influences availability. The overall availability
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
7.3 Availability of the Systems 121

could see a notable rise if backup systems can be repaired and refurbished while the primary
system carries out the vital function. Certain system configurations exhibit characteristics of reli-
ability and availability in steady-state. These configurations showcase notable enhancements in
reliability and steady-state availability through the implementation of redundancy. Nevertheless,
these scenarios are relatively uncomplicated due to the underlying assumption of a consistent
failure rate. Furthermore, for the standby redundancy situation, it is assumed:
1. The changeover system’s dependability is one.
2. There are no common-cause failures.
3. Failures are immediately discovered and remedied.
Certainly, these conditions might not consistently hold, particularly when dealing with standby
equipment. Such equipment needs to undergo periodic inspections to confirm its operational sta-
tus. The determination of its availability is based on the frequency of these tests. While methods
like BITE are at times employed for electronic devices, they do not guarantee a perfect detection
rate for all issues. In practical situations, these variables need to be taken into account, leading to
intricate analyses.

7.3.1 Availability of Repairable Systems

Maintaining a system’s ease of repair and its ability to undergo regular PM directly affects how
accessible the system is. The duration required for troubleshooting issues and conducting routine
upkeep can temporarily render the system unavailable. Consequently, a strong connection exists
between reliability and maintainability, each impacting the other, while also jointly influencing the
system’s availability and associated expenses. When the system has settled into a stable state, once
any temporary variations have diminished and maintenance procedures are expected to proceed
uniformly:
MTBF PMcycle
Availability (A) = ∗ (7.5)
MTBF + MTTR PMcycle + PMtime(total)
The PM cycle refers to the period during which all PM chores are finished. The visual representa-
tion of the aforementioned availability, maintainability, and reliability criteria can also be observed
in Figure 7.5, depicting availability as a function of mean time between failures (MTBF) and MTTR.
The boundary of the “solution space” for the system and mission requirements is determined by
the 0.990 availability line, a minimum MTBF of 300 hours, and a maximum MTTR of five hours.
Additionally, the constant availability lines within this graph are generated using average values of
MTBFs and MTTRs. MTTR and MTBF are commonly considered variables distributed with either
an exponential or normal distribution.

7.3.2 Availability of Nonrepairable Systems

Mean time to failure (MTTF) for nonrepairable systems is the average time it takes for a system or
component to fail, and it cannot be repaired or restored once it fails. MTTF is a measure of reliability
and is typically expressed in units of time, such as hours, days, or years. It represents the expected
lifespan of the system or component before it experiences a failure that renders it permanently
non-operational. Availability for nonrepairable systems can be calculated in terms of MTTF and
MTBF using the following formulas:
MTTF
A= (7.6)
MTBF + MTTR
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
122 7 Design for Maintainability and Availability Analysis for System Design

A is the availability of the system. MTTF is the mean time to failure, which represents the average
time until a failure occurs. MTTR is the mean time to repair, which is not applicable to nonre-
pairable systems because they cannot be repaired once they fail. For nonrepairable systems, MTTR
is essentially zero because there are no repair actions taken.

Problems

7.1 A power component has an MTTF = 34 hours and an MTTR = 2.5 hours. What is the
steady-state availability? If the MTTR is reduced to 1.5 hours, what MTTF can be tolerated
without decreasing the steady-state availability of the power component?
Solution:
MTTF 34
Steady-state availability = MTTF+MTTR = 34+2.5
= 0.931 MTTR is reduced to 1.5 hour
Now,
MTTF
= 0.931
MTTF + 1.5
MTTF = 0.931(MTTF + 1.5)

(1 − 0.931)MTTF = 0.931 × 1.5

0.931 × 1.5
MTTF = = 20.239 hours
1 − 0.931

7.2 A power system protection relay has a constant failure rate of 0.2 per day. Once it has failed,
the mean time to repair is 2.5 days with constant repair rate. What are the point availability at
the end of two days the interval availability over a two-day mission, starting from zero and the
steady-state availability? (i) If two protective relays operate in series, compute the availability
at the end of two days. (ii) If they operate in parallel, compute the steady-state availability of
the system. (iii) If one relay operates in a standby made with no failure in standby, what is the
steady-state availability?
Solution:
1
𝜆 = 0.2 per day ; = 2.5 ∴ 𝜇 = 0.4 per day
𝜇

(i) The point availability

𝜇 𝜆 −(𝜆+𝜇)
Ae (t) = ∓ e t
𝜆+𝜇 𝜆+𝜇
∴
0.4 0.2
Ap (2) = + e−(12)+0.4)×2 = 0.65 ± (0.33 × 0.30) = 0.7590
0.2 + 0.4 0.2 + 0.4
𝜇 𝜆 { }
Aj (T) = + 1 − e−(𝜆+𝜇)T
𝜆 + 𝜇 (𝜆 + 𝜇) × T
2

0.4 0.2 { }
∴ Al (T) = + 2
1 − e−(0.2+0.4)×2
0.2 + 0.4 (0.2 + 0.4) × 2
0.66 + 0.28 × (1 − 0.30) = 0.8560
𝜇 0.4
A(∞) = + = 0.66
𝜆 + 𝜇 0.2 + 0.4
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Bibliography 123

{ }2
(ii) As (2) = Ap (2) = (0.7590)2 = 0.5760
(iii) As (∞) = 1 − {1 − A(∞)}2 = 1 − {1 − 0.66}2 = 0.84
(iv) For the standby redundant system
𝜆1 𝜇 + 𝜇 2
As (∞) = ; Here 𝜆1 = 𝜆2 = 0.2 and 𝜆 = 0.4
𝜇 2 + 𝜆1 𝜇 + 𝜆1 𝜆 2
(0.2)(0.4) + (0.4)2
As (∞) = = 0.8571
(0.4)2 + (0.2)(0.4) + (0.2)(0.2)

7.4 Conclusion

The primary objective of a system maintenance program aligns with the mission success, a
paramount goal of every NASA System Reliability and Quality Assurance (SR&QA) office. An
effectively devised maintainability program offers increased availability at reduced expenses,
leading to the creation of designs featuring easily serviceable (and assembled) modules. When
maintenance is taken into account, it curtails the tendency to opt for lower-cost components at the
expense of reliability unless there are valid maintainability trade-offs. Moreover, maintainability
analysis compels the consideration of potential obsolescence and the necessity for upgrades, all
while diminishing overall maintenance hours and the total cost of ownership. The following are
the maintainability program advantages:

● Ensures system preparedness and accessibility.

● Defines the parameters that influence the trade-offs affecting the genuine reliability value.
● Ensures a precise evaluation of maintenance and logistics program costs.
● Facilitates the expansion of the system.
● Minimizes the need for maintenance personnel.
● Lowers operational expenses.
● Decreases the overall cost of ownership.

Bibliography

Diatte, K., O’Halloran, B., and Van Bossuyt, D.L. (2022). The integration of reliability, availability, and
maintainability into model-based systems engineering. Systems 10 (4).
Kothari, D.P., Nagrath, I.J., and Saket, R.K. (2022). Modern Power System Analysis. New Delhi: Tata
McGraw-Hill Publishing Company.
Lalli, V.R. and Packar, M.H. (2018). Designing for Maintainability and System Availability.
Leslie, R.E. (1986). System Analysis and Design, Methods and Inventions. Englewood Cliffs, NJ:
Prentice-Hall.
Muralidharan, K. and Syamsundar, A. (2012). Stastical Methods for Quality, Reliability and
Maintainability, Eastern Economy Edition. New Delhi: PHI Learning Private Limited.
O’Connor, P.P. (2010). Practical Reliability Engineering. Wiley India Edition.
Sarita, K., Saket, R.K., and Khan, B. (2023). Reliability, availability, and condition monitoring of
inverters of grid-connected solar photovoltaic systems. IET Renewable Power Generation 17 (7):
1635–1653.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
124 7 Design for Maintainability and Availability Analysis for System Design

Sayed, A., EL-Shimy, M., El-Metwally, M., and Elshahed, M. (2020). Impact of subsystems on the
overall system availability for the large scale grid-connected photovoltaic systems. Reliability
Engineering & System Safety 196: 106742.
Zhang, X., Sun, L., Sun, H. et al. (2016). Floating offshore wind turbine reliability analysis based on
system grading and dynamic FTA. Journal of Wind Engineering and Industrial Aerodynamics 154:
21–33.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
125

Genetic Algorithm and Artificial Neural Networks in Reliability-Based

Design Optimization
Heeralal Gargama, Sanjay Kumar Chaturvedi, and Rajiv Nandan Rai
Subir Chowdhury School of Quality and Reliability, Indian Institute of Technology Kharagpur, Kharagpur, West Bengal, India

8.1 Introduction
In the realm of reliability-based design optimization (RBDO), the traditional constraints undergo
a conversion into probabilistic constraints (as outlined by Rao 1992), and probability distributions
are frequently employed to depict the randomness and uncertainty in design variables. To assess the
likelihood of structural design failure, it is essential to construct a joint probability density function
(pdf ) encompassing all problem parameters and design variables. However, creating a joint pdf
can prove to be a daunting task, and in some cases, it may seem nearly impossible, particularly
when dealing with intricate computational requirements and limited data accessibility in complex
engineering systems (Kaymaz and McMahon 2005). To address this challenge and to estimate fail-
ure probability more efficiently, the researchers have developed approximate reliability techniques
and advanced simulation methods (Haldar and Mahadevan 2000; Valdebenito and Schuëller 2010).
These techniques have been integrated in various ways within optimization procedures, resulting
in the development of several RBDO approaches in recent years, as outlined by Deb et al. (2009).
In order to attain accurate approximations of failure probabilities, methods such as the
second-order reliability method (SORM) or first-order reliability method (FORM) are utilized to
ensure an elevated level of precision. In this process of conducting reliability analysis through
these techniques, the initial step entails converting the performance or limit state function from
its original space of random variables into a space of standardized normal variables. In the context
of the FORM, the limit state function is approximated using a first-order Taylor series expansion
centered at the design point, which is typically known as the Most Probable Point (MPP) of failure.
The approximation results in a linear function within the U-space. Conversely, SORM utilizes a
second-order surface to approximate the limit state surface (Youn and Choi 2004a). The process of
locating the MPP involves addressing a constrained optimization challenge and various methods
that can be utilized to accomplish this objective, including:
(a) Performance measure approach (Youn and Choi 2004b);
(b) Reliability index approach (Yu et al. 1998);
(c) Fast performance measure approach (Du and Chen 2001);
(d) Fast reliability index approach (Deb et al. 2009).
Researchers have integrated the aforementioned methods into optimization algorithms in
diverse ways, leading to the development of various methodologies that include:

Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
126 8 Genetic Algorithm and Artificial Neural Networks in Reliability-Based Design Optimization

(a) RBDO using double-loop techniques;

(b) RBDO using single-loop techniques;
(c) RBDO using decoupled techniques.
The above approaches allow designers to efficiently address the challenges posed by uncertainty
and randomness in complex engineering systems. Nevertheless, the FORM and SORM come with
certain limitations (Youn and Choi 2004a; Rahman and Wei 2008):
(a) Nonlinearity: These methods may not provide adequate accuracy when dealing with highly
nonlinear performance functions;
(b) Multiple MPPs: If multiple MPP exist within the system, FORM and SORM methods can
produce erroneous failure probability estimates;
(c) Derivative Requirements: FORM and SORM necessitate the availability of higher-order
derivatives, yet many engineering problems lack analytical derivatives, making their
application challenging.
To address these challenges in RBDO, simulation techniques have been employed, including
direct integration using optimization algorithms and meta-model approximations.
Direct integration involves computing gradient information using Monte Carlo simulation
(MCS). This gradient information can then be efficiently utilized within gradient-based optimiza-
tion algorithms to solve RBDO problems, as demonstrated by Jensen et al. 2009. MCS-based RBDO
techniques are known for their ease of implementation, robustness, and accuracy, which persist
unaffected by the problem’s dimension or intricacy. Furthermore, the demand for a substantial
number of MCS cycles to estimate low failure probabilities can be mitigated through various
methods of sampling, which include importance sampling, stratified sampling, latin hypercube
sampling (LHS), etc.
The MCS technique for evaluating the probability of failure (pof ) is recognized for its ease of
implementation and robustness. However, it does come with certain limitations (Gargama et al.
2014a,b):
(a) Numerical Noise: MCS can introduce numerical noise due to its reliance on a limited sam-
pling size, which may affect the precision of results;
(b) Low Target Probability: In RBDO, when aiming for a low target pof , accuracy of MCS can
deteriorate further with a limited sample size;
(c) Zero Probability: In cases where the pof is extremely low, MCS may erroneously provide a
zero pof .
Meta-models address the above challenges in conjunction with RBDO methods thereby offer-
ing an attractive approach to reduce computational burden. Meta-models are typically combined
with the design of experiments (DOE) to create approximate functions that depict performance
responses within the given boundaries of design space. Among various meta-models, the response
surface method (RSM) and its other variants have proven effective, as demonstrated by their suc-
cessful utilization with different inverse reliability measures (Foschi et al. 2002; Qu and Haftka
2003; Yoo et al. 2011). For smaller problems with few variables, polynomial response surfaces are
particularly appealing due to their simplicity and efficacy (Wang and Shan 2007). Nonetheless,
this approach may not be well-suited for problems featuring highly nonlinear responses or a large
number of random parameters/variables. Accurately representing all peaks and valleys can pose a
challenge. Fortunately, there are alternative meta-models suitable for addressing highly nonlinear
and large-scale optimization problems, including kriging, ANNs, radial basis functions, support
vector regression, and more (Zhu et al. 2011).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8.2 Reliability-based Design 127

The ANNs represent a promising paradigm in this context. They can function as surrogate mod-
els, providing simplified approximations of the limit state or objective function, which proves com-
putationally efficient when assessing failure probabilities. They can also be particularly useful
in combination with methods like first/second-order reliability methods or sampling methods.
Following this, double-loop structures in RBDO have been utilized for optimization purposes, as
exemplified by the research of Chau 2007, Cardoso et al. 2008, and Gondal and Lee 2012.
In this chapter, the author aims to approximate an accurate design response capable of estimating
relatively very low probabilities of failure. To achieve this, we employ the concept of an equivalent
safety factor known as the PSF. It is important to note that obtaining pof using approximated func-
tions is only meaningful if their accuracy is high. To obtain more accurate reliability estimates,
given the extremely nonlinear behavior of the electromagnetic interference (EMI) shielding design
issue, we suggest PSF as a surrogate measure of the pof and use of ANN to approximate PSF1 .
In the proposed approach, the author begins by generating various design points throughout the
entire design space. Subsequently, calculate the PSF at each of these design points using MCS. These
design vectors, in conjunction with their respective PSF values, are subsequently employed for the
training of an ANN. Following the ANN training, it is applied to the design optimization problem to
achieve the desired PSF value while adhering to the specified design variable constraints. To address
the optimization task, we employed a real-coded GA as outlined in the work by Deb et al. (2007)
This specific real-coded GA approach is illustrated through the design of electromagnetic shields
for military applications operating within the frequency range of 7.99–12.5 GHz in this chapter.
The chapter is illustrated as follows. The theoretical backgrounds required for the reliability-
based optimization have been introduced in Section 8.2. Wherein the definition of performance-
based reliability is presented in Section 8.2.1. RBDO formulation, PSF calculation, MCS approach,
and ANN approach are presented in Sections 8.2.2–8.2.5, respectively. The GA deployed for the
optimization is explained in Section 8.2.6. Section 8.3 presents the RBDO methodology stepwise to
solve the case study undertaken. A case study of shielding structure design has been illustrated in
Section 8.4. Equations used for the evaluation of shielding effectiveness are summarized in
Appendix 8.A.

8.2 Reliability-based Design

8.2.1 Performance-based Reliability

In the system or product development lifecycle, the primary goal is to ensure that the intended
functions are achieved with satisfactory performance levels. The design approach utilized has a
direct impact on both the functional attributes and the performance outcomes. The system’s design
capacity (also referred to as “capacity” or “resistance” in the literature) is analogized to the sys-
tem’s strength, while the customer’s requirements (referred to as the load or demand) are analo-
gized to the stress. This analogy allows us to calculate performance-based reliabilities using the
stress-strength theory (Tsai and Chang 2008). The performance requirements can alternatively be
articulated as an “acceptable level of damage defined by acceptable probability of failure.” To be more

1 The concept of PSF was initially introduced by Birger (1970) and has been evaluated in comparison to
conventional measures of pof and safety indices for application of reliability-based design by Qu and Haftka 2003.
Their research showed that PSF maintains accuracy, especially in the low pof region, and exhibits lesser variation
compared to traditional failure probability or safety indices. Consequently, approximating the limit state function
using PSF leads to better accurate results than using traditional failure probabilities.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
128 8 Genetic Algorithm and Artificial Neural Networks in Reliability-Based Design Optimization

Probability density Figure 8.1 Reliability assessments using “bigger the

better” PI.
Unreliability L
Reliability

Performance of the design

precise, performance-based design (PBD) can be defined as follows; “Design specifically intended to
limit the consequences of one or more perils to a predefined acceptable level” (Frydrýšek 2008). The
classification of PBD into four distinct categories is determined by the characteristics of perfor-
mance indices (PI), as outlined by Tsai and Chang (2008):

(a) Nominal the Best PI: In this category, the notion is that the provided performance is consid-
ered better when it aligns more closely with the target value.
(b) Smaller the Better PI: The category signifies that provided performance is deemed better
when it is smaller, with the optimal target value being zero.
(c) Bigger the Better PI: This classification implies that superior performance is associated with
larger values, with an emphasis on achieving the highest possible level of strength or a related
attribute. Consequently, the optimal target value in this scenario can be a positive maximum.
(d) Zero Point Proportional PI: In this category, the objective is to have the output and input
exhibit a closer, constant proportional relationship for better performance. The ideal outcome
in this category may be represented as a straight line passing through the origin.

In this chapter, the author focuses on the (c) “bigger the better” PI for reliability evaluation, as
depicted in Figure 8.1. In this context, when evaluating the design performance against a predefined
threshold value labeled as “L,” the reliability of the structures can be described as the likelihood
that the performance of the design surpasses the specified value. For the “bigger the better” PI, the
reliability evaluation interval would be from L to ∞.

8.2.2 RBDO Problem

Let B represent a vector comprising independently distributed random design variables
{B1 , B2 , … , BN }, and let C denote a vector containing independently distributed random
parameters {C1 , C2 , … , CM }. Additionally, D represents a vector of design variables with precise
specifications (i.e., deterministic variables). Each parameter Cl , l = 1, 2, 3, … , M and random
variable, Bj , j = 1, 2, 3, 4, … , N is characterized by its pdf , along with respective vectors indicating
standard deviation and mean, denoted as 𝜇Bj , 𝜇Cl and, SDBj , SDCl , respectively. As a result, the
optimization problem exhibits stochasticity arising from two distinct sets, namely B and C, while
both B and D serve as decision variable vectors. During the optimization progression, the standard
deviation of a particular design variable or parameter may be held constant or varied based on a
fixed coefficient of variation, based on the inherent randomness it possesses.
The general RBDO problem formulation, seeking the optimal solution to achieve a desired or
target reliability (R) for any design of the system, can be articulated as (Gargama et al. 2014a):
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8.2 Reliability-based Design 129

Maximize f (𝜇B , 𝜇B , D)
subject to Pr (Ei (B, C, D) ≤ 0) > (1 − 𝛼i ) or Ri , i = 1, 2, … , I
B L ≤ 𝜇B ≤ B U
DL ≤ D ≤ DU
𝛼i ∕R ∈ [0, 1] (8.1)
In the above formulation (8.1), design variables bounds are represented by the subscripts U and
L, while 𝛼 i = 1 − Ri , denotes the probability of meeting the ith constraint. To be noted that the
Ei (B, C, D) represents the ith performance function/performance criteria. The limit state function
for the ith performance criteria are defined as Ei (B, C, D) = 0, which serves as the boundary dis-
tinguishing between success and failure regions within the given design parameter constraints.
A design is considered success if Ei (B, C, D) > 0 and failure otherwise. The joint probability of
achieving feasible solutions in the presence of randomness or uncertainty, for all I constraints is
denoted term Pr(•) denotes. These constraints, Pr(Ei (B, C, D) ≤ 0), are commonly referred to as
chance constraints and pose challenges in solving RBDO problems as they are not straightforward
to address.2
The pof , denoted as pfi , for the solution with respect to the ith constraint is determined through
a multi-dimensional integration across the failure domain, as follows:

pfi = f𝜇X ,𝜇P (B, C) dBdC (8.2)

∫Gi (B,C,D)≤0
Here f𝜇B ,𝜇C represents the joint pdf of all random design variables and parameters. In many engi-
neering applications, the joint pdf might not have a readily available explicit expression, even when
the pdf of individual variables and parameters are known. In such cases, it becomes necessary to
employ simulation and approximation methods, such as SORM, FORM, MCS, ANN, RSM, etc.
Simulation and approximation approach used in this chapter are presented in Sections 8.2.4 and
8.2.6, respectively. The next section will go through how to compute the PSF, an alternative index
related to the factor of safety, which improves the precision of probability calculations, especially
in scenarios with low failure probabilities.

8.2.3 Probability Sufficiency Factor

Traditional design methods commonly utilize cautious scaling factors like safety factors (SF) or
safety margins to guarantee the security of design. The conventional SF is defined as follows:
SFi = gRi ∕gSi (8.3)
These SF are calculated using the nominal resistance, i.e., capacity or strength, denoted as gRi ,
and the nominal load, i.e., demand or stress, denoted as gSi , corresponding to the ith performance
criterion. However, relying solely on these SF may not be adequate, as they do not provide insight
into the actual safety margin in design reliability.
Hence, in the context of RBDO, reliability constraints are articulated using the pof and are struc-
tured as a stipulation on the SF, as follows:
p(SFi ≤ 1) = ptfi (8.4)

2 Most of the research in the field of RBDO is dedicated to efficiently and accurately estimating these constraints.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
130 8 Genetic Algorithm and Artificial Neural Networks in Reliability-Based Design Optimization

Figure 8.2 Pdf of the SF.

1.5

PSF = 1
1
Density

PSF = 0.8

0.5

0
0.5 1 1.5 2 2.5
Safety factor

where ptfi is the target pof . PSF for an RBDO application with the target pof denoted as ptfi is
solution to the following equation:
p(SFi ≤ PSFi ) = ptfi (8.5)
Note: Figure 8.2 illustrates the schematic pdf of the SF for a single performance criterion let’s call
it E1 (B, C, D) = 0. In this figure, the region beneath the curve on the left side of the vertical line SF ≤ 1
represents the pof . It is worth noting that by employing the inverse transformation described in
Eq. (8.5), we can determine a specific value for the SF corresponding to a given value of ptf .
For example, consider the region to the left of the vertical line where PSF = 0.8 (equivalent to SF = 0.8),
indicating an acceptable pof , the value 0.8 indicates that achieving this probability is possible through
either a 25% increase in capacity or a 20% reduction in load (calculated as 1/0.8-1).

8.2.4 MCS Approach

In MCS, the goal is to capture the probabilistic behavior of a system by repeatedly sampling each
random variable and parameter from their respective representative/estimated pdf . The MCS pro-
cess involves the following steps:
(a) Express the problem by incorporating all the parameters and random variables.
(b) Construct pdfs for each input random variable and parameter. These pdfs are commonly
derived from engineering estimations or the insights of experts.
(c) Generate random samples from respective pdfs of each random variable and parameter.
(d) Evaluate the system’s response for each set of values generated in Step (c).
(e) Extract probabilistic data by tallying the occurrences of simulation cycles where the perfor-
mance function fails. In other words, calculate the fraction of simulations where Ei (B, C, D) ≤ 0
occurs.
This fraction provides an unbiased estimator of the pof from the MCS. It is determined as follows:

1 ∑
N
Qf
pfi = Gb (Bb , Cb ) = b (8.6)
N b=1 N
Here, pfi represents the approximated pof for the ith performance criteria, and the value of indica-
tor Gb (Bb , Cb ) = 1 if Ei (B, C, D) ≤ 0 and 0, otherwise. The symbol N and Qfb denotes combined count
of simulation trials and the cumulative number of simulations resulting in failure, respectively.
The accuracy of MCS typically relies on the total number of simulation cycles. Each individual
experiment or simulation within MCS obeys to a binomial distribution, characterized by a certain
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8.2 Reliability-based Design 131

probability of failure or success, and is considered independent. By employing a normal distribu-

tion approximation to the binomial distribution, one can compute the anticipated error linked to
the number of cycles in the simulation. As a result, when considering a 94.99% confidence interval,
one can approximate the percentage deviation in the estimated pof as:
√( ) ( )
error% = 1 − pTf ∕ N × pTf × 200 (8.7)
where pTf is the true probability of the failure. A significant limitation of MCS is the necessity for a
substantial sample size, denoted as “N,” when calculating low failure probabilities. To mitigate the
computational expenses associated with estimating these low failure probabilities, various system-
atic sampling techniques have been devised. These methods include importance sampling, Subset
Simulation, LHS, Line Sampling, and the Horseracing Simulation Method, etc. (Zuev 2009). Here,
the LHS is applied as a DOE tool.

8.2.5 Latin Hyper Cube Design

The design of Experiments generates a collection of input design vectors within the design space,
often referred to as the experimental region. Subsequently, each set of values of the design vectors
is used to assess the limit state function. This procedure can be divided into two types: (i) physical
experiments and (ii) computer-generated experiments, depending on the nature of the simulation
model employed, which can be either deterministic and/or stochastic. The deterministic model
generates response values by applying mathematical relationships that align with the input design
or decision variables. Consequently, these responses are free from random errors and exclusively
encompass systematic errors, often arising from approximated models or model uncertainty.
To minimize systematic errors, various space-filling designs are suggested, including LHS
designs, factorial design, and central composite design. Among these, LHS design stands out
as a favored space-filling design and is particularly well-suited for DOE in computer-based
experiments. It excels at ensuring dense coverage of each variable’s entire range within the design
space (Myers et al. 2009). LHS design is capable of handling a substantial number of variables and
has been found to provide accurate estimates of output (i.e., means, variances, and distribution
function) compared to other sampling approaches such as random and stratified sampling (McKay
et al. 1979).
Using the LHS procedure, the generation of a set of sampling points k from the design variables
k1 , k2 , … , kn is carried out by the following:
(a) Divide the range of all the variable into k distinct, non-overlapping intervals, with the bound-
aries of these intervals determined based on equal probability of occurrences.
(b) Randomly select one value from individual interval, taking into account the probability density
within that interval.
(c) Pair up the k values obtained for k1 with k values from k2 in a random manner. Continue this
pairing process by combining each pair with k values from k3 , and so on, until the last design
variable is included.
As stated earlier, the nature of the EMI shielding problem is highly nonlinear, we suggest PSF as
a surrogate measure of the pof and use of ANN to approximate PSF.

8.2.6 Artificial Neural Networks

ANNs have found extensive utility in numerous practical applications within engineering and sci-
ence due to their numerous benefits. These advantages include the capacity to accumulate expertise
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
132 8 Genetic Algorithm and Artificial Neural Networks in Reliability-Based Design Optimization

through learning from examples and their inherent ability to generalize (Gondal and Lee 2012).
Single-layer ANNs have limitations when it comes to learning nonlinear functions. Therefore, addi-
tional layers of neurons are introduced between the output and input layers, and these layers are
trained effectively to model any well-defined function with remarkable precision.
Multilayer feedforward ANNs, often referred to as multilayer perceptrons, have gained significant
recognition and have proven successful in tackling a wide array of challenging problems, includ-
ing function approximation, pattern recognition, data classification, etc. A multilayer perceptron
consists of multiple layers, each comprising its set of neurons, connections with associated weights
(represented as a weight matrix), a bias vector, and an output vector. The layers responsible for pro-
ducing the final output are referred to as output layers, while the intermediary layers are known
as hidden layers. As highlighted by Hornik (1991), multilayer perceptron networks equipped with
sigmoid activation functions and linear activation functions in their hidden layers output layers,
respectively, possess remarkable power, enabling them to approximate any continuous function
with a high degree of precision.
In this chapter, the multilayer perceptron networks have an output layer comprising a single
neuron which represents the PSF with respect to the network’s inputs. In the hidden layer, the
determination of the number of neurons is achieved through a trial-and-error process, aiming to
obtain a multilayer perceptron model with the lowest mean squared error (MSE) and a linear cor-
relation coefficient (R-value) that approximates ≈1. For the training of the multilayer perceptron
networks, the back error propagation approach is employed. This technique minimizes the MSE
by adjusting the values of the interconnected weights. The equation employed for computing the
MSE is as depicted below:

1∑
C
MSE = (t − ai )2 (8.8)
C i=1 i
In this formula, ti represents the target output, ai signifies the network’s output for the ith training
dataset, and C denotes the total number of training data samples.

8.2.7 Real-coded GA
After approximating the design response surface in terms of PSF using ANN, an appropriate algo-
rithm is employed to solve the optimization problem presented in Eq. (8.1). GA optimizers typically
demand users to make several upfront choices prior to initiating the optimization process. These
choices encompass parameters like population size (N), crossover probability (pc ), mutation prob-
ability (pm ), selection strategy, specific characteristics of the fitness function, as well as the types
of crossover and mutation to be employed. According to previous considerations rooted in schema
processing (Goldberg et al. 1992), it was proposed that the population size should grow in tandem
with the complexity of the problem. Nevertheless, it is essential to recognize that the appropri-
ate population size should also factor in the signal-to-noise ratio inherent in a given problem. To
address this issue, Deb (2000) advocated for a straightforward calculation method: N = 10n, where
n denotes the variable number within the problem. Furthermore, comparatively larger populations
offer increased genetic diversity, leading to expedited convergence.
Regarding the probability of crossover, it has been observed that a value around 0.7 tends to yield
optimal results across a diverse range of problems. However, a higher crossover probability pc can
enhance the speed of exploration and serve as a primary means for a GA to uncover novel and
improved solutions. In the realm of real-coded GAs, it is worth noting that the mutation opera-
tor is the sole mechanism responsible for generating new values for variables. Consequently, it
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8.2 Reliability-based Design 133

is necessary to set a considerably higher mutation probability pm compared to binary-coded GAs

(typically ranging from 0.01 to 0.1) (Vukovć and Sopta 1999).
The real-coded GA retains the identical evaluation procedure and reproduction operator as the
binary-coded genetic algorithm. Nevertheless, it diverges when it comes to the crossover and muta-
tion processes. Without delving into intricate specifics, here is an overview of the steps involved
in performing mutation and crossover on the population during Steps 5 and 6 of the executed
real-coded GA.

Step 1: Choose N, pc , pm , tournament size (T n ) and maximum generation number (Gmax ).

Step 2: Create initial population of solutions by employing a uniform distribution method within
the predefined range of design parameters.
Step 3: Calculate the fitness function (as described by Deb 2000) by first computing the objective
function and checking for constraint violations in each solution vector, denoted as x⃗,
{
f (⃗x), if gj (⃗x) ≥ 0 ∀j = 1, 2, … , m,
F(⃗x) = ∑m (8.9)
fmax + j=1 ⟨gj (⃗x)⟩, otherwise.

Here, F(⃗x), f (⃗x) and gj (⃗x) are the fitness function, objective function and number of
constraints ( j = 1, 2, … , m), respectively. The value of bracket penalty operator ⟨gj (⃗x)⟩ =
gj (⃗x), when gj (⃗x) is zero and negative, otherwise. The objective function’s maximum
value, denoted as f max , represents the most unfavorable feasible solution within the entire
population. In the case of a minimization problem, it is common to transform and utilize
the fitness function in the following manner:

1
FT (⃗x) = (8.10)
1 + F(⃗x)

Step 4: Utilize the tournament selection operator with replacement to pick favorable strings from
the population for the purpose of creating the mating pool.
(a) Pairs formation: Generate a random set of competitors and subsequently create cou-
ple of competitors, ensuring that there are total of T n pairs.
(b) Winner selection: The winner among the pairs of competitors is determined by iden-
tifying the string associated with the competitor possessing the best fitness, denoted
as FT (⃗x).
(c) Repeat this method T n rounds to obtain the population size equal to N.
Step 5: Perform simulated binary crossover (SBX) operations on randomly chosen pairs from the
population. SBX manipulates offspring solutions around the parental solutions by lever-
aging a distribution index. To initiate this process, first, pick parents for the crossover
operation. Generate a series of random numbers ranging from 0 to 1, with the count of
random values matching the number of strings in the population, represented as N. Exam-
ine each random number to see if it is less than or equal to a predefined threshold, pc . If
it meets this condition, select the corresponding string for crossover; otherwise, place the
strings directly into an intermediate population. To compute the two child solutions, c1
and c2 , from the selected parents, p1 and p2 , adhere to the following steps:
(i) First, within the range of 0 to 1 a random number u is generated. Next, determine the
𝛽̄ ordinate using the polynomial probability distribution function according to the
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
134 8 Genetic Algorithm and Artificial Neural Networks in Reliability-Based Design Optimization

instructions provided below:

⎧ 1∕
⎪(2u) c , if u ≤ 0.5
(𝜂 +1)

⎨ ( ) 1∕(𝜂c +1)
1
⎪ 2(1−u) , otherwise.
⎩
The distribution index, 𝜂 c , is a parameter that can take on any positive value. A smaller
value of the distribution index allows to create the child solutions farther away from
the parent solutions, while a larger value brings them closer. For solving the problem,
undertaken in this chapter, we have considered 𝜂 c = 1.
(ii) The child solutions are calculated as:
̄ 2 − p1 |], c2 = 0.5[(p1 + p2 ) − 𝛽|p
c1 = 0.5[(p1 + p2 ) + 𝛽|p ̄ 2 − p1 |]

For calculating the c1 and c2 within the given constraints of pL and pU , 𝛽̄ needs to be
changed as:
⎧ 1
⎪ (𝛼u) ∕(𝜂c +1) , if u ≤ 𝛼1
𝛽̄ = ⎨( ) 1∕(𝜂c +1)
1
⎪ 2−𝛼u , otherwise,
⎩
2
where 𝛼 = 2 − 𝛽 −(𝜂c +1) and 𝛽 = 1 + c −c min [(p1 − pL ), (pU − p2 )]. Here, it is assumed
2 1
that p1 < p2 , moreover the above equation may be modified for p1 > p2 .
Step 6: With mutation probability of pm , perform mutation operation. To determine the mutation
value for the constraint variables, we establish a perturbation factor denoted as 𝛿̄ in the
following manner:
{[ ] 1∕
2u + (1 − 2u)(1 − 𝛿)𝜂m +1 𝜂m +1 − 1 if u ≤ 0.5;
[ 𝜂
] 1∕(𝜂m +1)
1 − 2(1 − u) + 2(u − 0.5)(1 − 𝛿) m +1 otherwise,
where u represents a random value ranging from 0 to 1, 𝜂 m is the distribution index,
𝛿 = min[(p − pL ), (pU − p)]/(pU − pL ) and the maximum permissible perturbance in the
parent value p is ΔU L max . Subsequently, the mutated child solution is calculated as
follows:
̄ max
c = p + 𝛿Δ
Step 7: Substitute the previous population with the newly generated population.
Step 8: Check for the termination criteria, which may be defined as reaching a maximum number
of generations Gmax , achieving a desired fitness level, or reaching a specified time limit for
calculation. If any of these termination criteria are satisfied, stop the optimization process.
Otherwise, repeat the optimization process starting from Step 3.

8.3 RBDO Methodology Using PSF and ANNs

After presenting the several aspects of RBDO problems in earlier sections and for the sake of brevity,
the entire optimization methodology described in this chapter follows these subsequent steps:
Step 1: Generate a collection of design points that encompass the entire design space.
Generate a collection of design points using the LHS technique within a specified range of
design variables, given in Section 8.2.5 as a DOE method from the design space.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8.3 RBDO Methodology Using PSF and ANNs 135

Step 2: Perform probabilistic assessment.

(a) Define pdf of each parameter and design variable.
(b) Specify the desired output performance and the acceptable pof for that performance.
(c) Assess the PSF by utilizing MCS at every design point generated through DOE, as out-
lined in Section 8.2.3 and Section 8.2.4, respectively.
Step 3: Approximate the PSF using ANN back error propagation architecture (Section 8.2.6) cor-
responding to design points generated via DOE.
In this chapter, ANN is developed and trained to offer a cost-effective means of verify-
ing the violation of probabilistic constraints necessary for conducting reliability analysis.
The ANN’s training process involves using a database consisting of various input design
variable combinations (input) and their corresponding PSF values, which were evaluated
during Step 2.
Step 4: Perform RBDO using GA (Section 8.2.7):
(a) Specify the objective function that needs to be minimized;
(b) Specify the ranges for design variables and deterministic constraints;
(c) Utilize an approximate ANN for checking probabilistic constraints.
Step 5: Verify the convergence conditions, ensuring they meet the reliability requirements. If they
are met, halt the process; otherwise, narrow down the design space and recommence from
Step 1.

8.3.1 Case Study: EMI Shielding Design for the Required Load of 80 dB
In this chapter, the transmission line modeling described in Appendix 8.A is used to calculate the
SE of the planar multilayer shielding structure. An alternative formulation to the RBDO problem
presented in Eq. (8.1) could be to attain the target pof by solving the constraint optimization prob-
lem with the decision variable vectors. Consequently, the design problem, which centers around a
single performance criterion, can be articulated as the task of minimizing the discrepancy between
the outputs of the trained ANN. This discrepancy pertains to the error between the approximated
PSF (PSFa ) and the target PSF (PSFt ). Therefore, the problem statement for the shielding design
can be articulated as follows:

f (𝜀1 , 𝜇1 , d1 , … 𝜀n , 𝜇n , dn ) = Min. ∣ (PSFt − PSFa ) ∣

Subject to d1 + · · · + dn = d or ≤ d
𝜀1L , … 𝜀nL ≤ 𝜀1 , … 𝜀n ≤ 𝜀1U , … 𝜀nU
𝜇1L , … , 𝜇nL ≤ 𝜇1 , … 𝜇n ≤ 𝜇1U , … 𝜇nU
d1L , … , dnL ≤ d1 , … dn ≤ d1U , … dnU (8.11)

where the desired total thickness of the shielding structure is denoted as d (in μm). Constrained
optimization techniques can be applied in both the formulations, i.e., Eqs. (8.1) and (8.11). To
implement RBDO formulation (8.11), sets of input design vectors are generated from the design
space. These vectors are used to compute the PSF using (8.5) equation for each set of values. Typ-
ically, identifying local behaviors in the design space necessitates a larger number of analyses
compared to detecting global responses. To ensure robust coverage and prevent potential inaccura-
cies near the corners of the design space, we generate 500 sampling points. This strategy helps avoid
issues that might arise from using LHS near certain corners, ultimately ensuring better accuracy
in the vicinity of these design points.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
136 8 Genetic Algorithm and Artificial Neural Networks in Reliability-Based Design Optimization

Table 8.1 Distribution properties of parameters and variables.

Parameters and
random variables Distribution type

𝜎1, 𝜎3 Normal distribution with coefficients of variation of 0.02

d1 , d3 Normal distribution with standard deviation of 3
Frequency Uniform distribution with limits from 8 GHz to 12.5 GHZ
∘
𝜃 Uniform distribution with limits from 0 to 89
Polarization Uniform distribution for occurrence of TE or TM

The objective is to determine the design variable values that yield an acceptable/target pof
represented as ptf , ensuring that the shielding responses remain below a specified limit. In the
shielding design problem, the failure mode occurs when the SE in decibels (dB) falls short of
the required SE (dB) due to variability in the shielding manufacturing and the randomness
in the material properties. Therefore, it is assumed that the thickness and conductivity of the
third and first layers within the three-layered shielding structure adhere to independent normal
distributions. Other external random parameters, such as polarization, frequency, and incidence
angle, are presumed to conform to uniform distributions. A summary of the probabilistic attributes
representing the variability of these parameters and variables is given in Table 8.1.
When setting ptf equal to 0.001, we conducted one million MCS cycles to collect the PSF dataset
for 500 designs. According to Eq. (8.7), the calculated percentage error, based on one million sim-
ulation cycles, is 6.29%. Consequently, there is a 95% likelihood that the estimated ptf , obtained
through these one million MCS cycles, will lie within the range of 1 × 10−3 ± 6.3214 × 10−5 .
The dataset, which comprises 500 design points along with their corresponding PSF, is approxi-
mated using a two-layered multilayer perceptron neural network. To divide the dataset for training
and evaluation, we employed a random split, allocating 70% of the data for training, 15% for valida-
tion, and another 15% for testing purposes. The training of the neural network was carried out in
batch mode, where all training set inputs were applied before updating the weights. This method
accelerates the training procedure while also resulting in a reduced error compared to incremental
training. For the ANN training, we opted for the Levenberg-Marquardt algorithm, which is known
for its speed and stability. Subsequently, we identified a network configuration with 13 neurons in
the hidden layer that demonstrated the best validation performance and correlation coefficient. The
MSE calculated for this network reached 1.9081 × 10−6 at epoch 125, as illustrated in Figure 8.3a.
Figure 8.3b depicts the R-value. Notably, the training process appeared optimal, as there was no sig-
nificant increase in the test curve before the validation curve started to rise, indicating the absence
of overfitting.
The design problem described in Eq. (8.11) with its associated constraints can now be formulated
as follows:
f (𝜎1 , d1 , 𝜎3 , d3 ) = Minimize|(1 − PSFa )|
subject to d1 + d3 = 500, 30 ≤ 𝜎1 ≤ 10,000,
30 ≤ 𝜎3 ≤ 10,000, 10 ≤ d1 ≤ 300,
10 ≤ d3 ≤ 300, d2 = 125,
𝜀1 = 0, 𝜀2 = 3.1,
𝜀3 = 0, 𝜎2 = 0, 𝜇1 , 𝜇2 , 𝜇3 = 𝜇0 .
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8.4 Conclusion 137

10–3

Output ~= 1*Target + –4.1e–005

Data
Train 1
Mean squared error (MSE)

Fit
Validation
Y=T
Test 0.8
10–4 Best All: R = 0.99999
0.6

10–5 0.4

0.2
10–6
0 50 100 0.2 0.4 0.6 0.8 1
Number of epochs Target
(a) (b)

Figure 8.3 (a) Learning curves and (b) regression plot of all design points of the data set.

Table 8.2 Reliability-based design parameters for EMI shielding structure.

Acceptable/
Material parameters target pof Achieved pof SE (dB)

𝜎 1 = 7815.8408, d1 = 252.3906 1 × 10−3 1.0406 × 10−3 107.1588

𝜎 3 = 8419.1538, d3 = 247.6144

In order to streamline optimization process, we utilized a real-coded GA with specific parameter

settings: with a population size of N = 100, a tournament size of T n = 4, crossover probability of
pc = 0.899, mutation probability of pm = 0.01, and maximum number of generations Gmax = 100.
The outcome of this optimization is displayed in the first column of Table 8.2, which includes both
an acceptable probability of failure and the actual achieved probability. The fourth column presents
the relevant SE when considering electromagnetic (EM) waves at an 8 GHz frequency, assuming
a perfectly optimized design point and excluding any uncertainty. Notably, considering the uncer-
tainty in the parameters and design variables, this analysis indicates that there is a requirement for
an extra margin of around 27.16 dB in the practical shielding design based on the specified scenario
and distribution parameters of the random variables.

8.4 Conclusion

In this chapter, an RBDO approach has been presented and demonstrated using cases related
to the design of shielding structures utilized in military applications. Utilizing RBDO in the
design of these structures allows for the deliberate inclusion of uncertainties originating from
both the process of manufacturing and external random variable’s impact. The methodology
commenced through the generation of the design points within design space utilizing LHS and
computed the PSF through the MCS approach. The PSF served as an additional safety measure,
enhancing the accuracy of probability calculations in regions of low failure likelihood. To approx-
imate the PSF values associated with LHS-generated design points, Artificial Neural Networks are
employed. Although constructing the deterministic response database and training the ANN can
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
138 8 Genetic Algorithm and Artificial Neural Networks in Reliability-Based Design Optimization

be time-intensive, one of the significant advantages is that the predicted PSF values align closely
with acceptable tolerances and can be generated efficiently with minimal computational resources
during the design optimization phase. The design parameters are used to estimate the pof , and
this estimated probability is then compared to the desired or target pof , showing that the design
successfully aligns with the specified requirements.

Appendix 8.A
Evaluation of Electromagnetic Shielding Effectiveness
Now, let’s contemplate a scenario where a plane wave is propagating in the z-direction and impinges
upon a multilayer planar dielectric medium with varying thicknesses represented by d1 , d2 , … , dn .
This occurs at an arbitrary incident angle 𝜃 0 , which implies the wave is incident at an angle other
than normal (i.e., oblique incidence). The boundaries between these layers are denoted as Z(0),
Z(1), … , Z(n), and this arrangement is depicted in Figure 8.A.1 for reference.
Each individual medium is assumed to exhibit homogeneity and isotropy, characterized by spe-
cific constitutive parameters for each layer, denoted as permittivity (𝜀k ), permeability (𝜇 k ), and
thickness (dk ). Notably, the 0th and (n + 1)th regions are considered semi-infinite and may possess
distinct properties. The plane wave fields within each layer are decomposed into two polarization
types: transverse electric (TE) and transverse magnetic (TM). It is important to emphasize that the
expressions for transmission coefficients vary depending on the polarization type of the electromag-
netic waves under consideration (Schulz et al. 1988). Assessing the shielding effectiveness under
normal incidence conditions (i.e., when 𝜃 0 = 0) is simply a specific instance within the broader
framework of oblique incidence (Oraizi and Afsahi 2007).
The characteristics of the kth layer, including permittivity, wave number (kk ), propagation con-
stant (𝛾 k ), and characteristic impedances (𝜂 kTE and 𝜂 kTM ), are provided as follows:
𝜀k = 𝜀′k − j𝜀′′k (8.A.1)
√
kk = 𝜔 𝜇k 𝜀k = 𝛽k − j𝛼k (8.A.2)
where 𝜔 = 2𝜋f and f is the frequency of incident wave in hertz (Hz). The real and imaginary part
𝜎
of the kth layer permittivity is denoted by 𝜀′ k and 𝜀′′ k , respectively, where 𝜀′′k = 𝜔𝜀k , 𝜎 k is the elec-
0
trical conductivity and 𝜀0 is the permittivity of free space (8.854 × 10−12 farads per meter (Fm−1 )).
Permeability is denoted by 𝜇 k = 𝜇 ′ k − j𝜇 ′′ k , where 𝜇 ′ k and 𝜇 ′′ k is the real and imaginary part of
the kth layer permeability, respectively. In the medium left to the shield, the wavenumber is real

Region 0 Region 1 Region 2 Region k Region n–1 Region n Region n+1

μ0, ε0
Transmitted wave
Reflected wave
d1 d2 dk dn–1 dn
Air
θ0
μ1, ε1 μ2, ε2 μk, εk μn–1, εn–1 μn, εn μn+1, εn+1
x
Incident wave Free space
Z(0) Z(1) Z(2) Z(k–1) Z(k) Z(n–2) Z(n–1) Z(n)
y z

Figure 8.A.1 Oblique incidence of plane wave on a planar multilayer shield.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 139

√
with the components kx = k0 sin 𝜃 0 , kz = k0 cos 𝜃 0 with k0 = 𝜔 𝜇0 𝜀0 where 𝜇 0 is the permeabil-
ity of the free space −7 −1
√ (4𝜋 × 10 Henries per meter (Hm )) and in the lossy medium, kkx = kx and
kkz = 𝛽 kz− j𝛼 kz = kk 2 − kkx 2 :
𝛾k = jkkz = 𝛼kz − j𝛽kz (8.A.3)
where 𝛼 kz is the attenuation constant and is measured in nepers per meter (Np/m), whereas 𝛽 kz is
the phase constant and is measured in radians per meter (rad/m):
𝜂kTE = (𝜔𝜇k ∕kkz ), 𝜂kTM = (kkz ∕𝜔𝜇k ) (8.A.4)
For the multilayer shield input impedance seen at the left boundary of the kth layer toward the
right side is given as:
Z(dk ) cos h(𝛾k dk ) + 𝜂k( TE ) sin h(𝛾k dk )
TM
Zin (dk ) = 𝜂k(TE∕TM) , k = 1, 2, … , n (8.A.5)
Z(dk ) sin h(𝛾k dk ) + 𝜂k(TE∕TM) cos h(𝛾k dk )
The transmission coefficient, which is defined as the ratio of the transmitted field to the incident
field, is provided as follows:
∏n
exp[−dk (𝛾k − 𝛾0 )]
Tm = Pm (8.A.6)
k=1
[1 − qk exp(−2𝛾k dk )]
where m represents either E or H field and
2n+1 𝜂1(TE∕TM) 𝜂2(TE∕TM) … 𝜂n(TE∕TM) ZL
PE = ,
(𝜂0 + 𝜂1(TE∕TM) )(𝜂1(TE∕TM) + 𝜂2(TE∕TM) ) … (𝜂n(TE∕TM) + ZL )
2n+1 𝜂0 𝜂1(TE∕TM) … 𝜂n(TE∕TM)
PH = ,
(𝜂0 + 𝜂1(TE∕TM) )(𝜂1(TE∕TM) + 𝜂2(TE∕TM) ) … (𝜂n(TE∕TM) + ZL )
(𝜂k(TE∕TM) − 𝜂(k−1)(TE∕TM) )(𝜂k(TE∕TM) − Z(dk ))
qk = , for k = 1, 2, … , n.
(𝜂k(TE∕TM) + 𝜂(k−1)(TE∕TM) )(𝜂k(TE∕TM) + Z(dk ))
Using Eq. (9.17), the total SE is determined as:
Sm = −20 log10 ∣Tm ∣ . (8.A.7)

References

Birger, I.A. (1970). Safety Factors and Diagnostics, Problems of Mechanics of Solid Bodies, 71–82.
Leningrad: Sudostroenve (in Russian).
Cardoso, J.B., de Almeida, J.R., Dias, J.M., and Coelho, P.G. (2008). Structural reliability analysis using
Monte Carlo simulation and neural network. Advances in Engineering Software 39 (6): 505–513.
Chau, K.W. (2007). Reliability and performance-based design by artificial neural network. Advances in
Engineering Software 38 (3): 145–149.
Deb, K. (2000). An efficient constraint handling method for the genetic algorithm. Computer Methods
in Applied Mechanics and Engineering 186 (2–4): 311–338.
Deb, K., Karthik, S., and Tatsuya, O. (2007). Self-adaptive simulated binary crossover for real-parameter
optimization. Proceedings of the Genetic and Evolutionary Computation Conference (GECCO-2007),
UCL London, 1187–1194.
Deb, K., Gupta, S., Daum, D. et al. (2009). Reliability based optimization using evolutionary algorithms.
IEEE Transaction on Evolutionary Computation 13 (5): 1054–1074.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
140 8 Genetic Algorithm and Artificial Neural Networks in Reliability-Based Design Optimization

Du, X. and Chen, W. (2001). A most probable point-based method for efficient uncertainty analysis.
Journal of Design and Manufacturing Automation 4 (1): 47–66.
Foschi, R.O., Li, H., and Zhang, J. (2002). Reliability and performance-based design: a computational
approach and applications. Structural Safety 24 (2–4): 205–218.
Frydrýšek, K. (2008). Performance-based design applied for a beam subjected to combined stress.
Annals of the Faculty of Engineering Hunedoara-International Journal of Engineering Fascicule 2
(Tome VI): 129–134.
Gargama, H., Chaturvedi, S.K., and Thakur, A.K. (2014a). Reliability-based design optimization of
electromagnetic shielding structure using neural networks and real-coded genetic algorithm.
Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering
Science 228 (18): 3471–3481.
Gargama, H., Chaturvedi, S.K., and Thakur, A.K. (2014b). Reliability-based design optimization
scheme for designing electromagnetic shielding structures. Journal of Electromagnetic Waves and
Applications 28 (6): 765–776.
Gondal, Z.R. and Lee, J. (2012). Reliability assessment using feed-forward neural network-based
approximate meta-models. Proceedings of the Institution of Mechanical Engineers, Part O: Journal of
Risk and Reliability 226 (5): 448–454.
Goldberg, D.E., Deb, K., and Clark, J.H. (1992). Genetic algorithms, noise, and the sizing of
populations. Complex Systems 6: 333–362.
Haldar, A. and Mahadevan, S. (2000). Reliability Assessment Using Stochastic Finite Element Analysis.
New York: John Wiley & Sons, Inc.
Hornik, K. (1991). Approximation capabilities of multilayer feedforward networks. Neural Networks 4
(2): 251–257.
Jensen, H., Valdebenito, M., Schuëller, G., and Kusanovic, D. (2009). Reliability-based optimization of
stochastic systems using line search. Computer Methods in Applied Mechanics and Engineering 198
(49–52): 3915–3924.
Kaymaz, I. and McMohan, C.A. (2005). A response surface method based on weighted regression for
structural reliability analysis. Probabilistic Engineering Mechanics 20 (1): 11–17.
McKay, M.D., Bechman, R.J., and Conover, W.J. (1979). A comparison of three methods for selecting
values of input variables in the analysis of output from a computer code. Technometrics 21 (2):
239–245.
Myers, R.H., Montgomery, D.C., and Anderson-Cook, C.M. (2009). Response Surface Methodology:
Process and Product Optimization Using Designed Experiments, 3e. Hoboken, New Jersey: John Wiley
& Sons, Inc.
Oraizi, H. and Afsahi, M. (2007). Analysis of planar dielectric multilayers as FSS by transmission line
transfer matrix method (TLTMM). Progress in Electromagnetics Research 74: 217–240.
Qu, X. and Haftka, R.T. (2003). Deterministic and reliability-based optimization of composite laminates
for cryogenic environments. AIAA Journal 41 (10): 2029–2036.
Rao, S.S. (1992). Reliability-Based Design. New York: McGraw-Hill.
Rahman, S. and Wei, D. (2008). Design sensitivity and reliability-based structural optimization by
univariate decomposition. Structural and Multidisciplinary Optimization 35 (3): 245–261.
Schulz, R.B., Plantz, V.C., and Brush, D.R. (1988). Shielding theory in practice. IEEE Transactions on
Electromagnetic Compatibility 30 (3): 187–201.
Tsai, Y.-T. and Chang, H.-C. (2008). Reliability-based optimum design for mechanical problems using
genetic algorithms. Proceedings of the Institution of Mechanical Engineers, Part C: Journal of
Mechanical Engineering Science 222: 1791–1799.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 141

Valdebenito, M.A. and Schuëller, G.I. (2010). A survey on approaches for reliability-based
optimization. Structural and Multidisciplinary Optimization 42 (5): 645–663.
Vukovć, S. and Sopta, L. (1999). Binary-coded and real-coded genetic algorithm in pipeline flow
optimization. Mathematical Communications 4: 35–42.
Wang, G.G. and Shan, S. (2007). Review of metamodeling techniques in support of engineering design
optimization. Journal of Mechanical Design 129 (4): 370–380.
Yoo, K.-S., Eom, Y.-S., Park, J.-Y. et al. (2011). Reliability-based topology optimization using successive
standard response surface method. Finite Elements in Analysis and Design 47 (7): 843–849.
Youn, B.D. and Choi, K.K. (2004a). Selecting probabilistic approaches for reliability-based design
optimization. AIAA Journal 42 (1): 124–131.
Youn, B.D. and Choi, K.K. (2004b). An investigation of nonlinearity of reliability-based design
optimization approaches. Journal of Mechanical Design 126 (3): 403–411.
Yu, X., Chang, K.H., and Choi, K.K. (1998). Probabilistic structural durability prediction. AIAA Journal
36 (4): 628–637.
Zhu, P., Zhang, Y., and Chen, G. (2011). Metamodeling development for reliability-based design
optimization of automotive body structure. Computers in Industry 62 (7): 729–741.
Zuev, K. (2009). Advanced stochastic simulation methods for solving high-dimensional reliability
problems. Ph.D. thesis, The Hong Kong University of Science and Technology, http://hdl.handle
.net/1783.1/5947.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
143

Parametric Estimation Models for Minimal and Imperfect Maintenance

Rajiv Nandan Rai, Sanjay Kumar Chaturvedi, and Heeralal Gargama
Subir Chowdhury School of Quality and Reliability, Indian Institute of Technology, Kharagpur, Kharagpur, West Bengal, India

9.1 Introduction

Maintenance encompasses a series of activities aimed at ensuring the functionality and operational
condition of equipment, systems, or assets. These tasks involve inspections, repairs, replacements,
and upgrades, all with the goal of preventing failures, reducing downtime, and prolonging the
lifespan of the assets. To conduct reliability analysis on maintained systems, many industries have
established different facilities to maintain, repair, and overhaul (MRO) these systems. Before delv-
ing into the rest of the content of this chapter, the authors find it necessary to provide a brief
introduction to MRO and its basics, providing a fundamental understanding (Keith Mobley 2014).
This primer on MRO will facilitate comprehending all the reliability assessment and analysis tech-
niques discussed later in this chapter, specifically tailored for maintained systems. MRO entails
life cycle maintenance, encompassing preventive maintenance (PM) carried out on a routine basis,
plan maintenance when the system is out-of-service, and repairs in correction, overhaul or rebuilds
for equipment damage. Industries primarily utilize MRO for costly, long-lasting products or equip-
ment, such as massive machines used in manufacturing, generation of electrical power, infrastruc-
ture and marine boats, vehicles used in mass transit, and vehicles and systems used in military.
Over time, systems used in industries naturally deteriorated due to long-term use, continuous
use, and exposure to environment, causing harm due to different factors. System deterioration even-
tually leads to its failure, which can cause various issues related to their safety, damage, quality, and
machine downtime unplanned. In the past, maintenance was viewed as a challenging task man-
aged reactively after failures occurred. However, it is now widely recognized as a critical element
of asset management and a vital commercial function. Industries are increasingly realizing the
impact of well-planned maintenance interventions on productivity and reliability, resulting in a
rise in PM activities that better align with other industrial processes, like scheduling of production
and management of spare parts. For example, companies in the process and chemical sectors can
significantly enhance profitability by inhibiting unscheduled breakdowns.
The continuous automation of production processes and growing market competition have
heightened awareness of the importance of effective maintenance planning. The primary objective
of maintenance professionals is to keep infrastructure, facility provider equipments, and tools
in optimal and efficient operating condition, thus preventing unexpected downtime or failure of
equipment. This is where the system’s reparability and maintainability play an essential role. The
terms “maintenance” and “repair” are used interchangeably, but these differ in meaning in the
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
144 9 Parametric Estimation Models for Minimal and Imperfect Maintenance

business of asset management. Repairs need to do restorative work when any break or damage in
asset occurs, or it ceases to function properly. On the other hand, maintenance includes the task
of routine basis repair correction and prevention, which is performed on assets to extend their
life expectancy by avoiding damage. Examples of maintenance activities include routine cleaning,
grease trap maintenance, air conditioning unit upkeep, painting, and regular inspections.
An overhaul is a comprehensive procedure of maintenance conducted on industrial equipment
or machines with the aim of preserving its functionality. Typically, maintenance service providers
carry out machinery overhauls, and the frequency of overhauls can be agreed upon, often planned
annually for regular maintenance. Overhauls usually commence with a thorough inspection by
skilled maintenance personnel, monitoring the machine’s operation while in use.1 Subsequently,
disassembly of the equipment takes place for further inspection and subsequent stages of over-
hauling such as repairs/replacement. A trained technician proficiently examines the machine to
identify parts requiring replacement or repair. Based on issues either identified, damaged parts of
the machine are replaced or machine is kept fixed, making overhauling a more efficient alternative
to replace entire equipment. However, replacement of parts may entail excessive delay if they need
to be acquired from the manufacturer. After successfully substituting any necessary spare parts, the
equipment is reassembled. Reassembling followed by testing is a crucial step in proper functioning
of the equipment. Testing ensures that the reassembled equipment operates correctly, and if any
issues arise, the procedure may need to restart from the inspection stage.
Most intricate systems, such as automobiles, communication, civil and military aviation, medi-
cal, locomotives, and more, undergo repairs once they experience failures. When these systems are
put into operation, they are influenced by three main factors: operations, maintenance, and logis-
tics. These factors must be purposefully managed based on the owners’ commercial plans. Evalu-
ation of performance parameters, characteristics, and reliability under these conditions becomes a
significant concern. Key areas of interest include assessment of anticipated failures during system
operation and period of warranty, maintenance of reliability threshold at minimum and within a
specified interval, address of rate of wear and tear, identification of optimal replacement, determi-
nation of overhaul timings, and minimization of cost of complete life cycle.
Traditional process whether parametric or nonparametric for analysis of accelerated test data and
reliability life, relies on a sample randomly drawn from a single population, by assuming that the
data collected for reliability analysis is identically distributed (i.i.d) and independent. This assump-
tion may hold true for the first failure occurs in units, that are identical, stemming from the same
design process, same manufacturing process, and operating in similar environment. The life data
for such items include time taken in a single failure (or in a very first failure or suspension) when
items are reparable. Recent literature on reliability covers such analysis of such data in plenty,
where times of failure of equipment are modeled via appropriate life distributions. However, in
case of systems that are repairable, successive failures occur in a single system, which often violates
the assumption of independence and identical distribution. Consequently, methods of reliability
assessment and analysis based on statistical approaches for repairable and nonrepairable systems
differ.
To address the characteristics, of complex systems, which are repairable and drawn for their reli-
ability, a process is employed rather than a distribution. For systems, which are repairable, the
time spent until the next failure occurs depends on the distribution of life and probability of the
first failure and after the first failure, the impact of actions done to maintain a system is taken into
consideration. The Power Law Process (PLP) (Rigdon & Basu 2000) is a widely used model for this

1 Regular inspections and frequent checks play a crucial role in preventing significant damage in older and more
complex machines.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
9.2 Maintenance Actions on Maintained Systems 145

purpose. Moreover, if the time taken in the first failure follows the Weibull type of distribution,
then the power law model adequately captures the phenomenon of minimal repair and effectively
governs each failure that occurs successively. In other words, the very first failure is addressed
by the Weibull type of distribution, and subsequent failures are addressed by PLP for a repairable
system. Consequently, the PLP can be seen as a generalization of the Poisson process and an exten-
sion of the Weibull type of distribution. The PLP offers the advantage of computational simplicity
by providing practical solutions. Practical solutions so obtained are well accepted and adopted in
real-world applications.

9.2 Maintenance Actions on Maintained Systems

Maintenance actions are of utmost importance in the total technical life (TTL) of a repairable sys-
tem, significantly impacting its overall reliability, availability, downtime, and operational costs.
When we deal with the systems, which are repairable, it becomes crucial to comprehend different
actions of maintenance usually they perform. These maintenance actions can be broadly catego-
rized into two main groups, as shown in Figure 9.1.
Reactive or Corrective Maintenance (CM): Corrective (or unplanned or unscheduled) type
of maintenance usually perform tasks on repairable systems to restore it to operating condition
in smallest possible time from failure state. It involves repairing of those components, which are
responsible for failure of a system. CM (Jezdimir Knezevic 1997) is done generally in the following
three steps:
Detection of a Problem: first step is to locate the component failed and then to detect the problem.
Repairing of Faulty Component(s): next step is to perform repairing or replacing actions on
faulty part of a system.
Verification: Once the component(s) is (are) repaired, it is verified by checking whether its oper-
ation is operating successfully or not.

Maintenance

Reactive or corrective
Proactive maintenance
maintenance (CM)

On failure Preventive maintenance Predictive maintenance

(PM)

On-condition
Scheduled preventive
Overhaul monitoring
maintenance (SPM)

Time based Age based

Figure 9.1 Maintenance actions on maintained systems.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
146 9 Parametric Estimation Models for Minimal and Imperfect Maintenance

Proactive Maintenance (PM): Proactive maintenance is a maintenance approach focused on

detecting and addressing potential causes of equipment failure in advance, thus preventing such
failures from occurring. The primary objective of proactive maintenance is to enhance asset reli-
ability and minimize the chances of downtime. It is essential to recognize that wear and tear are
natural components of equipment lifecycles and categorized into two: preventive type of mainte-
nance (PM) and predictive type of maintenance.
Preventive Maintenance: Preventive maintenance (PM) (or scheduled maintenance) involves
carrying out regularly scheduled maintenance tasks to proactively avoid future unforeseen fail-
ures during system operation. In simple terms, it entails addressing issues before they lead to a
breakdown or malfunction. The tasks are carried out either on fixed time of operation or age of an
item. Maintenance actions are scheduled based on wear out profile and characteristics of failure.
Overhaul PM activities are important and necessary one to perform a task on systems, which are
possible to repair TTL. In overhauling, we disassemble the whole system, inspect each component
and again reassemble the system. Overhauling rejuvenates the system and brings the system close
to as good as new condition. It is called “capital maintenance” in industrial terminology.
Predictive Maintenance: Predictive maintenance is done check operational conditions
of equipment during servicing, enabling the estimation of optimal maintenance timing. This
approach offers cost-saving advantages over preventive maintenance based on time and performed
on routine basis, since the task of maintenance is performed at the time of need only. Thus, it is
called maintenance based on condition (CBM), executed based on estimations of an item’s degra-
dation state. However, CBM is recommended, mainly for critical systems, because it may not be
cost-justifiable or feasible to apply on all the systems. The primary goal of predictive maintenance
is to facilitate well-timed scheduling of corrective actions and prevent unexpected equipment
failures. The key advantage lies in obtaining the right information about the equipment’s current
condition, leading to prolonged equipment lifespan, increased safety of plant, reductions in
accidents due to environmental impacts, and optimized handling of spare parts. What sets both
the strategies apart is reliance of predictive type of maintenance on real-time condition data of
equipment rather than relying on mean or expected life statistics solely to predict the future
maintenance time. Machine learning (ML) approaches are commonly adopted for defining the
actual condition of the system and forecasting its future states.
Several essential components are necessary for the successful implementation of predictive
maintenance, including health-indicating parameter(s), sensors, their types, collection of data, its
preprocessing, fault detection, time-to-failure prediction, early fault detection, resource optimiza-
tion, and maintenance scheduling. Predictive maintenance has also emerged as a driving force for
enhancing productivity and achieving “just-in-time” manufacturing practices. In this chapter, our
focus is solely on the investigation of corrective actions of maintenance for further analysis.

9.3 Classifications of Imperfect Maintenance Categories

9.3.1 Perfect, Imperfect, and Minimal Repairs

A repairable system, after losing its functionality, is restored to a functional state, without any need
to replace complete system. The repairing extent depends on factors like how critical was the com-
ponent failed, what is the status of operation a system, risk indices etc. Consequently, the level
of repair of the system is decided by the management. Two extremes of the repair are perfectly
repaired and minimally repaired. If it is perfectly repaired, it gets restored to “as-good-as-new”
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
9.3 Classifications of Imperfect Maintenance Categories 147

(AGAN) equipment, like it is being replaced by brand new system. This type of repair is typically is
performed for highly critical components that could otherwise compromise the system’s operation,
safety, and the well-being of personnel working with it. On the other hand, a system is deemed min-
imally repaired, if it is restored to “as-bad-as-old” (ABAO) state. This type of repair is undertaken
when there is a pressing need for the system to function for a limited time or when the system is
scheduled for PM activities shortly or when it will be discarded soon.
Imperfect repair encompasses any repair that falls outside the categories of perfect and mini-
mal repairs. In everyday systems, the majority of repairs are imperfect, meaning that the system is
neither fully restored nor left in a bad state from “as-good-as-new” to “ABAO.”
Figure 9.2 (Rajiv et al. 2020) illustrates the distinctions between perfect repair, minimal repair,
and general repair. In the case of perfect repair, the system is brought back to an “AGAN” state,
indicated by a zero starting point on the scale of time, demonstrating that the degradation of per-
formance has been entirely reversed. On the other hand, minimal repair leaves the system’s age
unchanged after the repair action, meaning there is no restoration of life below its previous age.
Regarding general repair, it renews a portion of the system’s life, allowing it to function again but
not fully restoring it to an “ABAO” or “AGAN” state. Instead, the system is brought back to an
intermediate state between these two extremes.
Figure 9.3 (Rajiv et al. 2020) summarizes various techniques for repairable and nonrepairable
systems, respectively. Counting is said to be renewal process (RP) in which the intervals between
occurrences are independent stochastically and have the same distribution, which can be any
arbitrary life distribution. In an RP, the time between failures (TBF) can be characterized by a
single distribution, and the repair frequency remains constant. However, when the frequency of
repairs increases (indicating deteriorating systems) or decreases (indicating system improvement),
it leads to nonrenewal behavior, which in turn influences the corresponding maintenance costs.
In homogeneous Poisson process (HPP), the random variables are exponential and are independent
and identically distributed (i.i.d) On the other hand, in nonhomogeneous Poisson process (NHPP)

Figure 9.2 Types of repairs. Perfect repair

1
Degradation

0.5

0
0 0.2 0.4 0.6 0.8 1
Time
Minimal repair
1
Degradation

0.5

0
0 0.2 0.4 0.6 0.8 1
Time
General repair
1
Degradation

0.5

0
0 0.2 0.4 0.6 0.8 1
Time
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
148 9 Parametric Estimation Models for Minimal and Imperfect Maintenance

Failure data
analysis

Live data analysis Recurrent event data analysis

(for non-repairable items) (for repairable items)

Non- Parametric Non-parametric Parametric

parametric distributional analysis analysis
analysis analysis

Kaplan-Meier Normal Mean Perfect renewal

estimator distribution cumulative process (PRP)
function (MCF)

Median rank Exponential

estimator distribution General renewal
process (GRP)

Weibull
distribution
Non-homogenous
poisson process
(NHPP)
Other
distribution

Figure 9.3 Various techniques for reliability analysis.

(Rigdon & Basu 2000; Rajiv et al. 2020) the random variables are neither independent statistically
nor distributed identically.
The NHPP finds frequent application in modeling repairable systems that undergo minimal
repair. The generalized renewal process (GRP) provides a framework that encompasses both
extremes of repair, ranging from RP to NHPP. The GRP proves especially valuable in studying
the patterns of failure of a specific unit and comprehending the impact of repair actions on the
system’s age. An ideal example of a system that suits the GRP model is one that receives repairs
after a failure but is only partially rejuvenated, achieving neither an RP nor an NHPP condition.
In such cases, it is essential to exercise caution when analyzing the data, as overlooking the actual
behavior may underestimate or overestimate engineering metrics.
Analyzing failure-repair scenarios using parametric methods necessitates a certain level of sta-
tistical expertise, proficiency in solving equations of complex nature, and validate the assumptions
of distributions. Additionally, equations require an iterative approach or specialized software to
provide solutions rejecting analytical approach. Moreover, parametric type of approaches demand
computations and may not be easily understandable for beginners or individuals with average sta-
tistical knowledge.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
9.4 Parametric Reliability Estimation Models for Maintained Systems 149

Whether system is able to repair or not able to repair, analysis of events follow a path from
nonparametric to versatile models, with complete analysis on graphical ground. The selection of
appropriate method relies heavily on data available and specific questions we aim to address.

9.4 Parametric Reliability Estimation Models for Maintained

Systems

Having described the classification of maintenance categories in preceding section, we are going to
discuss now parametric methods for estimation of reliability estimation with different performance
parameters for the cases of perfectly, minimally and imperfectly done maintenances.
Modeling of reliability for the systems that are repairable involves employing various parametric
and nonparametric methods. Here, we are focusing on parametric methods primarily considering
CM. Classification of these methods of maintenance is done, based on the restoring capability of the
system to its operational condition, namely: perfect, minimal, and imperfect maintenance (Rigdon
& Basu 2000; Rajiv et al. 2020; Rai & Bolia 2014). Now, let’s discuss parametric approaches for all
categories of maintenance. In the literature, one can find several parametric approaches available
for each maintenance category. However, our focus lie on HPP, NHPP and GRP in subsequent
sections. In the upcoming section, essential fundamental definitions related to repairable systems
are introduced.

9.4.1 Definitions
Consider few terms and definitions related to repairable systems that are reproduced from (Rigdon
& Basu 2000; Rajiv et al. 2020).

(1) Point Process: A stochastic model used to possess randomness inherent that describes time
of occurrence of events. These occurrences are thought of as point on the time continuum.
In general, neither the time period lapse between occurrences are distributed identically nor it
is independent. Therefore, “occurrences in time” is actually a time over which system stayed
failed.
(2) Counting Random Variable: Let N(t) denotes failure numbers occurred in the interval
[0, t] when N has its argument an interval N (a, b], it will result number of failures within that
interval. Here, N is the counting random variable. Failure number within (a, b] is given by:
N (a, b] = N (b) − N(a) (9.1)
Suppose, N(t1 ) = k1 , N (t2 ) = k2 , … .N (tn ) = kn (9.2)

Note:
• For all value of k, If P(N(t, t + s] = k) is independent of t a point process shows stationary
increments.
• A point process shows independent increment if for all n and for all a1 < b1 ≤ a2 < b2 ≤ …
an ≤ bn , the random variables N (a1 , b1 ], N (a2 , b2 ], … . . . . , N (an , bn ], are independent.
( ( ] ( ] ) ∏n ( ( ] )
In other words, P N a1 , b1 = k1 , … .. , N an , bn = kn = i=1 P N ai , bi = ki .
(3) Mean Function of Point Process: It is the expected value of counting random variable N(t)
through time, i.e., the expected number of failures through time,

Λ(t) = E(N(t)) (9.3)

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
150 9 Parametric Estimation Models for Minimal and Imperfect Maintenance

(4) Rate of Occurrence of Failures: When Λ(t) is differentiable then ROCOF is defined as:

d
u(t) = Λ(t) (9.4)
dt

(5) Intensity Function: Probability of failure over a small interval divided by interval length is
called intensity function of a point process, i.e.,

P (N (t, t + Δt] ≥ 1)
u(t) = lim (9.5)
Δt→0 Δt
● The intensity and ROCOF functions are measures of reliability a repairable system is having.
Both the functions are equal provided that failures simultaneously do not occur. (The simulta-
neous failures occurs at discontinuous mean function Λ(t).)
● Intensity and hazard rate functions are of similar kind. Notwithstanding, The intensity function
is unconditional probability of a failure over a small interval of time divided by interval length.
Similarly, hazard function is conditional probability of occurrence of one failure over a small
interval, divided interval length. Further, for a wearing out system, probability of failure within
interval (t0 , t0 + Δt0 ) conditioned on survival past time t0 will be < than the probability of failure
in a later interval (t1 , t1 + Δt1 ) conditioned on survival past time t1 .
● Comparing hazard and intensity function, the hazard function shows the limit of conditional
probability; while the intensity function is not.
● The ROCOF (Rate of Occurrence of Failures) or intensity function of many repairable systems
exhibits a bathtub-shaped curve, resembling the hazard rate curve often associated with
non-repairable systems. Despite the similarity in shape, they are different in interpretations.
In Figure 9.4, we can observe intensity function of bathtub-shaped. In the initial phase of the
ROCOF curve or intensity function, the system is prone to experience infant mortality failures,
which occur due to deficiencies in design and poor quality of manufacturing. With rectification
of infant mortality failures, intensity function decreases gradually. Subsequently, the ROCOF
remains constant, indicating that only random failures would occur during a phase. Finally,
as the system ages, deterioration sets in, leading to a reduction in the TBF due to wear and tear
in the system during this phase.

u(t)

t
Early failures Constant failures Deterioration

Figure 9.4 Bathtub-shaped intensity function.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
9.4 Parametric Reliability Estimation Models for Maintained Systems 151

(6) Complete Intensity Function: Consider conditional probability for some models given the
failure history of process. Let t history of failure process through time t. which is represented
by a set of failure times {ti : i = 1, 2, … , N(t)}. therefore the intensity function is:
P(N (t, t + Δt] ≥ 1 ||t )
u(t) = lim (9.6)
Δt→0 Δt

9.4.2 Parametric Analysis Approaches

9.4.2.1 Renewal Process
The random events are modeled stochastically and represents RP. Assuming RP is the time between
the successive arrivals of events that are independent and distributed identically. If a maintainable
system in operation can be restored to AGAN state after each breakdown and if the probability den-
sity function governing the TBF remains consistent across consecutive failures, then the sequence
of failures can be classified as an RP. A special case is HPP with function u. In other words, if TBF,
X 1 , X 2 . . . . . are independent and distributed identically in an exponential way and randomly vary-
ing, then N(t) corresponds to an HPP with u. The number of failures that are expected in [0, t],
would be:
t
E[N(t)] = u dt = ut (9.7)
∫0
Given that the intensity function remains unchanged, the HPP is unsuitable for representing sys-
tems that degrade or enhance over time, necessitating careful consideration before its application.
In such cases, opting for a Poisson process characterized by a varying intensity function could serve
as a more suitable and effective alternative.

9.4.2.2 Nonhomogeneous Poisson Process

A Poisson process (NHPP) of nonhomogeneous nature is characterized by an intensity function
varying nature. Let N(t) be the cumulative number of failures observed in cumulative test time t,
and u(t), the failure intensity. Under the NHPP model, u(t)Δt is the probability of a failure occurring
over the interval [t, t + Δt] for small Δt. Thus the expected number of failures within interval [0, t]
is given by:
t
E[N(t)] = u(t)dt (9.8)
∫0
u(t) is approximated by Power Law Model, i.e.,

u(t) = a × b × tb−1 ; a > 0, b > 0 (9.9)

where a is scale and b is shape parameter. For b = 1, u(t) = a, a stable system. For b > 1, u(t) is increas-
ing and indicating that system is deteriorating whereas when b < 1, u(t), is decreasing indicating
that system is improving.
The power law model finds practical foundation in minimal repair and popular one. It is used to
model the situation when a failed system repairing is just enough to operate the system again. If the
time taken in first failure is distributed Weibull type, then each failure occur in succeeding stages
is modeled by PLP. So, PLP model represents an extension of Weibull distribution. The expected
number of failures then becomes:
t
E[N(t)] = u(t)dt = atb (9.10)
∫0
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
152 9 Parametric Estimation Models for Minimal and Imperfect Maintenance

P(T ≤ t | T > t1)

t1
t

Figure 9.5 Conditional probability of occurrence of failures.

The inter-arrival times between failures occur successively and follow a conditional distribution
of probability Weibull type. The arrival of ith failure up to (i − 1)th failure is conditional on the
cumulative operating time. The system retains the condition of ABAO after the (i − 1)th repair. Thus,
the repair process does not restore any added life to the component or system.
Model parameters, of conditional probability are as given by (refer Figure 9.5):
F(t) − F(t1 ) R(t)
P(T ≤ t ∣ T > t1 ) = =1− (9.11)
R(t1 ) R(t1 )
where F (⋅) is probability of failure and R (⋅) is reliability of a component at the respective time.
Thus,
[ ]
F(ti ) = 1 − exp a(ti−1 )b − a(ti )b (9.12)

By using above two equation the probability density function of the ith failure given that (i − 1)th
failure occurred at time ti − 1 can be obtained as:
[ { }]
f (ti ∣ ti−1 ) = abtib−1 exp −a (ti )b − (ti−1 )b (9.13)

NHPP model parameters can be estimated by method called maximum likelihood estimation
(MLE) given below. The likelihood function is defined as:
∏
n
L= f (ti ∣ ti−1 );
i=1

where n is number of failures.

The likelihood function is given by:

∗b
∏
n
L = an bn e−at tib−1 (9.14)
i=1

where,
{
∗ tn if the test is failure terminated
t =
t > tn if the test is time terminated
Natural logarithm both sides give of Eq. (9.14)
∑
n
ln L = n ln a + n ln b − at∗ b + (b − 1) ln ti (9.15)
i=1
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
9.5 NHPP: Illustrative Example 153

Differentiate (9.15) with respect to a and equated to zero,

𝜕(ln L) n
= − t∗ b = 0
𝜕a a
n
̂
a= b (9.16)
t∗
Similarly, taking first derivative of (9.15) with respect to b and equated to zero gives,

𝜕(ln L) n ∑ n
= − at∗ b ∗ ln t∗ + ln ti = 0
𝜕b b i=1
̂ n
b= ∑n (9.17)
n ln t∗ − i=1 ln ti
The above notion can be extended to a fleet of K identical systems then,
[ ]
∏K
∏
nl
nl nl −at∗ b b−1
L= a b e ti
l=1 i=1
∑
K
∑
K
∑ ∑
K nl
ln L = ln a nl + ln b nl − at∗ b + (b − 1) ln ti (9.18)
l=1 l=1 l=1 i=1

where, l = 1, 2, …K.
Differentiating Eq. (9.18) with respect to a and b and equated to zero, we get
∑K
nl
a = l=1 b
̂ (9.19)
K ∗ t∗
∑K
nl
̂
b = ∑ ∑ l=1 ( ∗ ) (9.20)
K nl
l=1 i=1
ln tt
i

9.5 NHPP: Illustrative Example

Example 9.1
Consider the simulated Data for K=3 Systems operated for T=2000 hours (Crow 1974) in Table 9.1.
Estimate the Intensity Function, Instantaneous MTBF, Cumulative MTBF using NHPP. Comment on
the results.
Solution:
Here, n = 34 and K = 3. Using Eqs. (9.19) and (9.20), the values obtained are:

b = 0.453, a = 0.36224, u(t) = 0.36224 × 1.56 × t−0.547 ,

1
Instantaneous MTBF = u(t)
, MTBF (t = 2000 hours) = 389.56 hours,

CMTBF = b × MTBF = 176.47 hours.

From Figure 9.6, it can be observed that the intensity function is decreasing with time (also evi-
dent from the value of b = 0.453 which is less than 1). Hence the systems are in infant mortality
period of their lifecycle and improving.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
154 9 Parametric Estimation Models for Minimal and Imperfect Maintenance

Table 9.1 Failure times in hours for three systems.

System #1 1.2, 55.6, 72.7, 111.9, 121.9, 303.6, 326.9, 1568.4, 1913.5
System #2 1.4, 35.0, 46.8, 65.9, 181.1, 712.6, 1005.7, 1029.9, 1675.7, 1787.5, 1867.0
System #3 0.3, 32.6, 33.4, 241.7, 396.2, 480.8, 588.9, 1043.9, 1136.1, 1288.1, 1408.1, 1439.4, 1604.8

Intensity function Figure 9.6 Intensity function curve for

0.4000 Example 9.1.

0.3000
u(t)

0.2000
0.1000
0.0000
0 500 1000 1500 2000 2500
TTF

Example 9.2
The aircraft cabin’s temperature is controlled by the air-conditioning system, which guarantees a
pleasant environment for both passengers and crew. This system can adjust the air temperature,
providing cooling or heating as necessary, depending on external factors and the desired comfort
level. The air-conditioning system works in tandem with the aircraft’s pressurization system, ensur-
ing proper cabin pressure at different altitudes to avoid discomfort and potential health concerns
for those on board. Table 9.2 provides the TBF data of 13 Aircraft Air-Conditioning Systems with
the tail number of aircraft mentioned in the first column of the Table 9.2 (Blischke & Prabhakar
Murthy 2000).
Estimate the Intensity Function, Instantaneous MTBF, Cumulative MTBF using NHPP. Com-
ment on the results for the following cases: (1) All failures, all systems (2) Selected failures, all
systems, and (3) Selected failures, main systems
Solution:
Case 1. All failures all Systems: Here n = 212, K = 13. Using Eqs. (9.19) and (10.20), the computed
values obtained are: b = 0.6633, a=0.0928, u(t) = 0.0928 × 0.6633 × t−0.3367 , instantaneous MTBF =
1
u(t)
, MTBF (t = 2422 hours) = 230.458 hours, CMTBF = 152.863 hours.
Figure 9.7, shows that the intensity function decreases with time as also evident from the value of
b = 0.6633 which is less than 1. Hence the Aircraft Air-Conditioning Systems are in infant mortality
period of their lifecycle.
Case 2. Selected failures (Excluding system failures occurrences at the same time), all Systems
Using Eqs. (9.19) and (9.20), the computed values obtained are: b = 0.8315, a=0.0209,
1
u(t) = 0.0209 × 0.8315 × t−0.1685 , instantaneous MTBF = u(t) , MTBF (t = 2422 hours) = 215.78 hours
and CMTBF = 179.42 hours and
Figure 9.8, shows that the intensity function decreases with time as also evident from the value
of b = 0.8315 which is less than 1. Hence the Aircraft Air-Conditioning Systems are still in infant
mortality period of their lifecycle.
Case 3. Selected failures, Main systems (Tail numbers 7908, 7909 and 7910).
Using Eqs. (9.19) and (9.20), the computed values obtained are: b=1.1, a=0.00458, u(t) = 0.00458
1
× 1.1 × t0.1 , Instantaneous MTBF = u(t) , MTBF (t = 2422 hours) = 91.90 hours and CMTBF =
92.81 hours.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
9.5 NHPP: Illustrative Example 155

Table 9.2 Time between failures of aircraft air-conditioning system.

7907 194, 15, 41, 29, 33, 181

7908 413, 427, 485, 522, 622, 687, 696, 865, 1312, 1496, 1532, 1733, 1851, 1885, 1916, 1934, 1952,
2019, 2076, 2138, 2145, 2167, 2201
7909 90, 100, 160, 346, 407, 456, 470, 494, 550, 570, 649, 733, 777, 836, 865, 983, 1008, 1164, 1474,
1550, 1576, 1620, 1643, 1705, 1835, 2043, 2113, 2214, 2422
7910 74, 131, 179, 208, 710, 722, 792, 813, 842, 1228, 1287, 1314, 1467, 1493, 1819
7911 55, 320, 56, 104, 220, 239, 47, 246, 176, 182, 33, 15, 104, 35
7912 23, 261, 87, 7, 120, 14, 62, 47, 225, 71, 246, 21, 42, 20, 5, 12, 120, 11, 3, 14, 71, 11, 14, 11, 16, 9, 1,
16, 52
7913 97, 51, 11, 4, 141, 18, 142, 68, 77, 80, 1, 16, 106, 206, 82, 54, 31, 216, 46, 111, 39, 63, 18, 191, 18,
163, 24
7914 50, 44, 102, 72, 22, 39, 3, 15, 197, 188, 79, 88, 46, 5, 5, 36, 22, 139, 210, 97, 30, 23, 13, 14
7915 359, 9, 12, 270, 603, 3, 104, 2, 438
7916 50, 254, 5, 283, 35, 12
7917 130, 493
8044 487, 18, 100, 7, 98, 5, 85, 91, 43, 230, 3, 130
8045 102, 209, 14, 57, 54, 32, 67, 59, 134, 152, 27, 14, 230, 66, 61, 34

Intensity function
0.07
0.06
0.05
0.04
u(t)

0.03
0.02
0.01
0
0 500 1000 1500 2000 2500 3000
TTF

Figure 9.7 Intensity function curve for Example 9.2, Case 1.

Intensity function
0.02
0.015
u(t)

0.01
0.005
0
0 500 1000 1500 2000 2500 3000
TTF

Figure 9.8 Intensity function curve for Example 9.2, Case 2.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
156 9 Parametric Estimation Models for Minimal and Imperfect Maintenance

Intensity function
0.012
0.01
0.008
u(t)

0.006
0.004
0.002
0
0 500 1000 1500 2000 2500 3000
TTF

Figure 9.9 Intensity function curve for Example 9.2, Case 3.

From Figure 9.9 it can be observed that when main systems are selected then the intensity func-
tion shows an increase with time as also evident from the value of b = 1.10 which is greater than 1.
Hence the Aircraft Air-Conditioning Systems are facing slight wear out.

9.6 Generalized Renewal Process

As per the outlined framework GRP is rooted in two overarching strategies for addressing repairs
that are imperfect : Arithmetic Reduction of Age (ARA) and Arithmetic Reduction of Intensity
(ARI). Within ARA models, the reparative impact is elucidated by presuming a decrease in the
actual system age, resulting in the attainment of a designated “virtual age.” Among these, Kijima’s
virtual age models, which rest on the ARA concept, hold prominence as widely referenced and effi-
cacious constructs in existing literature. Conversely, in the ARI approach, the reparative influence
is evaluated based on alterations in the failure intensity pre- and post-failure. A concise introduc-
tion to ARA models is presented in the subsequent section.

9.6.1 Arithmetic Reduction of Age Models

At the core of this category of models lies the concept that repairs have a rejuvenating effect on the
system. The actual age of the system is denoted by its operational time, t, while the virtual age is
established as a constructive function of the real age, potentially influenced by previous failures.
The foundational premise underpinning Kijima’s virtual age models (Kijima 1989; Kijima & Sumita
1986) stems from the notion that repair interventions contribute to a reduction in the system’s age.
To understand these models, consider observation time of repairable system from time t0 = 0 and
t1 , t2 … are failure times in success. The TBF is X n = tn − tn − 1 . Effectiveness of repair is accounted
by the virtual age concept in the following way. If repair effectiveness (RE) is q and V n represents
virtual age after nth repair of the system with V 0 = 0. Kijima - I model (Rajiv et al. 2020) assume
that the nth repair can remove the damage incurred only during the time between (n−1)th and nth
failure, yielding virtual age as:
Vn = Vn−1 + qXn (9.21)
Vi = Vi−1 + qXi (9.22)
∑
i
Vi = q Xj (9.23)
j=1
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
9.6 Generalized Renewal Process 157

where the distribution of X n is given by:

F(x + y) − F(y)
Pr{Xn < x ∣ Vn−1 = y} = (9.24)
1 − F(y)
However, in practice the nth repair action may also decrease all damage accumulated up to nth
failure, yielding the Kijima - II model for virtual age:

Vn = q(Vn−1 + Xn ) (9.25)

Vi = q(Vi−1 + Xi ) (9.26)

∑
i
Vi = qi−j+1 Xj (9.27)
j=1

where the distribution of X n is given by Eq. (9.24).

In this model, after repair q = 0 converges to AGAN condition and modeled by RP. If we assume
q = 1 means restored component to the same condition what it was before the repair, i.e., ABAO
condition and modeled as NHPP. The RE (i.e., q) is index represents effectiveness and repair quality.
If q within interval 0 < q < 1 represent a state of the system better than old condition of the system
but worse than new. For q > 1, condition of a system is worse than old. The assumptions taken in
Kijima models are given below:

● TTFF distribution is known and is estimated using Weibull distribution MLEs form data avail-
able.
● The failures is viewed as a point processes if repair time is negligible.

GRP model parameters are estimated using MLE. Approach given in Yanez et al. (2002) is widely
used for GRP parameter estimation currently and most widely. The inter-arrival of failures are
assumed to follow the Weibull distribution, and f (.) and F(.) of the time to ith failure is given by:
[ ] [ { }]
f ((ti ∣ ti−1 )) = ab(Vi−1 + Xi )b−1 × exp a (Vi−1 )b − (Vi−1 + Xi )b

[ { }]
F(ti ) = 1 − exp a (Vi−1 )b − (Vi−1 + Xi )b

In above equations V i could be either from Kijima - I or Kijima - II model.

The likelihood, log-likelihood and MLEs for the failure terminated (single and multiple
repairable systems) and time terminated (single and multiple repairable systems) cases for system
data are given in the following sub-sections for both Kijima - I and Kijima - II models.

9.6.1.1 Kijima - I Model

The likelihood, log-likelihood functions and MLEs for failure terminated single repairable system
data set are given as follows (Rajiv et al. 2020):

n ⎡
( i−1 )b−1 ⎡ ⎧( i−1 )b ⎫⎤
∏ ∑ ⎤
⎢ ⎥ ⎢ ⎪ ∑ b ⎪⎥
L= ab q Xj + Xi × exp ⎢a ⎨ q Xj − (Vi−1 + Xi ) ⎬⎥ (9.28)
⎢
i=1 ⎣
⎥ ⎢ ⎪ j=1 ⎪⎥
j=1 ⎦ ⎣ ⎩ ⎭⎦
And, by taking log on both sides of Eq. (9.28).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
158 9 Parametric Estimation Models for Minimal and Imperfect Maintenance

ln L = n log(b) + n log a + (b − 1)
( ) ⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤
∑
n
∑
i−1
∑n
⎢ ⎪ ∑ ∑ ⎪⎥
× log q Xj + Xi + ⎢a ⎨ q Xj − q Xj + Xi ⎬⎥ (9.29)
i=1 j=1 i=1 ⎢ ⎪ j=1 j=1 ⎪⎥
⎣ ⎩ ⎭⎦
To obtain failure terminated MLEs, differentiate (9.29) with respect to a, b and q and equate
to zero.
The likelihood, log-likelihood functions and MLEs for time terminated single repairable system
data set are given as follows:

n ⎡
( i−1 )b−1 ⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤
∏ ∑ ⎤ ∑ ∑
⎢ab q X + X ⎢ ⎪
⎥ × exp ⎢a ⎨ q X ⎪⎥
L= − q Xj + Xi ⎬⎥
⎢
i=1 ⎣
j i
⎥ ⎢ ⎪ j=1
j
⎪⎥
j=1 ⎦ ⎣ ⎩
j=1
⎭⎦
⎡ ⎧( ∑n )b ⎫⎤
⎢ ⎪ q j=1 Xj − ⎪⎥
× exp ⎢a ⎨( ∑n )b ⎬⎥ (9.30)
⎢ ⎪ T − tn + q j=1 Xj ⎪⎥
⎣ ⎩ ⎭⎦
Taking log on both sides of Eq. (9.30), we obtain
( i−1 )
∑n
∑ ∑
n
ln L = n log(b) + n log a + (b − 1) log q Xj + Xi +
i=1 j=1 i=1

⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤ ⎡ ⎧( n )b ( )b ⎫⎤
⎢ ⎪ ∑ ∑ ⎪⎥ ⎢ ⎪ ∑ ∑n
⎪⎥
× ⎢a ⎨ q Xj − q Xj + Xi ⎬⎥ + ⎢a ⎨ q Xj − T − tn + q Xj ⎬⎥
⎢ ⎪ j=1 j=1 ⎪⎥ ⎢ ⎪ j=1 j=1 ⎪⎥
⎣ ⎩ ⎭⎦ ⎣ ⎩ ⎭⎦
(9.31)
To obtain time terminated MLEs, differentiate the above logarithm Eq. (9.31) of the likelihood
function with respect to a, b, and q and equate to zero.
The likelihood, log-likelihood functions and MLEs for failure terminated multiple repairable sys-
tem data set are given as follows:
⎡ n ⎡ ( i−1 )b−1
⎤ ⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤⎤
∏K
⎢∏ l ∑ ⎢ ⎪ ∑ ∑ ⎪⎥⎥
L= ⎢ ⎢ ab q Xl,j + Xl,i ⎥ × exp ⎢a ⎨ q Xl,j − q Xl,j + Xl,i ⎬⎥⎥ (9.32)
⎢
l=1 ⎢ i=1 ⎣
⎥ ⎢ ⎪ j=1 ⎪⎥⎥
⎣
j=1 ⎦ ⎣ ⎩
j=1
⎭⎦⎦
where K is number of systems (l = 1, 2, … , K). Taking log on both sides of Eq. (9.32)
( i−1 )
∑
K
∑
K
∑ ∑
K nl
∑
ln L = nl log(b) + nl log a + (b − 1) log q Xl,j + Xl,i
l=1 l=1 l=1 i=1 j=1

⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤
∑ ∑⎢ ⎪ ∑
K nl
∑ ⎪⎥
+ ⎢a ⎨ q Xl,j − q Xl,j + Xl,i ⎬⎥ (9.33)
l=1 i=1 ⎢ ⎪ j=1 j=1 ⎪⎥
⎣ ⎩ ⎭⎦
To get failure terminated MLEs, differentiate the above logarithm of the likelihood function with
respect to a, b and q and equate to zero. The likelihood, log-likelihood functions and MLEs for time
terminated multiple repairable system data set are given as follows:
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
9.6 Generalized Renewal Process 159

⎡ n ⎡ ( i−1 )b−1
⎤ ⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤
∏K
⎢∏ l ∑ ⎢ ⎪ ∑ ∑ ⎪⎥
L= ⎢ ⎢ab q X + X ⎥ × exp ⎢a ⎨ q X − q Xl,j + Xl,i ⎬⎥
⎢
l=1 ⎢ i=1 ⎣
l,j l,i
⎥ ⎢ ⎪ j=1
l,j
⎪⎥
⎣
j=1 ⎦ ⎣ ⎩
j=1
⎭⎦
⎡ ⎧( ∑nl )b ⎫⎤⎤
⎢ ⎪ q j=1 Xl,j − ⎪⎥⎥
× exp ⎢a ⎨( ∑nl )b ⎬⎥⎥ (9.34)
⎢ ⎪ T − tl,n + q j=1 Xl,j ⎪⎥⎥
⎣ ⎩ ⎭⎦⎦

Taking log on both sides of Eq. (9.34), we obtain:

( i−1 )
∑
K
∑
K
∑ ∑
K nl
∑
ln L = nl log(b) + nl log a + (b − 1) log q Xl,j + Xl,i
l=1 l=1 l=1 i=1 j=1

⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤
∑ ∑⎢ ⎪ ∑
K nl
∑ ⎪⎥
+ ⎢a ⎨ q Xl,j − q Xl,j + Xl,i ⎬⎥
l=1 i=1 ⎢ ⎪ j=1 j=1 ⎪⎥
⎣ ⎩ ⎭⎦
⎡ ⎧( n )b ( )b ⎫⎤
∑K
⎢ ⎪ ∑ l ∑nl
⎪⎥
+ ⎢a ⎨ q Xl,j − T − tl,n + q Xl,j ⎬⎥ (9.35)
l=1 ⎢ ⎪ j=1 j=1 ⎪⎥
⎣ ⎩ ⎭⎦

To obtain time terminated MLEs, differentiate the above logarithm of the likelihood function
with respect to a, b and q, and equate to zero.

9.6.1.2 Kijima - II Model

The likelihood, log-likelihood functions and MLEs for failure terminated single repairable system
data set are given as follows (Rajiv et al. 2020):

n ⎡
( i−1 )b−1 ⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤
∏ ∑ ⎤
⎢ i−j ⎥ ⎢ ⎪ ∑ i−j ∑
i−j ⎪⎥
L= ab q Xj + Xi × exp ⎢a ⎨ q Xj − q Xj + Xi ⎬⎥ (9.36)
⎢
i=1 ⎣
⎥ ⎢ ⎪ j=1 ⎪⎥
j=1 ⎦ ⎣ ⎩
j=1
⎭⎦

Taking log on both sides of Eq. (9.36)

( i−1 )
∑
n
∑
i−j
ln L = n log(b) + n log a + (b − 1) log q Xj + Xi
i=1 j=1

⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤
∑n
⎢ ⎪ ∑ i−j ∑ ⎪⎥
i−j
+ ⎢a ⎨ q Xj − q Xj + Xi ⎬⎥ (9.37)
i=1 ⎢ ⎪ j=1 j=1 ⎪⎥
⎣ ⎩ ⎭⎦

To obtain failure terminated MLEs, differentiate the above logarithm of the likelihood function
with respect to a, b and q, and equate to zero.
The likelihood, log-likelihood functions and MLEs for time terminated single repairable system
data set are given as follows:
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
160 9 Parametric Estimation Models for Minimal and Imperfect Maintenance

n ⎡
( i−1 )b−1 ⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤
∏ ∑ ⎤ ∑ ∑
⎢ab ⎢ ⎪
⎥ × exp ⎢a ⎨ ⎪⎥
L= qi−j Xj + Xi qi−j Xj − qi−j Xj + Xi ⎬⎥
⎢
i=1 ⎣
⎥ ⎢ ⎪ j=1 ⎪⎥
j=1 ⎦ ⎣ ⎩
j=1
⎭⎦
⎡ ⎧(∑n n−j+1 )b ⎫⎤
⎢ ⎪ j=1 q Xj − ⎪⎥
× exp ⎢a ⎨( ∑n n−j+1 )b ⎬⎥ (9.38)
⎢ ⎪ T − tn + j=1 q Xj ⎪⎥
⎣ ⎩ ⎭⎦
Taking log on both sides of Eq. (9.38), we obtain:
( i−1 )
∑n
∑
i−j
ln L = n log(b) + n log a + (b − 1) log q Xj + Xi
i=1 j=1

⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤
∑n
⎢ ⎪ ∑ i−j ∑ ⎪⎥
i−j
+ ⎢a ⎨ q Xj − q Xj + Xi ⎬⎥
i=1 ⎢ ⎪ j=1 j=1 ⎪⎥
⎣ ⎩ ⎭⎦
⎡ ⎧( n )b ( )b ⎫⎤
⎢ ⎪ ∑ n−j+1 ∑n
n−j+1 ⎪⎥
+ ⎢a ⎨ q Xj − T − tn + q Xj ⎬⎥ (9.39)
⎢ ⎪ j=1 j=1 ⎪⎥
⎣ ⎩ ⎭⎦
To obtain time terminated MLEs, differentiate the above logarithm of the likelihood function
with respect to a, b and q, and equate to zero.
The likelihood, log-likelihood functions and MLEs for failure terminated multiple repairable sys-
tem data set are given as follows:

⎡ n ⎡ ( i−1 )b−1
⎤ ⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤⎤
∏K
⎢∏ l ∑ ⎢ ⎪ ∑ ∑ ⎪⎥⎥
L= ⎢ ⎢ab qi−j Xl,j + Xl,i ⎥ × exp ⎢a ⎨ qi−j Xl,j − qi−j Xl,j + Xl,i ⎬⎥⎥
⎢
l=1 ⎢ i=1 ⎣
⎥ ⎢ ⎪ j=1 ⎪⎥⎥
⎣
j=1 ⎦ ⎣ ⎩
j=1
⎭⎦⎦
(9.40)

where, K is number of systems (l = 1, 2, . . . . . , K). Taking log on both sides of Eqs. (9.40) and (9.40),
we obtain:
( i−1 )
∑
K
∑
K
∑ ∑
K nl
∑
i−j
ln L = nl log(b) + nl log a + (b − 1) log q Xl,j + Xl,i
l=1 l=1 l=1 i=1 j=1

⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤
∑ ∑⎢ ⎪ ∑
K nl
∑ ⎪⎥
i−j i−j
+ ⎢a ⎨ q Xl,j − q Xl,j + Xl,i ⎬⎥ (9.41)
l=1 i=1 ⎢ ⎪ j=1 j=1 ⎪⎥
⎣ ⎩ ⎭⎦
To obtain failure terminated MLEs, differentiate the above logarithm of the likelihood function
with respect to a, b and q, and equate to zero.
The likelihood, log-likelihood functions and MLEs for time terminated multiple repairable sys-
tem data set are given as follows:
K ⎡⎡ nl ⎡
( i−1 )b−1
∏ ∏ ∑ ⎤⎤
L= ⎢⎢ ⎢ab i−j
q Xl,j + Xl,i ⎥⎥
⎢⎢ ⎢
l=1 ⎣⎣ i=1 ⎣
⎥⎥
j=1 ⎦⎦
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
9.7 GRP: Illustrative Examples 161

⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤
⎢ ⎪ ∑ i−j ∑
i−j ⎪⎥
× exp ⎢a ⎨ q Xl,j − q Xl,j + Xl,i ⎬⎥
⎢ ⎪ j=1 j=1 ⎪⎥
⎣ ⎩ ⎭⎦
(
⎡ ⎧ ∑ni n−j+1 )b ⎫⎤⎤
⎢ ⎪ j=1
q Xl,j − ⎪⎥⎥
× exp ⎢a ⎨( ∑ni n−j+1 )b ⎬⎥⎥ (9.42)
⎢ ⎪ T − tl,n + j=1 q Xl,j ⎪⎥⎥
⎣ ⎩ ⎭⎦⎦
Taking log on both sides of Eq. (9.42), we obtain
( i−1 )
∑
K
∑
K
∑ ∑
K nl
∑
i−j
ln L = nl log(b) + nl log a + (b − 1) log q Xl,j + Xl,i
l=1 l=1 l=1 i=1 j=1

⎡ ⎧( i−1 )b ( i−1 )b ⎫⎤
∑ ∑⎢ ⎪ ∑
K nl
∑ ⎪⎥
i−j i−j
+ ⎢a ⎨ q Xl,j − q Xl,j + Xl,i ⎬⎥
l=1 i=1 ⎢ ⎪ j=1 j=1 ⎪⎥
⎣ ⎩ ⎭⎦
⎡ ⎧( n )b ( )b ⎫⎤
∑K
⎢ ⎪ ∑ i ∑ ni
⎪⎥
n−j+1 n−j+1
+ ⎢a ⎨ q Xl,j − T − tl,n + q Xl,j ⎬⎥ (9.43)
l=1 ⎢ ⎪ j=1 j=1 ⎪⎥
⎣ ⎩ ⎭⎦
To obtain time terminated MLEs, differentiate the above logarithm of the likelihood function
with respect to a, b and q, and equate to zero.
Note: The MLE equations so obtained from Kijima - I and Kijima - II models cannot be easily solved
as these are complex and exhibit nonlinearity. These can be solved in MATLAB or alike. An easier
way to obtain estimators is by maximizing the log-likelihood functions. There are various methods to
maximize the objective function and any off the shelf software can be used for this.

9.6.1.3 Virtual Age based Reliability Metrics

Transforming reliability metrics from the actual time domain to a virtual time domain greatly sim-
plifies mathematical computations when estimating parameters of reliability. The virtual time scale
is converted back to the original real-time scale. Using virtual time scale all the equations for MTBF,
reliability, availability, intensity function, and expected number of failures (Rajiv et al. 2020; Rai &
Bolia 2014) are formulated and presented in Table 9.3 for ready reference.

9.7 GRP: Illustrative Examples

Example 9.3
Military aircraft utilize aero engines with formidable thrust capabilities to facilitate rapid ascent
and withstand substantial “G” forces during maneuvers. These engines are meticulously engi-
neered to avert surge and stall phenomena induced by the back pressure generated by rocket and
missile launches, which create considerable turbulence ahead of the engine. These power plants
endure extensive aerodynamic and thermal strains, thus rendering them susceptible to recurrent
failures. The recorded failure times in hours for one such aero engine, with a time interval between
overhauls set at 550 hours, are detailed below (Rajiv et al. 2020):
203, 477, 318, 536, 494, 213, 303, 525, 345, 299, 154, 230, 132, 321, 123, 351, 188, 49. 02, 267, 548,
380, 61, 160, 375, 550, 174, 176, 257, 102, 81, 541, 518, 533, 547, 299, 208, 326, 451, 349, 152, 509,
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
162 9 Parametric Estimation Models for Minimal and Imperfect Maintenance

Table 9.3 Virtual age-based reliability metrics.

S. no. Parameter Equation

1 Intensity function u(vi ) = a × b × (vi )b−1 (9.44)

2 Mean time between 1

MTBF(vi ) = (9.45)
failures (MTBF) u(Vi )
3 Reliability For first failure,
{ }
R(v1 ) = exp −a(v1 )b (Note: v1 = qt1 ) (9.46)

For subsequent failures,

R(vi + vi−1 ) [ ]
R (vi ∣vi−1 ) = = exp a (vi−1 )b − (vi + vi−1 )b (9.47)
R(vi−1 )
4 Availability MTBF(vi )
A(vi ) = (9.48)
MTBF(vi ) + MTTR

where MTTR stands for mean time to repair

5 Expected number of Vi
failures E[N(t)] = u(Vi )dV
∫0
i = 1, 2, … n (9.49)

249, 325, 261, 328, 48, 19, 142, 200, 426, 90, 522, 446, 338, 55, 549, 84, 342, 162, 250, 368, 96, 431, 14,
207, 324, 546.
The maintenance team’s task of determining if the aero engine experiences degradation within
a 550-hour span is of utmost importance. This assessment will also aid in evaluating the effective-
ness of the current PM strategy. Additionally, the team is keen on determining the characteristic
lifespan of the aero engine. Estimate repair effectiveness index, shape and scale parameters using
GRP Kijima - I model. Plot the intensity function curve. If MTTR = 528 hours of an aero engines.
Plot availability curve. What will be MTBF and availability of an aero engines at t = 550 hours?
Solution:
Maximizing Eq. (9.33), we obtain the following results:
a = 0.00022, b = 1.35, q = 0.75, MTTR = 528 hours (given)
Using Eqs. (9.44), (9.45) and (9.48), Intensity function, MTBF and availability can be estimated.
The intensity function equation from the values obtained works out to be
u(vi ) = 0.00022 × 1.35 × (vi )0.35
The intensity function curve is plotted at Figure 9.10.
MTBF (t = 550 hours) = 394.5602 hours
The availability curve is plotted at Figure 9.11. Availability (t = 550 hours) = 0.4277.

Example 9.4
For GRP application, let a system tested for T = 395.2 hours with 56 times failure as given in
Table 9.4 below. The first failure is recorded at 0.7 hours, the second failure is recorded 3 hours
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
9.7 GRP: Illustrative Examples 163

Intensity Function
Intensity function

0.003
0.002
0.001
0
TTF

Figure 9.10 Intensity function plot for Example 9.3.

Availability

0.8
Availability

0.6
0.4
0.2
0
0 100 200 300 400 500 600
TTF

Figure 9.11 Availability plot for Example 9.3.

Table 9.4 Time to failure Data (hrs) of Example 9.4.

0.7 0.63 125 244 315 366

3.7 72 133 249 317 373
1 99 151 250 320 379
1 99 163 260 324 389
1 100 164 263 324 394
2 102 174 273 342 395.2
4 112 177 274 350
5 112 191 282 355
5 120 192 285 364
5 121 213 304 364

later at 3.7. The last failure occurred at 395.2 hours into the test and thereafter system was removed
from the test. This failure data is truncated. Find out values of a, b, and q Kijima MLEs (Rajiv et al.
2020).
Solution:
Here n = 56, K = 1, and T = 395.2 hours
Maximizing Eq. (9.33), we obtain the following results for Kijima - I model: b = 0.9372, a = 0.2061,
q = 1.0
Maximizing Eq. (9.37), we obtain the following results for Kijima - II model: b = 0.24725,
a = 0.89442, q = 0.93
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
164 9 Parametric Estimation Models for Minimal and Imperfect Maintenance

Table 9.5 Time to failure data (hrs) of Example 9.5.

System 1 System 2 System 3 System 4 System 5 System 6

2227.08 772.9542 900.9855 411.407 688.897 105.824

2733.229 1034.458 1289.95 1122.74 915.101
3524.214 3011.114 2689.878
5568.634 3121.458 3928.824
5886.165 3624.158 4328.317
5946.301 3758.296 4704.24
6018.219 5052.586
7202.724 5473.171

Example 9.5
Suppose K = 6 systems are observed during [0, T i ] hours = 1, … , k. That is, the data are time
truncated with T 1 = 8760, T 2 = 5000, T 3 = 6200, T 4 = 1300, T 5 = 2650, T 6 = 500. Failure data are
given in Table 9.5 below. This failure data set is from multiple repairable systems. Find out a, b and
q values for the given data set using Kijima MLEs (Rajiv et al. 2020).
Solution:
Here, K = 6, T 1 = 8760, T 2 = 5000, T 3 = 6200, T 4 = 1300, T 5 = 2650, T 6 = 500.
Maximizing Eq. (9.35), we obtain the following results for Kijima - I model: b = 1.238, a = 0.00018,
q = 0.10
Maximizing Eq. (9.43), we obtain the following results for Kijima - II model: b = 1.358,
a = 0.000068, q = 0.55

9.8 Conclusion

A repairable system has the potential to land in conditions beyond the extremes of “As Good As
New” (AGAN) and “As Bad As Old” (ABAO) after repair. These conditions include being better than
old yet not AGAN, being superior to new, and conversely, being worse than old. To enhance accu-
racy in analyses and predictions, the GRP emerges as a valuable tool for diminishing the modeling
uncertainty stemming from repair assumptions.
This chapter took a concise look at fundamental terminology linked to repairable systems, fol-
lowed by an exploration of modeling and analysis within the repairable systems context, utilizing
the GRP-based ARA concept through Kijima - I and Kijima - II virtual age models. It furnished
likelihood and log-likelihood functions, along with maximum likelihood estimators (MLEs) for
NHPP, Kijima - I, and Kijima - II virtual models, encompassing both failure and time-terminated
data for single and multiple repairable systems. Moreover, reliability metrics based on virtual age
were presented to streamline mathematical computations. To foster a better grasp of the content,
all models have been illustrated through solved examples. Additional resources on ARI models are
suggested for readers with a keen interest in the subject. Though this chapter is oriented mainly on
Kijima based ARA models and its literature that provide a broad spectrum of the available models
on imperfect repair (Rajiv et al. 2020).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 165

Practice Problems

9.1 A Regional transit bus of six-year-old experiences minimal repair upon failure. Which had
intensity function u(t) = 0.0464 t2.1 with t measured in years. Estimate following: Ans: (a)
21.56 years. (b) 2.37
(a) MTBF (instantaneous)
(b) Expected number of failures over the coming year.

9.2 Consider the TBF of a compressor presented in Table Problem 9.2 below: (a) Estimate the
shape and scale parameters. (b) Also estimate the REI. Use Kijima - I model for this failure
terminated compressor failure time’s data (Yanez et al. 2002). Ans: (a) 𝛽 = 1.62, 𝛼 = 3072 hours
(b) q = 0.70.
Table Problem 9.2 Time between failures for a compressor.

No. of Time between No. of Time between

failures failures (h) failures failures (h)

1 3456 13 360
2 1584 14 998
3 236 15 656
4 516 16 180
5 1820 17 244
6 452 18 1528
7 432 19 44
8 1264 20 3064
9 3072 21 324
10 384 22 1528
11 2448 23 348
12 32 24 336

References

Blischke, W.R. and Prabhakar Murthy, D.N. (2000). Reliability Modeling, Prediction and Optimization.
Canada: John Wiley & Sons.
Crow, L.H. (1974). Reliability analysis for complex, repairable systems. SIAM Reliability & Biometry
379–410.
Jezdimir Knezevic, D.R. (1997). Systems Maintainability. London: Chapman & Hall.
Keith Mobley, R. (2014). Maintenance Engineering Handbook. New York: Mc Graw Hill.
Kijima, M. (1989). Results for Repairable Systems with General Repair. Journal of Applied Probability
26 (1): 89–102.
Kijima, M. and Sumita, U. (1986). A useful generalization of renewal theory: counting processes
governed by non-negative markovian increments. Journal of Applied Probability 23 (1): 71–88.
Rai, R.N. and Bolia, N. (2014). Availability based optimal maintenance policies for repairable systems
in military aviation by identification of dominant failure modes. I Mech E Part O: Journal of Risk and
Reliability 228 (1): 52–61.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
166 9 Parametric Estimation Models for Minimal and Imperfect Maintenance

Rai, R.N. and Bolia, N. (2014). Availability based methodologies for optimal maintenance policies in
military aviation. International Journal of Performability Engineering 10 (6): 641–648.
Rajiv, N.R., Chaturvedi, S.K., and Bolia, N. (2020). Repairable Systems Reliability Analysis: A
Comprehensive Framework. USA: John Wiley & Sons.
Rigdon, S.E. and Basu, A.P. (2000). Statistical Methods for the Reliability of Repairable Systems. New
York: Wiley.
Yanez, M., Joglar, F., and Modarres, M. (2002). Generalized renewal process for analysis of repairable
systems with limited failure experience. Reliability Engineering and System Safety 77: 167–180.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
167

Reliability Analysis of Transmission Systems

Section 3
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
169

10

Transmission System Reliability Evaluation Including Security

Pushpendra Singh 1 , Rajesh Arya 2 , Lakhan Singh Titare 3 , Mohd. Tauseef Khan 1 , and Sharat
Chandra Choube 4
1
Department of Electrical Engineering, Rajkiya Engineering College, Banda, Uttar Pradesh, India
2
Department of Information Technology, Acropolis Institute of Technology and Research, Indore, Madhya Pradesh, India
3
Department of Electrical Engineering, Government Engineering College, Ujjain, Madhya Pradesh, India
4
Department of Electrical & Electronics Engineering, University Institute of Technology, RGPV, Bhopal, Madhya Pradesh, India

10.1 Introduction

The power system security analysis and monitoring plays a crucial role in modern energy
management systems (EMS). However, implementing this analysis in real time poses chal-
lenges for power system engineers. It is essential for operational staff of a power system to
be aware of any disruptions or contingencies that may lead to limit breaches and emergency
conditions in order to ensure safe operation. Security assessments provide alerts to system
operator about the both secure and unsecured characteristics of the operating states, partic-
ularly in the event of unexpected incidents. This allows operators to take appropriate control
or corrective actions within a timely manner. Conducting a comprehensive examination of
every potential contingency in real time is impractical due to time constraints. Contingency
selection is employed to identify the most critical scenarios that have the potential to disrupt
the system and their likelihood of occurrence. Two common techniques used for ranking
techniques and screening approaches include contingency selection. Ranking methodologies
involve assigning a rough severity ranking to contingencies based on performance index values.
Performance indicators are quickly evaluated and expressed in terms of network characteristics,
providing a relative measure of contingency severity. Screening techniques utilize approxima-
tions of network solutions, such as AC/DC load flow, distribution parameters, local solution
techniques, linearized load flow, etc., to identify critical situations. These techniques aim to
reduce the number of unforeseen circumstances by screening and identifying contingencies
that may cause significant system disruptions. By employing both ranking methodologies and
screening techniques, power system operators can prioritize their attention and resources on the
most critical contingencies, allowing for efficient and effective real-time security analysis and
monitoring.
Contingency evaluation is a critical task for power system planning and operation engineers.
During planning, contingency analysis helps assess the requirement for transmission expansion
based on the power system’s performance to meet increased demand or generation. Operational
analysis of contingencies enables system operators to maintain the power system at safe operat-
” standby components is shown in Figure 5.1
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
170 10 Transmission System Reliability Evaluation Including Security

that meets acceptable quality standards. The screening and rating of contingencies aim to identify
potential contingency that could leads to system instability or violations of voltage stability criteria.
Various techniques and methods have been proposed for voltage security analysis using perfor-
mance indicators (PI) (Vaahedi et al. 1999; Naik et al. 2015; Singh et al. 2021; Singh et al. 2023).
Wang et al. (2020) introduced an incremental-risk based technique for filtering significant con-
tingency events in transmission network operation and design. Choube et al. (2001) proposed a
technique based on singular value decomposition for contingency selection. Arya et al. (2002) uti-
lized the concept of vanishing eigenvalues to develop a strategy for contingency selection. Zhu and
Liu (2022) proposed a heuristic method based on the triangle count approach for selection of the
branch contingency during the screening phase. Heidari et al. (2022) proposed a utilizing clustering
and similarity, a complete method for online contingency screening, providing a real-time list of
potential outcomes that may lead to system instability. Coelho et al. (2022) suggested a topological
technique to test critical transmission lines with buses for various transmission line contingencies.
Amjady and Esmaili (2003) presented a method to effectively assess the status of voltage stability in
both the current situation and the post-contingency phase. Zhao and Singh (2018) developed a con-
tingency ranking technique as a pre-selection technique for reliability assessment in line switching
operations. Nan et al. (2021) offered an N-1 contingency selection approach based on graph theory
to efficiently detect the most important contingencies for security-constrained part commitment
decisions. Hagh et al. (2018) introduced a sequential technique considering N-1 contingency cri-
teria to assess system security under post-outage transient stability conditions. Armaghani et al.
(2020) presented an approach for online vulnerability assessments with N-1 line outage in the large
transmissions system, particularly relevant for the estimation of post-contingency phases involving
wind power resources. These techniques and methods contribute to the assessment and selection of
contingencies, enabling power system engineers to make informed decisions for secure and reliable
power system operation.
Operators of control rooms can make use of the technique Ni et al. (2003) provided for eval-
uating online security using probabilistic risk-based indices. Based on the Bayes classifier, Kim
and Singh (2005) developed a method for evaluating the security of power systems. Verma and
Shrivastava (2005) presented an effective method for formulating the voltage control area consid-
ering the impact of contingencies, including network topology changes, P-V, Q-coupling, and their
effects. Zarate et al. (2006) developed a fast calculation approach using nonlinear programming
methods to assess security margins against voltage collapse. Arya et al. (2006) introduced tech-
niques for assessing voltage collapse risk using ANN and managing voltage security margins in
advance. Srivani and Swarup (2008) proposed a method for static security evaluation using the
ward network. Berizzi et al. (2009) utilized fuzzy logic for online evaluation of voltage security at
the present operating point. Bahmanyar and Karami (2014) developed an voltage stability margins
may be monitored online using an ANN-based technique. Lee et al. (2010) introduced a new indi-
cator called the “stability region area for the voltage” that provides valuable information for load
shedding, contingency screening, and ranking. ANN-based approaches have received significant
attention in the field of security assessment (Javan et al. 2013) due to their ability to handle data in
parallel while maintaining high accuracy and speed. These techniques and methods contribute to
the analysis and assessment of voltage stability, enabling operators and engineers to monitor and
ensure the security of the power system in real-time.
Indeed, deterministic criteria used in power system planned construction often overlook the
inherent randomness and uncertainty present in various components such as site resources, sys-
tem behavior, and customer expectations. To address this limitation, probabilistic approaches can
be employed to account for the inherent uncertainty in these factors. However, the interpretation
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10.2 Problem Formulation 171

and utilization of probabilistic reliability indices can pose challenges for power system planners and
designers. One way to tackle this challenge is by incorporating deterministic factors into probabilis-
tic assessments using the concept of well-being. By considering deterministic elements alongside
probabilistic analysis, a more comprehensive understanding of system reliability and performance
can be achieved. This approach allows for a better evaluation of the impact of uncertainties on
system behavior and the overall well-being of the power system. By combining deterministic and
probabilistic considerations, power system planners and designers can make more informed deci-
sions, considering both the average system performance and the associated uncertainties. This
integrated approach helps address the limitations of purely deterministic criteria and provides a
more realistic and robust assessment of power system reliability and performance.
This book chapter introduces a novel technique for evaluating the security of a power system,
considering uncertainty in control variables and line parameters. The proposed techniques consist
of two main part. In the first part, a Monte Carlo simulation is utilized to determine the system’s
security index, considering both the availability and unavailability of transmission lines. This sim-
ulation helps in assessing the system’s security under different scenarios and provides a ranking
of lines based on their security index values. The second part focuses on evaluating the using the
cut-set method for the probabilistic insecurity index (PISI). This approach considers single and
double-line outages and examines the impact of these outages on voltage stability limits. Continu-
ous power flow analysis is utilized to determine the voltage stability limits when specific lines are
out of service. To further enhance the accuracy and applicability of the technique, a multi-layer
recurrent neural network (RNN) is trained using the obtained PISI values for various single and
double-line outages at different loading scenarios. This trained RNN can then be used to predict
the PISI for any given operating conditions, enabling online applications of the proposed technique.
By combining Monte Carlo simulation, cut-set analysis, and RNN modeling, this technique offers
a comprehensive approach to evaluate the security of power systems considering uncertainties.
It provides valuable insights for system operators in real-time decision-making and enhances the
overall security assessment process.

10.2 Problem Formulation

Real-time system operation in power systems involves dealing with uncertainties in control vari-
ables and system parameters, which can be considered as arbitrary variables. Transmission line
parameters, in particular, are prone to errors for various reasons. One of the reasons for inaccuracies
in computed transmission line parameters is the truncation of the Taylor series expansion, which is
a mathematical approximation commonly used in calculations. Simplified modeling assumptions,
such as assuming a flat earth, fully transposed lines, or rounded conductors, can also contribute to
inaccuracies. Human errors during the data input phase can lead to significant mistakes in param-
eter values. This can occur due to manual data management issues, where incorrect values are
entered into the system. Weather factors, such as temperature variations, can affect the sagging of
conductors and introduce additional uncertainties. Furthermore, frequency fluctuations can also
impact the accuracy of parameter values. Studies have shown that these mistakes in parameter
values can range from 5 to 10%. In cases where human data entry errors occur, the magnitude
of the mistakes can be even larger. However, in the field of system planning and operation, it is
generally acceptable for a few percentage errors to be present, considering the overall complexity
and uncertainties in the system. To account for parameter uncertainty, power systems are typically
designed with a high stability margin and a sufficiently low likelihood of voltage collapse. This
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
172 10 Transmission System Reliability Evaluation Including Security

ensures that even with the worst-case consequences of parameter uncertainty, the system remains
stable and secure. Overall, uncertainties and errors in control variables and system parameters are
inevitable in real-time power system operation. However, through proper system design, model-
ing techniques, and margin considerations, the impacts of these uncertainties can be mitigated to
maintain the power system stable and reliable.
The following are the failure probabilities taking into consideration the control variable and line
parameter uncertainties:
⎡ ⎤
⎢ n ⎥
⎢∑ ∑
n,m
⎥
Pf = ⎢ P(Csi ) − (P(Csi ) ∗ P(Csj ))⎥ (10.1)
⎢ i=1 i=1 ⎥
⎢ ⎥
⎣ j=2 ⎦
maintain the power system stable and reliable. The following limitations under the base case sce-
nario and the following anticipated loading scenario, which takes into account the ambiguities in
the control variables and line parameters, are applicable:
(i) Power flow restrictions:
F(V, 𝛿) = 0 (10.2)
(ii) Constraints on the generated reactive power:
Qkg,min ≤ Qkg ≤ Qkg,max (10.3)

k = 1, 2, … , NG
(iii) Constraints on the generated active power:
k
Pg,min ≤ Pgk ≤ Pg,max
k
(10.4)

k = 1, 2, … , NG
(iv) Constraints on load bus voltages:
i
Vmin ≤ V i ≤ Vmax
i
(10.5)

i = (NG + 1), … … , NB

10.3 Monte Carlo Simulation for Evaluation of the Security Index:

With and Without Considering the Absence of Transmission Lines

In power systems, various variables and parameters, such as PV-bus voltages, line charging sus-
ceptances, line resistances, and line reactances, can be considered as random variables due to their
inherent uncertainties. These uncertainties arise from factors such as measurement errors, model
inaccuracies, environmental conditions, and other sources of variability. Determining the static
voltage stability limits and a continuous power flow approach can be employed. This approach
involves evaluating the power flow equations for different sampled values of parameters and vari-
ables, considering their probabilistic nature. By considering a range of possible values for these
random variables, the static voltage stability limit is established as a random variable as well. The
following steps involve the implementation of Monte Carlo simulation:
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10.3 Monte Carlo Simulation for Evaluation of the Security Index: With and Without Considering the Absence 173

Step 1: Using the corresponding probability density functions, obtain the sample vector z ∈ Z. For
constant power factor of loads, the “Z” vector comprises resistances to the line, reactances
to the line, susceptance to line charging, and generated bus voltages.
Step 2: Perform step 1 repeatedly, saying “N” each time.
Step 3: Outage on a certain line, i = 1
where,
i = 1, 2, 3, … … , NL.
Step 4: With the use of the continuous power flow approach (Ajjarapu and Christy 1992), for each
sampled condition “Z” determine the static voltage stability limit.
Step 5: At step 4, all “N” static voltage stability requirements are met. Get the value of X(Z j ) for
j = 1, 2, 3, 4, ……, N, which indicates whether voltage collapse has occurred or not.
If the nominal load level (Sd ) for this sampled condition exceeds the static voltage stability
limit, then X(Z j ) = 1.
= 0, Otherwise
Step 6: Calculate the security index as follows without taking the absence of a transmission line
into account:
[N ]
∑
̂
pj = X(Zj ) ∕N (10.6)
j=1
Step 7: Calculate the security index as follows, accounting for the absence of a transmission line
SI
SIj = ̂
p∗j Aj (10.7)
Step 8: For various line outage scenarios that reoccur starting with step 5, get estimates for ̂
pj and
SI j .
Step 9: For “N” times, Steps 1 through 8 are repeated.
It is obvious that a lot of samples may be produced using steps 1 through 9 and utilized to rank
of the transmission lines.
The two-state models for transmission lines functioning are commonly use to represent the oper-
ational status of a transmission line. In this model, each transmission line is considered to have two
states: the down state and the up state. The up state represent the normal operating condition of the
transmission line, where it is available and capable of transmitting power. The transmission line
is considered to be functioning and operational in this state. On the other hand, the “down” state
represents a failure or planned outage of the transmission line. In this state, the transmission line
is unavailable and cannot be used for power transmission. The down state can be due to various
reasons such as equipment failure, maintenance, or other planned events. The transmission line
failure and repair rates are depicted in Figure 10.1 as 𝜆j and 𝜇 j , respectively. By analyzing the failure
and repair rates of the transmission line, system operators can assess the reliability and availabil-
ity of the transmission infrastructure. This information is crucial for planning and managing the
transmission network, ensuring its resilience and minimizing disruptions in power supply.
This model is employed to determine the long-term availability and unavailability of transmission
lines. They have the following expressions (Rubinstein 1981):
Aj = 𝜇j ∕(𝜆j + 𝜇j ), Aj = 𝜆j ∕(𝜆j + 𝜇j ) (10.8)

The letters Aj , and Aj stand for the transmission line’s availability and unavailability, respectively.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
174 10 Transmission System Reliability Evaluation Including Security

λj
State up State down
μj

Figure 10.1 For every transmission line, there is a two-state model.

10.4 Evaluation of the Load Flow’s Minimal Eigenvalue Jacobian

Load flow’s minimal eigenvalue a critical proximity indication is the Jacobian for evaluating the
voltage stability of a certain system. The system load has an inverse relationship with this indicator’s
value. Here is how Jacobian is rated:
[ ][ ] [ ]
H N Δ𝛿 ΔPg
= (10.9)
M L ΔV ΔQg
Equation (10.9) may be used to solve for the incremental power flow at each solution point.
[ ][ ] [ ]
H M Δ 𝛿 Δ P ′
= when the load flow is used Jacobian J is
N L Δ V Δ Q
[ ]
′ H N
J = (10.10)
M L
The following describes the voltage collapse condition (Kundur 1994; Singh et al. 2017; Kothari
et al. 2022).
𝜆min [H] ≠ 0
𝜆min [J ′ ] = 0
[ ]
H M
′
J = 𝜆 ≠ 0 𝜆min - is minimum eigen value of [H] or [J].
N L min [H]
th
For the i eigen value, the modal solution to the incremental power flow Eq. (10.9) may be stated
as follows.

ΔVm,i = 𝜉i ∕𝜆i (10.11)

𝜆i - is the ith eigenvalue of the Jacobian for load flow.

ΔV m,i is the change in the load bus voltage vector as a result of modal variation, and 𝜉 i is the right
eigen vector corresponding to 𝜆i .
∑
ΔVm,i = (𝛼i 𝜉i )∕𝜆i
i
∑ ∑ 𝛼i 𝜉i
ΔVi = ΔVm,i = (10.12)
i i
𝜆i

According to Eq. (10.12), all of the load flow Jacobian’s eigenvalues should be positive for the
system to retain a stable voltage. When the minimal eigenvalue is extremely near to zero, a very
little change in load can cause a very significant change in voltage. From the perspective of load
flow, the operating point is thought of as voltage unstable. Thus, the minimum eigen value’s size
must be big enough to provide a sufficient stability buffer. The steps below are used to determine
the load flow Jacobian’s minimum eigen value:

Step 1: With the continuous power flow method, perform AC power flow and get load flow Jaco-
′
bian [J ]
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10.6 Evaluation of the PSI and the Cut-set Approach 175

Step 2: To determine the inverse load flow Jacobian.

[J ′ I] = [J ′ ]−1
Step 3: Select E0 = [1, 0, ……, 0]T
Step 4: The iteration count is set to k = 1.

10.5 Evaluation of Schur’s Inequality

As stated in (Arya et al. 2012), Schur’s inequality is as follows:
√∑
𝜆max ≤ a2ij (10.13)
ij

′
where, aij - is the ijth load flow element [ J ] for Jacobian.
𝜆max -maximum load flow eigen value Jacobian.
The magnitude of the most significant eigenvalue is less than or equal to the square root of the
squares of each matrix component. The lower constraint on the lowest eigenvalue of the load flow
Jacobian is derived using Eq. (10.13). The following is the sensitivity matrix [S]
[S] = [J ′ ]−1 (10.14)
Now, the higher limit on the highest eigenvalue of [S] is provided using inequality Eq. (10.13) as
follows:
√∑
S𝜆max ≤ s2ij (10.15)
ij

where sij is the element of [S] and S𝝀max is the greatest eigenvalue of [S]. The matrix theory reveals
that:
J′ 𝝀min = 𝟏∕(S𝝀max ) (10.16)
The load flow Jacobian’s minimum eigenvalue is J 𝜆min . Inequality Eq. (10.15) may be stated as
′

follows using Eq. (10.16):

⎛√∑ ⎞
J ′ 𝜆min ≥ 1∕ ⎜ s2ij ⎟ = 𝜏 (10.17)
⎜ ⎟
⎝ ij ⎠
A proximity indicator (𝜏) and a lower limit on the lowest eigenvalue of the load flow Jacobian
are what the right-hand side of Eq. (10.17) actually represents. The size of this proximity indicator
represents the separation between voltage collapse and the measured operating point, which has
been used to verify voltage stability.

10.6 Evaluation of the PSI and the Cut-set Approach

An effective method for evaluating the reliability of power system is a cut-set method. The tech-
nique is easy to program on a digital computer and provides a quick and effective check of any
general grid. Cut-sets also help identify the unique and identifiable types of failures in a system, as
they are closely related to the types of system failures. In determining a cut set, one selects the line
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
176 10 Transmission System Reliability Evaluation Including Security

or group of lines that is most important. A minimal cut set is a group of transmission lines where if
one line fails, the entire system fails; however, as long as all components in the set are functioning
properly, the system is not affected. For a system to fail, every element of a minimum cut-set must
be in the failure state. A power network’s failure likelihood can be expressed as follows:

Pf = [P(Cs1 + Cs2 + … + Csn ) − P(Cs1 Cs2 + Cs1 Cs3 + … + Csn Csm )] (10.18)

where the minimal cut-sets Cs1 , Cs2 , . . . . , Csn , and Csm are. Although this exact evaluation is
theoretically always feasible, it is a laborious and time-consuming procedure that is prohibitively
expensive with power networks.
In order to get around this issue, failure probability evaluations typically make use of approxima-
tions, which slow down the assessment process but increase speed. Keeping the first-order in terms
within the expansion of Eq. (10.18), the failure probability may be roughly calculated as follows.
This is because the error added is typically small and is within the range of the statistics on the
availability of transmission lines for power systems, which have high repair rates and low failure
rates for each transmission line.
⎡ ⎤
⎢∑ n
∑
n,m ⎥
Pf = ⎢ P(Csi ) − (P(Csi ) ∗ P(Csj ))⎥ (10.19)
⎢ i=1 ⎥
⎢ i=1 ⎥
⎣ j=2 ⎦
i≠j

This chapter takes into account cut sets up to the second order. This is because there is very little
chance that more than two transmission lines will go down. Under typical loading conditions, a
cut-set of a higher order may not be a minimum cut-set, which is necessary in Eq. (10.18). In order
to calculate the probability of failure (Pf ) under the aforementioned approximation, Eq. (10.19)
is used.
When a given load is compared to the static voltage stability limit in the base case and extra
line loss situations, the lowest cut-set (first & second order) is determined. The continuing flow of
power approach for used to compute the static voltage stability limit during single or double-line
outages (Arya and Verma 1996). An acceptable stability margin must be maintained for proper sys-
tem operation. The results of ongoing power flow may be used to build a capacity table is available,
which offers permissive loads & matching network condition. The capacity availability for various
load levels may thus be used to calculate the cut-set. Equation (10.19) can be used to determine the
probability of failure.
The PISI of the transmission system is the name given to this failure likelihood. It is further
emphasized that system factors and control variables affect load ability limit. In order to account for
uncertainties, failure probability, or PISI, is determined. The following is the suggested algorithm
for computing PISI:

Step 1: Calculate the static voltage stability limit using the base case load flow method.
Step 2: To establish the stability limit for each dependent circumstance, for single and double-line
outages, execute continuous power flow.
Step 3: Plan for the possibility of capacity failure table displaying the stability limit and potential
scenarios.
Step 4: For the ith loading condition, determine the cut-sets for the different peak load considera-
tions: Cs1 (i), Cs2 (i), Cs3 (i), Cs4 (i), ……, Csr (i).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10.7 Recurrent Neural Network (RNN) Assessment of Probabilistic Insecurity 177

Recurrent network

Input layer Hidden layer Output layer

Figure 10.2 Recurrent neural network.

where

(i = 1, 2, 3, ..NL)

Step 5: Use the equation to get the probability of failure (Pf ), using relation (10.10).
Step 6: Stop.

10.7 Recurrent Neural Network (RNN) Assessment of Probabilistic

Insecurity

In a recurrent neural network (RNN), as depicted in Figure 10.2, the architecture includes a looped
design that allows the network to retain the information from the previous input. This recurrent
structure enables the RNN to process sequential data and utilize prior knowledge when making
predictions or decisions. The inputs of the RNN are the total system load, generator bus voltage,
and the line parameters (resistances, reactances, and susceptance). Line parameters are equal to
the number of units in the input layer. The output layer consists of a single unit that generates
the probabilistic insecurity index (PISI) as the result of the RNN’s computation. Figure 10.3 (Sher-
stinsky 2020) illustrates the structure of a conventional RNN, where each repeating module in the
network contains a single activation layer utilizing the hyperbolic tangent function. This activation
layer performs calculations based on the input data & previously learned information, enabling the
RNN to capture temporal dependencies and patterns in the sequential data. RNNs are commonly
employed in applications involving sequential data, such as voice recognition and time series fore-
casting, where the order and context of the data play a crucial role. The equations (Haykin 2003;
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
178 10 Transmission System Reliability Evaluation Including Security

ht

Xt

Figure 10.3 Repeated module in a standard RNN. Source: Adapted from Haykin (2003) and Ajitha et al.
(2022).

Ajitha et al. 2022) below represent a basic formulation of a simple RNN:

h(t) = 𝜎(Whx ∗ X + Whh ∗ h(t−1) + bh ) (10.20)
where,
– W hx —is the matrix of weights between the i/p and the hidden layers.
– W hh —is the weight matrix that recurs between neighboring time steps’ hidden layers.
– X—is an element of the input vector.
– bh and by —bias vectors that let every node to learn and offset.
– h(t) —is the hidden layer’s neuronal output.

[X]T = [R1 , R2 , … ., RNL , XL1 , XL2 , … ., XLNL , Bc1 , Bc2 , … ., BcNL , V1 , V2 , … ., VNG &Sd ]T
(10.21)

Y (t) = sof max (Wyh ∗ h(t) + by ) (10.22)

10.8 Results and Discussions

In the 6-bus 7-line, 14-bus 20-line, and 25-bus 35-line IEEE standard test systems, the proposed
method has been implemented [Appendix 10.A]. Resistances to the line, reactances to the line,
susceptance to line charges, voltage of the PV bus and loads bus power factor make up the vector
of to be sampled random variables “Z.” Both the load bus and PV-bus voltage should be between
the ranges of 0.80 pu − 1.05 pu and 0.95 pu − 1.10 pu, respectively.

10.8.1 IEEE Six-Bus System

There are 2-PV buses and 4-PQ buses in this system. Lines 1 through 7 are presumed to be avail-
able at 0.99636033, 0.99636033, 0.99658703, 0.99545481, 0.99545481, 0.99658703, and 0.99658703
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10.8 Results and Discussions 179

Double line outage Single line outage

0.027
Probability of failure

0.022
0.017
0.012
0.007
0.002
0.6 0.9 1.2 1.5 1.8 2.1
Total system load in (pu)

Figure 10.4 Probability of failure shown against different levels of load for Ṽ 1 = 0.9625 pu,
Ṽ 2 = 0.9517 pu, RT = 1.4307 pu, and XT = 3.6482 pu for six-bus system.

(Billinton and Allan 1984). Using Eqs. (10.6) and (10.7), determine the security index with and
without taking transmission line availability into account, and then arrange the transmission lines
in severity order. The value of the security index has determined the line’s severity. Table 10.1 com-
pares the contingency ranking of lines based on the security index. The ranking was performed
considering various factors such as the availability of the line, voltage performance index, min-
imum eigenvalue, Schur’s inequality value, and line flow performance index (Verma and Niazi
2012; Singh et al. 2018). Equation (10.19) was used to calculate the PISI, which takes line uncer-
tainties into consideration, is the probability of failure characteristics and control factors under dif-
ferent loading conditions. Static voltage stability constraints derived using the predictor-corrector
approach (Arya and Verma 1996) were utilized. The PISI values were determined, which provide
an assessment of the system’s voltage stability. Table 10.2 lists the inputs, values for the input layer
in an RNN, PISI values generated from trained networks and cut-set calculations that take into
account single and double-line outages, as well as the % error. The buried layer’s neuron count
was increased to 17. The input layer’s variables include system load (R1 − R7 , X 1 − X 7 , V 1 − V 2 , and
Sd ) voltage of generator buses, and line characteristics (line resistances, reactances, and charging
susceptances). In addition to being utilized in training, trained RNN has undergone limited testing.
Percentage errors along with the test instances in Table 10.3, the PISI computed the uses of trained
network and cut-set approach taking into account the single and double-line outage is displayed.
The ratio of training to testing instances is not governed by any rules.
1000 training examples were chosen for the current study in order to represent a wide variety
of power system operating situations. In addition, outside of training cases, testing examples
have been chosen from this range. 1000 test instances have been used to test the trained network.
Table 10.3 only lists a select few with the greatest % error magnitudes. The estimated probabil-
ity of failure is plotted against different load levels in Figure 10.4, with Ṽ 1 = 0.9625 pu, Ṽ 2 =
0.9517 pu, RT = 1.4307 pu, and XT = 3.6482 pu taking into account single and double-line out-
ages. The percentage error in the failure probability as determined by trained RNN is less than ±1%.

10.8.2 IEEE 14-Bus System

Eleven load buses and three generator buses make up this system. It is assumed that lines
1 through 20 have the following availability values: 0.99545486, 0.99627723, 0.996336033,
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 10.1 Ranking of contingencies for the six-bus, seven-line system accounting unpredictability based on the security index with and without taking into
account the availability of transmission lines, the voltage performance index, the lowest eigen value, the value of Schur’s inequality, and the performance index
for line flows (S d = 0.7401 pu).

Transmission line
Security index without availability
taking into account consideration while
the absence of a evaluating Index of voltage Minimum eigen Schur’s inequality Index of line flow
transmission line security index performance (PIV ) value (𝝀min ) value (𝝉) performance (PIMW )

Line Line Line Line Line Line

S. no. p̂ number SI number PIV number 𝝀min number 𝝉 number PIMW number

1. 0.8936 7 0.003252 7 0.0089 7 0.2138 7 0.1849 7 0.8405 7

2. 0.4139 5 0.001506 5 0.3187 2 0.2140 5 0.2002 2 0.7162 5
3. 0.3611 2 0.001232 2 0.3264 5 0.2218 2 0.2037 5 0.6317 2
4. 0.1217 1 0.000553 1 0.4295 6 0.2577 1 0.2381 6 0.5641 1
5. 0.0997 6 0.000453 6 0.6056 1 0.4553 6 0.3432 1 0.5299 6
6. 0.0086 4 0.000029 4 0.6571 4 0.4765 3 0.3748 4 0.5073 3
7. 0.0032 3 0.000011 3 0.6708 3 0.4849 4 0.3788 3 0.4364 4
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 10.2 For the six-bus test system, training instances were used to train RNN, PISI produced using the cut-set approach, and RNN.

Outage on a single line Double-line outage

PISI PISI
calculated PISI calculated PISI
using the calculated using the calculated
Total resistance Total reactance Overall system cut-set using the cut-set using the
of the line (pu) of the line (pu) Voltage on the PV bus (pu) load (pu) technique RNN % Error technique RNN % Error
Sr. no. RT XT V1 V2 Sd

1 1.4667176 3.5405501 0.9647450 1.0855388 1.35 0.022203691 0.022227383 −0.106703691 0.021884957 0.021908688 −0.108436011
2 1.4723193 3.5616766 1.0843165 0.9579196 1.00 0.019951132 0.020002480 −0.257369584 0.019632398 0.019683788 −0.261759334
3 1.4032147 3.6191689 0.9847865 1.0155034 1.10 0.026780936 0.026753588 0.102116022 0.026462202 0.026434895 −0.241678302
4 1.4622588 3.6280009 1.0067415 0.9657292 1.10 0.023364876 0.023338060 0.114769662 0.023046142 0.023019367 0.116179858
5 1.4225665 3.5652915 1.0536958 0.9839854 1.00 0.01632311 0.016345186 −0.135242228 0.016004376 0.016026493 −0.138195707
6 1.4115554 3.5919799 0.9718635 1.0230430 1.40 0.026765424 0.026815515 −0.187146623 0.026446690 0.026496820 −0.189550538
7 1.3876319 3.6286525 0.9524766 1.0177007 1.10 0.018628132 0.018684821 −0.304321099 0.018309398 0.018366128 −0.309841752
8 1.4308110 3.5267878 1.0277902 0.9727332 1.00 0.023367966 0.023343840 0.103244472 0.023049232 0.023025148 0.104490897
9 1.4169602 3.6333706 0.9992843 0.9581896 1.85 0.026777846 0.026809088 −0.116671358 0.026459112 0.026490392 −0.118218883
10 1.4157303 3.6547260 1.0057756 1.0033403 1.80 0.026781967 0.026822293 −0.150570174 0.026463233 0.026503596 −0.152525520
11 1.4736682 3.6101888 1.0844996 1.0781777 1.70 0.019929694 0.019983579 −0.270374176 0.019610960 0.019664882 −0.274959809
12 1.4536465 3.6148612 0.9764595 0.9620212 1.00 0.015228614 0.015268276 −0.260446692 0.014909880 0.014949584 −0.266291191
13 1.4895729 3.6302745 1.0661153 0.9846217 1.05 0.019951132 0.019991827 −0.203972790 0.019632398 0.019673133 −0.207491116
14 1.4062879 3.6742212 1.0634236 1.0439772 1.40 0.019975202 0.019991816 −0.083173261 0.019656468 0.019673121 −0.084719108
15 1.4764972 3.6514651 0.9601764 0.9566103 1.70 0.026809901 0.026786442 0.087502874 0.026491167 0.026467745 0.088412517
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 10.3 Test cases created for the 6-bus system utilizing the cut-set technique to validate the trained network.

Single line outage Double-line outage

PISI PISI
calculated PISI calculated PISI
using calculated using the calculated
Total Total Total system the cut-set using the cut-set using the
resistance (pu) reactance (pu) Voltage on the PV bus (pu) load (pu) technique RNN % Error technique RNN % Error
Sr. no. RT XT V1 V2 Sd

1 1.4764972 3.6514651 0.9601764 0.9566103 1.65 0.021691506 0.02171474 −0.107112442 0.021559552 0.021582802 −0.107841196
2 1.4367804 3.6242044 0.9750771 1.0908399 1.45 0.019683454 0.019731262 −0.242884410 0.0195515 0.019599324 −0.244605042
3 1.4157303 3.6547260 1.0057756 1.0033403 1.25 0.023182217 0.023237565 −0.238752311 0.023050263 0.023105628 −0.240190538
4 1.4658493 3.5779551 1.0584796 0.9726279 1.15 0.026469674 0.026442507 0.102633762 0.026337720 0.026310570 0.103084486
5 1.4944772 3.5577229 1.0553987 1.0555223 1.80 0.026473749 0.026467346 0.024184745 0.026341795 0.026335408 0.024247355
6 1.4600478 3.5391203 1.0607959 1.0188818 1.85 0.021132127 0.021107740 0.115400543 0.021000173 0.020975802 0.116052008
7 1.4047314 3.5561451 0.9564821 0.9963408 1.75 0.016304293 0.016324544 −0.124207942 0.016172339 0.016192606 −0.125318601
8 1.3982844 3.6256708 1.0947264 1.0508896 1.65 0.023144236 0.023198757 −0.235568442 0.023012282 0.023066818 −0.236987690
9 1.4152725 3.5614248 1.0983200 1.0694296 1.70 0.016248296 0.016304573 −0.346353580 0.016116342 0.016172634 −0.349286516
10 1.4529598 3.5946015 0.9890439 0.9810556 1.05 0.018620484 0.018677469 −0.306034907 0.018488530 0.018545532 −0.308311016
11 1.4622588 3.6280009 1.0067415 0.9657292 1.05 0.023178096 0.02315436 0.102405757 0.023046142 0.023022423 0.102918230
12 1.4047314 3.5561451 0.9564821 0.9963408 1.80 0.026494156 0.02653479 −0.153368359 0.026362202 0.026402851 −0.154195393
13 1.4046704 3.5794307 1.0941954 1.0988784 1.70 0.019759457 0.019813717 −0.274603914 0.019627503 0.019681779 −0.276529728
14 1.4528591 3.6040027 0.9994537 0.9628956 1.00 0.019725126 0.019766744 −0.210991175 0.019593172 0.019634807 −0.212499922
15 1.4169602 3.6333706 0.9992843 0.9581896 1.80 0.026591066 0.026567732 0.087752429 0.026459112 0.026435793 0.088130980
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 10.4 Ranking of contingencies for the 14-bus, uncertainty in the 20-line method of accounting based on the security index with and without taking into
account the availability of transmission lines, the voltage performance index, the lowest eigen value, the value of Schur’s inequality, and the line flow
performance index (S d = 3.2057 pu).

Line Line Line Line Line Line

S. no. p̂ number SI number PIV number 𝝀min number 𝝉 number PIMW number

1. 1.0000000 10 0.0037617 10 0.0532 15 0.1300 10 0.1247 15 0.9753 8

2. 0.9835935 8 0.0030776 8 0.2075 10 0.2117 15 0.1627 10 0.7549 15
3. 0.9825183 15 0.0032821 13 0.2468 8 0.2861 8 0.2201 8 0.6938 10
4. 0.8957811 3 0.0027724 1 0.3283 13 0.3774 3 0.279 1 0.6265 3
5. 0.8693352 13 0.0016593 15 0.3751 3 0.3848 13 0.2912 13 6024 13
6. 0.8276328 1 0.0025521 3 0.4087 1 0.3899 2 0.2915 3 0.4537 16
7. 0.8267000 2 0.0014862 2 0.4164 2 0.4150 1 0.2934 16 0.4162 1
8. 0.8176957 16 0.0041268 12 0.4237 16 0.4246 12 0.308 12 0.3697 2
9. 0.6674546 12 0.0023984 16 0.4204 4 0.4400 6 0.3116 2 0.3678 12
10. 0.6607942 4 0.0042791 4 0.4365 12 0.4484 16 0.3126 6 0.2548 4
11. 0.6546000 9 0.0020839 6 0.4429 6 0.4529 9 0.3218 4 0.2457 9
12. 0.5964156 6 0.0029421 9 0.4538 9 0.4764 4 0.3265 9 0.2194 6
13. 0.5279563 17 0.0039513 17 0.4576 17 0.4809 5 0.3311 11 0.2081 5
14. 0.4967308 11 0.0035999 11 0.4832 5 0.4828 17 0.3358 17 0.1949 11
15. 0.4861905 5 0.0027908 5 0.4884 11 0.4911 20 0.3383 5 0.1508 17
16. 0.4556678 20 0.0022592 20 0.4925 20 0.4913 11 0.3393 18 0.1407 20
17. 0.3656452 18 0.0015340 18 0.5001 18 0.4918 19 0.3425 20 0.1383 19
18. 0.3269835 7 0.0011212 7 0.5017 19 0.4960 18 0.3486 7 0.1241 7
19. 0.3060000 19 0.0015552 19 0.5037 7 0.5075 7 0.3512 19 0.1005 18
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 10.5 For the 14-bus test system, training instances were used to train RNN, PISI produced using the cut-set approach, and RNN.

Single line outage Double-line outage

PISI PISI
Total Total Total calculated PISI calculated PISI
resistance reactance susceptance Total using the calculated using the calculated
of the line of the line of the system cut-set using the cut-set using the
e (pu) (pu) line (pu) Voltage on the PV bus (pu) load (pu) technique RNN % Error technique RNN % Error
Sr. no. RT XT BcT V1 V2 V3 Sd

1 1.6042 4.53288 0.2849 1.0985 1.0867 1.0357 3.80 0.051198207 0.050073527 2.19671664 0.050187265 0.051221869 −2.06148649
2 1.6090 4.48502 0.2911 1.0846 1.0080 0.9644 3.00 0.038371241 0.038895992 −1.36756387 0.038360351 0.038344849 0.04041105
3 1.5875 4.37574 0.2929 1.0713 1.0543 0.9603 3.70 0.068012148 0.066988146 1.50561669 0.066324552 0.066232799 0.13833890
4 1.6804 4.52455 0.3550 1.0559 1.0102 0.9667 3.50 0.073982852 0.074841303 −1.16033752 0.073602433 0.073970831 −0.50052405
5 1.6727 4.53445 0.3537 1.0930 1.0749 1.0366 3.20 0.020096817 0.020690922 −2.95621350 0.018478687 0.018667433 −1.02142648
6 1.5556 4.63175 0.3287 1.0768 1.0768 1.0079 3.90 0.027705047 0.027833410 −0.46332120 0.027366816 0.027618466 −0.91954552
7 1.6457 4.55146 0.2835 0.9716 0.9614 0.9523 2.70 0.070073542 0.070135728 −0.08874461 0.070043845 0.070098245 −0.07766603
8 1.5498 4.45702 0.2756 0.9954 0.9747 1.0190 2.90 0.032044691 0.032063200 −0.05776055 0.032021164 0.032054873 −0.10527244
9 1.5897 4.50539 0.3194 1.0153 0.9995 0.9817 2.30 0.011411055 0.011261375 1.31170690 0.011097978 0.011403143 −2.74973381
10 1.6804 4.52455 0.3550 1.0559 1.0102 0.9667 3.00 0.057929880 0.058511169 −1.00343578 0.057575880 0.058019316 −0.77017713
11 1.6496 4.49989 0.2837 1.0334 1.0206 1.0107 2.50 0.023621750 0.023611395 0.04383530 0.022975380 0.023607145 −2.74974790
12 1.5937 4.50971 0.3045 1.0927 1.0355 1.0911 3.60 0.050502396 0.048997483 2.97988455 0.047567311 0.048950519 −2.90789566
13 1.5630 4.52398 0.3604 1.0927 1.0513 1.0231 3.70 0.043050652 0.042890121 0.37288964 0.042608465 0.042949935 −0.80141350
14 1.6611 4.59325 0.3765 1.0868 1.0778 1.0360 3.00 0.030837685 0.030732716 0.34039355 0.030711506 0.030209043 1.63607398
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 10.6 Test cases created for the 14-bus system utilizing the cut-set technique to validate the trained network.

Single line outage Double-line outage

PISI PISI
Total Total Total calculated PISI calculated PISI
resistance reactance susceptance Total using the calculated using the calculated
of the of the of the system cut-set using the cut-set using the
line (pu) line (pu) line (pu) Voltage on the PV bus (pu) load (pu) technique RNN % Error technique RNN % Error
Sr. no. RT XT BcT V1 V2 V3 Sd

1 1.6804 4.5245 0.3550 1.0559 1.0102 0.9667 3.30 0.0739828520 0.074982862 −1.35167868 0.0727800820 0.074170974 −1.91108902
2 1.5889 4.5819 0.3185 1.0894 1.0698 0.9637 2.80 0.0149570180 0.014919278 0.25232268 0.0136388773 0.013428926 1.53935804
3 1.6811 4.5088 0.3066 1.0582 1.0110 1.0552 3.50 0.0670110200 0.065672101 1.99805785 0.0656672440 0.065562060 0.16017651
4 1.6611 4.5932 0.3765 1.0868 1.0778 1.0360 3.70 0.0598030410 0.058998621 1.34511562 0.0563377080 0.056274471 0.11224566
5 1.5608 4.4150 0.3487 1.0275 0.9885 0.9937 3.20 0.0367757500 0.037776034 −2.71995640 0.0362458115 0.036422585 −0.48770794
6 1.5937 4.6211 0.3576 1.0889 0.9650 0.9885 3.50 0.0287576780 0.029326858 −1.97922870 0.0285766392 0.028542475 0.11955184
7 1.6811 4.5088 0.3066 1.0582 1.0110 1.0552 3.30 0.0429028730 0.043094332 −0.44626157 0.0426598144 0.042976752 −0.74294112
8 1.6042 4.5329 0.2849 1.0985 1.0867 1.0357 3.60 0.0331167070 0.033616714 −1.50983411 0.0328075000 0.033200258 −1.19715961
9 1.5937 4.5097 0.3045 1.0927 1.0355 1.0911 3.40 0.0325348010 0.032934479 −1.22846344 0.0325348010 0.032969857 −1.33720130
10 1.5327 4.6164 0.3303 1.0639 0.9759 0.9527 2.80 0.0596568890 0.058609378 1.75589338 0.0596568890 0.058557856 1.84225586
11 1.6351 4.4319 0.3364 1.0550 1.0788 0.9504 4.00 0.0203750950 0.020775105 −1.96322980 0.0202975266 0.020100544 0.97047483
12 1.5937 4.5097 0.3045 1.0927 1.0355 1.0911 3.30 0.0238195870 0.023920014 −0.42161585 0.0237105532 0.023915106 −0.86270971
13 1.6384 4.5724 0.3241 1.0323 0.9942 1.0311 3.80 0.0318673690 0.032549425 −2.14029715 0.0315572695 0.032162061 −1.91648942
14 1.5630 4.5240 0.3604 1.0927 1.0513 1.0231 3.40 0.0430506520 0.042050654 2.32284001 0.0428745883 0.042314125 1.30721658
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
186 10 Transmission System Reliability Evaluation Including Security

Double line outage Single line outage

0.082
Probability of failure

0.072
0.062
0.052
0.042
0.032
0.022
0.012
0.002
2.0000 2.5000 3.0000 3.5000 4.0000
Total system load (pu)

Figure 10.5 Probability of failure shown against various load levels for Ṽ 1 = 1.0824 pu, Ṽ 2 = 1.0351 pu,
Ṽ 3 = 0.9522 pu, RT = 1.6427 pu, X T = 4.6137 pu and BcT = 0.2514 pu for 14-bus system.

0.99580437, 0.99658704, 0.99572088, 0.99545486, 0.99580437, 0.996336033, 0.99572088,

0.99580473, 0.99559204, 0.99545486, 0.99580473, 0.996336033, 0.99658704, 0.99572088,
0.99580473, 0.996336033, and 0.99658704 (Billinton and Allan 1984). Table 10.4 compares
the security index-based contingency rating of lines with and without taking into account the
uncertainty relating to the availability of the transmission line accounting, the voltage perfor-
mance index, minimum eigen value, Schur’s inequality value, and the performance index for line
flow (Verma and Niazi 2012; Singh et al. 2018). Buses 7 and 5 are connected by line number 14,
which is radial and whose failure causes islanding (Dester and Castro 2009). Table 10.5 lists the
inputs for the RNN’s input layer, values for PISI determined using the trained network and the
cut-set approach while accounting for single and double-line outages, as well as the percentage
error. Sixty-four neurons were added to the buried layer. The input layer’s variables include
line resistances (R1 − R20 ), line reactance (X 1 − X 20 ), line charging susceptances (Bc1 − Bc20 ),
PV-bus voltages (V 1 − V 3 ), and total system load (Sd ). Few testing cases, other from those used
in training, have been utilized to validate trained RNN. Table 10.6 displays using a trained
network and cut-set methodology, the PISI was calculated. While taking into account a single
and double-line outage along with the % errors for the test instance. In order to cover a wide
variety of operating circumstances for the power system, 1000 training examples were chosen for
the current study. Moreover, excluding training examples, testing cases have been chosen from
this range. For another 1000 test cases, the trained network has been tested. With the largest
% error magnitudes, only a select number are shown in Table 10.6. The predicted chance of
failure is plotted against different load levels in Figure 10.5, with Ṽ 1 = 1.0824 pu, Ṽ 2 = 1.0351 pu,
Ṽ 3 = 0.9522 pu, RT = 1.6427 pu, X T = 4.6137 pu, and BcT = 0.2514 pu taking into account single
and double-line outages. The percentage error for the failure probability as determined by trained
RNN is less than 3%.

10.8.3 IEEE 25-Bus System

Twenty load buses and of five generator buses make up this system. It is assumed that lines
1 through 35 have the following availability values: 0.99559204, 0.99658704, 0.99545486,
0.99580473, 0.996336033, 0.99559204, 0.99572088, 0.99658704, 0.99545486, 0.99627723,
0.99559204, 0.99658704, 0.99545486, 0.99572088, 0.99580473, 0.996336033, 0.99559204,
0.99627723, 0.99545486, 0.99580473, 0.99658704, 0.996336033, 0.99559204, 0.99658704,
0.99627723, 0.99572088, 0.99545486, 0.99545486, 0.99580473, 0.99580473, 0.99627723, 0.99627723,
0.99559204, 0.996336033, and 0.99572088 (Billinton and Allan 1984). Table 10.7 compares rating
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 10.7 Ranking of contingencies for the 25-bus, using the security index as the basis, 35-line system accounting uncertainty with and without taking into
account the availability of transmission lines, the voltage performance index, the lowest eigen value, the value of Schur’s inequality, and the line flow
performance index (S d = 14.25 pu).

Line Line Line Line Line Line

S. no. p̂ number SI number PIV number 𝝀min number 𝝉 number PIMW number

1. 1.0000000 5 0.00091594 35 0.1442 11 0.1736 11 0.1645 5 0.910046 35

2. 0.9143000 35 0.00256579 5 0.1593 35 0.1800 5 0.1646 35 0.858553 11
3. 0.8200000 11 0.00231678 13 0.1865 5 0.2114 35 0.1946 11 0.846557 3
4. 0.8000000 13 0.00366397 11 0.1998 13 0.2416 13 0.2162 3 0.843429 5
5. 0.5645120 3 0.00064187 3 0.2211 3 0.2736 3 0.2358 4 0.804561 13
6. 0.5522361 4 0.00074016 4 0.2458 4 0.2846 10 0.2409 13 0.770274 10
7. 0.5268493 10 0.00101597 33 0.2639 33 0.3038 4 0.2418 10 0.761828 4
8. 0.4846530 33 0.00196134 10 0.2911 10 0.3091 33 0.2464 33 0.745136 33
9. 0.3281250 14 0.00361453 14 0.2966 17 0.3248 17 0.2555 17 0.744512 14
10. 0.3090909 17 0.00363611 17 0.3059 14 0.3252 14 0.2590 14 0.733536 9
11. 0.2235294 9 0.00140409 9 0.3186 9 0.3260 1 0.2658 8 0.728637 17
12. 0.2168675 8 0.00084904 1 0.3425 1 0.3261 9 0.2659 9 0.727498 8
13. 0.2125000 21 0.00062811 20 0.343 21 0.3286 8 0.2681 1 0.724276 20
14. 0.2105263 20 0.00136246 15 0.3475 15 0.3327 21 0.2698 20 0.720549 15
15. 0.2077922 1 0.00072233 26 0.3498 20 0.3341 15 0.2717 21 0.717101 1
16. 0.2023810 15 0.00076702 29 0.3531 29 0.3343 20 0.2750 15 0.716267 22
17. 0.1940299 18 0.00088321 19 0.3532 18 0.3351 29 0.2752 29 0.713124 19
18. 0.1923077 22 0.00072525 8 0.3549 8 0.3356 25 0.2770 22 0.711710 29
19. 0.1917808 29 0.00070461 21 0.3555 22 0.3370 22 0.2789 34 0.706750 21
20. 0.1891995 24 0.00064612 18 0.3557 24 0.3371 18 0.2789 19 0.705310 18
21. 0.1891892 26 0.00064573 22 0.3564 34 0.3395 34 0.2802 18 0.618581 26
22. 0.1818182 34 0.00064744 34 0.3571 26 0.3428 26 0.2832 25 0.610767 34
23. 0.1739130 25 0.00080956 25 0.3585 25 0.3438 24 0.2834 26 0.595676 25
24. 0.1714286 16 0.00060759 23 0.3613 16 0.3446 19 0.2849 24 0.582575 24
25. 0.1687556 19 0.00080457 24 0.3626 19 0.3449 23 0.2850 16 0.574432 23
26. 0.1666667 31 0.00051162 7 0.3639 23 0.3452 7 0.2861 7 0.504538 16
27. 0.1500000 7 0.00062046 16 0.3655 7 0.3555 16 0.2871 27 0.446335 7
28. 0.1465812 23 0.00041194 31 0.3671 31 0.3585 27 0.2881 23 0.415604 27
29. 0.1336782 27 0.00213633 27 0.3677 27 0.3680 32 0.2916 31 0.342200 30
30. 0.1219512 30 0.00066618 30 0.3685 32 0.3933 30 0.2936 32 0.292021 31
31. 0.1106546 32 0.0039124 32 0.3698 30 0.3966 31 0.3067 30 0.182064 32
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 10.8 For the 25-bus test system, training instances were used to train RNN, PISI produced using the cut-set approach, and RNN.

Single line outage Double-line outage

PISI PISI
Total Total Total calculated PISI calculated PISI
resistance reactance susceptance Total using the calculated using the calculated
of the of the of the system cut-set using the cut-set using the
line (pu) line (pu) line (pu) Voltage on the PV bus (pu) load (pu) technique RNN % Error technique RNN % Error
Sr. no. RT XT BcT V1 V2 V3 V4 V5 Sd

1 3.2670 8.2441 1.8710 1.0385 1.0377 1.0246 0.9666 1.0390 16.5000 0.120877350 0.119634097 1.039213066 0.119993594 0.121999362 −1.671562965
2 3.0886 8.3500 1.7975 1.0072 0.9635 1.0868 1.0778 1.0360 16.0000 0.058184539 0.056979097 2.115587115 0.057184539 0.056102010 1.893044539
3 3.2642 8.3465 1.8327 0.9793 1.0199 1.0099 1.0020 1.0712 15.7500 0.117413828 0.116491478 0.791774879 0.112625830 0.109987978 2.342137477
4 3.2670 8.2441 1.8710 1.0385 1.0377 1.0246 0.9666 1.0390 14.7500 0.122171658 0.11904689 2.624821142 0.116697451 0.119394886 −2.311477131
5 3.2845 8.3421 1.8646 0.9508 1.0312 1.0426 1.0239 1.0369 12.5000 0.100421664 0.101478618 −1.041553954 0.101385916 0.102617698 −1.214943543
6 3.3596 8.3251 1.9022 0.9530 1.0818 0.9538 0.9954 0.9747 14.0000 0.073956824 0.074944386 −1.317727294 0.073386302 0.074901688 −2.064944416
7 3.2503 8.3769 1.8786 0.9868 1.0875 1.0350 0.9995 0.9909 14.2500 0.086792981 0.086534876 0.298266904 0.084946759 0.085880591 −1.099314605
8 3.2832 8.3042 1.8473 1.0058 1.0033 1.0865 1.0199 1.0139 16.0000 0.109911045 0.106909037 2.808002136 0.109100725 0.107460809 1.503120924
9 3.2900 8.1858 1.8506 0.9812 1.0404 0.9905 1.0971 1.0121 16.2500 0.062676858 0.06281575 −0.221110493 0.062645620 0.063584350 −1.498476958
10 3.1829 8.2850 1.8173 1.0885 1.0234 1.0109 1.0267 0.9548 15.7500 0.078953798 0.077518086 1.85209952 0.078653006 0.076337157 2.944387669
11 3.2012 8.3095 1.8615 0.9937 1.0703 1.0684 1.0514 1.0633 15.5000 0.065318686 0.067135056 −2.705546029 0.064117106 0.064917235 −1.247918051
12 3.3184 8.3020 1.8591 0.9751 1.0908 0.9682 1.0012 0.9643 14.5000 0.085620507 0.083250939 2.846295367 0.085357411 0.086357349 −1.171471263
13 3.2832 8.3042 1.8473 1.0058 1.0033 1.0865 1.0199 1.0139 15.0000 0.109708594 0.109613488 0.086764497 0.109100725 0.106248208 2.614571762
14 3.2900 8.1858 1.8506 0.9812 1.0404 0.9905 1.0971 1.0121 14.7500 0.062645620 0.062598318 0.075563533 0.060944003 0.060368534 0.944257934
15 3.2845 8.3421 1.8646 0.9508 1.0312 1.0426 1.0239 1.0369 15.2500 0.100819384 0.103226649 −2.332018836 0.098074931 0.099078070 −1.022828704
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 10.9 Test cases created for the 25-bus system utilizing the cut-set technique to validate the trained network.

Single line outage Double-line outage

PISI PISI
Total Total Total calculated PISI calculated PISI
Resistance Reactance Susceptance Total using the calculated using the calculated
of the of the of the system cut-set using the cut-set using the
line (pu) line (pu) line (pu) Voltage on the PV bus (pu) load (pu) technique RNN % Error technique RNN % Error
Sr. no. RT XT BcT V1 V2 V3 V4 V5 Sd

1 3.2454 8.4066 2.0789 1.0073 0.9651 1.0395 1.0849 1.0827 16.50 0.120682105 0.123456209 −2.247034462 0.119993594 0.121876130 −1.568864065
2 3.0886 8.3500 1.7975 1.0072 0.9635 1.0868 1.0778 1.0360 15.50 0.058873050 0.05764253 2.134743106 0.058184539 0.059128193 −1.621830277
3 3.1536 8.2514 1.8614 1.0318 1.0711 1.0946 0.9643 0.9663 14.25 0.089710726 0.087510516 2.514224023 0.089022215 0.087322020 1.909854958
4 3.0886 8.3500 1.7975 1.0072 0.9635 1.0868 1.0778 1.0360 14.00 0.054930041 0.054130317 1.477404509 0.054241530 0.053429167 1.497677331
5 3.2544 8.2392 1.8475 0.9614 0.9523 1.0579 1.0051 1.0490 13.50 0.083563305 0.082951596 0.737428659 0.082874794 0.080902830 2.379449119
6 3.2059 8.2117 1.7948 1.0364 1.0744 0.9599 0.9906 1.0550 16.50 0.087281368 0.087421567 −0.160371195 0.086592857 0.085414580 1.360708925
7 3.2035 8.2343 1.7687 1.0357 0.9977 1.0109 0.9704 1.0295 16.25 0.126366928 0.122822889 2.885487429 0.116033292 0.113843616 1.887110493
8 3.3596 8.3251 1.9022 0.9530 1.0818 0.9538 0.9954 0.9747 14.75 0.076525784 0.075145155 1.837282319 0.073386302 0.073806304 −0.572316376
9 3.3071 8.4263 1.8431 1.0684 1.0465 1.0657 1.0529 1.0169 15.50 0.101110175 0.100538867 0.568245829 0.100421664 0.097859115 2.551789102
10 3.3596 8.3251 1.9022 0.9530 1.0818 0.9538 0.9954 0.9747 15.00 0.074074813 0.07489636 −1.096912129 0.073386302 0.073152310 0.318849495
11 3.2900 8.1858 1.8506 0.9812 1.0404 0.9905 1.0971 1.0121 13.00 0.063365369 0.062126838 1.993552128 0.062676858 0.062381957 0.470510449
12 3.1327 8.1714 1.6052 1.0362 1.0559 1.0102 0.9667 1.0846 15.25 0.104670457 0.106974157 −2.153510594 0.103981946 0.105931273 −1.874677998
13 3.3045 8.2627 1.8763 1.0044 1.0735 0.9966 1.0986 1.0657 13.50 0.073688662 0.075810501 −2.798871915 0.073000151 0.074588455 −2.175754209
14 3.1829 8.2850 1.8173 1.0885 1.0234 1.0109 1.0267 0.9548 16.00 0.079341517 0.078519208 1.047270549 0.078653006 0.078522963 0.165337993
15 3.3513 8.3514 1.8515 1.0121 1.0048 1.0153 0.9995 0.9817 14.50 0.098763442 0.099103019 −0.342650502 0.098074931 0.095188206 2.943387076
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
190 10 Transmission System Reliability Evaluation Including Security

Double line outage Single line outage

0.1400
Probability of failure

0.1200
0.1000
0.0800
0.0600
0.0400
0.0200
0.0000
11.00 12.00 13.00 14.00 15.00 16.00 17.00
Total system load in (pu)

Figure 10.6 Probability of failure shown against various load levels for Ṽ 1 = 0.9793 pu, Ṽ 2 = 1.0199 pu,
Ṽ 3 = 1.0099 pu, Ṽ 4 = 1.0020 pu, Ṽ 5 = 1.0712 pu, RT = 3.2642 pu, X T = 8.3465 pu and BcT = 1.8327 pu for
25-bus system.

of lines’ contingency based on security index with and without taking into account the uncertainty
relating to the availability of the trans. line accounting, the voltage performance index, minimum
eigen value, Schur’s inequality value, and the performance index for line flows (Verma and Niazi
2012; Singh et al. 2018). Lines 2, 6, 12, and 28 are radial lines that connect buses 1 and 16 as
well as buses 1 and 25 and 15 and 16 (not included in Table 10.7), and when they are down,
the buses become isolated (Dester and Castro 2009). Table 10.8 lists the inputs for the RNN’s
input layer, values for PISI determined using the trained network and the cut-set approach while
accounting for single and double-line outages, as well as the percentage error. The buried layer’s
neuron count was increased to 111. Line resistance (R1 − R35 ), line reactance (X 1 − X 35 ), line
charging susceptances (Bc1 − Bc35 ), PV-bus voltages (V 1 − V 5 ), and total system load (Sd ). are
the variables of the input layer. Few testing cases, other from those used in training, have been
utilized to validate trained RNN. Table 10.9 includes the PISI, % errors for test cases, and trained
N/W and cut-set technique results when single and double-line outages are taken into account.
1000 training examples were chosen for the current study in order to represent a variety of power
system operating scenarios. In addition, outside of training cases, testing examples have been
chosen from this range. 1000 test instances have been used to test the trained network. Table 10.9
only includes a small number of items with significant % error magnitudes. In Figure 10.6, the
projected likelihood of failure is shown against different load levels with the values Ṽ 1 = 0.9793 pu,
Ṽ 2 = 1.0199 pu, Ṽ 3 = 1.0099 pu, Ṽ 4 = 1.0020 pu, Ṽ 5 = 1.0712 pu, RT = 3.2642 pu, X T = 8.3465 pu
and BcT = 1.8327 pu taking into account single and double-line outages. The proportion of
inaccuracy in the likelihood of failure as determined by trained ANN is less than 3%.

10.9 Conclusions
The novel approach presented in the book chapter addresses the voltage security evaluation by con-
sidering uncertainties in control variables and line parameters. It offers a comprehensive method-
ology that combines Monte Carlo simulation, cut-set analysis for contingency planning, and the
utilization of a recurrent neural network (RNN) for efficient and accurate results. The Monte Carlo
simulation is used to assess the security index, taking into account transmission line availability.
This helps in identifying the most critical lines in the system. The cut-set approach is then employed
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10.A.1 Data for IEEE six-bus, seven-line test system (100 MVA Base) 191

to evaluate the probability of failure under different considering voltage stability, loading situations,
and single- and double-line outages limits. To overcome the computational complexity associated
with probabilistic approaches, the research introduces the use of RNNs. The RNN is trained using
the obtained security index and PISI values for various contingencies and loading scenarios. This
trained network enables fast and precise evaluations, providing prompt preventive intervention.
The propose approaches effectiveness is demonstrated by comparing the contingency selection
results with established approaches. The RNN-based method offers accurate and efficient results,
which are crucial for ensuring the stability and reliability of complex power systems. The approach
has been implemented and tested on different test systems, including the 6-bus 7-line, 14-bus
20-line, and 25-bus 35-line IEEE standard test systems, showcasing its applicability in real-world
scenarios.

Appendix 10.A

10.A.1 Data for IEEE six-bus, seven-line test system (100 MVA
Base)

Table 10.A.1 Line data.

Bus no. Resistance to Reactance to Susceptance to Line flow

Line no. From To lines (R) (pu) lines (X) (pu) lines (Bc) (pu) Tapping limit (pu)

1 1 6 0.123 0.5180 0.0000 1.0000 3.00

2 1 4 0.0800 0.3700 0.0000 1.0000 3.00
3 4 6 0.0970 0.4070 0.0000 1.0000 3.00
4 6 5 0.0000 0.3000 0.0000 1.0000 1.00
5 5 2 0.2820 0.6400 0.0000 1.0000 2.50
6 2 3 0.7230 1.0500 0.0000 1.0000 1.50
7 4 3 0.0000 0.1330 0.0000 1.0000 2.50

Table 10.A.2 Bus data.

Load (pu) Generation (pu)

Bus number Active Reactive Active Reactive

1 0.0000 0.0000 0.4500 0.0600

2 0.0000 0.0000 0.5000 0.2000
3 0.2750 0.0650 0.0000 0.0000
4 0.0000 0.0000 0.0000 0.0000
5 0.1500 0.0900 0.0000 0.0000
6 0.2500 0.0050 0.0000 0.0000
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
192 10 Transmission System Reliability Evaluation Including Security

4
5

Figure 10.A.1 IEEE six-bus test system.

10.A.2 Data for IEEE 14-bus, 20-line system (100 MVA Base)
Table 10.A.3 Line data.

Line Bus number Line flow

number From To (R) (X) (Bc) Tapping limit (pu)

1 1 2 0.01938 0.05917 0.0264 1.0 0.92

2 1 8 0.05403 0.22304 0.0246 1.0 0.55
3 2 3 0.04699 0.19790 0.0219 1.0 0.41
4 2 6 0.05811 0.17632 0.0187 1.0 0.70
5 2 8 0.05695 0.17388 0.0170 1.0 0.50
6 6 3 0.06701 0.17103 0.0173 1.0 0.04
7 8 6 0.01335 0.04211 0.0064 1.0 0.42
8 6 7 0.00000 0.20912 0.0000 1.0 0.42
9 6 9 0.00000 0.55628 0.0000 1.0 0.20
10 8 4 0.00000 0.25202 0.0000 1.0 0.52
11 4 11 0.09498 0.19890 0.0000 1.0 0.09
12 4 12 0.12291 0.25581 0.0000 1.0 0.10
13 4 13 0.06613 0.13027 0.0000 1.0 0.20
14 7 5 0.00000 0.17615 0.0000 1.0 0.01
15 7 9 0.00000 0.08450 0.0000 1.0 0.42
16 9 10 0.03181 0.08450 0.0000 1.0 0.08
17 9 14 0.12711 0.27038 0.0000 1.0 0.12
18 11 10 0.08205 0.19207 0.0000 1.0 0.05
19 12 13 0.22092 0.19988 0.0000 1.0 0.025
20 13 14 0.17093 0.34802 0.0000 1.0 0.08
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10.A.2 Data for IEEE 14-bus, 20-line system (100 MVA Base) 193

Table 10.A.4 Data bus.

Load (pu) Generation (pu)

Bus number Active Reactive Active Reactive

1 0.0000 0.0000 1.6730 2.0000

2 0.2170 0.1270 0.7000 0.0000
3 0.9420 0.1900 0.5000 0.0000
4 0.1120 0.0750 0.0000 0.0000
5 0.0000 0.0000 0.0000 0.0000
6 0.4780 0.0390 0.0000 0.0000
7 0.0000 0.0000 0.0000 0.0000
8 0.0760 0.0180 0.0000 0.0000
9 0.2950 0.1660 0.0000 0.0000
10 0.0900 0.0580 0.0000 0.0000
11 0.0350 0.0180 0.0000 0.0000
12 0.0610 0.0160 0.0000 0.0000
13 0.1350 0.0580 0.0000 0.0000
14 0.0580 0.0580 0.0000 0.0000

13
12 14

11

1 4 7

10
8
5
6
9
3

Figure 10.A.2 IEEE 14-bus test system.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
194 10 Transmission System Reliability Evaluation Including Security

10.A.3 Data for IEEE 25-bus, 35 line system (100 MVA Base)
Table 10.A.5 Line data.

Bus number Line flow

Line number From To (R) (X) (Bc) Tapping limit (pu)

1 1 3 0.0720 0.2876 0.0179 1.0 0.460

2 1 16 0.0290 0.1379 0.0337 1.0 0.490
3 1 17 0.1012 0.2794 0.0148 1.0 0.170
4 1 19 0.1487 0.3897 0.0224 1.0 0.08
5 1 23 0.1085 0.2245 0.0573 1.0 0.360
6 1 25 0.0753 0.3593 0.0873 1.0 0.420
7 2 6 0.0617 0.2935 0.0186 1.0 0.300
8 2 7 0.0511 0.2442 0.0155 1.0 0.390
9 2 8 0.0597 0.2763 0.0175 1.0 0.380
10 3 13 0.0564 0.1487 0.0085 1.0 0.260
11 3 14 0.1183 0.3573 0.0185 1.0 0.490
12 4 19 0.0196 0.0514 0.0113 1.0 0.430
13 4 20 0.0382 0.1007 0.0220 1.0 0.440
14 4 21 0.0970 0.2547 0.0558 1.0 0.260
15 5 10 0.0479 0.2372 0.0577 1.0 0.370
16 5 17 0.0144 0.1269 0.1335 1.0 1.03
17 5 19 0.0929 0.2442 0.0140 1.0 0.530
18 6 13 0.0263 0.0691 0.0040 1.0 0.100
19 7 8 0.0529 0.1465 0.0078 1.0 0.075
20 7 12 0.0364 0.1736 0.0110 1.0 0.150
21 8 9 0.0387 0.1847 0.0118 1.0 0.075
22 8 17 0.0497 0.2372 0.0572 1.0 0.080
23 9 10 0.0973 0.2691 0.0085 1.0 0.110
24 10 11 0.0898 0.2359 0.0135 1.0 0.080
25 11 17 0.1068 0.2807 0.0161 1.0 0.0250
26 12 17 0.0460 0.2196 0.0135 1.0 0.026
27 14 15 0.0281 0.0764 0.0044 1.0 0.230
28 15 16 0.0256 0.0673 0.0148 1.0 0.130
29 17 18 0.0806 0.2119 0.0122 1.0 0.200
30 18 19 0.0872 0.2294 0.0132 1.0 0.050
31 20 21 0.0615 0.1613 0.0354 1.0 0.150
32 21 22 0.0141 0.1087 0.0238 1.0 0.200
33 22 23 0.2250 0.3559 0.0169 1.0 0.160
34 22 24 0.0970 0.2595 0.0567 1.0 0.095
35 24 25 0.0470 0.1458 0.0317 1.0 0.105
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10.A.3 Data for IEEE 25-bus, 35 line system (100 MVA Base) 195

Table 10.A.6 Bus data.

Load (pu) Generation (pu)

Bus no. Active Reactive Active Reactive

1 2.0000 0.6500 2.6730 3.0000

2 0.1000 0.0300 0.9930 1.0000
3 0.5000 0.1700 1.4719 1.0000
4 0.3000 0.1000 0.3910 1.0000
5 0.2500 0.0800 1.9300 1.0000
6 0.1500 0.0500 0.0000 0.0000
7 0.1500 0.0500 0.0000 0.0000
8 0.2500 0.0000 0.0000 0.0000
9 0.1500 0.0500 0.0000 0.0000
10 0.1500 0.0500 0.0000 0.0000
11 0.0500 0.0000 0.0000 0.0000
12 0.1000 0.0000 0.0000 0.0000
13 0.2500 0.0800 0.0000 0.0000
14 0.2000 0.0700 0.0000 0.0000
15 0.3000 0.1000 0.0000 0.0000
16 0.3000 0.1000 0.0000 0.0000
17 0.6000 0.2000 0.0000 0.0000
18 0.1500 0.0500 0.0000 0.0000
19 0.1500 0.0500 0.0000 0.0000
20 0.2500 0.0800 0.0000 0.0000
21 0.2000 0.0700 0.0000 0.0000
22 0.2000 0.0700 0.0000 0.0000
23 0.1500 0.0500 0.0000 0.0000
24 0.1500 0.0500 0.0000 0.0000
25 0.2500 0.0800 0.0000 0.0000
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
196 10 Transmission System Reliability Evaluation Including Security

3 13
14 6
16 15

1 17 12

19 18 11
25
24
5
10

22 21 20 7
23 9
4

Figure 10.A.3 IEEE 25-bus test system.

References

Ajitha, A., Goel, M., Assudani, M. et al. (2022). Design and development of residential sector load
prediction model during COVID-19 pandemic using LSTM based RNN. Electric Power Systems
Research 212: 108635. https://doi.org/10.1016/j.epsr.2022.108635.
Ajjarapu, V. and Christy, C. (1992). The continuation power flow: a tool for steady state voltage stability
analysis. IEEE Transactions on Power Systems 7 (1): 416–423. https://doi.org/10.1109/59.141737.
Amjady, N. and Esmaili, M. (2003). Voltage security assessment and vulnerable bus ranking of power
systems. Electric Power Systems Research 64 (3): 227–237. https://doi.org/10.1016/S0378-
7796(02)00196-7.
Armaghani, S., Naghshbandy, A.H., and Shahratash, S.M. (2020). Application of a new fast analytical
approach for post-contingency state estimation of “N-1” transmission line outage in online
vulnerability assessment of large-scale transmission networks with wind power resources.
International Journal of Electrical Power & Energy Systems 123: 106304. https://doi.org/10.1016/j
.ijepes.2020.106304.
Arya, L.D. and Verma, H.K. (1996). A method for tracing PV-curve for voltage stability analysis with
voltage dependent load. Electric Machines and Power System 24 (6): 583–596. https://doi.org/10
.1080/07313569608955696.
Arya, L.D., Pande, V.S., and Tare, R.S. (2002). Contingency selection based on vanishing eigen value
concept. Journal of the Institution of Engineers (India) 83: 213–217.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 197

Arya, L.D., Titare, L.S., and Kothari, D.P. (2006). Determination of probabilistic risk of voltage collapse
using radial basis function (RBF) network. Electric Power Systems Research 76 (6-7): 426–434. https://
doi.org/10.1016/j.epsr.2005.09.011.
Arya, L.D., Singh, P., and Titare, L.S. (2012). Optimum load shedding based on sensitivity to enhance
static voltage stability using differential evolution. Swarm and Evolutionary Computation 6: 25–38.
https://doi.org/10.1016/j.swevo.2012.06.002.
Bahmanyar, A.R. and Karami, A. (2014). Power system voltage stability monitoring using artificial
neural networks with a reduced set of inputs. International Journal of Electrical Power & Energy
Systems 58: 246–256. https://doi.org/10.1016/j.ijepes.2014.01.01.
Berizzi, A., Bovo, C., Cirio, D. et al. (2009). On-line fuzzy voltage collapse risk quantification. Electric
Power Systems Research 79 (5): 740–749. https://doi.org/10.1016/j.epsr.2008.10.010.
Billinton, R. and Allan, R.N. (1984). Reliability Evaluation of Power System. Plenum Press, Inc. (Book).
Choube, S.C., Arya, L.D., and Sakravadia, D.K. (2001). Contingency selection based on singular value
decomposition. Journal of the Institution of Engineers (India) 81 (4): 226–229.
Coelho, E.P.R., Alves, R.J.M., Paiva, M.H.M. et al. (2022). Topological multi-contingency screening
based on current flow betweenness. Electric Power Systems Research 203: 10709. https://doi.org/10
.1016/j.epsr.2021.107609.
Dester, M. and Castro, C.A. (2009). Multi-criteria contingency ranking method for voltage stability.
Electric Power System Research 79 (1): 220–225. https://doi.org/10.1016/j.epsr.2008.05.023.
Hagh, M.T., Gargari, M.Z., and Pakdel, M.J.V. (2018). Sequential analysis of optimal transmission
switching with contingency assessment. IET Generation, Transmission & Distribution 12 (6):
1390–1396. https://doi.org/10.1049/iet-gtd.2017.0435.
Haykin, S. (2003). Neural Network: A Comprehensive Foundation. Pearson Education, Second Edition,
Inc. (Book).
Heidari, H., Hagh, M.T., and Salehpoor, P. (2022). Accurate, simultaneous and real-time screening of
N-1, N-k, and N-1-1 contingencies. International Journal of Electrical Power & Energy Systems 136:
107592. https://doi.org/10.1016/j.ijepes.2021.107592.
Javan, D.S., Mashhadi, H.R., and Rouhani, M. (2013). A fast static security assessment method based
on radial basis function neural networks using enhanced clustering. International Journal of
Electrical Power & Energy Systems 44 (1): 988–996. https://doi.org/10.1016/j.ijepes.2012.08.014.
Kim, H. and Singh, C. (2005). Power system probabilistic security assessment using bayes classifier.
Electric Power Systems Research 74 (1): 157–165. https://doi.org/10.1016/j.epsr.2004.10.004.
Kothari, D.P., Nagrath, I.J., and Saket, R.K. (2022). Modern Epower System Analysis, 5e. New Delhi,
India: McGraw Hill (Book).
Kundur, P. (1994). Power System Stability and Control. McGraw Hill, Inc. (Book).
Lee, C.Y., Tsai, S.H., and Wu, Y.K. (2010). A new approach to the assessment of steady-state voltage
stability margins using the P–Q–V curve. International Journal of Electrical Power and Energy
Systems 32 (10): 1091–1098. https://doi.org/10.1016/j.ijepes.2010.06.005.
Naik, S.D., Khedkar, M.K., and Bhat, S.S. (2015). Effect of line contingency on static voltage stability
and maximum loadability in large multi bus power system. International Journal of Electrical Power
& Energy Systems 67: 448–452. https://doi.org/10.1016/j.ijepes.2014.12.002.
Nan, L., Liu, Y., Wu, L. et al. (2021). Graph theory based N-1 transmission contingency selection and its
application in security-constrained unit commitment. Journal of Modern Power Systems and Clean
Energy 9 (6): 1458–1467. https://doi.org/10.35833/MPCE.2019.000602.
Ni, M., McCalley, J.D., Vittal, V., and Tayyib, T. (2003). On-line risk-based security assessment. IEEE
Transactions on Power Systems 18 (1): 258–265. https://doi.org/10.1109/TPWRS.2002.807091.
Rubinstein, R.Y. (1981). Simulation and Monte-Carlo Methods. John Wiley and Sons, Inc. (Book).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
198 10 Transmission System Reliability Evaluation Including Security

Sherstinsky, A. (2020). Fundamentals of recurrent neural network (RNN) and long short-term memory
(LSTM) network. Physica D: Nonlinear Phenomena 404: 132306. https://doi.org/10.1016/j.physd
.2019.132306.
Singh, P, Purey, P., Titare, L.S. et al., (2017). Enhancement of voltage security by MW-generation
rescheduling based on sensitivities using black hole. 2017 International Conference on Information,
Communication, Instrumentation and Control (ICICIC), 1-5, https://doi.org/10.1109/ICOMICON
.2017.8279045.
Singh, P., Titare, L.S., Choube, S.C., and Arya, L.D. (2018). Security assessment accounting
uncertainties in line parameters and control variables with the considerations of transmission line
unavailability. Journal of Electrical Systems and Information Technology 5 (3): 576–593. https://doi
.org/10.1016/j.jesit.2017.10.002.
Singh, P., Arya, R., Titare, L.S., and Arya, L.D. (2021). Optimum value based under voltage load
shedding strategies. Australian Journal of Electrical and Electronics Engineering 18 (3): 146–160.
https://doi.org/10.1080/1448837X.2021.1945719.
Singh, P., Arya, R., Titare, L.S. et al. (2023). Value aided optimal load shedding accounting voltage
stability consideration employing crow search algorithm with modification based on lampinen’s
criterion. Applied Soft Computing 143: 110391. https://doi.org/10.1016/j.asoc.2023.110391.
Srivani, J. and Swarup, K.S. (2008). Power system static assessment and evaluation using external
system equivalents. International Journal of Electrical Power & Energy Systems 30 (2): 83–92. https://
doi.org/10.1016/j.ijepes.2007.06.008.
Vaahedi, E., Fuchs, C., Xu, W. et al. (1999). Voltage stability contingency screening and ranking. IEEE
Transactions on Power Systems 14 (1): 256–265. https://doi.org/10.1109/59.744541.
Verma, K. and Niazi, K.R. (2012). Supervised learning approach to online contingency screening and
ranking in power systems. International Journal of Electrical Power & Energy Systems 38 (1): 94–114.
https://doi.org/10.1016/j.ijepes.2011.12.025.
Verma, M.K. and Shrivastava, S.C. (2005). Approach to determine voltage control areas considering
impact of contingencies. IEE-Proceedings Generation, Transmission and Distribution 152 (3):
342–350. https://doi.org/10.1049/ip-gtd:20041134.
Wang, L., Hu, B., Xie, K. et al. (2020). Screening model of incremental risk events for reliability analysis
of transmission system. International Journal of Electrical Power & Energy Systems 120: 105995.
https://doi.org/10.1016/j.ijepes.2020.105995.
Zarate, L.A.L., Castro, C.A., Ramos, J.L.M., and Ramos, E.R. (2006). Fast computation of voltage
stability security margins using non-linear programming techniques. IEEE Transactions on Power
Systems 21 (1): 19–27. https://doi.org/10.1109/TPWRS.2005.860945.
Zhao, S. and Singh, C. (2018). A hybrid method for reliability evaluation of line switching operations.
Electric Power Systems Research 163, Part A: 365–374. https://doi.org/10.1016/j.epsr.2018.07.001.
Zhu, Y. and Liu, G. (2022). Application of triangle count in branch contingency screening.
International Journal of Electrical Power & Energy Systems 135: 107392. https://doi.org/10.1016/j
.ijepes.2021.107392.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
199

11

Probabilistic Voltage Security Assessment and Enhancement Using

Rescheduling of Reactive Power Control Variables
Lakhan Singh Titare 1 , Aanchal Singh S. Vardhan 2 , Liladhar Arya 3 , and Devkaran Sakravdia 1
1
Department of Electrical Engineering, Government Engineering College, Ujjain, Madhya Pradesh, India
2 Department of Electrical Engineering, National Institute of Technology, Jamshedpur, Jharkhand, India
3
Department of Electrical Engineering, Shri G.S. Institute of Technology & Science, Indore, Madhya Pradesh, India

11.1 Introduction
Power system security studies, encompassing evaluation and regulation, are an indispensable
aspect of contemporary energy management systems. Power utilities are confronting sub-
stantial obstacles in preserving the excellence and dependability of power provision due to the
ever-expanding interconnections and loads in expansive power network infrastructures. Economic
constraints have compelled power companies to run their generation and transmission systems at
full capacity. The enhanced efficacy of protective mechanisms, Static Var Systems, rapid turbine
valving, and automated voltage regulators has notably amplified the transient stability limits of
the system. Consequently, the power system can now transmit greater real power over extended
distances. Nevertheless, this has engendered fresh challenges in upholding the mandated bus
voltages, as the system is now constrained by voltage stability (Taylor 1994). Voltage security
pertains to a power system’s capability to sustain stable and acceptable voltage levels, ensuring the
smooth operation and functionality of electrical equipment and devices under various operating
conditions, including contingencies or adverse changes in the system. It is crucial for ensuring
the proper functioning and reliability of the power grid. If a power system’s condition restores to
the same or a state that is somewhat similar to the pre-disturbance state of operation after any
little disturbance, the operating point is considered to be small disturbance stable (Kundur 1994;
Concordia 1991; IEEE System Dynamic Performance Subcommitte n.d.). Additionally, it has
long been accepted practice to assess reliability indicators during the planning stage using a
probabilistic technique based on security evaluation including correctability (Sebastian Rios et al.
1998; Arya et al. 2009; Billinton and Wenyuan 1991). An approach for assessing a composite
system’s probability of failure on the basis of maximum load and system ability to repair proposed
in (Arya et al. 2000). If failure probability, or, as it is sometimes referred to, probability of load
shedding is lower, the system is more resistant to operating constraint violations. A large number
of power flow studies under different loading conditions and outages are required for probability
of load shedding evaluation. The security of the system in its operational state cannot be evaluated
using this type of indicator. Such failure probability indicator may be used in online monitoring
applications in power systems owing to advancements in artificial neural networks (Singh
et al. 2021; Arya et al. 2007). The majority of research studies that evaluated reliability indicator

Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
200 11 Probabilistic Voltage Security Assessment and Enhancement Using Rescheduling of Reactive Power Control Variables

did not consider voltage security from the perspective of static voltage stability. When evaluating
the composite power system’s adequate suitability while taking voltage and line flow limits into
account, Billinton and Khan (1992) employed the Linear Programming model. For evaluating
the reliability of a transmission system, Pereira and Pinto (1992) developed Monte-Carlo (MC)
sampling with a variance reduction technique. A novel method for evaluating the reliability of
interconnected power systems is introduced in the research paper (Deng and Singh 1992), which
includes planned outages and frequency calculations for load loss, while considering various
factors impacting power supply interruptions. Billinton and Li (1993) proposed a system state
transition sampling methodology to directly compute reliability of composite system. A basic
framework has been designed that can be used in the probabilistic risk estimation of cascading fail-
ures (Henneaux 2015). Kirschen and Jayaweera (2007) introduced a methodology for comparing
the probabilistic and deterministic assessment of transmission system security. Voltage stability
constraints were taken into account while evaluating the reliability indices (Melo et al. 1997;
Billinton and Aboreshaid 1998). The method for calculating voltage collapse-linked indicators
and their effect on power system soundness indicators based on regaining system-solving capacity
through load curtailment was probably first presented in (Melo et al. 1997). This has been
accomplished by employing a univariate search technique to optimize the static voltage stability
limit and a predictor-corrector-based continuation power flow technique to compute the threshold
value of stable voltage (Capitaanescu and Van Cutsem 2002). A novel collapse prediction index
(NCPI) is proposed in order to evaluate the power system’s voltage stability circumstances as
well as the critical line conditions. In addition, the sensitivity hypotheses of the currently used
indices and evaluates how they affect the forecasting of voltage collapse (Mokred et al. 2023). A
new methodology was developed that considers voltage stability into account while evaluating
the adequate capabilities of a transmission network (Xia et al. 2021). A novel sensitivity method
is employed to choose the optimal control variables for addressing the power system voltage
instability. In order to account for changes in power system operation, a sensitivity analysis is
carried out on a voltage stability index called Thevenin-Based Voltage Stability Margin (TVSM)
(Alzaareer et al. 2020). The methodology is built up to incorporate infeed uncertainty into the
evaluation of the security of the power system. It enables the evaluation of the operational security
of a power system’s potential state under uncertainty resulting from various contingencies and
loads (de Jong et al. 2018). Benidris and Mitra (2015) presented the formulation of expressions
and sensitivity evaluations of the severity index with respect to the voltage and reactive power
constraints and the implications of these constraints on the reliability of the power system. This
chapter introduces a novel methodology for assessing and controlling voltage security by adjusting
reactive power control variables and, when required, implementing load shedding. The technique
involves computing voltage stability limits for single- and double-line outage scenarios using
the continuation power flow method. Based on these results, the probabilistic insecurity index
(PII) is calculated using the cut-set technique. To proactively manage voltage security, the PII
sensitivity is determined concerning various settings of reactive power control variables and
loading conditions. This sensitivity analysis allows for preventive control measures. Developed
equations in Section 11.2 are used to calculate the reactive power control variable settings in
emergency situations and load-shedding conditions. The obtained results are utilized to train an
Artificial Neural Network (ANN) to predict PII for any operating condition. The trained ANN is
also employed to compute the sensitivity of PII with respect to reactive power control variables
and load shedding at different buses. Using the ANN and the derived sensitivities, voltage security
can be controlled effectively by adjusting reactive power control variables or implementing
load shedding at specific buses when necessary. The advantage of employing an ANN lies in its
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
11.2 Computation of Probabilistic Insecurity Index (PII) Using Cut-set Technique 201

suitability for online applications, making the proposed methodology practical and efficient for
real-time voltage security assessment and control.

11.2 Computation of Probabilistic Insecurity Index (PII) Using

Cut-set Technique

The computation of the PII involves assessing power system’s vulnerability and security across
various contingencies and operating conditions. The PII is a measure of the probability of voltage
instability or collapse in the system. The cut set is a widely used and powerful technique in the
field of network analysis, particularly in assessing the security and reliability of power networks.
The cut-set method helps in understanding the various modes of system failure. A cut set is a set
of transmission lines in the network such that their failure would lead to the disconnection of
the network into two or more separate sections. By identifying these cut sets, one can determine
the most vital lines or combinations of lines that, if interrupted, could cause system failure or a
significant reduction in its performance. The cut-set method is particularly useful for designing
robust and reliable power systems. It aids in determining redundancy requirements, assessing the
impact of component failures, and designing effective preventive control measures. Moreover, the
cut-set technique can be readily executed on a digital computer, allowing for efficient and practical
analyses of large and complex power networks. A minimal cut set refers to a set of transmission
lines that, when failed, lead to system failure. However, when any single element of the set remains
operational, it does not cause the system to fail. Consequently, every constituent of a minimum cut
set must be in a state of failure for the system failure to be initiated. The failure probability (F P ) of
a power network can be expressed as follows:

FP = P[Cs1 + Cs2 + Cs3 + … + Csn ] (11.1)

where

Cs1 , Cs2 ,…,Csn are minimal cut sets.

While it is theoretically possible to perform exact evaluations, doing so is often impractical and
time-consuming, especially for large power networks. Therefore, in practice, failure probability
assessments often resort to approximations, sacrificing some precision to achieve much faster
results. Power systems with low failure rates and high repair rates for individual lines, the
introduced error is typically negligible and does not significantly impact the availability data for
transmission lines. Thus, the failure probability is estimated by keeping the first-order terms in
the expansion of Eq. (11.1) as approximated below:
∑
n
FP = P(Csi ) (11.2)
j=1

The second approximation involves not considering cut sets with orders higher than a certain
number. The number of components within a cut set corresponds to its order. In this chapter, only
cut sets up to second order are taken into consideration. This decision is based on the very low
probability of more than two transmission lines failing simultaneously under usual loading condi-
tions. Additionally, a next-higher-order cut set may not necessarily be a minimal cut set, which is
required for Eq. (11.1) to be valid. Therefore, Eq. (11.2) is used to calculate the probability of failure
with this approximation. To model the operation of each transmission line, a two-state modal is
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
202 11 Probabilistic Voltage Security Assessment and Enhancement Using Rescheduling of Reactive Power Control Variables

λi Figure 11.1 Dual-state model for individual

transmission lines.
Up state Down state
μi

used, comprising an “up state” representing the line being in operation and a “down state” indicat-
ing that the line is either in a failed state or scheduled to be off. The failure and repair rates of the
transmission lines are denoted by the symbols “𝜆” and “𝜇,” respectively, as depicted in Figure 11.1.
This modal is utilized to determine the long-term availability and unavailability of transmission
lines. The corresponding expressions for availability and unavailability are provided in Billinton
and Allan 1983:
𝜇Ti
ATi = (11.3)
(𝜆Ti + 𝜇Ti )
𝜆Ti
ATi = (11.4)
(𝜆Ti + 𝜇Ti )
where ATi and ATi Indicate the unavailability and availability of the transmission line, respectively.
The determination of minimal cut sets (first and second order) for a given load is achieved through
a comprehensive analysis, wherein the load is compared against the static voltage stability limit in
the base case and also under diverse line outage scenarios. This detailed investigation allows us to
identify the critical cut sets that exert significant influence on the power system’s voltage stabil-
ity under various operating conditions. The static voltage stability limits under circumstances of
single or double-line outage are determined by applying the continuation power flow technique
(Arya and Verma 1996). For the purpose of securing proper system functioning and stability, it is of
paramount importance to sustain ample stability margins. Subsequently, utilizing the outcome of
the continuation power flow, prepare a table displaying capacity outages, indicating the permissi-
ble loading levels and the corresponding network conditions. By analyzing various load levels, one
can identify the relevant cut sets from the capacity availability table. The PII of the transmission
system is then computed using relation (11.2), representing the probability of failure. It is impor-
tant to emphasize that the loadability limit is influenced by the settings of reactive power control
variables. As a result, the failure probability (PII) is assessed for different configurations of reactive
power control variables. The computational procedure is visually presented in Figure 11.2 using
a comprehensive flow chart. This computational methodology provides a systematic and compre-
hensive analysis of the voltage security of the system. It aids in identifying potential stability issues,
assessing their probabilities, and optimizing the control variables to enhance the overall voltage
stability and security of the power transmission network.

11.3 Computation of Probabilistic Insecurity Index (PII) Sensitivity

using ANN

The ANN is trained using the backpropagation algorithm, enabling it to learn from a set of instances
obtained from Section 11.2 (Fu 1994). These instances cover a wide range of operating conditions,
offering a comprehensive understanding of the system’s behavior. The generic network architecture
consists of one hidden layer, one input layer, and one output layer. The input layer’s units are equal
to the number of reactive power control variables and total load buses, while the output layer has a
single unit that gives the PII value. Additionally, it is postulated that the neurons in the hidden layer
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
11.3 Computation of Probabilistic Insecurity Index (PII) Sensitivity using ANN 203

Start

Execute the AC power flow at the base case and acquire the static voltage
stability limit.

Conduct the continuation power flow analysis for single and double line
outage conditions to determine the static voltage stability limit for each
contingency.

Generate a probability of capacity outage table, including the static voltage

stability limit and various contingencies.

Compute the cut-sets Cis1, Cis2, ...,Cisr for ith loading condition for different
peak load considerations i = l, 2, ...NL

Calculate failure probability (Fp) using Eq. (2.2)

Iteratively modify the reactive power control setting one by one, increasing
and decreasing in steps, until reaching the limits V1r = V 0r + ΔVr

If
No all change
have been
made
?

Yes

Stop

Figure 11.2 Flowchart for computing probabilistic insecurity index.

adopt a sigmoidal activation function, while the neuron in the output layer follows a nonsigmoidal
(linear) activation function. Once trained, the ANN provides real-time online assessment of voltage
security, enabling quick identification of potential voltage instability issues. The network equations
corresponding to the architecture depicted in Figure 11.3 can be expressed as:

∑
m
O= (Wjk Hj ) (11.5)
j=1
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
204 11 Probabilistic Voltage Security Assessment and Enhancement Using Rescheduling of Reactive Power Control Variables

V1 W11 Σ H1
W11
Σ
V2 H2
W21
Σ
VNCB O
Σ

P1
W1j Wjk
P2

Σ
PNLB Wij Hj

Input layer Hidden layer Output layer

Figure 11.3 Generalized diagram of backpropagation network for the computation of probabilistic voltage
security.

The output of the neuron located at the jth position in the hidden layer is as follows:
1
Hj = ( ) (11.6)
1 + e−Netj
j = 1,2, 3, … , n

∑
n
Netj = Wij Ui (11.7)
j=1

where

W ij is the weight that connects the ith input node to the jth hidden neuron.
U i is the input variable at ith node (Haykin 2003). The following formulae provide the weight
change W jk .

ΔWjk = 𝜂𝛿Hj (11.8)

𝛿 = (T − O) (11.9)

T – is the network’s target value

O – is the network’s output value,
𝜂 is the learning rate that lies between (0, 1). Using the backpropagation approach, the following
weight change ΔW ij expression is given for the hidden layer:

ΔWij = 𝜂𝛿Ui (11.10)

where 𝛿 j is the slope of error and is written as follows:

𝛿j = 𝛿ΔWjk Hj (1 − Hj )Ui (11.11)

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
11.4 Voltage Security Enhancement using a Monovariable Approach 205

U i is the component of the input vector [U], where

[U]T = [V1 , V2 , … , VNCB , P1 , P2 , … , PNLB ]T
After successfully training the ANN, the network Eqs. (11.5), (11.6), and (11.7) are employed to
compute the sensitivities of the output variable “O” concerning the input variables, which can be
either the reactive power control variables or load shedding. The sensitivity expression is obtained
by applying the chain rule of differentiation. This mathematical technique allows us to compute
how changes in the input variables occur in the context of the trained Artificial Neural Network
(ANN). The chain rule enables us to analyze the complex relationship between the network’s layers
and efficiently determine the sensitivities of the output with respect to various input variables as
follows:
( )( )( )
𝜕PII ∑ 𝜕PII 𝜕Hj 𝜕Netj
m
= = SPIIi (11.12)
𝛿Ui j=1
𝜕Hj 𝜕Netj 𝜕Ui

where SPIIi is the sensitivity of PII with respect to U i . The expression for sensitivity given by alge-
braic manipulation of network equations is as follows:
∑
m
SPIIi = Wjk Hj (1 − Hj )Wij (11.13)
j=1

The evaluation of sensitivities based on Eq. (11.13) is characterized by its inherent simplicity and
efficiency. By interpreting expression (11.13) as the sum of all path gains originating from an input
node and leading to the output node. Each path gain is obtained by multiplying the weights and
the gain of the nodes {H j (1 − H j )} encountered along that specific path. The method for computing
sensitivity is highly versatile and can be extended to accommodate multiple hidden layers within
the neural network. In Section 11.4, algorithm not only ensures enhanced voltage stability but
also incorporates the capability of implementing load shedding as a viable and effective corrective
measure to safeguard the power system.

11.4 Voltage Security Enhancement using a Monovariable

Approach
This approach focuses on controlling a single variable to improve voltage security. The variable
chosen for control could be reactive power or load shedding. Trained ANN is employed for
real-time voltage security calculation. Subsequently, sensitivities are assessed to facilitate cor-
rective rescheduling. This is achieved by defining a threshold value for PII. Whenever the PII
surpasses this limit, the calculated sensitivities concerning reactive power control variables are
employed to perform corrective rescheduling. The reactive power control variable with the highest
sensitivity is selected for the rescheduling task. The adjustment for this reactive power control
variable is obtained as follows:
(PIIo − PIIth )
ΔVr = (11.14)
SPIIr
where
PIIo is the current insecurity index obtained from the ANN,
PIIth is the threshold value of the indicator, and
SPIIr is the sensitivity of indicator with respect to the “r” control variable.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
206 11 Probabilistic Voltage Security Assessment and Enhancement Using Rescheduling of Reactive Power Control Variables

In the event that the control correction obtained through Eq. (11.14) exceeds the limit, it can
be adjusted to the limiting value. Following that, modified sensitivities are reassessed, and an
alternative control variable is chosen for corrective action. The monovariable rescheduling pro-
cess continues until the PII is brought within acceptable limits. Subsequently, a continuation load
flow program is executed to verify that all voltage levels are within the prescribed bounds. If it is
observed that all control variables have been fully utilized and yet the PII still violates the limit,
then the sensitivities of PII with respect to load shedding at the load buses are evaluated, and
selection of a particular bus with the greatest sensitivity is made for load shedding, as outlined
below:
(PIIo − PIIth )
ΔPi = (11.15)
SPIIDi
SPIIDi is the sensitivity of indicator (PII) with respect to load shedding at ith bus.
The limit on the sensitivity of PII with respect to load shedding at a bus should not surpass a
certain maximum threshold (e.g. 80% of the total load at that bus, considering 20% as reserved
for emergency services). If the total load shedding at a bus is inadequate to meet this limit, the
next bus with the highest sensitivity is chosen for additional load shedding. This iterative process
continues until the PII value falls within the acceptable range. In essence, the trained ANN acts as
a system-wide relay, not only identifying insecure conditions but also providing recommendations
for corrective measures (Zarate et al. 2006). This intelligent approach ensures the system’s voltage
stability and enhances overall security.

11.5 Results and Discussion

The algorithm formulated for voltage security control, employing the rescheduling of reactive
power control variables or load shedding, has been applied to two IEEE standard test systems:
14-bus and 25-bus (Bijwe et al. 1993).

11.5.1 14-bus System

This system comprises five PV buses, and their voltages can be regulated between (0.95 ≤ V i ≤
1.15pu) to adjust the voltage of load buses. The availability of lines from line No. 1 to line No.
20 is considered as: 0.99725754, 0.99037152, 0.99027586, 0.99592631, 0.99097267, 0.99825389,
0.99772435, 0.99138266, 0.99633603, 0.99472818, 0.99927413, 0.99327504, 0.99733418, 0.99792642,
0.99078241, 0.99546628, 0.99662247, 0.99016989, 0.99906437, and 0.9908926 (Billinton and Allan
1984). Table 11.1 shows a set of samples used to train the ANN, which enabled the computation of
the PII using the cut-set technique and the ANN, along with the percentage errors. The network
underwent training with 150 instances, and only 18 instances with the highest error magnitudes
are listed in Table 11.1. Satisfactory results were achieved by incorporating two hidden layers, each
consisting of ten sigmoidal neurons. The output neuron is assumed to be linear. In Table 11.2,
various weights namely W j2, k , W j1, j2 , and W i, j1 , connect the neurons between the hidden layer-2
and the output layer, hidden layer-1 and hidden layer-2, and input node and hidden layer-1,
respectively. The input layer includes variables V 1 , V 2 , V 3 , and Pd namely PV-bus voltages
and overall load of the system. Table 11.3 shows the PII obtained using the trained network and
the cut-set approach, along with the percentage errors for test cases. There are no prescribed
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
11.5 Results and Discussion 207

Table 11.1 A set of samples for training of ANN. PII computed using the cut-set technique and
Backpropagation Algorithm Network for a 14-bus system.

PII (by cut-set

S. No V 1 (pu) V 2 (pu) V 3 (pu) P d (pu) technique) PII (by ANN) % Error

1 0.95 0.95 0.95 6.031 0.076020248 0.078049989 −2.67

2 0.95 0.95 1 6.0271 0.067941537 0.066929208 1.49
3 0.95 1 1 5.9883 0.064533306 0.062474694 3.19
4 1 1 1 6.8166 0.072356281 0.073050901 −0.96
5 1 1 1.05 6.7161 0.060812244 0.058239886 4.23
6 1 1.05 1.05 6.6445 0.056262153 0.055018759 2.21
7 1.05 1.05 1.05 7.5022 0.067948321 0.066895122 1.55
8 1.05 1.05 1.1 7.4461 0.056266172 0.056620649 −0.63
9 1.05 1.1 1.1 6.8277 0.044472754 0.042885077 3.57
10 1.1 1.1 1.1 83,266 0.064225551 0.063531915 1.08
11 1.1 1.1 1.15 9.0376 0.071610771 0.068288031 4.64
12 1.1 1.15 1.15 7.3169 0.048195524 0.048831705 −1.32
13 1.15 1.15 1.15 9.1367 0.060812591 0.060338253 0.78

regulations governing the ratio between training and testing instances. The training instances
chosen for this research have successfully encompassed a wide range of operational scenarios
for the power system. Within this scope, supplementary testing instances have been chosen,
alongside the training instances. The trained network was tested with 250 instances, and only
18 instances with the highest error magnitudes are listed in Table 11.3. In this range, additional
testing instances have been selected in addition to the training instances. The PII values must not
exceed a predefined threshold set at 0.060812590. This specific threshold is chosen to ensure that,
even during outage scenarios, the distance to the saddle-node bifurcation point (measured in mega
volt amperes (MVA)) remains at a minimum of 19.99% of the static voltage stability limit. This
criterion is indispensable and may vary based on the operational philosophy and specific needs
of the individual power utility. The threshold value holds significant importance in Eqs. (11.14)
and (11.15) as they are utilized to evaluate control measures. The percentage error of PII, as
computed by trained ANN, is less than ±5%. If further error reduction is desired, increasing the
size of the ANN is an option. This is achieved by augmenting the number of neurons in the hidden
layer and expanding the number of hidden layers. For a specific operating setting, corrective
reactive power rescheduling was demonstrated with total load Pd = 6.0271 pu and V 1 = 0.95 pu,
V 2 = 0.95 pu, V 3 = 0.95 pu. It was observed that the stability margin obtained was less than
20% with a PII of 0.072356281, exceeding the specified threshold value. Sensitivities of PII were
computed using Eq. (11.14), and it was found that Bus No. 2 is selected for voltage correction
(ΔV2 = 0.1614 pu) due to its highest sensitivity. This correction brought the PII within the desired
limit of 0.060812590 as the final value. Another case was considered with values: V 1 = 1.15 pu,
V 2 = 1.15 pu, V 3 = 1.15 pu, total load Pd = 9.3672 pu, and despite correction attempts using
Eq. (11.14) for rescheduling of reactive power at the most sensitive buses, the PII remained above
the specified limit (0.076020248). Hence, sensitivities were assessed concerning load shedding at
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
208 11 Probabilistic Voltage Security Assessment and Enhancement Using Rescheduling of Reactive Power Control Variables

Table 11.2 Final weights obtained after ANN training for the 14-bus system.

W j1, j2 (Weight W j1, j2 (Weight

W ij (Weight between between W j2, k (Weight
between j1th hidden j1th hidden between
ithinput layer-1 layer-1 j2th hidden
node and neuron and neuron and layer-2
jth hidden Values of j2th hidden Values of j2th hidden Values of neuron and Values of
layer-1 weights layer-2 weights layer-2 weights k th output weights
neuron) (W ij) neuron) (W j1, j2 ) neuron) (W j1, j2 ) node) (W j2, k )

W 1,1 −0.30023216 W 1,1 −0.2776832 W 6,1 −0.64809683 W 1,1 0.7230983

W 1,2 0.27647354 W 1,2 0.19835022 W 6,2 0.07503865 W 2,1 −0.769448
W 1,3 −0.18358764 W 1,3 −0.2341812 W 6,3 0.46713774 W 3,1 −0.722583
W 1,4 −0.08670065 W 1,4 −0.6902042 W 6,4 0.35403297 W 4,1 0.6882041
W 1,5 −0.84691089 W 1,5 −0.9776295 W 6,5 0.14505076 W ,5,1 0.8545416
W 1,6 −0.11793122 W 1,6 −0.5679249 W 6,6 −0.65689846 W 6,1 0.9073756
W 1,7 0.13485305 W 1,7 −0.365493 W 6,7 0.03919319 W 7,1 0.4347044
W 1,8 −0.37024051 W 1,8 0.34264155 W 6,8 −0.12120154 W 8,1 −0.961718
W 1,9 −0.18615765 W 1,9 −0.5132074 W 6,9 −0.5650661 W 9,1 0.1031879
W 1,10 0.86567297 W 1,10 0.37565473 W 6,10 −0.01778744 W 10,1 0.2004992
W 2,1 0.66145583 W 2,1 −0.6123977 W 7,1 0.47617357
W 2,2 0.90219146 W 2,2 0.91891559 W 7,2 −0.38991254
W 2,3 −0.52379505 W 2,3 0.67513838 W 7,3 −0.50879259
W 2,4 0.75761136 W 2,4 −0.4441866 W 7,4 −0.21676101
W 2,5 −0.52157099 W 2,5 −0.362877 W 7,5 −0.4396702
W 2,6 0.02811703 W 2,6 −0.322716 W 7,6 0.05882271
W 2,7 −0.74248271 W 2,7 −0.736477 W 7,7 0.05713673
W 2,8 0.44767 W 2,8 −0.2797636 W 7,8 −0.00745899
W 2,9 0.75771368 W 2,9 0.46671175 W 7,9 0.14822149
W 2,10 0.59574177 W 2,10 −0.37185 W 7,10 0.61642481
W 3,1 0.69399448 W 3,1 0.32263642 W 8,1 0.86812406
W 3,2 −0.24094788 W 3,2 0.13153567 W 8,2 −0.23425252
W 3,3 0.138715 W 3,3 −0.9109613 W 8,3 −0.87820354
W 3,4 0.48719812 W 3,4 0.07223889 W 8,4 0.20198286
W 3,5 0.86200771 W 3,5 −0.6365315 W 8,5 −0.85923602
W 3,6 0.11118879 W 3,6 −0.2011787 W 8,6 −0.54009888
W 3,7 0.7113249 W 3,7 0.63840616 W 8,7 −0.89531795
W 3,8 0.44375463 W 3,8 0.3039039 W 8,8 −0.67206656
W 3,9 0.00195087 W 3,9 0.45370143 W 8,9 0.84480996
W 3,10 −0.05467507 W 3,10 −0.7748062 W 8,10 0.17705815
W 4,1 0.4442245 W 4,1 0.61790615 W 9,1 −0.96609938
W 4,2 −0.02693093 W 4,2 0.23819518 W 9,2 −0.7293678
W 4,3 −0.83992177 W 4,3 −0.8211282 W 9,3 −0.13490159
W 4,4 −0.45336151 W 4,4 −0.5237951 W 9,4 0.47337608
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
11.5 Results and Discussion 209

Table 11.2 (Continued)

W j1, j2 (Weight W j1, j2 (Weight

W ij (Weight between between W j2, k (Weight
between j1th hidden j1th hidden between
ithinput layer-1 layer-1 j2th hidden
node and neuron and neuron and layer-2
jth hidden j2th hidden Values of j2th hidden Values of neuron and Values of
layer-1 Values of layer-2 weights layer-2 weights k th output weights
neuron) weights (W ij) neuron) (W j1, j2 ) neuron) (W j1, j2 ) node) (W j2, k )

W 4,5 0.5132074 W 4,5 −0.6083042 W 9,5 −0.05384209

W 4,6 −0.76093636 W 4,6 0.55057171 W 9,6 0.92812343
W 4,7 0.04174126 W 4,7 −0.654054 W 9,7 0.74402351
W 4,8 0.84932026 W 4,8 0.80304517 W 9,8 0.8716961
W 4,9 0.69136831 W 4,9 0.63062396 W 9,9 −0.3011121
W 4,10 0.58899559 W 4,10 0.85283625 W 9,10 0.04511855
W 5,1 0.14350462 W 10,1 0.50554263
W 5,2 −0.77981667 W 10,2 −0.9582004
W 5,3 0.53391204 W 10,3 −0.7336712
W 5,4 −0.44076614 W 10,4 −0.9653672
W 5,5 −0.56648787 W 10,5 −0.4090327
W 5,6 0.83536792 W 10,6 0.02880597
W 5,7 0.33216417 W 10,7 0.94920176
W 5,8 −0.24961992 W 10,8 −0.5693641
W 5,9 −0.28888735 W 10,9 0.5107654
W 5,10 0.78428229 W 10,10 −0.1438138

Table 11.3 Testing of trained ANN using a set of instances produced by the cut-set technique for a 14-bus
system.

S. No. V 1 (pu) V 2 (pu) V 3 (pu) P d (pu) PII(by cut-set technique) PII(by ANN) % Error

1 0.95 0.95 0.95 5.1923 0.043650384 0.042890867 1.74

2 0.95 0.95 1 5.8459 0.056269778 0.057434562 −2.07
3 0.95 1 1 6.2883 0.072388475 0.071527052 1.19
4 1 1 1 6.5168 0.056267451 0.054872018 2.48
5 1 1 1.05 6.0432 0.036514654 0.036887103 −1.02
6 1 1.05 1.05 7.1182 0.067945921 0.065662938 3.36
7 1.05 1.05 1.05 6.6639 0.044472754 0.043191939 2.88
8 1.05 1.05 1.1 7.4461 0.044472754 0.045597915 −2.53
9 1.05 1.1 1.1 7.3312 0.051859491 0.05095195 1.75
10 1.1 1.1 1.1 7.2623 0.028822574 0.028182713 2.22
11 1.1 1.1 1.15 7.4387 0.036780674 0.03732135 −1.47
12 1.1 1.15 1.15 8.972 0.067945801 0.06739544 0.81
13 1.15 1.15 1.15 9.3672 0.076020248 0.074385813 2.15
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
210 11 Probabilistic Voltage Security Assessment and Enhancement Using Rescheduling of Reactive Power Control Variables

Table 11.4 Voltage and load for the 14-bus test system at all buses, prior to and following load
shedding.

Voltage(pu) Load(pu)
Prior to load following load Prior to load following load
Bus No. shedding shedding shedding shedding

1 1.15 1.15 3.4688 3.4688

2 1.15 1.15 0.5459 0.5459
3 1.15 1.15 2.0865 2.0865
4 0.83 0.9486 0.2927 0.2927
5 0.8652 1.0041 0 0
6 0.9966 1.0585 1.0413 1.0413
7 0.8652 1.0041 0 0
8 1.0033 1.0595 0.1696 0.1696
9 0.8204 0.9839 0.7349 0.4410*
10 0.8011 0.9606 0.2324 0.2324
11 0.8057 0.9472 0.0856 0.0856
12 0.7869 0.9151 0.1369 0.1369
13 0.7744 0.9089 0.319 0.319
14 0.7408 0.9027 0.3471 0.3471

multiple load buses employing Eq. (11.15). Bus No. 9, with the highest sensitivity of 0.0843314, was
chosen for the load-shedding action. The amount of load shedding (0.3293922 pu) remained within
acceptable bounds, and the computed PII following the load shedding action was 0.060812589,
which is less than the threshold value. Table 11.4 provides information regarding the bus voltages
and bus loads before and after the load-shedding event.

11.5.2 25-bus System

This system comprises five PV buses, and their voltages can be regulated between (0.95 ≤ V i ≤
1.15 pu) to adjust the voltage of load buses. The availability of lines from line No. 1 to line No. 35
is considered as: 0.99548203, 0.99659806, 0.99655488, 0.99591478, 0.996347062, 0.99557426,
0.99652046, 0.99489641, 0.99137486, 0.99857835, 0.99263634, 0.99441704, 0.99035256, 0.99095274,
0.99073713, 0.99773568, 0.99949272, 0.99227775, 0.99685410, 0.99330443, 0.99868794, 0.990636047,
0.99119283, 0.99768705, 0.99137753, 0.99122069, 0.99865404, 0.99945461, 0.99160406, 0.99070482,
0.99177757, 0.99047761, 0.99899225, 0.995236073, and 0.99932094 (Billinton and Allan 1984).
Table 11.5 shows a set of samples used to train the ANN, which enabled the computation of the
PII utilizing the cut-set technique and the ANN, along with the percentage errors. The network
underwent training with 100 instances, and only 18 instances with the highest error magnitudes
are listed in Table 11.5. In Table 11.6, various weights, namely W j2, k , W j1, j2 , and W i, j1 , connect
the neurons between the hidden layer-2 and the output layer, hidden layer-1 and hidden layer-2,
and input node and hidden layer-1, respectively. The input layer includes variables V 1 , V 2 , V 3 ,
V 4 , V 5 , and Pd namely PV-bus voltages and overall load of the system. Table 11.7 illustrates the
PII acquired through the trained network and the cut-set methodology, along with the percentage
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
11.5 Results and Discussion 211

Table 11.5 Training samples and computed probabilistic insecurity index (PII) using cut-set technique and
artificial neural network for a 25-bus system.

PII(by cut-set
S. No V 1 (pu) V 2 (pu) V 3 (pu) V 4 (pu) V 5 (pu) P d (pu) technique) PII(by ANN) % Error

1 0.95 0.95 0.95 0.95 0.95 16.1245 0.129134758 0.125144494 3.09

2 0.95 0.95 0.95 0.95 1 15.7275 0.077344824 0.075604565 2.25
3 0.95 0.95 1 1 1 15.385 0.073622054 0.074409809 −1.07
4 0.95 1 1 1 1 15.9937 0.069214094 0.067947476 1.83
5 1 1 1 1 1 16.9108 0.113233571 0.110221558 2.66
6 1 1 1 1 1.05 16.1893 0.081008791 0.081673063 −0.82
7 1 1 1 1.05 1.05 17.3185 0.121191671 0.119385915 1.49
8 1 1 1.05 1.05 1.05 16.3393 0.048175967 0.046417544 3.65
9 1 1.05 1.05 1.05 1.05 17.2789 0.129134758 0.130568153 −1.11
10 1.05 1.05 1.05 1.05 1.05 16.5873 0.065550127 0.06711022 −2.38
11 1.05 1.05 1.05 1.1 1.1 17.6874 0.133542718 0.13092528 1.96
12 1.05 1.05 1.1 1.1 1.1 17.2962 0.097357431 0.100112646 −2.83
13 1.1 1.1 1.1 1.1 1.1 17.6705 0.129134758 0.126732851 1.86
14 1.1 1.1 1.1 1.1 1.15 17.9886 0.085302924 0.087972905 −3.13
15 1.1 1.1 1.1 1.15 1.15 18.6302 0.119991674 0.118470404 −1.52
16 1.1 1.1 1.15 1.15 1.15 18.8887 0.116646531 0.113205458 2.95
17 1.1 1.15 1.15 1.15 1.15 19.27 0.113233571 0.114943397 −1.51

discrepancies for the test cases. The trained network was tested with 200 instances, and only 18
instances with the highest error magnitudes are listed in Table 11.7. In this range, additional
instances for testing have been chosen in addition to the training instances. The values of the PII,
should not be higher than a predetermined level known as the threshold value, which has been
assumed as 0.109038301. The percentage error of PII, as computed by trained ANN, is less than
±5%. For a specific operating setting, corrective reactive power rescheduling was demonstrated
with: total load Pd = 24.7970 pu and V 1 = 1.05 pu, V 2 = 1.05 pu, V 3 = 1.05 pu, V 4 = 1.05 pu.
It was observed that the stability margin obtained was less than 20% with a PII of 0.152038301,
exceeding the specified threshold value. Sensitivities of PII were computed using Eq. (11.14),
and it was found that Bus No. 1 is selected for voltage correction (ΔV 1 = 0.08524 pu) due to its
highest sensitivity. This correction brought the PII within the desired limit of 0.1019038301 as the
final value. Another case was considered with values: V 1 = 1.15 pu, V 2 = 1.15 pu, V 3 = 1.15 pu,
V 4 = 1.15 pu, V 5 = 1.15 pu, total load Pd = 28.7239 pu and despite correction attempts using
Eq. (11.14) for rescheduling of reactive power at the most sensitive buses, the PII remained above
the specified limit (0.142145948). As a result, sensitivities were assessed for load shedding at
different load buses using Eq. (11.15). Among these buses, Bus No. 4, with the highest sensitivity
value of 0.2114285, was chosen for load shedding. The load-shedding amount (0.307905 pu)
remained within acceptable bounds, and the computed PII following the load-shedding action
was 0.10903828, falling below the specified threshold value. Table 11.8 provides details of all bus
voltages and bus loads before and after load shedding.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
212 11 Probabilistic Voltage Security Assessment and Enhancement Using Rescheduling of Reactive Power Control Variables

Table 11.6 Final weights obtained after ANN network training for the 25-bus system.

W ij (Weight W j1, j2 (Weight W j1, j2 (Weight W j2, k (Weight

between between j1th between j1th between
ithinput hidden layer-1 hidden layer-1 j2th hidden
node and neuron and neuron and layer-2
jth hidden Values of j2th hidden Values of j2th hidden Values of neuron and Values of
layer-1 weights layer-2 weights layer-2 weights kth output weights
neuron) (W ij ) neuron) (W j1, j2 ) neuron) (W j1, j2 ) node) (W j2, k )

W 1,1 0.2442573 W 1,1 0.7230983 W 6,1 −0.2776832 W 1,1 −0.19111826

W 1,2 0.7331331 W 1,2 −0.769448 W 6,2 0.19835022 W 2,1 0.216105829
W 1,3 0.0950225 W 1,3 −0.722583 W 6,3 −0.2341812 W 3,1 −0.56191311
W 1,4 −0.690432 W 1,4 0.6882041 W 6,4 −0.6902042 W 4,1 0.374601197
W 1,5 −0.773507 W 1,5 0.8545416 W 6,5 −0.9776295 W 5,1 −0.16724777
W 1,6 −0.404048 W 1,6 0.9073756 W 6,6 −0.5679249 W 6,1 0.589796739
W 1,7 −0.326991 W 1,7 0.4347044 W 6,7 −0.365493 W 7,1 0.976397132
W 1,8 −0.085284 W 1,8 −0.961718 W 6,8 0.34264155 W 8,1 −0.02965714
W 1,9 0.972212 W 1,9 0.1031879 W 6,9 −0.5132074 W 9,1 0.872369128
W 1,10 −0.654907 W 1,10 0.2004992 W 6,10 0.37565473 W 10,1 −0.39616986
W 2,1 0.5389484 W 2,1 0.0852845 W 7,1 −0.6123977
W 2,2 −0.084517 W 2,2 0.1914532 W 7,2 0.91891559
W 2,3 −0.381324 W 2,3 −0.609962 W 7,3 0.67513838
W 2,4 −0.847238 W 2,4 −0.34358 W 7,4 −0.4441866
W 2,5 −0.032479 W 2,5 −0.802622 W 7,5 −0.362877
W 2,6 0.1945016 W 2,6 0.2634518 W 7,6 −0.322716
W 2,7 −0.577581 W 2,7 0.2269542 W 7,7 −0.736477
W 2,8 −0.65358 W 2,8 −0.223834 W 7,8 −0.2797636
W 2,9 0.8746088 W 2,9 0.7524238 W 7,9 0.46671175
W 2,10 −0.115739 W 2,10 0.2191109 W 7,10 −0.37185
W 3,1 −0.9398377 W 3,1 0.8028368 W 8,1 0.32263642
W 3,2 0.5577976 W 3,2 0.5600327 W 8,2 0.13153567
W 3,3 0.8848459 W 3,3 −0.14538 W 8,3 −0.9109613
W 3,4 0.8298412 W 3,4 0.4290769 W 8,4 0.07223889
W 3,5 −0.9231917 W 3,5 0.4616857 W 8,5 −0.6365315
W 3,6 −0.5588921 W 3,6 0.1024955 W 8,6 −0.2011787
W 3,7 0.2056884 W 3,7 0.0465715 W 8,7 0.63840616
W 3,8 0.1129604 W 3,8 −0.171932 W 8,8 0.3039039
W 3,9 −0.0255147 W 3,9 0.5950119 W 8,9 0.45370143
W 3,10 0.8283314 W 3,10 0.5601836 W 8,10 −0.7748062
W 4,1 0.2134732 W 4,1 0.193971 W 9,1 0.61790615
W 4,2 −0.311213 W 4,2 0.2719943 W 9,2 0.23819518
W 4,3 −0.428993 W 4,3 −0.879007 W 9,3 −0.8211282
W 4,4 0.8494294 W 4,4 0.9177006 W 9,4 −0.5237951
W 4,5 0.3049794 W 4,5 0.2146476 W 9,5 −0.6083042
W 4,6 −0.116272 W 4,6 −0.494274 W 9,6 0.55057171
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
11.5 Results and Discussion 213

Table 11.6 (Continued)

W ij (Weight W j1, j2 (Weight W j1, j2 (Weight W j2, k (Weight

between between j1th between j1th between
ithinput hidden layer-1 hidden layer-1 j2th hidden
node and neuron and neuron and layer-2
jth hidden Values of j2th hidden Values of j2th hidden Values of neuron and Values of
layer-1 weights layer-2 weights layer-2 weights kth output weights
neuron) (W ij ) neuron) (W j1, j2 ) neuron) (W j1, j2 ) node) (W j2, k )

W 4,7 −0.549594 W 4,7 0.6106075 W 9,7 −0.654054

W 4,8 0.4564163 W 4,8 0.4472334 W 9,8 0.80304517
W 4,9 0.3039145 W 4,9 −0.339636 W 9,9 0.63062396
W 4,10 −0.335557 W 4,10 0.1843682 W 9,10 0.85283625
W 5,1 0.1410407 W 5,1 0.5262541 W 10,1 0.14350462
W 5,2 0.0759936 W 5,2 0.2055966 W 10,2 −0.77981667
W 5,3 0.5457719 W 5,3 −0.724611 W 10,3 0.53391204
W 5,4 −0.365993 W 5,4 0.5458727 W 10,4 −0.44076614
W 5,5 0.0861291 W 5,5 −0.307851 W 10,5 −0.56648787
W 5,6 0.2519786 W 5,6 0.3304626 W 10,6 0.83536792
W 5,7 0.1878709 W 5,7 −0.060892 W 10,7 0.33216417
W 5,8 −0.221088 W 5,8 0.8475661 W 10,8 −0.24961992
W 5,9 −0.305875 W 5,9 −0.486028 W 10,9 −0.28888735
W 5,10 0.4281549 W 5,10 −0.628961 W 10,10 0.78428229

Table 11.7 Testing of trained ANN using a set of instances produced by the cut-set technique for a 25-bus
system.

PII(by cut-set
S. No V 1 (pu) V 2 (pu) V 3 (pu) V 4 (pu) V 5 (pu) P d (pu) technique) PII(by ANN) % Error

1 0.95 0.95 0.95 0.95 0.95 16.1139 0.137950678 0.139799217 −1.34

2 0.95 0.95 0.95 0.95 1 16.2355 0.129263598 0.126872221 1.85
3 0.95 0.95 1 1 1 16.3004 0.121054491 0.117931285 2.58
4 0.95 1 1 1 1 16.3134 0.113446261 0.110655483 2.46
5 1 1 1 1 1 16.4435 0.109723491 0.111084062 −1.24
6 1 1 1 1 1.05 16.9398 0.124855638 0.12358211 1.02
7 1 1 1 1.05 1.05 17.2549 0.129134758 0.128114593 0.79
8 1 1 1.05 1.05 1.05 17.4399 0.133542718 0.138298348 −2.15
9 1 1.05 1.05 1.05 1.05 17.4789 0.137950678 0.136143524 1.31
10 1.05 1.05 1.05 1.05 1.05 16.5721 0.061271007 0.061951115 −1.11
11 1.05 1.05 1.05 1.1 1.1 17.6874 0.133542718 0.132260707 0.96
12 1.05 1.05 1.1 1.1 1.1 17.2619 0.081008791 0.079688347 1.63
13 1.1 1.1 1.1 1.1 1.1 17.495 0.121191671 0.118670884 2.08
14 1.1 1.1 1.1 1.1 1.15 18.7971 0.113233571 0.111954031 1.13
15 1.1 1.1 1.1 1.15 1.15 18.4765 0.078238604 0.077111968 1.44
16 1.1 1.1 1.15 1.15 1.15 17.2082 0.105315531 0.106316028 −0.95
17 1.1 1.15 1.15 1.15 1.15 20.07 0.125599631 0.122946157 2.11
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
214 11 Probabilistic Voltage Security Assessment and Enhancement Using Rescheduling of Reactive Power Control Variables

Table 11.8 Voltage and load for the 25-bus test system at all buses, prior to and following load
shedding.

Voltage (pu) Load (pu)

Prior to load following load Prior to load following load
Bus No. shedding shedding shedding shedding

1. 1.1500 1.1500 5.6986 5.6986

2. 1.1500 1.1500 0.2829 0.2829
3. 1.1500 1.1500 1.4311 1.4311
4. 1.1500 1.1500 0.8569 0.4285*
5. 1.1500 1.1500 0.7113 0.7113
6. 1.0490 1.0617 0.4285 0.4285
7. 1.0352 1.0499 0.4285 0.4285
8. 1.0249 1.0413 0.6775 0.6775
9. 1.0015 1.0157 0.4285 0.4285
10. 1.0355 1.0476 0.4285 0.4285
11. 1.0019 1.0237 0.1355 0.1355
12. 1.0038 1.0262 0.2710 0.2710
13. 1.0524 1.0636 0.7113 0.7113
14. 0.9653 0.9696 0.5742 0.5742
15. 0.9574 0.9620 0.8569 0.8569
16. 0.9965 1.0003 0.8569 0.8569
17. 0.9785 1.0109 1.7139 1.7139
18. 0.9952 1.0245 0.4285 0.4285
19. 1.0825 1.1059 0.4285 0.4285
20. 1.0338 1.0524 0.7113 0.7113
21. 0.9232 0.9704 0.5742 0.5742
22. 0.8037 0.8847 0.5742 0.5742
23. 0.9146 0.9747 0.4285 0.4285
24. 0.7344 0.8302 0.4285 0.4285
25. 0.7603 0.8515 0.7113 0.7113

11.6 Conclusions
A novel algorithm has been introduced, offering a new perspective has been introduced to monitor
and improve voltage security based on the probability of system failure. By employing the cut-set
technique and accounting for the static voltage stability limit, the algorithm computes PII under
various operating conditions. These collected samples were utilized to train an ANN using the
backpropagation algorithm, enabling the online application of the PII. Moreover, the ANN was
employed to assess the sensitivity of the PII concerning reactive power control variables and load
shedding, ensuring that the index remains at or above the threshold value. Taking into account
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 215

that voltage instability is a gradual event as pointed out in reference (Zarate et al. 2006), it log-
ically ensues that carrying out evaluations and putting into action preventive measures can be
accomplished in real-time employing this ANN.

References

Alzaareer, K., Saad, M., Mehrjerdi, H. et al. (2020). A new sensitivity approach for preventive control
selection in real-time voltage stability assessment. International Journal of Electrical Power & Energy
Systems 122: 106212. https://doi.org/10.1016/j.ijepes.2020.106212.
Arya, L.D. and Verma, H.K. (1996). A method for tracing PV-curve for voltage stability analysis with
voltage dependent loads. Int Journal of Electric machines and Power Systems 24: 583–596.
Arya, L.D., Choube, S.C., and Saket, R.K. (2000). Composite system reliability evaluation based on
static voltage stability limit. Journal of Institution of Engineers India Part El Electrical Engineering
Division 80: 133–139.
Arya, L.D., Titare, L.S., and Kothari, D.P. (2007). Probabilistic assessment and preventive control of
voltage security margins using artificial neural network. International Journal of Electrical Power &
Energy Systems 29 (2): 99–105. https://doi.org/10.1016/j.ijepes.2006.05.005.
Arya, L.D., Titare, L.S., and Kothari, D.P. (2009). An approach to mitigate the risk of voltage collapse
accounting uncertainties using improved particle swarm optimization. Applied Soft Computing 9 (4):
1197–1207. https://doi.org/10.1016/j.asoc.2009.02.005.
Benidris, M. and Mitra, J. (2015). Reliability and sensitivity analysis of composite power systems
considering voltage and reactive power constraints. IET Generation Transmission and Distribution 9
(12): 1245–1253.
Bijwe, P.R., Kothari, D.P., and Arya, L.D. (1993). Alleviation of line overloads and voltage violations by
corrective rescheduling. Proceedings of the IEE - Part C: Monographs 140 (4): 249–255.
Billinton, R. and Aboreshaid, S. (1998). Voltage stability considerations in composite power system
reliability evaluation. IEEE Transactions on Power Systems 13 (2): 655–660.
Billinton, R. and Allan, R.N. (1983). Reliability Evaluation of Engineering Systems: Concepts and
Techniques’, (Book). PITMAN Books Limited.
Billinton, R. and Allan, R. (1984). Reliability Evaluation of Power Systems. Plenum Press.
Billinton, R. and Khan, E. (1992). A security based approach to composite power system reliability
evaluation. IEEE Transactions on Power Systems 7: 65–70.
Billinton, R. and Li, W. (1993). A system state transition sampling method for composite system
reliability evaluation. IEEE Transactions on Power Systems 8 (3): 761–766.
Billinton, R. and Wenyuan, L. (1991). Hybrid Approach for reliability evaluation of composite
generation and transmission systems using Monte-Carlo simulation and enumeration technique.
IEE Proceedings-C 138 (3): 233–241.
Capitaanescu, F. and Van Cutsem, T. (2002). Preventive control of voltage security margins: a
multi-contingency sensitivity based approach. IEEE Transactions on Power Systems 17 (2): 358–364.
Concordia, C. (1991). Voltage instability. Journal of Electrical Power and Energy Systems 13 (1): 14–24.
Deng, Z. and Singh, C. (1992). A new approach to reliability evaluation of interconnected power system
including planned outages and frequency calculations. IEEE Transactions on Power Systems 7 (2):
734–743.
Fu, L.M. (1994). Neural Network in Computer Intelligence. Mc Graw Hill Inc.
Haykin, S. (2003). Neural Network: A Comprehensive foundation. Pearson Education.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
216 11 Probabilistic Voltage Security Assessment and Enhancement Using Rescheduling of Reactive Power Control Variables

Henneaux, P. (2015). Probability of failure of overloaded lines in cascading failures. International

Journal of Electrical Power & Energy Systems 73: 141–148.
IEEE System Dynamic Performance Subcommitte (1990). Voltage stability of power systems concepts,
Analytical tools and industry experience. IEEE document 90th 0358-2-PWR-1990.
de Jong, M., Papaefthymiou, G., and Palensky, P. (2018). A framework for incorporation of infeed
uncertainty in power system risk-based security assessment. IEEE Transactions on Power Systems 33
(1): 613–621.
Kirschen, D.S. and Jayaweera, D. (2007). Comparison of risk-based and deterministic security
assessments. IET Generation Transmission and Distribution 1 (4): 527–533. https://doi.org/10.1049/
iet-gtd:20060368.
Kundur, P. (1994). Power System Stability and Control. New York: Mc Graw-Hill.
Melo, A.C.G., Mello, J.C., and Granville, S. (1997). Effects of voltage collapse problems in the reliability
evaluation of composite systems. IEEE Transactions on Power Systems 12 (1): 480–487.
Mokred, S., Wang, Y., and Chen, T. (2023). A novel collapse prediction index for voltage stability
analysis and contingency ranking in power systems. Protection and Control of Modern Power Systems
8 (7): https://doi.org/10.1186/s41601-023-00279-w.
Pereira, M.V.F. and Pinto, L.M.V.G. (1992). A new computational tool for composite reliability
evaluation. IEEE Transactions on Power Systems 7 (1): 258–263.
Sebastian Rios, M., Victor, V.P., and Kiguel, D.L. (1998). Bus based reliability indices and associated
costs in the bulk power system. IEEE Transactions on Power Systems 13 (3): 719–723.
Singh, P., Arya, R., Titare, L.S., and Arya, L.D. (2021). Optimal load shedding to avoid risks of voltage
collapse using black hole algorithm. Journal of The Institution of Engineers (India): Series B 102:
261–276. https://doi.org/10.1007/s40031-021-00543-3.
Taylor, C.W. (1994). Power System Voltage Stability. Mc Graw Hill. Inc.
Xia, C., Zheng, X., Guan, L., and Baig, S. (2021). Probability analysis of steady-state voltage stability
considering correlated stochastic variables. International Journal of Electrical Power & Energy
Systems 131: 107105. https://doi.org/10.1016/j.ijepes.2021.107105.
Zarate, L.A.L., Castro, C.A., Ramos, J.L.M., and Ramos, E.R. (2006). Fast computation of voltage
stability security margins using nonlinear programming techniques. IEEE Transactions on Power
Systems 21 (1): 19–27.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
217

Reliability Analysis of Distribution Systems

Section 4
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
219

12

Modern Aspects of Probabilistic Distributions for Reliability Evaluation

of Engineering Systems
Aanchal Singh S. Vardhan 1 , Aanchal Verma 2 , Jyotsna Ogale 3 , R. K. Saket 2 , and
Stuart Galloway 4
1
Department of Electrical Engineering, National Institute of Technology, Jamshedpur, Jharkhand, India
2
Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
3
Department of Electronics Engineering, Samrat Ashok Technological Institute, Vidisha, Madhya Pradesh, India
4
University of Strathclyde, Glasgow, UK

12.1 Introduction

The reliability concept was developed in 1949 when 70% of Navy electronics equipment was not
operated properly, and during the Second World War, 80% of airborne equipment was damaged
during transportation. The economic values of these equipment’s were so high that they imme-
diately drew scientist’s attention toward the reliability assessment and dependability engineering.
The concept of reliability was primarily formulated to address situations in which nonrepairable
components, equipment, devices, and systems are involved and where the initial occurrence of
failure holds significant importance (Arya et al. 2000).
With development of different aspects of reliability of enormous amount of components, fail-
ure data was collected, analyzed, and published in the field of electronics engineering while at the
same time, repairable system development of analytical methods for reliability assessment was at
a slow pace. Nowadays, almost all branches of engineering, science, and technology studies are
going on different aspects of reliability for both types of systems, which can be repaired and can-
not be repaired. The first formal reliability evaluation was reported to occur when an explanation
was sought for the poor performance of the missiles during the Second World War. The missiles
were constructed of large number of components considered highly reliable. Over time, problems
were identified related to reliability evaluation of components or systems (Arya et al. 2001). The
availability of data was lacking, computational resources were limited, there were no reliability
techniques of realistic nature available, the use of probabilistic techniques was averted, and under-
standing was lacking of the meaning, significance, and inferences drawn from probabilistic models
and risk indices. As mentioned, all the problems vanished in 21st century because now we have
valid and sufficient data availability and in-depth understanding of well-developed probabilistic
and reliability evaluation techniques. In this chapter, discussions are made on different techniques
used for evaluation of reliability, which are applicable to power system analysis and explain the
significance of the various reliability indices (Bansal et al. 2002).
Whenever unreliable components, devices, and systems occupy a place in human lives, a large
number of issues related to safety considerations arise, and reliability becomes the most important
” standby components is shown in Figure 5.1
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
220 12 Modern Aspects of Probabilistic Distributions for Reliability Evaluation of Engineering Systems

the imposed constraint of cost of the system, that might be the operating cost or planning cost
including initial investment, reliability consideration becomes an important part of engineering
design (Bharti et al. 2019). Not only this, but the analysis of reliability also considers factors like
operating environment, time of operation, and quality of service rendered by the system. Naturally,
under these considerations, the usability, applicability, usefulness, or goodness of the system must
be judged not only qualitatively but also quantitatively by taking numerical measures based on
probability theory known as reliability indices since any qualitative judgment is meaningless for
any engineer who is designing or planning a system. The operating environment must get due
consideration for dependability assessment, which affects the rate of system failure (Billinton and
Zhang 1998). Take an example, a transmission line installed in coastal areas and those in the region
having fair weather condition will have different failure as well as repair rate, hence, will have
different probability of success and failure. Therefore, any reliability modeling must account for
the operating environment. The time of operation of a repairable or nonrepairable component is
also of great concern. The probability of a system operating successfully decreases with an increase
in the intended time of operation of a system. Quality of service requires special attention because
the system may give service, but the service may not satisfy the consumer. For example, electricity
supply may be available at some consumer’s premises, but supply voltage and frequency may be
too low, too high, or too fluctuating to use it effectively in a production process. Hence, reliability is
classically defined and states that it is the probability of a component performing its own function,
which is intended over a period specified under the conditions set for its operation (Kothari et al.
2022).

12.2 Life Distribution of Power Components: An Overview

In principle, any probability distribution can be employed to simulate the lifespan of the equipment.
Practically, it is found that the hazard functions are monotonic and seem to be most realistic (Kumar
et al. 2019). There are only a few that provide reasonable models for reliability of a device. The most
commonly used models for distribution of life are given below.

12.2.1 Binomial Distribution

Let us have an experiment doing a trial, which generates only two outcomes with possibility of
getting successful or failing. If p denotes the success probability, then (1 − p) = q will be the fail-
ure probability. If x is varying randomly and denotes the count of successful outcomes in a series
of n trials. The probability of obtaining exactly x successes can be articulated as b(x, n, p) = n Cx
n!
q(n−x) px , where x = 0,1,2,3……n and combinatorial quantities n Cx = (n−x)!x! are viewed as coef-
ficients in binomial expansions. This statistical distribution is commonly known as a binomial
distribution.
Corollary: Addition of all given below
∑
n
∑
n
n
f (x) = Cx px qn−x = qn + n C1 p1 qn−1 + … pn = (q + p)n = 1 (12.1)
x=0 x=0

To evaluate the number of gun firing rounds in military system for object detection by radar, in
the production process calculation of number of defective samples, and in the interconnected bulk
power system for estimation of reliability, the binomial distribution is usually applied (Kumar et.
al 2020).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12.2 Life Distribution of Power Components: An Overview 221

12.2.2 Exponential Distribution

The exponential function mainly uses the probability function to create models for the evaluation
of reliability. Its algebraic simplicity and tractability make it a well-known model for assessment of
reliability of a component and device and acting as a symbol of the operational duration within the
complete lifespan of the device. Just before entering the wear-out phase, the device is typically in
a state of relative stability, making the use of constant hazard rates a suitable choice for modeling
device reliability as well.
If nonnegative values are assumed by a random variable of continuous nature, X then an expo-
nential distribution will have parameter 𝜶 > 0, and the PDF is articulated as follows:
{ −ax
𝛼e , x ≥ 0
f (x) =
0, otherwise
In integral form, this function is shown below
x
F(x) = 𝛼e−ax dx (12.2)
∫0

12.2.3 Poisson Distribution

This distribution can be considered as a limiting case of PDF of binomial type subjected to
the conditions p → 0 and n → ∞. Therefore, Poisson distribution is nothing but the relative
frequency-dependent distribution of a variable x:
e−m mx ∑∞
∑∞ −m x
e m ∑∞
mx
f (x, m) = and f (x, m) = = e−m = e−m em = 1. (12.3)
x! x=0 x=0
x! x=0
x!
It is the probability exponentially distributed with a single parameter “m.”
Some of the application areas of Poisson distribution are the evaluation of bomb hits spatial distri-
bution, estimation of fragments in total number while a shell hits a target, evaluation of defective
vehicles based on arrival pattern in an engineering workshop, spare parts demand pattern, and
evaluation of power components reliability (Saket 2013).

Example 12.1
If 5% of MOSFET products are defectively manufactured and sold in boxes of 100 with the guarantee
of 10 defective pieces per box. So find the approximate probability of failure of a box to attain the
quality, which is guaranteed at the time of purchase?
Solution:
Given:- Number of products in boxes (n) = 100
Let p be the defective MOSFET probability = 5% = 0.05
m = Average of MOSFETs which are found defective in a box of n samples = No. of products in
boxes × probability of defective product (n × p) = 100 × 0.05 = 5
Since probability (p) is less, Poisson distribution may be used.
Within a set of 100 items, the probability of x defective products is calculated as follows:
e−m mx e−5 5x
P(X = x) = =
x! x!
where, x = 0, 1, 2, …
The probability of getting failed to attain the quality that is guaranteed at the time of purchase,
∑10 −5 x
e 5 ∑10
5x
P(X > 10) = 1 − P(X ≤ 10) = 1 − = 1 − e−5
x=0
x! x=0
x!
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
222 12 Modern Aspects of Probabilistic Distributions for Reliability Evaluation of Engineering Systems

[ ]
𝟓𝟎 𝟓𝟏 𝟓 𝟐 𝟓𝟑 𝟓𝟏𝟎
[1 − p(X ≤ 10)] = 1 − e −5
+ + + + ……………. +
𝟎! 𝟏! 𝟐! 𝟑! 𝟏𝟎!
[1 − p(X ≤ 10)] = 1 − e−5 [1 + 5 + 12.5 + 20.8333 + 26.0417 + 26.0417 + 21.7014
+15.5001 + 9.6881 + 5.3823 + 2.6911]

[1 − p(X ≤ 10)] = 1 − e−5 [146.3797] = 1 − 0.00673 × 146.3797

= (1 − 0.985135) = 0.0148646

Example 12.2
1
While turning electric power switches in a factory, there is a small chance of defectiveness 500 exists.
If power switches are packed in a group of 10. With Poisson’s distribution calculate number of pack-
ets approximately, which contains (i) no defective piece, (ii) single defective piece, and (iii) double
defective pieces in a 1000 packets consignment.

Solution:
1
Given probability of defect p = 500
;n = 10, and N = 1000
1 1
∴ Mean m = n × p = 10 × 500
= 50 = 0.02.
e−m mx e−0.02 (0.02)x
Poisson type distribution is expressed as, p (x) = x!
= x!

e−0.02 (0.02)0
(i) p (probability with zero defective piece) = P(0) = = 0.980198
0!
In a consignment of 1000 packets, total packets in number contains no defective
piece = p(0) × N = 1000 × 0.98019 = 9802
(ii) p (probability with single piece defect) = p(1) = 0.01960 ∴ In a 1000 packets consignment, total
packets with single defect in piece will be = p(1) × N = 1000 × 0.01960 = 20
(iii) p (probability when double pieces of switches are in defect) = p(2) = 0.001960

In a consignment of 1000 packets, total number of packets having double defective

pieces = p(2) × N = 1000 × 0.001960 = 2

Example 12.3
An industrial machine manufactures internal parts of an induction generator in large quantity. The
mean number of defective units in a sample of 20 parts is 2. If an Indian power industry purchases
1000 such samples for installation of a wind power plant, find the chances for having atleast 3
defective parts?
Solution:
Average of defective units (m) = 2
Here, X is distributed in a Poisson’s way whose function of probability is given as,
e−2 2x
p(x) =
x!
here, x represents total defective units.
P [of a part containing at least three defective units]

= p [X = 3] + p [X = 4] + p [X = 5] + p [X = 6] + p [X = 7] … … . + p [X = n]

= 1 − {p [X = 0] + p [X = 1] + p [X = 2]}
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12.2 Life Distribution of Power Components: An Overview 223

{ }
e−2 20 e−2 21 e−2 22
=1− + +
0! 1! 2!
= 1 − (0.135 + 0.270 + 0.270) = 0.325

In 1000, the parts having at least three defective units are 0.3250 × 1000.0 = 325

Example 12.4
A machine produces defective items with a probability of 0.05 for modern power system applica-
tions. Such items are packed in a group of 20 for international marketing. Find the total packet
number, which contains almost, exactly and minimum of two defective items in a shipment con-
taining 1000 packets. Approximate the binomial distribution by Poisson approach.
Solution:
Given number of items in a group (n) = 20, and p = 0.05. Then m = n × p = 20 × 0.05 = 1.00
If X represents the overall count of defective items and its distribution follows poisons law.
Then the probability function will be as given below
e−m mx e−1
P(X = x) = =
x! x!
P [packets having at least two defective items]

= [X ≥ 2] = 1 − p [X < 2]

= 1 − {p [X = 0] + p [X = 1]}

= 1 − {e−1 + e−1 } = 1 − {0.3678 + 0.3678} = (1 − 0.7356) = 0.2644

Packets containing at least 2 defective items = 1000 × 0.2644 = 264

P [a packet with two defective items]
e−1
= p[X = 2] = = 0.1839
2!
The number of packet with the exactly two item is defective = 1000 × 0.1839 = 184

P [X ≤ 2] = p [X = 0] + p [X = 1] + p [X = 2]
e−1
= e−1 + e−1 + = 0.9193
2!
Packets having almost 2 items defective = 1000.0 × 0.9193 = 919.

12.2.4 Geometric Distribution

If trial number n is not fixed in an experiment and first success achieved on the Xth trial with prob-
ability p, preceded by (x − 𝟏) failures, where in (x − 𝟏) trials, the probability is (𝟏 − p)(x − 𝟏). If a
quantity varying randomly and assumes positive values only i.e X then it is distributed as geometric
distribution and the mass function is expressed as:
{ x−1
pq , x = 1, 2, 3, … 0 < p ≤ 1, q = (1 − p)
P(X = x) = (12.4)
0, otherwise
Above equation shows the geometric distribution of probability.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
224 12 Modern Aspects of Probabilistic Distributions for Reliability Evaluation of Engineering Systems

Example 12.5
A circuit breaker operates in independent fashion in an interconnected composite electric power
system. The probability of successful operation of the circuit breaker during any one fault is 0.70.
(i) What will be the probability of protection on 10th attempt?
(ii) Find the probability of breaker to attempt <4?
(iii) What will be the probability of an even number of attempts?
Solution:
Given p = 0.70 and q = (1–p) = 0.30
−1
(i) Success probability on the 10th trial = (0.70) (0.30)10 = (0.70)(0.30)9 = 0.0000139.
(ii) The probability of taking less than four attempts is
∑
4−1
∑
3
P(X < 4) = pqn−1 = (0.7)(0.3)n−1
n=1 n=1
= (0.7)[(0.3)0 + (0.3)1 + (0.3)2 ] = (0.7)[1 + 0.3 + 0.09]
= (0.7)(1.39) = 0.973
(iii) The probability of taking an even number of attempts.
∑
∞
n−1 ∑
∞
n−2 1
(0.7)(0.3)2 = (0.7)(0.3) (0.3)2 = (0.7)(0.3) = 0.23
n=1 n=1
1 − (0.3)2

12.2.5 Weibull Distribution

The Weibull and exponential distributions are both closely related. Both distributions provide a
valid model for the entire length of life for very many devices. For reliability, modeling Weibull
distribution is widely used because of its flexibility of algebraic tractability and modeling differ-
ent types of hazard behavior. One representation of the extreme value distribution is the Weibull
distribution (Saket et al. 2006).
The formula representing the Weibull PDF for a random variable X is given as follows:
{ 𝛽
𝛼𝛽x𝛽−1 e−𝛼x , for x > 0, 𝛼 > 0, 𝛽 > 0
f(x) = (12.5)
0, elsewhere
If a random variable follows a Weibull distribution and has a value below the integral
a 𝜷
∫𝟎 𝜶𝜷x𝜷−𝟏 e−𝜶x dx, then probability is evaluated by changing the variable, y = x 𝜷 we get
𝜶𝜷 𝜷
∫𝟎 𝜶e−𝜶y dy = 𝟏 − e−𝜶𝜶 which shows that random variable y is also distributed exponentially.

12.2.5.1 Weibull Distribution Mean and Variance

If the Weibull distribution’s parameters are represented as 𝛼 and 𝛽, then the mean can be deter-
mined by calculating the integral.
x ∞
𝜷 𝜷
Mean 𝝁 = x(𝜶𝜷x𝜷−𝟏 e−𝜶x )dx = 𝜶𝜷x𝜷 e−𝜶x
∫𝟎 ∫𝟎
Put u = 𝜶x𝜷 ⇒ du = 𝜶𝜷x𝜷−𝟏 dx
Also if x = ∞ ⇒ u = ∞ and x = 𝟎 ⇒ u = 𝟎
∞ ( )
𝜶𝜷 u −u 𝟏 (𝟏∕𝜷)−𝟏
= 𝟏∕𝜷 e u du
𝜶 ∫𝟎 𝜶 𝜷
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12.2 Life Distribution of Power Components: An Overview 225

∞ ∞ ( )
𝟏
+𝟏 −𝟏
= 𝜶 −𝟏∕𝜷 u𝟏∕𝜷 e−u du = 𝜶 −𝟏∕𝜷 e−u u 𝜷 du
∫𝟎 ∫𝟎
( )
𝟏
= 𝜶 −𝟏∕𝜷 𝚪 𝟏 + (12.6)
𝜷
∞
(∵Γn =) e−x xn−1 dx
∫0
Use the substitution u = 𝛼x𝛽 to evaluate the variance
∞
𝜷
E(X𝟐 ) = 𝜶𝜷x𝜷+𝟏 e−𝜶x dx
∫𝟎
( )
𝟏+ 𝜷𝟐
𝜶 −𝟐∕𝜷 𝚪 check carefully
Then, variance
[ ( ) ( ( ))2 ]
2 1
𝜎 = E[(X − 𝜇) ] = E[X ] − 𝜇 = 𝛼
2 2 2 2 −2∕𝛽
Γ 1+ − Γ 1+ (12.7)
𝛽 𝛽

12.2.6 Normal Distribution

A continuous random variable X is distributed according to the Gaussian curve or normal curve
using 𝜇 and 𝜎, and the PDF is expressed as follows:
( )2
1 −1 x−𝜇
f (x) = √ e 2 𝜎 ; −∞ < x < ∞ (12.8)
𝜎 2𝜋
X has a standard deviation of 𝜎 and a mean of 𝜇, and is distributed as N(𝜇, 𝜎). When the
normal curve covers the entire area under the curve and the X-axis, the area beneath the curve
and between two ordinates X = a and X = b, where a < b, signifies the probability of X falling
[between a] and b. The probability is represented as P[a < X < b]. If X is a variable and expressed as
𝟐
X−𝝁 𝟏 −Z
Z= 𝝈
, Eq. (12.8) is replaced as 𝝓(Z) = √ e 𝟐 , −∞ < Z < ∞ it is known as standard normal
𝟐𝝅
distribution. After substituting 𝝁 = 0 and 𝝈 = 1 and changing x into Z this is achieved. Now x has
distribution as ( )
N(𝝁, 𝝈) then Z = X−𝝁 𝝈
has the distribution N(0, 1).
Given by Eq. (12.8) are a few characteristics of the normal distribution, which are as follows:
Mean = 𝝁, Variance = 𝝈 2 , Standard deviation = 𝝈
∞
√
2 4
Average deviation about the mean = ∣ x − 𝜇 ∣ f (x)dx = 𝜎 = 𝜎 (approx.)
∫−∞ 𝜋 5
2
Quartile deviation = 𝜎
3

Function of N (𝜇, 𝜎) is moment generating.

( )
X −𝜇
∵Z =
𝜎
Mx (t) = M𝜎Z + 𝜇(t)
= e𝜇t Mz (𝜎t)
= e𝜇t ⋅ e𝜎tt∕2 (since Mz(t) = et
2 ∕2
verify it)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
226 12 Modern Aspects of Probabilistic Distributions for Reliability Evaluation of Engineering Systems

( ) ( ) ( )2
𝜎2t 𝜎2t
2
t 𝜇+ 𝜎2 t t t2
=e =1+ 𝜇+ + 𝜇+ + ……
1! 2 2! 2
∴ E(X) = 𝜇; E(X 2 ) = 𝜎 2 + 𝜇2, (12.9)

Thus we get 𝜇 1 = 0, 𝜇 2 = 𝜎 2 , 𝜇 3 = 0; 𝜇 4 = 3𝜎 4 , etc.

The normal distribution approach is used to compute the probability of hitting a shot, to estimate
and extract the statistical parameters in every field of engineering and technology. This approach
is used to calculate the errors made unknowingly in data measurement by experiments and to
evaluate the reliability of composite power systems.

Example 12.6
A particular type of randomly chosen diode is used in power converter applications in modern
power system. The breakdown voltage X of the diode is normally distributed with 𝜇 = 40 V and
𝜎 = 1.50 V. Find the probability of break down voltage (i) lies in the range of 39 V and 42 V, and
(ii) between 40 V and 43 V.
Solution:
Here X follow N (40 V, 1.50 V).

(i) To determine p[39 V ≤ X ≤ 42 V], express the event 39 V ≤ X ≤ 42 V in a standardized form:

39 − 40 X − 40 42 − 40
39 V ≤ X ≤ 42 V comes to be ≤ ≤
1.5 1.5 1.5
X−40
i.e. −0.67 ≤ Z ≤ 1.33Z = 1.5

P(39 V ≤ X ≤ 42 V) = p(−0.67 ≤ Z ≤ 1.33)

= 𝜙(Z = 1.33) − 𝜙(Z = −0.67)
= 0.9082 − 0.2514 = 0.6568
[ ]
40 − 40 X − 40 43 − 40
(ii) . p(40 V ≤ X ≤ 43 V) = P ≤ ≤
1.5 1.5 1.5
= 𝜙 [0 ≤ Z ≤ 2] = 𝜙 [Z = 2] − 𝜙 [Z = 0]
= 0.4773 − 0.0000 = 0.4773

Example 12.7
If the mean of lifetimes of certain kind of electronic devices is 300 hours and a standard deviation
is 25 hours. Assume the distribution of lifetime follows the normal curve and is measured to the
nearest hour.

(i) Determine the probability of a devices having life time > 350 hrs.
(ii) What proportion of items will exhibit lifetimes of 300 hours or less?
(iii) What percentage of items will have lifetimes ranging between 220 and 260 hours?
Solution:
Here X follows N(𝜇 = 300, 𝜎 = 25)

(i) p(X > 350)

[ ]
X − 300 350 − 300
= >
25 25
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12.3 Failure Distribution Functions for Reliability Evaluation 227

= 𝜙 [Z > 2] = 1 − 𝜙 (Z ≤ 2)

= 1 − 0.9772 = 0.0228

The resulting percentage is 2.28%

[ ]
X − 300 300 − 300
(ii) p(X = 300) = >
25 25
= 𝜙 (Z = 0) = 0.5000

The resulting percentage is 50%

(iii) p(220 ≤ X ≤ 260) = 𝜙(−3.2 ≤ Z ≤ − 1.6)
= 𝜙 (−1.6) − 𝜙 (−3.2)
= [1 − 𝜙 (1.6)] − [1 − 𝜙 (3.2)]
= [1 − 0.9452] − [1 − 0.9903]
= 0.0548 − 0.0007 = 0.0541

The resulting percentage is 5.41%

12.2.7 Gamma Distribution

X is a random variable of a continuous nature, and its distribution is according to the law of prob-
ability is given as:
⎧ e−x x𝜶−𝟏
⎪ ; 𝜶 > 𝟎, 𝟎 < x < ∞
f(x) = ⎨ 𝚪(𝜶) (12.10)
⎪ 𝟎, otherwise
⎩
It is named Gamma variable with 𝜶 as a parameter and Γ(𝛼) as a variant. The distribution of the
function is termed as Gamma distribution.
Notes: The probability function is defined above. Since,
∞ ∞
1
f (x)dx = e−x x𝛼−1 dx = 1 (12.11)
∫0 Γ(𝛼) ∫0
The PDF of a random variable varying continuously and following gamma distribution with
parameter (𝛼, u) is given as:
⎧ 𝛼u
⎪ e−ax xu−1 ; 𝛼, u > 0,0 < x < ∞
f (x) = ⎨ Γ(u) (12.12)
⎪ 0, otherwise
⎩

12.3 Failure Distribution Functions for Reliability Evaluation

Principally, to model equipment longevity, any distribution function may be used. In practice, the
most realistic approach seems to have monotonic hazard functions of distribution, and in this cat-
egory, only a few provide reasonable models of device reliability. The following are the main life
distribution models used:
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
228 12 Modern Aspects of Probabilistic Distributions for Reliability Evaluation of Engineering Systems

12.3.1 Evaluation of Reliability Based on Exponential Distribution

In the context of reliability modeling, the exponential distribution is widely utilized. It possesses
algebraic simplicity and tractability and denotes the device life cycle’s functional life interval (Saket
et al. 2007). If failure times are exponentially distributed, results of great importance can be derived
by connecting the mean time between failures (MTBF) of systems connected serially and parallelly.
To begin with, for any component, let’s initially establish a connection between reliability and ser-
vice time T. Utilize the fact that
t ( )
dF(t)
R(t) = 1 − F(t) = 1 − f (x)dx ∵ f (t) =
∫0 dt
The PDF of the time to failure T is given if it follows exponential distributions with parameter 𝛼,
f (t) − 𝛼.e𝛼t , t ≥ 0.
Then,
t [ ]t
e−ax
R(t) = 1 − 𝛼e−𝛼x dx = 1 − 𝛼 = 1 + [e−𝛼t − 1] = e−𝛼t (12.13a)
∫0 −𝛼 o

Conversely, when
f (t) 𝛼e𝛼t
Z(t) = = −𝛼t = 𝛼, constant
R(t) e
from the equation, we get,
t
f (t) = 𝛼e∫0 𝛼dx

⇒ f (t) = 𝛼e−𝛼t , t ≥ 0 (12.13b)

This characteristic signifies that, in the context of reliability, the exponential distribution main-
tains a consistent failure rate distribution.
1
MTTF = E(T) = (12.14a)
𝛼
1
var(T) = 𝜎T 2 = 2
𝛼
R(T0 + t) e−𝛼(T0 +t)
R(t∕T0 ) = = −𝛼T = e𝛼t (12.14a)
R(T0 ) e 0
This signifies that the time taken by a component to fail does not depend on time duration when
the component functions well. Reliability remains the same regardless of components, new brand-
ing, and long operation time. This is referred to as the constant failure rate (CFR) distribution’s
memory-less attribute.

Example 12.8
A power component has MTTR = 20.0 hours and MTBF = 100.0 hrs. If failure and repair dis-
tributions follow the exponential path, evaluate the availability and unavailability of the power
component after a long time.
Solution:
1
Given MTBF = = 𝜆
= 100 ⇒ 𝜆 = 0.01
1
MTTR = = 20 ⇒ 𝜇 = 0.05
𝜇
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12.3 Failure Distribution Functions for Reliability Evaluation 229

Availability of component
1
A(∞) =
MTBF
= 𝜆 =
100
= 0.83
MTBF + MTTR 1 1 100 + 20
+
𝜆 𝜇
The unavailability of component
𝜆 0.01 0.01
A(∞) = = = = 0.1666
𝜆+𝜇 0.01 + 0.05 0.06

Example 12.9
The failure rate of an electro-mechanical energy conversion device is 0.023 per hour. The repair
time is exponential with a mean of 10 hours. Determine the reliability for 3 days and MTTF, if two
devices are operating in an active redundant configuration.
Solution:
Within an active redundancy setup
m1 m2
R(t) = em2 t = em1 t
m1 − m2 m1 − m2
√
where m1 , m2 = 12 {−(3𝜆 + 𝜇) ± 𝜆2 + 6𝜆𝜇 + 𝜇 2 }
1
Here 𝜆 = 0.023 and 𝜇 = 10 = 0.1
1 √
m1 , m2 = {−(0.069 + 0.1) ± (0.023)2 + 6 × (0.023) × (0.1) + (0.1)2 }
2
= −0.0065, −0.1625

Putting values in equation,

−0.0065 −0.1625t 0.1625 −0.0065t
R(t) = e + e
0.156 0.156
3 days =72 hours.

∴ R(72) = −0.0417 × e−11.7 + 1.0417 × e−0.468 = 0.6524

3𝜆 + 𝜇 3 × 0.023 + 0.1
MTTF = = = 159.7 hours.
2𝜆2 2 × (0.023)2

Example 12.10
The failure rate of a power component is 0.02 / 1000 hrs. For a constant failure rate, determine the
probability of satisfactory operation of a component for atleast 20,000 hrs? If a series carries 4 such
components is connected in series, determine reliability of the power component for 5000 hr.
Solution:
The power component experiences a constant failure rate, and its failure time distribution follows
an exponential distribution with a parameter of 𝜆 = 0.02.
The PDF of the power component is f (t) = 0.02e−0.02xt , t ≥ 0

(i) . Pr [t ≥ 20]
∞
= f (t)dt
∫20
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
230 12 Modern Aspects of Probabilistic Distributions for Reliability Evaluation of Engineering Systems

∞
= 0.02e−0.02xt dt
∫20

= −[e0.02×t ]∞
20 = e
−0.02×20
= 0.6703
(ii) The reliability function of the component is R(t) = e−𝜆xt . There is a series of 4 components
connected in series. The reliability of this combination is given below.
∴ R(t = 5) = e−5.0×4.0×0.2 = 0.183

Problem 12.11
A production company makes repairs on a power transformer oil tank by employing a single welder.
The welder begins working with three arc-welding units with the aim to increase welding operation
reliability. The failure rate of each welding unit is 0.32/month. If the units are utilized in a standby
mode by a welder with the assumption that unit failures are independent. Calculate the MTTF, and
standard deviation of the time to failure. What is the probability that a welder won't be without a
welding unit for the whole six-month welding work. The rule is that the broken welding equipment
will not be brought back to the project after repair.
Solution:
Given rate of failure 𝜆 = 0.320/month.
T is the time to failure, and the rate of failure = 𝜆 with exponential distribution.
Therefore the PDF is as f (t) = xe−0.32t , ≥ 0
= 0.32
and R(t) = e−𝜆t = e0.32t
Utilizing three arc-welding units, the welder begins the job.
∞
MTTF = 3e−0.32t dt = 9.375
∫o
3 3
Var(T) = 𝜎T 2 = = = 29.297
𝜆2 (0.32)2
P(T > 6) = R(t = 6) = e−0.32×6 = 0.1466.

12.3.2 Reliability Evaluation Based on Weibull Distribution

The strength of the tensile test specimen is observed and described by this distribution. When the
rate of failure of a component increases or decreases with time, the Weibull distribution describes
𝛽
the failure times with parameters 𝛼 and 𝛽. The distribution is given below f (t) = 𝛼𝛽t𝛽−1 e−𝛼t ,
t > 0, 𝛼 > 0, 𝛽 > 0; the Weibull type failure-time distribution’s reliability function may be stated as
the following:
𝛽
R(t) = e−𝛼t (12.15a)
Z (t) = 𝛼𝛽t𝛽−1 Z (t) = t−1 provides the failure rate associated with the Weibull distribution.
f (t) 𝛼𝛽t−1 e−𝛼t𝛽
Then 𝜆(t) = = (12.15b)
R(t) e−𝛼t𝛽
(
𝛽 t ) 𝛽−1 1
𝛼𝛽t𝛽−1 = where 𝜃 𝛽 =
𝜃 𝜃 𝛼
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12.3 Failure Distribution Functions for Reliability Evaluation 231

Thus, we have
( )
1
MTTF = E(T) = 𝜃Γ 1 + (12.16a)
𝛽
MTTF is the Weibull model of mean time to failure.
The Weibull model variance is
{ ( ) [ ( )2 ]}
2 1
Var(T) = 𝜎t = 𝜃
2 2
Γ 1+ − Γ 1+
𝛽 𝛽
R(t + T0 )
Now R (t∕T0 ) =
R(T0 )
[ ( )𝛽 ]
t + T0
exp − [ ( ) ( )𝛽 ]
𝜃 t + T0 𝛽 T0
= [ ( ) ] = exp − + (12.16b)
T 𝛽 𝜃 𝜃
exp − 0
𝜃

The Weibull distribution is commonly employed for reliability modeling, offering the ability to
represent various hazard behaviors and exhibiting algebraic simplicity. Furthermore, by utilizing
two adjustable parameters, it can be effectively adapted to fit numerous real-world scenarios.

Example 12.12
If a power system has a scale parameter of 550 days and a shape parameter of 1.4 and it shows
Weibull type of failure distribution, then find: (i) R (100 days); (ii) MTTF; (iii) standard deviation;
and (iv) the design life assuming 90% reliability.
Solution:
PDF of the Weibull distribution is expressed as follows:
{ ( )𝛽 }
𝛽 ( t )𝛽−1 t
f (t) = exp − ,t ≥ 0
𝜃 𝜃 𝜃
Now 𝜃 = 550 days and 𝛽 = 1.4
(i)
{ ( )𝛽 }
t
R(t) = exp −
𝜃
{ ( ) }
100 1.4
∴ R(100) = exp − = 0.9122.
550
(ii)
( )
1
MTTF = 𝜃 × Γ 1 +
𝛽
( )
1
= 550 × Γ 1 + = 550 × 0.91057
1.4
= 500.8 days

(iii)
{ ( ) ( ( ))2 }
2 1
Var = 𝜃2
Γ 1+ − Γ 1+
𝛽 𝛽
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
232 12 Modern Aspects of Probabilistic Distributions for Reliability Evaluation of Engineering Systems

{ ( ) ( ( ))2 }
2 1
= 5502 × Γ 1+ − Γ 1+
1.4 1.4
= 5502 × {Γ(2.43) − (Γ(1.71))2 }

= 5502 × ⌊(1.43 × 0.8860) − (0.91057)2 ⌋

S.D. = 550 × 0.66174 = 363.96 days.

(iv) Let tD be the required design life for R = 0.90
{ ( ) }
tD 1.4
∴ R(tD ) = exp − = 0.90
550
( )1.4
t
⇒ D = 0.10536
550
⇒tD = 550 × (0.10536)1∕1.4 = 110.2 days

12.3.3 Normal Distribution-based Reliability Evaluation

Normal distribution also contributes to model device life length and popular one. For structural
components of power system, it is the most appropriate model to evaluate the reliability. The PDF
of T is given by, if T (time to failure) has a normal distribution with N(𝜇, 𝛼):
[ ]
1 −(t − 𝜇)2
f (t) = √ exp , −∞ < t < ∞. (12.17a)
𝜎 2𝜋 2𝜎 2
This case is Var (T) = 𝜎 T 2 = 𝜎 2 and MTTF = E(T) = 𝜇.
∞
Standard normal integrals and normal tables can be used to determine R(t) = ∫t f (t)dt.
f (t)
The conditional reliability determined by is thus called 𝜆(t) = R(t) hazard function or instanta-
neous failure rate of any component:
{ }
T +t T0
R(t∕T0 ) = P T > 0 > T0 = e− ∫T 𝜆(t)dt (12.17b)
T

12.4 Use of Exponential Model to Evaluate Reliability and MTBF

Very well, accepted modeling methodology for reliability assessment is the exponential distribution
(Saket et al. 2009). With this model the reliability and MTBF of systems connected in parallel and
series, can be evaluated as follow.

12.4.1 Components Connected in Serial

Let’s suppose a system has n components connects in series, and the components have their failure
rate 𝛼 1 , 𝛼 2, ..𝛼 n , respectively. Therefore, the law of product of reliability can be expressed as
∑n
𝛼i
Rs (t) = e−t i=1 (12.18)
The MTBF for the series system is given as,
1
𝜇s = (12.19)
1 1 1
+ + … .. +
𝜇1 𝜇2 𝜇n
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12.5 Probabilistic Methods For Reliability Evaluation 233

where MTBF of the ith component is 𝜇 i . If special case exists with CFR 𝛼 for all n components and
have the same 𝜇, the system’s MTBF and its failure rate n𝛼 is 1/𝜇𝛼 = 𝜇/n.

12.4.2 Parallel System

Obtaining the mean time to failure for a parallel system is generally a challenging task, but if
we consider the special case with the same failure rate 𝛼, of all the components are useful and
interesting, it is possible to get a result. The system reliability function for this unique case can be
considered as,
( ) ( )
n −𝛼t n −2𝛼t
Rp (t) = 1 − (1 − e−𝛼t )n = e − e + … … + (−1)n−1 e−n𝛼t
1 2
Utilizing that,
fp (t) = R′p (t),
( ) ( )
n −𝛼t n −2𝛼t
We obtain, fp (t) = 𝛼 e − 2𝛼 e + … … + (−1)n−1 n𝛼e−n𝛼t (12.20)
1 2
∞
and gives the failure-time distribution’s mean 𝜇p = ∫0 t.fP (t)dt
( ) ∞ ( ) ∞ ∞
n n
𝛼 te−𝛼t dt − 2𝛼 te−2𝛼t dt + … … + (−1)n−1 n𝛼 te−n𝛼t dt
1 ∫0 2 ∫0 ∫0
( ) ( )
1 n 1 n 1
= − + … . + (−1)n−1 (12.21)
𝛼 1 2𝛼 2 n𝛼
(
Then MTBF for system in which components are connected parallelly is 𝜇p = 𝛼1 1 + 12 + … …
)
… … + n1 . In MTBF system consists of “n” component with the same failure rate 𝛼 therefore
whenever the system fails, each defective component is replaced. Using two parallel components
when only one rathers. Instead of being doubled, the pair’s mean time to failure is 50% longer than
that of the single component.

12.5 Probabilistic Methods For Reliability Evaluation

If a system performs its function adequately over an intended period of time subjected to some
constraints called operating conditions then its probability is called reliability. The basic attributes
of reliability are probability of success, time period, adequate function and operating conditions.
The probabilistic methodologies for estimation of reliability functions have been described in this
section successfully. The probabilities of success and failure and indices of reliability are evalu-
ated by applying the Gaussian distribution method (GDA), safety factor concept (SFC), peak load
consideration (PLC), and Simpson 1/3rd rule.

12.5.1 Gaussian Distribution Approach (GDA)

Over a specified time interval, load model for Gaussian distribution is given by:
( )2
Pd −Pd
1 −0.5 𝜎d
f (Pd ) = e (12.22)
𝜎d√2𝜋
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
234 12 Modern Aspects of Probabilistic Distributions for Reliability Evaluation of Engineering Systems

where
Pd = Demand on power generation station.
Pd = Average load on the power plant.
𝜎 d = Standard deviation of load demand.
The model of total generation capacity of is Gaussian distributed systems.
[ ]2
1 −0.5 C−C
f (C) = √ e 𝜎c (12.23)
𝜎c 2𝜋
where
C = Generation capacity of power plant
C = Mean generation capacity of power plant
𝜎 c = Standard deviation for generation capacity.
The description of the probability of failure (PF ) for the load model of the developed system and
the generation mentioned above is as follows:
PF = (1 − Ps ) (12.24)
The probability of success (Ps ) of develop the power system is using GDA can be expresses as:
∞ C [ ]2 ( )2
Pd−Pd
1 −0.5 C−C
−0.5
Ps = e 𝜎c e 𝜎d
dc ⋅ dPd (12.25)
∫−∞ ∫−∞ 2𝜋𝜎c 𝜎d
where
PF = Failure probability of power generation station.
PS = Success probability of power generation station.
C = Generation capacity of power plant.
C = Mean generation capacity of power plant.
C−C Pd −Pd
Putting, x = 𝜎c
and y = 𝜎d
According to condition of substitution above, equation can be written in simplified way as given
below.
∞ [ z ]
2 2
Ps = e−0.5(x +y ) dy dx (12.26)
∫−∞ ∫−∞
𝜎c x + C − Pd
where Z =
𝜎d
Further, consider the substitutions made as follows,
x′ = y.sin 𝜃 + x.cos 𝜃
y′ = y.cos 𝜃 − x.sin 𝜃
With the above transformation, the axis rotates by an angle of 𝜃. Equation (12.26) is modified by
following substitutions.
𝜎 𝜎
Let sin 𝜃 = √ 2c 2 and cos 𝜃 = √ 2d 2
Pd +𝜎C Pd +𝜎C

The above substitution is shown as follows in equation (12.27). The probability of success after
above substitution comes:
∞ 𝛽
1 −0.5(x′2 +y′2 ) ′ ′
Ps = e dy dx (12.27)
∫−∞ ∫−∞ 2𝜋
C − Pd
where 𝛽 = √
𝜎d2 + 𝜎C2
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12.5 Probabilistic Methods For Reliability Evaluation 235

where 𝛽 is the limit, and it becomes out to be x-independent. It is further simplified as follows:
[ ]
𝛽 ∞
1 −0.5(x′2 ) ′ −0.5(y′2 ) ′
Ps = √ e dx e dy
∫−∞ ∫−∞ 2𝜋
+𝛽
1 −0.5(y′2 ) ′
Ps = √ e dy = 𝜑(𝛽) (12.28)
∫−∞ 2𝜋
If we have normally distributed curve and the curve’s area under it has a standard deviation
𝛽
∫−∞ [N(0,1)] = 1 and mean of 0, it gives numerical value of 𝜑 (𝛽), which is conveniently obtained
from the table of standard normal distribution. Curves illustrating the relationship between
failure probability and generating capacity have been plotted for different capacity and load
models.
By predicting the various load levels Pdi daily variations in the load can be accounted for. If the
relative frequency of occurrences is Lo , L1 , L2 , … , Li and the frequency of occurrences are 𝛼 0 𝛼 1
𝛼 2 , … , 𝛼 I for different levels, the total failure probability is provided by, and each load level failure
probability may be calculated as follows:
∑
PF = 𝛼i Pfi (12.29)
i

where

𝛼 I = Relative frequency of occurrences.

Pfi = Probability of failure of Li th load level.
From Eq. (12.29) it follows that

Pfi = 1 − 𝜙(𝛽N ) (12.30)

Hence, the overall failure probability is as shown below

∑
PF = 𝛼i [1 − 𝜙(𝛽i )] (12.31)
i

Equation (12.31) is used to calculate probabilities of failures. It is visible from the equation that
PF depends on both the parameters 𝛼 i and 𝛽 i and these parameters are expressed as

𝛼i = ni ∕n
Ci− − Li
𝛽i = (12.32)
𝜎d2 + 𝜎C2
i i

where

n = total study duration.

N i = time duration for occurrence of “Li th load level.”
Ci = Li th load level mean generation capacity.
Li = Mean Li th load level mean.

12.5.2 Safety Factor Concept (SFC) and Peak Load Consideration (PLC)-based
Evaluation of Reliability
The generation capacity of a plant shows the Gaussian distribution of probability therefore PDF is
Gaussian. Low-level loading is dominated by peak loading of the system, whereas, at the low load
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
236 12 Modern Aspects of Probabilistic Distributions for Reliability Evaluation of Engineering Systems

level conditions, PF is almost negligible. If the system’s peak load is Pdmax , the safety factor “S” is
defined as:
S = C∕Pdmax (12.33)
If S is a random variable and the generation capacity “C” is distributed normally, then at a con-
stant Pdmax , “S” is a safety factor distribution that will also be distributed normally. The function of
a safety factor is related to these parameters as follows:
( )
Pdmax s−C
P −0.5
fs = √dmax e 𝜎c
(12.34)
2𝜋𝜎c
The Gaussian mean distribution of the safety factor S.
C 𝜎
S= and 𝜎s = c (12.35)
Pdmax Pmax
The system’s failure probability is expressed as follows:
⎡ C ⎤
1
Pdmax −0.5
(
s−c∕Pdmax𝜎s
)2 ⎢1 − P ⎥
PF = √ e 𝜎s ds = 𝜙 ⎢ 𝜎c
d max ⎥ (12.36)
∫−∞ 2𝜋𝜎s ⎢ ⎥
⎢ P ⎥
⎣ d max ⎦
The probability of failure is obtained at different peak loads and capacity of generation using
equation (12.36) as described in example 12.13. Using concept of safety factor PF v∕s P C plots can
dmax
be drown for the adequacy assessment and reliability evaluation of the modern power system.

12.5.3 LOLP Evaluation using Simpson’s 1/3rd rule

The load duration curve is a realistic model to evaluate the failure probability of the modern power
system. The daily peak loads are arranged in descending order to prepare a cumulative load model.
The daily peak load variation curve (DPLVC) is drawn for the comprehensive analysis of load vari-
ations. This analysis results a load duration curve (LDC) model for reliability assessment. Normal
distribution function based on this system uses the generation model. Maximum average gener-
ation capacity contributes to the evaluation of reliability (Varshney and Saket 2014). The loss of
load probability (LOLP) occurs when the load probability exceeds the generation capacity. When
capacity of generation exceeds the system load level, a “loss of load” occurs. To build a multi-step
model, LDC is split into a number of steps. The approximation by increasing the number of approx-
imation stages, the quantity can be decreased. For estimation of the LOLP of a composite power
system, 100 steps of the stepped load duration curve (SLDC) are used. The different steps of the
SLDC show the load’s fluctuations over a 24-hour period. In this section, the Simpson’s 1/3rd rule
is used with the following approximation for evaluation of the LOLP index. The modern power
system’s LOLP applying SLDC is expressed as follows:
[ ( )2
]
100 Pd (t)
t 1 −0.5 c−C
LOLP = √ e 𝜎c dc dt (12.37)
∫0 100 ∫−∞ 2𝜋𝜎 c
C−C
Putting, =Z
𝜎C
Using Eq. (12.37) for 100 steps of SLDC, the LOLP is expressed as follows:
[ ]
100 Pd (t)
t 1 −0.5(z)2
LOLP = √ e dz dt (12.38)
∫0 100 ∫−∞ 2𝜋𝜎 c
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12.5 Probabilistic Methods For Reliability Evaluation 237

( )
100
t Pd (t) − C
LOLP = 𝜑 dt (12.39)
∫0 100 𝜎c
Area under the normal distribution curve [ [N ](0,1)] represents the inner integral for different
P (t)−C
value of ‘t’, and it has been replace by 𝜙 d 𝜎 function. The expression of Eq. (12.39) can be
C
evaluated by any method of the area evaluation. By assuming LDC as SLDC with a maximum of
100 steps, LOLP is evaluated. It is possible only if a multilevel representation of the daily load
duration curve is used. This approach is similar to the Markov load model used in calculation of the
frequency and duration. The electric load varies many times in steps, if the power system operates
for 24 hour per day. The SLDC is an important attribute for reliability evaluation and planning of the
modern power system. The following numerical example 12.13 is taken for reliability evaluation of
an interconnected power system using various probabilistic approaches.

Example 12.13
The generation capacities of an interconnected composite power system are given below:
(i) 2 × 200 = 400 MW Nuclear power plant
(ii) 10 × 50 = 500 MW Thermal power plant
(iii) 3 × 100 = 300 MW Hydro power plant
Daily peak load variation is from 300 MW to 1000 MW, which is measured annually.
(a) Prepare the generation capacity state table including the standard deviation data for graphical
illustration and power system reliability evaluation.
(b) Calculate the failure probabilities for (i) constant mean load of 480 MW, (ii) constant generation
capacity of 500 MW, and (iii) provide the graphical illustrations for reliability assessment of the
power system using Gaussian distribution approach.
(c) Calculate the failure probabilities of power system considering peak loads and safety factor
concept for (i) constant load of Pd = 480 MW, (ii) peak load variation from Pdmax = 480 MW
to Pdmax = 510 MW, and (iii) provide the graphical illustration for reliability analysis of power
system.
(d) Evaluate the LOLP of power system for Pdmax = 1000 MW and Pdmin = 300 MW for 10 steps of
the stepped load duration curve (SLDC) and draw the LOLP v/s generation capacity curves for
analysis of the modern power system.
Solution:
(a) The generation capacity, state data and data of standard deviation of power stations are as given
in the table below. The subscript 0, 1, 2, 3, 4, etc., present the generation unit availability in
power system due to nuclear (X), thermal (Y ), and hydro (Z) power plants (Table 12.1).
(b) Failure probability evaluation using Gaussian distribution approach
(i) Evaluation of failure probability at constant 480 MW mean load
Here, we are evaluating failure probability by keeping the mean load constant at 480MW. GDA
equation and data table of standard deviation are utilized. The standard deviation for the mean load
and generating capacity is assumed to be 10% of the mean loads and generating capacity, respec-
tively. The failure probability at generation capacity C = 490 MW is Pf = 0.3853. Similarly, using
this equation, the failure probability can be calculated considering available generating capacity
states at 480 MW fixed mean load.
Many plots of generation capacity versus failure probability are drawn for different values of
standard deviations. The failure probabilities are illustrated in curve A (𝜎 d = 5% of load and 𝜎 c = 5%
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
238 12 Modern Aspects of Probabilistic Distributions for Reliability Evaluation of Engineering Systems

Table 12.1 The generation capacity state data and standard deviation data of power stations.

Generating
Capacity capacity available
Sr. no. state in MW 𝝈 c = 5% 𝝈 c = 10% 𝝈 c = 15% 𝝈 c = 20% 𝝈 c = 25%

1 X 1 Y 0 Z0 200 10 20 30 40 50
2 X 2 Y 0 Z0 400 20 40 60 80 100
3 X 2 Y 1 Z0 450 22.5 45 67.5 90 112.5
4 X 2 Y 2 Z0 500 25 50 75 100 125
5 X 2 Y 3 Z0 550 27.5 55 82.5 110 137.5
6 X 2 Y 4 Z0 600 30 60 90 120 150
7 X 1 Y 5 Z2 650 32.5 65 97.5 130 162.5
8 X 1 Y 6 Z2 700 35 70 105 140 175
9 X 2 Y 7 Z0 750 37.5 75 112.2 150 187.5
10 X 1 Y 8 Z2 800 40 80 120 160 200
11 X 2 Y 9 Z0 850 42.5 85 127.5 170 212.5
12 X 1 Y 10 Z 2 900 45 90 135 180 225
13 X 2 Y 10 Z 1 1000 50 100 150 200 250
14 X 2 Y 10 Z 2 1100 55 110 165 220 275
15 X 2 Y 10 Z 3 1200 60 120 180 240 300

of capacity), curve B (𝜎 d = 15% of load and 𝜎 c = 5% of capacity), curve C (𝜎 d = 5% of load and

𝜎 c = 15% of capacity), and curve D (𝜎 d = 15% of load and 𝜎 c = 15% of capacity). The curves are
labeled for different combinations of 𝜎 c and 𝜎 d . By comparing A and B, it is found that PF at the
same generating capacity with A is less than B. This is because of the large amount of uncertainty
associated with the distribution function of generating capacity. Similarly, for the same generating
capacity, if we observe the curves A and C, it can be seen that C has a lower failure probability than
B, and so on. The failure probability for total generating capacity is shown by curve D, the highest
among all others. With an increase in both 𝜎 c and 𝜎 d , or both simultaneously, the chance of failure
rises (Figure 12.1).
(ii) Evaluation of failure probability keeping constant generation capacity
In this section, the PF is evaluated using GDA, keeping generation capacity at a constant value
i.e. 500 MW. For mean load and generation capacity, the standard deviation is assumed to be 5%
of both. The failure probability at generation capacity C = 500 MW is Pf = 0.4432. Then, at a mean
capacity = 500 MW (fixed), the failure probability is calculated by increasing the load steps by 5 MW.
In the figure given below, different plots of PF versus capacity of generation at different values of
standard deviation are shown. The failure probabilities as per the curve: There are four curves:
curve A (𝜎 d = 10% of load and 𝜎 c = 10% of capacity), curve B (𝜎 d = 20% of load and 𝜎 c = 10% of
capacity), curve C (𝜎 d = 10% of load and 𝜎 c = 20% of capacity), and curve D (𝜎 d = 20% of load and
𝜎 c = 20% of capacity). If the mean load and generating capacity are both equal, the PF and success
are precisely equal to 50%. It is the same in all cases, and the independents of the standard deviation
is shown in the figure given below. Below the point of intersections PF increases with an increase
− −
of 𝜎 c and 𝜎 d above the points; it varies oppositely. Using GDA for 𝜎 d = 10% of L and 𝜎 C = 10% of C
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12.5 Probabilistic Methods For Reliability Evaluation 239

Failure probability versus Generating capacity

0.6
A
0.5 B
C
D
Failure probability

0.4

0.3

0.2

0.1

0
485 490 495 500 505 510 515 520 525
Generating capacity

Figure 12.1 P F versus C plot at constant load of the developed system.

Failure probability versus Mean load

0.7
A
0.65 B
C
Failure probability

D
0.6

0.55
X: 500
Y: 0.5
0.5

0.45

485 490 495 500 505 510 515 520 525

Mean load

Figure 12.2 Plot of P F v/s generation capacity at constant capacity.

the overall PF is evaluated as PF = 0.0594. If 𝜎 c and 𝜎 d are increased from 10 to 20%, the PF overall
also increases by the same amount of capacity of generation and load level at fix value of frequency
− −
of occurrence. At 𝜎 c = 20% of C and 𝜎 d = 20% of L, the failure probability PF is 0.0986 (Figure 12.2).
(c) Using SFC and PLC evaluate the PF
See Table 12.2.
(i) At constant load evaluation of PF
This section discusses the adequacy evaluation of the capacity of generation based on worst-case
loading condition and safety factor. During evaluation, it is assumed that the maximum load on
to the system is Pdmax = 480 MW. Considering the approximations, curves have been plotted at dif-
ferent values of standard deviations between the safety factor (C/Pdmax ) and PF. Curve: A is for
𝜎 c = 10% of C, curve B is for 𝜎 c = 12% of C, curve C is for 𝜎 c = 15% of C, and curve D is for
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
240 12 Modern Aspects of Probabilistic Distributions for Reliability Evaluation of Engineering Systems

Table 12.2 Mean capacity for different load levels.

−
Li 1000 Mw 1200 Mw 1500 Mw 1800 Mw 2000 Mw
−
Ci 1250 Mw 1500 Mw 1875 Mw 2250 Mw 2500 Mw
𝛼i 0.1 0.2 0.4 0.2 0.1

Failure probability versus Safety factor

0.55
0.5
A
0.45 B
Failure probability

C
D
0.4
0.35
0.3
0.25
0.2
0.15
1 1.01 1.02 1.03 1.04 1.05 1.06 1.07 1.08 1.09 1.1
Safety factor

Figure 12.3 Plot of failure probability v/s safety factor at constant load.

𝜎 c = 18% of C. Observations of the curves A and B state that, the PF with 𝜎 c = 12% of C is more
than 𝜎 c = 10% of C for the same factor of safety and capacity of generation, if we increase 𝜎 c for
any value of safety factor and generation capacity Pf also increases. For all generation capacity
values, PF decreases with an increase in safety factor. For any value of factor of safety factor of
capacity of generation, the PF increases as the standard deviation rises. Given figure presents PF
versus estimate of the safety factor at changing capacity of generation and constant load. For the
10.5% factor of safety, PF increases with an increase in standard deviation of capacity of generation
(Figure 12.3).

(ii) Evaluation of failure probability under varying load

If load Pdmax is increased from 480 MW to 510 MW, keeping the safety factor’s value (C/Pdmax ),
different curves are drawn as below: Curves A and B are for 𝜎 c = 10% of C, Curve C is for 𝜎 c = 12% of
C, Curve D is for 𝜎 c = 15% of C, and Curve E is for 𝜎 c = 18% of C. With an increase in 𝜎 c, PF increases
at the same value of capacity of generation because of large amount of uncertainty involved in
the distribution factor of capacity of generation. When the safety factor is 1, at variable load PF is
constant and equal to 50% in all the cases as drawn in curves. If factor of safety is below one, the
PF for 𝜎 c = 10% of C is more than 𝜎 c = 12% of C. If safety factor is greater than 1, the probability
of failure varies oppositely. Ratio of capacity of generation to peak load demand is called factor of
safety, and it is one for equal values of peak load and capacity of generation. At this instant, both the
probabilities are 50%, as shown in figure. If capacity of generation is more than load, PF of curve
more than B, C, and D, A will be. Similar to this, curve A has a lower failure probability than curve
B and vice versa if load exceeds generating capacity (Figure 12.4).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12.5 Probabilistic Methods For Reliability Evaluation 241

Failure probability versus Safety factor

0.75
0.7 A
B
0.65 C
D
Failure probability

0.6
0.55 X: 1
Y: 0.5
0.5
0.45
0.4
0.35
0.3
0.95 0.96 0.97 0.98 0.99 1 1.01 1.02 1.03
Safety factor

Figure 12.4 Plot of probability of failure v/s safety factor at variable load.

0.1
A
0.09 B
D C
0.08 D
C
Loss of load probability

0.07
0.06
0.05 B
A
0.04
0.03
0.02
0.01
0
500 600 700 800 900 1000 1100 1200 1300 1400
Generating capacity (vv)

Figure 12.5 Plot of LOLP v/s capacity of generation for 10 steps.

(d) Simpson 1/3rd rule to evaluate LOLP

To evaluate reliability uses the Simpson 1/3rd rule Pdmax = 1000 MW and Pdmin = 300 MW is
taken. LDC becomes straight line. LDC becomes a straight line. In the mentioned example of SLDC,
a maximum of 100 steps are used using the Simpson 1/3rd rule approximation. During 24 hours of
operation of power system load of the systems varies in many steps. Let’s have a look at the following
figure, which represents the relation between the capacity of generation and LOLP for 10 steps of
SLDC with various 𝜎 c . The curves A, B, C and D as shown in Figure 12.5 indicate the relationship of
LOLP with capacity of generation at different values of 𝜎 c for ten steps of the SLDC. The percentage
values are 10%, 15%, 20%, and 25% respectively. The LOLP increases with an increase in 𝜎 c because
large uncertainty involved in the distribution function of capacity of generation. Also, at a particular
value of capacity of generation, LOLP increases as standard deviation increases. Consider 10 steps
of SLDC; during a 24-hour operation, there were a total of 10 load variations (Figure 12.5).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
242 12 Modern Aspects of Probabilistic Distributions for Reliability Evaluation of Engineering Systems

12.6 Additional Solved Examples

Example 12.14
The MTBF of a protective relay is 0.8 year with random failure occurrence.

(a) Calculate the probability of survival of the relay for 1 year without failure.
(b) Find the probability of more than 2 failures of the relay within a year.
(c) Find the per year expected number of failure.
Solution:
The number of failures during the interval of t follows a Poisson process since failures
−𝜆t
are random events, given by P[N(t) = n] = e n!(𝜆t)n , n ≥ 0 exponential distribution with a mean of
1/𝜆 the time between failures.
Now
1
MTBF = = 0.8 year
𝜆
1
∴𝜆 = 0.8
per year =1.25 per year
(a)
e−𝜆 𝜆0
P[N(1) = 0] = = e−1.25 = 0.2865
0!
(b)
[ ] [ ]
𝜆 0 𝜆1 𝜆2 (1.25)2
P[N(1) > 2] = 1 − e−𝜆 + + = 1 − 0.2865 1 + 1.25 + = 0.1315
0! 1! 2! 2
(c)

E[N(t)] = 𝜆t

∴ E {number of failure per year} = 𝜆 = 1.25.

Example 12.15
Six power components with identical nature are interconnected in bulk power system with constant
failure rates. The components are associated with (a) High redundancy with three components
per sub-system, and (b) Low redundancy with two components per sub-system. In each example,
determine the MTTF to necessarily get the 90% reliability in100 hours of operation.
Solution:
R = e−𝜆t for each component, if the failure rate for each component is constant.

(a) For high redundancy:

Rs (t) = 1 − [1 − {R(t)}3 ]2 = [1 − (1 − e−3𝜆t )2 ]

∴ Rs (100) = [1 − (1 − e−300𝜆 )2 ] = 0.90
⇒ (1 − e−300𝜆 )2 = 0.10
⇒ e−300𝜆 = 0.6837
∴ 300𝜆 = 0.3801 ⇒ e−300𝜆 = 0.3162
1 300
∴ MTTF of each component = = 789.2 hours.
𝜆 0.3801
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12.6 Additional Solved Examples 243

(b) For low redundancy:

Rs (t) = [1 − {1 − [R(t)]}2 ]3 = [1 − {1 − e−𝜆t }2 ]3
∴ Rs (100) = [1 − {1 − e−100𝜆 }2 ]3 = 0.90
1 − (1 − e−100𝜆 )2 = 0.9654, = (1 − e−100𝜆 )2
= 0.345, = 1 − e−100𝜆 = 0.1857, = e−100𝜆 = 0.8142,
= 100𝜆 = 0.20551
1 100
∴ MTTF of each components = = = 486.6
𝜆 0.20551

Example 12.16
If two units, standby and primary, are carried by a health monitoring system employing power
transformer, and if MTTF of standby unit is 333 hours and MTTF of primary unit is 1000 hours
with number of failures in standby mode of backup unit. Estimate the design life for dependability
at 90% if the primary unit requires repairs at a rate of 0.01 per hour while the standby unit is in use.
Solution:
For the standby redundant system,
m1 m2
R(t) = em2 t − em1 t (12.40)
m1 − m2 m1 − m2
where roots of the equation are m1 and m2
m2 + (𝜆1 + 𝜆2 + 𝜇)m + 𝜆1 𝜆2 = 0 (12.41)
1 1
Here 𝜆1 = = 0.001∕hour ; 𝜆2 =
1000
= 0.003∕hour and 𝜇 = 0.01/hour
333
Substituting the above values in Eq. (12.41), we get
m2 + 0.014 m + 0.000003 = 0
√
−0.014 ±(0.0 − 14)2 − 4 × 0.000003
∴ m1 , m 2 = = (0.00022 − 0.01378)
2
Put the values of m1 and m2 in Eq. (12.40)
R(t) = −0.01622 × e−0.01378t + 1.01622 × e−0.00022t
At 90% reliability, design the life of D is given by 1.01622 × e−0.00022t − 0.01622 × e−0.01378D = 0.90,
solve the equation by trial, we get D = 550 hours

Example 12.17
If a newly purchased distribution transformer fails with a constant rate of 0.03 per day. If we assume
continuous use and constant rate of repair equal to 0.2 per day.
(a) Calculate the steady-state availability and interval availability for the first 30 days.
(b) Define the standby unit’s steady-state availability. In standby, assume no failures.
(c) What is the availability at steady state if both units are active?
Solution:
𝜇 = 0.2 per day, 𝜆 = 0.03 per day,
𝜇 𝜇
A1 (T) + + [1 − e−(𝜇+𝜆)T ]
𝜇 + 𝜆 T × (𝜆 + 𝜇)2
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
244 12 Modern Aspects of Probabilistic Distributions for Reliability Evaluation of Engineering Systems

0.2 0.03
∴ A1 (30) + [1 − e−(0.2×0.03)×30 ]
0.2 + 0.03 30 × (0.03 + 0.2)2
(a)
𝜇 𝜇 [ ]
A1 (T) + + 1 − e−(𝜇+𝜆)T
𝜇 + 𝜆 T × (𝜆 + 𝜇)2
0.2 0.03 [ ]
∴ A1 (30) + 2
1 − e−(0.2×0.03)×30
0.2 + 0.03 30 × (0.03 + 0.2)

= 0.8696 + 0.0031 + 0.8727

𝜇
A (∞) = = 0.8696
𝜆+𝜇
(b) For the standby redundant system
𝜆𝜇 + 𝜇 2 (0.03 × 0.2) + (0.2 × 0.2)
As (∞) = =
𝜇2 + 𝜆𝜇 + 𝜆2 (0.2)2 + (0.03 × 0.2) + (0.03)2
0.046
=> = 0.9808
0.0469
(c) For the active redundant system
Ai (∞) = 1 − [1 − A(∞)]2 = 1 − [1 − 0.8696]2 = 0.9830

12.7 Conclusion
The presented analysis here provides a means to relate reliability of power system to reliability of a
component for different equipment. States of binary component is used in the configuration of the
system. An independent component using a reductionist approach to reliability analysis, and it is
component-level studies because the reliability of the system is a function of reliability of a compo-
nent. This fact is very well-defined and accepted. To achieve a high level of reliability in composite
systems, focus on reliability of independent and individual components is very much essential. To
apply the methods few examples are taken to explain the benefits and drawbacks of different meth-
ods. In addition, case studies are taken from the experience of the author’s consultancy, research,
and academics. The basic concepts of theory of probability and the redundancy of the system to
improve reliability are also covered comprehensively. Structure of power component for evalua-
tion of reliability, Hazard model for analysis of component failure, different types of distribution
functions of probability to illustrate reliability, and various solved, unsolved numerical examples
are described in each section and sub-section of this chapter.

References

Arya, L.D., Choube, S.C., and Saket, R.K. (2000). Composite system reliability evaluation based on
static voltage stability limit. Journal of the Institution of Engineers (India): Series-B 80: 133–140.
Arya, L.D., Choube, S.C., and Saket, R.K. (2001). Generation system adequacy evaluation using
probability theory. Journal of the Institution of Engineers (India): Series-B 81: 170–174.
Bansal, R.C., Bhatti, T.S., and Kothari, D.P. (2002). Discussion of “Bibliography on the application of
probability methods in power system reliability evaluation”. IEEE Transaction on Power System 17
(03): 924.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 245

Bharti, O.P., Saket, R.K., and Nagar, S.K. (2019). Reliability assessment and performance analysis of
DFIG based WT for wind energy conversion system. International Journal of Reliability and Safety,
Inderscience Publishers (UK) 13 (04): 235–266.
Billinton, R. and Zhang, W. (1998). Algorithm for failure frequency and duration assessment of
composite power systems. IEE Proceedings: Generation, Transmission and Distribution 145 (2):
117–122.
Kothari, D.P., Nagrath, I.J., and Saket, R.K. (2022). Modern Power System Analysis, 5e, 01–850. New
Delhi (India): McGraw Hill.
Kumar, S., Sachin Kumar, K.S., Kumar, A. et al. (2019). Probabilistic evaluation and design aspects for
reliability enhancement of induction motor. International Journal of Reliability and Safety,
Inderscience Publishers (UK) 13 (04): 267–290.
Saket, R.K. (2013). Design aspects and probabilistic approach for generation reliability evaluation of
MWW based micro-hydro power plant. Renewable and Sustainable Energy Reviews, Elsevier 28:
917–929.
Saket, R.K., Bansal, R.C., and Singh, G. (2006). Generation capacity adequacy evaluation based on peak
load considerations. The South Pacific International Journal on Natural Sciences, University of the
South Pacific, Suva, Fiji Island 24: 38–44.
Saket, R.K., Bansal, R.C., and Singh, G. (2007, 2007). Reliability evaluation of power system considering
voltage stability and continuation power flow. Journal of Electrical Systems, Engineering and
Scientific Research Groups, France 3 (2): 48–60.
Saket, R.K., Bansal, R.C., and Singh, G. (2009). Power systems component modelling and reliability
evaluation of generation capacity. International Journal of Reliability and Safety, Inderscience
Publishers (UK) 03 (04): 427–441.
Varshney, L. and Saket, R.K. (2014). Reliability evaluation of SEIG rotor core magnetization with
minimum capacitive excitation for unregulated renewable energy applications in remote areas. Ain
Shams Engineering Journal, Elsevier 5 (03): 751–757.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
247

13

Reliability Enhancement of Electrical Distribution Systems Considering

Active Distributed Generations
Kalpesh B. Kela 1 , Bhavik N. Suthar 2 , Smriti Singh 3 , Rajesh Arya 4 , and Liladhar Arya 5
1
Department of Electrical Engineering, L.D. College of Engineering, Ahmedabad, Gujarat, India
2
Department of Electrical Engineering, Government Engineering College, Bhuj, Gujarat, India
3
Department of Electrical Engineering, Electrical Department of Electrical Engineering, Indian Institute of Technology (BHU),
Varanasi, Uttar Pradesh, India
4
Department of Information Technology, Acropolis Institute of Technology and Research, Indore, Madhya Pradesh, India
5
Department of Electrical Engineering, Shri G.S. Institute of Technology and Science, Indore, Madhya Pradesh, India

13.1 Introduction
Infrastructure for delivering power and active resources are combined in active distribution sys-
tems, which also include passive infrastructure. The active distribution network is designed to
create a network that is reliable, high-quality, and energy-efficient. DG, or distributed generation,
is now the best option for electrical distribution firms looking to improve the reliability of their
power distribution systems. DGs boost system performance by minimizing losses, improving volt-
age profile, and reliability. In addition, the use of DGs with low capacity at customer locations is
growing for a number of reasons, such as the depletion of fossil fuel resources, increased demand,
environmental concerns, and so forth.
By altering the rate of failures and time of repairs for various system segments, the reliability of
the system of electrical distribution can be raised. By including DGs at the customer’s end, the extra
cost to accomplish this can be reduced. However, buying energy from DGs may come at a premium
price. If the DG switching-over period is shorter, the interruption time of the reliability indicators
can be decreased to a brief interruption.
DG placement and size have a significant impact on reliability along with other metrics in distri-
bution systems. The distribution system’s reliability has risen due to DGs placed in specified areas
(Arya et al. 2012). According to numerous studies, DG implantation decreases loss. However, reli-
ability could be more crucial for some loads than loss and profile of voltage. Many costs relating
to investment, maintenance, operations, and other expenses are associated with introducing DGs
into the distribution network. Therefore, the costs and advantages of deploying DGs for customers
and utilities must be assessed.
In recent years, a number of academics have sought to increase the reliability of DG-based dis-
tribution networks. The best locations for DGs based on reliability indices can be found through
sequential Monte Carlo simulation, according to Yousefian and Monsef (2011). The authors of
Al-Muhaini and Heydt (2013) examined the reliability of the distribution system in relation to con-
ventional and sustainable DG. For the evaluation of DG reliability, a combined Markov system
simulation was used, which takes into account DG sufficiency regarding DG physical breakdown
and beginning and changeover probabilities. The placement of transportable distributed electrical
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
248 13 Reliability Enhancement of Electrical Distribution Systems Considering Active Distributed Generations

power units in distribution systems was recommended by Awad et al. (2014) as a way to increase
system reliability and reduce costs. In order to improve reliability in massive primary networks of
distribution, it is necessary to provide the appropriate location and size for numerous dispersed
generators; for that, Kumar et al. (2014) employed a method of random search known as cat swarm
optimization.
In the existence of storage systems, Abbasi and Hosseini (2016) looked at how the size, shape,
and DG allocation of the distribution network affected the system’s reliability. Reliability is one
of the features of the distribution network expansion strategy that Bagheri et al. (2015) give that
integrates DGs. In order to increase reliability and take into account the whole cost of the power
used by the system, Battu et al.’s (2015) proposal for where to place distributed generators in the
distribution network was successful. In order to evaluate consumer- and energy-oriented reliability
indices for energy distribution networks with DG, Arya (2016) established a methodology that took
into consideration the impact of eliminating random allowed interruption periods at load sites. The
Bootstrapping method and Monte Carlo simulation were used to complete it. With and without
DGs, the ideal value for the main reliability indices in the distribution system was found by the
authors of Ray et al. (2015). To do this, a differential search method was employed. DG placement
in distribution systems that maximize profit and value was found by Kansal et al. (2017).
To maximize system reliability, the inventors of Rahiminejad et al. (2016) adopted DG scheduling.
Here, a cost–benefit evaluation is carried out while taking the various costs related to the compu-
tations into account. In Wang (2017), Wang Yue presented a method for evaluating the analytical
distribution network’s reliability that takes into consideration both the constraints placed on net-
work reconfiguration and stochastic load point disruption durations. The Roy Billinton reliability
test system case studies are used in the research study to highlight the benefits of the recommended
method above the conventional sector technique-based reliability evaluation method. Kela et al.
(2019) suggests a strategy for boosting reliability by maximizing the entire reliability expenses asso-
ciated with electrical distribution systems; both utility and consumer costs are taken into account.
In order to analyze the SAIDI (system average interruption duration index) and SAIFI (system
average interruption frequency index) indices based on network structure and remotely managed
reclosers and switches, the authors of Lekbich et al. (2019) presented a method of analysis based
on graph theory. The study Pattabiraman et al. (2019) offers a feeder reconfiguration technique for
reducing reliability indices. The reconfiguration is handled by Binary Particle Swarm Optimization,
which chooses the best tie switches to reduce both the duration and the number of interruptions.
A reliability assessment method for the integration of large-scale PV systems into a distribution
system is provided in Su et al. (2019). In Okoye et al. (2020), researchers simulated and established
a link between load increase and generator capacity using power production and demand data
from the IEEE reliability test system (IEEE RTS ’96 standard). An economic and reliability assess-
ment of photovoltaic solar power systems for rural electricity supply was made by the authors in
Raghuwansh and Arya (2019). The authors of Kumar et al. (2020) provide a thorough analysis
of the effect of combined renewable sources on the assessment of electrical network reliability.
Mahmood et al. (2020) introduces a novel multistage hybrid method based on the tie-set method
and graph reduction techniques for two-terminal reliability assessment problems. The reliability
of the distribution network was examined by the researchers of Idowu et al. (2021) in relation
to DGs. Here, the network’s reliability with and without DG units is compared. To improve dis-
tributor line reliability and voltage profile, the researchers of Agajie et al. (2021) advised placing
DERs, or distributed energy resources, in the best possible locations. The article (Byk et al. 2020)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
13.2 Electrical Distribution Reliability Indices: Customer and Energy Based 249

demonstrates the efficiency of operating the system in a state without a power supply to it. It has
shown an increase in power supply reliability as well as economic efficiency. In Ahmad and Asar
(2021), authors have increased the reliability of a distribution network by injecting DGs at specific
locations. The optimal locations are found with the help of the artificial neural network (ANN)
technique. The researchers Mahboob et al. (2021) have found the optimal locations and sizes of
DGs to improve the reliability of a distribution system. Here, power losses and the voltage pro-
file of the system have also been assessed. The authors Xu et al. (2023) have suggested a reliability
evaluation method for an active distribution system. Here, the uncertainty of the DGs has been con-
sidered, and the model has been developed. In Nasser Hussain et al. (2023), the reliability of the
distribution network has been enhanced by reconfiguration of the system and optimal placement
of capacitors.
According to the previous discussions, the objective of this study is to increase the distribution
system’s reliability by adjusting the rate of failure and times to repair its various components while
taking DGs into account. An objective function created here was optimized to achieve this. How-
ever, rather than starting from any presumed positions, this optimization was carried out after find-
ing the right places of DGs according to reliability enhancement features. Naturally, an evaluation
of costs and benefits was done to support the choice. Some background knowledge on reliability
indices is provided in Section 13.2. The mathematical approach to the issue is described in Section
13.3. A brief summary of the flower pollination (FP) optimization approach is given in Section 13.4.
The solution to the problem is discussed in Section 13.5. The outcomes and their interpretation are
covered in Section 13.6. Section 13.7 comes to a conclusion.

13.2 Electrical Distribution Reliability Indices: Customer

and Energy Based

Three customer-oriented reliability indices, known as SAIFI, SAIDI, and CAIDI (index that repre-
sents the customer’s average interruption duration) are commonly used by utilities (Billinton and
Allan 2013):
∑
𝜆sys,i Ni
SAIFI = ∑ (13.1)
Ni
∑
Usys,i Ni
SAIDI = ∑ (13.2)
Ni
∑
Usys,i Ni
CAIDI = ∑ (13.3)
𝜆sys,i Ni
AENS (index that represents the average energy not supplied), one of the most essential reliability
indices that is energy based, is defined as follows:
∑
Li Usys,i
AENS = ∑ (13.4)
Ni
The following are expressions for calculating 𝜆sys,i and Usys,i for each load point:
∑
𝜆sys,i = 𝜆k (13.5)
∑
Usys,i = 𝜆k r k (13.6)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
250 13 Reliability Enhancement of Electrical Distribution Systems Considering Active Distributed Generations

13.3 Defining the Problem

Choosing the right DG placements and lowering failure rates and maintenance times in various
system components are the two main objectives of this chapter, which is to increase distribution
system reliability. Two objective functions were examined in this approach.

13.3.1 DG Location Determination

Enhanced reliability is the main criterion for choosing the location of DGs; the objective function
examined is as follows:
SAIFI SAIDI CAIDI AENS
J= + + + (13.7)
SAIFIt SAIDIt CAIDIt AENSt
The sum of the normalized quantities of SAIFI, SAIDI, CAIDI, and AENS is the objective func-
tion. The normalization is done in relation to the indices’ respective target values. As a result,
in the procedure, there will be equal weighting for each index.
It is first decided what J will be after connecting DG to load position 1. Here, J is the system’s
total reliability. Gradually, DGs are linked separately at each of the remaining load sites, and the
value of J is determined. The system’s reliability improves when DG is connected to every load
point one by one. The improvement in reliability with respect to J is arranged and graded across
all load locations. Choosing the top two values in the ranking, connecting the DGs at these load
places, and calculating J’s value are the next steps. The same methodology is then applied to the
top three rankings. Until the minimum standards of reliability is attained, this procedure can be
repeated. Section 13.5 goes into great length about it.
After finding the DG locations, the optimized reliability index values are found as below.

13.3.2 Connecting DGs in the System as Standby Units

The distribution firm may own the DGs, or it could encourage major consumers to become own-
ers of the DGs. The customer-owned and operated DGs, as well as those administered by other
agencies, may end up being crucial in boosting system reliability. Incorporating DGs reduces out-
age time as well as the cost associated with lowering failure and repair rates. It is also possible
to reduce the total expense of interruption for the consumer. On the other hand, the cost of the
energy borrowed through DGs could be substantial because of its higher costs (Arya et al. 2012).
As a result, an objective function is written as follows:
∑
Nc
∑
Nc
∑
Nc
F= 𝛼k ∕𝜆2k + 𝛽k ∕rk + ADCOST(EENSO-EENSD) + CIC (13.8)
k=1 k=1 k=1

𝜆k × rk × Li × Cpk = CIC (13.9)

There are four components in the relation (13.8). The initial two terms refer to the costs of chang-
ing the rate of failure and average times for repairs for each of the distribution system components,
which are, correspondingly, maintenance activity adjustment costs. The lower the value associated
with these criteria, the larger the investments or expenditures utilities must make in preventa-
tive care and maintenance programes in order to meet them (Kela and Arya 2014). Both of these
variables are derived from Duane’s (Ebeling 2019) reliability growth model. The third term in the
same relation represents the additional cost (ADCOST) to be borne by the owners of the DGs when
expected energy is not supplied by them while DGs are not connected (EENSO) and when DGs are
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
13.3 Defining the Problem 251

present (EENSD). Here, ADCOST is taken as Rs. 2 per kWH. The fourth component of the rela-
tion (13.8), customer interruption cost (CIC), represents the expenses incurred at the customer’s
end in terms of loss during a power outage. When a utility is in the occupation of providing elec-
tricity to businesses and industries, the high expenses associated with outages in electricity are of
course magnified. Any load site’s overall disruption expense can be estimated by adding up the
expenses of all section outages, as shown in relation (13.9). The overall cost of all the customer
interruptions can therefore be calculated. Here Cpk in relation (13.9) is the cost of interruptions for
a downtime duration rk related to the kth distribution sector in Rs. for each kilowatt. The cost char-
acteristics associated with the consumers at a single customer load site completely determines the
customer cost at that load site. The aggregate of the expenses experienced by all types of customers
as a result of the distribution outage represents the total of the consumer’s costs related to any load
location due to any interruption (Chowdhury and Koval 2011). The calculations are done for all the
distribution sectors Nc . Table 13.3 shows the results for Roy Billinton test system bus-2 (RBTS-2)
(Figure 13.1). The price of the DGs and the expense of repair plus customer interruptions are thus
balanced by the function defined in (13.8). The other two costs are anticipated to decrease when
the price of DGs increases. By minimizing the function specified in (13.8), this is achieved:
∑
EENS = Li Usys,i (13.10)

where EENS is the index that represents expected energy not supplied. Usys,i , without DG is calcu-
lated as per (13.6). The load may be shifted to DG if the power supply to load terminals from the
source is disrupted while DG has been installed in the system as an on-standby unit. In this case, it
is important to take into account the switch’s failure rate and changeover time as it shifts the loads
to DG.
By connecting the transfer switch to the system in series and the DG in parallel, the reliability
simulation can be demonstrated. Assuming DG is used as a standby unit for the load sites, the
estimated calculations (Arya et al. 2012) give the relevant rate of failure and maintenance time.
( )
𝜆eq = 𝜆s 𝜆dg rs + rdg + 𝜆sw (13.11)
𝜆s • 𝜆dg • rs • rdg + 𝜆sw • s
req = ( ) (13.12)
𝜆s • 𝜆dg rs + rdg + 𝜆sw
Ueq = 𝜆eq req (13.13)

where
𝜆eq = an equivalent rate of failure
req = equivalent duration of interruption
𝜆s = overall failure rate from the source up to the load point
rs = average time between interruptions up to the load point, from the source
𝜆dg = DG failure rate
rdg = average time of a DG outage
𝜆sw = rate of switch failure when shifting load to the DG
s = time for switching or service restoration with DG
Ueq = the load points’ equivalent annual outage time when DGs are connected to the system.

The following restrictions are placed on the minimization of the objective function given
in (13.8).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
252 13 Reliability Enhancement of Electrical Distribution Systems Considering Active Distributed Generations

(i) Constraints on the decision variables

𝜆k,min ≤ 𝜆k ≤ 𝜆k,max (13.14)

rk,min ≤ rk ≤ rk,max k = 1, … , Nc (13.15)

(ii) Inequality constraints for the energy and customer-based reliability indices
SAIFI ≤ SAIFIt (13.16)
SAIDI ≤ SAIDIt (13.17)
CAIDI ≤ CAIDIt (13.18)
AENS ≤ AENSt (13.19)
The optimization procedure achieves the lowest possible failure rate and repair time for the
kth sector. The reliability growth model (Ebeling 2019) is used to analyze failure and repair
information as well as related expenses to create these lower limit values.
SAIFIt , SAIDIt , CAIDIt , AENSt , and SARFIt are each of the indices’ target/threshold values.
They are determined by managerial or administrative decisions.
The FP optimization approach is used to solve the formulated problem (Yang 2012). Other opti-
mization methods, such as teaching learning-based optimization (TLBO) (Rao et al. 2012) and
differential evolution (DE) (Price et al. 2006), have confirmed this. The method described in this
study was tested on the RBTS-2. The crew may be informed of the intended values to increase
system reliability using the optimum values obtained after addressing the problem. They are deter-
mined by managerial/administrative decisions.

13.3.3 Analysis of Costs and Benefits

Cost–benefit analysis can be used to determine whether both the fixed and variable amounts put
toward DGs integrated into the distribution system will generate a benefit in the long run. It sup-
ports positioning a particular number of DGs in the locations mentioned in the present research.
To evaluate the full costs and advantages of connecting a specified number of DGs at particular
locations across the course of an economic cycle, the cumulative present value (CPV) technique
(Raoofat and Malekpour 2011) is applied. The CPV approach, commonly referred to as the time
value of money strategy, is based on the notion that someone who invests would prefer to get a
specific amount of revenue now instead of the same quantity in the future. The costs and benefits
of the plan are transferred in this manner to the very first year of its functioning. The concept is
represented by the equations below:
Benefit = CICDGO − CICDG − ADCOSTDG − CostMDG − DGI − DGM (13.20)
where
CICDGO = CICDGO × CPV1 (13.21)
CICDGO = Cost of customer interruption when DGs are not attached
CICDG = CIC × CPV1 (13.22)
CICDG = Cost of customer interruption when DGs are attached
ADCOSTDG = ADCOST × CPV1 (13.23)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
13.4 The Flower Pollination Algorithm Overview 253

ADCOSTDG = Additional cost for DG operation in rupees per kWh

CostMDG = CostMDG × CPV2 (13.24)
CostMDG = Cost of system maintenance needed to achieve the specified reliability
∑
n
DGI = DGIi × CPV2 (13.25)
i=1

DGI = Cost of installation for all connected DGs

∑
n
DGM = DGMi × CPV2 (13.26)
i=1

DGM = The price of maintaining every DG linked to the system (13.27)

(1−PVEL )
CPV1 = 1−PV1 (13.28)
( 1)
( )
1 + Iinf × (1 + LG)
PV1 = ( )
1 + Iint
(1−PVEL )
CPV2 = 1−PV2 (13.29)
( 2)

(1+Iinf )
PV2 = (13.30)
(1+Iint )
The economic lifespan of equipment is denoted by EL. LG stands for load growth rate. Iinf and
Iint are the inflation and interest rates, respectively.

13.4 The Flower Pollination Algorithm Overview

Invented in 2012 by Xin-She Yang (2012), the FP algorithm was modeled after the flow pollination
technique used by flowering. The process is governed by the following basic principles: (i) global
pollination processes include biological pollination and cross-pollination; (ii) Lévy flight pattern
is used by pollen carriers; (iii) pollinators, like insects, are capable of producing flower reliabil-
ity, which is a measure of how closely related two flowers are, and serves as the probability of
reproduction; (iv) pollination control is determined by a switching probability of [0, 1].
Because of closeness and other conditions, such as wind, local pollination accounts for a consid-
erable fraction p of total pollination activity. The notations used to describe the FP are as follows:
M, flower/pollen gamet population; D, various variables, kmax , maximum allowed number of gen-
erations; and p, switch probability of all pollination activities belongs to [0, 1].
Step A: Initialization: This generates a size “M” beginning population
[ ]
S0 = U10 , U20 , … … , UM 0
(13.31)
[ 0 T
]
Ui0 = Ui10 , Ui20 , … , UiD (13.32)
where
M = number pollen gametes.
D = number of decision vector.
Uij0 = from a uniform distribution the jth parameter of the vector Ui results as follows:
( )
Uij0 = Uj,min + Uj,max − Uj,min randj (13.33)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
254 13 Reliability Enhancement of Electrical Distribution Systems Considering Active Distributed Generations

where
Uj,max and Uj,min = higher and lower limits on the variable Uj , respectively.
randj = random number between 0 and 1.
Step B: Vectors updated by both local and global pollinators

⎧ k ( )
(k)
⎪ Ui + ∝ ×L Ubest − Ui , if rand < p
k
Ui(k+1) = ⎨ (k) ( ) (13.34)
(k) (k) (k) (k)
⎪ Ui + ∈ Uj − Uk , otherwise. where, Uj ≠ Uk
⎩

where
(k)
Ubest = current best option identified among all options.
∝ = scaling factor used to manage step sizes smaller than zero.
L = Lévy distribution.
A Lévy flight can effectively reflect an insect’s ability to move over longer distances with
a variety of distance increments (Pavlyukevich 2007); (Yang 2010).
𝜎x (𝛽)
L=v× (13.35)
𝜎y (𝛽)

randx
v= (13.36)
| |1∕𝛽
|randy |
| |
where
( ) 1∕𝛽
⎡ 𝜋𝛽 ⎤
⎢ Γ(1 + 𝛽) × 2 ⎥
𝜎x (𝛽) = ⎢ ( ) (
𝛽−1
)⎥ (13.37)
⎢ Γ 1+𝛽 × 2 2 ⎥
⎣ 2 ⎦

𝜎y (𝛽) = 1 (13.38)

randx and randy are two random variables that are normally distributed and have stan-
dard deviations of 𝜎x (𝛽) and 𝜎y (𝛽), 𝛽 is the range of the distribution factor from 0.3 to 1.99,
and Γ( • ) is the function of the gamma distribution.
Step C: Assessment of the changed vectors’ appropriateness in comparison to the original vectors
{ ( ) ( )
(k+1) Uik+1 , if f Uik+1 < f Ui(k)
Ui = (13.39)
Ui(k) , otherwise
(k)
The fitness of the determined best solution, Ubest , is then found. Until the best solution is
( )
discovered or the predetermined number of iterations kmax is reached, this procedure is
repeated for each target vector index (i).

13.5 Solution Approach

These are the stages to complete the entire problem’s solution.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
13.5 Solution Approach 255

13.5.1 Evaluation of the DG Locations (Steps)

Step 1: Data input 𝜆k , rk , 𝜆dg , rdg , 𝜆sw , s, Cpk , Ni , Li , SAIFIt , SAIDIt , CAIDIt , and AENSt .
Step 2: Failure rate and repair time should be evaluated.
Step 3: Determine ADCOST and CIC along with SAIDI, SAIFI, CAIDI, AENS, and EENS.
Step 4: Compute the value of J as per relation (13.7).
Step 5: Connect DG only at load point 1, then calculate the results for all indexes and the function
J given by relation (13.7).
Step 6: At the subsequent load point, link the DG alone, and determine all the indexes and J as
per (13.7). Similar to this, attach DG at all loads point one at a time and figure out what
each index and J are worth.
Step 7: Considering the quantity of J in (13.7), calculate the improvement in reliability brought
on by DG at each load point.
Step 8: Sort and order the increases in load location reliability (in relation to J as in (13.7)) in a
decreasing sequence.
Step 9: Connect the DGs to the load locations of the first two rankings’ numbers. For the DGs
connected at these load sites, determine all the indices and the quantity of J.
Step 10: Pick the top three figures on the list. Obtain the system’s indices in addition to J, then
attach DGs to those load locations.
Step 11: By doing so, you can calculate every index and J, as well as increasing the ranking number
and linking DGs at the appropriate load points.
Step 12: Repeat until you reach the desired degree of reliability.

13.5.2 Evaluating the Optimized Solution with the Help of Flower Pollination
Method (Steps)
( )
Step 1: Input data 𝜆k,max , rk,max , 𝜆k,min , rk,min , and cost of interruption Cpk and the target values
of SAIFI, SAIDI, CAIDI, and AENS.
Step 2: By applying relation (13.33), make a set of variables of size “M” for every failure rate
“𝜆” and repair time “r,” where every vector of the pertinent population comprises the
failure rate and maintenance time of every component. According to relations (13.14)
and (13.15), these values are generated by picking uniformly among the smallest one and
maximum boundaries.
Step 3: Compute 𝜆sys,i , rsys,i , and Usys,i at every load location.
Step 4: Compute customer and energy indexes as shown in the Eqs. (13.1)–(13.4), respectively
for all the vectors of the whole population. ( )
Step 5: Determine the value of the function F for each population vector i.e. F Ui(k) ,
i = 1, … , M, as mentioned in (13.8) and (13.9).
Step 6: For each vector in the population, evaluate the inequality constraints from the relations
(13.16)–(13.19). Vectors that satisfy these requirements are feasible; otherwise, they are
(k)
not feasible. Determine the optimal solution vector Ubest from the spectrum of feasible
vectors depending on how much the objective function’s value is.
Step 7: Set k = 1 as the generation counter.
Step 8: Select the vector with i = 1 as a target.
Step 9: Use relation (13.34), and find the latest value of the vector.
Step 10: Examine the fitness value of the modified vectors with the value of the initial vectors
using relation (13.39), and keep the better ones.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
256 13 Reliability Enhancement of Electrical Distribution Systems Considering Active Distributed Generations

Step 11: Repeat processes 3 through 6 for the revised vector.

Step 12: The target vector is increased to i = i + 1. Repeat from Step 9 if i ≤ M; otherwise, raise
the generation count to k = k + 1.
Step 13: Go to step 8 if the target optimal level is not determined or k ≤ kmax .

13.5.3 Analyzing the Costs and Benefits

Use the relationships (13.20)–(13.30) to do a cost–benefit analysis for various planning horizons,
such as 5, 10, and 15 years, and then make the necessary judgments.
33/11 kV

F1 F2 F3 F4
LP1 2 1 12 LP10 16 26 27 LP16
LP8 17
LP2 13 LP17
3 28
D.G. 14
LP9 LP11 18 29 30 LP18
LP3 5 4 19
15 D.G.
D.G. LP4 LP12 LP19
6 D.G. 20 31
32
LP5 D.G.
8 7 LP13 22 21 33 LP20
LP6
9 LP14 23
34 35 LP21
10 24
LP7 LP15 25 LP22
11 36

Figure 13.1 RBTS-2.

Table 13.1 Average repair times and failure rates for various RBTS-2 components.

Repair/ Rate of
Transformer/feeder sections/ replacement failure per
breakers/busbars Section number time (h) year (𝝀)

Transformers {1–22 (except 8,9)} 10 0.01500

Feeder elements/ {2,6,10,14,17,21,25,28,30,34} 5 0.06500
sections (0.6 km)
Feeder sections {1,4,7,9,12,16,19,22,24,27,29,32,35} 5 0.06500
(0.75 km)
Feeder elements/ {3,5,8,11,13,15,18,20,23,26,31,33,36} 5 0.06500
sections (0.80 km)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
13.5 Solution Approach 257

Table 13.2 RBTS-2 system information.

Distribution/
segment 𝝀0k (yr) 𝝀k,min (yr) rk0 (h) rk,min (h) 𝜶k (Rs.) 𝜷k (Rs. × 102 )

#1 0.048750 0.036650 5.000 2.252252 2.564 22.649

#2 0.015000 0.011270 10.000 4.504504 3.205 17.589
#3 0.052000 0.039090 10.000 4.504504 1.923 4.291
#4 0.048000 0.036090 5.000 2.252252 1.282 4.291
#5 0.015000 0.011270 10.000 4.504504 2.564 13.258
#6 0.015000 0.011270 10.000 4.504504 3.045 6.647
#7 0.048750 0.036650 5.000 2.252252 3.205 5.394
#8 0.015000 0.011278 10.000 4.504504 2.564 22.649
#9 0.015000 0.011278 10.000 4.504504 3.205 17.589
#10 0.039000 0.029323 5.000 2.252252 1.923 4.291
#11 0.015000 0.011278 10.000 4.504504 1.282 4.291
#12 0.048750 0.036650 5.000 2.252250 2.564 22.649
#13 0.052000 0.039097 5.000 2.252250 3.205 17.589
#14 0.039000 0.029323 5.000 2.252250 2.564 22.649
#15 0.052000 0.039097 5.000 2.252250 3.205 17.589
#16 0.048750 0.036650 5.000 2.252250 2.564 22.649
#17 0.015000 0.011278 10.000 4.504500 3.205 17.589
#18 0.052000 0.039097 5.000 2.252250 1.923 4.291
#19 0.015000 0.011278 10.000 4.504500 1.282 4.291
#20 0.015000 0.011278 10.000 4.504500 2.564 13.258
#21 0.039000 0.029323 5.000 2.252250 3.045 6.647
#22 0.015000 0.011278 10.000 4.504500 2.564 22.649
#23 0.015000 0.011278 10.000 4.504500 3.205 17.589
#24 0.048750 0.036650 5.000 2.252250 1.923 4.291
#25 0.015000 0.011278 10.000 4.504500 1.282 4.291
#26 0.052000 0.039097 5.000 2.252250 2.564 22.649
#27 0.015000 0.011278 10.000 4.504500 3.205 17.589
#28 0.015000 0.011278 10.000 4.504500 1.923 4.291
#29 0.048750 0.036650 5.000 2.252250 1.282 4.291
#30 0.015000 0.011278 10.000 4.504500 2.564 13.258
#31 0.015000 0.011278 10.000 4.504500 3.045 6.647
#32 0.048750 0.036650 5.000 2.252250 3.205 5.394
#33 0.015000 0.011278 10.000 4.504500 2.564 22.649
#34 0.039000 0.029323 5.000 2.252250 3.205 17.589
#35 0.015000 0.011278 10.000 4.504500 1.923 4.291
#36 0.015000 0.011278 10.000 4.504500 1.282 4.291
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
258 13 Reliability Enhancement of Electrical Distribution Systems Considering Active Distributed Generations

13.6 Discussions and Outcomes

The methods described in this research was applied on RBTS-2. The FP algorithm was used to solve
the problem, and the results were compared to those obtained by TLBO and DE. MATLAB R2019b
software, an Intel (R), Core (TM), i5-4590S CPU running at 3.00 GHz and 4.00 GB RAM were used
to run the proposed algorithm.

13.6.1 Roy Billinton Test System (RBTS-2)

The test system used in this study is the RBTS-2 (Allan et al. 1991), as shown in Figure 13.1. The
average repair times and failure rates for different RBTS-2 components are shown in Table 13.1.
Table 13.2 lists the current and lowest reachable rate of failure, current and lowest reachable aver-
age repair times, and cost coefficients for the rate of failure and time to repair the different RBTS-2
segments. Table 13.3 gives various customer information for RBTS-2, like the average load at differ-
ent load points, the interruption costs there, etc. Table 13.4 gives information about energy-based
and customer-based reliability indices, interruption costs for customers, additional costs on DGs,
and the value of the function J when a single DG is connected to various load points one by one.
The connected DGs are treated as backup units in this situation. The average downtime and failure
rate for the DG used in this study are 13.25 hours and 0.5 failures per year, respectively. The DG
changeover switch’s rate of failure and the restoration of service time are 0.1 failures per year and
0.5 hours, respectively. According to the improvement in distribution system reliability, the func-
tion values J reported in Table 13.4 are arranged in Table 13.5 in the order of decrement. The rating
of reliability with several DGs connected at various load sites is displayed in Table 13.6. The places
of DGs are identified in Table 13.6. In this study, the function F in relation (13.8) is optimized using
FP, DE, and TLBO. DGs are linked at load points 2, 3, 11, 12, and 18. Table 13.7 lists the control
settings for the three optimization techniques. Table 13.8 lists the optimized values of primary reli-
ability indices for the various RBTS-2 parts using all the optimization techniques used in this study.
In Table 13.9, the optimized values for the RBTS-2’s cost of maintenance, consumer interruption
cost, extra expenses for buying electricity from the associated generators, and function (F) are pre-
sented. These values were obtained using the three optimization techniques. Table 13.10 shows the
optimized quantities of the energy- and customer-based reliability indices created with each of the
three methodologies. An evaluation of costs and benefits has been done based on Section 13.3.3.
The cost of installing and maintaining each connected DG in the distribution system is estimated

Table 13.3 Customer information for the Roy Billinton test system-2.

Load average Type of Size of Interruption

Load point at every load customers/ customers cost (Cpk )
number point in MW premise in numbers (Rs./kW)

1–3,10,11 0.5350 Residential purpose 210.0 22.290

12,17–19 0.4500 Residential purpose 200.0 6.0450
8 1.0000 Small scale user 1.0 251.420
9 1.1500 Small scale user 1.0 251.420
4,5,13,14,20,21 0.5660 Government institutes 1.0 23.980
6,7,15,16,22 0.4540 Commercial purpose 10.0 192.980
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 13.4 The parameter values for RBTS-2’s various load points with and without DG connections.

Extra
expenditure Function
Load incurred on CIC (J) as
points SAIFI SAIDI CAIDI AENS EENS DG (rupees) (rupees) per (13.7)

Without DG 0.09860 0.58820 5.96660 4.66410 8899.00 — 828,680.00 5.663

1 0.1025760 0.5503740 5.3655230 4.5677580 8715.28200 367.49390 824,578.400 5.4326470
2 0.0985090 0.5096820 5.1739540 4.4640890 8517.48200 763.09450 820,168.400 5.2011770
3 0.0972990 0.5239790 5.3852710 4.5005130 8586.97900 624.09970 821,717.900 5.2878150
4 0.0985720 0.5878700 5.9638380 4.4910360 8568.89800 660.26240 820,756.800 5.6038370
5 0.0985470 0.5877420 5.9640870 4.4187840 8431.03900 935.97880 817,450.300 5.5791990
6 0.0982620 0.5838420 5.9416680 4.4673190 8523.64500 750.76740 756,231.600 5.5767130
7 0.0980580 0.5828210 5.9436260 4.4209550 8435.18200 927.69350 739,159.600 5.5568200
8 0.0985780 0.5879380 5.9641800 4.4264460 8445.66000 906.73800 714,689.000 5.5826290
9 0.0985580 0.5878360 5.9643800 4.2733290 8153.51200 1491.03400 641,237.100 5.5311470
With DG 10 0.1025750 0.5503710 5.3655320 4.5677490 8715.26500 367.52650 827,564.200 5.4326280
(at load points) 11 0.0968570 0.5217740 5.3870350 4.4948950 8576.26000 645.53770 826,723.900 5.2757400
12 0.0969390 0.5249360 5.4150990 4.5217720 8627.54100 542.97560 827,033.900 5.3005720
13 0.0985500 0.5877570 5.9640580 4.4272860 8447.26200 903.53280 817,839.400 5.5820980
14 0.0985500 0.5877570 5.9640580 4.4272860 8447.26200 903.53280 817,839.400 5.5820980
15 0.0980370 0.5827150 5.9438290 4.4161790 8426.07000 945.91660 863,237.800 5.5547720
16 0.0987520 0.5862900 5.9370070 4.5784670 8735.71500 326.62730 795,406.900 5.6244090
17 0.1020450 0.5504690 5.3943950 4.5792210 8737.15400 323.74950 825,945.500 5.4377430
18 0.0969390 0.5249360 5.4150990 4.5217720 8627.54100 542.97560 825,282.900 5.3005720
19 0.0969390 0.5249360 5.4150990 4.5217720 8627.54100 542.97560 826,898.800 5.3005720
20 0.0985450 0.5877320 5.9641080 4.4128350 8419.68900 958.68040 815,427.100 5.5771700
21 0.0985240 0.5876290 5.9643070 4.3550280 8309.39300 1179.27100 812,781.600 5.5574580
22 0.0980370 0.5827150 5.9438290 4.4161790 8426.07000 945.91660 735,650.200 5.5547720
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
260 13 Reliability Enhancement of Electrical Distribution Systems Considering Active Distributed Generations

Table 13.5 The load points are shown in descending

order of reliability enhancement, from largest to smallest.

Load point (LP) ranking based

on reliability improvement Reliability increment
when linked to DG in (%) (in terms of J )

2 8.8792020
11 7.3403990
3 7.0952830
12 6.8375170
18 6.8375170
19 6.8375170
10 4.2405200
1 4.2401700
17 4.1424800
9 2.3838250
15 1.9483860
22 1.9483860
7 1.9108000
21 1.8990990
6 1.5472740
20 1.5389500
5 1.5020310
13 1.4493110
14 1.4493110
8 1.4396570
4 1.0557670
16 0.6861260

Table 13.6 Based on load point rating, reliability found for the system with multiple connected generators.

Extra
DGs expenditure Value of
linked incurred function
at load on DG CIC ( J) as
points $LP) SAIFI SAIDI CAIDI AENS EENS (rupees) (rupees) per (7)

2,11 0.09678 0.44328 4.579908 4.294923 8194.713 1408.63 818,217.3 4.810259

2,3,11 0.09550 0.379084 3.969128 4.131375 7882.663 2032.73 811,260.1 4.427674
2,3,11,12 0.093869 0.315844 3.364737 3.989085 7611.175 2575.70 809,619 4.051987
2,3,11,12,18 0.09223 0.252604 2.738863 3.846796 7339.687 3118.68 807,977.9 3.670929
2,3,11,12,18,19 0.09059 0.189365 2.090338 3.704507 7068.199 3661.65 806,336.7 3.284208
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
13.7 Conclusion 261

Table 13.7 FP, TLBO, and DE control parameters

for Roy Billinton test system, bus-2.

Parameter
Parameter description values

Size of population for TLBO, 40.0

DE, and FP
Maximum predefined 1000.0
generation (kmax ) for TLBO, DE,
and FP
Updated size of step (𝛼) for FP 0.010
Factor of distribution for (𝛽) FP 1.50
Switching probability for FP 0.80
Size of step for DE 0.80
Cross-over rate for DE 0.70

at 25 rupees per kW and 5.94 rupees per kW, respectively. The assumed rates for load growth, inter-
est, and inflation are 5%, 10%, and 5%, respectively. By using all three optimization techniques,
calculations for economical life span estimates of 5, 10, and 15 years have been made.

13.6.2 Comparison Research

To validate the findings acquired by FP, a comparison of the FP results with those obtained by the
other two optimization techniques, TLBO and DE, was conducted. All of the strategies yield similar
values for the objective function (F), but FP yields better results. The reliability indices based on
customers and energy that were obtained using the three approaches are also compared. The results
for FP have been proven to be superior to those of the other two approaches. The benefit gained
by FP is greatest for the economic lifecycles taken into account in this work in the cost–benefit
analysis as well (Table 13.11).

13.7 Conclusion

In this study, the connection of DGs boosted the reliability of a distribution system (RBTS-2). The
locations of DGs have been determined using a technique that prioritizes reliability enhancement.
With these locations now identified, three optimization techniques—FP, TLBO, and DE—have
been used to optimize the objective function stated in this research. Of the three approaches, the
FP has been found to be the best. The installation of DGs has been supported by a long-term
analysis of costs and benefits, which also shows that the benefits of the values generated by
FP are greater.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
262 13 Reliability Enhancement of Electrical Distribution Systems Considering Active Distributed Generations

Table 13.8 RBTS-2’s optimized rates of failure and time for repair as determined by the three optimization
methodologies.

Rate of failure per year Time for repair in hours

Distributor
element By DE By TLBO By FP By DE By TLBO By FP

1 0.039839 0.036650 0.036653 2.254974 2.252253 2.252322

2 0.014873 0.015000 0.014995 4.568424 4.504504 4.504545
3 0.047409 0.052000 0.039095 4.647431 4.504504 4.508767
4 0.036107 0.036090 0.036091 2.297369 2.252256 2.252605
5 0.014789 0.015000 0.014999 7.138270 10.000000 9.997818
6 0.014874 0.015000 0.015000 8.800289 4.504504 4.513614
7 0.037194 0.036650 0.036654 2.270862 2.252252 2.252840
8 0.014694 0.015000 0.014999 4.890354 4.504542 4.516031
9 0.014840 0.015000 0.014999 4.515139 4.504507 4.505314
10 0.029675 0.029323 0.038991 2.252385 2.252252 2.252424
11 0.014999 0.015000 0.014982 4.551617 4.504504 4.505028
12 0.036650 0.036650 0.036650 2.260009 2.252250 2.252534
13 0.039132 0.039097 0.039116 2.253733 2.252250 2.257926
14 0.029329 0.029323 0.038999 2.433488 2.252250 2.255338
15 0.039671 0.039097 0.039136 2.326117 2.252250 2.252986
16 0.037130 0.036650 0.036651 2.507309 2.252250 2.268080
17 0.014994 0.015000 0.014999 5.097232 4.504500 9.987020
18 0.039415 0.039097 0.039100 2.359125 2.252250 2.262472
19 0.014222 0.015000 0.015000 5.019546 4.504500 4.510011
20 0.014976 0.015000 0.014984 9.993431 4.504500 4.504950
21 0.031561 0.029323 0.029323 2.256693 2.252250 2.252478
22 0.014541 0.015000 0.015000 5.578698 4.504500 4.508283
23 0.014421 0.015000 0.014999 4.632058 4.504500 4.504863
24 0.037447 0.036650 0.048749 2.557199 2.252276 2.256059
25 0.014903 0.015000 0.015000 4.542109 4.504500 4.506840
26 0.040029 0.039097 0.039156 2.271917 2.252250 2.252871
27 0.014648 0.015000 0.014996 4.681604 4.504500 4.505772
28 0.014113 0.015000 0.014977 8.478175 4.504500 9.982366
29 0.036795 0.036650 0.048745 2.253501 2.252250 2.253396
30 0.014993 0.015000 0.014980 9.423377 4.504500 4.505367
31 0.014868 0.015000 0.015000 4.571811 4.504500 4.538311
32 0.041654 0.036650 0.036651 2.301898 2.252250 2.252895
33 0.014998 0.015000 0.014999 9.948056 4.504500 9.979963
34 0.029505 0.030576 0.029326 2.945996 2.252250 2.252948
35 0.014890 0.015000 0.014999 4.663455 4.504500 9.938406
36 0.014664 0.015000 0.014999 5.889663 4.504500 4.505902
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 263

Table 13.9 Current and optimized objective function (J) quantities for RBTS-2 calculated by DE, TLBO,
and FP.

Values after
Current optimization in rupees
values in
rupees DE TLBO FP
∑Nc ∑Nc
Cost of maintenance ( k=1 𝛼k ∕𝜆2k + k=1 𝛽k ∕rk ) 232,680 258,719.8 251,732.1 254,692.6
∑Nc
Customer interruption cost ( k=1 CIC) 828,680 306,879.7 300,608.9 284,052.8
ADCOST (extra expenditures when
generators are linked) — 11,355.11 4230.645 12,398.98
Objective function (J) 1,064,500 576,954.6 556,571.6 551,144.3

Table 13.10 RBTS-2 reliability indices, both current and optimized.

Optimized values
Threshold/
Indices Current values FP TLBO DE target values

SAIFI 0.098600 0.082604 0.085918 0.084452 0.0860

SAIDI 0.588200 0.174741 0.132302 0.111135 0.40
CAIDI 5.966600 2.163416 1.586033 1.301081 4.00
AENS 4.664100 1.688404 1.560067 1.414852 2.20

Table 13.11 Analysis of costs and benefits.

Benefit in rupees
Method of optimization
Economic lifecycle
management
in years DE TLBO FP

5.0 972,060.00 1,173,100.00 1,128,600.00

10.0 2,280,500.00 2,540,400.00 2,658,400.00
15.0 3,895,700.00 4,137,300.00 4,438,500.00

References

Abbasi, F. and Hosseini, S.M. (2016). Optimal DG allocation and sizing in presence of storage systems
considering network configuration effects in distribution systems. IET Generation, Transmission &
Distribution 10 (3): 617–624.
Agajie, T.F., Khan, B., Guerrero, J.M., and Mahela, O.p (2021). Reliability enhancement and voltage
profile improvement of distribution network using optimal capacity allocation and placement of
distributed energy resources. Computers and Electrical Engineering 93: 107295.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
264 13 Reliability Enhancement of Electrical Distribution Systems Considering Active Distributed Generations

Ahmad, S. and Asar, A.u. (2021). Reliability enhancement of electric distribution network using
optimal placement of distributed generation. Sustainability 13 (20): 11407.
Allan, R.N., Billinton, R., Sjarief, I. et al. (1991). A reliability test system for educational purposes-basic
distribution system data and results. IEEE Transactions on Power systems 6 (2): 813–820.
Al-Muhaini, M. and Heydt, G.T. (2013). Evaluating future power distribution system reliability
including distributed generation. IEEE Transactions on Power Delivery 28 (4): 2264–2272.
Arya, R. (2016). Determination of customer perceived reliability indices for active distribution systems
including omission of tolerable interruption durations employing bootstrapping. IET Generation,
Transmission & Distribution 10 (15): 3795–3802.
Arya, R., Choube, S.C., and Arya, L.D. (2012). Reliability evaluation and enhancement of distribution
systems in the presence of distributed generation based on standby mode. International Journal of
Electrical Power and Energy Systems 43 (1): 607–616.
Awad, A.S.A., El-Fouly, T.H.M., and Salama, M.M.A. (2014). Optimal distributed generation allocation
and load shedding for improving distribution system reliability. Electric Power Components and
Systems 42 (6): 576–584.
Bagheri, A., Monsef, H., and Lesani, H. (2015). Integrated distribution network expansion planning
incorporating distributed generation considering uncertainties, reliability, and operational
conditions. International Journal of Electrical Power and Energy Systems 73: 56–70.
Battu, N.R., Abhyankar, A.R., and Senroy, N. (2015). Dg planning with amalgamation of economic and
reliability considerations. International Journal of Electrical Power and Energy Systems 73: 273–282.
Billinton, R. and Allan, R.N. (2013). Reliability Evaluation of Power Systems. Springer Science and
Business Media.
Byk, F., Kakosha, Y., and Myshkina, L. (2020). Distributed power generation and power supply
reliability improvement. In: E3S Web of Conferences, vol. 216, 01013. EDP Sciences.
Chowdhury, A. and Koval, D. (2011). Power Distribution System Reliability: Practical Methods and
Applications. John Wiley and Sons.
Ebeling, C.E. (2019). An Introduction to Reliability and Maintainability Engineering. Waveland Press.
Idowu, K., Uhunmwangho, R., Okafor, E., and Big-Alabo, A. (2021). Reliability improvement study of a
distribution network with distributed generation. Applications of Modelling and Simulation 5: 53–65.
Kansal, S., Tyagi, B., and Kumar, V. (2017). Cost–benefit analysis for optimal distributed generation
placement in distribution systems. International Journal of Ambient Energy 38 (1): 45–54.
Kela, K.B. and Arya, L.D. (2014). Reliability optimization of radial distribution systems employing
differential evolution and bare bones particle swarm optimization. Journal of The Institution of
Engineers (India): Series B 95 (3): 231–239.
Kela, K.B., Suthar, B.N., and Arya, L.D. (2019). Reliability optimization of electrical distribution
systems considering expenditures on maintenance and customer interruptions. Indonesian Journal
of Electrical Engineering and Computer Science 14 (3): 1057–1064.
Kumar, D., Samantaray, S.R., Kamwa, I., and Sahoo, N.C. (2014). Reliability-constrained based optimal
placement and sizing of multiple distributed generators in power distribution network using cat
swarm optimization. Electric Power Components and Systems 42 (2): 149–164.
Kumar, S., Saket, R.K., Dheer, D.K. et al. (2020). Reliability enhancement of electrical power system
including impacts of renewable energy sources: a comprehensive review. IET Generation,
Transmission & Distribution 14 (10): 1799–1815.
Lekbich, A., Belfqih, A., Ouaderhman, T. et al. (2019). An analytical multicriteria model based on
graph theory for reliability enhancement in distribution electrical networks. International Journal of
Electrical and Computer Engineering 9 (6): 4625.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 265

Mahboob, A., Mehdi, H., Noor, A., and Khan, M.A. (2021). Reliability based optimal distributed
generation placement for a radial distribution network. In: 2021 International Conference on
Engineering and Emerging Technologies (ICEET), 1–6. IEEE.
Mahmood, M.K., Uçan, O.N., Zaidan, Z., and Karim, S.M. (2020). Hybrid algorithm for two-terminal
reliability evaluation in communication networks.
Nasser Hussain, A., Abed, W.R., and Yaqoob, M.M. (2023). Dual strategy of reconfiguration with
capacitor placement for improvement reliability and power quality in distribution system. Journal of
Engineering 2023: 1943535.
Okoye, M.O., Yang, J., Lei, Z. et al. (2020). Predictive reliability assessment of generation system.
Energies 13 (17): 4350.
Pattabiraman, S., Sampath, K., Kannan, M. et al. (2019). Reliability improvement in radial distribution
system through reconfiguration. In: 2019 IEEE Innovative Smart Grid Technologies-Asia (ISGT Asia),
1963–1967. IEEE.
Pavlyukevich, I. (2007). Lévy flights, non-local search and simulated annealing. Journal of
Computational Physics 226 (2): 1830–1844.
Price, K., Storn, R.M., and Lampinen, J.A. (2006). Differential Evolution: A Practical Approach to Global
Optimization. Springer Science and Business Media.
Raghuwansh, S.S. and Arya, R. (2019). Economic and reliability evaluation of hybrid photovoltaic
energy systems for rural electrification. International Journal of Renewable Energy Research (IJRER)
9 (1): 515–524.
Rahiminejad, A., Hosseinian, S.H., Vahidi, B., and Shahrooyan, S. (2016). Simultaneous distributed
generation placement, capacitor placement, and reconfiguration using a modified
teaching-learning-based optimization algorithm. Electric Power Components and Systems 44 (14):
1631–1644.
Rao, R.V., Savsani, V.J., and Vakharia, D.P. (2012). Teaching–learning-based optimization: an
optimization method for continuous non-linear large scale problems. Information Sciences 183 (1):
1–15.
Raoofat, M. and Malekpour, A.R. (2011). Optimal allocation of distributed generations and remote
controllable switches to improve the network performance considering operation strategy of
distributed generations. Electric Power Components and Systems 39 (16): 1809–1827.
Ray, S., Bhattacharya, A., and Bhattacharjee, S. (2015). Differential search algorithm for reliability
enhancement of radial distribution system. Electric Power Components and Systems 44: 1–14. https://
doi.org/10.1080/15325008.2015.1091864.
Su, S., Hu, Y., He, L. et al. (2019). An assessment procedure of distribution network reliability
considering photovoltaic power integration. IEEE Access 7: 60171–60185.
Wang, Y. (2017). Analytical complex distribution system reliability evaluation considering stochastic
interruption durations and network reconfigurations. Electric Power Components and Systems 45
(19): 2151–2163.
Xu, W., Zeng, S., Du, X. et al. (2023). Reliability of active distribution network considering uncertainty
of distribution generation and load. Electronics 12 (6): 1363.
Yang, X.-S. (2010). Nature-Inspired Metaheuristic Algorithms. Luniver Press.
Yang, X.-S. (2012). Flower pollination algorithm for global optimization. In: International Conference
on Unconventional Computing and Natural Computation, 240–249. Springer.
Yousefian, R. and Monsef, H. (2011). DG-allocation based on reliability indices by means of Monte
Carlo simulation and AHP. In: 2011 10th International Conference on Environment and Electrical
Engineering, 1–4. IEEE.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
267

14

Reliability Enhancement Strategy for Electrical Distribution Systems

Considering Reward and Penalty
Kalpesh B. Kela 1 , Bhavik N. Suthar 2 , Liladhar Arya 3 , and Rajesh Arya 4
1 Department of Electrical Engineering, L.D. College of Engineering, Ahmedabad, Gujarat, India
2
Department of Electrical Engineering, Government Engineering College, Bhuj, Gujarat, India
3
Department of Electrical Engineering, Shri G.S. Institute of Technology & Science, Indore, Madhya Pradesh, India
4
Department of Information Technology, Acropolis Institute of Technology and Research, Indore, Madhya Pradesh, India

14.1 Introduction
In order for distribution companies to function as efficiently as possible, several countries world-
wide are gradually transitioning away from traditional price-based (on the rate of return or the
cost of the service) management and toward performance-oriented management/regulations. In
performance-based rules, incentives are set by the companies in order to encourage high efficiency
(in terms of profit). The standard of services offered to customers may suffer as a result (Fumagalli
et al. 2007). Many performance-oriented rules are interwoven with quality regulations in order to
satisfy these requirements, adopting either direct or indirect quality assurance measures (CEER
2011). Under indirect quality control, customers are informed about the performance quality of
a distribution company, whereas under direct quality control, the output is evaluated in accor-
dance with the financial rewards provided by regulatory bodies to ensure proper service quality
(Alvehag and Awodele 2013). Three criteria are used to characterize the services quality provide
by the distribution company: voltage quality, commercial quality, and reliability of operations or
continuation of supply (Fumagalli et al. 2007). The regulations governing service reliability are the
main topic of this chapter. Various measures put in place by regulators have regulated service relia-
bility. For regulatory purposes in distribution networks, the rewards and penalties scheme (RPS), a
direct control of quality mechanism, is a successful and popular solution. Service reliability indices
like System Average Interruption Frequency Index (SAIFI), System Average Interruption Duration
Index (SAIDI), and Average Energy Not Supplied (AENS) are regulated by the regulator using this
method. Based on the attainment of these indices within their intended levels, a reward or penalty
is agreed upon for the distribution company.
For distribution companies to keep or raise their level of quality, RPS creates financial incentives.
In order to reduce the overall reliability cost to society using RPS, the regulator works to maintain
a level of reliability that is socioeconomically optimal (Alvehag and Awodele 2013). In Billinton
and Pan (2004), the idea of employing indices of reliability in RPS was covered. Fotuhi et al. (2006)
presents various reward-penalty models with their applications and features with real system
reliability information gathered by Tehran Regional Electrical Company. RPS coupled with clus-
tering analysis is a method for increasing the operational reliability of distribution networks that

Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
268 14 Reliability Enhancement Strategy for Electrical Distribution Systems Considering Reward and Penalty

was first introduced in Mohammad Nezhad-Shourkaei et al. (2011). Here, the utilities have been
categorized, and the effectiveness of the utility in one group has been evaluated in comparison
to that of the other cluster members. The key conditions for developing and putting into practice
a successful RPS have been examined in Mohammad Nezhad-Shourkaei and Fotuhi-Firuzabad
(2011). Here, various RPS varieties have been in-depth detailed. By combining to calculate the RPS
parameters for each electric business, a method is presented in Simab et al. (2012) that combines
data envelopments analysis (DEA) with fuzzy C-mean grouping (FCM) based on the average
system interruption time index. The essential elements of RPSs across the electric business were
determined using an algorithm reported in Simab and Haghifam (2012) employing DEA that
is based on reliability indices and consumers’ perceived value of interruptions. Based on the
idea of yardstick theory, a strategy for developing processes for reward or penalty methods has
been put forward in Jooshaki et al. (2014). A new incentive program for supply continuity that
was adopted in Sweden from 2016 to 2019 has been addressed in Grahn et al. (2016), where a
reward or penalty is based on the DSO’s performance and the financial benefit provided to the
DSO. Jooshaki et al. (2017) focuses on a regulatory framework based on RPS for the construction
of a distribution system, taking into account the incorporation of distributed generators. The
clustering using fuzzy c-means and the analytical hierarchy process (AHP) is used in Ghasemi and
Dashti (2017) to offer a model for determining the variables of the rewards and penalty program in
performance-oriented management for all electricity distribution firms. Sumesh et al. (2018) has
put out a performance-oriented penalty function approach that significantly affects the energy pro-
ducers’ state of mind to act dynamically to commit profit, resulting in predictable energy-sharing
actions. A novel approach has been put forth in Wang et al. (2018) to assess load-based reliability
indices in electrical distribution networks utilizing data from sophisticated metering equipment.
With the aid of the recently adopted rewards and penalties regulatory framework for utility
regulators, they are employed to offer reliability oversight. Jooshaki et al. (2018) discusses a novel
fuzzy-based design procedure to apply reward-penalty systems in the distribution sector more
successfully. A model for making decisions to ensure that the RPS for energy distribution compa-
nies (EDCs) is properly implemented is a fresh fuzzy-based design approach for the distribution
sector’s implementation of reward-penalty systems that work better. For EDCs, a decision-making
model for the effective implementation of the RPS is described in Ghasemi and Dashti (2018).
During each regulatory term, this model might modify the RPS’s parameters, improving the
program’s efficiency. A Cubic Reward-Penalty structure is suggested in Sumesh et al. (2019) to
increase the service reliability of electricity distribution firms. The employment of a reward and
penalty structure that fluctuates in a cubic pattern based on performance based on reliability in the
scheme’s innovative characteristic. A method that utilizes relative performance to improve reli-
ability regulation is put out in Khonakdar-Tarsi et al. (2021). Here, the feeders of the distribution
system are grouped according to factors that the utilities are not in control of the reward-penalty
system, which rewards feeders who perform better than the standard and penalizes feeders whose
performance falls below the benchmark, is the foundation upon which the quality regulation
is designed. In Janjic et al. (2021), a brand-new approach to RPS design is put out that depends
on the spatial distribution of reliability performance. With this approach, the analysis provides
a set of reward and penalty criteria that are more visible. An incentive-based demand response
program is suggested in Ghorashi et al. (2020). Customers who minimize or reschedule their
needs during the desired hours are rewarded under the initiative. If the clients’ demands go above
certain limits, they could also face penalties. Lu et al. (2021) explores an inventive reward-penalty
mechanism created expressly to encourage net zero energy structures and communities. In order
to apply the suggested reward-penalty system, two distinct strategies, the building level-based
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
14.2 Reward and Penalty System (RPS) 269

approach and the strategy based on community level, have been created. A paradigm to enhance
the asset management techniques for public lighting is provided in Ghotboddini and Toossian
Shandiz (2022). The reward and penalty mechanisms in the regard use the guaranteed duration
and high correction time. In Amirrezai et al. (2023), the impact of distributing generations on the
distribution system power factor has been studied. By improving the distribution system power
factor, the penalty costs to the distribution companies have been eliminated.
By integrating rewards and penalties applied to them by the regulator, the topic of this chapter
is improving distribution systems’ reliability. In this case, the reliability indices’ desired values can
be attained using the optimal values of reward and penalty. A distribution company’s financial risk
and unneeded tariff changes for customers can be reduced by determining the best values for both
rewards and penalties.
Briefly describing the RPS theory is Section 14.2. Section 14.3 provides a description of the issue
this study aims to address. Rao algorithms are briefly described in Section 14.4. The methods used
by Rao algorithms to solve the problem are discussed in Section 14.5. The results for the two sample
systems are discussed in Section 14.6. In Section 14.7, conclusions are presented.

14.2 Reward and Penalty System (RPS)

To assess the system’s reliability, the regulator uses system quality indicators in RPS. The utilities
must comply with set targets for these initiatives. The utility will receive a reward or a penalty
depending on whether or not it meets these objectives. The utility benefits increase with reliability
level. Regulators thus attempt to imitate market-like conditions. The average SAIFI and SAIDI,
both from the perspective of reliability, are frequently used quality indicators. Both indexes are
customer-based (Billinton and Allan 1996). It is also possible to integrate these indices with
energy-based reliability indicators like Expected Energy Not Supplied (EENS) (Billinton and
Allan 1996).
Through RPS, the regulator aims to attain a reliability level that is socioeconomically optimal
(Fumagalli et al. 2007). Network costs and customer interruption costs are related to socioeco-
nomic reliability. High customer interruption costs result from poor reliability. Higher system
reliability lowers interruption expenses for the users, but it also necessitates higher capital and
maintenance costs, which are eventually covered by the users through tariffs. The level of reliabil-
ity that minimizes the overall cost to society of reliability for networks and customers is known as
the socioeconomic optimum reliability level, and it lies somewhere in the middle (Billinton and
Li 1994).The cost-versus-reliability graph in Figure 14.1 represents the level of reliability that is
socioeconomically optimal. When the rate of rise in network expenses is equal to the rate of fall in
costs associated with customer disruption, the optimal level of reliability Ropt is attained:
𝜕Cnetwork | 𝜕CIC ||
| =− (14.1)
|
𝜕R |R=Ropt 𝜕R ||R=Ropt
The socioeconomic optimal may or may not coincide with the utility’s optimal because the util-
ity aims to set the optimal at its minimum cost. By integrating the costs of customer interruptions
into their own cost functions, optimal RPS provides incentives for achieving socioeconomic relia-
bility (Sappington 2005). In reality, the regulator is unaware of the socioeconomic reliability level;
instead, a threshold or target value is established for the level of quality index that assesses reli-
ability. The utility sets the cost of customer interruption owing to a divergence in reliability from
this target value in terms of either a reward or a penalty. When a utility achieves or falls short of its
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
270 14 Reliability Enhancement Strategy for Electrical Distribution Systems Considering Reward and Penalty

Cost curves Socioeconomically optimal reliability

Total reliability cost for society level, Ropt
Annual cost

Society
CTotal = Cnetwork + CIC

Customer
Interruption 𝛛CIC
costs, CIC 𝛛R

Penalty CRP > 0 𝛛Cnetwork

4
𝛛R

Network costs,
Cnetwork Reward CRP < 0

System reliability, R

Figure 14.1 The cost versus reliability graph shows the amount of reliability that is socially and
economically optimal.

target reliability level, it bears financial risk, which results in either a profit or a loss for the utility
(cost of reward or penalty,CRP ). Here, the utility is being rewarded when CRP < 0 and being penal-
ized when CRP > 0. The ultimate objective is to achieve the socioeconomic reliability level Ropt at
the minimal value of the utility’s total reliability cost curve while taking the impact of RPS into
account. Figure 14.2 (Alvehag and Awodele 2013; Ajodhia 2006) depicts the various RPS schemes.
This article is primarily concerned with a continuous scheme in which incentives and penalties
rise as deviations from the predetermined target level do.

Target level Rtarget

1. Minimum standards
Penalty CRP > 0

System
reliability, R
Reward CRP < 0

4. Dead Band

3. Capped
Dead band
width
2. Continuous

Figure 14.2 Different RPS configurations.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
14.3 Problem Identification 271

14.3 Problem Identification

The objective function is stated as follows with the intention of improving reliability by determining
the optimal reward and penalty values:
∑
Nc
∑
Nc
∑
Nc
∑
Nc
F= 𝛼k ∕𝜆2k + 𝛽k ∕rk + ADCOST(EENSO − EENSD) + CIC + CRP (14.2)
k=1 k=1 k=1 k=1

where

𝜆k × rk × Li × Cp k = CIC (14.3)

The relation (14.2) consists of five elements. The first two terms, which are called maintenance
activity adjustment costs, refer to the expenses associated with adjusting the failure rate and the
average time required for repairs for every component of the distribution systems. The more invest-
ments or expenditures utilities must make in preventive care and maintenance programs in order
to achieve these standards, the lower the value associated with them (Kela and Arya 2014). Both
of these variables are drawn from the reliability growth model proposed by Duane (Ebeling 2004).
When expected energy is not delivered by the owners of the DGs while the DGs are not connected
(EENSO) or when the DGs are present (EENSD), an additional cost (ADCOST) must be incurred
by them (Arya et al. 2012). This cost is represented by the third term in the same relation. ADCOST
is calculated here as Rs. 2/kWH. The costs suffered at the level of the consumer in terms of loss dur-
ing an outage are represented by the fourth component of the relation (14.2), known as customer
interruption cost (CIC). The substantial costs linked to electricity outages are obviously amplified
when a utility’s business is supplying electricity to businesses and industries. By adding together
the costs associated with each section outage, as indicated in relation (14.3), one may calculate the
entire interruption expense for every load site. This makes it possible to estimate the total cost of all
customer interruptions. Here Cpk in (14.3) is the cost of interruptions for a downtime duration r k
related to the kth distribution sector in Rs. for each kilowatt. At a single customer load site, the cost
features of the consumers are wholly responsible for determining the customer cost. The overall
cost of the consumer’s charges associated with any load location due to any interruptions is rep-
resented by the sum of the expenses incurred by all different types of customers as a result of the
distribution outage. The calculations are done for all the distribution sectors N c . The fifth compo-
nent has to do with reward/penalty costs, which were briefly discussed in Section 14.2 (Alvehag
and Awodele 2013). As a result, balances the objective functions of the costs associated with DGs,
maintenance, customer interruption, and reward/penalty. By minimizing the objective function,
as stated by (14.2), this is accomplished.
The following restrictions are placed on the minimization of the objective function given in
(14.2).

(i) Constrained on the decision variable

𝜆k,min ≤ 𝜆k ≤ 𝜆k,max (14.4)

rk,min ≤ rk ≤ rk,max (14.5)

k = 1, … , Nc

(ii) Constraints on inequality for the energy and customer-oriented reliability indexes

SAIFI ≤ SAIFIt (14.6)

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
272 14 Reliability Enhancement Strategy for Electrical Distribution Systems Considering Reward and Penalty

SAIDI ≤ SAIDIt (14.7)

CAIDI ≤ CAIDIt (14.8)

AENS ≤ AENSt (14.9)

The optimization procedure achieves the lowest possible failure rate and repair time for the kth
sector. The reliability growth model (Ebeling 2004) is used to analyze failure and repair information
as well as related expenses to create these lower limit values. SAIFIt ,SAIDIt , CAIDIt , AENSt and
SARFIt are each of the indices’ target/threshold values. They are determined by managerial or
administrative decisions. The reliability level (R) was created in this research, taking into account
the influence of all energy and customer-oriented reliability indexes because all of these indexes are
interrelated. The sum of the normalized results for SAIFI, SAIDI, CAIDI, and AENS is represented
here by R. The normalization is done in relation to the various indices’ target values. As a result,
the method will give each index equal weight:
SAIFI SAIDI CAIDI AENS
R= + + +
SAIFIt SAIDIt CAIDIt AENSt
The primary aim of the study is to determine the failure rate and repairing time that will raise
the distribution reliability system, with the secondary goal of selecting the best reward/penalty
for the utility for the set/target reliability level. Constrained optimization of the objective function
in relation (Section 14.2) is used to achieve it in this case. In this case, the socioeconomic reliability
level described in Section 14.2 is the optimal reliability level. The following relations are used to
determine this target/set reliability level.
For an optimal RPS, the following can be demonstrated in accordance with (Ajodhia 2006):
𝜕CRP 𝜕CIC
= ,∀ R (14.10)
𝜕R 𝜕R

CRP = CIC − K, ∀ R (14.11)

K is a randomly chosen constant.

According to (14.10), the cost of customer interruption at a certain reliability level determines
the optimal reward or penalty. To obtain the optimal reward or penalty, the regulator must be able
to measure and reconstruct (Ajodhia 2006). As seen in Figure 14.2, a continuous RPS scheme with
a constant slope has been taken into consideration for this study. As a result, there is a linear rela-
tionship b/w reliability of the system and customer interruption costs. The amount of money for
each unit of system performance indicator for reliability and, consequently, a reward rate for utility
are provided by the slope of RPS in this case. The utility’s and consumers’ transactions are handled
by the value of K. All costs associated with customer disruption will be passed on to the utility if K
is set to zero. In light of this, the utility will run on overall reliability cost. In this situation, a utility
that seeks to maximize profits may suffer a loss because the permitted only the stated target value
generates enough income to pay all of the company’s dependability expenses. Due to this, deter-
mining the value of K in a continuous RPS is equivalent to deciding what level of system quality
indicator is considered to be reliable (Alvehag and Awodele 2013). Revenue only covers the overall
dependability costs incurred by the organization for the specified goal value. In light of this, deter-
mining the value of K in a continuous RPS corresponds to deciding what degree of system quality
indicator for dependability:

K = CIC ( R = Rtarget ) (14.12)

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
14.4 Rao Algorithms: An Overview 273

Therefore, regardless of the reliability target value, the profit-maximizing utility can still attain
the optimum level of reliability provided the rate of incentive is adjusted based on the cost of cus-
tomer interruption. As a result, the utility will receive no reward or penalty (incentive rate) for
meeting the target or specified reliability level. Thus, the cost of customer interruption determines
the target reliability level and, consequently, the incentive or penalty.
The FP optimization method (Yang 2013) and Rao optimization techniques (Rao 2020) are used
to address the formulated problem. Testing has been done on Roy Billinton Test System Bus 2
(RBTS-2) using the methodology described in this study. To increase reliability, the crew may be
provided with the optimized failure rates and repair times that were so obtained as target values.
While optimizing the objective function for the predetermined target level of reliability (R), the
optimal reward and penalty values are found as well.

14.4 Rao Algorithms: An Overview

The Rao algorithms are three recently created, straightforward, parameter-less, metaphor-free opti-
mization algorithms by R.V. Rao (2020) that can be used to solve both constrained and uncon-
strained optimization problems.
Step (a) Initialization: The following is used to produce a population of dimension “M”:
[ ]
S0 = U10 , U20 , … , UM 0
(14.13)
[ ] T
Ui0 = Ui10 , Ui20 , … ., UiD
0
(14.14)
Uij0 i.e. jth a uniform distribution-derived parameter of the U i vector is obtained
Uij0 = Uj,min + (Uj,max − Uj,min )randj (14.15)
U j,min and U j,max are low and high limits on decision variable U j . A random number in the [0, 1]
rang is randj . D in (14.13) shows the total decision variables to be optimized. Here, decision vari-
ables are the rate of failure 𝜆 and repairing time r for different sections of the distribution systems
considered in this chapter.
Step (b) Modifying/Updating vectors
( k )
Ui(k+1) = Uik + r1,i Ubest k
− Uworst (14.16)
( ) ( )
| | | |
Ui(k+1) = Uik + r1,i Ubest
k k
− Uworst + r2,i |Uik or Ulk | − |Ulk or Uik | (14.17)
| | | |
( ) (
| k | | | ( )
Ui(k+1) = Uik + r1,i Ubest
k
− |Uworst | + r2,i |Uik or Ulk | − Ulk or Uik (14.18)
| | | |
Relations (14.15)–(14.17) give the updated/modified vectors for (k + 1) iteration for there are three
different algorithms: Rao 1, 2, and 3. Among all the solutions obtained in the current generation
or iterations, U best (k) is the current best, and U worst (k) is current worst. Uik and Ulk are the different
candidate vectors of the population of M candidates.
Step (c) Evaluating the fitness between the updated and original vectors
{ ( k+1 ) ( )
(k)
U k+1
, if f U < f U
Ui(k+1) = i i i (14.19)
Ui(k) otherwise
(k)
The current best solution Ubest is then discovered, together with its fitness. Until the best solution
is found or the predetermined number of generations (kmax ) have been carried out, this process is
carried out for each target vector index i, and a new population is generated.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
274 14 Reliability Enhancement Strategy for Electrical Distribution Systems Considering Reward and Penalty

14.5 Steps to Solve the Problem

Data input 𝜆k,max , r k,max , 𝜆k,min , r k,min , 𝜆dg , r dg , 𝜆sw , cost interruptions (Cpk ), N i , Li , SAIFIt , SAIDIt ,
CAIDIt , and AENSt . 𝜆dg , r dg , 𝜆sw , N i , and Li are the fail rate of DG, repairs the rate of DG, the failure
rate of the switch connecting the standby DG to the distributions system at the time of power failure,
no. of customer at load point i and total load is connected at the load point i, respectively.
1. Discover K at the indices’ target values.
2. Initialization: By using relation (14.14), create “M” size is the population for each failure rate 𝜆
and repair time r, where each vector of the population contains the corresponding failure rate
and repair time of each component. These values are derived via a uniform sampling procedure
between the lower and higher bounds, as stated in relation (14.4).
3. In accordance with (Ghorashi et al. 2020), determine the system failure rate 𝜆sys,i , average repair
time r sys,i , and annual outage duration U sys,i at each load position.
4. For the vectors of the population, evaluate the energy and customer-oriented reliability indexes
for the system as indicated in Ghorashi et al. (2020). ( )
5. Determine the value of the population’s vectors’ objective function i.e. F Ui(k) , i = 1, … , M
′ ′

as stated by relations (14.2) and (14.3).

6. Determine the overall reliability R-value for each vector in the population using the formula
from Section 14.3.
7. For each population vector, evaluate the inequality constraints resulting from the relations
(14.5), (14.6), (14.7), and (14.8). Vectors that meet these restrictions will be feasible; otherwise,
(k) (k)
they will not be. Determine the best solution vector Ubest and the worst solution vector Uworst
value is determined by the objective functions, from each vector that can be considered viable.
8. Place k = 1 in the generation counter.
9. Choose the target vector with i = 1.
10. By using the relation (14.15)–(14.17), determine the vector’s updated value.
11. Utilizing relation (14.18), compare the updated initial vectors’ fitness for vectors’ fitness to keep
the good one.
12. Repeat steps 4 through 8 to get the modified or updated vector.
13. Target vector is increased to i = i + 1. Repeat from Step 11 if i ≤ M; else, raise the generation
counts k = k + 1.
14. If k ≤ kmax or the required optimum value cannot be found, start over at Step 11.

14.6 A Discussion of the Findings

On RBTS-2, the methods presented in this study have been used. The FP and Rao algorithms have
successfully solved the issue. The algorithms were programmed in MATLAB R2019b software, a
4.00 GB RAM and a 3.00 GHz Intel (R), Core (TM), i5-4590S CPU.

14.6.1 Test System-Bus 2 by Roy Billinton (RBTS-2)

There is a test system called the Roy Billinton Test System-Bus 2 that was employed in this study
(Allan et al. 1991). It is seen in Figure 14.3. The information on average maintenance durations
and failure rates of the various RBTS-2 components has been gathered from Allan et al. (1991) and
Arya (2016).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
14.6 A Discussion of the Findings 275

33/11 kV
F1 F2 F3 F4
LP1 2 1 12 LP10 16 26 27 LP16
LP8 17
LP2 13 LP17
3 28
D.G. 14
LP9 LP11 18 29 30 LP18
LP3 5 4 19
15 D.G.
D.G. LP4 D.G. LP12 LP19
6 20 31
D.G. 32
LP5
8 7 LP13 22 21 33 LP20
LP6
LP14
9 23
34 35 LP21
10 24
LP7 LP15 25 LP22
11 36

Figure 14.3 RBTS-2.

( )
Table 14.1 provides current failure rates 𝜆0k ∕year and the lowest achievable failure rates
( 0 )
(𝜆k,min /year), current average repair times rk (h) and the lowest achievable average repair times
(r k,min (h)), as well as cost coefficients 𝛼 K and 𝛽 K for the various sections of the RBTS-2 (Table 14.2).
Table 14.3 shows the optimized values for maintenance expenses, additional DG costs for energy
purchases, costs for customer interruption, reward/penalty costs, and objective function (F) as
determined by the FP and Rao algorithms. In this study, the DGs for RBTS-2 are located at sites 2,
3, 11, 12, and 18. Table 14.4 displays the amount of customer and energy-oriented reliability indices
that were optimized using the FP and Rao algorithms. Here, the comparison of the enhancements
is presented in relation to the corresponding threshold or target levels.
Table 14.5 shows optimized values for maintenance costs and extra costs incurred for DGs in
relation to various reliability levels attained around a target reliability level. Table 14.6 displays the
network, utility, and society’s optimal costs taking the effect of continuous RPS on the utility into
account. The results of Tables 14.5 and 14.6 were derived using the following targets for a reliability
index based on consumers and energy SAIFIt = 0.084746, SAIDIt = 0.198559, CAIDIt = 2.435000,
and AENSt = 1.806670. For various reliability levels, optimized values of rewards and penalties as
well as customer interruption charges are also displayed. The data from this table has been used to
plot Figure 14.4.
Table 14.7 shows the optimized rates of failure and times to repair for the various RBTS-2 sections
as determined through the FP and Rao algorithms. In Table 14.5, the Cnetwork has been made while
taking ADCOST and maintenance costs into account. The FP and Rao algorithms have been used to
solve the problem with the system under examination. Following are the control parameters used:
A maximum generation (kmax ) size of 1000 is defined, with a population size of 30. Updated step
size (∝), distribution factor (𝛽), and switch probability for FP are 0.01, 1.5, and 0.8, respectively.
The connected DGs are treated as backup or standby units in this situation. The DGs used in this
study’s average downtime and failure rate are 13.25 hours and 0.5 failures per year, respectively.
The changeover switch at DGs has a failure rate of 0.1 failure per year and a service restoration
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
276 14 Reliability Enhancement Strategy for Electrical Distribution Systems Considering Reward and Penalty

Table 14.1 RBTS-2 System-related data.

Distributor
segment 𝝀0k ∕year 𝝀k,min /year rk0 (h) r k,min (h) 𝜶 K (Rs.) 𝜷 K (Rs. × 102 )

#1 0.048750 0.036650 5.000 2.252252 2.564 22.649

#2 0.015000 0.011270 10.000 4.504504 3.205 17.589
#3 0.052000 0.039090 10.000 4.504504 1.923 4.291
#4 0.048000 0.036090 5.000 2.252252 1.282 4.291
#5 0.015000 0.011270 10.000 4.504504 2.564 13.258
#6 0.015000 0.011270 10.000 4.504504 3.045 6.647
#7 0.048750 0.036650 5.000 2.252252 3.205 5.394
#8 0.015000 0.011278 10.000 4.504504 2.564 22.649
#9 0.015000 0.011278 10.000 4.504504 3.205 17.589
#10 0.039000 0.029323 5.000 2.252252 1.923 4.291
#11 0.015000 0.011278 10.000 4.504504 1.282 4.291
#12 0.048750 0.036650 5.000 2.252250 2.564 22.649
#13 0.052000 0.039097 5.000 2.252250 3.205 17.589
#14 0.039000 0.029323 5.000 2.252250 2.564 22.649
#15 0.052000 0.039097 5.000 2.252250 3.205 17.589
#16 0.048750 0.036650 5.000 2.252250 2.564 22.649
#17 0.015000 0.011278 10.000 4.504500 3.205 17.589
#18 0.052000 0.039097 5.000 2.252250 1.923 4.291
#19 0.015000 0.011278 10.000 4.504500 1.282 4.291
#20 0.015000 0.011278 10.000 4.504500 2.564 13.258
#21 0.039000 0.029323 5.000 2.252250 3.045 6.647
#22 0.015000 0.011278 10.000 4.504500 2.564 22.649
#23 0.015000 0.011278 10.000 4.504500 3.205 17.589
#24 0.048750 0.036650 5.000 2.252250 1.923 4.291
#25 0.015000 0.011278 10.000 4.504500 1.282 4.291
#26 0.052000 0.039097 5.000 2.252250 2.564 22.649
#27 0.015000 0.011278 10.000 4.504500 3.205 17.589
#28 0.015000 0.011278 10.000 4.504500 1.923 4.291
#29 0.048750 0.036650 5.000 2.252250 1.282 4.291
#30 0.015000 0.011278 10.000 4.504500 2.564 13.258
#31 0.015000 0.011278 10.000 4.504500 3.045 6.647
#32 0.048750 0.036650 5.000 2.252250 3.205 5.394
#33 0.015000 0.011278 10.000 4.504500 2.564 22.649
#34 0.039000 0.029323 5.000 2.252250 3.205 17.589
#35 0.015000 0.011278 10.000 4.504500 1.923 4.291
#36 0.015000 0.011278 10.000 4.504500 1.282 4.291
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
14.6 A Discussion of the Findings 277

Table 14.2 Information regarding customers connected to Roy Billinton Test System-Bus 2.

Load Load average Size of

point at every load Type of customers Interruption
Sr. no number point in MW customers/premise in numbers cost (C pk ) (Rs./kW)

1 1–3,10,11 0.5350 Residential purpose 210.0 22.290

2 12,17–19 0.4500 Residential purpose 200.0 6.0450
3 8 1.0000 Small scale user 1.0 251.420
4 9 1.1500 Small scale user 1.0 251.420
5 4,5,13,14,20,21 0.5660 Government institutes 1.0 23.980
6 6,7,15,16,22 0.4540 Commercial purpose 10.0 192.980

Table 14.3 Results for the FP, Rao-1, and Rao-2-derived objective function (F) for RBTS-2: both current and
optimized.

Optimized Optimized Optimized

values (Rs.) values (Rs.) values (Rs.)
Current
Sr. no. values (Rs.) By FP By Rao-1 By Rao-2

1 (∑
Maintenance cost
∑Nc ) 232 680 306 348.71 266 826.3 276 386.1
Nc
𝛼
k=1 k
∕𝜆2
k
+ 𝛽 ∕r
k=1 k k

2 Customer
(∑ )interruption cost 807 980 302 267.59 301 950.8 301 808.6
Nc
k=1
CIC
3 Additional cost (ADCOST) to be 3118 11 434.22 11 496.36 11329.66
paid while generators are linked
with the system
(∑ )
Nc
4 Reward/penalty k=1
CRP 490 450 −15 264.67 −15 581.5 −15 723.6
5 Objective function (F) 1 534 228 604 785.85 564 692 573 800.7

Table 14.4 Reliability indices for RBTS-2 determined by FP, Rao-1, and Rao-2: both current and optimized.

Optimized Optimized Optimized

measures measures measures
Current Threshold
Sr. no. Indices measures By FP By Rao-1 By Rao-2 measures

1 SAIFI (interruptions/customer 0.09860 0.080462 0.080971 0.081635 0.084746

per year)
2 SAIDI (h/customer per year) 0.58820 0.177953 0.168077 0.191913 0.198559
3 CAIDI (h/customer interruption 5.96660 2.2301 2.075657 2.398312 2.435000
per year)
4 AENS (kW/customer) 4.66410 1.667670 1.651387 1.695072 1.806670
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
278 14 Reliability Enhancement Strategy for Electrical Distribution Systems Considering Reward and Penalty

Table 14.5 Values of maintenance cost and ADCOST, as determined by FP, and related values of overall
reliability R (before and after the target values) (considering SAIFIt = 0.084746, SAIDIt = 0.198559,
CAIDIt = 2.435000, AENSt = 1.806670) for Roy Billinton Test System (Bus-2).

Sr. Maintenance
no. cost (Rs.) ADCOST (Rs.) SAIFI SAIDI CAIDI AENS F (Rs.) R

1 292 425.50 10 128.00 0.0860 0.2250 2.6230 2.010 667 137.20 3.1320
2 290 868.10 10 385.00 0.0850 0.2230 2.660 1.9430 656 514.00 3.0940
3 295 194.80 10 630.00 0.0820 0.2260 2.7570 1.8780 658 002.00 3.060
4 264 349.60 10 650.00 0.0860 0.2040 2.4170 1.8730 617 472.50 2.9640
5 275 661.70 10 687.00 0.0860 0.1980 2.2860 1.8640 623 918.50 2.9160
6 281 435.60 10 933.70 0.08290 0.2030 2.3220 1.7980 620 153.90 2.8710
7 266 436.70 11 206.00 0.0810 0.190 2.3830 1.7280 574 835.80 2.80
8 266 249.70 11 252.00 0.0830 0.1870 2.2530 1.7150 574 703.20 2.7750
9 270 691.60 11 340.00 0.0830 0.1760 2.1560 1.6920 576 717.20 2.7120
10 2 831 670 11 400.00 0.0830 0.1780 2.140 1.6770 585 878.10 2.7070
11 295 181.60 11 375.00 0.0820 0.1750 2.1560 1.6830 597 618.40 2.6980
12 306 348.70 11 434.00 0.08040 0.1780 2.2300 1.6680 604 785.90 2.6960
13 268 835.90 11 47s1.00 0.0840 0.1720 2.1150 1.6580 560 790.90 2.6910
14 287 743.70 11 574.00 0.0830 0.170 2.0560 1.6310 585 959.70 2.6450
15 275 668.90 11 618.00 0.0820 0.1660 2.0740 1.620 571 269.70 2.6230
16 271 853.20 11 595.00 0.080 0.1660 2.1210 1.6260 562 966.30 2.6190
17 275 059.60 11 605.00 0.0810 0.1660 2.0910 1.6230 568 568.0 2.6160
18 287 907.50 11 526.00 0.080 0.1660 2.0630 1.6440 578 949.50 2.6140

time of 0.5 hours, respectively. Figure 14.4 depicts the implications of an ideal continuous RPS for
the expenses associated with different parameter values for RBTS-2. The optimal cost for the n/w,
utilized, social, reward or penalty, and consumer interruption are graphically illustrated here for
different reliability levels.
Table 14.7 shows the optimized rates of failure and times to repair for the various RBTS-2 sections
as determined through the FP and Rao algorithms. In Table 14.5, the Cnetwork has been made while
taking ADCOST and maintenance costs into account. The FP and Rao algorithms have been used to
solve the problem with the system under examination. Following are the control parameters used:
A maximum generation (kmax ) size of 1000 is defined, with a population size of 30. Updated step
size (∝), distribution factor (𝛽), and switch probability for FP are 0.01, 1.5, and 0.8, respectively.
The connected DGs are treated as backup or standby units in this situation. The DGs used in this
study’s average downtime and failure rate are 13.25 hours and 0.5 failures per year, respectively.
The changeover switch at DGs has a failure rate of 0.1 failure per year and a service restoration
time of 0.5 hours, respectively. Figure 14.4 depicts the implications of an ideal continuous RPS for
the expenses associated with different parameter values for RBTS-2. The optimal cost for the n/w,
utilized, social, reward or penalty, and consumer interruption are graphically illustrated here for
different reliability levels.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
14.6 A Discussion of the Findings 279

Table 14.6 The effects of continuous RPS on utility (considering SAIFIt = 0.0847460, SAIDIt = 0.1985590,
CAIDIt = 2.4350000, AENSt = 1.8066700) for RBTS-2 as obtained by FP.

utility society
Sr no. CIC C RP C network Ctotal = Cnetwork + CRP Ctotal = Cnetwork + CIC R

1 341 057.99 23 525.732 302 553.5096 326 079.2416 643 611.5094 3.132077
2 336 396.39 18 864.131 301 253.4363 320 117.5669 637 649.8347 3.0939145
3 334 854.53 17 322.268 305 825.2102 323 147.4785 640 679.7463 3.0603485
4 330 002.67 12 470.406 274 999.4572 287 469.8636 605 002.1314 2.963884
5 327 551.30 10 019.036 286 348.1753 296 367.2114 613 899.4791 2.9163098
6 322 658.38 5 126.11 292 369.40 297 495.5196 615 027.7874 2.8714413
7 307 362.90 −10169.36 277 642.2999 267 472.9406 585 005.2084 2.8000841
8 307 366.70 −10 165.56 277 502.0912 267 336.532 584 868.7998 2.7751445
9 306 108.78 −11 423.48 282 031.8602 270 608.3757 588 140.6435 2.7116754
10 304 421.57 −13 110.69 294 567.1973 281 456.5055 598 988.7733 2.7073760
11 304 297.26 −13 235 306 556.1356 293 321.1365 610 853.4043 2.6984982
12 302 267.59 −15 264.68 317 782.9447 302 518.268 620 050.5358 2.6960579
13 299 008.20 −18 524.07 280 306.7189 261 782.6536 579 314.9213 2.6909185
14 302 086.95 −15 445.31 299 318.1045 283 872.7911 601 405.0588 2.6446134
15 300 757.62 −16 774.65 287 286.6803 270 512.0334 588 044.3011 2.6231037
16 298 525.19 −19 007.07 283 448.1982 264 441.1262 581 973.394 2.6188200
17 299 718.06 −17 814.21 286 664.1614 268 849.9537 586 382.2215 2.6159990
18 298 523.97 −19 008.3 299 433.8125 280 425.5158 597 957.7836 2.6138074

700 000

600 000

500 000
Optimum annual cost

400 000 CRP

CIC
300 000 Cnetwork
Ctotal_utility = Cnetwork + CRP
200 000 Ctotal_society = Cnetwork + CIC
R
100 000

0
3.132
3.094
3.060
2.964
2.916
2.871
2.800
2.775
2.712
2.707
2.698
2.696
2.691
2.645
2.623
2.619
2.616
2.614

–100 000
Optimum reliability

Figure 14.4 Effect of an optimal continuous RPS on the price of various RBTS-2 parameters.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
280 14 Reliability Enhancement Strategy for Electrical Distribution Systems Considering Reward and Penalty

Table 14.7 FP, Rao-1, and Rao-2’s optimized values for rate of failure and time to repair obtained for the
RBTS-2.

Failure Failure Failure Repair Repair Repair

rates rates rates times times times
Distributor (/year) (/year) (/year) (in h) (in h) (in h)
system
segments With FP With Rao-1 With Rao-2 With FP With Rao-1 With Rao-2

1 0.03679 0.03666 0.03665 2.25225 2.25227 2.29738

2 0.01497 0.01499 0.01500 4.50497 4.68206 9.97628
3 0.03910 0.04089 0.03909 4.60288 4.55572 9.99998
4 0.03609 0.03666 0.03613 2.25237 2.26249 2.25225
5 0.01127 0.01468 0.01421 4.96719 9.84338 4.50450
6 0.01500 0.01478 0.01500 9.34639 4.93965 4.50609
7 0.03665 0.03759 0.03668 2.25229 2.35412 2.25344
8 0.01237 0.01461 0.01500 9.43417 9.93448 4.51630
9 0.01500 0.01473 0.01499 4.50613 4.53304 6.57998
10 0.03899 0.03880 0.02958 2.25525 2.25225 2.35627
11 0.01128 0.01500 0.01128 4.50491 4.50529 4.50487
12 0.03665 0.03732 0.03673 2.27395 2.26746 2.26344
13 0.03943 0.03922 0.03910 2.44635 2.25259 2.26420
14 0.03102 0.03261 0.02932 2.25353 2.25240 2.25227
15 0.03911 0.03910 0.03910 2.25225 2.26204 2.25225
16 0.04238 0.03704 0.03665 2.25229 2.25495 2.25225
17 0.01499 0.01454 0.01500 9.97048 6.29397 4.50451
18 0.03910 0.03944 0.03910 2.27024 2.25240 2.26192
19 0.01500 0.01495 0.01500 9.97133 4.79284 9.97537
20 0.01485 0.01468 0.01500 9.98236 8.97340 4.90722
21 0.02934 0.03895 0.02938 2.75800 2.25750 2.25662
22 0.01260 0.01224 0.01226 5.03652 4.53734 4.50469
23 0.01136 0.01500 0.01489 5.11156 9.99853 4.51759
24 0.04780 0.03692 0.03665 2.25635 2.31743 2.25255
25 0.01500 0.01129 0.01128 4.50450 4.57636 4.52306
26 0.03910 0.03919 0.04350 2.31682 2.25227 2.35457
27 0.01481 0.01447 0.01500 4.52148 4.56738 4.50450
28 0.01128 0.01478 0.01487 9.85532 4.72852 4.50450
29 0.03665 0.03874 0.03951 2.25238 2.25255 2.45385
30 0.01500 0.01476 0.01498 9.53818 4.50896 5.46092
31 0.01500 0.01500 0.01500 9.99991 4.51428 4.57499
32 0.04875 0.03665 0.04875 2.25870 2.27121 2.28080
33 0.01283 0.01464 0.01500 4.50450 5.37821 10.00000
34 0.02996 0.02963 0.02933 2.25408 2.57953 2.31643
35 0.01153 0.01494 0.01145 5.59178 4.50900 6.79013
36 0.01128 0.01499 0.01391 4.50450 4.68041 4.50546
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 281

14.7 Conclusion

By including a reward or penalty for the utilities by the regulator in this research, the distribution
system reliability has improved. In light of the target values for the customer and energy-oriented
reliability indexes, the best values for rewards and penalties have been identified. The results show
that utility is rewarded when each of these indicators falls below the desired or target level, and
it is penalized when they are above. The Flower Pollination and Rao algorithms have been used
to optimize the objective function that was created. In addition to producing optimized values
for all reliability indices, other components of the objective function, including costs associated
with maintenance, customer disruptions, and additional expenses required to reach the level of
reliability have also been produced with optimized values. All optimization approaches produced
extremely similar optimized outcomes.

References

Ajodhia, V.S. (2006). Regulating beyond price. Integrated price-quality regulation for electricity
distribution networks. PhD Dissertation, Delft University of Technology, Delft.
Allan, R.N., Billinton, R., Sjarief, I. et al. (1991). A reliability test system for educational purposes-basic
distribution system data and results. IEEE Transactions on Power Systems 6 (2): 813–820.
Alvehag, K. and Awodele, K. (2013). Impact of reward and penalty scheme on the incentives for
distribution system reliability. IEEE Transactions on Power Systems 29 (1): 386–394.
Amirrezai, M., Rezaie, H., and Goetz, S.M. (2023). Feasibility study of incorporating static
compensators in distribution networks containing distributed generation considering system power
factor. Electric Power Systems Research 219: 109253.
Arya, R. (2016). Determination of customer perceived reliability indices for active distribution systems
including omission of tolerable interruption durations employing bootstrapping. IET Generation
Transmission and Distribution 10 (15): 3795–3802.
Arya, R., Choube, S.C., and Arya, L.D. (2012). Reliability evaluation and enhancement of distribution
systems in the presence of distributed generation based on standby mode. International Journal of
Electrical Power & Energy Systems 43 (1): 607–616.
Billinton, R. and Allan, R.N. (1996). Reliability Evaluation of Power Systems. Springer International
Edition.
Billinton, R. and Li, W. (1994). Reliability Assessment of Electrical Power Systems using Monte Carlo
Methods. New York, NY, USA: Plenum.
Billinton, R. and Pan, Z. (2004). Historic performance-based distribution system risk assessment. IEEE
Transactions on Power Delivery 19 (4): 1759–1765.
CEER (2011). 5th CEER benchmarking report on the quality of electricity supply, Council of European
Energy Regulators, Technical Report, Brussels, Belgium.
Ebeling, C.E. (2004). An Introduction to Reliability and Maintainability Engineering. Tata McGraw-Hill
Education.
Fotuhi, M., Shourkaei, H.M., Kharazi, M.B., and Salimi, A. (2006). Reliability assessment of utilities
using an enhanced reward-penalty model in performance based regulation system. In: 2006
International Conference on Power System Technology, 1–6. IEEE.
Fumagalli, E., Schiavo, L., and Delestre, F. (2007). Service Quality Regulation in Electricity Distribution
and Retail. Springer Science & Business Media.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
282 14 Reliability Enhancement Strategy for Electrical Distribution Systems Considering Reward and Penalty

Ghasemi, M. and Dashti, R. (2017). A risk-based model for performance-based regulation of electric
distribution companies. Utilities Policy 45: 36–44.
Ghasemi, M. and Dashti, R. (2018). Designing a decision model to assess the reward and penalty
scheme of electric distribution companies. Energy 147: 329–336.
Ghorashi, S.M., Rastegar, M., Senemmar, S., and Seifi, A.R. (2020). Optimal design of reward-penalty
demand response programs in smart power grids. Sustainable Cities and Society 60: 102150. https://
doi.org/10.1016/j.scs.2020.102150.
Ghotboddini, R. and Toossian Shandiz, H. (2022). Reward and penalty model for the lighting of public
thoroughfares contracts: an empirical study in a distribution company. Journal of AI and Data
Mining 10 (1): 127–138.
Grahn, E., Ström, L., Alvehag, K. et al. (2016). Incentivizing continuity of supply in Sweden. In: 2016
13th International Conference on the European Energy Market (EEM), 1–5. IEEE.
Janjic, A., Velimirovic, L.Z., and Vranic, P. (2021). Designing an electricity distribution reward-penalty
scheme based on spatial reliability statistics. Utilities Policy 70: 101211.
Jooshaki, M., Abbaspour, A., Fotuhi-Firuzabad, M. et al. (2014). A new reward-penalty mechanism for
distribution companies based on concept of competition. In: IEEE PES Innovative Smart Grid
Technologies, Europe, 1–5. IEEE.
Jooshaki, M., Ali Abbaspour, M.F.F., and Moeini-Aghtaie, M. (2017). Designing a regulatory
framework for efficient integration of distributed generation technologies. Turkish Journal of
Electrical Engineering and Computer Sciences 25 (5): 4117–4130.
Jooshaki, M., Abbaspour, A., Fotuhi-Firuzabad, M. et al. (2018). Designing a new procedure for reward
and penalty scheme in performance-based regulation of electricity distribution companies.
International Transactions on Electrical Energy Systems 28 (11): e2628.
Kela, K.B. and Arya, L.D. (2014). Reliability optimization of radial distribution systems employing
differential evolution and bare bones particle swarm optimization. Journal of The Institution of
Engineers (India): Series B 95 (3): 231–239.
Khonakdar-Tarsi, I., Fotuhi-Firuzabad, M., Ehsan, M. et al. (2021). Reliability incentive regulation
based on reward-penalty mechanism using distribution feeders clustering. International
Transactions on Electrical Energy Systems 31 (8): e12958.
Lu, Y., Khan, Z.A., Gunduz, H. et al. (2021). Comparison of two strategies of reward-penalty
mechanism for promoting net zero energy buildings. Sustainable Energy Technologies and
Assessments 47: 101347.
Mohammad Nezhad-Shourkaei, H. and Fotuhi-Firuzabad, M. (2011). Principal requirements of
designing the reward-penalty schemes for reliability improvement in distribution systems. 21st
International Conference on Electricity Distribution Measured Indices, no. 1207, pp. 6–9.
Mohammad Nezhad-Shourkaei, H., Fotuhi-Firuzabad, M., and Billinton, R. (2011). Integration of
clustering analysis and reward/penalty mechanisms for regulating service reliability in distribution
systems. IET Generation Transmission and Distribution 5 (11): 1192–1200.
Rao, R. (2020). Rao algorithms: Three metaphor-less simple algorithms for solving optimization
problems. International Journal of Industrial Engineering Computations 11 (1): 107–130.
Sappington, D.E.M. (2005). Regulating service quality: A survey. Journal of Regulatory Economics 27
(2): 123–154.
Simab, M. and Haghifam, M.R. (2012). Quality performance based regulation through designing
reward and penalty scheme for electric distribution companies. International Journal of Electrical
Power & Energy Systems 43 (1): 539–545.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 283

Simab, M., Alvehag, K., Söder, L., and Haghifam, M.-R. (2012). Designing reward and penalty scheme
in performance-based regulation for electric distribution companies. IET Generation Transmission
and Distribution 6 (9): 893–901.
Sumesh, S., Potdar, V., and Krishna, A. (2018). Clustered prosumer penalty framework on smart grid.
In: Advances in Smart Grid and Renewable Energy, 1–12. Singapore: Springer.
Sumesh, S., Potdar, V., and Krishna, A. (2019). Cubic reward penalty structure for power distribution
companies. International Journal of Systems Assurance Engineering and Management 10 (3): 350–368.
Wang, B., Camacho, J.A., Pulliam, G.M. et al. (2018). New reward and penalty scheme for electric
distribution utilities employing load-based reliability indices. IET Generation Transmission and
Distribution 12 (15): 3647–3654.
Yang, X.S. (2013). Flower pollination algorithm for global optimization. In: International Conference on
Unconventional Computing and Natural Computation, 240–241. Berlin Heidelberg: Springer.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
285

15

Reliability Analysis of Composite Distribution System Using Frequency

Duration Concept
Atul Koshti 1 , Eram Taslima 2 , Pradeep Purey 3 , Liladhar Arya 4 , and Sharat C. Choube 5
1
Department of Electrical Engineering, GES’S R.H. Sapat College of Engineering, Management Studies & Research, Nashik,
Maharashtra, India
2 Department of Electrical Engineering, Indian Institute of Availabilities on annual basis Technology (BHU), Varanasi,

Uttar Pradesh, India

3
Department of Electronics Engineering, Maharaja Ranjit Singh College of Professional Sciences, Indore, Madhya Pradesh, India
4
Department of Electrical Engineering, Shri Govindram Seksaria Institute of Technology and Science, Madhya Pradesh, India
5 Department of Electrical Engineering, UIT, RGPV, Madhya Pradesh, India

15.1 Introduction

Reliability indices assessment is qualitative analysis required for power system adequacy assess-
ment. This becomes very essential in distribution system with integration of DG i.e. in composite
distribution system (CDS) (Billinton and Allan 1992; Pereira and Balu 1992). DG is usually inte-
grated at 11 KV and possesses advantages (El-Khattam and Salama 2004) such as relieving of system
loading, improved voltage, etc. Also due to the uncertain environment condition i.e. stochastic
nature of DG, availability of distribution system substation capacity, increased failure rate of com-
ponents creates challenges for the distribution system planning in view of maintaining desired
level of reliability. Adequacy review of CDS is conferred in Hegazy et al. (2003), Arya and Koshti
(2008, 2012), Costa and Matos (2009), and Xu and Singh (2012). Sample-based reliability evaluation
methods are discussed in Kothari et al. (2022), Probability Methods Subcommittee (1979), Billinton
and Li (1993), and Wang et al. (2013). Well-being evaluation for power system is used for adequacy
assessment (Wangdee and Billinton 2006; Amaral et al. 2010; Ahmadi-Khatir et al. 2009; Billinton
and Lian 1994). Statistical analysis for CDS is carried out in Arya et al. (2013), Halve et al. (2023),
and Koshti et al. (2023). Reliability planning is discussed in detail and advances with design
(Khezri and Mahmoudi 2020; Roy et al. 2022; Ranjbar et al. 2022; Yahaya et al. 2020; Hamzeh
and Vahidi 2020) for hybrid renewable energy systems (HRES). Integration of renewable energy
systems (RES) for sustainable development is addressed in Hassan et al. (2022).
It is observed that there is rising interest in further segregation of reliability in three states i.e.
(H), (M), and (R). A three-state model can give further insight into the reliability planning as per
the deterministic criteria.
Therefore, this chapter suggests deterministic criteria-based reliability indices evaluation plan-
ning which uses a ratio of supply capacity available to the load capacity is driven as one of the
criteria that segregate reliability in three state. Threshold values of this ratio are used to select CDS
states H, M, or R. Moreover, the frequency and duration concept is used to evaluate frequencies of
H, M, and R states along with duration indices in H, M, and risky states e.g. MHT, MRT, and MMT.
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
286 15 Reliability Analysis of Composite Distribution System Using Frequency Duration Concept

Thus Section 15.1 presents an introduction. Modeling of different components in CDS such
as substation capacity, DG capacity and load is discussed in Section 15.2. Section 15.3 presents
a frequency-duration concept for reliability indices evaluation. Section 15.4 presents algorithm
using Monte Carlo simulation (MCS)-based reliability analysis of CDS. Section 15.5 presents result
section. Section 15.6 presents illustrative examples. Section 15.7 presents conclusions.

15.2 Components Modeling in Composite

Distribution System (CDS)
15.2.1 Capacity Modeling
Capacity of a CDS is given as:
CT = CS,S + CdgUS (15.1)
where CS,S and CdgUS are the capacity available due to distribution substation (DS) and DG units
and their capacity samples are obtained as follows
( )2
CS,S −CS,S
1 −0.5 𝜎S
f (CS,S ) = √ e (15.2)
2𝜋𝜎S
( )2
CdgU,i −CdgU,i
1 −0.5 𝜎i
f (CdgU,i ) = √ e (15.3)
2𝜋𝜎i
Availability and unavailability of ith DG unit are obtained in terms of repair and failure rates 𝜇i
and 𝜆i follows:
𝜇i
Ai = (15.4)
𝜆i + 𝜇i
𝜆i
Ai = (15.5)
𝜆i + 𝜇i
And total capacity from DG is calculated as
∑
CdgUS = CdgU,p (15.6)
p

p 𝜖 set of available units.

15.2.2 Load Modeling

The load is treated as normally distributed of mean ld and standard deviation 𝜎ld . Load at a bus in
an interval is not constant. Truly speaking it is a random variable. It has been a natural choice to
assume Normal distribution as probability distribution function of load at a bus:
( ) ( √ ) −0.5( ld−ld )2
𝜎ld
f li = 1∕( 2𝜋𝜎ld ) e (15.7)

15.3 Frequency-Duration Concept for Reliability Indices Evaluation

Figure 15.1 represents state transition diagram for three state model which is represented based on
deterministic criteria as separation of H, M, or R states by a lines between state also it is possible
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
15.3 Frequency-Duration Concept for Reliability Indices Evaluation 287

Healthy state

qki
Risky state
qji i qik
Marginal state k
qij
qjk

j
qkj

Figure 15.1 State transition diagram for three-state model.

that there may be transition from individual state to other states between each other due to failure
of single component. Here if a system exists in a particular state then due to state change of the
component (on/off) the state of system transition (Arya et al. 2013; Halve et al. 2023) will take
place. Based on this frequency of the system and duration indices can be calculated in H, M, and R
state given as follows
∑ ∑( )
fH = pj qjk (15.8)
j∈H k∈M
k∈R
∑ ∑( )
fM = pj qjk (15.9)
j∈M k∈H
k∈R
∑ ∑( )
fR = pj qjk (15.10)
j∈R k∈H
k∈M
∑
pH = pj (15.11)
j∈H
∑
pM = pj (15.12)
j∈M
∑
pR = pj (15.13)
j∈R
PH
MHT = (15.14)
fH
PM
MMT = (15.15)
fM
PR
MRT = (15.16)
fR
AX = 8760 • pX (15.17)

where X will be H, M, or R and qjk are the transition rates from jth to kth state.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
288 15 Reliability Analysis of Composite Distribution System Using Frequency Duration Concept

15.4 MCS-Based Reliability Indices Evaluation of CDS

The well-being analysis framework is used for reliability indices evaluation of CDS using MCS
enlighten in the following steps:
S1: The (DS) is assumed to be available forever and capacity is evaluated using (15.2).
S2: The total DG capacity CdgUS can be evaluated with (15.6).
S3: For CDS the CT is obtained with (15.1).
S4: Sample for load (ld) is acquired using relation (15.7)
S5: Subsequently state is identified with:
System state Deterministic criteria
Healthy state if (CT ∕l) ≥ 1.2
Marginal state if 1.2 > (CT ∕l) ≥ 1.1
Risky state Otherwise

S6: kth system state duration is attained as:

1
Tk = − • lnu (15.18)
q
u is random numeral [0, 1] and q is total rate of transition (Arya et al. 2013).
S7: State transition method (Arya et al. 2013; Halve et al. 2023) is used for system transition due
to the state change of a DG unit.
S8: Probability indices in H, M, and R states are calculated for ith sample as follows:
TH,i
pH,i = (15.19)
TD
TM,i
pM,i = (15.20)
TD
TR,i
pR,i = (15.21)
TD
where TH,i , TM,i , and TR,i are the duration sample for H, M, and R states, respectively, and
TD = TH,i + TM,i + TR,i .
S9: Convergence is ascertained with calculation of 𝛽 in terms of p and NS
√
(1 − p)
β= (15.22)
p • NS
If 𝛽 ≤ 0.04, then the solution is converged; otherwise, do again from steps 2 to 8 for Sample
(say NS = 10,000).
S10: Frequency of H, M, and R states given as:
(nHM + nHR )
fH = pH,i (𝜆HM + 𝜆HR ) = (15.23)
TD
(nMH + nMR )
fM = pM,i (𝜆MH + 𝜆MR ) = (15.24)
TD
(n + nRM )
fR = pR,i (𝜆RH + 𝜆RM ) = RH (15.25)
TD
where 𝜆XY is the total transition rate and nXY is the number of transitions from state X to Y
where X and Y may be H, M, or R.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
15.5 Result and Discussion 289

S11: Then, mean time for H, M, and R states evaluated correspondingly:

p
MHT = Hi (15.26)
fH
pMi
MMT = (15.27)
fM
pRi
MRT = (15.28)
fR

S12: Availabilities on annual basis in H, M, and R states are obtained as:

AX = 8760 • PXi (15.29)
where X will be H, M, or R.

15.5 Result and Discussion

A CDS (Arya et al. 2013) is considered with (DS) data CS,S = 2800 KW, 𝜎S = 140 KW and shown in
Figure 15.2. And reliability data of DG’s are shown in Table 15.1 (Arya et al. 2013).
Load is assumed as normally distributed with mean ld = 2700 KW and standard deviation 𝜎ld =
135 KW. The simulation is run using algorithm developed in Section 15.4, and results are obtained
and given in Table 15.2 with deterministic criteria explained in Section 15.4. And samples are
obtained for simulation and reliability analysis is carried out to evaluate various reliability indices
along with frequency and duration indices. Comparing the results shows a significant decrease
in risky state probability with the addition of DG capacity is observed and also notable change

33 KV substation bus

G1

11 KV

G2 G3

11 KV

G5

G4

Figure 15.2 Sample composite distribution system.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
290 15 Reliability Analysis of Composite Distribution System Using Frequency Duration Concept

Table 15.1 Reliability data of DG.

No. I II III IV V

𝜆 (per hour) 1 × 10−3 2.4 × 10−3 3 × 10−3 4 × 10−3 7 × 10−3

μ (per hour) 3 × 10−3 5 × 10−3 6 × 10−3 4 × 10−3 5 × 10−3
A 0.750 0.676 0.666 0.500 0.417
Mean capacity (KW) 100.00 100.00 100.00 100.00 100.00
Standard deviation (KW) 5.000 5.000 5.000 5.000 5.000

Source: Arya et al. (2013)/with permission of Elsevier.

Table 15.2 Reliability indices of CDS.

Sr. No. Reliability indices Without DG With DG

1. PH 0.017400 0.616800
2. PM 0.188700 0.335500
3. PR 0.793900 0.051700
4. fH (occ/yr) 0.000365 0.004050
5. fM (occ/yr) 0.002100 0.002780
6. fR (occ/yr) 0.00300 0.000450
7. MHT (h/yr) 47.67120 152.1200
8. MMT (h/yr) 88.65000 120.5000
9. MRT (h/yr) 264.6333 112.7200
10. AH (h/yr) 152.4240 5403.168
11. AM (h/yr) 1653.000 2938.980
12. AR (h/yr) 6954.600 452.8920

is observed for increase in healthy and marginal state probability which indicates that samples
obtained based on deterministic criteria are shifted in H state from R state results in increase of PH ,
PM , and reduction in PR . which gives more mean healthy time (MHT), mean marginal time (MMT),
and less mean risky time (MRT). Availability AH , AM is observed improved is whereas significant
reduction is observed in AR .

15.6 Illustrative Examples

15.1 A distribution substation feeds a load via a distributor segment. Capacity of substation is
normally distributed with mean and standard deviation C = 180 MW, 𝜎C = 20 MW. And
Load is also normally distributed with mean and standard deviation l = 100 MW, 𝜎l =
10 MW. Then calculate reliability assuming the distributor reliability is unity.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
15.6 Illustrative Examples 291

Solution:
The reliability is given (Arya and Koshti 2008) as
𝛽
R = 𝜑(𝛽) = N(0,1) dt
∫−∞
CT − l 80
𝛽= √ = √ = 3.577
𝜎c + 𝜎l
2 2 500
From table of normal distribution [5] the reliability is obtained as
R = 𝜑(3.577) = 0.9997674
15.2 Problem statement is same as above Exercise 15.1 but reliability of distributor segment is
assumed to be 0.95. Obtain probability of meeting the load.
Solution:

R = 𝜑(𝛽) • (Reliability of distributor)

Product law follows due to independent event in series.
R = 0.9997674 × 0.95
R = 0.94977903
15.3 A distribution substation meets the load having distributed generation (DG) with follow-
ing data. Capacity of substation is normally distributed with C = 100 MW, 𝜎C = 10 MW.
And Load is also normally distributed with l = 80 MW, 𝜎l = 10 MW. DG is normally
distributed with capacity Cdg = 20 MW and 𝜎dg = 5 MW. Obtain reliability of meeting
the load.
Solution:
Total average capacity
CT = C + Cdg = 120 MW
l = 80 MW
𝜎l = 10 MW
√
𝜎T = 𝜎C2 + 𝜎dg 2
𝜎T = 11.18 MW
Safety Index is written as
CT − l
𝛽= √
𝜎T 2 + 𝜎l 2
120 − 80
𝛽= √
11.182 + 102
40
𝛽= = 2.67
14.999
Therefore
R = 𝜑(𝛽) = 𝜑(2.67)
R = 0.996207
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
292 15 Reliability Analysis of Composite Distribution System Using Frequency Duration Concept

15.4 Now assume same capacity and load data in Exercise 6.3. Except the load is fed via a dis-
tributor with reliability 0.99. How the reliability is affected.
Solution:
R = 𝜑(𝛽) • (Reliability of distributor)
R = 𝜑(2.67) × 0.99
R = 0.98215

15.5 A distribution substation feeds a load via a distributor having reliability 0.98. Capacity of
distribution substation and load is normally distributed with
C = 100 MW.
𝜎C = 10 MW
l = 70 MW
𝜎l = 10 MW
It is connected to a distribution generation have availability 0.9 and rated capacity 30 MW.
Obtain reliability.
Solution:
Mean capacity of distribution generation
Cdg = 0.1 × 0 + 0.9 × 30
Cdg = 27 MW
Total mean capacity
CT = C + Cdg
CT = 100 + 27
CT = 127 MW
𝜎dg 2 = 0.1 × (0 − Cdg )2 + 0.9 × (30 − Cdg )2
𝜎dg 2 = 0.1 × (0 − 27)2 + 0.9 × (30 − 27)2
𝜎dg 2 = 81
Therefore
𝜎dg = 9 MW
√
𝜎T = 𝜎C 2 + 𝜎dg
2

√
𝜎T = 100 + 81
𝜎T 2 = 181
CT − l
𝛽= √
𝜎T 2 + 𝜎l 2
127 − 70
𝛽= √
181 + 100
57
𝛽= = 3.400
16.76
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
15.6 Illustrative Examples 293

Therefore, 𝜑(𝛽) = 𝜑(3.400)

R = 0.996631 × 0.98
R = 0.97669

15.6 For the deterministic criteria given below:

System state Deterministic criteria

Healthy state if (C∕l) ≥ 1.2
Marginal state if 1.2 > (C∕l) ≥ 1.1
Risky state Otherwise

Determine the system state for the following cases from (a) to (c). Also determine the num-
ber of transitions from one state to another when moving from case (a) to (c) sequentially.
What if determine the number of transitions from one state to another when moving from
case (c) to (a) sequentially i.e. in reverse order. Where Capacity of substation is normally
distributed with mean and standard deviation C, 𝜎C . And Load is also normally distributed
with mean and standard deviation l, 𝜎l .
(a)
C = 130 MW
𝜎C = 10 MW
l = 100 MW
𝜎l = 10 MW
(b)
C = 115 MW
𝜎C = 10 MW
l = 100 MW
𝜎l = 10 MW
(c)
C = 105 MW
𝜎C = 10 MW
l = 100 MW
𝜎l = 10 MW
Solution:
(a)
(C∕l) = 130∕100
(C∕l) = 1.3
As (C∕l) ≥ 1.2
(Therefore system state is healthy state (H).)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
294 15 Reliability Analysis of Composite Distribution System Using Frequency Duration Concept

(b)
(C∕l) = 115∕100
(C∕l) = 1.15
As 1.2 > (C∕l) ≥ 1.1

Therefore system state is marginal state (M).

(c)
(C∕l) = 105∕100
(C∕l) = 1.05
As (C∕l) < 1.1

Therefore system state is risky state (R).

As it is observed that moving from Case (a) to (b) system state is changed from H to M
therefore
nHM = 1 (number of transition from H to M)

Further moving from Case (b) to (c)

nMR = 1 (number of transition from M to R)

If moves from case (c) to (a) i.e. in reverse order sequentially then it is observed that
moving from Case (c) to (b) system state is changed from R to M therefore
nRM = 1 (number of transition from R to M)

Further moving from Case (b) to (a)

nMH = 1 (number of transition from M to H)

15.7 The Capacity of distribution substation and load is normally distributed with
C = 100 MW
𝜎C = 10 MW
l = 95 MW
𝜎l = 10 MW
Distribution substation is integrated with a DG unit having normally distribution with
Cdg = 10 MW and 𝜎dg = 1 MW and assuming availability as 1. Then determine the system
state as per deterministic criteria given Exercise 15.6 with and without DG unit.
Solution:
Without DG unit
(C∕l) = 100∕95
(C∕l) = 1.05263

As
(C∕l) < 1.1
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
15.6 Illustrative Examples 295

Therefore system state is risky state.

Now with single DG unit: Total system capacity

CT = C + Cdg = 110 MW
(CT ∕l) = 110∕95
(CT ∕l) = 1.15789
As 1.2 > (CT ∕l) ≥ 1.1

Therefore system state is marginal state.

It is observed that system state is changed from risky state to marginal state with addition
of single DG unit of given capacity.

15.8 In the example Exercise 15.7 a second identical DG unit is integrated then determine the
system state.
Now with single DG unit: Total system capacity

CT = C + Cdg = 120 MW
(CT ∕l) = 120∕95
(CT ∕l) = 1.26315
As (C∕l) ≥ 1.2

Therefore system state now in healthy state.

Note: It is observed that with addition of one DG units the system state is changed from
risky to marginal then further addition of DG unit system state is shifted to healthy state.

15.9 For a distribution system depending on load and supply capacity availability, the system
states are represented by three-state model (H, M, R) as discussed in chapter given for 10
samples of distribution substation capacity and load distribution sequentially is given as
follows:

S = [H M R H R M H R H M]

Then determine the number of transitions from one state to another nHM , nHR , nMH , nMR ,
nRH , and nRM .
Solution:
The number of transitions from one state to another is obtained sequentially as follows:

nHM = 2
nHR = 2
nMH = 1
nMR = 1
nRH = 2
nRM = 1
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
296 15 Reliability Analysis of Composite Distribution System Using Frequency Duration Concept

15.10 A three-state model (Healthy [H], Marginal [M], and Risky state [R]) is presented for reli-
ability analysis for a composite distribution system with the following data:

𝜆HM = 0.5
𝜆HR = 0.3
𝜆MR = 0.4

The probabilities are given as follows for the three states:

pH = 0.2
pM = 0.7
pR = 0.1

Calculate
i) Transition rates 𝜆MH , 𝜆RH , 𝜆RM
ii) Frequencies fH , fM , fR
iii) Mean time in three states MHT, MMT, MRT
iv) Availabilities on annual basis in three states AH , AM , AR .
Refer Figure 15.3 for three-state model.
Solution:
From Frequency balance approach (Arya et al. 2013)

PH 𝜆HM = PM 𝜆MH

Therefore

𝜆MH = (PH 𝜆HM )∕PM

H
λMH λHR

λRH
λHM

M R

λRM λMR

Figure 15.3 Three-state model.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
15.6 Illustrative Examples 297

0.2 X 0.5
𝜆MH =
0.7
𝜆MH = 0.1428
PH 𝜆HR = PR 𝜆RH
Therefore
𝜆RH = (PH 𝜆HR )∕PR
0.2 × 0.3
𝜆RH =
0.1
𝜆RH = 0.6
PM 𝜆MR = PR 𝜆RM
Therefore
𝜆RM = (pM 𝜆MR )∕PR
0.7 X 0.4
𝜆RM =
0.1
𝜆RM = 2.8
Now
fH = PH (𝜆HM + 𝜆HR )
= 0.2 (0.5 + 0.3) = 0.16
fM = PM (𝜆MH + 𝜆MR )
= 0.7 (0.1428 + 0.4) = 0.37996
fR = PR (𝜆RH + 𝜆RM )
= 0.1 (0.6 + 2.8) = 0.34
Then mean time for H, M, and R states evaluated correspondingly:
p 0.2
MHT = H = = 1.25 h/yr
fH 0.16
p 0.7
MMT = M = = 1.84229 h/yr
fM 0.37996
p 0.1
MRT = R = = 0.29411 h/yr
fR 0.34
Availabilities on annual basis in H, M, and R states are obtained as:
APH = 8760 • pH = 8760 × 0.2 = 1752 h/yr
APM = 8760 • pM = 8760 × 0.7 = 6132 h/yr
APR = 8760 • pR = 8760 × 0.1 = 876 h/yr

15.11 The Capacity of distribution substation and load is normally distributed with
C = 100 MW
𝜎C = 10 MW
l = 95 MW
𝜎l = 10 MW.
Find the range of samples for distribution substation capactiy and load.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
298 15 Reliability Analysis of Composite Distribution System Using Frequency Duration Concept

Solution:
The distribution substation capacity sample range is given as follows:
C − 4𝜎C to C + 4𝜎C
Therefore range will be
((100 − 40), (100 + 40)) = (60,140) MW
The distribution substation load demand sample range is given as follows:
l − 4𝜎l to l + 4𝜎l
Therefore range will be ((95 − 40), (95 + 40)) = (55,135) MW.

15.7 Conclusions

This chapter presents a deterministic criterion-based reliability analysis of CDS for three-state
model with evaluation of probabilities in H, M, and R state using MCS. The frequency-duration
concept is used to assess frequencies of H, M, and R states along with duration indices in H, M, and
risky states e.g. MHT, MMT, and MRT using the simulation. The results show a reliability analysis
with DG to a sample composite distribution (CDS) system. Significant Improvement is observed in
the performance with the addition of DG capacity such as reduction in risky state probability and
mean risky state time, increase in healthy state probability and mean healthy state time based on
the frequency indices. Also, illustrative examples are discussed to understand the concept.

References

Ahmadi-Khatir, A., Fotuhi-Firuzabad, M., and Goel, L. (2009). Customer choice of reliability in
spinning reserve procurement and cost allocation using well-being analysis. Electric Power Systems
Research 79 (10): 1431–1440.
Amaral, T.S., Borges, C.L.T., and Rei, A.M. (2010). Composite system well-being evaluation based on
non-sequential Monte Carlo simulation. Electric Power Systems Research 80 (1): 37–45.
Arya, L.D. and Koshti, A. (2008). Distributed generation capacity reliability evaluation using safety
index. Journal of The Institution of Engineers (India) 89: 3–7.
Arya, L.D. and Koshti, A. (2012). Probabilistic simulation approach for distributed generation (DG)
capacity evaluation using artificial neural network representation of load duration curve. Journal of
The Institution of Engineers (India) 93: 57–62.
Arya, L.D., Koshti, A., and Choube, S.C. (2013). Frequency-duration analysis of composite distribution
system using a non-sequential Monte Carlo simulation. International Journal of Electrical Power &
Energy Systems 46: 17–25.
Billinton, R. and Allan, R.N. (1992). Reliability Evaluation of Engineering Systems, vol. 792. Springer.
Billinton, R. and Li, W. (1993). A system state transition sampling method for composite system
reliability evaluation. IEEE Transactions on Power Systems 8 (3): 761–770.
Billinton, R. and Lian, G. (1994). Composite power system health analysis using a security constrained
adequacy evaluation procedure. IEEE Transactions on Power Systems 9 (2): 936–941.
Costa, P.M. and Matos, M.A. (2009). Assessing the contribution of microgrids to the reliability of
distribution networks. Electric Power Systems Research 79 (2): 382–389.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 299

El-Khattam, W. and Salama, M.M.A. (2004). Distributed generation technologies, definitions and
benefits. Electric Power Systems Research 71 (2): 119–128.
Halve, S.S., Koshti, A., and Arya, R. (2023). A sampling method based on system state transition for
distribution system adequacy assessment using distributed generation. Journal of Operation and
Automation in Power Engineering 11 (4): 249–257.
Hamzeh, M. and Vahidi, B. (2020). Reliability evaluation of distribution transformers considering the
negative and positive effects of rooftop photovoltaics. IET Generation, Transmission & Distribution 14
(15): 3063–3069.
Hassan, R., Das, B.K., and Hasan, M. (2022). Integrated off-grid hybrid renewable energy system
optimization based on economic, environmental, and social indicators for sustainable development.
Energy 250: 123823.
Hegazy, Y.G., Salama, M.M.A., and Chikhani, A.Y. (2003). Adequacy assessment of distributed
generation systems using Monte Carlo simulation. IEEE Transactions on Power Systems 18 (1): 48–52.
Khezri, R. and Mahmoudi, A. (2020). Review on the state-of-the-art multi-objective optimisation of
hybrid standalone/grid-connected energy systems. IET Generation, Transmission & Distribution 14
(20): 4285–4300.
Koshti, A.S., Verma, A., Arya, R. et al. (2023). Tolerable random interruption duration based reliability
estimation of stand alone hybrid renewable energy system by network reduction and sequential
Monte Carlo simulation. IET Generation, Transmission & Distribution.
Kothari, D.P., Nagrath, I.J., and Saket, R.K. (2022). Modern Power System Analysis. Tata McGraw-Hill
Publishing Company.
Pereira, M.V.F. and Balu, N.J. (1992). Composite generation/transmission reliability evaluation.
Proceedings of the IEEE 80 (4): 470–491.
Probability Methods Subcommittee (1979). IEEE reliability test system. IEEE Transactions on Power
Apparatus and Systems 6: 2047–2054.
Ranjbar, H., Saber, H., and Sharifzadeh, M. (2022). Bi-level planning of distributed energy resources
into existing transmission grids: pathway to sustainable energy systems. IET Generation,
Transmission & Distribution 16 (24): 4963–4979.
Roy, P., He, J.B., Zhao, T., and Singh, Y.V. (2022). Recent advances of wind-solar hybrid renewable
energy systems for power generation: a review. IEEE Open Journal of the Industrial Electronics
Society 3: 81–104.
Wang, Y., Guo, C., and Wu, Q.H. (2013). A cross-entropy-based three-stage sequential importance
sampling for composite power system short-term reliability evaluation. IEEE Transactions on Power
Systems 28 (4): 4254–4263.
Wangdee, W. and Billinton, R. (2006). Bulk electric system well-being analysis using sequential Monte
Carlo simulation. IEEE Transactions on Power Systems 21 (1): 188–193.
Xu, Y. and Singh, C. (2012). Adequacy and economy analysis of distribution systems integrated with
electric energy storage and renewable energy resources. IEEE Transactions on Power Systems 27 (4):
2332–2341.
Yahaya, A.A., AlMuhaini, M., and Heydt, G.T. (2020). Optimal design of hybrid dg systems for
microgrid reliability enhancement. IET Generation, Transmission & Distribution 14 (5): 816–823.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
301

Reliability Analysis of Distribution Systems Integrated With

Renewable Energy Systems
Section 5
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
303

16

Reliability Assessment of Distribution Systems Integrated with

Renewable Energy Systems
Sachin Kumar 1 , Sandeep Kumar 2 , Aanchal Singh S. Vardhan 3 , R. K. Saket 4 , and
P. Sanjeevikumar 5
1
Department of Electrical Engineering, G.B. Pant Institute of Engineering and Technology, Pauri-Garhwal, Uttarakhand, India
2
Department of Electronics and Communication Engineering, G.B. Pant Institute of Engineering and Technology, Pauri-Garhwal,
Uttarakhand, India
3
Department of Electrical Engineering, National Institute of Technology, Jamshedpur, Jharkhand, India
4
Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
5
Department of Electrical Engineering, IT and Cybernetics, University of South-Eastern Norway, Porsgrunn, Norway

16.1 Introduction
Power system reliability (PSR) assessment is the process of evaluating the ability of an electrical
power system to deliver electricity to consumers without interruptions or disturbances. It involves
analyzing various factors that can affect the reliability of the system and identifying potential vul-
nerabilities or areas for improvement. Some key points to consider when conducting reliability
assessments for such integrated systems are as follows.
The first step in assessing the reliability of a distribution system integrated with renewable energy
systems is to model the renewable energy sources (RES)s accurately. This involves understand-
ing the characteristics of renewable resources, such as solar irradiance or wind speed, and their
variability over time. Various tools and techniques, such as weather data analysis and time series
modeling, can capture the intermittent nature of renewable energy generation (Kumar et al. 2020a).
Reliability assessment for integrated distribution systems should incorporate probabilistic analy-
sis methods. Traditional deterministic approaches may not capture the uncertainty associated with
renewable energy generation and its impact on system reliability. Probabilistic methods, such as
Monte Carlo simulation or stochastic optimization, can account for the variability and uncertainty
in renewable energy availability and their effects on system performance.
The configuration and design of the distribution system play a crucial role in its reliability. This
includes the layout of RESs, such as placing solar panels or wind turbines, and integrating energy
storage systems. Proper system design can help mitigate the impacts of renewable energy variability
and enhance system resilience (Kumar et al. 2020b).
Integrating renewable energy and distribution systems requires careful coordination and control
mechanisms. Grid integration strategies, such as advanced power electronics, demand response,
and smart grid technologies, can improve the reliability and stability of the overall system.
These strategies enable efficient utilization of renewable energy, effective load management, and
enhanced grid resiliency.
Several reliability metrics can be used to assess the performance of integrated distribution
systems. Typical metrics include the Expected Energy Not Supplied, Average Energy Not Supplied,
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
section of RAMPS illustrates the reliability analysis methodologies of modern distribution systems
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
304 16 Reliability Assessment of Distribution Systems Integrated with Renewable Energy Systems

System Average Interruption Duration Index, System Average Interruption Frequency Index, and
Average Service Availability Index, referred to as EENS, AENS, SAIDI, SAIFI, and ASAI, respec-
tively, which are the reliability indices taken into account for the EPDN reliability assessment.
These metrics quantify the frequency and duration of power interruptions and can be used to
evaluate the impact of renewable energy integration on system reliability.
Sensitivity analysis helps understand the influence of various factors on system reliability. Param-
eters such as renewable energy capacity, energy storage capacity, and demand patterns can signif-
icantly impact the reliability of integrated distribution systems. Sensitivity analysis identifies the
critical parameters and their impact on system performance, aiding decision-making and system
improvement (Kumar et al. 2021).
Reliability assessments should consider resiliency planning for integrated distribution systems.
This involves identifying vulnerabilities, developing contingency plans, and implementing mea-
sures to mitigate the impact of extreme events or system disturbances. Resiliency planning ensures
the system can quickly recover and restore normal operations in the face of disruptions.
Assessing the reliability of distribution systems integrated with renewable energy systems
requires a comprehensive approach that accounts for the intermittent nature of RESs and
their impact on system performance. Considering the factors mentioned above and employing
appropriate modeling and analysis techniques, evaluating and enhancing the reliability of such
integrated systems is possible.
Load analysis involves studying the patterns and characteristics of electrical demand within the
power system. By analyzing historical data and conducting load flow studies, the system opera-
tor can identify peak demand periods, load growth trends, and potential areas of congestion or
overload. This information helps design a reliable system to meet the expected demand.
Fault analysis examines the behavior of the power system during abnormal conditions such as
short circuits or equipment failures. It involves simulating and analyzing various fault scenarios to
determine their impact on system reliability. This analysis helps identify weak points in the system,
assess the adequacy of protection schemes, and optimize fault clearance strategies.
The adequacy assessment evaluates the capacity of the power system to meet the load demand
under normal operating conditions. It considers factors such as the availability of generation
resources, transmission capacity, and the reliability of essential equipment. Adequacy assessment
helps identify potential capacity shortages and determines the need for system upgrades or
additional resources.
PSR assessment often involves creating a detailed power system model, including generation
units, transmission lines, substations, and distribution networks. This model performs various
simulations and studies to assess the system’s reliability. Computer tools like power system anal-
ysis software (e.g. PSS/E, PSCAD, and DIgSILENT) are commonly used for system modeling and
analysis.
Risk assessment involves evaluating the potential risks and vulnerabilities that can impact the
reliability of the power system. This includes analyzing factors like extreme weather events, equip-
ment aging, cyber threats, and human errors. System operators can implement preventive measures
and develop contingency plans to minimize disruptions and improve system reliability by assessing
these risks.
Based on the findings of the reliability assessment, various measures can be implemented to
enhance PSR. These measures may include infrastructure upgrades, system reinforcement, equip-
ment maintenance and replacement strategies, grid modernization initiatives, and the implemen-
tation of advanced monitoring and control systems. PSR assessment is ongoing as power systems
evolve and new challenges arise. It helps system operators and planners make informed decisions
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
16.2 Reliability Functions 305

to ensure a reliable and resilient consumer electricity supply. For reliability assessment, in continu-
ation, the chapter deals with essential reliability functions, various RESs, optimization, and control
strategies, and some practical case studies with challenges and future directions. The chapter’s flow
can be easily depicted in Figure 16.1.

Chapter begins
16.2 Reliability Functions

By analyzing the reliability or survival function (Re(t)), cumulative dis- Literature on reliability
tribution function (fc u(t)), failure density function (ff (t)), and hazard assessment of renewable
rate function (Hz(t)), one could evaluate the reliability of a system. The integrated electrical
distribution systems
reliability of a component, or the likelihood that a system or piece of
equipment won’t malfunction within a certain amount of time (0, t),
is a method for describing a component’s failure probability. This rela-
Some basics on
tionship has been expressed using Eq. (16.1):
distribution system’s
Re(t) = Probability Pr {T ≥ t} (16.1) reliability

The fcu (t) is defined in Eq. (16.2), which is the failure probability that
occurs before time “t”:
Glimpse on crucial
fcu (t) = 1 − Re(t) = Pr {T < t} (16.2) renewable energy sources
The derivative of fc (t) is termed ff (t), as given in Eq. (16.3), and the area
under the entire f (t) is calculated to be 1:
dfc (t) dRe(t) Study on optimization
ff (t) = =− (16.3)
dt dt and control related
strategies
Equation (16.4) states that R(t) depends on the term failure rate func-
tion or hazard rate function (H(t)) and that a higher H(t) will result in
a lower Re(t):
t
Re(t) = e− ∫0 Hz(t)dt (16.4) Highlights on case studies
performed in the literature
where
Re(t) − Re(t + Δt)
● Hz(t) = Pr{t ≤ T ≤ t + Δt|T ≥ t} =
Re(t)
Challenges and future
directions
ff (t) dRe(t) 1
Hz(t) = =− × (16.5)
Re(t) dt Re(t)
From Eq. (16.2), Infant mortality is an example of an early life fail- Chapter ends
ure, and Hz(t) describes these failures as well as wear-out failures, in
which the failure rate significantly rises. A bathtub curve may be used Figure 16.1 The flow of
to demonstrate this. Hz(t) is a different method of expressing a failure the chapter.
distribution. When Hz(t) is rising, decreasing, or constant, failure rates
may be described as increasing, decreasing, or constant.
To simulate equipment lifetime, any distribution may be utilized. In reality, distribution functions
with monotonic hazard functions seem to be the most realistic, and a handful within that class
are widely considered to offer the most plausible system reliability models. It is important to note
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
306 16 Reliability Assessment of Distribution Systems Integrated with Renewable Energy Systems

that the following distributions can be used to obtain the life distribution of an electrical power
system or component: Binomial, Exponential, Poisson, Geometric, normal, Gamma, and Weibull
Distributions.
The basic parameter used in the reliability evaluation is ‘forced outage’ at some distant time in
the future. This probability parameter is defined as the system or component unavailability. It is
known as the forced outage rate in power systems, as mentioned in Eqs. (16.6) and (16.7):

𝜆f
Unavailability = (16.6)
𝜆f + 𝜇r
𝜇r
Availability = (16.7)
𝜆f + 𝜇r

where
1
● 𝜆f = expected failure rate =
Mean-Time-To-Failure (MTTF)
1
● 𝜇r = expected repair rate =
Mean-Time-To-Repair (MTTR)

The aforementioned reliability parameters operate as reliability indices. At load point “L,” the
reliability parameters can be estimated.
The reliability assessment of a distribution network is as crucial as other components of the
EPDN. The IEEE guide for EPDN reliability is based on the IEEE guide, referenced in standard
1366-2012. The reliability of an EPDN can be evaluated using a few reliability indices under the
provided specification. According to Equations, these indexes are divided into two categories
(16.8)–(16.14):
∑
EENS = [(load at Lth load) × (Annual Dl at Lth load)] (16.8)
∑
(EENS at Lth load)
AENS = (16.9)
Total no. of customers at all loads
∑
[(annual DL at Lth load) × (No. of customers at Lth load)]
SAIDI = (16.10)
Total no. of customers at all loads
∑
[(average 𝜆f at Lth load) × (No. of customers at Lth load)]
SAIFI = (16.11)
Total no. of customers at all loads
SAIDI
ASAI = 1 − (16.12)
8760
∑nL
U L NL
CAIDI = ∑L=1
nL hour/customer/interruption (16.13)
L=1
𝜆f N L

ASUI = 1 − ASAI p.u. (16.14)

where

● 𝜆f = failure/year
● UL = hour/year
● DL = repair rate in hour
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
16.3 Renewable Energy Sources 307

16.3 Renewable Energy Sources

Renewable energy is derived from naturally replenished sources and can be sustained over time
without depleting finite resources. These energy sources are considered environmentally friendly
because they impact the planet less than fossil fuels can contribute to climate change by releasing
greenhouse gases. RESs include the following sources.

16.3.1 Solar Energy

This is generated by capturing sunlight using photovoltaic cells (solar panels) to convert it directly
into electricity or using solar thermal systems to produce heat for various applications. IEEE
1562-2021 recommends practice for sizing stand-alone photovoltaic (PV) systems. Solar PV
systems convert sunlight directly into electricity using semiconductor materials in solar cells, as
demonstrated in Figure 16.2.
The basic operating principle involves the creation of electron-hole pairs when photons from
sunlight strike the semiconductor material. These electron-hole pairs generate an electric current,
which can be harnessed for various applications.
The fundamental equation that describes the current–voltage relationship in a solar cell is the
diode equation, which combines the current components due to the light-generated current and
the diode-like behavior of the semiconductor material, as depicted in Figure 16.3 and the related
equation is given as:
( ( ) )
V
I = IL − ID exp −1 (16.15)
VT
where
● I = the solar cell output current.
● IL = the light-generated current ∝ intensity of incident light.
● ID = the diode current, dependent on the voltage across the solar cell.
● V = the voltage across the solar cell terminals.
● VT = the thermal voltage, a characteristic temperature-related voltage.
An equivalent circuit can represent a PV cell and model its behavior under different conditions.
The equivalent circuit is an electrical representation that simplifies the complex physical processes

Sunlight

Front contact
Anti reflecting
coating

Specially treated
semi-conductor Back contact
material

Figure 16.2 PV cell: basic operation.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
308 16 Reliability Assessment of Distribution Systems Integrated with Renewable Energy Systems

Sunlight Figure 16.3 PV effect: a schematic diagram.

n-type
p-n junction
p-type

Solar Photons
panel
Electron flow

Hole flow

occurring within the PV cell. However, the most popular comparable circuit model for a PV cell is
the one-diode model; double- and triple-diode models have also been implemented.
By multiplying the current and voltage at each point along the curve, one can determine the
power output of a solar cell, also known as the power-voltage (P − V) curve:

P =I×V (16.16)

According to Figure 16.4, the P − V curve of a solar photovoltaic (PV) module represents the
relationship between the power output and the voltage across the device’s terminals. This curve is
essential for understanding the behavior and performance of a solar cell or module under different
operating conditions.
The shape of the P − V curve can vary based on factors such as temperature, shading, and the
characteristics of the solar cell or module. In most cases, the curve has a characteristic “hump”
shape, with the MPP located on the descending side of the curve, typically at around 70–80% of the
open-circuit voltage.
Solar PV systems are designed to ensure that they operate near the MPP under varying conditions
(e.g. changing sunlight intensity and temperature) to maximize energy production and efficiency.
MPP Tracking algorithms in solar inverters or charge controllers often achieve this.
Engineers use more complex models that consider factors such as semiconductor material prop-
erties, shading effects, and system efficiency for detailed analysis and design of solar PV systems.
In addition, reliability and availability are crucial factors in designing, operating, and maintaining

80 T = 25°C
Pm
60 2
/cm T = 50°C
Power (W)

mW T = 25°C
40 100 2
/cm
60 m
W T = 50°C
20 T = 25°C
m2
20 mW/c
T = 50°C
0
14 16 18 Vm 22
Panel voltage (V)

Figure 16.4 Power-voltage curve of a solar PV.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
16.3 Renewable Energy Sources 309

grid-connected solar PV systems. These systems aim to generate renewable energy consistently and
contribute to the overall energy supply.

16.3.2 Wind Energy

Wind turbines capture the kinetic energy of moving air and turn it into electricity. When exposed to
the wind, turbine blades rotate, producing wind power. The IEEE 1834-2019 standard is available
for wind turbine rotor systems.
A wind energy conversion system (WECS) is designed to harness the wind’s kinetic energy and
convert it into electrical energy. Several components and technologies are involved in a typical
WECS: wind turbine, generator, transmission system, power electronics, yaw and pitch control
systems, tower and foundations, nacelle, anemometer, wind vane, and monitoring and control sys-
tems. A typical DFIG-based WECS is illustrated in Figure 16.5.
The air density (kg/m3 ), the WT rotor radius (m), the wind speed (m/s), and the power coefficient
all affect the mechanical and electrical power outputs. The coefficient, which depends on the blade
pitch angle and tip speed ratio, is calculated as the WT power’s ratio to the wind stream’s power, as
represented by Eq. (16.17):
⎧0; 0 ≤ V ≤ Vc,in
⎪
⎪ or
⎪ V ≥ Vc,out
PW =⎨ ( ) (16.17)
V − Vc,in
⎪PW,r × ; Vc,in ≤ V ≤ Vr
⎪ Vr − Vc,in
⎪P ; Vr ≤ V ≤ Vc,out
⎩ W,r
where PW = output power, PW,r = rated output power, Vr = rated wind speed, Vc,in , and V c,out =
cut-in and cut-out speeds, respectively.
Wind energy conversion systems come in various sizes, from small residential turbines to large
utility-scale wind farms with multiple turbines. The technology has advanced significantly over
the years, improving efficiency, reducing costs, and increasing reliability, making wind energy an
increasingly important contributor to the global energy mix.
A WECS, also commonly referred to as a wind turbine or wind power system, is a technology that
converts the wind’s kinetic energy into electrical energy. Wind turbines are a critical component

ω r > ωs Ps Machine
ωr < ω s terminal
Wind Main power
turbine Pr Pr Circuit system
ωr < ω s ωr > ωs
breaker

Stator
Slip
ring Rotor Grid
DC link
Wind

Gear side side

Rotor
box converter converter
DC to AC AC to DC Trans.
Stator

Control system
CLOE PLOE

Figure 16.5 DFIG-based WECS structure. ⇒𝜔r and 𝜔s are the rotor and synchronous speeds, Pr and Ps are
rotor and stator active powers. CLOE and PLOE are the complete and partial loss of excitations.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
310 16 Reliability Assessment of Distribution Systems Integrated with Renewable Energy Systems

of WECS and are used to capture the energy from the wind and transform it into a usable form of
electricity. Here are the key components and aspects of a typical Wind Energy Conversion System:
1. Wind Turbine: The wind turbine is the primary component capturing wind energy. There
are two main types of wind turbines: horizontal-axis wind turbines (HAWT) and vertical-axis
wind turbines (VAWT). HAWTs are the most common and resemble traditional windmills,
while VAWTs have a vertical rotor axis and are less common.
2. Rotor Blades: In HAWTs, the rotor blades are attached to the rotor hub and capture the wind’s
kinetic energy. The blades, such as fiberglass or composite materials, are typically lightweight
and durable.
3. Rotor Hub: The rotor hub connects the blades to the main shaft, allowing the blades to rotate.
It also houses various control systems and sensors.
4. Main Shaft: The main shaft is connected to the rotor hub and transfers the rotational energy
from the blades to the generator.
5. Generator: The generator converts mechanical energy from the rotating shaft into electrical
energy. Most wind turbines use synchronous, asynchronous, or permanent magnet generators.
6. Tower: The tower supports the wind turbine and elevates it to a height where it can capture
stronger and more consistent winds. Taller towers are often used to access higher wind speeds
at greater heights.
7. Nacelle: The nacelle is a protective housing that encloses the generator and other critical
components. It is usually mounted on the tower and contains control systems, gearboxes (if
needed), and other necessary equipment.
8. Anemometer and Wind Vane: These instruments are used to measure wind speed and direc-
tion to optimize the orientation of the wind turbine blades for maximum energy capture.
9. Control System: A control system regulates the wind turbine’s operation. It adjusts the pitch
angle of the blades and the rotor speed to ensure efficient energy capture and protect the turbine
in varying wind conditions.
10. Power Electronics: Power electronics components, such as inverters and converters, convert
the generator’s variable voltage and frequency output into grid-compatible electricity.
11. Grid Connection: The electricity generated by the wind turbine is typically connected to the
electrical grid. This allows for the distribution of wind-generated power to homes, businesses,
and industries.
12. Energy Storage (Optional): Some WECS may include energy storage systems, such as bat-
teries, to store excess energy when wind conditions are favorable and release it when the wind
is not blowing.
Further, increasing the availability and reliability of wind turbine generator systems is crucial for
optimizing their performance and reducing downtime. Here are some strategies and approaches
that can help achieve this goal: regular maintenance and inspections, condition monitoring, remote
monitoring and control, redundancy and backup systems, lifecycle planning, grid integration, and
energy storage (Kumar et al. 2021).

16.3.3 Vehicle Charging-Discharging

Integrating electric vehicle (EV) charging and discharging into the distribution system can enhance
its reliability, but it also introduces new challenges that must be carefully managed. Here are some
ways in which EV charging and discharging, as represented in Figures 16.6 and 16.7, respectively,
can impact distribution system reliability (DSR).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
16.3 Renewable Energy Sources 311

Figure 16.6 EV charging structure.

EV
Charging
station

Low voltage network

Power
conversion
Grid Transformer system

EV Home

Charging/discharging power flow

for V2H, V2G, V2D, and G2V

⇒ V2H - vehicle to home, V2G - vehicle to grid, V2D - vehicle to

distribution, G2V - grid to vehicle.

Figure 16.7 Strategies related to of EV-to-everything. ⇒ V2H — vehicle to home, V2G – vehicle to grid,
V2D — vehicle to distribution, G2V — grid to vehicle.

1. Demand Response and Peak Shaving: EVs can be used for demand response, which involves
adjusting their charging or discharging patterns to help balance electricity demand with supply.
During times of high demand, EV charging can be temporarily reduced or shifted to off-peak
hours, alleviating stress on the distribution grid and preventing grid overload during peak
periods.
2. Energy Storage: Some EVs have substantial battery capacities, particularly electric buses and
larger vehicles. These batteries can be utilized for energy storage, absorbing excess energy from
the grid during periods of low demand and injecting it back into the grid during peak demand,
effectively acting as distributed energy storage systems.
3. Grid Ancillary Services: EVs can provide ancillary services to the grid, such as frequency reg-
ulation and voltage support. These services help maintain the stability and reliability of the grid
by quickly responding to fluctuations in supply and demand.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
312 16 Reliability Assessment of Distribution Systems Integrated with Renewable Energy Systems

4. Voltage Regulation: Strategically placing EV charging infrastructure across the distribution

network can help regulate voltage levels, ensuring that they remain within acceptable limits
and minimizing voltage fluctuations.
5. Islanded Operation: In the case of a localized power outage, EVs with bidirectional charg-
ing capability can potentially power critical infrastructure or homes during islanded operation,
enhancing resiliency and reliability.
Some primary EV charging and discharging equations are depicted in Eqs. (16.18)–(16.22). The
modeling and other descriptions can be observed as follows.
1. Charging Power (Pc ): The power at which an EV is being charged is given by:
Pc = Vc × Ic (16.18)
where
● P = charging power (W)
c
● Vc = charging voltage (V)

● I = charging current (A)

c
2. Battery Charging Time (Tc ): The time it takes to charge an EV’s battery is calculated using:
Tc = Ecap ∕Pc (16.19)
where
● Tc = charging time (h)

● E
cap = energy capacity of the battery (Wh)
3. State of Charge (SoC): The State of Charge represents the current energy level of the battery
as a percentage. It can be calculated using:
SoC = (Ecur ∕Em ) × 100 (16.20)
where
● Ecur = current energy level of the battery (Wh)
● E
m = maximum energy capacity of the battery (Wh)
4. Discharging Power (Pdis ): The power drawn from the battery during discharging is given by
the same equation as the charging power:
Pdis = Vdis × Idis (16.21)
where
● P
dis = discharging power (W)
● V
dis = discharging voltage (V)
● I
dis = discharging current (A)
5. Battery Discharging Time (Tdis ): The time for which the battery can discharge at a certain
power before reaching empty is calculated using:
Tdis = Ecel ∕Pdis (16.22)
where
● T
dis = discharging time (h)
● Ecel = current energy level of the battery (Wh)

These equations explain power, energy, and time relationships during EV charging and discharg-
ing. In practice, efficiency losses, charging profiles, and other factors might affect the actual charg-
ing and discharging times. Modern charging systems might incorporate advanced features like fast
charging, variable charging rates, and smart charging algorithms.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
16.4 Optimization and Control 313

However, there are challenges associated with integrating EVs into the distribution system as
well (Kumar et al. 2020a):

1. Load Management: Sudden and widespread EV charging during peak hours can lead to local-
ized grid congestion and overloads. Proper load management strategies are needed to ensure
that charging is distributed intelligently across time and locations.
2. Infrastructure Upgrades: Integrating many EVs might necessitate infrastructure upgrades
in some areas to accommodate increased demand. This can be expensive and requires careful
planning.
3. Bidirectional Charging: While bidirectional charging (vehicle-to-grid, or V2G) can offer ben-
efits, it requires sophisticated control systems to manage the bidirectional power flow safely and
effectively.
4. Communication and Control: To coordinate EV charging and discharging to benefit the grid
while minimizing interruptions to car owners, reliable communication and control systems are
also essential.
5. User Behavior: Predicting and influencing user behavior regarding EV charging patterns can
be complex, and incentives may be required to encourage the desired behavior.

In summary, while integrating EVs into the distribution system can enhance reliability and pro-
vide grid support, careful planning, advanced technologies, and effective policies are necessary to
manage the associated challenges and optimize the benefits.

16.3.4 Other Renewable Energy Sources

It is mentioned in the literature that several RESs, such as hydropower, biomass, geothermal, and
ocean energy, have the potential to significantly contribute to our global energy needs while mini-
mizing environmental impact. These RESs offer several advantages, including reduced greenhouse
gas emissions, minimal air pollution, long-term sustainability, and improved PSR. However, each
source also has its own set of challenges.
The critical advantage of RESs is that they are sustainable and do not produce harmful emissions
during energy generation. They help mitigate climate change, reduce greenhouse gas emissions,
and rely less on limited fossil resources. However, challenges like intermittent availability (solar
and wind) and the need for advanced technology and infrastructure remain to be addressed for the
broader adoption of these energy sources. Simultaneously, dealing with these uncertain RESs is
crucial by implementing various optimization and control techniques.

16.4 Optimization and Control

Optimization and control techniques enhance DSR by effectively managing grid operations, miti-
gating faults, and minimizing downtime. Optimizing DSR involves employing various techniques
and methodologies to improve the stability and resilience of the electrical grid (Kumar et al. 2020b).
Here are some optimization techniques specifically focused on enhancing DSR:

1. Reliability-Centered Maintenance (RCM): RCM is a methodical technique that identifies

crucial assets, evaluates potential failure modes, and chooses the most efficient maintenance
strategies. Utilities can minimize downtime and reduce the risk of outages by prioritizing main-
tenance efforts on the most critical components.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
314 16 Reliability Assessment of Distribution Systems Integrated with Renewable Energy Systems

2. Volt-VAR Optimization (VVO): VVO adjusts voltage levels and reactive power flows to
optimize system performance and minimize energy losses. This technique maintains voltage
within desired limits, improving equipment efficiency and reducing the risk of voltage-related
outages.
3. Fault Location, Isolation, and Service Restoration (FLISR): FLISR techniques use
real-time data and automation to quickly detect faults, isolate affected areas, and restore
power to unaffected zones. These systems minimize outage duration and improve service
reliability.
4. Optimal Placement of Distributed Energy Resources: Using optimization algorithms,
determine the best locations for integrating RESs, energy storage systems, and microgrids.
This approach can enhance grid resilience by reducing dependency on centralized generation
and improving local supply during outages.
Advanced control techniques manage the operation of DERs, such as solar panels, wind tur-
bines, and energy storage systems. These techniques optimize the utilization of renewable
resources and improve grid resilience.
5. Load Balancing and Load Shedding: Employ optimization algorithms to balance loads
across distribution feeders and substations. In overloads or emergencies, load-shedding
algorithms can identify and disconnect non-critical loads to maintain grid stability.
6. Demand Response Optimization: Develop algorithms that optimize demand response pro-
grams, considering customer preferences, load profiles, and grid constraints. This ensures effi-
cient load reduction during peak periods.
7. Microgrid Operation Optimization: Microgrid optimization techniques maximize the use
of local resources (generation, storage, and demand) to maintain critical loads during grid out-
ages. These algorithms balance supply and demand within the microgrid.
8. Stochastic Optimization: Stochastic optimization accounts for uncertainties in load fore-
casts, equipment failures, and renewable energy generation. It helps utilities make robust deci-
sions in the face of varying conditions.
9. Dynamic Voltage and Reactive Power Optimization: These techniques adjust real-time
voltage levels and reactive power settings to maintain optimal voltage profiles and minimize
energy losses. Proper voltage control enhances equipment lifespan and reduces the risk of
voltage-related outages.
10. Risk Assessment and Mitigation: Use risk assessment techniques to identify vulnerabilities
in the distribution system and prioritize mitigation strategies. This can involve identifying weak
points, single points of failure, and potential sources of disruption.
11. Data Analytics and Predictive Maintenance: Implement machine learning and data ana-
lytics to predict equipment failures based on historical data. Predictive maintenance can sched-
ule interventions before issues lead to outages.
12. Optimal Switching and Reconfiguration: Algorithms for optimal switching and reconfig-
uration help reroute power during faults or outages, minimizing the affected customers and
reducing the duration of outages.
13. Wide-Area Monitoring and Control (WAMC): WAMC systems use synchronized data from
multiple points in the distribution network to make real-time decisions that improve grid sta-
bility and reliability.
14. Scenario Analysis and Contingency Planning: Conduct scenario analyses to model various
failure scenarios and their impacts on the distribution system. Develop contingency plans that
outline actions to be taken for specific events.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
16.5 Case Study 315

15. Communication and Control Networks: Implement robust communication systems to

ensure quick and reliable information exchange between control centers, substations, and
field devices.
16. Cybersecurity Measures: Utilize control techniques to secure communication networks and
protect control systems from cyber threats that could compromise grid reliability.

The effectiveness of these techniques often depends on integrating them into an overarching
control and optimization framework that considers the dynamic nature of distribution networks
and the diverse range of operating conditions. This approach enables utilities to manage, adapt,
and respond to various challenges while maintaining a reliable distribution system.
Each optimization technique requires a combination of advanced data analytics, modeling, simu-
lation, and control strategies. The choice of technique depends on the specific goals and challenges
of the distribution system, and a holistic approach often involves integrating multiple techniques
for comprehensive reliability improvement.

16.5 Case Study

Any nation’s ability to flourish economically and socially depends on the reliability of its power
distribution infrastructure. Industrial, commercial, residential, and other institutions providing
services are expanding quickly. All of these organizations require reliable power sources. As a
result, utilities must make a solid effort to guarantee that customers receive a reliable power sup-
ply. However, the essential issues with any power distribution system are higher power interruption
frequency and longer interruption duration, which impact customers’ everyday lives.
Indeed, let’s delve into a case study demonstrating how DSR can be enhanced using various
strategies. A suburban region with a growing population has been experiencing frequent power
outages due to aging infrastructure, adverse weather conditions, increasing electricity demand,
etc. The local utility company aims to improve DSR to ensure consistent and high-quality power
supply to its customers. Therefore, the following objectives must be guaranteed while assessing the
reliability improvement of the power distribution system.

1. Minimize the frequency and duration of power outages.

2. Improve response and restoration times after outages.
3. Enhance the overall resilience of the distribution network.

Overall, the Roy Billinton Test System (RBTS) and IEEE Reliability Test System (RTS) signifi-
cantly advance the understanding of PSR and develop and test new reliability assessment methods.
Both are widely used in academic research, industrial applications, and educational settings within
the power engineering community. Some case studies have been tabulated in Table 16.1 and dis-
cussed in Sections 16.5.1–16.5.3.

16.5.1 IEEE Reliability Test System-79

The IEEE RTS is another commonly used benchmark system for studying the reliability and per-
formance of power systems. The IEEE RTS provides a standardized platform for researchers and
practitioners to evaluate the effectiveness of various reliability analysis methods and techniques.
The IEEE RTS is a simplified power system network representation consisting of generators,
transmission lines, loads, and other components. The IEEE Reliability Test System Task Force
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
316 16 Reliability Assessment of Distribution Systems Integrated with Renewable Energy Systems

Table 16.1 Case studies for assessing the various distribution systems’ reliability.

S. No. References Case study

1 Ranjbar et al. (2022) Practical power system – installed capacity =

40,270 MW, maximum load = 36,872 MW, and
installed capacity of RESs = 3010 MW with ratio of
WT to PV = 2:3
2 Taheri Otaghsara and Asghari IEEE RTS 24 buses
Gharakheili (2022)
3 Alizadeh et al. (2022) RBTS4 and Finnish Distribution System
4 Karimi et al. (2021) 34 feeders in Ahvaz city, Iran
5 Sinishaw et al. (2021) Bata 34 bus feeder of Bahir Dar distribution system,
Ethiopia
6 Afzal et al. (2020) IEEE 69-bus system
7 Bagen et al. (2019) 29-buses actual utility substation in the Province of
Manitoba, Canada
8 Zare et al. (2018) IEEE 69-bus system
9 Chen et al. (2016) RBTS and IEEE 15-bus system
10 Benidris and Mitra (2015) IEEE RTS 24 buses

created it to make testing and contrasting various reliability assessment approaches easier. The
system has undergone several updates and versions to reflect power system technology and
practice changes.
It’s worth mentioning that the IEEE RTS also comes in various versions, as described later in this
section, each representing different levels of complexity and realism.
The application of probability methods (APM) Subcommittee of the Power System Engineer-
ing Committee created and released the initial IEEE Reliability Test System (RTS-79) in 1979. To
meet the need for a uniform model to test and compare outcomes from various PSR evaluation
approaches, RTS-79 was created. The model was created primarily to study the effects of transmis-
sion and generation outages on the system and is represented in Figure 16.8.
In 1986, the RTS-79 noticed a minor update (RTS-86). The improvement of the information about
the creation of feet received special consideration in this version. Unit derates (partial outages), load
forecast uncertainty and unit scheduled maintenance are essential features incorporated in this
release. The model was increased to allow users to examine multi-area interconnection effects and
improvements relating to reliability indices, including loss of load probability, loss of load expecta-
tion, and loss of energy expectation. The RTS-86 connected two identical RTS-79 versions using a
300 MW-rated 230 kV tie line. The RTS-86 also provides precise system reliability indices for bench-
marking and case study purposes.
The RTS received its most recent upgrade in 1996 (RTS-96). This update significantly increased
the size and complexity of the model; the RTS-96 is made up of three associated RTS-79 models. The
industry at the time saw a sharp rise in interregional transactions and advancements in computing
capacity. Thus, a tractable analysis of larger multi-area models might be performed. The most sig-
nificant change to the RTS-96 was adding information on operating costs and restrictions for the
generating units: unit start-up heat input, net plant incremental heat rates, unit cycling restrictions,
ramping rates, and unit emissions data.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
16.5 Case Study 317

BUS 18
BUS 17
BUS 21
BUS 22
BUS 23

BUS 16

BUS 19 BUS 20

BUS 14
BUS 15 Synchronous
condenser
230kV BUS 13

BUS 24
BUS 11 BUS 12

BUS 10
BUS 3 BUS 6
BUS 9 Cable

138kV BUS 4
BUS 5

BUS 8

Cable

BUS 1 BUS 2 BUS 7

Figure 16.8 Reliability case study structure IEEE RTS-79. Source: Benidris and Mitra (2015)/The Institution
of Engineering and Technology.

The IEEE RTS-79 system is a classical PSR analysis case consisting of 24 buses, 32 generating
units, and 42 transmission lines. Four more lines expand the original system. Figure 16.8 depicts
the redesigned network topology of the IEEE RTS-79 system. The MTTF and MTTR of units for
various capacities are provided via the modified IEEE RTS-79 system. In this scenario, the settings
are slightly changed to improve the reliability of the larger-capacity units. It’s a widely used case
study system for evaluating the reliability and performance of power systems.
It’s important to note that while the IEEE RTS 24-bus version provides a more detailed represen-
tation than smaller versions, it’s still a simplified power system model. This allows for controlled
testing and comparison of methods while avoiding the complexity of real-world networks.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
318 16 Reliability Assessment of Distribution Systems Integrated with Renewable Energy Systems

Researchers and engineers have widely adopted the IEEE RTS-79 version in the power industry
to develop and validate new reliability assessment techniques and explore the impact of various
factors on PSR.
New sensitivity criteria that can be helpful for site selection are discussed. The technique may
integrate the available DGs by determining the optimal sites in a power system.

16.5.2 Roy Billinton Test System

The RBTS is a widely used case study for evaluating power systems’ reliability and risk assessment.
Roy Billinton and his colleagues developed it to provide a standardized platform for researchers
and practitioners to test and compare reliability assessment methods and techniques, represented
as Figure 16.9. The RBTS consists of a simplified representation of a power system network, includ-
ing generators, transmission lines, and loads. It’s designed to represent various aspects of power
system behavior, such as generation, transmission, and distribution, which are crucial for assess-
ing reliability and risk. The RBTS includes different scenarios with varying levels of complexity to
determine the performance of other reliability analysis techniques.
Using a Monte Carlo simulation-based method, examined the effects of the stochastic nature
of distributed and renewable resources, including solar PV, battery energy storage, and electric
vehicles, on the reliability of distribution networks. The results are illustrated, and customer-side

1×40 MW
4×20 MW G
2×5 MW

Bus 2
2×40 MW
1×20 MW G
1×10 MW L3
20 MW + j9.6 MVar
L2 L7
Bus 1

L1 L6 L4

Bus 3 Bus 4

L5 L8

85 MW + j41.2 MVar 40 MW + j9.4 MVar

Bus 5

L9
20 MW + j9.6 MVar
Bus 6

20 MW + j9.6 MVar

Figure 16.9 Reliability case study structure RBTS. Source: Adapted from Chen et al. (2016).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
16.5 Case Study 319

dependability indices, such as SAIDI, SAIFI, and CAIDI, are computed using a modified RBTS,
considering capacity considerations.

16.5.3 Bahir Dar’s Power Distribution Test System

Substation-1 66/45/15 kV and Substation-2, including 230/66/15 kV and 230/132/15 kV substa-
tions, supply the electrical loads for Bahir Dar city. The electricity supply system for Bahir Dar
Town includes the Bata, Ghion, Papyearus, Air Force, Sematate, and Industry feeders. The chapter
focuses on the Bata feeder, which has a more significant yearly interruption duration and annual
interruption frequency. A single-line diagram of the Bata feeder distribution network is displayed
in Figure 16.10. The figure shows that the network contains 34 buses and is configured in a radial
pattern. Node 1 is the slack bus out of the entire number of buses. The final four buses or nodes
are common coupling nodes, while the remaining 29 are coupled to loads through step-down
distribution transformers. According to the single-line diagram of the Bata feeder, the distribution
line nodes, which correspond to the numbers up to 34, are represented in dark gray.
In contrast, the sectional lines of the feeder, which correspond to the numbers up to 33, are rep-
resented by black color. The operational glitches are at their peak and need to be reduced. The Bata
feeder’s distribution system is modeled, and the system’s reliability indicators are computed.
Using smart grid technology, the reliability is improved. In recent years, generation and trans-
mission networks have principally conceived about and implemented smart grid applications.
Three smart grid technology applications enhance the distribution system’s reliability, although
the conceptual framework of smart grid technology is quite broad. These technologies include
quick power restoration, effective network reconfiguration, and efficient switching device location.

230kV Bahir Dar Substation with Bata feeder.

• Nodes in dark gray numbers 34
• Load 33
33
• Sectionalizing Lines in black numbers
32
32

27 26 25 24 28 29 31
23
26 25 24 23 27 28
22 29
22
18 17 16 15 14
21 30
30
17 16 15 14 13 21 31
20
11 13 20
10 19
12 12
10 19
18
1 2 3 94 5 6 117 8 9

1 2 3 4 5 6 7 8

Figure 16.10 34-bus Bahir Dar’s Bata feeder. Source: Sinishaw et al. (2021)/with permission of Elsevier.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
320 16 Reliability Assessment of Distribution Systems Integrated with Renewable Energy Systems

Rapid power restoration reduces the annual frequency and length of disruptions for consumers
by placing switching devices optimally and reconfiguring the network optimally. So, improving
Ethiopia’s electricity distribution systems’ reliability through smart grid technology is preferable.
A Bata feeder from the Bahir Dar power distribution system with 34 buses is considered to
analyze and evaluate the applicability of smart grid technology. The duration and frequency
of interruption statistics from Ethiopian Electric Utility were gathered and examined during
2017–2019. The Bata feeder experiences 712.43 hours of average interruption each year, with an
average frequency of 604 interruptions per year, according to the interruption data studied. The reli-
ability indices derived are SAIFI = 131.15 int/yr, SAIDI = 245.84 h/yr, CAIDI = 1.88 h/cust.int, and
ASAI = 0.972 pu.

16.6 Challenges and Future Directions

Ensuring DSR in the presence of RESs presents challenges and opportunities for the energy indus-
try. Here are some of the key challenges and future directions in this area.

16.6.1 Challenges

1. Intermittency and Variability: RESs like solar and wind are intermittent and variable.
This can lead to fluctuations in power generation, making it challenging to maintain a stable
supply-demand balance in the distribution grid.
2. Grid Stability: The intermittent nature of RESs can affect grid stability by causing voltage
and frequency fluctuations. Sudden changes in power output can lead to voltage violations and
potential blackouts if not appropriately managed.
3. Integration and Penetration Limits: Integrating a high penetration of renewable energy into
the distribution system requires careful planning to avoid overloading the grid. The existing
infrastructure might not be designed to handle the bidirectional power flow and variable gener-
ation associated with RES.
4. Limited Control: Operators have limited control over renewable energy generation, unlike tra-
ditional power plants. This lack of control can make managing and optimizing the distribution
system’s performance challenging.
5. Storage and Energy Management: Effective energy storage solutions are essential to store
excess renewable energy during periods of high generation and release it during periods of low
generation. Developing cost-effective and efficient storage technologies is crucial.
6. Grid Planning and Resilience: Integrating renewable sources requires rethinking grid plan-
ning and resilience strategies. The intermittent nature of these sources makes traditional plan-
ning approaches less effective, necessitating new methods to ensure reliable operation.

16.6.2 Future Directions

1. Advanced Forecasting: Improving forecasting models for renewable energy generation can
help grid operators anticipate fluctuations and proactively balance supply and demand.
2. Energy Storage Development: Research and development of better energy storage technolo-
gies, such as advanced batteries, compressed air energy storage, and pumped hydro storage, can
efficiently utilize surplus renewable energy.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Problems 321

3. Demand Response and Flexibility: Implementing demand response programs and flexible
load management can help match energy consumption with intermittent renewable energy
generation patterns.
4. Microgrids and Decentralization: Creating microgrids and decentralized energy systems
can enhance reliability by allowing localized generation, consumption, and storage, reducing
the dependency on central grids.
5. Smart Grid Solutions: Integrating smart grid technologies, including advanced sensors, com-
munication networks, and automation, can enable real-time monitoring and control of the
distribution system, improving its resilience.
6. Hybrid Energy Systems: Combining multiple RESs (solar, wind, hydro, etc.) can mitigate
intermittency, providing a more consistent power output.
7. Grid-Forming Inverters: Developing grid-forming inverter technologies that allow RESs to
contribute to grid stability, even during grid disturbances, can enhance system resilience.
8. Policy and Regulation: Governments and regulatory bodies need to create supportive policies
that encourage the integration of RESs and provide incentives for investing in grid reliability
enhancement.
9. Advanced Control Algorithms: Implementing advanced control algorithms, such as predic-
tive control and machine learning, can help manage the complex interactions between renew-
able generation, storage, and demand.
10. Collaborative Research: Continued collaboration between research institutions, academia,
industry, and government agencies is essential to address the technical and operational chal-
lenges of integrating RESs while maintaining the distribution system’s reliability.
While integrating RESs into distribution systems poses challenges, innovative technologies,
strategies, and supportive policies can pave the way for a more reliable, resilient, and sustainable
energy future.

Problems
16.1 What is the AENS if SAIDI is given as A hour/year and all customers have the same annual
consumption, B kWhr/year/customer?

16.2 CAIDI is given as A hour/failure, and SAIDI is B hour/year/customer. What is SAIFI?

16.3 If ASAI is 0.9999, and all failures always last exactly one hour and affect the whole system,
what are SAIDI, SAIFI, and ASUI?

16.4 Why does CAIDI sometimes increase when a redundancy eliminating some failures is
introduced?

16.5 How is SAIFI affected if the disconnecting switches are removed?

16.6 A transformer and distributor are part of a radial system that is 400 m long. The transformer
and distributor have annual failure rates of 0.80 and 8.2 failures per kilometer, respectively.
Transformer and distributor repairs take six and five hours, respectively. Could you find the
system’s failure rate, the duration of each outage, and the total annual outage time?
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
322 16 Reliability Assessment of Distribution Systems Integrated with Renewable Energy Systems

16.7 The 132/11 kV transformer, 11 kV transformer, 11 kV feeder, and 11/0.44 kV distribution

transformer constitute a distribution system. It functions in a two-state fluctuation environ-
ment where normal and stormy weather durations are anticipated to last 120 and 3 hours,
respectively. The following table displays the outage rates and repair times:

Equipment 𝝀 𝝀′ 𝝀′′ r r ′′

132/11 kV transformer 0.1 0.1 0.5 6 8

11 kV transformer 0.15 0.15 0.4 4 5
11 kV feeder 1.2 5.2 2 5 3
11/0.44 kV distribution transformer 0.3 0.3 0.6 5 4

Find the system’s yearly outage rate, downtime per outage, and annual total outage time.
The following variables are provided:
𝜆 = Normal weather rate per year, 𝜆′ = Stormy weather rate per year, 𝜆′′ = component outage
rate per year, r = expected repair time forced outage hours/year, r ′′ = expected maintenance
time hours/year, N = expected value for normal hours/year, S = expected value for stormy
hours/year.

16.8 A double circuit 220 kV line operates in a two-state fluctuating mode environment having
expected normal weather and stormy weather durations of 110 and 4 hours, respectively.
Both circuits are similar. The outage data for each circuit is as follows:
𝜆 = 3 outages/year of normal weather, 𝜆′ = 9 outages/year of stormy weather, 𝜆′′ = 1
outage/year, r = 6 hours, r ′′ = 8 hours.
Find annual outage rate, downtime per outage, and total outage time per year.

16.9 Explain the difference between primary and secondary failures in a power system.

Solution
16.1 A × B; kWh per year per customer.

16.2 B/A; failure per year per customer.

16.3 SAIDI = 0.876 hours per year = SAIFI, ASUI = 0.0001.

16.4 CAIDI is a measure of the average length of interruption. Suppose an interruption that
lasts less than the average interruption time of the system is eliminated. In that case, SAIFI
and SAIDI decreases, but the average interruption length, CAIDI, increases: CAIDI =
SAIDI/SAIFI, so when SAIFI decreases more than SAIDI, CAIDI increases.

16.5 SAIFI is unaffected since the same customers will have the same failures in both cases.
Disconnecting switchers will only shorten the average failure length, affecting, for example,
SAIDI, CAIDI, and AENS, but not SAIFI.

16.6 failure rate = 4.1 per year, downtime = 5.1 hours per outage, total outage time = 21.2 hours
per year.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 323

16.7 annual outage rate = 5.35, outage per year = 5.86 hours, total outage time per year =
31.3 hours.

16.8 annual outage rate = 0.023, downtime per outage = 3.1 hours, total outage time per year =
0.07 hours.

16.9 A primary failure is when a component fails independently of the other components in the
system, for example, due to wear or external effects such as falling trees. Secondary failures
are those caused by other failures somewhere else in the system, for example, caused by
overload or by a circuit breaker that does not open in a situation where it could stop a failure
from causing further effects on the system.

16.7 Conclusion

In conclusion, the reliability assessment of distribution systems integrated with renewable energy
systems is critical in ensuring modern power grids’ stability and resilience. The chapter has shed
light on the complex interplay between conventional and renewable energy sources through
comprehensive analysis and evaluation, highlighting challenges and opportunities for enhancing
system reliability. Integrating renewable energy resources brings significant benefits, including
reduced carbon emissions, enhanced energy independence, and potential cost savings. However,
it also introduces unique operational and reliability considerations due to the intermittency and
variability of renewable generation. Advanced modeling techniques, innovative control strategies,
and robust grid infrastructure are essential to address these challenges. Moreover, effective
coordination among stakeholders, including utilities, regulators, and technology providers, is
crucial for maximizing the reliability and performance of distribution systems with renewable
energy integration. Three case studies on the 24-bus IEEE Reliability Test System-79, the 6-bus
Roy Billinton Test System, and the 34-bus Bahir Dar’s Power Distribution Test System highlight
the importance of power system reliability’s significance. Ongoing research and development
efforts are needed to refine reliability assessment methodologies further, incorporate real-time
data analytics, and optimize system planning and operation. By embracing these advancements,
we can pave the way for a more sustainable and resilient energy future, where renewable energy
plays a central role in powering our communities while ensuring the reliability and stability of our
electrical power grids.

References

Afzal, S., Mokhlis, H., Illias, H.A. et al. (2020). State-of-the-art review on power system resilience and
assessment techniques. IET Generation, Transmission & Distribution 14(25): 6107–6121.
Alizadeh, A., Kamwa, I., Moeini, A., and Mohseni-Bonab, S.M. (2022). An exact MILP model for joint
switch placement and preventive maintenance scheduling considering incentive regulation. IET
Generation, Transmission & Distribution 16(23): 4672–4688.
Bagen, B., Huang, D., and Fattal, K. (2019). Enhanced probabilistic approach for substation reliability
assessment. IET Generation, Transmission & Distribution 13(12): 2488–2495.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
324 16 Reliability Assessment of Distribution Systems Integrated with Renewable Energy Systems

Benidris, M. and Mitra, J. (2015). Reliability and sensitivity analysis of composite power systems
considering voltage and reactive power constraints. IET Generation, Transmission & Distribution
9(12): 1245–1253.
Chen, Y., Zheng, Y., Luo, F. et al. (2016). Reliability evaluation of distribution systems with mobile
energy storage systems. IET Renewable Power Generation 10(10): 1562–1569.
Karimi, H., Niknam, T., Dehghani, M. et al. (2021). Automated distribution networks reliability
optimization in the presence of DG units considering probability customer interruption: a practical
case study. IEEE Access 9: 98490–98505.
Kumar, S., Saket, R.K., Dheer, D.K. et al. (2020a). Reliability enhancement of electrical power system
including impacts of renewable energy sources: a comprehensive review. IET Generation,
Transmission & Distribution 14(10): 1799–1815.
Kumar, S., Sarita, K., Vardhan, A.S.S. et al. (2020b). Reliability assessment of wind-solar PV integrated
distribution system using electrical loss minimization technique. Energies 13(21): 5631.
Kumar, S., Sarita, K., Saket, R.K. et al. (2021). Reliability assessment for DFIG-based WECS
considering the impact of 3-phase fault and lightning impulse voltage. International Transactions on
Electrical Energy Systems 31(8): e12952.
Ranjbar, H., Saber, H., and Sharifzadeh, M. (2022). Bi-level planning of distributed energy resources
into existing transmission grids: pathway to sustainable energy systems. IET Generation,
Transmission & Distribution 16(24): 4963–4979.
Sinishaw, G.Y., Bantyirga, B., and Abebe, K. (2021). Analysis of smart grid technology application for
power distribution system reliability enhancement: a case study on Bahir Dar power distribution.
Scientific African 12: e00840.
Taheri Otaghsara, S.S. and Asghari Gharakheili, M. (2022). An optimum budget-based development in
maintenance for electric power transmission networks. IET Generation, Transmission & Distribution
16(22): 4520–4527.
Zare, M., Azizipanah-Abarghooee, R., Hooshmand, R.-A., and Malekpour, M. (2018). Optimal
reconfigurattion of distribution systems by considering switch and wind turbine placements to
enhance reliability and efficiency. IET Generation, Transmission & Distribution 12(6): 1271–1284.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
325

17

Reliability Evaluation and Performance of Hybrid Photovoltaic Energy

Systems for Rural Electrification Using Markov Process
Santosh S. Raghuwanshi 1 , Smriti Singh 2 , Akanksha Singh S. Vardhan 3 , Rajesh Arya 4 , and
R. K. Saket 2
1
Department of Electrical Engineering, Medi-Caps University, Indore, Madhya Pradesh, India
2
Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
3
Department of Electrical Engineering, Shri GS Institute of Technology and Science, Indore, Madhya Pradesh, India
4
Department of Information Technology, Acropolis Institute of Technology and Research, Indore, Madhya Pradesh, India

17.1 Introduction

Energy plays a crucial role in propelling and nurturing the progress and development of society.
Every sector of the economy requires energy input in some manner. Unfortunately, the genera-
tion and utilization of energy give rise to various environmental issues. Different nations and the
international community confront a range of problems related to demand and environmental con-
cerns. These encompass local and regional environmental consequences, potential global climate
change and sea level rise due to the greenhouse effect, and threats to international relations under
oil supply conditions leading to nuclear proliferation. Consequently, each country ought to take
actions to limit environmental harm and energy shortages (Akhtar et al. 2021; Raghuwanshi and
Arya 2020b).
So required alternative energy resources called renewable energy resources. The alternative
energy resources like sun, wind, biomass, ocean, etc., promise clean and abundant sources of
energy for all times to cane. Many believe that alternative energy sources will form the foundation
of energy supplies in the next century. Biomass has dictionary been used very widely for cooking
in developing countries. Expansion of alternative sources faces three obstacles viz., technical,
economic, and reliability. Moreover, the available energy fluctuates with natural cycles. In view
of these, the contribution from these sources will remain marginal in the foreseeable future.
Nevertheless, the global prospects for photovoltaic (PV) cells and wind energy utilization can
increase sustainability if cost are reduced. The cost of PV system would have to reduce by more
than 50% before they can penetrate more into the energy scene. Similarly, if the cost of wind energy
utilization can be brought down by about 30%, the contribution of wind energy can be sustainable
(Alanazi et al. 2022; Ostovar et al. 2021).
Solar energy is the most abundant and constant stream of energy. It is available in both direct
and indirect forms of energy. The design process for the solar PV system encompasses several
key steps, such as calculating the load requirements, estimating losses, choosing an array size that
aligns with the load, and determines the appropriate storage battery capacity. Additionally, a con-
troller is essential to prevent both battery overcharging and excessive discharging. PV systems offer
” standby components is shown in Figure 5.1
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
326 17 Reliability Evaluation and Performance of Hybrid Photovoltaic Energy Systems for Rural Electrification

of reliability, preservation of conventional resources, and pollution reduction, leading to savings.

The uses of the solar PV system encompass a wide range of applications, such as illuminating vil-
lage streets, powering railways, supporting offshore rooftops and oil platforms, facilitating battery
charging, providing domestic lighting, enabling community center illumination, supporting TV
and radio systems, serving defense needs, powering microwaves and repeater stations, facilitat-
ing water pumping, running solar refrigerators, and supporting satellite operations, among others.
For efficient functioning and continue energy supplies used renewable hybrid PV energy system
(RHPVES). RHPVES is includes wind/biomass/fuel cell/solar batteries and conventional diesel
engine (Raghuwanshi and Arya 2019a; Kundur and Malik 2022).
The effectiveness of a system is determined by its ability to perform the intended task and utilize
resources efficiently. The system’s ability to perform its task is primarily dependent on its reliability
and quality. Reliability refers to the probability of failure occurring in a given time interval (Saket
2011). Product reliability is the probability that a unit will adequately perform its intended function
for a specified duration, under specific environmental conditions (Niu et al. 2021).
The reliability of systems is often compromised due to poor design and manufacturing. To
improve reliability, it is important to plan and design new systems with reliability as a key param-
eter. With increasing automation and complexity, subsystems become more integral, leading to
potential conflicts and decreased reliability. This highlights the importance of stringent reliability
standards to ensure the success of these systems in performing critical tasks. Additionally, vari-
ations in temperature and irradiance can impact panel performance, which must be considered
by the International Commission on Accreditation (ICA) (Abunima and Teh 2020; Tawfiq et al.
2021).
The reliability of a system is dependent on the reliability of its individual parts and components.
The failure of a single component can cause the entire system to fail, making it crucial to consider
the reliability of each component when assessing system reliability. To achieve this, it is necessary
to analyze the functional relationships between components and their failure models (Saket 2011).
While it may be difficult to establish failure rates for all working conditions, it is possible to predict
reliability under specific stress conditions through extrapolation or interpolation of available data
(Kebede et al. 2021).
This chapter presents the Markov process and calculates important reliability indices, such as
LOLP, availability, failure and repair time, and EENS. The hybrid renewable energy system is con-
sidered for the evaluation of these reliability indices.
The chapter’s structure unfolds as follows: Section 17.1 serves as the chapter’s Introduction.
Section 17.2 provides an explanation of Reliability Indices. Section 17.3 delves into the Markov pro-
cess. Section 17.4 explores the Reliability of Series, Parallel, and Complex Systems, along with an
explanation of the components of the Renewable Hybrid PV System. This section also includes illus-
trative numerical examples with solutions to enhance comprehension. Lastly, Section 17.5 brings
the chapter to a Conclusion.

17.2 Reliability Indices

● Availability: The performance of maintained equipment can be measured by its availability,
which takes into account both reliability and maintainability factors. Availability is determined
by the frequency of failures and the speed at which faults are corrected. Steady-state availabil-
ity refers to the duration of time that equipment is available for use (Postnikov and Mednikova
2022).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
17.3 Markov Process 327

● Repairability: Repairability is the likelihood that a failed equipment can be repaired and
restored to operable condition within a specified downtime period, assuming maintenance is
performed under optimal conditions.
● Failure: A device or system experiences a failure when its properties are altered to the extent that

it cannot function properly, either partially or completely. While some components have clearly
defined failure modes, others do not.
● EENS Index: It is a measure of the amount of energy that is not delivered to customers due to

outages in a power system. It is calculated by multiplying the probability of an outage by the

amount of energy that would have been delivered if the outage had not occurred. The EENS
index is typically expressed in megawatt hours (MWh) (Kuo et al. 2022).
● LOLP: LOLP is a measure of the probability that a power system will not be able to meet the

demand for electricity. It is calculated by multiplying the probability of each component in the
system failing by the amount of load that each component can supply. The LOLP is typically
expressed as a percentage. A lower LOLP indicates a more reliable power system. A high LOLP
could indicate that there are problems with the system’s infrastructure, such as aging equip-
ment or inadequate maintenance. It could also indicate that the system is not operating at its full
capacity (Hou et al. 2023).
These reliability indices help engineers, analysts, and decision-makers understand and improve
the performance, maintenance, and design of systems by identifying potential weaknesses or areas
for improvement in terms of reliability and dependability.

17.3 Markov Process

Markov analysis examines a system’s various states, including when all components are functioning
and when one component has failed while others continue to operate. The fundamental concept
of Markov analysis is that the likelihood of a system moving from one state to another is deter-
mined exclusively by its present state, with no consideration of past states. This property, similar
to exponential distribution in finance, makes Markov analysis ideal for predicting failure times. To
analyze a system’s reliability, Markov process can be used, with the system’s N components falling
into either an operating or failed state. The system’s state can be defined as one of the 2n possible
combinations of functioning and nonfunctioning components (Kuo et al. 2022).
In a Markov process, there are two key components:
1. States: A set of distinct states that the system can be in. These states can represent various
situations, conditions, or configurations of the system being modeled. For example, in weather
forecasting, states could be “sunny,” “cloudy,” or “rainy” (Kalantar-Neyestanaki and Cherkaoui
2021).
2. Transition Probabilities: The probabilities linked to transitions between different states are
usually depicted using a transition matrix or transition diagram. Within this matrix, each indi-
vidual element signifies the likelihood of moving from one state to another within a single time
interval (Rayati et al. 2023).
A Markov process can be employed to represent the unpredictable behavior of a PV system, and
a Markov state diagram can be employed for its depiction. The main advantage of this approach
is that it offers a comprehensive representation of all possible states and the transitions between
them. The occurrences of PV string(s), module(s), and inverter(s) failures and subsequent repairs
are responsible for the shifts between different Markov states. Consequently, this strategy can be
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
328 17 Reliability Evaluation and Performance of Hybrid Photovoltaic Energy Systems for Rural Electrification

especially advantageous for managing individual components (Singh ƛ

et al. 2023). Examine a repairable single-component system as illustrated
in the state-transition diagram. In Figure 17.1, the state-transition dia- 0 1
gram for the Markovian model can be seen. Each component in this Up Down
system can exist in one of two operational states. When the component
is in state 1, it functions as intended, while in state 0, it is considered μ
failed. The component’s failure rate is denoted as 𝜆, and its repair rate
Figure 17.1 State-
is represented by 𝜇. Thus, the condition of the component, whether it transition diagram for
is operational or faulty, can be depicted using a Markov chain featuring single repairable
two states: 𝜆 and 𝜇 (as shown in Figure 17.2). component system.
The steady-state probabilities is given by Eqs. (17.1) and (17.2):
𝜇
Pup = (17.1)
𝜆+𝜇
𝜆
Pdown = (17.2)
𝜆+𝜇
Similarly, for a two repairable component system the staedy state probabilities is given by Eqs.
(17.3) and (17.4):
𝜇1 𝜇2
Pup = ( )( ) (17.3)
𝜆1 + 𝜇 1 𝜆 2 + 𝜇 2
𝜆1 𝜆2
Pdown = ( )( ) (17.4)
𝜆1 + 𝜇 1 𝜆 2 + 𝜇 2
The key assumption in a Markov process is that the future behavior of the system depends only
on its current state and not on how it arrived at that state. This property is known as the Markov
property.
Markov processes have several important properties:
1. Homogeneity: The transition probabilities remain constant over time. This means that the
probability of transitioning from one state to another does not change as time progresses.
2. Irreducibility: In a finite number of steps, it’s feasible to reach any state from any other state
with a nonzero probability. In other words, there are no isolated subsets of states that cannot be
reached from other states (Kou et al. 2022).
3. Stationarity: The long-term behavior of a Markov process reaches an equilibrium distribution
called the stationary distribution or steady-state distribution. In this distribution, each state has
a fixed probability of being occupied.
Markov processes find applications in various fields such as finance (stock market modeling),
biology (population dynamics), physics (particle movements), and computer science (algorithm
analysis). They are also used in machine learning algorithms like Hidden Markov Models (HMMs)
for speech recognition and natural language processing (Spertino et al. 2020).

ƛ1 Figure 17.2 State-transition diagram for two repairable component

system.
1U&2U 1D&2U
μ1
ƛ2 μ2 μ2 ƛ2
ƛ1
1U&2D 1D&2D
μ1
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
17.4 Reliability of the System 329

The steps of a Markov process for calculating reliability indices are as follows:
1. Step 1: Define the system: Identify the components and their states that make up the system.
Determine the possible states of each component, such as functioning or failing.
2. Step 2: Construct the state-transition diagram: Create a diagram that represents the possible
transitions between states for each component. This diagram shows how the system can move
from one state to another based on component failures or repairs.
3. Step 3: Assign probabilities: Assign probabilities to each transition in the state-transition dia-
gram. These probabilities represent the likelihood of moving from one state to another, such as
the probability of a component failing or being repaired.
4. Step 4: Calculate steady-state probabilities: Solve for the steady-state probabilities of each state
in the system using mathematical equations or numerical methods, such as matrix operations
or iterative algorithms. These steady-state probabilities represent the long-term behavior of the
system and indicate its reliability (Spertino et al. 2020).
5. Step 5: Compute reliability indices: Once you have obtained the steady-state probabilities, you
can calculate various reliability indices, such as availability, LOLP, EENS, and system reliability.
6. Step 6: Analyze results and make decisions. Analyze the calculated reliability indices to assess
the performance and dependability of the system. Use these results to make informed decisions
(Samy et al. 2020).

17.4 Reliability of the System

The system components may be connected in three ways: series, parallel, and complex
(series–parallel) configuration. In series, all components must be working for the system to
work. In a parallel configuration, at least one component must work for system to work.

17.4.1 System with Series Components

In series system, all the components are necessary to operate for system to work. The block diagram
in Figure 17.3 illustrates the representation of two series components within the system. If any of
the n components fail, the information will not reach the OUT end in the system.
The reliability of series components system is (Samy et al. 2020; Xu et al. 2022):
∑
n
∏
n
− 𝜆j ∗t
−𝜆j t
R(t)series = e =e j=1
(17.5)
j=1

The unavailability, failure rate, and interruption duration are as follows (Samy et al. 2020; Xu
et al. 2022):
∑
n
Useries = 𝜆j ∗ r j (17.6)
j=1
∑n
𝜆series = 𝜆j (17.7)
j=1

Now,
∑n
j=1 𝜆j ∗ rj
rseries = ∑N (17.8)
j=1 𝜆j

where n is number of system components.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
330 17 Reliability Evaluation and Performance of Hybrid Photovoltaic Energy Systems for Rural Electrification

1 2 n
In Out

Figure 17.3 Series system.

17.4.2 System with Parallel Components

Two or more components are in parallel configuration if all units must fail for the system fail. If one
components or units operate system will operate. Two parallel components system are represented
by the block diagram shown in Figure 17.4 shows that there are n parallel paths between the IN
and OUT ends, with each component connected across each other.
The reliability of parallel components system is as follows (Billinton et al. 1992):
∏
n
R(t)parallel = 1 − [1 − e−𝜆j t ] (17.9)
j=1

To calculate the unavailability, failure rate, and interruption duration of a parallel system with
two components, the following equations can be utilized (Billinton et al. 1992; Abunima and Teh
2020).
𝜆 1 𝜆2
Uparallel = (17.10)
(𝜆1 + 𝜇1 )(𝜆2 + 𝜇2 )
Under the assumptions that 𝜆1 ≪ 𝜇1 , 𝜆2 ≪ 𝜇2 , approximate relations can be derived for the
indices:

Uparellel = (𝜆1 r1 ) ∗ (𝜆2 r2 ) (17.11)

r1 r2
rparellel = (17.12)
r1 + r2
𝜆parallel = 𝜆1 𝜆2 (r1 + r2 ) (17.13)

17.4.2.1 Complex (Series–Parallel) System

Systems are not always series or parallel configuration. A complex system is the combination may
be series or parallel or series–parallel configuration shown in Figure 17.5. For the complex system
successful operation of the system, there should be at least one continuous path between IN and
OUT terminals.

Figure 17.4 Parallel system.

1

2
In Out

n
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
17.4 Reliability of the System 331

Figure 17.5 Series–parallel system.

1

In Out
3

Series–parallel system indices are as follows (Khan et al. 2022):

𝜆comp = (𝜆1 𝜆2 ) ∗ (r1 + r2 ) ∗ 𝜆3 (17.14)

Ucomp = (𝜆parellel •rparallel + 𝜆3 •r3 ) (17.15)

(𝜆parallel •rparallel ) + 𝜆3 •r3

rcomp = (17.16)
𝜆1 •𝜆2 •𝜆3 •(r1 + r2 )
EENS is as follows Raghuwanshi and Arya (2019b) and Ghania et al. (2022):
EENS = Lavg •Usystem kWh/yr (17.17)
where Lavg is average load in kW. LOLP is Raghuwanshi et al. 2023 and Raghuwanshi and Arya
(2020a):
Usystem
LOLP = (17.18)
8760

17.4.3 Renewable Hybrid PV System

Figure 17.6 depicts the RHPVES, which is comprised of PV arrays, a converter, an inverter, a charge
controller, a solar battery, a DG set, and a load. The PV arrays and solar batteries are connected to
the direct current (DC) bus via the charge controller, while the inverter is connected between the
DC bus and alternating current (AC) bus. The AC bus is connected to the DG set via a changeover
switch (COS). The farm house has two types of loads: household and irrigation.

DC bus AC bus

Farmhouse load
DC/DC

PV panel Converter
DC/AC
Irrigation
Inverter

CC

Charge COS
Battery bank controller Change over
switch DG set

Figure 17.6 Renewable hybrid PV energy system.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
332 17 Reliability Evaluation and Performance of Hybrid Photovoltaic Energy Systems for Rural Electrification

17.4.3.1 PV Arrays
PV effect is the direct conversion of sun energy into electrical energy. Solar cells, which are devices
that employ this effect to convert sunlight into electricity, can be made up of a single converter cell,
also known as a photovoltaic cell, or a combination of cells called a solar array. These cells contain
semiconductors that generate voltage proportional to the amount of light or radiation they receive.
Iron selenium cells are a type of PV cell commonly employed, despite being very rare (Alfawaz et al.
2023).
PV arrays generator power is as follows (Alfawaz et al. 2023):
[ ( ( ) )]
eV
Pavg = ILg − Ir exp −1 ×v (17.19)
kT
In this context, ILg and Ir correspond to the light-generated current and reverse saturation cur-
rent, respectively. Additionally, e represents the electron charge, k denotes Boltzmann’s constant,
T stands for temperature, and v signifies the output voltage of the cells.

17.4.3.2 Converter
A converter is an electronic device that can transform a fixed voltage or DC voltage from a PV panel
into a variable, average voltage that is suitable for a specific load. This is achieved by inserting a
high-speed switch, such as an IGBT, GTO, SCR, MOSFET, or another similar component, between
the DC source and the load. In a solar PV energy system, the converter is usually installed between
the PV panels and the DC bus. Buck, boost, and buck–boost converters are link between PV and
inverter or PV-battery-inverter. Generally, buck–boost converter topology is used.
The output voltage of the buck–boost converter is as outlined below Raghuwanshi and Khare
(2018),
d
Vo = V (17.20)
1 − d PV
where d and VPV are the duty cycle and PV voltage, respectively.

17.4.3.3 Inverter
An inverter is an electronic circuit designed to convert DC power into AC power. This conversion
process can be achieved using either transistors or silicon-controlled rectifiers (SCRs). Transis-
torized inverters are ideal for low and medium power outputs, while SCRs are more suitable for
high-power outputs. This is because SCRs are capable of handling significantly higher load cur-
rents than transistors. As a result, SCRs have become the preferred choice for high-power output
applications. In solar energy systems, an inverter is typically installed between the DC bus and the
AC bus. This enables the converted AC power generated by the solar panels to be supplied to the
AC loads in the building or fed back to the utility grid.
The inverter power is as follows Raghuwanshi and Khare (2018),
PLoad
Pinverter = (17.21)
𝜂inverter
where PLoad and inverter are the load power and inverter efficiency, respectively.

17.4.3.4 Charge Controller and Solar Batteries

Charge controller is circuit that connects the DC bus to solar batteries. This charge controller is
controlling the charging and discharging of the solar batteries. The charge controller also prevents
the overcharging and deep charging of the battery (Billinton 1970). Electrical Storage batteries
are devices that can store and release electricity and are comprised of individual cells that work
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
17.4 Reliability of the System 333

together. Each cell is made up of a material and an electrolyte that works together to produce
electrochemical energy storage. When a battery is connected to an electric load, separation of
charge occurs at the interface between the electrodes and the electrolyte, creating free electrons
and ions. The electrons flow through the external load, while the ions flow through the electrolyte
and combine with the other electrode to create a circuit. This process allows the battery to convert
chemical energy into electrical energy that can be used to power various devices (Raghuwanshi
and Khare 2018).
Energy storage ability in battery is as follows,
(Eavg × DA)
Ebattery = (17.22)
𝜂inverter × 𝜂battery × DOD

where Eavg and DA are average demand energy and autonomy, respectively, DOD is the depth of
discharge, 𝜂inverter and 𝜂battery are inverter and battery efficiency, respectively.

17.4.3.5 DG Set
A DG is a device that consists of a diesel engine and an alternator, working together to convert
mechanical energy into electrical energy. Diesel engine operates by repeating a cycle of four stages
or strokes during which the Piston move up and down. Diesel engine is connected to synchronous
generator with couple via shaft. A diesel engine generates mechanical power, which is then trans-
formed into electrical power through an electrical generator. The DG is linked to an AC load.
DG-generated power is as follows (Raghuwanshi et al. 2022),
𝜋
PDG = P × L × A × N × (17.23)
30
where P and L are the operative pressure and length of stroke, respectively, A and N are the area of
cylinder bore and RPM, respectively.

Problem 17.1
Design a Reliability Block Diagram (RBD) of the hybrid energy system is shown in Figure 17.6.
Calculate the following reliability indices of PV-only and DG-only system: failure rate, unavailabil-
ity, EENS and LOLP. Table 17.1 provides the failure rate and repair time data for each individual
component of the energy system. Consider Lavg as 5 kW.

Table 17.1 Failure rate and repair time information for each component within the system

S. No. System apparatuses Failure rate (𝛌) in yr−1 Repair time (r) in h

1 PVG 0.055 40
2 Converter 0.0357 70
3 Solar inverter 0.0945 65
4 Charge controller 0.0312 55
5 Solar batteries 0.012 25
6 DG set 0.499 90
7 Switch 0.011 15
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
334 17 Reliability Evaluation and Performance of Hybrid Photovoltaic Energy Systems for Rural Electrification

Switch

PVG DC/DC Inverter

Household load

Irrigation load

Figure 17.7 RBD of PV-only system.

Solution:

1. The PV-only system includes PV arrays, a DC–DC converter, a solar inverter, a switch, and a
load, all connected in series. The RBD of the PV-only system is depicted in Figure 17.7.
The reliability indices are calculated using Eqs. (17.5)–(17.18). The failure rate, unavailability,
EENS, and LOLP are as follows:
𝜆sys = 0.1955 yr−1
Usys = 10.995 h∕yr
rsys = 56.24 h
EENS = 54.97 kWh/yr
LOLP = 0.001255 yr−1
In series system, if any one component is not working, the complete system is shutdown. The
electricity is not supplied to the load. This system is not reliable system.
2. DG Only System: DG only system consists diesel-generator set, switch, and load. The RBD of
DG only system is shown in Figure 17.8. The system components are connected in series.
The failure rate, unavailability, EENS, and LOLP are as follows:
𝜆sys = 0.51 yr−1
Usys = 45.075 h∕yr
rsys = 88.38 h

DG Household load

Switch

Irrigation load

Figure 17.8 RBD of DG only.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
17.4 Reliability of the System 335

EENS = 225.375 kWh/yr

LOLP = 0.005145 yr−1
Reliability indices shows the system failure rate is high and system is not reliable. DG is failed,
energy not supplied the load. This system configuration is not adequate system in remote areas.

Problem 17.2
Design a Reliability Block Diagram (RBD) of the hybrid energy system shown in Figure 17.6. Calcu-
late the following indices of PV/battery, DG/battery, and PV/DG energy system. Table 17.1 provides
the failure rate and repair time data for each individual component of the energy system.
Solution:

1. PV with Battery Backup System: The PV/battery system comprises PV arrays, a DC–DC
converter, a charge controller, a solar batteries, a solar inverter, a switch, and a load. These com-
ponents are interconnected in both series and parallel configurations, as shown in Figure 17.9.
The PV arrays generate electricity which is converted into DC power by the DC–DC converter.
This power is then sent to the charge controller which regulates the charging of the solar batter-
ies. The solar batteries stores the energy for later use. When energy is needed, the solar inverter
converts the DC power stored in the solar batteries into AC power for the load to use. The switch
is used to control the power flow between the PV arrays, the solar batteries, and the load.
The failure rate, unavailability, EENS, and LOLP are calculated using Eqs. (17.5)–(17.18).
𝜆sys = 0.21 yr−1
Usys = 7.86 h∕yr
rsys = 39 h
EENS = 39.3 kWh/yr
LOLP = 0.000897 yr−1
This system is not feasible in terms of load demand. Inverter is failed the overall system is shut
down and not supplied the electricity.

PVG DC/DC Inverter

Switch Household load

Charge controller

Battery

Irrigation load

Figure 17.9 RBD of PV system with battery.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
336 17 Reliability Evaluation and Performance of Hybrid Photovoltaic Energy Systems for Rural Electrification

Inverter

Household load
Charge
controller
Switch

Battery

DG Irrigation load

Figure 17.10 RBD of DG/battery.

2. Diesel with Battery Backup System: The DG/battery system is composed of a DG generator,
a charge controller, a solar-batteries, a solar inverter, a switch, and a load. The block diagram
of the system can be seen in Figure 17.10. The DG generator is connected in parallel with the
solar batteries in order to ensure that the power generated can be efficiently stored. The charge
controller regulates the charging of the solar batteries to ensure that it is fully charged and ready
for use. When energy is required, the solar inverter converts the DC energy stored in the solar
batteries into AC power for the load to use. The switch is used to control the power flow between
the DG generator, the solar batteries, and the load. This system provides reliable backup power,
especially in areas where there are frequent power outages.
The failure rate, unavailability, EENS, and LOLP are calculated using Eqs. (17.5)–(17.18).

𝜆sys = 0.0318 yr−1

Usys = 0.0992 h∕yr
rsys = 9.88 h
EENS = 0.496 kWh/yr
LOLP = 0.0000113 yr−1

3. PV with DG System: The PV/DG system comprises of PV arrays, a DC–DC converter, a solar
inverter, a DG generator, a switch, and a load. As seen in Figure 17.11, the PV arrays are con-
nected in series to ensure effective power generation. The PV system is connected in parallel with
the DG generator to ensure optimal utilization of the power generated. The DC–DC converter
is responsible for regulating the voltage levels of the DC power generated by the PV arrays. The
solar inverter then converts the DC power into AC power, which can be used directly by the load.
The switch is used to control the flow of power between the PV arrays, the DG generator, and
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
17.4 Reliability of the System 337

PVG DC/DC Inverter

Household load

Switch

DG Irrigation load

Figure 17.11 RBD of PV/diesel energy system.

the load. This system is ideal for areas with ample sunlight and unreliable grid power, providing
a stable and consistent source of electricity.
PV/diesel energy system reliability indices are calculated using Eqs. (17.5)–(17.18):

𝜆sys = 0.0245 yr−1

Usys = 1.023 h∕yr
rsys = 10.20 h
EENS = 5.105 kWh/yr
LOLP = 0.0001167 yr−1

This system is adequate system, if one system is repair on that time second system is supply the
electricity.

Problem 17.3
Design a Reliability Block Diagram (RBD) of the hybrid energy system shown in Figure 17.12.
Calculate the following Reliability Indices of PV/DG/battery: failure rate, unavailability, EENS,
and LOLP. Table 17.1 provides the failure rate and repair time data for each individual component
of the energy system.
Solution:
PV/DG with battery backup: The PV/DG/battery system is made up of several essential compo-
nents, including PV arrays, a DC–DC converter, a charge controller, a solar inverter, a DG generator,
a switch, and a load. This system is illustrated in the block diagram shown in Figure 17.12.
The PV arrays are connected in series, whereas the PV/battery system is connected parallel with
the DG set. Moreover, the PV system is connected parallel with solar-batteries, ensuring that any
excess energy generated is stored for future use. The charge controller is responsible for regulating
the charging of the battery, ensuring it does not overcharge or undercharge. Overall, this system is
ideal for remote areas where there is no access to the power grid, providing efficient and reliable
electricity for various applications. The indices are:
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
338 17 Reliability Evaluation and Performance of Hybrid Photovoltaic Energy Systems for Rural Electrification

PV DC/DC Inverter

Household load
Charge
controller
Switch
Battery

Irrigation load
DG

Figure 17.12 RBD of PV/DG/battery energy system.

𝜆hybrid = 0.0821 yr−1

Uhybrid = 0.0603 h∕yr
rhybrid = 0.8923 h
EENS = 0.3015 kWh/yr
Uhybrid
LOLP = = 0.000006883
8760
The indices results show that RHPVES is suitable and more feasible energy system in remote
areas.

17.5 Conclusion
This chapter’s primary focus was on assessing the reliability of standalone solar PV energy systems.
To achieve this objective, six different system configurations were analyzed, including PV-only,
diesel-only, PV/diesel, PV/battery, PV/diesel/battery, and diesel/battery. The development of the
system’s operational strategy was based on various indices, such as failure rate, repair time, unavail-
ability, LOLP, and EENS. By utilizing the Markov process method, a reliability evaluation was
conducted, which allowed for a thorough analysis of the system’s suitability for the intended appli-
cation. The system configurations were analyzed based on their unavailability, failure rate, EENS,
and LOLP. Among all configurations considered, the hybrid PV/battery/DG system demonstrated
the lowest values across all indices and proved to be the most reliable and sustainable option. The
reliability performance of a system can be better understood through these indices, and as such,
the PV/battery/DG system showcases superior reliability and sustainability compared to other con-
figurations. Therefore, it can be concluded that PV/battery/DG system is the most reliable and
sustainable option among all system configurations.

References

Abunima, H. and Teh, J. (2020). Reliability modeling of PV systems based on time-varying failure rates.
IEEE Access 8: 14367–14376.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 339

Akhtar, I., Kirmani, S., and Jameel, M. (2021). Reliability assessment of power system considering the
impact of renewable energy sources integration into grid with advanced intelligent strategies. IEEE
Access 9: 32485–32497.
Alanazi, A., Alanazi, M., Nowdeh, S.A. et al. (2022). An optimal sizing framework for autonomous
photovoltaic/hydrokinetic/hydrogen energy system considering cost, reliability and forced outage
rate using horse herd optimization. Energy Reports 8: 7154–7175.
Alfawaz, O., Khedr, A.M., Alwasel, B., and Osamy, W. (2023). Reliability evaluation for chain routing
protocols in wireless sensor networks using reliability block diagram. Journal of Sensor and Actuator
Networks 12 (2): 34.
Billinton, R. (1970). Power System Reliability Evaluation. Taylor & Francis.
Billinton, R., Allan, R.N., Billinton, R., and Allan, R.N. (1992). Discrete Markov chains. In: Reliability
Evaluation of Engineering Systems: Concepts and Techniques, 260–279. Boston, MA: Springer.
Ghania, S.M., Mahmoud, K.R.M., and Hashmi, A.M. (2022). A reliability study of renewable energy
resources and their integration with utility grids. Engineering, Technology & Applied Science Research
12 (5): 9078–9086.
Hou, K., Tang, P., Liu, Z., and Dong, Z. (2023). An efficient reliability assessment approach for multiple
energy systems based on shadow price. Energy Reports 9: 829–836.
Kalantar-Neyestanaki, M. and Cherkaoui, R. (2021). Coordinating distributed energy resources and
utility-scale battery energy storage system for power flexibility provision under uncertainty. IEEE
Transactions on Sustainable Energy 12 (4): 1853–1863.
Kebede, F.S., Olivier, J.-C., Bourguet, S., and Machmoum, M. (2021). Reliability evaluation of renewable
power systems through distribution network power outage modelling. Energies 14 (11): 3225.
Khan, A.A., Minai, A.F., Pachauri, R.K., and Malik, H. (2022). Optimal sizing, control, and
management strategies for hybrid renewable energy systems: a comprehensive review. Energies 15
(17): 6249.
Kou, L., Chu, B., Chen, Y., and Qin, Y. (2022). An automatic partition time-varying Markov model for
reliability evaluation. Applied Sciences 12 (12): 5933.
Kundur, P.S. and Malik, O.P. (2022). Power System Stability and Control. McGraw-Hill Education.
Kuo, T.-C., Pham, T.T., Bui, D.M. et al. (2022). Reliability evaluation of an aggregate power conversion
unit in the off-grid PV-battery-based DC microgrid from local energy communities under dynamic
and transient operation. Energy Reports 8: 5688–5726.
Niu, M., Xu, N.Z., Kong, X. et al. (2021). Reliability importance of renewable energy sources to overall
generating systems. IEEE Access 9: 20450–20459.
Ostovar, S., Esmaeili-Nezhad, A., Moeini-Aghtaie, M., and Fotuhi-Firuzabad, M. (2021). Reliability
assessment of distribution system with the integration of photovoltaic and energy storage systems.
Sustainable Energy, Grids and Networks 28: 100554.
Postnikov, I. and Mednikova, E. (2022). A reliability analysis of fuel supply for district heating systems
based on statistical test method. Energy Reports 8: 304–311.
Raghuwanshi, S.S. and Arya, R. (2019a). Renewable energy potential in India and future agenda of
research. International Journal of Sustainable Engineering 12 (5): 291–302.
Raghuwanshi, S.S. and Arya, R. (2019b). Reliability evaluation of stand-alone solar PV energy system
for irrigation. Proceedings of Recent Advances in Interdisciplinary Trends in Engineering &
Applications (RAITEA).
Raghuwanshi, S.S. and Arya, R. (2020a). Design and economic analysis of a stand-alone hybrid
photovoltaic energy system for remote healthcare centre. International Journal of Sustainable
Engineering 13 (5): 360–372.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
340 17 Reliability Evaluation and Performance of Hybrid Photovoltaic Energy Systems for Rural Electrification

Raghuwanshi, S.S. and Arya, R. (2020b). Reliability evaluation of stand-alone hybrid photovoltaic
energy system for rural healthcare centre. Sustainable Energy Technologies and Assessments 37:
100624.
Raghuwanshi, S.S. and Khare, V. (2018). Sizing and implementation of photovoltaic water pumping
system for irrigation. IAES International Journal of Artificial Intelligence 7 (1): 54.
Raghuwanshi, S.S., Masih, A., and Raghuwanshi, P. (2022). Design and optimisation of stand-alone
hybrid energy system for rural areas. International Journal of Ambient Energy 43 (1): 8707–8721.
Raghuwanshi, S.S., Raghuwanshi, P., Masih, A., and Singh, P. (2023). Modeling and optimization of
hybrid renewable energy with storage system using flamingo swarm intelligence algorithms. Energy
Storage 5 (7): e470.
Rayati, M., Bozorg, M., Carpita, M., and Cherkaoui, R. (2023). Stochastic optimization and Markov
chain-based scenario generation for exploiting the underlying flexibilities of an active distribution
network. Sustainable Energy, Grids and Networks 34: 100999.
Saket, R.K. (2011). Reliability evaluation of defence support systems. In: Innovations in Defence Support
Systems-2: Socio-Technical Systems, vol. 338 (ed. L.C. Jain, E.V. Aidman, and C. Abeynayake),
241–286. Berlin, Heidelberg: Springer-Verlag.
Samy, M.M., Mosaad, M.I., El-Naggar, M.F., and Barakat, S. (2020). Reliability support of
undependable grid using green energy systems: economic study. IEEE Access 9: 14528–14539.
Singh, S., Saket, R.K., and Khan, B. (2023). A comprehensive review of reliability assessment
methodologies for grid-connected photovoltaic systems. IET Renewable Power Generation 17 (7):
1859–1880.
Spertino, F., Chiodo, E., Ciocia, A. et al. (2020). Maintenance activity, reliability, availability, and
related energy losses in ten operating photovoltaic systems up to 1.8 MW. IEEE Transactions on
Industry Applications 57 (1): 83–93.
Tawfiq, A.A.E., El-Raouf, M.O.A., Mosaad, M.I. et al. (2021). Optimal reliability study of
grid-connected PV systems using evolutionary computing techniques. IEEE Access 9: 42125–42139.
Xu, X., Tang, J., and Xiang, H. (2022). Data transmission reliability analysis of wireless sensor networks
for social network optimization. Journal of Sensors 2022: 1–12.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
341

18

Probabilistic Distribution and Monte Carlo Approach for Reliability

Evaluation of SEIG-Based Micro Hydro Power Generation System
Lokesh Varshney 1 , Kanhaiya Kumar 2 , Gautam Singh Dohare 3 , Udaya M. Bhaskara Rao 4 , and
Jitendra Singh Shakya 5
1
Department of Electrical, Electronics & Communication Engineering, Galgotias University, Greater Noida, Uttar Pradesh, India
2 Electrical
& Electronics Engineering Department, Galgotias College of Engineering & Technology, Greater Noida,
Uttar Pradesh, India
3
Department of Electrical, Electronics & Communication Engineering, Galgotias University, Greater Noida, Uttar Pradesh, India
4
Engineering Division, CSIR National Metallurgical Laboratory, Jamshedpur, Jharkhand, India
5 Department of Electrical Engineering, Samrat Ashok Technical Institute, Vidisha, Madhya Pradesh, India

18.1 Introduction

Compared to an alternator, a self-energized induction generator (SEIG) has various benefits.

However, it has certain issues, like reactive energy use and weak voltage regulation. The SEIG
works either as the SEIG was constructed by the engineer or as an appropriate level of capacitor
is attached in shunt with the induction motor (IM) contacts. In comparison to the IM, the acces-
sibility of developed SEIG is not very common. But in highlands or isolated places, uncontrolled
renewable energy (RE) sources like micro hydro and wind are fairly common. SEIG has a direct
link to a turbine that has a renewable energy source connected to it. Dams, gear-pulleys, and
various other mechanical attachments are not required. An IM’s effectiveness when functioning
as a SEIG with an appropriate capacitance level is unsatisfactory. The choice of the parameter
values made throughout the design stage, which optimized the efficacy of the device as a motor
instead of as a power source, is what causes the poor performance. Understanding that an IM is a
magnetic circuit and the way hysteresis will affect it is crucial for illustrating this notion. In light
of this hysteresis, an IM has two operational areas. In particular, an IM will function in either
a linear or nonlinear saturation zone. In the majority of instances, it is not advisable to run an
IM at saturation because doing so lowers the iron’s relative permeability and raises the motor’s
magneto motive force (MMF) need. However, working in the linear zone only partially makes
use of the iron’s abilities, making this strategy uneconomical. As a consequence, the knee of the
curve of saturation represents the ideal point of operation for an IM. These ideas maximize iron
use while reducing saturation. A SEIG’s functioning is steady when its magnetic circuit reaches
saturation, though. As a result, an IM’s voltage at the terminal is raised till the magnetic circuit
becomes saturated with the goal to use the IM as a generator (Nakorn et al. 2021; Silva et al. 2018).
As a consequence, IM work prior to the knee of the saturate curve, whereas SEIG works after the

Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
342 18 Probabilistic Distribution and Monte Carlo Approach for Reliability Evaluation

knee. The IM’s rotor core is designed with less hysteresis loss in mind; hysteresis loss is dependent
on the size of the hysteresis loop. Because of this circumstance, an IM’s rotors core is made of soft
magnetic material, resulting in a smaller hysteresis loop. On the opposite hand, the rotating core
of the planned SEIG, wherein hysteresis loop is wide, uses hard magnetic materials. As a result,
the rotor of this machine has a greater hysteresis impact. As a result, the rotating core possesses
enough remaining magnetism to cause the SEIG’s first activation. When using an IM as a SEIG
under these circumstances, the issue of excitation loss may arise. But as compared to how SEIG
was designed, IM is more accessible and affordable. Consequently, utilizing IM as a SEIG is fairly
common. It is necessary to assess the machine’s rotor’s dependability because it will function
as SEIG.
The structure of this research is as follows: Section 18.2 has information on residual magnetism
losses and recovery. These three techniques for restoring residual magnetism are briefly described
in this section. The idea, the causes, and variables that contribute to failure functioning of SEIG are
described in Section 18.3. The measured minimal capacitance value for SEIG excitation is evaluated
in Section 18.4. The reliability of the SEIG’s rotors core magnetization is assessed in Section 18.5
along the lines of the curve, hazard rate, failure density function, survivor function, cumulative
failure distribution, and utilizing Monte Carlo simulation (MCS), the odds of either succeeding or
failing have been assessed. The work’s summary is presented in Section 18.6.

18.2 Residual Magnetism in SEIG: Restoration and Loss

18.2.1 Residual Magnetism Losses

The short-circuiting state and coupling of an additional reactive loading are to blame for the
depletion of residual magnetism in SEIG. These circumstances are what lead to the SEIG’s abrupt
endpoint voltage reduction and the reduction in remaining magnetism in the rotating core.
Figure 18.1 depicts the tested configuration with reactive loads. In order to temporarily excite the

iga iaM

Cse
ic
ilc i;a
Prime
mover SEIG
IM
igb
ibM
ilb

igc Cse icM

Csh

Figure 18.1 Reactive loading configuration for a study.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
18.4 SEIG Tests with Lowest Capacitive Excitation 343

core of iron and restore the remaining magnetism, one may employ (any one of) the methods
described in Section 2.2.

18.2.2 Residual Magnetism Restoration

In broadly, three techniques are effective for restoring residual magnetism.
1. For a period of 10–15 minutes, use the machine as a motor powered by an exciting ac unit.
2. By bringing in capacitors with charge terminals. In most cases, even with a degaussed rotor,
the current at the discharge point will be enough to trigger self-excitation if the capacitors are
charged to a very high voltage, such as the rated machine voltage.
3. By raising the machine velocity over its certified value, which exceeds the resonant frequency at
weak magnetization and starts self-excitation (keep in mind that the rotors and bearing of the
unit must be certified for the increased velocity).
4. Attach six-volt batteries across two of the equipment’s connections when it is at resting and
leave it there for 10–15 minutes. Or, enough of residual magnetism may be ensured by running
a DC current across the equipment before it is brought up to speed. With the power grid being
available, the initial option is superior to the other strategies. The third option is helpful when
the grid is unavailable (Chaturvedi and Goel 2019).

18.3 Problems with SEIG Excitation Failure in RE Systems

In this study, an IM serving as a SEIG with an appropriate capacitor size is utilized. Shaft velocity,
residual magnetism, lower permeability at lower magnetization, and the capacity of the linked
capacitor all affect SEIG production. By raising the rate, the rating of the capacitor, or both of them,
the self-excitation reliability has to be extremely high (Bu et al. 2020; Ion 2020). While the SEIG
produces active energy; capacitors supply reactive energy to the reactive load (QL ) and to the SEIG
for excitation. With a proper size for the capacitor, SEIG is initiated without any load. During reli-
ability assessments, this simulated SEIG condition has been taken into account (Saket et al. 2022;
Saket 2011; Varshney et al. 2013a, 2021; Varshney and Saket 2014). The minimum required termi-
nal capacitor, which is necessary for SEIG to withstand voltage deposits, was used. It is advised to
compare the reliability of various SEIGs using a small amount of capacitor.

18.4 SEIG Tests with Lowest Capacitive Excitation

The SEIG IM has the following specifications: 12.3 A, 3-phase, 50 Hz, 4-pole, 230 V, and 5.5 hp.
In terms of units, the IM’s per-phase equivalent variables are Rs = 0.04960, Rr = 0.03500,
Xs = 0.13440, and Xr = 0.13440. Figure 18.2 depicts the estimated magnetization gradient of the
IM. The capacitor’s estimated lowest possible value is 24.48 μF (Chaturvedi 2018).
The use of Vg/F for the evaluation machine is shown in Figure 18.3. IM working as a SEIG is
assessed through the examination of junction voltage at various primary mover speeds. Figure 18.4
displays the differences in produced delta linked no load terminals voltages with varied prime
mover speeds at 25 μF and 36 μF capacitors delta linked. As can be observed, the junction voltage’s
match between observed and calculated values is enough to support the reliability of the Khan and
Khan (2022) assessment.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
344 18 Probabilistic Distribution and Monte Carlo Approach for Reliability Evaluation

Experimental no load terminal voltage (V)

By equation no load terminal voltage (V)
350
300 1.0
300
250
250 0.8
200 200
0.6
150 150
100 100 0.4
50
50
Test at synchronous speed 1500 RPM 0.2
0
0
–50 0.0
0 1 2 3
Magnetization current (A)

Figure 18.2 IM magnetization graph without load.

Equation Vg/F (Volts)

Experimental Vg/F (Volts)

200 200 1.0

150 150 0.8

Vg/F (V)

0.6
100 100

0.4
50 50
0.2
0 0
0.0
70 80 90 100 110 120 130
Xm (Ω)

Figure 18.3 Vg/F variation with Xm.

18.5 Rotor Core Magnetization of SEIG Reliability Assessment

Using Least Capacitor Score

Degradation of excitation was only taken into account in this work when the SEIG’s production
failed. The primary reason for a breakdown of machine performance is thought to be the lack
of excitement. Throughout the testing phase, other SEIG system breakdown factors, such as gen-
erating failure caused by design/manufacturing flaws and operational factors, are not taken into
account (Saket and Varshney 2012; Varshney et al. 2016; Varshney and Saket 2013). Utilizing the
smallest size for the capacitor, the reliability of SEIG excitation was evaluated. Figure 18.5 lists the
evaluated dependability indices.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
18.5 Rotor Core Magnetization of SEIG Reliability Assessment Using Least Capacitor Score 345

25 μF-Experiment (Voltage)
36 μF-Simulation (Voltage)
36 μF-Experiment (Voltage)
300 25 μF-Simulation (Voltage) 300
260
280
260 240 280
240 220
260
Voltage (V)

220
200
200 240
180 180
160 160 220
140 140
120 200
120
0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15
Speed (p.u)

Figure 18.4 Change of the voltage at the terminal for without a load with respect to time as well as
capacitance.

18.5.1 Assessment of Reliability-Related Functions

Probability distribution curves are used here to characterize the reliability assessment of SEIG exci-
tation. In the beginning, SEIG (Varshney et al. 2013b) was the subject of over 32 tests on the first
day. It indicates a 32-fold rise in reactive burden. The overall number of tests was determined the
following day by adding up all successful tests and all unsuccessful ones from the preceding time
period (Kumar et al. 2021; Saket 2011). In tests, the total number of fails is tallied across seven
periods. Figure 18.5 describes the probability indices such as the hazard function, failure density
function, survival function and cumulative failure distribution. The following is the process for
evaluating dependability indices:
1. Computation for each interval’s overall failures. The time span is one day.
2. Assessment of reliability metrics include the risk rate, failed density function, and cumula-
tive failed distribution. Figures 18.6–18.9 show a graphical representation of each dependability
index, accordingly.

18.5.2 Probabilities of Failure and Success

In this work, the probability of failure and success have been assessed utilizing the MCS tech-
nique. The method used to assess the dependability indicators differs among the analytic and
simulation-based approaches (Kumar et al. 2021). Analytic approaches use a mathematical frame-
work for expressing the data, which is frequently streamlined and employ direct mathematical
approaches to assess the model’s reliability scores. However, MCS calculates the reliability indexes
by mimicking the system’s actual procedure and random characteristics. As a result, the technique
approaches the issue as a set of genuine tests carried out over a period of modeled time. By record-
ing the instances of an occurrence, it may calculate probability along with other indices. The setting
up for lab tests began with a least capacitance of 25 μF. Utilizing MCS, it is being practically tested if
the IM will successfully excite when used as a SEIG. In entirety, 71 experiments were run over the
course of 8 consecutive days. Figure 18.10 displays different test findings (success/failure excitation)
from a study on the SEIG.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
346 18 Probabilistic Distribution and Monte Carlo Approach for Reliability Evaluation

rval
each inte
30
25 Number of failures in each interval
Cumulative failures

failures in
Number of experiments
20 Falure density
Cummulative failure distribution
Survivor function
15 Hazard rate
10

f
Number o
5
0
Haza
r
Survi d rate
v
Cum or functio
m n
Failu ulative fa
re de ilure
n
7

Num sity distri

6

ber o butio
5

Tim Cum f exp n

e int
4

erva ulativ erim

3

l e e n t s
2

Num failu
1

ber o res
0

f fail
ures
in ea
ch in
terva
l

Figure 18.5 Data on actual failures and an assessment of reliability indexes.

Failure density function

0.5 0.47

0.4
Failure density function

0.35259

0.3

0.2 0.17733

0.1 0.0923
0.0477 0.04874 0.04874
0.01037
0.0
0 1 2 3 4 5 6 7
Time interval (Day)

Figure 18.6 Graph of a failed density function.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
18.5 Rotor Core Magnetization of SEIG Reliability Assessment Using Least Capacitor Score 347

Cumulative failure distribution curve

0.5
Cumulative failure distribution curve

0.4
0.32795
0.30572 0.3165 0.31448
0.3 0.28687
0.24175

0.2
0.13805

0.1

0.00202
0.0
0 1 2 3 4 5 6 7
Time interval (Day)

Figure 18.7 Distribution curve for cumulative failures.

Survivor function
0.5

0.4 0.37953
Survivor function

0.3

0.22078
0.2
0.1414
0.10171 0.10171
0.1 0.06946

0.0
0 1 2 3 4 5 6 7
Time interval (Day)

Figure 18.8 Curve for the survivor function.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
348 18 Probabilistic Distribution and Monte Carlo Approach for Reliability Evaluation

2.26212
Hazard rate

2.0

1.5
Hazard rate

1.0
0.73896 0.76912
0.69372 0.64847
0.5781
0.5 0.47253

0.03016 0.01488
0.0
0 1 2 3 4 5 6 7 8
Time interval (Day)

Figure 18.9 Hazard graph.

S-Success, F-Failure
F
Experiment result outcome

Probability success
Probability failure

me 70
o utco 60
lt 5
esu s 40 0 t
ent
r es 30 en
im ucc eri
m
per ys 20
Ex ilit re 10 xp
ab lu fe
P rob fai o.
o
ity N
abil
P rob

Figure 18.10 Results of the SEIG’s development of failures and successes.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
18.6 Discussion and Outcomes 349

18.6 Discussion and Outcomes

The SEIG’s chances of failure and success with the lowest capacitor size have been assessed utilizing
Monte Carlo modeling. The simulation parameters are displayed in Figures 18.11 and 18.12. The
simulation findings lead to the following conclusions:

1. SEIG experiments are run frequently, which improves the results in terms of failure and success
probabilities.
2. After enough testing, the chance of failure and success fluctuates about its true value. The true
value cannot be accurately estimated using the mean of the failure and success probabilities.
3. As the quantity of experiments rises, the estimation of the likelihood of failure and success tends
toward the true value.
4. Based on the outcomes of the simulations, the probabilities of failure and success were inclined
toward 0.55 and 0.45, correspondingly.

Employing a statistical approach, the probability of success (Ks) and failure (Kf) has been deter-
mined. The proportion of the quantity of successful/failed assessments to the total number of
results is the probability of success (Ks)/failure (Kf). The ratings are listed below.

Success probability Ks = s/J

Ks = 39∕71 = 0.549

Similarly,

Failure probability Kf = f/J

Kf = 32∕71 = 0.451

1.1 Probablity of success

1.0

0.9
Probablity of success

0.8

0.7

0.6

0.5

0.4

0.3
0 10 20 30 40 50 60 70 80
Number of experiment

Figure 18.11 Probability for a successful simulation using the Monte Carlo method.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
350 18 Probabilistic Distribution and Monte Carlo Approach for Reliability Evaluation

Probablity of failure
0.8

0.7

0.6
Probablity of failure

0.5

0.4

0.3

0.2

0.1

0.0

–0.1
0 10 20 30 40 50 60 70 80
Number of experiment

Figure 18.12 Chance of fail simulation using the Monte Carlo method.

18.7 Conclusion

Despite the least amount of capacitive stimulation, the SEIG’s performance dependability is noted.
This study correctly assesses the possibilities of failure and success of 3-phase, 5.5 hp, and 25 μF. The
analytical outcomes are Kf = 0.452 and Ks = 0.549. The results go toward Ks = 0.55 and Kf = 0.45
and are centered on MCS. The findings of the simulations as well as the analysis are nearly identical.
The outcomes of experiments have been used to derive the failure density slope, survival graph,
cumulative graph, and hazard graph. The area under the failure density graph is one, and also
production failure diminishes as the quantity of experiments is reduced.

References

Bu, F., Liu, H., Huang, W. et al. (2020). Induction-machine-based starter/generator systems:
techniques, developments, and advances. IEEE Industrial Electronics Magazine 14 (1): 4–19.
Chaturvedi, Y.K. (2018). A dissertation report on modelling of self excited induction generator for
minimum capacitance requirement.
Chaturvedi, Y. and Goel, A. (2019). Wind powered electricity generation through self excited induction
generator. World Scientific News 121: 90–100.
Ion, C.P. (2020). A comprehensive overview of single–phase self-excited induction generators. IEEE
Access 8: 197420–197430.
Khan, M.F. and Khan, M.R. (2022). Analysis of a nine-phase self excited induction generator equipped
with optimum excitation capacitances. In: 2022 International Conference for Advancement in
Technology (ICONAT), 1–6. IEEE.
Kumar, K., Varshney, L., Ambikapathy, A. et al. (2021). Solar tracker transcript—a review.
International Transactions on Electrical Energy Systems 31 (12): e13250.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 351

Nakorn, P., Machot, P., Kinnares, V., and Manop, C. (2021). Study of three-phase self-excited induction
generator operating as single-phase induction generator supplying non-linear load. In: 2021 18th
International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and
Information Technology (ECTI-CON), 806–809. IEEE.
Saket, R.K. (2011). Reliability evaluation of defence support systems. In: Innovations in Defence Support
Systems-2: Socio-Technical Systems (ed. L.C. Jain, E.V. Aidman, and C. Abeynayake), 241–286. Berlin,
Heidelberg: Springer-Verlag.
Saket, R.K. and Varshney, L. (2012). Self excited induction generator and municipal waste water based
micro hydro power generation system. International Journal of Engineering and Technology 4 (3): 282.
Saket, R.K., Kothari, D.P., and Nagrath, I.J. (2022). Modern Power System Analysis. Tata McGraw-Hill
Publishing Company.
Silva, F.B., da Silva Gonçalves, F.A., Vanço, W.E. et al. (2018). Application of bidirectional switches in
the development of a voltage regulator for self-excited induction generators. International Journal of
Electrical Power & Energy Systems 98: 419–429.
Varshney, L. and Saket, R.K. (2013). Reliability evaluation of residual magnetism in rotor of SEIG. In:
2013 Students Conference on Engineering and Systems (SCES), 1–4. IEEE.
Varshney, L. and Saket, R.K. (2014). Reliability evaluation of SEIG rotor core magnetization with
minimum capacitive excitation for unregulated renewable energy applications in remote areas.
Ain Shams Engineering Journal 5 (3): 751–757.
Varshney, L., Saket, R.K., and Eslamian, S. (2013a). Power estimation and reliability evaluation of
municipal waste water and self-excited induction generator-based micro hydropower generation
system. International Journal of Hydrology Science and Technology 3 (2): 176–191.
Varshney, L., Varshney, V., Newwel, A., and Saket, R.K. (2013b). Squirrel cage rotor design for safety
and reliability improvement of a three phase induction machine. In: Proceedings of the 3rd
International Conference on Trends in Information, Telecommunication and Computing, 267–273.
Springer.
Varshney, L., Gupta, C., and Debnath, D. (2016). Performance improvement of squirrel cage rotor by
design aspects. In: 2016 International Conference on Recent Advances and Innovations in Engineering
(ICRAIE), 1–4. IEEE.
Varshney, L., Vardhan, A.S.S., Vardhan, A.S.S. et al. (2021). Performance characteristics and reliability
assessment of self-excited induction generator for wind power generation. IET Renewable Power
Generation 15 (9): 1927–1942.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
353

19

Reliability and Mean Life Assessment of Solar Panel by Cooling

Rahul Agrawal 1 , Jyotsna Ogale 2 , Nga T. T. Nguyen 3 , R. K. Saket 4 , and Joydeep Mitra 5
1
Department of Electrical Engineering, Indian Institute of Technology (BHU) Varanasi, Uttar Pradesh, India
2 Department of Electronics Engineering, Samrat Ashok Technological Institute, Vidisha Madhya Pradesh, India
3
Department of Electrical Engineering and Computer Science, University of Wyoming, Wyoming, USA
4
Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
5
Department of Electrical Engineering, Michigan State University, Michigan, USA

19.1 Introduction
The world is widely harnessing solar power for its energy needs and decreasing its dependence on
diminishing fossil fuels. Solar PV panel works in the sunlight the higher the exposure to the sun,
the more energy the solar cell will produce. But as a consequence, the temperature of the Solar
cell rises due to rising in the ambient temperature and irradiance (Markvart 2000); due to the rise
in the temperature of the solar cell, its lifetime gets reduced (Hacke et al. 2015). So, solar panels
reliability and mean life are essential parameters for the more extensive acceptability of solar power
generation.
This work focuses on the environmental effects (temperature and humidity) on the reliability of
solar panels and how they affect the mean life of solar panels (Park et al. 2013; Luque and Hegedus
2011). Mean life and reliability were studied and compared with and without cooling of the panel
(Sharaf et al. 2022; Teo et al. 2012; Sohani et al. 2021; Mawoli et al. 2020). The environmental impact
on the solar cell tends to reduce the lifetime of the panel, which will decrease its reliability. As a part
of environmental impact, temperature and relative humidity are the prime factors that cause the
degradation of solar panels (Park et al. 2013; Chikate et al. 2015). To reduce the temperature impact,
various cooling methods are being introduced to improve the lifetime of solar panels (Hacke et al.
2015; Sohani et al. 2021). This research aims to compare the improvement in mean life and com-
pare the reliability of solar panels by cooling and without cooling (Park et al. 2013). The proposed
work studied the two degradation models, namely, Erying and Peck, to calculate the degradation
after cooling. The cooling methods for which the reliability assessment is done are water cool-
ing (Moharram et al. 2013), Air cooling using extended fin area (Ravi and Porpatham 2017), Wick
structure with nanofluids using CuO/water nanofluids (Kusiak et al. 2013), and passive cooling by
heat pipe (Akbarzadeh and Wadowski 1996). Corresponding to the mean life obtained from these
models, a reliability study of solar panels is done, and the reliability comparison for without cooling
(Teo et al. 2012) and cooling (considering the Erying and Peck model) (Park et al. 2013; Firoozzadeh
et al. 2019) is done.

Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
354 19 Reliability and Mean Life Assessment of Solar Panel by Cooling

It is observed that the output will be maximum if the irradiance falling on the PV surface is maxi-
mum. As a consequence of high irradiance, cell temperature also increases as irradiance increases,
reducing the lifetime of the solar cell. It is observed that the efficiency improvement of solar PV
systems by applying cooling techniques (Irwan et al. 2015; Cuce et al. 2011) used to be the prime
focus of research. At the same time, reliability is also a prime factor that should be considered
after using the cooling methods (Firoozzadeh et al. 2019; Nižetić et al. 2017). The reliability of
subsystems of solar cells without cooling is calculated in (Makrides et al. 2018) using statistical
techniques Ordinary Least Squares (OLS), Classical Seasonal Decomposition, and the Year-on-Year
(YOY) comparative technique to obtain the degradation rates (Meeker et al. 2022; Aminisharifabad
et al. 2021). As observed from the literature survey, some of the cooling methods and the parameters
they found are tabulated in Table 19.1.
In (Moharram et al. 2013), it is calculated that using the water-cooling method for a five-minute
operation of the cooling system is capable of reducing the solar cell temperature by 10 ∘ C and
increasing the efficiency by 12.5%. The wet cotton wick formation that is placed on the back sur-
face of the panel is estimated to cool the solar panel due to the capillary action of the cotton wick
(Kusiak et al. 2013).
For the work to be carried out, data for the location in central India, longitude 23.181 and latitude
79.986, that is for the city Jabalpur in Madhya Pradesh, is taken. This data contains atmospheric
temperature, relative humidity, and solar irradiance in one-hour intervals taken from March 3,
2021, to April 27, 2022. The plots for atmospheric temperature and relative humidity are shown in
Figures 19.1 and 19.2.

Table 19.1 Summary of improvements by cooling.

Researcher Cooling methods Temperature reduction Parameters calculated

Moharram et al. Water cooling Significant Efficiency improved by 12.5% for

(2013) a temperature reduction of 10 ∘ C
Ravi and Air cooling with 16 ∘ C Adding fins increases thermal
Porpatham (2017) extended fins area efficiency to 15.5% and electrical
efficiencies to 10.5%
Kusiak et al. Wick structure was Up to 11% using Maximum efficiency attained is
(2013) placed in the back CuO/water and 17% 10.4%
panel with by Al2 O3 /water
nanofluids
Akbarzadeh and Passive cooling Significant Passive cooling with a heat pipe
Wadowski (1996) with heat pipe can reduce the temperature by
38 ∘ C
Hasan et al. (2017) PCM-RT42 Annual avg. by Electrical output incremented by
10.5 ∘ C 5.9%
Karthick et al. PCM-eutectic Maximum 9 ∘ C Average energy gains 103.6 Wh
(2020)
Salem et al. (2019) PCM – a mixture of Significant Efficiency increments by 10.9% to
water and Al2 O3 22.7%
Tana et al. (2017) Organic paraffin By 15 ∘ C Efficiency improved by 5.39%
wax

Source: Sohani et al. 2021; Mawoli et al. 2020; Moharram et al. 2013; Karthick et al. 2020; Dwivedi et al. 2020; Salem
et al. 2019.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19.2 Methodology 355

317
312
Atmospheric temperature in °K

307
302
297
292
287
282
277
3/31/2021 7/29/2021 11/26/2021 3/26/2022
Date

Figure 19.1 Atmospheric temperature in ∘ K at longitude 23.181 and latitude 79.986. Source: Anonymous
(n.d.). Adapted from Live and Forecast API.

120

100
Relative humidity

80

60

40

20

0
1

1
21

21

21

28 22

31 2
2
02

02

02

02

02

02

02

02

02
20

20

20

0
/2

/2

/2

/2

/2

/2

/2

/2

/2

/2
1/

0/

1/
31

30

31

30

31

31

30

31
/3

/3

/3
3/

4/

5/

6/

7/

8/

9/

1/

2/

3/
10

11

12

Date

Figure 19.2 Relative humidity at longitude 23.181 and latitude 79.986. Source: Anonymous (n.d.). Adapted
from Live and Forecast API.

Only 4–17% of incoming solar radiation is typically converted into electricity by PV modules.
The type of solar cells used, as well as the working conditions, influence conversion efficiency. As a
result, more than half of the incoming solar power is transformed to heat. Short term losses are
caused by an increase in cell temperature. A reduction in electrical rate of returns, and reduces
reliability and solar panel efficiency too (Moharram et al. 2013).

19.2 Methodology

The workflow of the proposed work starts with defining a location to work upon, followed by the
atmospheric condition in that location, which affects the change in solar panel parameters, and
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
356 19 Reliability and Mean Life Assessment of Solar Panel by Cooling

Obtain atmospheric temperature and relative humidity at desired

location

Obtain Solar irradiance at desired location

Calculation of Cell temperature without cooling and with cooling

Calculation of Degradation using Erying and Peck models

Calculation of mean life

Calculation of failure rate

Calculation of reliability, PDF and CDF using the failure rate

obtained

Comparison of reliability with cooling and without cooling

Figure 19.3 Flowchart of proposed work.

the degradation through those parameters is calculated, followed by the calculation of failure rates,
mean life and reliability (Meeker et al., 2022). The flowchart of the work is shown in Figure 19.3.

19.2.1 Cell Temperature

The nominal operating conditions temperature (NOCT) and standard temperature conditions
(STC), which depend upon the atmospheric temperature, can be estimated by using the equation
for the temperature of the PV cell. The equation of cell temperature is shown below (Markvart
2000)
NOCT − 20
Tcel = Tamb + ∗S
800
– T cel – Temperature of solar cell
– T amb – Ambient temperature of the atmosphere
– NOCT – nominal operating conditions temperature
where at STC, the temperature is 20 ∘ C, and the irradiance is 800 W/m2 , while at NOCT, NOCT
is given by the solar cell manufacturer, and it lies in the range 45 ± 2 ∘ C (Sun et al. 2020).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19.2 Methodology 357

345
Cell temperature in °K

335
325
315
305
295
285
275
30 1

31 1

30 1

31 1

31 1

30 1
10 021

21

31 1
2/ 022

31 2
2
2
02

02

02

02

02

11 02

02

02
0

20

20
/2

/2

/2

/2

/2

/2

/2

/2

/2

/2
1/

0/

1/
31

28
/3

/3

/3
3/

4/

5/

6/

7/

8/

9/

1/

3/
12
Date

Figure 19.4 Cell temperatures in ∘ K without cooling.

The temperature is taken 46 ∘ C according to requirement of the experimental setup. In the ideal
condition, the maximum irradiance that can hit the PV surface is 1000 W/m2 . So, from the
atmospheric temperature and irradiance from March 31, 2021, to April 27, 2022, the temperature
of the solar cell is calculated by putting the values in equation A. The cell temperature comes out
to be as shown in Figure 19.4.

19.2.2 Degradation Equations

This work is performed considering two degradation models:

i. Erying
ii. Peck

Erying degradation model (Park et al. 2013; David 2002)

( )
E b
RD Erying = A exp − act − (19.1)
kB T rh
where

S – Solar irradiance in W/m2

Eact – Thermal activation energy of the degradation Process (eV)
K B – Boltzmann constant (8.62* 10-5 eV/K)
T – Temperature (∘ K)
rh – Relative humidity (%)
A and b – Two constants dependent on the failure mode
RD (%/h) – Inverse to the meantime to failure (MTTF).

According to the Peck model (Park et al. 2013),

The degradation rate is given as,
( )
−Eact
(rhn )
RD Peck = Be KB T
(19.2)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
358 19 Reliability and Mean Life Assessment of Solar Panel by Cooling

where

B and n – Constants dependent on failure mode

19.2.3 Mean Life Without Cooling

The mean life of solar panels is 25 years without cooling (Sharaf et al. 2022; Kim et al. 2021; Guérin
et al. 2006). Therefore for the without cooling temperature in this work, the mean life equation is
optimized such that to find constants. Since, the solar data from March 31, 2021, to April 27, 2022,
contains values per hour, so there are 9430 of the total values.
By the Erying method using Eq. (19.2), mean life is
∑
9430
RD Eryingi = RD Erying1 + RD Erying2 + RD Erying3 + … + RD Erying9430
i=1
9430
Mean life = ∑9430
i=1 RDi

This equation is optimized using the SciPy library method, which contains the SLSQP optimiza-
tion technique.
To maximize:
9430
Mean life = ∑9430
i=1 RD Eryingi

Constraints:
9430
∑9430 ≤ 25
i=1 RD Eryingi

A = (2 ∗ 106 , 4.5 ∗ 106 )

b = (1, 4)

Such that the constants A and b obtained are:

A = 4.00∗ 106

b = 3.02793706

Similarly, for Peck equation, as mentioned in Eq. (19.3),

∑
9430
RD Pecki = RD Peck1 + RD Peck2 + RD Peck3 + … + RD Peck9430
i=1
9430
mean life = ∑9430
i=1 RD Pecki

This equation is optimized using the SciPy library method, which contains the SLSQP optimiza-
tion technique.
To maximize:
9430
mean life = ∑9430
i=1 RD Pecki
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19.2 Methodology 359

Constraints:
9430
∑9430 ≤ 25
i=1 RD Pecki

B = (1 ∗ 106 , 10 ∗ 106 )
n = (10−6 , 10)
Such that the constants B and n obtained are:
B = 3.4789 ∗ 106
n = 1.9∗ 10−6

19.2.4 Mean Life After Cooling

For the calculation of mean life using four cooling techniques that are water cooling (Moharram
et al. 2013), Air cooling using extended fin area (Ravi and Porpatham 2017), Wick structure with
nanofluids Using Al2 O3 /water and CuO/ water nanofluids (Kusiak et al. 2013), Passive cooling by
heat pipe (Akbarzadeh and Wadowski 1996), the mean life is calculated in three different cases:
when cooling is performed to cool cell temperature not less than 25 ∘ C (298 ∘ K), cooling is per-
formed to cool cell temperature not less than 35 ∘ C (308 ∘ K), cooling is performed to cool cell
temperature not less than 45 ∘ C (318 ∘ K). Which implies that with the constraints of reducing the
temperature for the cooling methods, the temperature above 25, 35, and 45 ∘ C, the cooling will be
done as per the capability of technique, but for the cell temperature less than these temperatures
naturally, then the cooling method will not reduce the temperature further.

19.2.4.1 Mean Life with Cooling up to 25 ∘ C

When water cooling (Moharram et al. 2013), air cooling using extended fin area (Ravi and
Porpatham 2017), Wick structure with nanofluids Using Al2 O3 /water and CuO/ water nanofluids
(Kusiak et al. 2013), passive cooling by heat pipe (Akbarzadeh and Wadowski 1996), cooling
methods are applied such that to bring down the temperature not less than 25 ∘ C that is 298 ∘ K.
For different cooling techniques taken into consideration and applying cooled temperature
according to constraints of those cooling techniques, the following Figures 19.5–19.8 are obtained.
These figures compare the cell temperature before and after applying the cooling techniques.

345
Tcell Tcell wc
335
Temperature in °K

325
315
305
295
285
275
30 1

31 1

30 1
1

30 1
10 021

11 021

12 021

31 1
2/ 022

2
02

02

02

02

02

02

02

02
20
/2

/2

/2

/2

/2

/2

/2

/2

/2

/2
1/

0/

1/
31

31

31

28

31
/3

/3

/3
3/

4/

5/

6/

7/

8/

9/

1/

3/

Date

Figure 19.5 Cell temperature after water cooling.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
360 19 Reliability and Mean Life Assessment of Solar Panel by Cooling

345 Tcell Tac

Cell temperature in °K

335
325
315
305
295
285
275
30 1
1

30 1
1

30 1
10 021

11 021

12 021

31 1
2/ 022

2
02

02

02

02

02

02

02

02
20
/2

/2

/2

/2

/2

/2

/2

/2

/2

/2
1/

0/

1/
31

31

31

31

28

31
/3

/3

/3
3/

4/

5/

6/

7/

8/

9/

1/

3/
Date

Figure 19.6 Cell temperature after air cooling with extended fins.

345 Tcell Tws

Cell temperature in °K

335
325
315
305
295
285
275
1

21

21

21

28 2

31 2
2
02

02

02

02

02

02

02

02

02

02
20

20

20
/2

/2

/2

/2

/2

/2

/2

/2

/2

/2
1/

0/

1/
31

30

31

30

31

31

30

31
/3

/3

/3
3/

4/

5/

6/

7/

8/

9/

1/

2/

3/
10

11

12

Date

Figure 19.7 Cell temperature after cooling by wick structure.

Using the cooled temperature and putting it in Eq. (19.1) of Eyring and Eq. (19.2) of Peck, to
obtain the mean life after cooling and also the failure rate is obtained, the failure rate is reciprocal
of mean life for a constant failure rate function (Ebeling 2004), as shown in Table 19.2.

19.2.4.2 Mean Life with Cooling up to 35 ∘ C

For different cooling techniques taken into consideration and applying cooled temperature up to
35 ∘ C and according to constraints of those cooling techniques, the following Figures 19.9–19.12
are obtained. The estimated data are shown in Table 19.3 for graphical illustration. These figures
compare the cell temperature before and after applying the cooling techniques.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19.2 Methodology 361

345 Tcell Tpc

335

325
Cell temperature in °K

315

305

295

285

275
1

21

21

21

2
02

02

02

02

02

02

02

02

02

02
20

20

20
/2

/2

/2

/2

/2

/2

/2

/2

/2

/2
1/

0/

1/
31

30

31

30

31

31

30

31

28

31
/3

/3

/3
3/

4/

5/

6/

7/

8/

9/

1/

2/

3/
10

11

12

Date

Figure 19.8 Cell temperature after cooling with passive cooling heat pipe.

Table 19.2 Mean life and failure rate calculated when cooled up to 25 ∘ C.

Failure rate (𝝀)

Cooling technique Degradation equation Mean life failures per year

Without cooling 25 0.04

Water cooling Erying 59.6170 0.01677
Peck 61.5025 0.01625
Air cooling extended fin Erying 44.6956 0.02237
Peck 44.5812 0.02243
Wick surface Erying 58.9043 .01698
Peck 60.5570 .01651
Passive cooling heat pipe Erying 58.8606 0.01698
Peck 60.5322 0.01652

19.2.4.3 Mean Life with Cooling up to 45 ∘ C

For different cooling techniques taken into consideration and applying cooled temperature up to
45 ∘ C and according to constraints of those cooling techniques, the following Figures 19.13–19.16
 362
Cell temperature in °K 3/ Cell temperature in °K
Cell temperature in °K 3/
31 31
/2

275
285
295
305
315
325
335
0 345
3/ /2

275
285
295
305
315
325
335
345
31 2

Figure 19.9
02 4/
/2

Figure 19.11
Figure 19.10
4/ 1 30 1

275
285
295
305
315
325
335
345
30 /2
4/ 021 /2 02
30 02 5/
/2 5/
31 1 31 1
5/ 021 /2 /2
31 02 6/
02
/2 6/
30 1 30 1
6/ 021 /2 /2
30 02 02
/2 7/ 1 7/
31 1
31
7/ 021 /2 /2
31 02 02
/2 8/ 1 8/
8/ 021 31 31 1
31 /2 /2
/2 02 02
9/
30 1 9/
30 1
9/ 021 /2 /2
30 02
/ 10
/3 1 10 021
10 202 /3
1 1/

Tcell
/3 20 1/
2

Cell temperature after water cooling.

1/ 11 Date
2

Date
21 11 02
0 /3 1

Date
Tcell
11 0/ /3
/3 21
Series1

20 0/
0/ 12 21 20
/3 12 21
19 Reliability and Mean Life Assessment of Solar Panel by Cooling

12 202 /3
/3 1 1/
20 1/
1/ 20

Cell temperature after cooling by wick structure.

20 1/ 21 2

Tws
1/ 2 31 1/
/2 Tac
31 1

Cell temperature after air cooling with extended fin.

31 1 0 /2
/2 2/
28 22 2/ 022
Twc

2/ 022 /2 28
28 /2
/2 3/ 022 02
3/ 022 31 3/
31 /2 31 2
/2 02 /2
02 2 02
2 2

Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19.2 Methodology 363

345 Tcell Tpc

335
Cell temperature in °K

325
315
305
295
285
275
4/ 021

5/ 021

6/ 021

7/ 021

8/ 021

9/ 021

31 1
2/ 022

3/ 022

2
10 202

11 202

12 202

02
20
/2

/2

/2

/2

/2

/2

/2

/2

/2
/
1/

0/

1/
31

30

31

30

31

31

30

28

31
/3

/3

/3
3/

1/
Date

Figure 19.12 Cell temperature after cooling by passive cooling heat pipe.

Table 19.3 Mean life and failure rate calculated when cooled up to 35 ∘ C.

Cooling technique Degradation equation Mean life Failure rate (𝝀)

Without cooling 25 0.04

Water cooling Erying 42.2947 0.02364
Peck 43.1613 0.02317
Air cooling extended fin Erying 37.5643 0.02662
Peck 37.6175 0.02658
Wick surface Erying 42.2947 0.02346
Peck 43.1613 0.02317
Passive cooling heat pipe Erying 42.2943 0.02364
Peck 43.1605 0.02317

345 Tcell Twc

Cell temperature in °K

335
325
315
305
295
285
275
4/ 021

31 1
6/ 021

31 1

31 1
9/ 021

1
21

31 1
2/ 022

31 2
2
02

02

02

10 02

11 02

02

02
20

20
/2

/2

/2

/2

/2

/2

/2

/2

/2

/2
1/

0/

1/
31

30

30

30

28
/3

/3

/3
3/

5/

7/

8/

1/

3/
12

Date

Figure 19.13 Cell temperature after water cooling.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
364 19 Reliability and Mean Life Assessment of Solar Panel by Cooling

345 Tcell Tac

Cell temperature in °K

335
325
315
305
295
285
275
4/ 021

31 1
6/ 021

31 1

31 1
9/ 021

1
21

31 1
2/ 022

31 2
2
02

02

02

10 02

11 02

02

02
20

20
/2

/2

/2

/2

/2

/2

/2

/2

/2

/2
1/

0/

1/
31

30

30

30

28
/3

/3

/3
3/

5/

7/

8/

1/

3/
12
Date

Figure 19.14 Cell temperature after air cooling with extended fins.

345 Tcell Tws

Cell temperature in °K

335
325
315
305
295
285
275
1

1
21

21

21

28 22

31 2
2
02

02

02

02

02

02

02

02

02
20

20

20

0
/2

/2

/2

/2

/2

/2

/2

/2

/2

/2
1/

0/

1/
31

30

31

30

31

31

30

31
/3

/3

/3
3/

4/

5/

6/

7/

8/

9/

1/

2/

3/
10

11

12

Date

Figure 19.15 Reliability comparison of reliabilities of cooling method calculated by Erying and without
cooling.

345 Tcell Tpc

Cell temperature in °K

335
325
315
305
295
285
275
30 1

31 1
6/ 021

31 1

31 1
9/ 021

1
21

31 1
2/ 022

31 2
2
02

02

02

02

10 02

11 02

02

02
20

20
/2

/2

/2

/2

/2

/2

/2

/2

/2

/2
1/

0/

1/
31

30

30

28
/3

/3

/3
3/

4/

5/

7/

8/

1/

3/
12

Date

Figure 19.16 Reliability comparison of reliabilities of cooling method calculated by Peck and without
cooling.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19.3 Reliability Assessment 365

Table 19.4 Mean life and failure rate calculated when cooled up to 45 ∘ C.

Failure rate (𝝀)

Cooling technique Degradation equation Mean life (years) (failures per year)

Without cooling 25 0.04

Water cooling Erying 33.2692 0.03006
Peck 33.5838 0.02978
Air cooling extended fin Erying 32.2318 0.03102
Peck 32.3162 0.03094
Wick surface Erying 33.2692 0.03006
Peck 33.5838 .02978
Passive cooling heat pipe Erying 33.2692 0.3006
Peck 33.5838 0.02978

are obtained. The estimated data are shown in Table 19.4 for graphical illustration. These figures
compare the cell temperature before and after applying the cooling techniques.

19.3 Reliability Assessment

The probability of a system or the components of a system to perform a desired task or function for
a provided period of time when used under specified operating conditions. It is the probability of
not failing for a given period of time (Ebeling 2004). Mathematically, it can be expressed as for any
random time T is the time of failure; T ≥ 0. Then, the reliability is given as:

R(t) = Py (T ≥ t)

where

R(t) – reliability function

Py – probability
R(t) ≥ 0
R(0) = 1

Considering that system is not failing at T = 0

lim R(t) = 0
t→∞

and reliability will lie in the range R(t) ∈ [0, 1]R(t) is the probability that the time to fail is greater
than or equal to t.
Considering the constant failure rate (Ebeling 2004), therefore taking the exponential distribu-
tion curve, Therefore,

R(t) = e−𝜆t

where 𝜆 is the failure rate (Ebeling 2004), the reliability equations calculated are shown in
Table 19.5.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
366 19 Reliability and Mean Life Assessment of Solar Panel by Cooling

Table 19.5 Reliability R(t) calculations.

Reliability for Reliability for Reliability for

Degradation cooling up to cooling up to cooling up to
Cooling equation 25 ∘ C 35 ∘ C 45 ∘ C

Without cooling e−0.04t e−0.04t e−0.04t

−0.01677t −0.02364t
Water cooling Erying e e e−0.03006t
Peck e−0.01625t e−0.02317t e−0.02978t
−0.02237t −0.02662t
Air cooling with extended fins Erying e e e−0.03102t
Peck e−0.02243t e−0.02658t e−0.03094t
−0.01698t −0.02364t
Wick surface Erying e e e−0.03006t
Peck e−0.01651t e−0.02317t e−0.02978t
−0.01698t −0.02364t
Passive cooling heat pipe Erying e e e−0.03006t
Peck e−0.01652t e−0.02317t e−0.02978t

19.3.1 Cooling up to 25 ∘ C
As observed from Figure 19.17 by the Erying degradation equation, the reliability with cooling
improves significantly with all four cooling techniques; reliability improvement with water cooling,
wick structure, and passive cooling with heat pipe are similar, but with air cooling extended fins
cooling reliability, curve lies below from the curve of other cooling which implies that the reliability
improvement using air cooling with extended fins is less than the others.
As observed from Figure 19.18, the reliability with cooling improves significantly; reliability
improvement with water cooling, wick structure, and passive cooling with heat pipe are similar,
but with air cooling extended fins cooling reliability, improvement is less than the others.

1
0.9
0.8
0.7
0.6
Reliability

0.5
0.4
0.3
0.2
0.1
0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
R(t) Ewater cooling R(t) Eaf R(t) Ews
R(t) Epc R(t) without cooling

Figure 19.17 Cell temperature after cooling by wick structure.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19.3 Reliability Assessment 367

1
0.9
0.8
0.7
0.6
Reliability

0.5
0.4
0.3
0.2
0.1
0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
R(t) Pwater cooling R(t) Paf R(t) Pws
R(t) Ppc R(t) without cooling

Figure 19.18 Cell temperature after cooling by passive cooling heat pipe.

19.3.2 Cooling up to 35 ∘ C
As observed in Figure 19.19, the reliability with cooling improves significantly. Reliability improve-
ment with water cooling, wick structure, and passive cooling with heat pipe are similar, but with
air cooling, extended fins cooling improvement is less than the others. Up to 35 ∘ C cooling, the
reliability difference becomes lesser than 25 ∘ C cooling. For both Erying and Peck equations (by
Figure 19.20), a similar observation is made.

1
0.9
0.8
0.7
Reliability

0.6
0.5
0.4
0.3
0.2
0.1
0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
R(t) Ewater cooling R(t) Eaf R(t) Ews R(t) Epc R(t)

Figure 19.19 Reliability comparison of reliabilities of cooling method calculated by Erying and without
cooling.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
368 19 Reliability and Mean Life Assessment of Solar Panel by Cooling

1
0.9
0.8
0.7
0.6
Reliability

0.5
0.4
0.3
0.2
0.1
0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
R(t) Pwater cooling R(t) Paf R(t) Pws R(t) Ppc R(t)

Figure 19.20 Reliability comparison of reliabilities of cooling method calculated by Peck and without
cooling.

19.3.3 Cooling up to 45 ∘ C
Up to 45 ∘ C cooling, the reliability improvement is lesser than 25 and 35 ∘ C cooling but still signif-
icant, as seen by the Erying equation in Figure 19.21. And in this case, all four cooling methods’
reliability curves overlap.

1
0.9
0.8
0.7
0.6
Reliability

0.5
0.4
0.3
0.2
0.1
0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
R(t) Ewater cooling R(t) Erying air fin R(t) Ews R(t) Epc R(t)

Figure 19.21 Reliability comparison of reliabilities of cooling method calculated by Erying and without
cooling.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19.4 Probability Density Function 369

1
0.9
0.8
0.7
0.6
Reliability

0.5
0.4
0.3
0.2
0.1
0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
R(t) Pwater cooling R(t) Paf R(t) Pws R(t) Ppc R(t)

Figure 19.22 Reliability comparison of reliabilities of cooling method calculated by Peck and without
cooling.

Observation from the Peck equation is also similar to Erying, as observed in Figure 19.22. The
distance between cooling and without curves reduces but is still significant. And in this case, all
four cooling methods’ reliability curves overlap.

19.4 Probability Density Function

This PDF function describes the shape of the failure distribution (Ebeling 2004).
Where f PDF (t) – Probability density function (PDF).
dFCDF (t) dR(t)
fPDF (t) = =−
dt dt
The range of PDF is f (t) ∈ [0, ∞)
∞
fPDF (t)dt = 1
∫0
For constant failure rate

fPDF (t) = 𝜆e−𝜆t

The equations formed are shown in Table 19.6.

19.4.1 Cooling up to 25 ∘ C
The area under the probability distribution curve till time t shows the probability of failure until
time “t” (Ebeling 2004). As seen in Figures 19.23 and 19.24, up to any time “t,” the PDF curve
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
370 19 Reliability and Mean Life Assessment of Solar Panel by Cooling

Table 19.6 Probability density function f (t) calculations.

PDF for PDF for PDF for

Degradation cooling cooling cooling
Cooling equation up to 25 ∘ C up to 35 ∘ C up to 45 ∘ C

Without cooling 0.04 * e−0.04t 0.04000 * e−0.04000t 0.04 * e−0.04t

−0.01677t −0.02364t
Water cooling Erying 0.01677 * e 0.02364 * e 0.03006 * e−0.03006t
Peck 0.01625 * e−0.01625t 0.02317 * e−0.02317t 0.02978 * e−0.02978t
−0.02237t −0.02662t
Air cooling with Erying 0.02237 * e 0.02662 * e 0.03102 * e−0.03102t
extended fins Peck 0.02243 * e−0.02243t 0.02658 * e−0.02658t 0.03094 * e−0.03094t
−0.01698t −0.02364t
Wick surface Erying 0.01698 * e 0.02364 * e 0.03006 * e−0.03006t
Peck 0.01651 * e−0.01651t 0.02317 * e−0.02317t 0.02978 * e−0.02978t
−0.01698t −0.02364t
Passive cooling heat Erying 0.01698 * e 0.02364 * e 0.03006 * e−0.03006t
pipe Peck 0.01652 * e−0.01652t 0.02317 * e−0.02317t 0.02978 * e−0.02978t

0.04

0.035

0.03

0.025
PDF

0.02

0.015

0.01

0.005

0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
PDF Ewater cooling PDF Eaf PDF Ews
PDF Epc PDF without cooling

Figure 19.23 PDF plot for the various cooling methods by Erying equation for cooling up to 25 ∘ C.

without cooling will have an area more than that of any of the four cooling methods, which implies
that the probability of failure is higher without cooling, which le it is also seen for air cooling with
fins cooling method has a higher area; therefore, it will have a higher chance of failure than the
other three cooling techniques.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19.5 Cumulative Distribution Function 371

0.04

0.035

0.03

0.025
PDF

0.02

0.015

0.01

0.005

0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
PDF Pwater cooling PDF Paf PDF Pws PDF Ppc PDF without cooling

Figure 19.24 PDF plot for the various cooling methods by Peck equation for cooling up to 25 ∘ C.

19.4.2 Cooling up to 35 ∘ C
From Figures 19.25 and 19.26, as the cooled temperature increased to 35 ∘ C from 25 ∘ C, the cooling
curves overlapping each other, as well as they became closer to the PDF curve without cooling,
implying that with failure probability of 35 ∘ C cooling is more than that of 25 ∘ C cooling.

19.4.3 Cooling up to 45 ∘ C
From Figures 19.27 and 19.28, it is observed that all the PDF curves of cooling overlap each other,
which implies that the failure probability with all four types of cooling will be approximately the
same, as well as the failure probability of 45 ∘ C, will be higher than both 25 and 35 ∘ C cooling. From
time 0 to any time “t,” the area under the curve of PDF gives the failure probability, while the area
under the curve from time t to infinity defines the reliability of the system or component. So, if
the intersection of two curves is considered as without cooling and with air cooling, then the area
under the curve before the intersection time shows that the failure probability without a cooling
system is higher than that with cooling techniques and after the intersection time it is seen that
area under the curve of without cooling is less which implies that the reliability of with cooling is
lower. Similar results are seen in Figures 19.23–19.28.

19.5 Cumulative Distribution Function

The cumulative distribution function (CDF) is the probability up to any time T, which is in between
the zero time and the maximum time. CDF does signify that the failure occurs before time t (Ebeling
2004):
FCDF (t) = 1 − R(t) = Py(T < t)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
372 19 Reliability and Mean Life Assessment of Solar Panel by Cooling

0.04

0.035

0.03

0.025
PDF

0.02

0.015

0.01

0.005

0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
PDF Ewater cooling PDF Eaf PDF Ews PDF Epc PDF

Figure 19.25 PDF plot for the various cooling methods by Erying equation for cooling up to 35 ∘ C.

0.04

0.035

0.03

0.025
PDF

0.02

0.015

0.01

0.005

0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year

PDF Pwater cooling PDF Paf PDF Pws PDF Ppc PDF

Figure 19.26 PDF plot for the various cooling methods by Peck equation for cooling up to 35 ∘ C.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19.5 Cumulative Distribution Function 373

0.04

0.035

0.03

0.025
PDF

0.02

0.015

0.01

0.005

0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
PDF Ewater cooling PDF Eaf PDF Ews PDF Epc PDF

Figure 19.27 PDF plot for the various cooling methods by Erying equation for cooling up to 45 ∘ C.

0.04

0.035

0.03

0.025
PDF

0.02

0.015

0.01

0.005

0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
PDF Pwater cooling PDF Paf PDF Pws PDF Ppc PDF

Figure 19.28 PDF plot for the various cooling methods by Peck equation for cooling up to 45 ∘ C.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
374 19 Reliability and Mean Life Assessment of Solar Panel by Cooling

Table 19.7 Cumulative distribution function F_CDF(t) calculations.

CDF for CDF for CDF for

Degradation cooling cooling cooling
Cooling equation up to 25 ∘ C up to 35 ∘ C up to 45 ∘ C

Without cooling 1 − e−0.04t 1 − e−0.04t 1 − e−0.04t

−0.01677t −0.02364t
Water cooling Erying 1−e 1−e 1 − e−0.03006t
Peck 1 − e−0.01625t 1 − e−0.02317t 1 − e−0.02978t
−0.02237t −0.02662t
Air cooling with Erying 1−e 1−e 1 − e−0.03102t
extended fins Peck 1 − e−0.02243t 1 − e−0.02658t 1 − e−0.03094t
−0.01698t −0.02364t
Wick surface Erying 1−e 1−e 1 − e−0.03006t
Peck 1 − e−0.01651t 1 − e−0.02317t 1 − e−0.02978t
−0.01698t −0.02364t
Passive cooling heat Erying 1−e 1−e 1 − e−0.03006t
pipe Peck 1 − e−0.01652t 1 − e−0.02317t 1 − e−0.02978t

where
F CDF (t) – CDF

FCDF (0) = 0

lim FCDF (t) = 1

t→∞

Implies the range of CDF F(t) ∈[0, 1]

F(t) is the probability that a failure occurs before time t.
For constant failure rate,
FCDF (t) = 1 − e−𝜆t
The equations for PDF are shown in Table 19.7.

19.5.1 Cooling up to 25 ∘ C
From Figures 19.29 and 19.30, it seems that the chance of occurrence of failure before any time is
greater without cooling compared to cooling. The chance of failure of air cooling with fins also has
a higher chance of failure than the other three cooling methods.
All the other three methods have their curve overlapping, implying that the chance of failure for
water cooling, wick structure cooling, and passive cooling with heat pipe is nearly the same.

19.5.2 Cooling up to 35 ∘ C
From Figures 19.31 and 19.32, it is observed that the difference between the chance of failure with-
out cooling and with cooling is less than 25 ∘ C cooling but still significant. Also, the chance of
failure of air cooling is higher than in other cooling methods.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19.5 Cumulative Distribution Function 375

1
0.9
0.8
0.7
0.6
CDF

0.5
0.4
0.3
0.2
0.1
0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year

CDF Ewater cooling CDF Eaf CDF Ews CDF Epc CDF without cooling

Figure 19.29 CDF plot for the various cooling methods by Erying equation for cooling up to 25 ∘ C.

1
0.9
0.8
0.7
0.6
CDF

0.5
0.4
0.3
0.2
0.1
0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year

CDF Pwater cooling CDF Paf CDF Pws CDF Ppc CDF without cooling

Figure 19.30 CDF plot for the various cooling methods by Peck equation for cooling up to 25 ∘ C.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
376 19 Reliability and Mean Life Assessment of Solar Panel by Cooling

0.9

0.8

0.7

0.6
CDF

0.5

0.4

0.3

0.2

0.1

0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
CDF Ewater cooling CDF Eaf CDF Ews CDF Epc CDF

Figure 19.31 CDF plot for the various cooling methods by Erying equation for cooling up to 35 ∘ C.

0.9

0.8

0.7

0.6
CDF

0.5

0.4

0.3

0.2

0.1

0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
CDF Pwater cooling CDF Paf CDF Pws CDF Ppc CDF

Figure 19.32 CDF plot for the various cooling methods by Peck equation for cooling up to 35 ∘ C.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19.5 Cumulative Distribution Function 377

19.5.3 Cooling up to 45∘ C

From Figures 19.33 and 19.34, it can be seen that the chance of failure for all the cooling methods is
nearly similar and is less than that of a chance of failure without cooling, but the chance of failure
for 45 ∘ C is more than that of 25 and 35 ∘ C. So, the results obtained from CDF are just another
representation of the probability of failure observed by the area of the PDF till time t. By all the
three entities measured, that is, reliability, PDF, and CDF, the more cooling is done, the more the
lifetime and the probability of failure is high if less cooling is done.

1
0.9
0.8
0.7
0.6
CDF

0.5
0.4
0.3
0.2
0.1
0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
CDF Ewater cooling CDF Eaf CDF Ews CDF Epc CDF

Figure 19.33 CDF plot for the various cooling methods by Erying equation for cooling up to 45 ∘ C.

1
0.9
0.8
0.7
0.6
CDF

0.5
0.4
0.3
0.2
0.1
0
1 6 11 16 21 26 31 36 41 46 51 56 61
Year
CDF Pwater cooling CDF Paf CDF Pws CDF Ppc CDF

Figure 19.34 CDF plot for the various cooling methods by Peck equation for cooling up to 45.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
378 19 Reliability and Mean Life Assessment of Solar Panel by Cooling

19.6 Results
The calculations of the rate of degradation using Eq. (19.1), and Erying degradation using equation
Eq. (19.2) have been illustrated in details in Section 19.5. The Peck degradation equation is used
for evaluation of the mean life of the system. Mean life as calculated for various cases is shown in
Table 19.8. The mean life of solar panels is 25 years without cooling (Kim et al. 2021; Guérin et al.
2006); therefore, the mean life after cooling is shown in Table 19.9.

19.7 Conclusion
From both the degradation equations, Erying and Peck, the mean life obtained is nearly equal in
every case. It is observed that less the cooling temperature of the cell, higher mean life, higher
reliability, and less probability of failure, as observed in Table 19.9. Therefore, incorporating cooling
methods increase the mean life and reliability of solar panels, and as Table 19.1 seen, that increases

Table 19.8 Mean life by different cooling methods.

Mean life Mean life Mean life

Degradation by cooling by cooling by cooling
Cooling technique equation up to 25 ∘ C up to 35 ∘ C up to 45 ∘ C

Water cooling Erying 59.6170 42.2947 33.2692

Peck 61.5025 43.1613 33.5838
Air cooling Erying 44.6956 37.5643 32.2318
extended fin Peck 44.5812 37.6175 32.3162
Wick surface Erying 58.9043 42.2947 33.2692
Peck 60.5570 43.1613 33.5838
Passive cooling heat Erying 58.8606 42.2943 33.2692
pipe Peck 60.5322 43.1605 33.5838

Table 19.9 Increment in mean life under various conditions.

Increment in Increment in Increment in

mean life mean life mean life
Degradation by cooling by cooling by cooling
Cooling technique equation up to 25 ∘ C up to 35 ∘ C up to 45 ∘ C

Water cooling Erying 34.6170 17.2947 8.2692

Peck 36.5025 18.1613 8.5838
Air cooling Erying 19.6956 12.5643 7.2318
extended fin Peck 19.5812 12.6175 7.3162
Wick surface Erying 33.9043 17.2947 8.2692
Peck 35.5570 18.1613 8.5838
Passive cooling heat Erying 33.8606 17.2943 8.2692
pipe Peck 35.5322 18.1605 8.5838
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 379

the efficiency, too. As seen from the above study, cooling increases the mean life and reliability
of solar panels significantly, so cooling techniques should be implemented in solar power plants.
Mean life and reliability for some of the different cooling methods are compared, and it can be
extended to many other best cooling techniques to obtain the best cooling method in terms of
reliability improvement. Reliability should also be taken into consideration while calculating the
reliability of solar power systems along with the efficiency and performance of PV systems. Eco-
nomic aspects of the cooling system should also be incorporated to find out if installing the solar
panel with the cooling system is economically viable or not.

References

Akbarzadeh, A. and Wadowski, T. (1996). Heat pipe-based cooling systems for photovoltaic cells under
concentrated solar radiation. Applied Thermal Engineering 16 (1): 81–87.
Aminisharifabad, M., Yang, Q., and Wu, X. (2021). A deep learning-based reliability model for complex
survival data. IEEE Transactions on Reliability 70 (1): 73–81.
Chikate, B.V., Sadawarte, Y., and Sewagram, B.D.C.O.E. (2015). The factors affecting the performance
of solar cell. International Journal of Computer Applications 1 (1): 0975–8887.
Cuce, E., Bali, T., and Sekucoglu, S.A. (2011). Effects of passive cooling on performance of silicon
photovoltaic cells. International Journal of Low Carbon Technologies 6 (4): 299–308.
David, L. (2002). Handbook of Batteries. New York: McGraw-Hill.
Dwivedi, P., Sudhakar, K., Soni, A. et al. (2020). Advanced cooling techniques of PV modules: a state of
art. Case studies in thermal engineering 21: 100674.
Ebeling, C.E. (2004). An Introduction to Reliability and Maintainability Engineering. Tata McGraw-Hill
Education.
Firoozzadeh, M., Shiravi, A., and Shafiee, M. (2019). An experimental study on cooling the
photovoltaic modules by fins to improve power generation: economic assessment. Iranian (Iranica)
Journal of Energy & Environment 10 (2): 80–84.
Guérin, F., Lantieri, P., and Dumon, B. (2006). Applying accelerated life models to halt testing.
Reliability Modeling, Analysis And Optimization 9: 107.
Hacke, P., Spataru, S., Terwilliger, K. et al. (2015). Accelerated testing and modeling of
potential-induced degradation as a function of temperature and relative humidity. IEEE Journal of
Photovoltaics 5 (6): 1549–1553.
Hasan, A., Sarwar, J., Alnoman, H., and Abdelbaqi, E.S. (2017). Yearly energy performance of a
photovoltaic-phase change material (PV-PCM) system in hot climate. Solar Energy 146: 417–429.
Irwan, Y.M., Leow, W.Z., Irwanto, M. et al. (2015, May). Comparison of solar panel cooling system by
using dc brushless fan and dc water. In: Journal of Physics: Conference Series, vol. 622, No. 1, 012001.
IOP Publishing.
Karthick, A., Ramanan, P., Ghosh, A. et al. (2020). Performance enhancement of copper indium
diselenide photovoltaic module using inorganic phase change material. Asia-Pacific Journal of
Chemical Engineering 15 (5): e2480.
Kim, J., Rabelo, M., Padi, S.P. et al. (2021). A review of the degradation of photovoltaic modules for life
expectancy. Energies 14 (14): 4278.
Kusiak, A., Zhang, Z., and Verma, A. (2013). Prediction, operations, and condition monitoring in wind
energy. Energy 60: 1–12.
Luque, A. and Hegedus, S. (ed.) (2011). Handbook of Photovoltaic Science and Engineering. Wiley.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
380 19 Reliability and Mean Life Assessment of Solar Panel by Cooling

Makrides, G., Theristis, M., Bratcher, J. et al. (2018). Five-year performance and reliability analysis of
monocrystalline photovoltaic modules with different backsheet materials. Solar Energy 171: 491–499.
Markvart, T. (ed.) (2000). Solar Electricity, vol. 6. Wiley.
Mawoli, M., Yayha, H.N., Danshehu, B.G. et al. (2020). Development and performance evaluation of
solar photovoltaic module’s surface-to-rear temperature controlled valve for cooling application.
Nigerian Journal of Technological Development 17 (1): 20–27.
Meeker, W.Q., Escobar, L.A., and Pascual, F.G. (2022). Statistical Methods for Reliability Data. Wiley.
Moharram, K.A., Abd-Elhady, M.S., Kandil, H.A., and El-Sherif, H. (2013). Enhancing the
performance of photovoltaic panels by water cooling. Ain Shams Engineering Journal 4: 869–877.
Nižetić, S., Papadopoulos, A.M., and Giama, E. (2017). Comprehensive analysis and general
economic-environmental evaluation of cooling techniques for photovoltaic panels, part I: passive
cooling techniques. Energy Conversion and Management 149: 334–354.
Park, N.C., Oh, W.W., and Kim, D.H. (2013). Effect of temperature and humidity on the degradation
rate of multicrystalline silicon photovoltaic module. International Journal of Photoenergy 2013.
Ravi, K. and Porpatham, E. (2017). Effect of piston geometry on performance and emission
characteristics of an LPG fuelled lean burn SI engine at full throttle condition. Applied Thermal
Engineering 110: 1051–1060.
Salem, M.R., Elsayed, M.M., Abd-Elaziz, A.A., and Elshazly, K.M. (2019). Performance enhancement
of the photovoltaic cells using Al2 O3 /PCM mixture and/or water cooling-techniques. Renewable
Energy 138: 876–890.
Sharaf, M., Yousef, M.S., and Huzayyin, A.S. (2022). Review of cooling techniques used to enhance the
efficiency of photovoltaic power systems. Environmental Science and Pollution Research 29:
26131–26159.
Sohani, A., Shahverdian, M.H., Sayyaadi, H. et al. (2021). Selecting the best nanofluid type for a
photovoltaic thermal (PV/T) system based on reliability, efficiency, energy, economic, and
environmental criteria. Journal of the Taiwan Institute of Chemical Engineers 124: 351–358.
Sun, V., Asanakham, A., Deethayat, T., and Kiatsiriroat, T. (2020). A new method for evaluating
nominal operating cell temperature (NOCT) of unglazed photovoltaic thermal module. Energy
Reports 6: 1029–1042.
Tan, L., Date, A., Fernandes, G. et al. (2017). Efficiency gains of photovoltaic system using latent heat
thermal energy storage. Energy Procedia 110: 83–88.
Teo, H.G., Lee, P.S., and Hawlader, M.N.A. (2012). An active cooling system for photovoltaic modules.
Applied Energy 90 (1): 309–315.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
381

20

Reliability Assessment of Different Topologies in Photovoltaic System

Laxman Chaudhary 1 , Aanchal Verma 1 , Ramesh C. Bansal 2 , and R. K. Saket 1
1 Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
2
Department of Electrical Engineering, University of Sharjah, Sharjah, United Arab Emirates

20.1 Introduction
The fundamental issue is global climate change, which is caused by the detrimental impact of car-
bon emissions produced by traditional electrical energy generation methods. Many governments
and organizations around the world are financially supporting the implementation of renewable
energy such as wind and solar (Patel 2005). PV system is an environmentally favorable electri-
cal energy generation technique because it produces no carbon dioxide and does not need fossil
fuels. It is currently the world’s most rapidly expanding product. Total global installed PV capacity
expanded fast from 41.5 to 1177 GW in 2010–2022 and is expected to hit 5400 GW by 2030 (Solar
Power Europe 2021, IRENA report).
An electric power distribution is a massive, intricate web of electrical components that deliv-
ers electricity from power plants to users. Its major goal is to constantly meet load demand at
the lowest possible cost while maintaining the level of reliability and environmental compliance.
Therefore, it is preparing for future electrical demands using environmentally friendly energy sup-
plies, guaranteeing that load requirements are met. Moreover, To reduce the risk of an electrical
service disruption, it is important to evaluate the power systems reliability. For developing and
running reliable power systems, the creation of energy supply reliability models is crucial. The
capacity of an electrical power network to provide a consistent and reliable electricity supply is
assessed through power system reliability assessments (Billinton and Allan 1992). Because of its
size, number of components, variables, interrelations, and various purposes and aims of the sys-
tems, evaluating the reliability of the complete power system is intrinsically difficult. Hence, The
reliability studies regarding this topic are divided into three hierarchical levels (HL): HL-1 (genera-
tion), HL-2 (transmission), and HL-3 (distribution). Generation management is critical for electric
power companies because generation facilities determine the economy of power systems (Billinton
and Allan 1984, 1996). The study presented in this paper focuses on generation systems at the HL-I
level, incorporating different solar energy methods.
With rising electrical consumption, the use of fossil fuels has grown. Fossil fuels are in a limited
amount, lasting in the next few years (Li and Billinton 1994)—so they are looking for alternate
sources to generate electricity. Solar power demands are increasing to generate electricity because

Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
382 20 Reliability Assessment of Different Topologies in Photovoltaic System

it is freely available and easily accessible. The power system needs for PV systems rise, so their
dependency will also grow. Their involvement in the generating system leads to an increase in
their failure mechanism, so they focus on how to minimize their failure rates and try to make the
system more reliable. In most studies, PV systems are a core component, according to the reliability
assessment of PV systems, which does not focus on sub-components. The focus of this research is
on a sub-component of PV systems (Peyghami et al. 2021).
The PV system topology can be classified based on component arrangement are central-inverter,
string-inverter, and micro-inverter (Zhang et al. 2013; Jana et al. 2017). The structural parts of
central and string structures are identical from electrical point of view, but their connections are
different. The central PV systems are composed of many PV arrays linked to only one inverter, and
one inverter is linked to each string in a string PV system. In micro-inverter, there is one inverter
per solar panel. Developing detailed reliability models is required to incorporate these components
to find the overall system reliability. The different topologies in PV system with their pros and cons
comparison as shown in Table 20.1.

20.1.1 Basic Reliability Estimation Concepts

The probability that system will fulfill a defined function under typical operating conditions for a
given time is referred to as reliability (Denson 1998; Ebeling 1997). The continuous random vari-
able is used to mathematically express this relationship. The term “reliability” can be defined as in
Eq. (20.1):
Re(t) = Pr (T ≥ t) (20.1)
0 ≤ Re(t) ≤ 1
where T is time taken for component to fail.
Cumulative distribution function is chance that failure occurs before time (Ebeling 1997). Math-
ematically shown in Eq. (20.2),
Fa(t) = 1 − Re(t) = Pr (T < t) (20.2)
0 ≤ Fa(t) ≤ 1
The probability density function describes the failure distribution’s shape. Mathematically
shown in Eq. (20.3),
dFa(t) dRe(t)
fp(t) = =− (20.3)
dt dt
When computing reliability, failure probability, or a visual representation of the failure distribution,
the function Re(t), Fa(t), and fp(t) are utilized (Ebeling 1997).

20.1.2 Hazard Rate Function

Failure rate function often knows as hazard rate function and represented by 𝜆(t). It delivers an
instantaneous failure rate. The hazard rate function is used to define the reliability function in
Eq. (20.4):
t
Re(t) = exp(− ∫0 𝜆(u) du) (20.4)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20.1 Introduction 383

Table 20.1 Summary of different topologies.

Topology Pros Cons

Multi-stage inverter Make smoother DC input High cost

with HFT (Bose et al. 1985) Better power quality and Less reliability
decoupling
Less harmonic
Multi-stage boost Better reliability and efficiency Lower power decoupling
inverter by GEC (Bose et al. 1985) Lower cost and harmonic
Multi-stage boost inverter Lower cost and size Poor power decoupling
Pseudo-dc-link Good reliability More harmonic
Half bridge topology Less conversion stage Poor power decoupling
with dc–dc-boost Good reliability and quality
Easily handle power levels
Fly-back inverter with Improve flexibility of PV Small life-span
active power decoupling (Shimizu generation system Lower efficiency
et al. 2006)
Less harmonic and cost problem of leakage energy
Cascaded H-bridge Improve flexibility of PV modules Extra power loss
multilevel inverter Lower switching frequency Unbalanced power supply
(Xiao et al. 2014; Villanueva et al. 2009) Increase efficiency Higher cost
Two-state power Enhance voltage gain, Large size, cost
conversion (Dogga and Pathak 2019) design flexibility and weight
Lesser conversion stage Unbalanced power supply
Increase energy More losses
Traditional buck inverter Less input volt. Larger size
and line frequency Lesser conversion stage High cost
transformer (Kjaer et al. 2005; Xue Large input electrolytic capacitor Low efficiency
et al. 2004)
Central-inverter Lower cost High replacement cost
(Solar Mango 2022) More credibility High-risk factor
Produce high power
String-inverter Robustness Shorter lifespan
(Sino Voltaics 2022) Durability Less efficient
Quick troubleshooting Complex system
Micro-inverter Longer lifespan High installation and
(Baker Home Energy 2022) High Quality, Small Size replacement cost
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
384 20 Reliability Assessment of Different Topologies in Photovoltaic System

Suppose that the hazard rate function value is time-independent. In that case, that rate is called
the constant failure such that 𝜆(t) = 𝜆 for t ≥ 0, then the reliability functions will be evaluated with
the help of exponential distribution (Ebeling 1997).
Re(t) = exp(−𝜆×t) (20.5)
(−𝜆×t)
Fa(t) = 1 − exp (20.6)
fp(t) = 𝜆exp (−𝜆×t)
(20.7)

20.1.3 Reliability Block Diagram

It is a graphical representation of a system’s components that allows the systems reliability to be
calculated based on the components reliability. The reliability performance structure is reflected in
the system model, which includes series, parallel, and mixed combinations. Each block in a relia-
bility block diagram (RBD) represents a system component or subsystem. The RBD of the system is
indicated by the connecting lines, which may or may not correlate to the functional block diagram
(Ebeling 1997; Catelani et al. 2019).
All the components must be operated to perform the whole system in series. If one of the sys-
tems fails, the entire systems fail during the operation. The series relationship is represented in
Figure 20.1.
The series equivalent Reliability equation is:

Res (t) = ReA (t) ∗ ReB (t) ∗ … ∗ Ren (t) (20.8)

In a parallel configuration, at least one component must be operational. If all components must
fail for whole system fail. The parallel relationship is represented in Figure 20.2.
The parallel equivalent Reliability equation is:

Rep (t) = 1 − (1 − ReA (t)) ∗ (1 − ReB (t)) ∗ … ∗ (1 − Ren (t)) (20.9)

Systems are arranged in the series and parallel combination and use the above results of series
and parallel to get complete system reliability (Ebeling 1997). The combined RBD is in Figure 20.3.

A B n

Figure 20.1 Reliability block for elements in series.

Figure 20.2 Reliability block for elements in

A parallel.
B

n
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20.2 Reliability Modeling of PV Topology 385

Figure 20.3 Reliability block for components in

combination. RA

RC

RB

The equivalent Reliability equation is:

Reeq. (t) = [1 − (1 − ReA (t)) ∗ (1 − ReB (t))] ∗ ReC (t)
This chapter applies the above block diagram technique to evaluate the overall PV system reliability
of different solar topologies.

20.2 Reliability Modeling of PV Topology

It is necessary to create a generation model for the entire system, including the PV system power
output as well as the traditional generation method. In this study, the failure rates of the criti-
cal components of the PV system are capacitors, diodes, switches, and converters. PV systems
are modeled using various topologies like central, string and micro-inverter (Ristow et al. 2008).
A probabilistic methodology is used to construct PV system reliability model. The reliability is eval-
uating these different PV topologies by considering three assumptions, and these assumptions are
(Anand et al. 2014; Billinton et al. 2001; Allan et al. 1994):
1. The entire systems will fail, if any one of them fails.
2. During operation, the failure rates of each component stay constant.
3. The entire system is modeled as a series and parallel RBD.
This study gives a summary of the reliability models for central, string, and micro-inverter PV
systems. Reliability is calculated by considering individual component failure rate, component
RBD and their arrangement in the combination of series and parallel. The different PV topologies
reliability modeling and their RBDs are shown in Figures 20.4–20.6.

20.2.1 Reliability Modeling of Central PV Topology

Every part of the central PV system is represented by two possibilities in which it is either oper-
ational or fails with a specific probability. The key components of a central PV system are solar
array, Bulk DC-Link capacitor, inverter, and filter (Alferidi and Karki 2017; Song and Wang 2013;
Wikstrom et al. 2000). The central PV systems are shown in Figure 20.4.

20.2.2 Reliability Modeling of String PV Topology

Modeling of string PV system topology is almost the same as central but different in the connec-
tion of the power electronics components, and one extra component is used, which is a DC–DC
converter. The key component of a string PV system is solar array, two parallel DC–DC converters,
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
386 20 Reliability Assessment of Different Topologies in Photovoltaic System

Part-A

DC AC

Figure 20.4 Schematic diagram of central-inverter.

Part-A

DC DC
DC AC

DC DC

Figure 20.5 Schematic diagram of string-inverter.

Part-A

DC DC DC AC

Figure 20.6 Schematic diagram of micro-inverter.

Bulk DC-Link capacitor, an inverter and filter (Alferidi and Karki 2017; Atasoy and Tezcan 2021).
A string-inverter PV system is shown in Figure 20.5.

20.2.3 Reliability Modeling of Micro PV Topology

The key component of a micro-inverter PV system topology is solar array, DC–DC converters, Bulk
DC-Link capacitor, inverter, and filter. Its modeling arrangement is shown in Figure 20.6 (Alferidi
and Karki 2017; Tariq et al. 2018; Li and Wolfs 2008).
The above shows the reliability modeling of different topologies. As earlier mentioned, the most
critical components are capacitors, diodes, switches, and converters, so focus on these compo-
nent’s failure rates and how the failure rates of these components depend on several factors. The
MIL-HDBK-217F second edition of the military handbook provides a failure database for the power
electronic elements. This information is used in this research to estimate the failure rate of power
electronic components, and the results of this stage are used to investigate component’s reliability
(Military Handbook, 217 F Notice 2 1995).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20.3 Estimation of Failure Rate 387

20.3 Estimation of Failure Rate

20.3.1 DC-Link Capacitor

Its goal is to provide a stable DC voltage and limit the fluctuation of the system. DC-link capacitors
can fail for a variety of reasons, including design flaws, material wear, operational temperature,
voltage, current, moisture, and mechanical stress (Wang and Blaabjerg 2014). As for the consid-
eration of the system requirements electrolytic capacitor has a voltage rating of 250 V and a peak
temperature of 50∘ C (Harb and Balog 2012). Capacitor failure rate equation is:
𝜆cap = n ∗ 𝜆base ∗ 𝜋CF ∗ 𝜋QF ∗ 𝜋ES ∗ 𝜋TF (20.10)
𝜋CF = 0.34 ∗ C0.18 (20.11)
1 1
−4061.74∗( T +273 − 298 )
𝜋TF = exp j (20.12)
where
n = Number of capacitor in DC-Link
𝜆base = Capacitor basic failure rate
𝜆cap = Capacitor failure rate
𝜋ES = Effect of environment stress
𝜋QF = Quality factor
𝜋CF = Capacitance factor
𝜋TF = Temperature factor
C = Capacitor value in micro Faraday
Tj = Junction temperature in degree Celsius

20.3.2 Diodes
The equation of failure rates of diodes as show (Military Handbook, 217 F Notice 2 1995):
𝜆diode = n ∗ 𝜆base ∗ 𝜋CCF ∗ 𝜋QF ∗ 𝜋ES ∗ 𝜋TF ∗ 𝜋ESF (20.13)
1 1
−3091∗( T +273 − 298 )
𝜋TF = exp j (20.14)
{
0.054, for VSr ≤ 0.3
𝜋ESF = (20.15)
2.43
VSr , for 0.3 ≤ VSr ≤ 1

operated voltage
VSr = (20.16)
rated voltage
where
n = Number of diodes
𝜆base = Diode basic failure rate
𝜆diode = Diode failure rate
𝜋ES = Effect of environment stress
𝜋QF = Quality factor
𝜋CCF = Contact Construction factor
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
388 20 Reliability Assessment of Different Topologies in Photovoltaic System

𝜋TF = Temperature factor

VSr = Voltage stress factor
Tj = Junction temperature in degree Celsius

20.3.3 Switch
The equation of failure rates of switch as show (Military Handbook, 217 F Notice 2 1995):
𝜆switch = n ∗ 𝜆base ∗ 𝜋QF ∗ 𝜋ES ∗ 𝜋TF (20.17)
1 1
−1925∗( T +273 − 298 )
𝜋TF = exp j (20.18)

where
n = Number of switch
𝜆base = switch basic failure rate
𝜆switch = switch failure rate
𝜋ES = Effect of environment stress
𝜋QF = Quality factor
𝜋TF = Temperature factor
Tj = Junction temperature in degree Celsius

20.3.4 Filter
LC filter is used for flattening the AC signal in PV system. As for the consideration of system require-
ment MIL-F-18327 LC filter is used (Military Handbook, 217 F Notice 2 1995). The equation of
failure rate of filter as show:
𝜆filter = n ∗ 𝜆base ∗ 𝜋QF ∗ 𝜋ES (20.19)
where
n = Number of filter
𝜆base = filter basic failure rate
𝜆Filter = filter failure rate
𝜋ES = Effect of environment stress
𝜋QF = Quality factor
The component failure rate is calculated using these failure rate equations. The exponential
distribution is used to find the reliability of component (Military Handbook, 217 F Notice 2
1995). The calculation of failure rates needs some fundamental value of failure rate, it is shown in
Table 20.2.

20.4 Reliability Estimation Using RBD

Equations (20.10)–(20.19) are used to calculate the failure rate. The RBD and their series-parallel
combinations are used to find the overall system reliability. This chapter first evaluates the single
block of reliability and similarly finds all other block reliability, and finally uses a parallel system to
get the reliability of the parallel blocks after that apply series system to estimate the overall system
reliability.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20.4 Reliability Estimation Using RBD 389

Table 20.2 Basic failure rate of components.

Components Failure rate (occur/yr)

Switches 0.012
Capacitor 0.0314
Diodes 0.025
Circuit-breaker 0.003
Filter 0.0011
Transformer 0.006

20.4.1 Central Topology

In Figure 20.4 of the central inverter, the reliability estimation of Block-A consists of DC-Link
capacitance, inverter, and filter. Use Eqs. (20.10)–(20.20) to get their failure rates. Similarly, the sec-
ond and third blocks are arranged. After that, all these blocks are parallel connected to the series
arrangement of circuit breaker and transformer (Obeidat and Shuttleworth 2017; Lindquist et al.
2008). The whole arrangement of the Central topology is shown in Figure 20.7.
DC-Link capacitor failure rate is estimated from Eqs. (20.10)–(20.12), where the capacitor value
is 330 μF (Harb and Balog 2012), the junction temperature is 50∘ C (Harb and Balog 2012), n = 3,
𝜆base = 0.0314, 𝜋ES = 1, 𝜋QF = 1, 𝜋CF = 0.9656, and 𝜋TF = 2.8718. These values are considered as per
the system requirement (Military Handbook, 217 F Notice 2 1995):

𝜆cap = 3 ∗ 0.0314 ∗ 0.9656 ∗ 1 ∗ 1 ∗ 2.8718

𝜆cap = 0.2613 occur∕yr

Part-A

DC AC

Part-B Towards
transmission
line
DC AC

Part-C

DC AC

Generation

Figure 20.7 Central-inverter topology.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
390 20 Reliability Assessment of Different Topologies in Photovoltaic System

A two-level three-phase voltage source inverter is used in which six switches and diodes are used
(Chan and Calleja 2011). The inverter failure rates are estimated with equation is:
𝜆inverter = 6 ∗ (𝜆diode + 𝜆switch ) (20.20)
Diode failure rate is estimated from Eqs. (20.13)–(20.16), where operating voltage = 607 V, rated
voltage = 690 V (Abdi et al. 2009), the junction temperature is 50∘ C (Harb and Balog 2012), n =
1, 𝜆base = 0.025, 𝜋ES = 1, 𝜋QF = 0.7, 𝜋CCF = 1, 𝜋TF = 2.232, and 𝜋ESF = 0.7324. These values are
considered as per the system requirement (Military Handbook, 217 F Notice 2 1995):
𝜆diode = 1 ∗ 0.025 ∗ 1 ∗ 0.7 ∗ 1 ∗ 2.232 ∗ 0.7324
𝜆diode = 0.0286 occur∕yr
Switch failure rate is estimated from Eqs. (20.17) and (20.18), where the junction temperature is
50∘ C (Harb and Balog 2012), n = 1, 𝜆base = 0.012, 𝜋ES = 1, 𝜋QF = 0.7, and 𝜋TF = 1.6487. These values
are considered as per the system requirement (Military Handbook, 217 F Notice 2 1995; Wang and
Blaabjerg 2014).
𝜆switch = 1 ∗ 0.012 ∗ 0.7 ∗ 1 ∗ 1.6487
𝜆switch = 0.0138 occur∕yr
Now the failure rate of inverter using Eq. (20.20) (Alferidi and Karki 2017):
𝜆Inverter = 6 ∗ (0.0286 + 0.0138)
𝜆Inverter = 0.2544 occur∕yr
Filter failure rate is estimated from Eq. (20.19), where n = 1, 𝜆base = 0.0011, 𝜋ES = 1, and 𝜋QF = 1.
These values are considered as per the system requirement (Alferidi and Karki 2017):
𝜆filter = 1 ∗ 0.0011 ∗ 1 ∗ 1
𝜆filter = 0.0011 occur∕yr
Reliability estimation of Block-A system using Eq. (20.8)
ReA = ReCapacitor ∗ ReInverter ∗ ReFilter
ReA = exp(−0.2613∗t) ∗ exp(−0.2544∗t) ∗ exp(−0.0011∗t)
ReA = exp(−0.5168∗t)
Similarly, the Block-B and Block-C reliability values are:
ReB = ReC = exp(−0.5168∗t)
After that, these blocks are parallel connected so their parallel equivalent reliability using Eq. (20.9):
ReParallel eq. = 1 − (1 − ReA ) ∗ (1 − ReB ) ∗ (1 − ReC )
ReParallel eq. = 1 − (1 − exp(−0.5168∗t) )3
The overall reliability of central topology using Eq. (20.8):
ReCen (t) = ReParallel eq. ∗ ReCkt. Breaker ∗ ReTrans
ReCen (t) = 1 − (1 − exp(−0.5168∗t) )3 ∗ exp(−0.009∗t)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20.4 Reliability Estimation Using RBD 391

The cumulative distribution function of central topology is evaluated from Eq. (20.2):

FaCen (t) = 1 − ReCen (t)

FaCen (t) = 1 − (1 − (1 − exp(−0.5168∗t) )3 ∗ exp(−0.009∗t) )

The probability density function of central topology is evaluated from Eq. (20.3):
dReCen (t)
fpCen (t) = −
dt
[ ]
1 − (1 − exp(−0.5168∗t) )3 ∗ exp(−0.009∗t)
fpCen (t) = −
dt
fpCen (t) = 1.5504 ∗ (1 − exp(−0.5168∗t) )2 ∗ exp(−0.5168∗t)
+ 0.009 ∗ 1 − (1 − exp(−0.5168∗t) )3 ∗ exp(−0.009∗t)
Figures 20.8–20.10 represent the reliability function, Cumulative distribution function and prob-
ability density function, respectively, of the Central-Inverter topology.

20.4.2 String Topology

Reliability estimation of Block-A in Figure 20.5 of string-inverter PV system consists of Two par-
allel DC–DC converters, Bulk DC-Link capacitance and inverter. Use Eqs. (20.10)–(20.21) to get
their failure rates. Similarly, the second and third blocks are arranged. After that, all these blocks
are parallel connected to the series arrangement of circuit breaker and transformer (Obeidat and

Figure 20.8 Reliability plot of central PV

system. 1
Reliability Rc(t)
0.9
Junction temp. at 50°C
0.8

0.7

0.6
Reliability

0.5

0.4

0.3

0.2

0.1

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
392 20 Reliability Assessment of Different Topologies in Photovoltaic System

Figure 20.9 Cumulative distribution plot of

1.2 central PV system.
CDF Fc(t)

Junction temp. at 50°C

1

0.8
CDF

0.6

0.4

0.2

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)

Figure 20.10 Probability density function

0.25 plot of central PV system.
PDF fc(t)

Junction temp. at 50°C

0.2

0.15
PDF

0.1

0.05

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20.4 Reliability Estimation Using RBD 393

Part-A Generation
DC DC
DC AC

DC DC
Part-B Towards
DC DC transmission
line
DC AC

DC DC
Part-C
DC DC
DC AC

DC DC

Figure 20.11 String-inverter topology.

Shuttleworth 2017; Lindquist et al. 2008; Kothari et al. 2022). The whole arrangement of the string
topology is shown in Figure 20.11
DC-Link capacitor failure rate is estimated from Eqs. (20.10)–(20.12), where the capacitor value
is 330 μF (Harb and Balog 2012), Junction temperature is 50∘ C (Harb and Balog 2012), n = 1, 𝜆base =
0.0314, 𝜋ES = 1, 𝜋QF = 1, 𝜋CF = 0.9656, and 𝜋TF = 2.8718. These values are considered as per the
system requirement (Military Handbook, 217 F Notice 2 1995):
𝜆cap = 1 ∗ 0.0314 ∗ 0.9656 ∗ 1 ∗ 1 ∗ 2.8718
𝜆cap = 0.0871 occur∕yr
Diode failure rate is estimated from Eqs. (20.13)–(20.16), where operating voltage = 607 V, rated
voltage = 690 V (Abdi et al. 2009), the junction temperature is 50∘ C (Harb and Balog 2012), n = 1,
𝜆base = 0.025, 𝜋ES = 1, 𝜋QF = 1, 𝜋CCF = 1, 𝜋TF = 2.232, and 𝜋ESF = 0.7324. These values are considered
as per the system requirement (Military Handbook, 217 F Notice 2 1995):
𝜆diode = 1 ∗ 0.025 ∗ 1 ∗ 1 ∗ 1 ∗ 2.232 ∗ 0.7324
𝜆diode = 0.0409 occur∕yr
Switch failure rate is estimated from Eqs. (20.17) and (20.18), where the junction temperature is
50∘ C (Harb and Balog 2012), n = 1, 𝜆base = 0.012, 𝜋ES = 1, 𝜋QF = 1, and 𝜋TF = 1.6487. These values
are considered as per the system requirement (Military Handbook, 217 F Notice 2 1995; Wang and
Blaabjerg 2014):
𝜆switch = 1 ∗ 0.012 ∗ 1 ∗ 1 ∗ 1.6487
𝜆switch = 0.0198 occur∕yr
Now the failure rate of inverter using Eq. (20.20) (Alferidi and Karki 2017):
𝜆Inverter = 6 ∗ (0.0409 + 0.0198)
𝜆Inverter = 0.3642 occur∕yr
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
394 20 Reliability Assessment of Different Topologies in Photovoltaic System

DC–DC converters are not treated as a single block; rather, it is a combination of components, and
their failure rates are estimated with the given equation (Alferidi and Karki 2017):

𝜆DC–DC Converter = (𝜆diode + 𝜆Switch )

𝜆DC–DC Converter = (0.0409 + 0.0198) (20.21)
𝜆DC–DC Converter = 0.0607 occur∕yr

Filter failure rate is estimated from Eq. (20.19), where n = 1, 𝜆base = 0.0011, 𝜋ES = 1, and 𝜋QF = 1.
These values are considered as per the system requirement (Alferidi and Karki 2017):

𝜆filter = 1 ∗ 0.0011 ∗ 1 ∗ 1
𝜆filter = 0.0011 occur∕yr

Reliability estimation of Block-A system using Eqs. (20.8) and (20.9):

ReA = (1 − (1 − ReDC–DC Con. )2 ) ∗ ReCap. ∗ ReInv. ∗ ReFil.

ReA = (1 − (1 − exp(−0.0607∗t) )2 ) ∗ exp(−0.0871∗t) ∗ exp(−0.3642∗t) ∗ exp(−0.0011∗t)
ReA = (1 − (1 − exp(−0.0607∗t) )2 ) ∗ exp(−0.4524∗t)

Similarly, the Block-B and Block-C and their reliability values are:

ReB = ReC = (1 − (1 − exp(−0.0607∗t) )2 ) ∗ exp(−0.4524∗t)

After that, these blocks are parallel connected so their parallel equivalent reliability using Eq. (20.9):

RePar. = 1 − (1 − ReA ) ∗ (1 − ReB ) ∗ (1 − ReC )

RePar. = 1 − [(1 − (1 − (1 − exp(−0.0607∗t) )2 ) ∗ exp(−0.4524∗t) )3 ]

The overall reliability of string topology using Eq. (20.8):

ReStr (t) = ReParallel eq. ∗ ReCkt. Breaker ∗ ReTrans

ReStr (t) = (1 − [(1 − (1 − (1 − exp(−0.0607∗t) )2 ) ∗ exp(−0.4524∗t) )3 ]) ∗ exp(−0.009∗t)

The cumulative distribution function of central topology is evaluated from Eq. (20.2):

FaStr (t) = 1 − ReStr (t)

FaStr (t) = 1 − (1 − [(1 − (1 − (1 − exp(−0.0607∗t) )2 ) ∗ exp(−0.4524∗t) )3 ] ∗ exp(−0.009∗t) )

The probability density function of central topology is evaluated from Eq. (20.3):
dReStr (t)
fpStr (t) = −
dt
fpStr (t) = (0.009 ∗ (1 − (1 − [1 − (1 − exp(−0.0607∗t) )2 ∗ exp(−0.4524∗t) ]3
∗ 3 ∗ (1 − [1 − (1 − exp(−0.0607∗t) )2 ∗ exp(−0.4524∗t) ]2 ) ∗ (0.4524 ∗ exp(−0.4524∗t) )
∗ 1 − (1 − exp(−0.0607∗t) )2 + 0.1214 exp(−0.5131∗t) − 0.1214 ∗ exp(−0.5738∗t) )))
∗ exp(−0.009∗t)

Figures 20.12–20.14 represent the reliability function, Cumulative distribution function and prob-
ability density function, respectively, of the string-inverter topology.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20.4 Reliability Estimation Using RBD 395

Figure 20.12 Reliability plot of string PV

system. 1
Reliability Rs(t)
0.9
Junction temp. at 50°C
0.8

0.7

0.6

Reliability
0.5

0.4

0.3

0.2

0.1

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)

Figure 20.13 Cumulative distribution plot

of string PV system. 1.2
CDF Fs(t)

Junction temp. at 50°C

1

0.8
CDF

0.6

0.4

0.2

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Time (yr)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
396 20 Reliability Assessment of Different Topologies in Photovoltaic System

Figure 20.14 Probability density function

0.2 plot of string PV system.
PDF fs(t)
0.18
Junction temp. at 50°C
0.16

0.14

0.12
PDF

0.1

0.08

0.06

0.04

0.02

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)

20.4.3 Micro Topology

Reliability estimation of Block-A in Figure 20.6 of micro-inverter solar system is DC–DC converters,
Bulk DC-Link capacitance, inverter, and filter. Use Eqs. (20.10)–(20.21) to get failure rates. Simi-
larly, the second and third blocks are arranged. After that, all phases are parallel connected to the
series arrangement of circuit breaker and transformer (Obeidat and Shuttleworth 2017; Lindquist
et al. 2008). The whole arrangement of the Micro topology is shown in Figure 20.15.

Part-A Generation

DC DC DC AC

Part-B Towards
transmission
DC DC DC AC line

Part-C

DC DC DC AC

Figure 20.15 Micro-inverter topology.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20.4 Reliability Estimation Using RBD 397

DC-Link capacitor failure rate is estimated from Eqs. (20.10)–(20.12), where the capacitor value
is 330 μF (Harb and Balog 2012), the junction temperature is 50∘ C (Harb and Balog 2012), n = 1,
𝜆base = 0.0314, 𝜋ES = 1, 𝜋QF = 1, 𝜋CF = 0.9656, and 𝜋TF = 2.8718. These values are considered as per
the system requirement (Military Handbook, 217 F Notice 2 1995).
𝜆cap = 1 ∗ 0.0314 ∗ 0.9656 ∗ 1 ∗ 1 ∗ 2.8718
𝜆cap = 0.0871 occur∕yr
Diode failure rate is estimated from Eqs. (20.13)–(20.16), where operating voltage = 607 V, rated
voltage = 690 V (Abdi et al. 2009; Kothari et al. 2022; Verma et al. 2023), the junction temperature
is 50∘ C (Harb and Balog 2012), n = 1, 𝜆base = 0.025, 𝜋ES = 1, 𝜋QF = 0.7, 𝜋CCF = 1, 𝜋TF = 2.232, and
𝜋ESF = 0.7324. These values are considered as per the system requirement (Military Handbook, 217
F Notice 2 1995):
𝜆diode = 1 ∗ 0.025 ∗ 1 ∗ 0.7 ∗ 1 ∗ 2.232 ∗ 0.7324
𝜆diode = 0.0286 occur∕yr
Switch failure rate is estimated from Eqs. (20.17) and (20.18), where the junction temperature is
50∘ C (Harb and Balog 2012), n = 1, 𝜆base = 0.012, 𝜋ES = 1, 𝜋QF = 0.7, and 𝜋TF = 1.6487. These values
are considered as per the system requirement (Military Handbook, 217 F Notice 2 1995; Wang and
Blaabjerg 2014):
𝜆switch = 1 ∗ 0.012 ∗ 0.7 ∗ 1 ∗ 1.6487
𝜆switch = 0.0138 occur∕yr
Now the failure rate of inverter using Eq. (20.20) (Alferidi and Karki 2017):
𝜆Inverter = 6 ∗ (0.0286 + 0.0138)
𝜆Inverter = 0.2544 occur∕yr
Now the failure rate of DC–DC converter using Eq. (20.21) (Alferidi and Karki 2017):
𝜆DC–DC Converter = (0.0286 + 0.0138)
𝜆DC–DC Converter = 0.0424 occur∕yr
Filter failure rate is estimated from Eq. (20.19), where n = 1, 𝜆base = 0.0011, 𝜋ES = 1, and 𝜋QF = 1.
These values are considered as per the system requirement (Alferidi and Karki 2017):
𝜆filter = 1 ∗ 0.0011 ∗ 1 ∗ 1
𝜆filter = 0.0011 occur∕yr
Reliability estimation of Block-A system using Eq. (20.8):
ReA = ReDC–DC Con. ∗ ReCap ∗ ReInv. ∗ ReFil.
ReA = exp(−0.0424∗t) ∗ exp(−0.0871∗t) ∗ exp(−0.2544∗t) ∗ exp(−0.0011∗t)
ReA = exp(−0.385∗t)
Similarly, the Block-B and Block-C and their reliability values are:
ReB = ReC = exp(−0.385∗t)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
398 20 Reliability Assessment of Different Topologies in Photovoltaic System

After that, these blocks are parallel connected so their parallel equivalent reliability using Eq. (20.9):

ReParallel eq. = 1 − (1 − ReA ) ∗ (1 − ReB ) ∗ (1 − ReC )

ReParallel eq. = 1 − (1 − exp(−0.385∗t) )3

The overall reliability of central topology using Eq. (20.8):

ReMicro (t) = ReParallel eq. ∗ ReCkt. Breaker ∗ ReTrans

ReMicro (t) = 1 − (1 − exp(−0.385∗t) )3 ∗ exp(−0.009∗t)

The cumulative distribution function of central topology is evaluated from Eq. (20.2):

FaMicro (t) = 1 − ReMicro (t)

FaMicro (t) = 1 − (1 − (1 − exp(−0.385∗t) )3 ∗ exp(−0.009∗t) )

The probability density function of central topology is evaluated from Eq. (20.3):
dReMicro (t)
fpMicro (t) = −
dt
[ ]
1 − (1 − exp(−0.385∗t) )3 ∗ exp(−0.009∗t)
fpMicro (t) = −
dt
fpMicro (t) = 1.155 ∗ (1 − exp(−0.385∗t) )2 ∗ exp(−0.385∗t)
+ 0.009 ∗ 1 − (1 − exp(−0.385∗t) )3 ∗ exp(−0.009∗t)
Figures 20.16–20.18 represent the reliability function, cumulative distribution function and prob-
ability density function, respectively, of the micro-inverter topology.

Figure 20.16 Reliability plot of micro PV

1.2 system.
Reliability Rm(t)
Junction temp. at 50°C
1

0.8
Reliability

0.6

0.4

0.2

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20.4 Reliability Estimation Using RBD 399

Figure 20.17 Cumulative distribution plot

of micro PV system. 1.2
CDF Fm(t)
Junction temp. at 50°C
1

0.8

CDF
0.6

0.4

0.2

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)

Figure 20.18 Probability density function

plot of micro PV system. 0.2
PDF fm(t)
0.18
Junction temp. at 50°C
0.16

0.14

0.12
PDF

0.1

0.08

0.06

0.04

0.02

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
400 20 Reliability Assessment of Different Topologies in Photovoltaic System

20.5 Results

In Section 20.5, plot the graphs of the reliability function, cumulative distribution function,
and probability density function for various topologies in PV system at a junction temperature
of 50∘ C. Similar, evaluation at junction temperature of 25 and 75∘ C of these topologies. This
section now shows a comparison plot of these junction temperatures for various topologies in PV
system.
The reliability function plots of 25, 50, and 75∘ C temperatures are shown in Figures 20.19–20.21,
respectively. The reliability value decreases at a slow rate at low temperature compared to high
temperature so the system is more reliable at low temperature.
The cumulative distribution function plots of 25, 50, and 75∘ C temperatures as shown in
Figures 20.22–20.24, respectively. The failure probability value increasing slow rate at low
temperature compare to high temperature so the systems get failure early at high temperature.
The probability density function plots of 25, 50, and 75∘ C temperatures as shown in
Figures 20.25–20.27, respectively. At a particular time “t” before that time, it represents fail-
ure probability and, after that time, it represents reliability function nature. In all these probability
density function plot represents micro has less failure before that time “t” and more reliable after
that time compared to all topologies.

Figure 20.19 Reliability plots at 25∘ C of

1.2
different topologies.
Junction temp. at 25°C

0.8
Reliability

0.6

0.4

0.2

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)
Reliability Rc(t) Reliability Rs(t) Reliability Rm(t)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20.5 Results 401

Figure 20.20 Reliability plots at 50∘ C of 1.2

different topologies.
Junction temp. at 50°C
1

0.8

Reliability
0.6

0.4

0.2

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)
Reliability Rc(t) Reliability Rs(t) Reliability Rm(t)

Figure 20.21 Reliability plots at 75∘ C of

0.9
different topologies.
Junction temp. at 75°C
0.8

0.7

0.6
Reliability

0.5

0.4

0.3

0.2

0.1

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)
Reliability Rc(t) Reliability Rs(t) Reliability Rm(t)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
402 20 Reliability Assessment of Different Topologies in Photovoltaic System

Figure 20.22 Cumulative distribution plots

1 at 25∘ C of different topologies.
Junction temp. at 25°C
0.9

0.8

0.7

0.6
CDF

0.5

0.4

0.3

0.2

0.1

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)
CDF Fc(t) CDF Fs(t) CDF Fm(t)

Figure 20.23 Cumulative distribution plots

1.2
at 50∘ C of different topologies.
Junction temp. at 50°C
1

0.8
CDF

0.6

0.4

0.2

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)
CDF Fc(t) CDF Fs(t) CDF Fm(t)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20.5 Results 403

Figure 20.24 Cumulative distribution plots

at 75∘ C of different topologies.
1.2
Junction temp. at 75°C

0.8

CDF
0.6

0.4

0.2

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)
CDF Fc(t) CDF Fs(t) CDF Fm(t)

Figure 20.25 Probability density plots at

25∘ C of different topologies.
0.12

Junction temp. at 25°C

0.1

0.08
PDF

0.06

0.04

0.02

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)
PDF fc(t) PDF fs(t) PDF fm(t)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
404 20 Reliability Assessment of Different Topologies in Photovoltaic System

Figure 20.26 Probability density plots at

0.25
50∘ C of different topologies.
Junction temp. at 50°C

0.2

0.15
PDF

0.1

0.05

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)
PDF fc(t) PDF fs(t) PDF fm(t)

Figure 20.27 Probability density plots at

0.6 75∘ C of different topologies.
Junction temp. at 75°C

0.5

0.4
PDF

0.3

0.2

0.1

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (yr)
PDF fc(t) PDF fs(t) PDF fm(t)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 405

20.6 Conclusions

From the reliability assessment of central-inverter, string-inverter, and micro-inverter on 25, 50,
and 75∘ C it is found that micro-inverter topology shows the better result in comparison with
central-inverter and String-inverter.
The most crucial component in the most reliable solar topology, according to this study, is an
inverter, which has a failure rate of 0.2544 per year, which is high when compared to all other
components utilized in the micro-inverter solar topology. The author proposed a decrement in the
failure rate of an inverter will significantly change system reliability by optimizing the inverter
architecture.

References

Abdi, B., Ranjbar, A.H., Gharehpetian, G.B., and Milimonfared, J. (2009). Reliability considerations for
parallel performance of semiconductor switches in high-power switching power supplies. IEEE
Transactions on Industrial Electronics 56: 2133–2139.
Alferidi, A. and Karki, R. (2017). Development of probabilistic reliability models of photovoltaic system
topologies for system adequacy evaluation. Applied Sciences 7(2): 176.
Allan, R.N., Billinton, R., Breipohl, A.M., and Grigg, C.H. (1994). Bibliography on the application of
probability methods in power system reliability evaluation: 1987–1991. IEEE Transactions on Power
Systems 9(1): 41–49.
Anand, S., Gundlapalli, S.K., and Fernandes, B.G. (2014). Transformer-less grid feeding current source
inverter for solar photovoltaic system. IEEE Transactions on Industrial Electronics 61: 5334–5344.
Atasoy, B. and Tezcan, S.S. (2021). Comparison of string and central inverter for 10MW PV plant. 2021
5th International Symposium on Multidisciplinary Studies and Innovative Technologies (ISMSIT),
182–186. https://doi.org/10.1109/ISMSIT52890.2021.9604588.
Baker Home Energy (2022). https://bakerhomeenergy.com/blog/2016-06-14/advantages-and-
disadvantages-of-microinverters/ (accessed 06 April 2022).
Billinton, R. and Allan, R.N. (1984). Power-system reliability in perspective. Electronics and Power
30(3): 231–236.
Billinton, R. and Allan, R.N. (1992). Reliability Evaluation of Engineering Systems: Concepts and
Techniques. New York: Plenum Press.
Billinton, R. and Allan, R.N. (1996). Reliability Evaluation of Power Systems, 2e. New York and London:
Plenum Publishing.
Billinton, R., Fotuhi-Firuzabad, M., and Bertling, L. (2001). Bibliography on the application of
probability methods in power system reliability evaluation 1996–1999. IEEE Transactions on Power
Systems 16(4): 595–602.
Bose, B.K., Szczesny, P.M., and Steigerwald, R.L. (1985). Microcomputer control of a residential
photovoltaic power conditioning system. IEEE Transactions on Industry Applications 5: 1182–1191.
Catelani, M., Ciani, L., and Venzi, M. (2019). RBD model-based approach for reliability assessment in
complex systems. IEEE Systems Journal 13(3): 2089–2097. https://doi.org/10.1109/JSYST.2018
.2840220.
Chan, F. and Calleja, H. (2011). Reliability estimation of three single-phase topologies in
grid-connected PV systems. IEEE Transactions on Industrial Electronics 58: 2683–2689.
Denson, W. (1998). The history of reliability prediction. IEEE Transactions on Reliability 47(3):
SP321–SP328. https://doi.org/10.1109/24.740547.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
406 20 Reliability Assessment of Different Topologies in Photovoltaic System

Dogga, R. and Pathak, M.K. (2019). Recent trends in solar PV inverter topologies. Solar Energy
183: 57–73.
Ebeling, C.E. (1997). An Introduction to Reliability and Maintainability Engineering, 1e. Tata
McGraw-Hill Education.
Harb, S. and Balog, R.S. (2012). Reliability of candidate photovoltaic module-integrated-inverter
(PV-MII) topologies–a usage model approach. IEEE Transactions on Power Electronics 28(6):
3019–3027.
Jana, J., Saha, H., and Bhattacharya, K.D. (2017). A review of inverter topologies for single-phase
grid-connected photovoltaic systems. Renewable and Sustainable Energy Reviews 72: 1256–1270.
Kjaer, S.B., Pedersen, J.K., and Blaabjerg, F. (2005). A review of single-phase grid-connected inverters
for photovoltaic modules. IEEE Transactions on Industry Applications 41(5): 1292–1306.
Kothari, D.P., Nagrath, I.J., and Saket, R.K. (2022). Modern Power System Analysis. Tata McGraw-Hill
Publishing Company.
Li, W. and Billinton, R. (1994). Reliability Assessment of Electrical Power Systems Using Monte Carlo
Methods. New York: Plenum Press.
Li, Q. and Wolfs, P. (2008). A review of the single phase photovoltaic module integrated converter
topologies with three different DC link configurations. IEEE Transactions on Power Electronics 23(3):
1320–1333.
Lindquist, T.M., Bertling, L., and Eriksson, R. (2008). Circuit breaker failure data and reliability
modelling. IET Generation, Transmission & Distribution 2(6): 813–820.
Military Handbook, 217 F Notice 2 (1995). Reliability Prediction of Electronic Equipment.
Obeidat, F. and Shuttleworth, R. (2017). PV inverters reliability prediction. World Applied Sciences
Journal 35(2): 275–287.
Patel, M.R. (2005). Wind and Solar Power Systems: Design, Analysis, and Operation. CRC Press.
Peyghami, S., Blaabjerg, F., and Palensky, P. (2021). Incorporating power electronic converters
reliability into modern power system reliability analysis. IEEE Journal of Emerging and Selected
Topics in Power Electronics 9(2): 1668–1681.
Ristow, A., Begovic, M., Pregelj, A., and Rohatgi, A. (2008). Development of a methodology for
improving photovoltaic inverter reliability. IEEE Transactions on Industrial Electronics 55:
2581–2592.
Shimizu, T., Wada, K., and Nakamura, N. (2006). Flyback-type single-phase utility interactive inverter
with power pulsation decoupling on the DC input for an AC photovoltaic module system. IEEE
Transactions on Power Electronics 21(5): 1264–1272.
Sino Voltaics (2022). https://sinovoltaics.com/learing-center/inverters/string-inverters/ (accessed 05
April 2022).
Solar Mango (2022). https://www.solarmango.com/ask/2015/10/23/what-are-central-inverters/
(accessed 04 April 2022).
Solar Power Europe (2021). Global Market Outlook for Solar Power 2021–2025.
Song, Y. and Wang, B. (2013). Survey on reliability of power electronic systems. IEEE Transactions on
Power Electronics 28(1): 591–604. https://doi.org/10.1109/TPEL.2012.2192503.
Tariq, M.S., Butt, S.A., and Khan, H.A. (2018). Impact of module and inverter failures on the
performance of central-, string-, and micro-inverter PV systems. Microelectronics Reliability 88:
1042–1046.
Verma, A., Singh, A., Anand Kumar, K. et al. (2023). Reliability analysis of multilevel and matrix
converters used in more electric aircraft. IET Electrical Systems in Transportation 13(2): e12078.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 407

Villanueva, E., Correa, P., Rodriguez, J.É., and Pacas, M. (2009). Control of a single-phase cascaded
H-bridge multilevel inverter for grid-connected photovoltaic systems. IEEE Transactions on
Industrial Electronics 56(11): 4399–4406.
Wang, H. and Blaabjerg, F. (2014). Reliability of capacitors for DC-link applications in power electronic
converters–an overview. IEEE Transactions on Industry Applications 50(5): 3569–3578. https://doi
.org/10.1109/TIA.2014.2308357.
Wikstrom, P., Terens, L.A., and Kobi, H. (2000). Reliability, availability, and maintainability of
high-power variable-speed drive systems. IEEE Transactions on Industry Applications 36(1): 231–241.
Xiao, B., Hang, L., Mei, J. et al. (2014). Modular cascaded H-bridge multilevel PV inverter with
distributed MPPT for grid-connected applications. IEEE Transactions on Industry Applications 51(2):
1722–1731.
Xue, Y., Chang, L., Kjær, S. et al. (2004). Topologies of single-phase inverters for small distributed
power generators: an overview. IEEE Transactions on Power Electronics 19: 1305–1314. https://doi
.org/10.1109/TPEL.2004.833460.
Zhang, P., Li, W., Li, S. et al. (2013). Reliability assessment of photovoltaic power systems: review of
current status and future perspectives. Applied Energy 104: 822–833.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
409

Reliability Analysis of Power Electronics Components and Systems for

Modern Power System Applications
Section 6
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
411

21

Reliability Evaluation of Power Electronics Converters for Modern

Power System Applications
Amit Kumar 1 , Sachin Kumar 2 , Sunil K. Singh 3 , R. K. Saket 4 , and P. Sanjeevikumar 5
1
Flameproof and Equipment Safety Department, CSIR-Central Institute of Mining and Fuel Research, Dhanbad, Jharkhand, India
2
Department of Electrical Engineering, G.B. Pant Institute of Engineering and Technology, Pauri-Garhwal, Uttarakhand, India
3
Department of Electrical Engineering, Shri Ramswaroop Memorial University, Lucknow, Uttar Pradesh, India
4
Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
5
Department of Electrical Engineering, IT and Cybernetics, University of South-Eastern Norway, Porsgrunn, Norway

21.1 Introduction
Power electronics systems are widely used in various applications such as variable speed drives,
electric vehicles, renewable energy systems, and power quality control stations (Kumar et al. 2020;
Singh et al. 2023). Reliability is essential for power electronics systems due to increased dependabil-
ity in many crucial areas like energy and transportation. Reliability is the ability of any instrument
to perform as desired for a certain period. Reliability is the maintainability represented by the prob-
ability of an item working in a healthy condition after a period. The primary purpose of reliability
engineering is to prevent the occurrence of faults (Kothari et al. 2022). Power electronics system
components are individually responsible for the reliable and safe operation of the complete system.
A single component failure may cause a total shutdown, downtime, and a huge maintenance cost.
Therefore, throughout the lifetime, reliable operation is essential. The chapter covers the reliability
evaluation and enhancement methods of the power electronics system and their fragile compo-
nents like semiconductor switches, capacitors, and sensors. According to a survey, power electron-
ics semiconductors are the most vulnerable among the other components (Sarita et al. 2023). In
almost 31% of cases, the failure is due to some fault in semiconductors. In contrast, 60% of the fail-
ure in the semiconductors is due to thermal stress. The chance of failure of power semiconductor
devices increases twice for every 10∘ C rise in working temperature.
Following the explanation above, semiconductor switches provide a significant role in the reli-
ability of the electrical converter system. Various methodologies have been developed to address
the challenges related to semiconductor reliability, such as remaining useful life estimation, active
thermal control, and condition monitoring.
In a PV system, PV inverters efficiently convert DC voltage to AC voltage suitable for the load or
grid. These PV inverters have the highest failure rate among all the other constituents of the system.
Inverter failure for five years in the PV production plant was estimated to account for 37% of all
unscheduled maintenance and 59% of the associated expenditures. Throughout the PV system’s
life, inverters must be replaced approximately three to five times.

Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
412 21 Reliability Evaluation of Power Electronics Converters for Modern Power System Applications

In a wind generation system, generated power has high fluctuations. Power electronic converters
regulate the power and maximize the generation from wind. Since even a single wind turbine fail-
ure significantly affects the grid. Generators have a lesser failure rate but higher downtime; on the
other hand, power converters are prone to failure, and repair time is less than a wind generator
(Kumar et al. 2021).

21.2 Failures in Power Electronics Converters

Any power electronics converter consists mainly of semiconductor switches, gate drivers,
electrolytic capacitors, sensors, and microcontrollers. Semiconductor switches have the highest
failure rate. The reason behind the higher failure rate is thermal degradation. Since semiconductor
switches are made up of layers of different materials with different temperature coefficients. Two
types of failures are observed in the converter: random and systematic. Random failures, such
as hardware degradation and human error. Manufacturing processes, design modifications, or
operational procedures are the only ways to eliminate systematic failures. Figure 21.1 provides the
failure mechanism of a semiconductor switch.

21.2.1 Catastrophic Failure

Failure at a consistent rate is referred to as catastrophic failure. Long-term historical failure infor-
mation collected from the use of a product in a similar operating environment can be used to esti-
mate the failure rate. The failure rate for similar items can be taken from the handbook whenever
historical data is unavailable. For electronic components, the military handbook MIL-HDBK-217F
provides a generic failure rate. The handbook method does not consider the physics of failure;

Failure type/failure rate Failure mechanism

1. Device failure in
gate driver.
2. Bond wire rupture due
Catastrophic to IGBT short circuit
failure/constant
failure 1. High voltage
breakdown
2. Dynamic latch up
Semiconductor failure 3. High temperature
4. Impact ionization
1. Chip solder joint
Wear-out cracking
failures/non-constant 2. Baseplate solder joint
failure 3. Wire bonds loft
off/cracking

Figure 21.1 Failure mechanism of a semiconductor switch.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
21.2 Failures in Power Electronics Converters 413

therefore, reliability may sometimes need to be revised. The FIDES approach was used, which con-
sidered the factor of physics of failure. Equations (21.1)–(21.5) can be used to forecast the failure
rate (𝜆) using the FIDES technique:
𝜆 = Πpm Πprocess 𝜆phy (21.1)

∑ tannual
phase
𝜆phy = Π𝜆 (21.2)
i=1
8760 i i
where
Πpm = factor for manufacturing quality impact
Πprocess = factor for all processes such as operation and maintenance
𝜆phy = Failure rate due to physical contributions
tannual = the ith phase’s duration in a year
Πi = the ith phase’s factor for induced electrical, mechanical, and thermal overstress
𝜆i = Failure rate of ith phase
( )0.511 ln(cg )
Πi = Πplacement ΠApp ΠRugg (21.3)

∑
k
𝜆i = 𝜆0k Πk (21.4)
1

where
Πplacement , ΠApp , and ΠRugg = constituent factors of Πi
𝜆0k = base failure rate
Πk = factor physical constraints experienced during operation.
Using the above equations, we can predict the constant failure rate of the two most vulnerable
components, semiconductor (𝜆phy_scd ) and capacitor (𝜆phy_cap ). Therefore, the total constant failure
rate (𝜆C_useful )
∑( )
𝜆C_useful = 𝜆phy_scd + 𝜆phy_cap (21.5)

21.2.2 Wear-Out Failure

Stress strength analysis is used to estimate wear-out failure. A solid grasp of the failure mechanics
is required to estimate the strength or lifetime of the device connected to the failure mechanism.
Based on the likelihood that the applied stress will exceed the device’s strength, the probability of
failure is estimated (Rani and Kumar 2022). Equation (21.6) has previously analyzed the lifespan
model of the electrolytic capacitor (Lcap ):
( )
T0 −Tcap Vcap −n2
Lcap = L0 2 n 1 (21.6)
V0
where L0 denotes lifetime under voltage V0 , V0 is nominal voltage, To is nominal temperature, Vcap
and Tcap are the operating voltage and temperature for the Capacitor lifetime Lcap , and n1 and n2
are constants. The Number cycle to failure (Nf ) may be estimated as described in Eq. (21.7):
( )( )−0.3
𝛽 ton

Nf = A ⋅ ΔTj𝛼 e Tjn 1.5

(21.7)
where the constants for curve fitting are A, 𝛼, and 𝛽, respectively, Tjn is junction temperature, ΔTj
is the deviation in junction temperature, and ton is the rise time of the temperature cycle.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
414 21 Reliability Evaluation of Power Electronics Converters for Modern Power System Applications

The wear-out failure rate estimation process evaluation method using a mission profile is preva-
lent (Miveh et al. 2015). The mission profile is translated into stress on the component. Dividing by
the applied stress, the device’s damage is calculated. This can be interpreted as how many opera-
tions must be completed before failure. Therefore, the damage distribution function for a mission
profile may be predicted using Monte Carlo simulation. The failure rate and wear-out failure den-
sity function can be calculated using this technique.
The failure rate for wear-out (𝜆C_wearout ) is provided by Eq. (21.8):
∑
𝜆C_wearout = (𝜆wearoutscd + 𝜆wearoutcap ) (21.8)
Therefore, the total failure rate in a converter (𝜆C_converter ) can be calculated and depicted as
Eq. (21.9):
𝜆C− converter = 𝜆C− useful + 𝜆C− wearout (21.9)
So, the converter’s Total reliability R(t) is calculated using Eq. (21.10):
t
R(t) = e− ∫0 𝜆C_converter (𝜏)d𝜏 (21.10)

21.3 Estimation and Monitoring of Junction Temperature

The wear-out failure mechanism is the most common in power semiconductor devices due
to high thermo-mechanical stress during thermal and power cycles. Apart from this change
in junction temperature, high-frequency operation of semiconductor switches leads to high
temperatures in the nearby areas. Therefore, a better understanding of junction temperature
helps design the converter to ensure reliable operation. The junction temperature information
helps predict the remaining useful life and assists in developing thermal management algorithms
for optimal loss management of IGBT modules during degraded operation (Choi et al. 2016).
Temperature-dependent electrical parameters (TDEPs), such as infrared or thermocouples, can
indirectly be used to determine junction temperature. The junction temperature measurement by
the direct approach is only possible with some changes in package modification.

21.3.1 Direct Method

Modern-day semiconductor packaging is done so that direct contact with the thermocouple, ther-
mistor, and sensors is impossible. Direct methods can be used if direct contact with the switches is
possible. The coating material determines the spatial resolution of a thermocouple probe, and the
thermal capacitance of the probes regulates the response time. In laboratory testing-based convert-
ers having IGBT ICs with exposed legs, a direct method can be used. However, indirect electrical
methods for junction temperature estimation in commercial compact converters with modules are
preferred.

21.3.2 Indirect Method

Electrical methods are the most widely used methods to determine junction temperature. The
monitoring of electrical parameters is more accessible than the physical direct methods due to
practicability and fast response time. Continuous monitoring of electrical parameters on which
junction temperature depends is an excellent method to estimate junction temperature.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
21.3 Estimation and Monitoring of Junction Temperature 415

21.3.2.1 Estimation by Vceon(Ihigh)

The Vceon(Ihigh) is nonlinearly related to the junction temperature and collector current. For error
within an acceptable bound limit, the relationship based on electro-thermal modeling is approxi-
mated by Eq. (21.11):
( ) [ ( )]
Vceon(Ihigh) = Vceo + 𝛼vceo Tj − Tjo + Ic ro + 𝛽ro Tj − Tjo − 𝛾ge ΔVge (21.11)

where
Vceon(Ihigh) = collector-emitter voltage under high current
Vceo = collector-emitter voltage at Tjo
Tjo = Base junction temperature
𝛼vceo = Temperature coefficient of Vce
IC = Collector current
r0 = Series resistance at Tjo
𝛽ro = Temperature coefficient of ro
𝛾ge = Temperature coefficient of ΔVge
ΔVge = Gate emitter voltage variation
During normal operation, Vge barely varies. Therefore, ΔVge may be ignored in the equation.
Considering the series module resistance, Tj is represented in Eq. (21.12):
[ ( )]
Vceon(Ihigh) − Vceo − IC ro + 𝛽ro Tj − Tjo
Tj = + Tjo (21.12)
𝛼vceo
Assuming temperature distribution is homogeneous, the estimated junction temperature (Tjest ) can
be expressed as Eq. (21.13):
Vceon(Ihigh) − Vceo − IC ro
Tjest = + Tjo (21.13)
𝛼vceo
Therefore, the estimated junction temperature using Vceon(Ihigh) is highly sensitive to the tempera-
ture coefficients 𝛼vceo , 𝛽ro , and 𝛾ge . In one of the studies, the sensitivity range is from 1 to 5 mV∕∘ C
for 300–1000 A, respectively.
This measurement method provides an option to measure junction temperature in each switch-
ing period. Many researchers came up with different measurement configurations to estimate the
Vceon(Ihigh) of semiconductor devices (Falck et al. 2018; Lelis et al. 2014). However, continuous high
ohmic drop across modules at high currents is a big disadvantage and leads to significant errors in
temperature measurement. Moreover, Vceon is heavily affected by the aging of the power module.
As the device ages, correlation factors need to be updated occasionally. A circuit to measure Vceon
is shown in Figure 21.2.

21.3.2.2 Estimation by Threshold Voltage (Vthr )

It is suggested by the researchers if the TDEP parameters have a linear dependency on the tem-
perature, it becomes a favorable condition for temperature estimation. When the device is turned
on, the gate emitter voltage is equivalent to the threshold voltage. The authors explained that the
threshold voltage highly depends upon the junction temperature because parameters like carrier
concentration and oxide layer charges depend on the manufacturing process and remain fixed. It
is described in Eq. (21.14):
√ √ { }
{ } √
( ) k NA Qf + Qm + Qot 4 ⋅ 𝜀s NA ⋅ k √
√ NA
Vthr Tj = ln ⋅ Tj − + ⋅ ln ⋅ Tj (21.14)
q ND Cox Cox ni Tj
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
416 21 Reliability Evaluation of Power Electronics Converters for Modern Power System Applications

Vc

Vcc D

S
Vout

Upper IGBT

Vcc

Figure 21.2 Circuit for measurement of Vceon .

where k = Boltzmann constant, q = elementary charge, NA = acceptors concentration, ND denotes

donors concentration, Cox = oxide layer capacitance, and Qf , Qm , and Qot are the fixed oxide charge,
mobile ion charge, and intrinsic charge of oxide, respectively. 𝜀s represents silicon permittivity. For
continuous monitoring of Vthr , the voltage developed in the parasitic inductance due to the turn-on
current is compared with the reference voltage. The choice of the reference voltage is crucial and
should be done carefully, as the transient present in the turn-on process is affected by the junction
temperature. The pulse generated by the comparator is used to create the measuring pulse with
the help of the NAND gate and driver output voltage. The measuring pulse is applied to held Vthr
by sample and hold circuit, as shown in Figure 21.3. Based on temperature sensitivity, the Vthr
monitoring method is better than the Vceon because the external gate resistance is less than the
internal gate resistance. Also, lesser resistance from the internal gate side may lead to the induction
of synchronization jitter, which may give a significant error.
The Vthr calculation can be incorporated in the gate-driver circuit, as the reference of the ground
for both circuits is standard. From the graph, it can be concluded that temperature-dependent volt-
age variation is independent of the collector current for a limited range.

21.3.2.3 Estimation by Short Circuit Current (SCC)

A relation between temperature and short circuit current (SCC) must be established to measure
the junction temperature by the SCC method, independent of the gate voltage. A short circuit and
an additional bypass IGBT switch are developed to achieve this. This switch is turned on for short
intervals when the switch under test (SUT) is in conducting mode. This arrangement requires an
additional protection circuit, which increases the complexity of the circuit. The SCC of IGBT is
calibrated in terms of temperature, and a curve is plotted between SCC and temperature. This curve
gives the junction temperature for a particular SCC. The IGBT switch will experience high voltage
and high current stress concurrently throughout the test, so assessing its short-circuit handling
capacity is crucial before the test. The curve in Figure 21.4 (SCC decreases with temperature rise)
shows that the SCC exhibits a negative temperature coefficient.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
21.3 Estimation and Monitoring of Junction Temperature 417

C1 C2
Vg

–
+ Vge + Vth
R1 R2
–
C3 C4 C5
Ve
+–

R3 R4 +
– Vref

Figure 21.3 Circuit to measure Vthr .

220
IGBT short circuit current (A)
210 Linear (IGBT short circuit current (A))
IGBT short circuit current (A)

200

190
y = –3.7091x + 212.3
R2 = 0.9857
180

170

160
30 40 50 60 70 80 90 100 110 120 130
Junction temperature (°C)

Figure 21.4 Short circuit current versus junction temperature.

21.3.2.4 Estimation by Turn-On/Off Delay Time

Tj can be estimated by turn-on delay time tdon also. The switch’s tdon depends upon the external
and internal gate resistance, gate capacitance, and threshold voltage, as Eq. (21.15) describes. The
capacitance (C) includes depletion capacitance and oxide capacitance and RGtotal the external and
internal gate resistance
( )
( ) Vth
tdon = RGtotal ⋅ C ⋅ ln 1 − (21.15)
ΔU
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
418 21 Reliability Evaluation of Power Electronics Converters for Modern Power System Applications

1150
300 V 400 V 500 V

1100
Turn -on- delay time (ns)

1050

1000

950

900

850
30 37 44 51 58 65 72 79 86 93 100 107 114
Chip temperature (°C)

Figure 21.5 tdon versus chip temperature.

Depletion capacitance is highly governed by the carrier concentrations, which are the function
of temperature. As the temperature increases, carrier concentration increases, which leads to an
increase in the depletion capacitance. A turn-on delay time has a linear relationship with the
overall gate capacitance, and as temperature increases, tdon will also increase. To estimate tdon ,
an additional timing-counter circuit is used, and a plot between the tdon and Tj is depicted. In
this method, tdon is measured when the current in the switch is zero. The plot between tdon and
Tj is given in Figure 21.5, which also explains that the tdon decreases as the Vdc increases and is
independent of almost independent of the current in the switch (Luo et al. 2015). The method is
more suitable compared to other current dependent methods, and the sensitivity of this method
follows in the range of 1–2 ns per degree centigrade. For high current injection, temperature
measurement of the SUT can be done by Vceon(Ihigh) , which has good temperature sensitivity in the
range of 0.5–3.5 mV∕∘ C. This temperature sensitivity varies according to the load current. But for
a lower range of current, this method is not suitable. However, the turn-off delay tdoff approach
has a relatively constant temperature sensitivity, approximately 4 ns/ ∘ C. Therefore, the tdoff
method is appropriate for the lower current range but cannot be employed for the power diodes
(Luo et al. 2015).
The plot between junction temperature and turn-off delay is depicted in Figure 21.6. It can be
concluded from the plot that turn-off delay time monotonically rises with a rise in junction tem-
perature. Also, the junction temperature for a particular DC link voltage increases for the specific
turn-off time if the load current increases. An increase in DC link voltage also shifts the curve
upwards, which means that for particular junction temperatures, turn-off delay time increases with
an increase in DC link voltage.

21.3.2.5 Estimation by Rgint with the Help of Igpeak

In this method, the temperature measurement of the internal thermo-sensitive gate internal resis-
tance (Rgint ) is positioned in the module’s center. In this method, the Junction temperature can be
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
21.3 Estimation and Monitoring of Junction Temperature 419

1100
300 V@94 A 500 V@172 A 500 V@85 A

1050
Turn -off time (ns)

1000

950

900

850
6 51 71 91 120
Chip temperature (°C)

Figure 21.6 tdoff versus chip temperature.

measured by measuring the temperature of the internal gate resistor in the middle of the module.
This Rgint is thermos-sensitive. During the running conditions, it is nearly impossible to open the
power module and measure the temperature. To eliminate this problem, a researcher proposed a
method to measure the gate current variation because of a thermos-sensitive internal gate resis-
tance change by measuring the voltage drop across external gate resistance (Rgext ). R-L-C network
for gate drive circuit is used (Baker et al. 2015), which operates in overdamped condition. Therefore,
gate current (Ig ) is estimated by Eq. (21.16).
−t
V
Ig = e Rtotal C (21.16)
RTotal
where Rtotal = Rgext + Rgint . This study assumes that the changing temperature does not affect exter-
nal resistance; therefore, Rgint is calculated using Eq. (21.17).

Vgneg − Vgpos
Rgint = Vpeak
− Rgext (21.17)
Rgext

Tj can be estimated based on calibration. The experiment conducted by the researcher obtained a
temperature sensitivity of +3 mΩ∕∘ C, which gives a linear relation between the external temper-
ature and internal resistance, as shown in Figure 21.7a,b. The main advantage of this method is
that there is no loading effect on the measurement circuit, which is an integral part of the gate
driver. However, errors may appear due to different assumptions, and the aging of the module cor-
rection factor needs to be calibrated from time to time. Other than these above-discussed methods,
there are various parameters like rate of change of collector current (dIC /dt), Forward voltage drop,
Voltage between Kelvin emitter and power emitter, Miller plateau voltage, Saturation current, etc.,
which can be monitored continuously to estimate junction temperature.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
420 21 Reliability Evaluation of Power Electronics Converters for Modern Power System Applications

Peak Analog to digital

detector converter

Memory
capacitor
Reset/control
Rg,ext +
Q1

Gate driver

(a)

4.7
2.82 mΩ/°C Linear (2.82 mΩ/°C)

4.6

4.5
RG,int (Ω)

4.4

4.3
25 38 44 50 58 63 78 82 98 105 118 125 135 140
Junction temperature (°C)
(b)

Figure 21.7 (a) Circuit for detecting the gate peak voltage and (b) Rg versus Tj plot.

21.4 Reliability of a Modern Power System

The uninterrupted ability of the MPS to satisfy its electrical power demand is a measure of its
reliability. The availability of the component is considered an important parameter to measure
reliability. Availability is the probability that any item operates properly at any instant of time “t”
after starting at t = 0. Failure rates may be constant or time-varying. Markov’s principle is used to
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
21.4 Reliability of a Modern Power System 421

obtain the availability of the exponentially distributed system. Availability is obtained as stated in
Eq. (21.18):
𝜇
A=1−U =1− (21.18)
𝜆+𝜇
where
A = Availability
U = Unavailability
𝜆 = Failure rate
𝜇 = Repair rate.
According to Markov’s principle, availability (A) is the probability of being in the operating state
“1.” “2” is downstate in the state space model. Markov’s principle cannot be used for a system with
nonexponential failures. In such cases, the failure rate is separated into two failure rates, which are
constant and variable (𝜆(t) = 𝜆constant + 𝜆wear-out (t)) as presented in Figure 21.8a,b. An inconsistent
operation or impulse overstressing might lead to the random failure of converter components like
switches and capacitors. In contrast, wear-out failures happen as a result of long-term deterioration.
Therefore, these two failures can be split into two different failure rates. As Eq. (21.19) describes,
total availability can be calculated:

Atotal (t) = Aconstant ⋅ Awearout (t) (21.19)

Figure 21.8 Markov’s models. (a) For constant

failure rate. (b) For time-varying failure rate. λ = constant

1 2

(a)
λwear-out(t)
Success

1 2
λ(t)
μ
1 2 λconstant

μ 1 2

(b)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
422 21 Reliability Evaluation of Power Electronics Converters for Modern Power System Applications

where
Atotal (t) = Availability of total system
Aconstant (t) = Availability associated with time-constant failure rate
Awearout (t) = Availability associated with time-varying failure rate
Semi-Markov’s process can be used to calculate the availability of time-varying failure systems.

21.4.1 Power Converter Availability Model

The converter’s reliability can be modeled using its component’s reliability. The failure rate of
capacitor and switches are very high. Series relation and Markov’s model of the converter are
shown in Figure 21.9. When all the components are available, the availability can be obtained using
Eq. (21.20) (Falck et al. 2018):
Aconverter (t) = Aswitch (t) ⋅ Acapacitor (t) ⋅ Aoc (21.20)
where
Aconverter (t) = Availability of the converter
Aswitch (t) = Availability of semiconductor switches
AOC = Availability of other components of converter.

21.4.2 HVDC Availability Model

A sending end converter, a DC transmission line, and a receiving end resistor are the three major
components of the HVDC system. Modeling the system as a series network is possible, as seen
in Figure 21.10. Equation (21.21) can be used to determine the availability of the HVDC system
AHVDC (t):
AHVDC (t) = ASEC (t) ⋅ AREC (t) ⋅ ADC (21.21)

Power converter Success λswitch(t)

λswitch(t) = λW-switch(t) + λc-switch

Power 1 2
semiconductor
devices μswitch
λcapacitor(t)
λcapacitor(t) = λW-capacitor(t) + λc-capacitor
Capacitors 1 2

μcapacitor
λOC

Other
components
1 2

μOC

Figure 21.9 Power converter availability block diagram with Markov’s model.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
21.4 Reliability of a Modern Power System 423

HVDC system Success λSEC(t)

λSEC(t) = λW-SEC(t) + λC-SEC

1 2
Sending end
converters
μSEC
λDC

Transmission 1 2
line
μDC
λREC(t)

Receiving end 1 2 λREC(t) = λW-REC(t) + λC-REC

converters

μREC

Figure 21.10 HVDC system availability block diagram with Markov’s model.

where
ASEC (t) = Availability of sending end converter
AREC (t) = Availability of receiving end converter
ADC = Availability of DC transmission line.

21.4.3 Reliability Model of Modern Power System

MPS reliability is measured with the help of probabilistic indices. LOLE is the number of days
per hours in a specified time when power cannot be supplied due to energy shortage, which is
represented as Eq. (21.22):
∑
n
( )
LOLE = Px ⋅ Cx − Lx (21.22)
x=1
where
Cx = Available capacity
Lx = Peak load forecast
Px = Probability of loss of load
EENS = curtailed energy (Ex ) because of generation shortage, calculated as Eq. (21.23):
∑
n
EENS = Px ⋅ Ex (21.23)
x=1

A wear-out failure rate of the wind turbine converter is estimated for the wind farm power system
reliability model under a specific mission profile. Then, the wind turbine converter’s availability
is evaluated using random and wear-out failure rates. Availability of the wind turbine system is
evaluated using components and power availability. The availability of wind farm HVDC can be
estimated using historical data and failure rates. Combining the models for the transmission lines
and the wind farm’s turbine system yields the wind farm reliability model. Similarly, the avail-
ability of the HVDC transmission line and conventional generators can also be estimated using
converter chance failure and individual availability, respectively. The combined reliability is shown
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
424 21 Reliability Evaluation of Power Electronics Converters for Modern Power System Applications

Mission profile model

Converter Wind turbine Wind power

Reliability data
reliability reliability uncertainty
and mission
model model model
profile
Wind turbine
Line converter generation
reliability model model

Wind farm generation

model
Converter reliability
model/HVDC system
model Generation System load
system model
Generators model
Convolution
model of generation
and system
load model

LOLE/EENS

Figure 21.11 Reliability evaluation in MPS.

in Figure 21.11. The generation system reliability and load can be convolved to produce power sys-
tem reliability indices like LOLE and EENS.

21.5 Challenges and Future Directions

The chapter has discussed the types of failures (constant and time-varying) common in a power
electronics-based system. Temperature effect on the performance of switches is crucial. There-
fore, continuously monitoring temperature-dependent parameters is an excellent option to increase
system reliability. The reliability modeling of a MPS with the help of Markov’s model has been
discussed.
Condition monitoring of TDEPs is a low-cost and affordable solution for reliability enhance-
ment. Still, it is necessary to distinguish temperature-dependent from damage-dependent electrical
parameters. A more acute study of the physics of failure is needed to accomplish this goal. Further
advancement of condition monitoring methods at the converter level is required to identify more
failure indices based on power system output parameters. The model-based lifespan estimation
approach is preferred over the data-driven based model due to its simplicity and accuracy. Real-time
monitoring for power electronics converter systems will be effective for better mission profile data,
ultimately enhancing the system’s availability and reliability.

References

Baker, N., Munk-Nielsen, S., Iannuzzo, F., and Liserre, M. (2015). IGBT junction temperature
measurement via peak gate current. IEEE Transactions on Power Electronics 31 (5): 3784–3793.
Choi, U.-M., Blaabjerg, F., Jørgensen, S. et al. (2016). Reliability improvement of power converters by
means of condition monitoring of IGBT modules. IEEE Transactions on Power Electronics 32 (10):
7990–7997.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 425

Falck, J., Felgemacher, C., Rojko, A. et al. (2018). Reliability of power electronic systems: an industry
perspective. IEEE Industrial Electronics Magazine 12 (2): 24–35.
Kothari, D.P., Nagrath, I.J., and Saket, R.K. (2022). Modern Power System Analysis. Tata McGraw-Hill.
Kumar, S., Saket, R.K., Dheer, D.K. et al. (2020). Reliability enhancement of electrical power system
including impacts of renewable energy sources: a comprehensive review. IET Generation,
Transmission & Distribution 14 (10): 1799–1815.
Kumar, S., Sarita, K., Saket, R.K. et al. (2021). Reliability assessment for DFIG-based WECS
considering the impact of 3-phase fault and lightning impulse voltage. International Transactions on
Electrical Energy Systems 31 (8): e12952.
Lelis, A.J., Green, R., Habersat, D.B., and El, M. (2014). Basic mechanisms of threshold-voltage
instability and implications for reliability testing of SiC MOSFETs. IEEE Transactions on Electron
Devices 62 (2): 316–323.
Luo, H., Chen, Y., Sun, P. et al. (2015). Junction temperature extraction approach with turn-off delay
time for high-voltage high-power IGBT modules. IEEE Transactions on Power Electronics 31 (7):
5122–5132.
Miveh, M.R., Rahmat, M.F., Ghadimi, A.A., and Mustafa, M.W. (2015). Power quality improvement in
autonomous microgrids using multi-functional voltage source inverters: a comprehensive review.
Journal of Power Electronics 15 (4): 1054–1065.
Rani, P. and Kumar, S. (2022). Probabilistic analysis of wind integrated power system considering
3-phase line fault. In: 2022 IEEE 10th Power India International Conference (PIICON), 1–6. IEEE.
Sarita, K., Saket, R.K., and Khan, B. (2023). Reliability, availability, and condition monitoring of
inverters of grid-connected solar photovoltaic systems. IET Renewable Power Generation 17 (7):
1635–1653.
Singh, S., Saket, R.K., and Khan, B. (2023). A comprehensive state-of-the-art review on reliability
assessment and charging methodologies of grid-integrated electric vehicles. IET Electrical Systems in
Transportation 13 (1): e12073.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
427

22

Reliability Assessment of Sub-components of Electric Vehicle for

Performance Enhancement Grid Integrated Power System
Saumya Singh 1 , Dhawal Dwivedi 1 , Sandeep K. Soni 2 , R. K. Saket 1 , and Dwarkadas P.
Kothari 3
1
Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
2
Department of Electrical Engineering, Shiv Nadar Institution of Eminence, Delhi, NCR, India
3
Department of Electrical Engineering, Indian Institute of Technology, Delhi, India

22.1 Introduction
EVs and the demand for sustainable mobility are transforming the global automotive sector. The
broad adoption of EVs is essential to reducing greenhouse gas emissions and combating climate
change. EV zero-tailpipe pollution and energy efficiency will change urban and long-distance
travel. The success of EVs is not solely dependent on their electrification but also on the seamless
integration of these vehicles into existing power systems. Grid integration represents a crucial
milestone in the EV revolution, as it allows for efficient charging infrastructure, vehicle-to-grid
(V2G) capabilities, and optimal power management to improve overall system performance.
Achieving a harmonious coexistence between EVs and the electric grid requires a comprehensive
understanding of the reliability of their sub-components. This chapter serves as a comprehensive
guide for researchers, engineers, and policymakers seeking to understand the multifaceted aspects
of reliability assessment and performance enhancement in the context of EVs and their integration
into grid-connected power systems. Through synthesizing theoretical insights and practical
applications, we aim to contribute to the evolving landscape of sustainable transportation and
its pivotal role in shaping our future. When individuals are in the process of acquiring a vehicle,
their primary concerns often revolve around the vehicle’s reliability, safety, cost, and ease of
maintenance. Plug-in electric vehicles (PEVs) are engineered with many electrical components
and systems, including batteries, motor drives, controllers, and energy management systems.
These intricate electrical systems inherently possess a higher susceptibility to failure. Conse-
quently, PEVs typically exhibit lower reliability than mechanically driven internal combustion
engine-based vehicle systems.
The subsequent sections of this chapter are organized as follows: Section 22.2 presents a compre-
hensive discussion on exploring and analyzing various sub-components within EVs. Section 22.3
includes charging infrastructure, various types of drive motors used in EVs, reliability studies of
sub-components in EVs, methodologies of reliability studies, load management, and the benefits
and challenges of grid integration. Section 22.4 exhibits reliability assessment techniques in EVs
and battery management systems (BMS). Section 22.5 explains reliability parameters, distribution
network (DN) reliability assessment, reliability evaluation using a V2G approach and high EV pen-
etration. Finally, Section 22.6 brings the chapter to a conclusion.
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
428 22 Reliability Assessment of Sub-components of Electric Vehicle for Performance Enhancement Grid Integrated

22.2 Electric Vehicles and Grid Integration

Grid integration for electric cars integrates them into their existing electrical systems. To maintain
an uninterrupted and steady power supply, regulating EV’s impact on the grid is crucial as more
people embrace them. EVs are becoming integral components of modern power systems, pivotal
in the transition toward more sustainable transportation and energy consumption. This chapter
explores the intricate interplay between EVs and power systems, focusing on the grid integra-
tion aspects from a power system perspective. The following are some EV grid integration aspects
described in terms of the power system (Kothari et al. 2022).
Future Solutions for Grid Integration

a. Vehicle-to-Grid (V2G) Technology: Two-way power transmission between EVs and the grid
is made possible by V2G technology. EVs have the ability to release electricity back to the grid
during peak demand, aiding in the reduction of strain on the grid. V2G can also provide a source
of distributed energy storage to enhance grid resilience. EVs can use V2G to draw electricity from
the grid and feed surplus energy back into it, serving as distributed energy resources to improve
grid stability, as discussed in Singh et al. (2023b).
b. Grid Modernization: The ongoing modernization of electrical grids with advanced sensors,
automation, and communication technologies will enable more efficient load management and
better grid stability.
c. Renewable Energy Integration: By synchronizing the charging of EVs with the availability of
renewable energy sources (RES), such as solar and wind power, we can achieve two significant
advantages:
(i) Reduced Environmental Impact: When EVs are charged using clean, renewable energy
(RE), their carbon emissions are greatly diminished. This means that the electricity pow-
ering these vehicles comes from eco-friendly sources, resulting in a lower overall environ-
mental footprint for EV usage.
(ii) Enhanced Grid Efficiency: Coordinating the charging of EVs with periods of high RE
production helps maintain a well-balanced electrical grid. During times when wind or solar
power generation is abundant, we can encourage more EV charging, thus relieving strain
on the grid and ensuring efficient utilization of RES.
Coordinating EV charging with RE generation, such as wind and solar, can reduce the car-
bon footprint of EV charging while optimizing grid usage. Integrating EVs with RES, such
as solar and wind, can enhance the sustainability of the transportation sector. EV charg-
ing can be encouraged during high RE generation periods, promoting a greener and more
balanced power system.
d. Interconnected Networks and Distribution System Planning: Grid operators are explor-
ing ways to share load and resources between neighboring grids to mitigate the localized impact
of EV charging.
EV adoption estimates must be considered by utilities and grid operators when developing their
long-term distribution systems. As a result of the aforementioned, the grid is guaranteed that
it will be able to support both user’s overall battery-powered cars and regular power users with
the same level of efficiency.
e. Data and Communication Systems: Successful management of EV grid integration depends
on maintaining reliable data and communication systems. Grid operators can make informed
judgments and optimize charging schedules using real-time data on EV charging routine, grid
load, and RE availability.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
22.2 Electric Vehicles and Grid Integration 429

Integrating EVs into the electrical grid is a complex task that requires careful planning, invest-
ment, and innovative solutions. Load management and grid stability are critical aspects that
must be addressed to ensure a smooth transition to a cleaner and more sustainable transporta-
tion system. Collaboration among government agencies, utilities, technology providers, and EV
manufacturers is essential to successfully manage the challenges and opportunities presented
by the electrification of transportation (Das et al. 2020).

22.2.1 Charging Infrastructure

EV grid integration involves establishing charging infrastructure in several locations to meet EV
charging demands. This infrastructure can include public charging stations, fast chargers, home
chargers, and workplace chargers. Charging infrastructure plays a pivotal role in successfully inte-
grating EVs into our transportation ecosystem. As EVs continue to gain popularity and become a
significant part of the global vehicle fleet, developing an extensive and reliable charging network
is crucial for promoting widespread adoption and addressing concerns related to range anxiety. In
this discussion, we will delve into the various aspects of charging infrastructure, its importance,
and the challenges as discussed in Yilmaz and Krein (2012).
Types of Charging Infrastructure: Charging infrastructure encompasses a range of technolo-
gies and charging speeds to cater to the diverse needs of EV users. The following levels of charging
comprise the main categories of charging infrastructure:
● Level 1: This level is the slowest charging option, using a standard 120-V household outlet. Level
1 chargers are suitable for overnight charging at home and provide an average of 2–5 miles of
range/h.
● Level 2: This charging level uses a 240-V power source and is commonly found in residential and
commercial settings. They offer faster charging speeds, providing around 10–30 miles of range
per hour of charging.
● Level 3 (DC Fast Charging): DC fast chargers are high-power chargers that can deliver a sub-
stantial amount of energy quickly, providing up to 80% charge in 30–60 minutes. These are typi-
cally located along highways and major travel routes.
Importance of Charging Infrastructure
– Range Confidence: An extensive charging network helps alleviate range anxiety, a common
concern among potential EV buyers. Knowing that charging stations (CS) are readily available
enhances driver confidence in the viability of EVs for long trips.
– Accessibility: Charging infrastructure is essential to ensure that EVs can be conveniently
charged at home, work, or on the road. This accessibility is essential for EV owners who may
not have access to home charging.
– Urban Mobility: In urban areas, public charging infrastructure is critical for residents who
may not have private garages or dedicated parking spots. It encourages the use of EVs in
densely populated regions.
– Reduced Carbon Emissions: Promoting EV adoption through charging infrastructure
reduces greenhouse gas emissions from the transportation sector, thereby addressing
environmental concerns.

22.2.2 Load Management and Grid Stability

EVs increase grid demand, especially during peak charging. Load management must balance
this additional load without overloading the grid. Intelligent charging, time-of-use (TOU) rates,
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
430 22 Reliability Assessment of Sub-components of Electric Vehicle for Performance Enhancement Grid Integrated

and demand response programs may encourage off-peak charging. EV adoption is on the rise
worldwide, and while this transition to cleaner transportation is essential for reducing greenhouse
gas emissions, it also challenges the stability and management of electrical grids. The variable and
sometimes unpredictable nature of EV charging can impact the grid’s stability. Localized voltage
fluctuations and power quality concerns can result from rapidly charging several EVs in a crowded
location. Grid operators must regulate voltage and power factors to preserve grid stability.
Load management and grid stability are critical to integrating EVs into the existing power infras-
tructure. This discussion will explore the implications, strategies, and technologies involved in
managing EV loads and ensuring grid stability.
Impact of EV Charging on Grids
● Increased Demand: Electricity demand will increase as more EVs are put on the road, particu-
larly during peak charging hours. This can strain local grids and lead to overloads.
● Load Distribution: The distribution of EVs across regions and neighborhoods can vary signifi-
cantly, creating localized demand spikes that challenge grid operators.
● Charging Speeds: The speed at which EVs charge, from slow Level 1 charging to fast DC charg-
ing, affects the grid differently. High-speed chargers draw a substantial amount of power quickly,
requiring careful management.
Load Management Strategies
● Time-of-Use (TOU) Pricing: TOU pricing serves as an incentive for EV owners to choose
off-peak hours for charging when electricity rates are more affordable, ultimately helping to
relieve grid congestion during peak demand periods.
● Demand Response Programs: Grid operators can communicate with EVs to temporarily
reduce or delay charging during grid stress events, ensuring grid stability.
● Smart Charging: The use of smart charging systems empowers EVs to communicate with the
grid, making real-time adjustments to their charging rates based on grid conditions, the avail-
ability of RES, and user-defined preferences.
● Fleet Charging Optimization: For commercial fleets of EVs, advanced software can optimize
charging schedules to minimize operational disruptions and reduce peak load impact.
Grid Stability Considerations
1. Voltage and Frequency Control: Rapid and uncontrolled EV charging can disrupt grid volt-
age and frequency. Grid operators need advanced monitoring and control systems to maintain
stability.
2. Distribution System Upgrades: In some cases, distribution systems may require upgrades to
handle the additional load from EVs. This includes transformers, substations, and grid infras-
tructure improvements.
3. Energy Storage Integration: The deployment of energy storage systems (ESS), such as
large-scale batteries, can absorb excess energy during low-demand periods and discharge it
when needed, helping stabilize the grid.

22.2.3 Benefits and Challenges of Grid Integration

The development of a reliable and consistent charging infrastructure for EVs is made possible
through grid integration. The important features that will enhance grid integration are efficient
charging infrastructure, V2G capabilities, enhanced RE integration, demand response and load
flexibility, grid support during emergencies, etc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
22.3 Sub-components of EVs 431

Some of the common challenges of grid integration are:

● Grid Capacity and Infrastructure Upgrades,
● Peak Demand Management,
● Battery Degradation and Lifetime,
● Interoperability and Communication Standards,
● Regulatory and Policy Frameworks, and
● Cybersecurity Concerns.
Addressing the above challenges and reaping the benefits of grid integration will be critical in
promoting the effective integration of EVs into the larger energy ecosystem, thereby contributing
to a more sustainable and electrified transportation future.
Challenges in Charging Infrastructure Development
Cost: Establishing a comprehensive charging network requires significant investment in infras-
tructure development, including installation, maintenance, and grid upgrades.
Grid Capacity: Rapid adoption of EVs can strain the electrical grid, necessitating upgrades to
accommodate the increased demand for electricity during peak charging times.
Standardization: Ensuring compatibility between different EV models and charging stations is
essential. Establishing universal charging standards, such as CCS (Combined Charging System)
and CHAdeMO, helps in this regard.
Balancing Demand: Managing peak demand for electricity during charging times is crucial to
avoid grid instability and high energy costs (Giri et al. 2019).

22.3 Sub-components of EVs

The powertrain subsystem of a battery electric vehicle (BEV) is primarily composed of key
components, including a battery system, a power distribution unit (PDU), a motor controller, and
a drive motor, as illustrated in Figure 22.1. When BEVs usually function, the electric energy kept
in the battery system is fed into the PDU and then, via the PDU, to the motor controller. Powering
the motor system allows for converting electric energy into mechanical energy, which is necessary
for the BEV’s operation.
The battery system, which comprises many battery cells connected in series and parallel, mostly
stores electrical energy. In addition to battery cells, the battery system includes components such
as a battery management system (BMS) controller, power electronics elements, and other related
components. The BMS controller is in charge of managing and monitoring the battery modules.

Power train system Driving force direction

Power
Motor Propulsion Propeller
Battery distribution Electric vehicle
control unit motor shaft
unit (PDU) wheel

Direction of regenerative power

Figure 22.1 Sub-components of EVs.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
432 22 Reliability Assessment of Sub-components of Electric Vehicle for Performance Enhancement Grid Integrated

Individual battery cells’ voltage, current, and temperature can all be measured. By analyzing the
collected data, we devise a suitable control method to safeguard against irregular conditions in
the battery compartment, which includes challenges like overcharging, excessive discharging, and
temperature spikes. The components of power electronics act as a shield for battery cells. They avert
harm from high currents, handle the initiation and termination of the electrical system, and sever
the power link to the BEV motor during urgent scenarios.
● The PDU channels electric power from the battery system and provides interfaces to various
components or systems within the BEV.
● The motor controller primarily supervises the drive motor. It ensures steady and dependable
performance and delivers instant feedback about the drive system, including the motor and its
controller, to the main vehicle controller.
● The drive motor in EVs converts electrical energy from the battery into mechanical energy to
propel the vehicle.
● The drive motor’s efficiency, power output, and control systems significantly influence an EV’s
performance, range, and overall driving experience.
Concept of drive motors utilized in EVs can be described as follows:
The drive motor in EVs converts electrical energy from the battery into mechanical energy to propel
the vehicle. It is an essential part of the electric drivetrain and serves as the primary power source
for the wheels.
The drive motor operates based on the principles of electromagnetism, where electrical current
flowing through coils generates a magnetic field. This magnetic field interacts with permanent
magnets or other magnetic components, resulting in a rotational motion that drives the wheels.
There are various types of drive motors used in EVs.
● Permanent Magnet Synchronous Motor (PMSM): PMSMs have permanent magnets in the
rotor, and the flow of direct current generates the stator’s magnetic field. They are efficient, have
a high torque-to-weight ratio and less acoustic noise, offer high power density and are widely
used in EVs (Krishnan 2017).
● Induction Motor (IM): IMs have rotor windings that induce a current when exposed to the
stator’s rotating magnetic field. They are robust, reliable, and cost-effective but may have slightly
lower efficiency than PMSMs.
● Switched Reluctance Motor (SRM): SRM uses the principle of magnetic reluctance, where the
rotor aligns itself with the minimum magnetic reluctance path. SRMs are relatively simple in con-
struction and can offer high torque at low speeds. PMSM and Synchronous Reluctance Machines
(SynRel) typically require fewer conductor connections than traditional three-phase IMs. Addi-
tionally, external control for PMSM and SynRel motors is more intricate when compared to
induction and DC motors of similar types. SynRel machines excel in high-speed operations and
have low inertia, making them ideal actuators and traction drive choices. Furthermore, SRM
is naturally resilient to faults because each phase within the machine operates independently
electrically. Therefore, if one phase experiences a short-circuit fault, it does not impact the other
phases (Jayasawal et al. 2022; Krishnan and Materu 1990).
● Brushless DC Motor (BLDC): BLDC motors have permanent magnets on the rotor, similar to
PMSM, but they use electronic commutation instead of brushes. They provide good efficiency
and control but may require more complex electronics. The utilization of BLDC motors fea-
turing trapezoidal back-EMF has become prevalent across industrial, consumer, and military
sectors due to their superior torque capabilities when compared to PMSM with sinusoidal
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
22.3 Sub-components of EVs 433

VCU I/P Motor rotation

(vehicle (instrument direction OBD (on-board
control unit) panel) control diagnostics)

Auxiliary Motor controller

Current
signal
power Control
IGBT
module
DC/DC
converter Motor rotor position signal

Drive
motor
Power Air
distribution unit conditioner
(PDU)
PTC heater
Fast charging
interface Powertrain
system
Controller Power battery pack
Area network BMS (battery
management
system)

Powertrain system

Figure 22.2 Structural diagram of power-train system in a BEV.

back-EMF. This popularity is attributed to the BLDC motor’s significant advantages, including
a high torque-to-weight ratio, exceptional efficiency, and reduced acoustic noise (Dwivedi et al.
2022).
As outlined previously and illustrated in Figure 22.2, the BEV’s power-train system is intricate
with many components, and a malfunction in any one of them could compromise the whole propul-
sion mechanism.

22.3.1 Reliability Study of Sub-components in EVs

Reliability refers to the capacity of an item or system to perform its intended functions within
predetermined conditions and over a specified period. The adoption of EVs has seen a significant
rise due to their environmental benefits and increasing interest in sustainable transportation.
However, ensuring the reliability of key sub-components in EVs is crucial for their widespread
acceptance and long-term success. EV usage has increased significantly due to its environmental
benefits and growing interest in sustainable mobility. However, assuring the dependability
of essential subsystems in EVs is critical for their general acceptance and long-term success.
This research aims to thoroughly evaluate the dependability of critical EV subsystems such
as battery packs, electric motors, power electronics, charging systems, and thermal manage-
ment. The current investigation attempts to identify possible deficiencies, failure patterns, and
improvement possibilities in EV dependability by utilizing real-world data, models, and accel-
erated testing. As the automotive industry transitions toward more sustainable and eco-friendly
transportation solutions, EVs have gained significant attention due to their potential to reduce
carbon emissions and dependence on fossil fuels. One critical aspect of ensuring EVs’ success
and widespread adoption is their reliability. Reliability studies of sub-components within EVs
play a crucial role in establishing the overall dependability of these vehicles (Gandoman et al.
2019).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
434 22 Reliability Assessment of Sub-components of Electric Vehicle for Performance Enhancement Grid Integrated

Importance of Reliability Studies

Reliability refers to the ability of a component, system, or vehicle to perform its intended function
consistently and without failure over a specified period (Sarita et al. 2023). In the context of EVs,
reliability is paramount because any unexpected breakdown or failure can lead to inconvenience
for the owner, increase maintenance costs, and potentially negatively impact the public perception
of electric cars.
Scope of Reliability Studies

1. Battery Systems: The battery pack is one of the most critical components of an EV. Its reliability
directly impacts the vehicle’s driving range, performance, and overall user experience. Studying
the battery’s degradation rate, thermal management system effectiveness, and safety measures
against thermal runaways are essential to its reliability assessment.
2. Electric Motors: The electric motor is another key component that drives the vehicle’s move-
ment. Studying the motor’s durability, efficiency, cooling mechanisms, and potential wear and
tear under various driving conditions is crucial for ensuring its reliability.
3. Power Electronics: Power electronics control the flow of electricity between the battery and the
motor. The reliability of inverters, converters, and other power electronic components is critical
to maintaining a smooth and efficient power transfer. Numerous scenarios, such as operating
electric motors or feeding power into the electrical grid, there is a frequent need for a consistently
high and stable voltage level (Soni et al. 2023).
4. Charging Infrastructure: The reliability of charging stations (CS) is equally important, as it
directly affects the convenience of EV ownership. Studying factors such as charging speed, con-
nectivity, and the ability to withstand different environmental conditions is vital.
5. Connectivity and Software: Many modern EVs have advanced software and connectivity fea-
tures. Ensuring the reliability of infotainment systems, over-the-air updates, and autonomous
driving capabilities involves rigorous testing and validation.
6. Auxiliary Systems: Heating ventilation and air conditioning (HVAC), lighting, and other aux-
iliary systems contribute to the overall comfort and functionality of the vehicle. Reliability stud-
ies in these areas address performance, longevity, and energy efficiency (Kamruzzaman and
Benidris 2019).

22.3.1.1 Methodologies of Reliability Studies

Reliability studies involve various methodologies, including:

1. Accelerated Testing: Components are subjected to extreme conditions and stresses to simu-
late years of use within a shorter time frame. This helps identify potential failure points and
weaknesses.
2. Field Testing: Real-world usage data is collected from a fleet of vehicles to analyze component
performance over extended periods and under diverse conditions.
3. Failure Analysis: Investigating the root causes of component failures provides insights into
design improvements and preventive measures.
4. Statistical Analysis: Data from tests and field studies are analyzed statistically to assess failure
rates, mean time between failures (MTBF), and other reliability metrics.
Benefits of Reliable Sub-components in EVs
● Lower Maintenance Costs: Reliable components result in fewer breakdowns and repairs,

reducing maintenance expenses for EV owners.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
22.4 Reliability Assessment Techniques in EVs 435

● Sustainability: Longer-lasting components lead to a longer vehicle lifespan, contributing to

the overall environmental benefits of EVs.
● Brand Image: Automakers prioritize reliability to establish a positive brand image and gain
a competitive edge in the market.
In conclusion, conducting reliability studies on sub-components within EVs is essential for
ensuring EVs’ long-term success and acceptance in the automotive landscape. These studies
enhance the consumer experience and contribute to a more sustainable transportation future.
Manufacturers and researchers continue to refine their approaches to reliability assessment to
provide EV owners with vehicles that are not only environmentally friendly but also dependable
and trustworthy.

22.4 Reliability Assessment Techniques in EVs

Reliability assessment uses investigative approaches based on traditional methods and computa-
tional simulations. Several novel techniques have been recently suggested for reliability analysis.
To gain deeper insight into the reliability of the BEV power train, a review of the reliability metrics
was carried out for both the complete system and its separate elements. These techniques ensure
that EVs meet safety, durability, and performance standards. Here are some common reliability
assessment techniques used in the EV industry. Figure 22.3 presents the categorization of reliability
evaluation methods.

22.4.1 Markov Model

Markov models are frequently employed to analyze the reliability of engineering and systems
that are resilient to faults. The Markov model outperforms alternative approaches when assessing

Classification of reliability
assessment technique

Network
Analytically Numerically
modeling

Markov models: Methods: Monte Carlo simulation

method:
based on mathematical 1. State space method
Through simulation and
methods
2. Contingency the physical behavior
enumeration method
3. Minimal cut-set method

Figure 22.3 Classification of reliability assessment technique.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
436 22 Reliability Assessment of Sub-components of Electric Vehicle for Performance Enhancement Grid Integrated

Start

Define failure events

Define critical system

components associated with the
events

Combine system
Define Markov components with
Model
model elements equivalent failure
effects

Assign failure rate of

Markov Model
model elements

Applying
Compute Markov model state simplifying
transition based on redundancy techniques to
management reduce
reducenumber
no of
ofstates
states

Define initial state probabilities

and run Markov model

Evaluate overall
system
reliability

Figure 22.4 Markov model design process.

the reliability of a complex system where components are meticulously modeled. This modeling
technique is particularly effective for components that can be accurately characterized using
mathematical methods. The Markov model can be depicted as discrete, encompassing temporal
and spatial aspects. In evaluating reliability for electrical systems, using stationary Markov models
is fundamental. These models exhibit discreteness in spatial representation while maintaining
continuous temporal. Figure 22.4 presents an overview of the design process for the Markov model
analysis.

22.4.2 Monte Carlo Simulation (MCS)

The utilization of Monte Carlo Simulation (MCS) provides the option to incorporate more advanced
component models, encompassing the effects of component aging. The application of the MCS
method in managing EV charging within the electric power system (EPS) operation has a notable
impact on the planning of the electric market. The EV charging load mainly drives the influence.
Such an effect emerges because of the irregular distribution of EVs both in space and time.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
22.4 Reliability Assessment Techniques in EVs 437

Optimal Evaluate
Calculate Highest profit overall
Start Input data configuration
prosperity economic capacity system
consider capacity reliability

Figure 22.5 Monte Carlo simulation design process.

The MCS is the premier method for predicting EV charging duration and power. It is rooted
in the traditional probability distribution of vehicular data. In forecasting the EV charging load
using the MCS method, a Probability Distribution Function (PDF) is crafted considering starting
charging moments and conditions, factoring in aspects like the count of EVs, battery size, and other
pertinent factors. Following this, MCS utilizes mathematical simulations to sum up the charging
demands of each separate EV to compute the overall charging demand. Numerous statistical tests
are conducted through the MCS methodology to tackle both computational and tangible issues.
This specific technique simulates EVs’ driving and charging actions to determine charging
time and power based on the PDF of EV attributes. The starting moment for charging is typically
depicted, as illustrated in Eq. (22.1):
[ ( )]
⎧ 1 ( y − 𝜇x )2
⎪ √ exp − , 𝜇x − 12 < y ≤ 24
⎪ 𝜎x 2𝜋 2𝜎x2
fx ( y) = ⎨ [ ( )] (22.1)
⎪ 1 ( y + 24 − 𝜇x )2
⎪ √ exp − , 0 < y ≤ 𝜇x − 12
⎩ 𝜎x 2𝜋 2𝜎x2

Furthermore, the PDF for the daily mileage is established, as depicted in Eq. (22.2):
[ ( )]
1 (ln y − 𝜇z )2
fz ( y) = √ exp − (22.2)
y𝜎 2𝜋 2𝜎z2
z

in which the pertinent shape parameters for the PDFs are denoted as 𝜎x , 𝜇x , 𝜎z , and 𝜇z , enhanc-
ing reliability. When applying the MCS technique to address challenges, the outcomes frequently
manifest as mathematical forecasts for a designated random factor. This factor is produced via a
conceptual test using exact technical information. The issue is fundamentally tackled by using the
average value of this particular factor. The significant need for random numbers in MCS cannot be
overstated. Figure 22.5 illustrates the design process of the MCS.

22.4.3 Contingency Enumeration Method

Another analysis method is the contingency enumeration technique, which gauges reliability by
scrutinizing a set number of contingencies. This method is particularly useful for assessing the
trustworthiness of electrical systems. The contingency enumeration procedure can be broken down
into four consecutive phases. Figure 22.6 illustrates the contingency enumeration method. The ini-
tial phase outlines the analysis framework, encompassing system boundaries and the chosen load
flow technique. Moving on to the subsequent phase, the focus shifts to handling contingencies
involving various outage combinations. Here, it is essential to account for all conceivable contin-
gencies, even in cases where this leads to prolonged computational efforts. These calculations are
crucial in identifying potential electrical system issues arising from these contingencies. Ultimately,
the reliability metrics are computed, and the cumulative annualized indices are derived by aggre-
gating results across all relevant operational scenarios and potentialities.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
438 22 Reliability Assessment of Sub-components of Electric Vehicle for Performance Enhancement Grid Integrated

Specification of
Start Selection of
operating
contingencies
framework

Evaluation of
Evaluate overall Calculation of contingencies and
system reliability reliability indices consequences of
remedial actions

Figure 22.6 Contingency enumeration method process.

22.4.4 State Space Method

The state space approach generally hinges on the ideas of “upstate” and “downstate.” Fundamen-
tally, this approach is anchored in three core parameters, as depicted in Figure 22.7.
In Figure 22.8, m represents the standard operational time of the system, while r signifies the
duration of failure, and T stands for the average time between instances of failure.
The diagram in Figure 22.8 illustrates the state space of an individual component within a larger
system. As in this context, t, 𝜆 represents the rate of failures, while 𝜇 pertains to the rate of repairs
(Ghosh et al. 2013).
In the context of reliability analysis using state space methods, MTTF (mean time to failure
[uptime]), MTBF (mean time between failures-cycle time: [sum of the uptime and downtime]), and
MTTR (mean time to repair-downtime) are important reliability metrics. These metrics help assess
and understand the performance and availability of systems or components, as already discussed
in Chapters 20 and 21.

m r m r

0
t

Figure 22.7 State space method process.

λ = 1/m Figure 22.8 Diagram depicting the state space method for an
individual element within a system.

U D

μ = 1/r
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
22.4 Reliability Assessment Techniques in EVs 439

22.4.5 Electric Vehicles Trend

Based on data from the transport ministry, India’s car sector presently holds the fifth position world-
wide and is forecasted to climb to the third spot by 2030. In the quest to reduce greenhouse gas
releases, EVs emerge as vital solutions. Utilizing RESs for recharging or generating energy onboard
can eradicate fuel expenses. Based on background information, categorization, and adoption rates,
EVs can be divided into hybrid and battery EVs. The literature has addressed challenges, including
inadequate battery charging infrastructure, extended charging durations, limited driving ranges,
and performance shortcomings (Singh et al. 2023a).
Numerous obstacles must be addressed, encompassing battery expenses, efficient charging tech-
niques, and enhancing charging station effectiveness and efficiency. Ensuring dependable charging
speeds stands out as a critical factor. For bridging the link between the electric grid and EV battery
setups, affordable and dependable power converters play a crucial role in enhancing EV charging
techniques. Recent research predicts that the EV market was valued at 170 billion in 2021 and is
projected to exceed 1103 billion by 2030, as illustrated in Figure 22.9. With such substantial global
sales growth expected in a decade, it would be reasonable to anticipate that EVs will become com-
monplace in everyday life.
In India, statistics show that roughly 1 in every 125 vehicle is electric, predominantly
two-wheelers and three-wheelers. The main elements impacting EV efficiency have been steady
for the last 20 years. These elements encompass limited product advancements, cost issues, and
an inadequate density of charging facilities, collectively hindering the broad acceptance of EVs in
the country (Rivera et al. 2021).

22.4.6 Fundamental Considerations in Assessing the Reliability of EVs

Car manufacturers have undergone substantial policy and strategy adjustments with the increasing
global demand for EVs. Within this context, a critical emphasis lies on assessing reliability, which

Other
India EV sales share
50 Japan PHEV share in EVs
Sustainable development scenario

United States
Europe
China

40
Stated policies scenario
Millions

30

20

10

0
2020 2025 2030 2025 2030
Years

Figure 22.9 Predictive data for global EV sales from 2020 to 2030.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
440 22 Reliability Assessment of Sub-components of Electric Vehicle for Performance Enhancement Grid Integrated

can be divided into two principal domains:

1. The evaluation of reliability in contemporary electrical grids incorporating EVs.

2. The analysis of reliability concerning the internal systems of EVs.

When assessing the reliability of EVs, various viewpoints need to be considered, such as:

● Understanding reliability from the perspective of the customer.

● Interpreting reliability from the standpoint of the manufacturer.
● Grasping reliability from the viewpoint of the seller.

The evaluation of reliability from the aforementioned perspectives holds substantial impor-
tance in the strategic planning and overall operational lifespan of various components within
EVs, encompassing aspects like power electronic converters, battery packs, and electric motors.
Within the operational phase of EVs, three distinct areas (production, sales, and customers) carry
relevance from a reliability standpoint. Figure 22.10 visually represents these three zones, demon-
strating the implementation of each element in assessing EV reliability. The utmost significance
of the reliability concept for the components of EVs.
Due to the presence of multiple sub-systems within EVs, conducting reliability and safety anal-
yses becomes notably challenging. Moreover, the intricate interdependence among components
adds uncertainty, prompting the utilization of a logical framework to explore the uncertain rela-
tionships of failures in EV components. To evaluate the reliability of electric vehicles (EVs), it is
essential to consider the failures that arise in crucial electrical components. Figure 22.11 depicts
the categorization of primary failures in the electrical components of EVs. The crucial factors to be
considered in the reliability assessment of EVs are outlined as follows:

● Understanding the functioning of power components within EVs.

● Pinpointing component malfunctions in EVs.
● Analyzing the sequence of failures.
● Proposing a framework to depict these malfunctions.
● Choosing an approach to gauge the reliability of EVs.
● Evaluating reliability from the manufacturing angle is complex, given the myriad of procedures
in battery cell production.

To investigate the reliability of the battery pack, it is imperative to address four significant
concerns.

Planning side

Manufacture side Customer side

Operation side Operation side

Seller side

Production Selling Customer

Lifetime of EV

Figure 22.10 The utmost significance of the reliability concept for the components of EVs.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
22.4 Reliability Assessment Techniques in EVs 441

Overvoltage failures

Low voltage failures

Voltage High-voltage insulation failures

failure
High-voltage loop failures
Current
Short circuit failures
Battery failure
system Low temperature failures
failures Temperature
failure Over temperature failures

Pre-charge failures
State-of-Charge
failure Overcharge failures

Over-discharge failures

Current vector trajectory

Open switch
failure 3-phase current mean value voltage
Power
components’

electronics
failures in
Electrical

failures Phase current

EVs

Device current
Short switch
failure Gate voltage
Broken rotor bar

Rotor-related Cracked rotor end rings

failures
Electrical Shorted rotor field windings
failure
Stator- The open or short circuit of stator windings
related
The abnormal connection of stator windings
failures
Electric motor Outer bearing race defect
failures
Bearing Inner bearing race defect
failures
Ball defect
Mechanical Train defect
failures
Bend shaft
Eccentricity-
related Static air-gap irregularities
failures
Dynamic air-gap irregularities

Figure 22.11 Categorization of primary electrical component failures in EVs.

1. Chemical Issues: During the battery manufacturing process, it is essential to account for the
electrochemical factors contributing to battery failures.
2. Thermal Issues: An additional crucial aspect in evaluating the reliability of the battery pack
pertains to maintaining safe conditions through thermal management. The thermal analysis for
battery pack heating is subdivided into two key components: battery cell thermal analysis and
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
442 22 Reliability Assessment of Sub-components of Electric Vehicle for Performance Enhancement Grid Integrated

comprehensive battery pack thermal analysis. Various cooling systems have been employed to
regulate the temperature of the cells.
3. Mechanical Issues: The primary mechanical-related failure that can influence the reliabil-
ity of EV batteries is the result of mechanical stresses experienced during different modes of
operation and collisions. These failures encompass stress-based and strain-based mechanisms.
Notably, advancements are being made in both testing and modeling the mechanical character-
istics of the battery pack, along with its individual components, such as cathodes, anodes, and
separators.
4. Electrical Issues: Several critical electrical parameters within the battery are pivotal in address-
ing reliability concerns.
(i) Over-Voltage: The condition where the battery charging voltage surpasses the established
standard, leading to issues like battery plating and overheating, is called over-voltage.
(ii) Under-Voltage: Another problem that results in battery pack failure is under-voltage dur-
ing the charging process.
5. Power Electronic Converter/Battery Charger: Central roles of the battery charger encom-
pass channeling energy from the electric grid to the battery, refining the charging sequence, and
halting the charge when the battery reaches full capacity. Any functional interruptions directly
influence the dependability of the battery charger.
One approach to mitigate the aforementioned issues is to restrict the State of Charge (SoC)
within a specific range. Operating the battery beyond this range can harm battery lifespan, reli-
ability, and safety.

22.4.7 EVs Battery Management Systems (BMS)

The BMS recharges the batteries of the EVs, also known as EV charging points or EV charging
systems. The critical functions of an automotive BMS encompass important aspects such as mon-
itoring the battery’s state of charge (SoC), gauging its state of health (SoH), overseeing voltage,
temperature, and current levels, and ensuring proper cell balancing for lithium-ion (Li-ion) bat-
teries. The BMS incorporates a range of essential indicators to guarantee the secure and efficient
functioning of EV batteries over their intended lifespan, including the following:

(i) SoC (State of Charge)

(ii) SoH (State of Health)
(iii) SoT (State of Temperature)

1. State of Charge (SoC): SoC indicates the amount of energy left in the battery as a percentage of
its total capacity. It is essential to accurately estimate the remaining driving range and prevent
over-discharge, which can damage the battery.
2. State of Health (SoH): SoH reflects the battery’s overall health and degradation level. It helps
assess the battery’s capacity reduction over time and ensures that it is replaced or serviced when
its capacity falls below a certain threshold.
3. State of Temperature (SoT): Temperature significantly influences and is highly responsive
to the dynamic attributes of a battery. Additionally, due to the requirement for high-energy
ESS, there is a consistent demand for battery packs with elevated energy density. Consequently,
high-energy-density battery packs inherently encounter challenges related to thermal manage-
ment due to the heat produced during charge and discharge operations.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
22.5 Evaluation of Distribution Systems Reliability with Integrated EVs 443

22.5 Evaluation of Distribution Systems Reliability

with Integrated EVs
The evaluation of distribution systems’ reliability with the integration of EVs is a critical aspect
in the field of power system planning and management. As the adoption of EVs continues to rise,
their interaction with the electrical grid introduces both opportunities and challenges, particularly
concerning the overall reliability of the DN. This evaluation aims to assess how the presence of EVs
influences the reliability of the distribution system and to develop strategies for ensuring a robust
and resilient power supply
Importance of Distribution System: Reliable electricity distribution is essential for the
smooth functioning of various societal and economic activities. Any disruptions in power supply
can lead to inconveniences, financial losses, and even safety hazards. Effective power system
planning, control, and maintenance hinge on the synchronized management of power flow within
the power grid and are fundamental aspects of power system planning and modeling (Giri et al.
2019).
● The integration of EVs into the distribution system adds a dynamic load and introduces bidi-
rectional power flow due to charging and discharging, thereby necessitating a comprehensive
reliability assessment.
● Impact of EVs on Reliability: The presence of EVs impacts the distribution system in several
ways.
(i) Load Variability: EV charging introduces additional load variations, increasing stress on the
distribution infrastructure during peak charging times.
(ii) Bidirectional Power Flow: V2G and V2H functionalities allow EVs to inject power back
into the grid. While this can provide grid support, it also introduces complexities in managing
power flows.
(iii) Infrastructure Overload: Rapid adoption of EVs in specific regions can potentially overload
local distribution infrastructure if not properly managed.
Reliability Assessment Techniques
Reliability assessment involves analyzing the potential failure points, evaluating the impact of
these failures on the system, and devising strategies to enhance the system’s reliability. Key assess-
ment techniques include:
(i) Load Flow Analysis: Analyzing power flows to identify potential bottlenecks and overloads
(ii) Fault Analysis: Simulating fault conditions to determine the system’s response and the
impact on reliability
(iii) Monte Carlo Simulation: Incorporating uncertainty in load and EV charging patterns to
assess the system’s performance under varying conditions.
(iv) Sensitivity Analysis: Evaluating how changes in EV penetration levels and charging behav-
iors affect reliability metrics.
Strategies for Enhancing Reliability: To ensure the reliability of distribution systems inte-
grated with EVs, various strategies can be employed:
● Advanced Metering Infrastructure (AMI): Smart metering allows real-time monitoring of
EV charging and grid conditions, enabling better load management.
● Demand Response: Implementing demand response programs to shift EV charging away from
peak load times, reducing stress on the grid.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
444 22 Reliability Assessment of Sub-components of Electric Vehicle for Performance Enhancement Grid Integrated

● Energy Storage: Integrating ESS with EV charging infrastructure can help balance supply and
demand and mitigate load fluctuations.
● Grid Upgrades: Proactive grid upgrades, including transformer replacements and infrastructure
reinforcement, can accommodate the increased load from EVs.

Future Directions: As EV technology advances and their adoption becomes more widespread,
continued research is necessary to understand the evolving dynamics between EVs and distribution
system reliability. This includes exploring optimized charging strategies, incorporating RES, and
developing advanced control algorithms to manage EV-grid interactions effectively.

22.5.1 Distribution Network (DN) Reliability Assessment

The reliability of a DN is a measure of its capability to provide an uninterrupted power supply. A
DN comprises interconnected elements such as busbars, distributors, service mains, cables, isola-
tors, circuit breakers, protecting devices, and transformers (Manoj et al. 2015). From a reliability
perspective, these components are arranged in a series configuration, meaning that each element
must remain operational to ensure continuous power supply. The expected failure rates influence
the overall reliability of the DN 𝜆 and interruption times 𝜏 of these individual components. Addi-
tionally, a range of customer interruption metrics has been developed in the literature to assess the
impact and severity of system failures and to conduct reliability forecasting assessments. This study
utilizes several metrics, including ASAI, SAIFI, SAIDI, CAIDI, and ENS. The following analytical
probabilistic models, chronological probabilistic models and their scientific interpretation, math-
ematical description, and importance are illustrated in Chapters 20 and 21. The above description
of these indices is also discussed in Billinton et al. (1996).

22.5.2 Reliability Evaluation Using a V2G Approach and High EV Penetration

Reliability assessments can indeed be conducted using analytical or chronological probabilistic
models, depending on the specific context and the nature of the system or process being analyzed.
These models help evaluate the likelihood of a system or component functioning correctly over a
given period or under certain conditions.

1. Analytical Probabilistic Models

● Probability Density Functions (PDFs): Analytical models often use probability density

functions to describe the distribution of failure or reliability over time. For example, the expo-
nential distribution might be used to model the failure rate of a component.
● Weibull Distribution: The Weibull distribution is frequently used in reliability engineering

to model the failure rates of components, with its shape parameter allowing for modeling
various failure patterns.
● Monte Carlo Simulation: Analytical models can also incorporate MCSs to account for com-

plex, interconnected systems or situations where exact mathematical models are not available.
This involves generating random inputs and analyzing the output distribution to assess relia-
bility.
2. Chronological Probabilistic Models
● Markov Models: These models are used when analyzing systems with discrete states and

transitions. Markov models consider the probability of transitioning from one state to another
over time, allowing for the assessment of system reliability and availability.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
22.5 Evaluation of Distribution Systems Reliability with Integrated EVs 445

● Fault Tree Analysis (FTA): FTA is a top-down deductive approach used to identify and eval-
uate the probability of failure events in a system. It uses a graphical tree structure to represent
how different events or failures can lead to system failure.
● Event Tree Analysis (ETA): ETA is complementary to FTA and is used to assess the con-
sequences of various events or failures, considering different possible outcomes and their
probabilities.
● Reliability Block Diagrams (RBD): RBDs are graphical representations that break down
a system into individual components and their interconnections, allowing for the analysis of
system reliability by combining component reliability data.

To address the requirement for chronological modeling of EV consumption, the study will
employ a method known as “chronology.” This method involves generating virtual scenarios for
both electricity generation and demand, which will then be utilized to compute reliability indices.
Traditional generating units are characterized by two conditions: By applying the exponen-
tial CDF to create a virtual scenario for the time-to-failure (TTF), time-to-repair (TTR), the
mean-time-to-failure (MTTF), mean-time-to-repair (MTTR) are stated, respectively, in years.
Uf is consistently between 0 and 1, corresponding to interruption f , dispersed random number
(Negarestani et al. 2016). There may be problems with the generation units that cause load
interruptions. The annual system and interruption indices can be used to separate the various
reliability indices (Talukdar and Deka 2021). There are many available annual system indices:

(i) Loss of load expectation (LOLE) in h/yr

(ii) Loss of energy expectation (LOEE) in megawatt hours/yr,
(iii) Loss of load frequency (LOLF) in interruptions/yr (Sandelic et al. 2022).

Numerous Interruption indices are available

(i) Energy Not Served per Interruption (ENSPI) in megawatt hours/interruption

(ii) System Average Interruption Frequency Index (SAIFI) in interruption/customer (int./cu.)
(iii) System Average Interruption Duration Index (SAIDI) in h/cu.
Various reliability indices for NI interruptions over N years can be determined in Eqs.
(22.3)–(22.8):

∑
NI
hi
LOLE = (22.3)
i=1
N
∑
NI
ENSi
LOEE = (22.4)
i=1
N
NI
LOLF = (22.5)
N
∑
NI
ENSi LOEE
ENSPI = = (22.6)
i=1
N LOLF
The number of customers interruptions
SAIFI = (22.7)
Total customers served
Total durations of customers interruptions
SAIDI = (22.8)
Total customers served
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
446 22 Reliability Assessment of Sub-components of Electric Vehicle for Performance Enhancement Grid Integrated

22.5.3 Reliability Parameters

EVs possess the dual capability of serving as an energy provider and an electricity user, attributed
to their engagement in V2G operations (Xu and Chung 2015). The extensive interplay between
their V2G and G2V functionalities and their time-space characteristics necessitates recognizing
this unique aspect and formulating fresh criteria for EV reliability. In alignment with this goal, the
reliability metrics for EVs predominantly bifurcate into two categories, illustrated in Figure 22.12.
A more detailed categorization and explanation can be derived from recent research findings (Hou
et al. 2016).

22.5.3.1 Reliability Parameters of EVs

1. Average Interruption Frequency of EV Charging Index (EAIFI , t/yr):

∑N
𝜆i
EAIFI = i=1 (22.9)
N •Y
The symbol 𝜆i denotes the count of charge interruptions over the period of EV simulation. Here,
N signifies the total number of EVs, while Y indicates the simulation duration in years.
Reliability parameters for electric vehicle (V2G and G2V)

Average Average extra System average

interruption time for interruption
frequency of EV charging frequency index
charging index (EAETC, min/ (SAIFI, time/
(EAIFI, times/ yr) (household·year))
yr)

Average
Grid system
frequency of EV Electric reliability
participating in vehicle parameters
V2G (EAFPV, reliability
time/yr) parameters
System average
Average extra interruption
length for Average duration index
charging (EAELC, interruption (SAIDI,
km/yr) duration of EV min/
charging index (household·year))
(EAIDI, h/yr)

Figure 22.12 Reliability parameters for EVs.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
22.5 Evaluation of Distribution Systems Reliability with Integrated EVs 447

2. Average Interruption Duration of EV Charging Index (EAIDI , h/yr):

∑N
ti
EAIDI = i=1 (22.10)
N •Y
In this context, ti stands for the cumulative charging disruption time during the simulation
period for the ith EV.
3. Average Extra Length for Charging (EAELC , km/yr):
∑N
Δli
EAELC = i=1 (22.11)
N •Y
In this scenario, li denotes the distance covered by the ith EV due to a power outage when SOCi <
Si , as it searches for an alternative charging station throughout the simulation duration.
4. Average Extra Time for Charging (EAETC , min/yr):

∑
N
tiE
EAETC = (22.12)
i=1
N •Y

Here, (tiE ) represents the time taken by the ith EV due to a power failure when SOCi < Si as it
seeks a new charging station over the course of the simulation interval.
5. Average Frequency of EV Participating in V2G (EAFPV , t/yr):
∑N
ci
EAFPV = i=1 (22.13)
N •Y
Here, ci indicates the cumulative count of instances the ith EV engaged in V2G activities during
the simulation duration.

22.5.3.2 Grid System Reliability Parameters

Although EVs are transportable and interrupt load for the distribution system, the impact of power
loss on them is relatively minimal. In order to remove an electric car, the existing DN described
below is referred to by system reliability indices.

1. System Average Interruption Frequency (SAIFI , t/(household•yr)):

∑M
fi
SAIFI = i=1 (22.14)
MY
•

In this context, M refers to the total number of households, and f ii denotes the count of power
interruptions experienced by the ith user throughout the simulation period.
2. System Average Interruption Duration Index (SAIDI , min/(household•yr)):
∑N s
i=1 ti
SAIDI = (22.15)
M •Y
Here, tsi signifies the total duration of the power outage experienced by the user.
3. System Expected Energy Not Supplied (SEENS , MWh/(household•yr)):
Y
∫i=0 Pc (t)dt
SEENS = (22.16)
Y
Here, Pc (t) indicates the system’s load decrease at time t and is set to zero when the network
operates normally.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
448 22 Reliability Assessment of Sub-components of Electric Vehicle for Performance Enhancement Grid Integrated

4. Average Extra Length for Charging (EAELC , km/yr):

∑N
Δli
EAELC = i=1 (22.17)
NY•

The methodology and Reliability parameters outlined above necessitate a profound understand-
ing of the mechanical, electrothermal, and lifespan characteristics of the vehicle’s drive-train and
its sub-components (Blaabjerg et al. 2021).

22.6 Conclusion
In conclusion, evaluating the reliability of distribution systems in the context of integrated EVs
is a multifaceted task that requires a holistic approach. It involves considering the impact of EV
charging behaviors, bidirectional power flow, and load variations on the DN. In the context of the
advancements in EVs, it is crucial to emphasize the critical importance of evaluating the reliability
and safety aspects concerning the longevity of EVs’ electrical components, especially when it comes
to planning, servicing, and maintenance. Among the various electrical components in EVs, the
battery pack, power electronic components, and the electric motor stand out as the most pivotal
elements requiring a thorough assessment from both reliability and safety perspectives. By utilizing
advanced assessment techniques and implementing strategies to manage the challenges posed by
EV integration, distribution system operators can ensure a reliable and efficient power supply while
accommodating the growing demand for electric mobility.

References

Billinton, R., Chen, H., and Ghajar, R. (1996). Time-series models for reliability evaluation of power
systems including wind energy. Microelectronics Reliability 36 (9): 1253–1261.
Blaabjerg, F., Wang, H., Vernica, I. et al. (2021). Reliability of power electronic systems for EV/HEV
applications. Proceedings of the IEEE 109 (6): 1060–1076. https://doi.org/10.1109/JPROC.2020
.3031041.
Das, H.S., Rahman, M.M., Li, S., and Tan, C.W. (2020). Electric vehicles standards, charging
infrastructure, and impact on grid integration: a technological review. Renewable and Sustainable
Energy Reviews 120: 109618.
Dwivedi, D., Singh, S., and Kalaiselvi, J. (2022). Impact of PWM and duty ratio control on voltage for
SiC fed three phase BLDC motor drive. 2022 IEEE International Conference on Power Electronics,
Smart Grid, and Renewable Energy (PESGRE), 1–5. https://doi.org/10.1109/PESGRE52268.2022
.9715861.
Gandoman, F.H., Ahmadi, A., Van den Bossche, P. et al. (2019). Status and future perspectives of
reliability assessment for electric vehicles. Reliability Engineering & System Safety 183: 1–16.
Ghosh, R., Kim, D.S., and Trivedi, K.S. (2013). System resiliency quantification using non-state-space
and state-space analytic models. Reliability Engineering & System Safety 116: 109–125.
Giri, A.K., Singh, S., Dash, R. et al. (2019). A review on application of ANN and machine learning
algorithm for the optimal placement of STATCOM. 2019 Innovations in Power and Advanced
Computing Technologies (i-PACT), volume 1, 1–5. https://doi.org/10.1109/i-PACT44901.2019
.8960138.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 449

Hou, K., Xu, X., Jia, H. et al. (2016). A reliability assessment approach for integrated transportation and
electrical power systems incorporating electric vehicles. IEEE Transactions on Smart Grid 9 (1):
88–100.
Jayasawal, K., Rana, A.K., and Teja, A.V.R. (2022). Computationally efficient model predictive torque
control of switched reluctance motor drives. IECON 2022 - 48th Annual Conference of the IEEE
Industrial Electronics Society, 1–6. https://doi.org/10.1109/IECON49645.2022.9968991.
Kamruzzaman, M.D. and Benidris, M. (2019). Reliability-based metrics to quantify the maximum
permissible load demand of electric vehicles. IEEE Transactions on Industry Applications 55 (4):
3365–3375.
Kothari, D.P., Nagrath, I.J., and Saket, R.K. (2022). Modern Power System Analysis. Tata McGraw-Hill
Publishing Company.
Krishnan, R. (2017). Permanent Magnet Synchronous and Brushless DC Motor Drives. CRC Press.
Krishnan, R. and Materu, P.N. (1990). Design of a single-switch-per-phase converter for switched
reluctance motor drives. IEEE Transactions on Industrial Electronics 37 (6): 469–476. https://doi.org/
10.1109/41.103450.
Manoj, K.N., Sethi, V.K., and Singh, S. (2015). Impact of distributed generation in electrical power
system. International Journal of Electrical and Electronics Research 3 (3): 146–152.
Negarestani, S., Fotuhi-Firuzabad, M., Rastegar, M., and Rajabi-Ghahnavieh, A. (2016). Optimal sizing
of storage system in a fast charging station for plug-in hybrid electric vehicles. IEEE Transactions on
Transportation Electrification 2 (4): 443–453.
Rivera, S., Kouro, S., Vazquez, S. et al. (2021). Electric vehicle charging infrastructure: from grid to
battery. IEEE Industrial Electronics Magazine 15 (2): 37–51. https://doi.org/10.1109/MIE.2020
.3039039.
Sandelic, M., Peyghami, S., Sangwongwanich, A., and Blaabjerg, F. (2022). Reliability aspects in
microgrid design and planning: status and power electronics-induced challenges. Renewable and
Sustainable Energy Reviews 159: 112127.
Sarita, K., Saket, R.K., and Khan, B. (2023). Reliability, availability, and condition monitoring of
inverters of grid-connected solar photovoltaic systems. IET Renewable Power Generation 17 (7):
1635–1653.
Singh, S., Saket, R.K., and Khan, B. (2023a). A comprehensive state-of-the-art review on reliability
assessment and charging methodologies of grid-integrated electric vehicles. IET Electrical Systems in
Transportation 13 (1): e12073.
Singh, S., Soni, S.K., Singh, K.A., and Saket, R.K. (2023b). Sliding mode control for bidirectional
DC-DC power converter in electric vehicle charger for G2V and V2G applications. 2023 IEEE 3rd
International Conference on Sustainable Energy and Future Electric Transportation (SEFET), 1–6.
https://doi.org/10.1109/SeFeT57834.2023.10245726.
Soni, S.K., Singh, S., Singh, K.A. et al. (2023). Event-triggered control for LPV modeling of DC-DC
boost converter. IEEE Transactions on Circuits and Systems II: Express Briefs 70 (6): 2062–2066.
https://doi.org/10.1109/TCSII.2022.3230418.
Talukdar, B.K. and Deka, B.C. (2021). An approach to reliability, availability and maintainability
analysis of a plug-in electric vehicle. World Electric Vehicle Journal 12 (1): 34.
Xu, N.Z. and Chung, C.Y. (2015). Reliability evaluation of distribution systems including
vehicle-to-home and vehicle-to-grid. IEEE Transactions on Power Systems 31 (1): 759–768.
Yilmaz, M. and Krein, P.T. (2012). Review of battery charger topologies, charging power levels, and
infrastructure for plug-in electric and hybrid vehicles. IEEE Transactions on Power Electronics 28 (5):
2151–2169.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
451

23

Reliability Assessment of Multilevel Inverter for Modern Power

System Applications
Saumya Singh 1 , Dhawal Dwivedi 1 , Kumari Sarita 2 , R. K. Saket 1 , and P. Sanjeevikumar 3
1
Electrical Engineering Department, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
2
Department of Electrical Engineering, Government Engineering College, Aurangabad, Aurangabad, Bihar, India
3
Department of Electrical Engineering, IT and Cybernetics, University of South-Eastern Norway, Porsgrunn, Norway

23.1 Introduction

Around the world, fossil fuels are the primary source of electricity. But using fossil fuels in power
plants creates greenhouse gases like carbon dioxide that cause climate change. Many nations use
renewable energy as an alternative energy source due to the problems associated with burning fos-
sil fuels. Several renewable energy sources (RES) have been created to generate power, including
wind, tidal, wave, and photovoltaic (PV). Shortly, it is anticipated that these RES, particularly pho-
tovoltaic, will overtake all other forms of renewable energy (Surana and Jordaan 2019; Kothari et al.
2022). The three most important performance parameters for grid-connected solar systems are con-
version efficiency, maximum power point tracking (MPPT), and total harmonic distortion (THD).
The inverter’s control method and topology selection significantly impact these performance met-
rics. MLI produces a terminal voltage using multiple DC voltage sources at different levels, typically
in steps, allowing for reduced THD and better output waveform quality than traditional two-level
inverters. The common configurations include:
● Capacitor-Clamped Flying Configuration
● Neutral-Point Clamped Diode Configuration
● H-Bridge Cascade Configuration
Consequently, MLIs primarily serve as a bridge connecting different electrical equipment
inverters (DC/AC) and (DC/DC) converters. This is necessary because the loads or system cannot
directly generate electricity from RES. The grid frequency has to match the frequency of the
injected signal and can be fed by a voltage or current with less harmonic content, or a pure
sinusoidal wave can power it (Li et al. 2014). The MLI with a reduced LC filter size produces
sinusoidal three-phase voltage and current with lower harmonics. Voltage sources for DC input
to the MLIs can be created using RES, such as PV arrays with DC/DC converters, induction
generator/synchronized generator-fed wind turbines using rectifier configurations, and fuel cells
as converters (DC/DC) (Mahalakshmi and Thampatty 2015). MLIs deliver enhanced power quality
and more significant levels in their terminal voltage waveforms than two-level inverters.

Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
section of RAMPS illustrates the reliability analysis methodologies of modern distribution systems
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
452 23 Reliability Assessment of Multilevel Inverter for Modern Power System Applications

Vdc Vdc Output voltage

Line to line Voltage

Line to line Voltage

Output voltage
Fundamental wave Fundamental wave
Vdc/2

0
–Vdc/2

–Vdc –Vdc

(a) (b)

Figure 23.1 Output line voltage, and fundamental waveform (a) two-level, and (b) three-level inverters.

MLIs offer numerous benefits:

● Enhanced output waveforms with minimized THD.

● Lower stress on power electronic devices due to reduced voltage steps.
● Modularity and scalability.
● Improved efficiency in certain configurations.

The output voltage waveforms of MLIs have higher levels than those of two-level inverters, as
illustrated in Figure 23.1. This means that the power quality is better. Commencing with three-level
converters and progressing through various MLI topologies for several years, the first MLI appli-
cation existed in 1975. MLI switches are capable of blocking lower voltages due to the distribu-
tion of DC-link stress voltage across various switching units. Consequently, creating power semi-
conductor devices with reduced voltage specifications enables their utilization in a broader spec-
trum of products, thereby enhancing their accessibility and adaptability (Wang et al. 2023). The
increased switching voltages will also increase power losses on high-voltage switches. Higher effi-
ciency, lower dV/dt, and less electromagnetic interference (EMI) are further benefits of MLIs.
Higher dv/dt values result in issues with insulation and EMI. The conventional MLI, includes
neutral-point clamped (NPC) (Nabae et al. 1981), flying capacitor (FC) (Kim et al. 2019), cascaded
H-bridge (CHB) (Malinowski et al. 2010). The two other most popular MLIs are modular multilevel
converter (MMC) and T-type. Due to its modular design, lack of energy-storing components, and
capacity to isolate faults, the CHB is the most prevalent subclass (Franquelo et al. 2008; Villanueva
et al. 2009). In the case of conventional MLIs, more components are needed to achieve higher out-
put voltage levels. External controllers and switches, clamped diodes, are needed for balancing the
voltage across the capacitors in the case of NPC, the no. of levels in terminal voltage increases. The
presence of clamped capacitors in the FC design imposes limitations and necessitates the use of
controllers to maintain the capacitor voltage balance. However, as the number of levels increases
in the output voltage for CHB, need for switches and DC sources increases. Because CHB lacks a
voltage balancing circuit, it is better suited for substantial-power applications (Malinowski et al.
2010). The inclusion of sub-module capacitor voltage balancing in MMC introduces complexity
to its control, yet its modular design renders it well-suited for a range of high-power applications,
particularly in high-voltage direct current (HVDC) transmission (Debnath et al. 2015).
The researchers have developed a switched-capacitor-based MLI, which can generate boosted
multilevel output voltage because the traditional MLI topologies cannot increase voltage, which
can raise the terminal voltage to appropriate grid voltage by utilizing switched-capacitor voltages
found in the circuits with minimize components. In the matter of conventional MLI, extra
high-gain DC/DC converter DC/DC converter is required to raise the voltage of the dispersed
power sources to required level for grid integration purposes. As the no. of voltage levels rises, the
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
23.2 Reliability Assessment Techniques 453

circuit’s count of the switched capacitors also increases, contributing to an elevation in the output
voltage. Suitable switching must be given to make self-voltage balance less voltage ripple since
this switched capacitor draws an enormous magnitude of charging current as the ripple voltage
across the capacitor voltage increases (Tayyab et al. 2022). Various techniques, such as hysteresis
current control and CHB inverters, have been employed to optimize the harmonic characteristics
of voltage and current waveforms for effective grid integration (Yao and Xiao 2012; Rashid 2011).
In addition to its fault-tolerant and versatile traits, it produced a higher number of voltage levels
and a well-matched sinusoidal voltage waveform.
Additionally, has examined new suggested MLIs with fewer components in light of comparative
variables, component selection, and the acceptability and availability of the components (Agrawal
and Jain 2017). Finally, research studies have suggested the design of a modular CHB MLI in sin-
gle or three-phase grid-connected applications to increase the flexibility and efficiency of power
systems and maximize power extraction and efficiency (Xiao and Tolbert 2014; Bana et al. 2019).
The number of power semiconductor devices employed in DC/AC converters increased to obtain
high-quality voltage waveforms, making them more prone to failure. Additionally, MLIs provide
economical ways for electric aircraft to improve reliability, as some MLI topologies may function in
faulty conditions without additional hardware components. According to the research presented
above, it is evident that additional work is needed to build a fault-tolerant MLI topology with
the express purpose of implementing fault-tolerance characteristics by adopting the lowest device
count possible. As a result, reliability is one of the main obstacles to the widespread use of MLIs in
industrial applications.
The factors influencing the reliability:
● Component Failures: Every electronic component within the inverter has its failure rate. Fac-
tors like operating temperature, switching frequency, and stress can accelerate these failures
(Sarita et al. 2023).
● Control Complexity: Higher-level multilevel inverters (MLI) require more complex control
strategies, potentially increasing the likelihood of control-related issues.
● Thermal Management: Thermal stress can accelerate component degradation.Exploring the
application of artificial neural networks for optimizing battery temperature control to improve
system efficiency and minimize energy consumption. Effective cooling systems and designs that
distribute thermal load evenly can significantly influence reliability (Saket 2011; Singh et al.
2023).
The following Sections 23.2–23.4 in this chapter are structured as follows: Section 23.2 empha-
sizes the use of reliability assessment techniques (RAT) for the mathematical evaluation of relia-
bility within the framework of MLIs, Section 23.3 presents the different types of MLIs popular in
grid-connected inverters, and their configurations are discussed in depth, Section 23.4 covers the
comparative reliability assessment of MLIs, and Section 23.5 concludes the chapter.

23.2 Reliability Assessment Techniques

Reliability engineering is an intricate discipline that employs a variety of mathematical approaches
to evaluate the probability of a system performing without failure over a specified time. In this
section, some basic mathematical formulas and metrics related to reliability are discussed briefly
that can be applied to the MLIs. Previous chapters provide a detailed description of mathematical
formulas, related computations, and the applicability of these techniques.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
454 23 Reliability Assessment of Multilevel Inverter for Modern Power System Applications

Reliability R(t):
R(t) = e−𝜆t (23.1)
where 𝜆 = failure rate. Failure rate 𝜆: The rate at which failures occur in a system. If f (t) represents
probability density function (PDF) of failures, then:
f (t)
𝜆(t) = (23.2)
R(t)
● Series Systems: For components connected in series (all components must work for the system
to work):
Rsystem (t) = R1 (t) × R2 (t) × · · · × Rn (t) (23.3)
where Ri (t) = Reliability of the ith component.
● Parallel Systems (Redundancy): For components connected in parallel (only one component
must work for the system to work):
Rsystem (t) = 1 − [(1 − R1 (t)) × (1 − R2 (t)) × · · · × (1 − Rn (t))] (23.4)
● Mean time to failure (MTTF) and mean time between failures (MTBF):
– For nonrepairable systems:
∞
MTTF = R(t)dt (23.5)
∫0
– For repairable systems:
MTBF = 1∕𝜆 (23.6)

● Availability: For repairable systems, Availability A is the proportion of time a system is

operational:
MTBF
A= (23.7)
MTBF + MTTR
where MTTR represents as mean time to repair. The above following parameters have been dis-
cussed in detail in previous chapters.
● Reliability Application to MLIs:
– Diode-Clamped: Given the nature of diode-clamped inverters, components are typically in
series. Thus, the reliability of the inverter would be a product of the individual reliabilities.
– Flying Capacitor: Capacitors can be thought of as parallel components (redundant paths).
The reliability would therefore be evaluated using the formula for parallel systems.
– Cascaded H-Bridge: Modules can act in parallel for redundancy purposes, but components
are typically in series within each module.
– T-type: Components within a T-type inverter are often in series.
● Failure Mode and Effects Analysis (FMEA): A systematic approach that evaluates potential
failure modes of each component and estimates the impact on the overall system (Kara-Zaitri
et al. 1991).
FMEA is a step-by-step methodology used to identify potential failure modes at the earliest stages
and evaluate their potential impact. While FMEA is primarily a qualitative technique, it uses
quantitative metrics such as Risk Priority Number (RPN), which is calculated as follows:
RPN = Severity × Occurrence × Detection (23.8)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
23.2 Reliability Assessment Techniques 455

where Severity (S) rates the seriousness of the failure mode effect on a scale of 1 (not severe) to 10
(very severe), Occurrence (O) estimates the likelihood of the failure mode occurring, rated from 1
(unlikely) to 10 (very likely), and Detection (D) assesses the probability of the failure mode being
detected before it affects the customer, rated from 1 (highly likely) to 10 (very unlikely).
● Markov Chain Analysis: Useful in modeling the reliability of systems where different states
(like healthy, degraded, or failed) are present and transition probabilities can be determined (Hou
et al. 2015).
Markov chain analysis models systems as they transition between states based on certain proba-
bilities. In the context of MLI, the states might represent different operational conditions such as
healthy, degraded, and failed. Let’s assume there are n states for the inverter’s health. The tran-
sition probability matrix (P) represents the likelihood of transitioning from one state to another.
For reliability assessment, if one state represents “failed,” the probability of being in this state
after a certain number of transitions (time steps) can be determined. This can be calculated by
raising the matrix (P) to the power of the number of transitions and examining the appropriate
matrix entry. The reliability is assessed by evaluating the long-term state probabilities, which are
the stable probabilities of being in each state as time tends to infinity.
For inverters, factors affecting the transition probabilities might include:
– Component degradation,
– Operational stresses (e.g. temperature and load variations),
– Control strategies implemented.
● Life Cycle Cost Analysis (LCCA): By considering the total cost of ownership, including repair
and maintenance, this analysis provides insight into the long-term reliability and economic fea-
sibility of different inverters (Dhillon 1989).
LCCA evaluates the total cost of ownership of a product, considering not just the initial cost but
all costs over its lifecycle. The total Life Cycle Cost (LCC) is given by:
LCC = CC + OC + MC + FC − RV (23.9)
where
– CC (Capital Costs): Initial costs, including purchase, transportation, and installation.
– OC (Operating Costs): Costs associated with energy, water, and other utilities required for the
system’s operation.
– MC (Maintenance Costs): Predicted maintenance, repairs, and replacement costs.
– Failure Costs: Costs related to system downtime, reactive maintenance, and potential penal-
ties for failing to meet energy delivery commitments.
– RV (Residual Value): Salvage value or end-of-life disposal costs.
For reliability, the Failure Costs term is crucial. If a multilevel inverter has a higher probability of
failure due to its design or operational conditions, the Failure Costs will be higher, thus increasing
the LCC. By comparing the LCCs of different inverter designs or operational strategies, one can
determine which option is most cost-effective over the long run, taking reliability into account.
Integrating Markov Chain Analysis with LCCA makes it possible to estimate the Failure Costs
more accurately. For instance, knowing the probability of failure from the Markov model can esti-
mate the expected downtime and associated costs for the inverter over its lifecycle. This holistic
approach comprehensively views the reliability and economic implications of different multilevel
inverter designs and operational strategies.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
456 23 Reliability Assessment of Multilevel Inverter for Modern Power System Applications

23.3 Types of Multilevel Inverters (MLIs)

The ability of MLIs to meet specific requirements for power rating and electric power quality has
led to an increase in their use in systems with low levels of electronic radiation and harmonic
distortion. The commercial viability of MLIs for high-power quality applications and dynamic per-
formance needs is now being studied. For these applications, the power ranges from 1 to 30 MW.
MLIs are excellent for medium to high-voltage applications because they provide higher voltages
and may produce output voltage waveforms with lower THD even with constrained device ratings
(Rodriguez et al. 2007). The flying capacitor (FC), NPC, CHB, T-type, and MMC topologies are now
the most popular and commonly used topologies in the commercial sector. As shown in Figure 23.3,
there are two ways that MLIs can be connected to the grid.
For MLI, there are various types of modulation strategies. The different modulation methods
are space vector modulation (SVM), selective harmonic elimination (SHE), third harmonic injec-
tion, and sinusoidal pulse-width modulation (SPWM) (Kouro et al. 2010; Ahmed and Borghate
2014) (Figure 23.2). The most well-known of these traditional techniques is SPWM, which is fur-
ther divided into level-shifted pulse-width modulation (LSPWM) and phase-shifted pulse-width
modulation (PSPWM) (Carrara et al. 1992). In this, the LSPWM technique is used for MLIs.

23.3.1 Cascaded H-Bridge Multilevel Inverters

Baker and Bannister introduced the CHB inverter in 1975. A generalized CHB MLI is demon-
strated using a single-phase full-bridge inverter that is series-linked and has a separate DC supply. A
series connection aims to improve the output voltage levels while lowering harmonic distortions.
Figure 23.3 shows the five-level CHB MLI architecture, which consists of two full-bridge invert-
ers with a distinct DC supply whose values can be the same or different. Each H-bridge output
voltage is added together to create the output voltage waveform. An MLI arrangement is consid-
ered to be symmetrical if the magnitude of the input side DC supply is same for every H-bridge,
otherwise, it is said to be asymmetrical. The generation of output voltage for a five-level CHB is
shown in Figure 23.4. Table 23.1 displays switching states to achieve appropriate voltage values for
five-level CHB.
There are different forms of modulation techniques for MLI. Generally, two signals are used in
the pulse width modulation technique: the reference (modulating)signal and the carrier signal,
but in MLI, multiple carriers are used. For five-level CHB (5-1), i.e. four carriers are used with
one modulating signal. To generate a PWM, the modulating signal is compared with carriers. If
the triangle carrier waveform has a peak amplitude of Ac with a frequency of fc and the sinusoidal
reference waveform has a peak amplitude of Am with a frequency of fm , respectively. Then, the

+ +

Grid Grid

– –
PV Single- PV Two-stage conversion
stage MLI
(a) (b)

Figure 23.2 Types of inverter in accordance with stage of conversion (a) one-stage or direct conversion
and (b) dual-stage conversion.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
23.3 Types of Multilevel Inverters (MLIs) 457

+Ve
S11 S12

Vdc/2 Cd A V1
S13 S14

–Ve
VAB
+Ve
S21 S22

Vdc/2 Cd V2
B
S23 S24

–Ve

Figure 23.3 Single phase five-level CHB inverter.

100
V1 (V)

–100
100
V2 (V)

0
–100
200
VAB (V)

–200
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Times (s)

Figure 23.4 The generation of output voltage for five-level CHB.

Table 23.1 Switching states for five-level CHB.

Switching signals Output voltage

States S11 S12 S13 S14 S21 S22 S23 S24 VAB

1 OFF ON ON OFF OFF OFF OFF OFF −Vdc /2

2 OFF ON ON OFF OFF ON ON OFF −Vdc
3 OFF OFF OFF OFF OFF OFF OFF OFF 0
4 ON OFF OFF ON OFF OFF OFF OFF +Vdc /2
5 ON OFF OFF ON ON OFF OFF ON +Vdc
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
458 23 Reliability Assessment of Multilevel Inverter for Modern Power System Applications

amplitude modulation index and frequency modulation ratio are given as:
Am
Ma = (23.10)
Ac
mm
mf = (23.11)
mc

23.3.2 Neutral-Point Clamped Three-Level Inverter (NPC)

A three-level converter, known as an NPC or diode-clamped, comprises two capacitor voltages con-
nected in series with the middle tap as the neutral. Baker released another diode-clamped inverter
in 1980. This inverter splits the DC bus voltage in a set of voltage levels using a series-connected
semiconductor device. Then, the most effective topology in MLI is the NPC introduced by Nabae
et al. (1981). The excellent capabilities in medium-voltage applications and the relatively high effi-
ciency of NPC inverters have led to widespread adoption. Furthermore, the neutral point gives a
zero-voltage level, resulting in three separate voltage levels. NPC inverters are typically used in
electrical drives, uninterruptible power supply (UPS), and medium-voltage grid-connected con-
verters because they are efficient in the fundamental frequency switching range. Entire frequency
switching increases voltage and current THD, necessitating extra reactors to minimize the prob-
lem, which might raise the implementation cost. On the other hand, increasing the voltage levels
necessitates expanding the number of clamping diodes and capacitors, which may compromise
the overall structure’s durability. As opposed to that, this topology has the advantage of lowering
the harmonic content of the produced voltage by adding the zero-voltage level. Higher levels can
be achieved by increasing switching devices, clamping diodes, and capacitors. The output phase
voltage of an n-level diode-clamped inverter is n-level, while the output line voltage is (2n − 1).
Figure 23.5 shows the diagram of a three-level NPC. The output voltage of the NPC three-level
inverter varies between (+Vdc ∕2) and (0) or (−Vdc ∕2) and (0), while a conventional inverter varies
between (+Vdc ∕2) and (−Vdc ∕2). Table 23.2 lists the NPC converter’s switching states and the output
voltages. Figure 23.6 presents the outcomes for phase and line voltage, with Figure 23.7 showing
the capacitor balancing of NPC topology.

23.3.3 Flying Capacitors Three-Level Inverter

Meynard and Foch (1992) created a new topology, the FC; this design uses the floating capacitor to
clamp the voltage. The topologies of the diode-clamped multilevel inverter (DCMLI) or NPC and

Idc Figure 23.5 Three-level NPC inverter.

+Ve
S1

C1

S2
N
Vdc A
S3
C2

S4
–Ve
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
23.3 Types of Multilevel Inverters (MLIs) 459

Table 23.2 Switching states for three-level NPC.

Device switching status Terminal voltage

States S1 S2 S3 S4 VAN

1 OFF OFF ON ON −Vdc /2

2 ON OFF ON OFF 0
3 ON ON OFF OFF +Vdc /2

100
Voltage (V)

50
Phase

0
–50
–100

200
Voltage (V)
Line

–200
0 0.1 0.2 0.3 0.4 0.5 0.6
Times (s)

Figure 23.6 Phase and line voltages waveform of three-level NPC inverter.

100.4
Voltage (V)

100.2
C1

100
99.8
100.2
Voltage (V)

100
C2

99.8
99.6
0.4 0.42 0.44 0.46 0.48 0.5 0.52 0.54 0.56 0.58 0.6
Times (s)

Figure 23.7 Balancing of capacitor voltage for three-level NPC inverter.

the flying capacitor multilevel inverter (FCMLI) are very similar, with the only difference that the
DCMLI employs clamping diodes while the FCMLI uses floating capacitors. Figure 23.8 depicts
the inverter with three capacitor clamping levels. One DC source can power all the FCs because of
switching state redundancy inside the phase of the FC inverter. The capacitor clamping requires
many bulk capacitors to clamp the voltage, just like the diode-clamped inverter. An N-level con-
verter will need (N − 1) ∗ (N − 2)∕2 clamping capacitors per phase. Additionally, the N − 1 main
DC bus capacitors, provided that each capacitor used has the same voltage rating as that of the
power switch. The rating of the capacitors, which have significant amounts of the DC bus voltage
across them, is a downside of the topology. The switches S1 and S4 are complementary. Thus, if S1
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
460 23 Reliability Assessment of Multilevel Inverter for Modern Power System Applications

Idc Figure 23.8 The three-level FCMLI.

+Ve
S1

C1

S2
Vdc N
C3 A
S3
C2

S4
–Ve

Table 23.3 Switching states for three-level FCMLI.

Switching status Terminal voltages

States S1 S2 S3 S4 VAN C3

1 OFF OFF ON ON −Vdc /2 NC

2 OFF ON OFF ON 0 ↓
3 ON OFF ON OFF 0 ↑
4 ON ON OFF OFF +Vdc /2 NC

is ON, S4 is OFF, and vice-versa. Similarly, the switches S2 and S3 are complementary. The state of
the FC corresponding to the selected switching combination is also shown in Table 23.3. The NC
indicates the capacitor is not charged or discharged in this mode. ↑ and ↓ signifies the charging and
discharging of the capacitors. The switching states are specified for the positive half of the current
waveform, i.e. the load current is going out. The capacitor’s state (↑ and ↓) will get reversed for the
negative half cycle of the current. The equation of output phase voltage for FCMLI is given below:
VC
VAN = S1 (VC − VC3 ) + S2 VC3 − (23.12)
2
where, VC1 =VC2 =VC . Figure 23.9 presents the outcomes for phase and line voltage, with
Figures 23.10 and 23.11 showing the capacitor balancing of FCMLI topology.

23.3.4 Three-Level T-Type Inverter

Depending on the multilevel inverter topology’s structure, they can have issues with voltage balanc-
ing, isolated power supply requirements (for CHB inverters), large capacitor requirements (for FC
inverters), and massive capacitor requirements (for NPC inverters). In addition to these topologies,
a unique variety of NPC inverter called the T-type inverter is gaining popularity. T-type inverters,
which have advantages like simple operation and fewer switches, can replace complex three-level
inverter topologies since they do not require an isolated source, a large capacitor, or clamping
diodes. This architecture is also more effective at medium frequencies than the two- and three-level
NPC inverters (Schweizer and Kolar 2011, 2013). The clamping diodes are eliminated despite the
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
23.3 Types of Multilevel Inverters (MLIs) 461

200
Voltage (V)

100
Line

0
–100
–200

100
Voltage (V)
Phase

–100
3 3.05 3.1 3.15 3.2 3.25 3.3 3.35 3.4
Times (s)

Figure 23.9 Line and phase voltages of three-level FCMLI.

101
Voltage (V)
C1

100

99
Voltage (V)

100.5
100
C2

99.5
99
98.5
3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8
Times (s)

Figure 23.10 Balancing of DC-link capacitor voltage for three-level FCMLI.

102
Voltage (V)
C3

100

3.3 3.4 3.5 3.6 3.7 3.8 3.9 4

Times (s)

Figure 23.11 Balancing of flying capacitor voltage for three-level FCMLI.

inverter having the same active switches as an NPC inverter. Figure 23.12 shows the three-level
T-type converter (3LT 2 C) fundamental topology. A dynamic, bidirectional switch is added to the
two-level, traditional VSC architecture to extend it to the DC-link midpoint (Schweizer and Kolar
2013). A less isolated gate driver is required than the other three-level topologies and the neutral
point is also connected to the common emitter switches, which operate at low frequency. The bidi-
rectional switches used in T-type are of common emitter or collector configuration. The 3LT 2 C
can use the same modulation techniques as the three-level NPC converter. When the bidirectional
switch (combination of S4 and S3 ) is ON, the voltage from the load terminal to the neutral point is
zero, −Vdc /2 when S4 is ON, and Vdc /2 when S1 is ON. The switching states and output voltages for
the 3LT 2 C are shown in Table 23.4. Figure 23.13 presents the outcomes for phase and line voltage,
with Figure 23.14 showing the capacitor balancing of 3LT 2 C topology.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
462 23 Reliability Assessment of Multilevel Inverter for Modern Power System Applications

+Ve Figure 23.12 Three-level T-type topology with

common emitter configuration.
S1
C1

Vdc N A
S4 S3
S2
C2

–Ve

Table 23.4 Switching states for three-level T-type MLI.

Switching signals Output voltage

States S1 S2 S3 S4 VAN

1 OFF OFF ON ON −Vdc /2

2 OFF ON ON OFF 0
3 ON ON OFF OFF +Vdc /2

100
Voltage (V)

50
Phase

0
–50
–100

200
Voltage (V)
Line

–200
0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8
Times (s)

Figure 23.13 Phase and line voltages of three-level T-type converter.

102 C 1 C2
Voltage (V)

101
Capacitor

100
99
98
0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Times (s)

Figure 23.14 Balancing of DC-link capacitor voltage for three-level T-type converter.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
23.4 Comparative Reliability Assessment of MLIs 463

23.4 Comparative Reliability Assessment of MLIs

MLI have become increasingly popular in contemporary power systems because they can gen-
erate high-quality output voltages, Minimize harmonic interference, and function effectively at
elevated voltage levels. This section delves into the reliability of four prominent configurations:
CHB, diode-clamped or NPC, flying capacitor, and T-type.

● Diode-Clamped (Neutral-Point Clamped)

– Advantages:
* Simple structure and widely used in medium-voltage applications.
* Reduced voltage stress across each switch due to multiple clamping diodes.
– Reliability Concerns:
* Voltage balancing issues for the clamping diodes, especially at higher levels.
* A higher number of clamping diodes increases component count, potentially affecting reli-
ability.
● Flying Capacitor (Capacitor-Clamped)
– Advantages:
* Offers the possibility of redundant paths, which can be a boon for reliability.
* Natural voltage balancing under certain conditions.
– Reliability Concerns:
* Requires voltage balancing control, which adds complexity.
* Capacitors are prone to aging and require maintenance and periodic replacement.
● Cascaded H-Bridge (CHB)
– Advantages:
* Modular structure allows for increased system reliability; a failure in one module doesn’t
incapacitate the entire system.
* Uses conventional two-level switching devices, simplifying design and control.
– Reliability Concerns:
* Requires separate DC sources for each H-bridge, which might be a limitation in some appli-
cations.
* Increase in the number of semiconductor devices can raise concerns for maintenance and
potential points of failure.
● T-type (Three-level inverter)
– Advantages:
* Fewer components compared to other three-level topologies.
* Efficient utilization of the DC supply voltage, allowing for higher output voltages.
– Reliability Concerns:
* Neutral-point voltage balancing can be challenging, similar to the diode-clamped topology.
* Potential for shoot-through (short-circuit) if both switches in a leg turn on simultaneously.
Requires meticulous control.

A comparative summary of the MLIs based on different important parameters is given:

● Component Count: Generally, as the component count increases, potential failure points also
increase. CHB has the highest component count, while T-type tends to have the least among
three-level inverters.
● Voltage Balancing: Both diode-clamped and T-type configurations face challenges with voltage
balancing, while FC inverters require intricate voltage balancing control.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
464 23 Reliability Assessment of Multilevel Inverter for Modern Power System Applications

● Redundancy: FC inverters offer redundant switching paths, providing an inherent reliability

advantage. The modularity of CHB also offers redundancy benefits.
● Control Complexity: T-type inverters have simpler control strategies than Diode-Clamped or
FC configurations, making them less prone to control-induced failures.
● Maintenance: FC inverters might need more frequent maintenance due to capacitor aging,
while CHB designs, due to their high component count, can have more potential maintenance
points.

23.5 Conclusion
In conclusion of this book chapter, it becomes evident that the increasing need for efficient and
reliable power conversion in contemporary power systems magnifies the importance of MLI. A
comprehensive understanding of their reliability, influenced by design, control, and environmental
factors, ensures that these systems can effectively serve the evolving needs of the energy landscape.
No single inverter configuration is universally the most reliable. The choice depends on the specific
application, operational environment, and reliability challenges that need addressing. For instance,
a T-type or CHB inverter might be preferable in climates where maintenance is complex due to their
modularity and lower component count. Meanwhile, applications that require redundancy might
benefit more from the FC design. As always, careful design, control, and operational practices will
significantly influence the reliability of any chosen topology.

References

Agrawal, R. and Jain, S. (2017). Comparison of reduced part count multilevel inverters (RPC-MLIs) for
integration to the grid. International Journal of Electrical Power & Energy Systems 84: 214–224.
https://doi.org/10.1016/j.ijepes.2016.05.011.
Ahmed, I. and Borghate, V.B. (2014). Simplified space vector modulation technique for seven-level
cascaded H-bridge inverter. IET Power Electronics 7: 604–613. https://doi.org/10.1049/iet-pel.2013
.0135.
Bana, P.R., Panda, K.P., Naayagi, R.T. et al. (2019). Recently developed reduced switch multilevel
inverter for renewable energy integration and drives application: topologies, comprehensive analysis
and comparative evaluation. IEEE Access 7: 54888–54909. https://doi.org/10.1109/ACCESS.2019
.2913447.
Carrara, G., Gardella, S., Marchesoni, M. et al. (1992). A new multilevel PWM method: a theoretical
analysis. IEEE Transactions on Power Electronics 7 (3): 497–505. https://doi.org/10.1109/63.145137.
Debnath, S., Qin, J., Bahrani, B. et al. (2015). Operation, control, and applications of the modular
multilevel converter: a review. IEEE Transactions on Power Electronics 30 (1): 37–53. https://doi.org/
10.1109/TPEL.2014.2309937.
Dhillon, B. (1989). Life Cycle Costing: Techniques, Models and Applications. Routledge.
Franquelo, L.G., Rodriguez, J., Leon, J.I. et al. (2008). The age of multilevel converters arrives. IEEE
Industrial Electronics Magazine 2 (2): 28–39. https://doi.org/10.1109/MIE.2008.923519.
Hou, K., Jia, H., Xu, X. et al. (2015). A continuous time Markov chain based sequential analytical
approach for composite power system reliability assessment. IEEE Transactions on Power Systems 31
(1): 738–748.
Kara-Zaitri, C., Keller, A.Z., Barody, I., and Fleming, P.V. (1991). An improved FMEA methodology. In:
Annual Reliability and Maintainability Symposium. 1991 Proceedings, 248–252. IEEE.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 465

Kim, S.-H., Kim, R.-Y., and Kim, S.-I. (2019). Generalized model predictive control method for
single-phase N-level flying capacitor multilevel rectifiers for solid state transformer. IEEE
Transactions on Industry Applications 55 (6): 7505–7514. https://doi.org/10.1109/TIA.2019.2917869.
Kothari, D.P., Nagrath, I.J., and Saket, R.K. (2022). Modern Power System Analysis. Tata McGraw-Hill
Publishing Company.
Kouro, S., Malinowski, M., Gopakumar, K. et al. (2010). Recent advances and industrial applications of
multilevel converters. IEEE Transactions on Industrial Electronics 57 (8): 2553–2580. https://doi.org/
10.1109/TIE.2010.2049719.
Li, W., Ruan, X., Bao, C. et al. (2014). Grid synchronization systems of three-phase grid-connected
power converters: a complex-vector-filter perspective. IEEE Transactions on Industrial Electronics 61
(4): 1855–1870. https://doi.org/10.1109/TIE.2013.2262762.
Mahalakshmi, R. and Thampatty, K.C.S. (2015). Grid connected multilevel inverter for renewable
energy applications. Procedia Technology 21: 636–642. https://doi.org/10.1016/j.protcy.2015.10.076.
Malinowski, M., Gopakumar, K., Rodriguez, J., and Pérez, M.A. (2010). A survey on cascaded
multilevel inverters. IEEE Transactions on Industrial Electronics 57 (7): 2197–2206. https://doi.org/10
.1109/TIE.2009.2030767.
Meynard, T.A. and Foch, H. (1992). Multi-level conversion: high voltage choppers and voltage-source
inverters. In: PESC ’92 Record. 23rd Annual IEEE Power Electronics Specialists Conference, vol. 1,
397–403. IEEE https://doi.org/10.1109/PESC.1992.254717.
Nabae, A., Takahashi, I., and Akagi, H. (1981). A new neutral-point-clamped PWM inverter. IEEE
Transactions on Industry Applications IA-17 (5): 518–523. https://doi.org/10.1109/TIA.1981.4503992.
Rashid, M.H. (2011). Power Electronics: Circuits, Devices, and Application (for Anna University). Pearson
Education India.
Rodriguez, J.É., Bernet, S., Wu, B. et al. (2007). Multilevel voltage-source-converter topologies for
industrial medium-voltage drives. IEEE Transactions on Industrial Electronics 54 (6): 2930–2945.
https://doi.org/10.1109/TIE.2007.907044.
Saket, R.K. (2011). Reliability Evaluation of Defence Support Systems, 241–286. Berlin, Heidelberg:
Springer-Verlag. ISBN 978-3-642-17764-4. https://doi.org/10.1007/978-3-642-17764-4_8.
Sarita, K., Saket, R.K., and Khan, B. (2023). Reliability, availability, and condition monitoring of
inverters of grid-connected solar photovoltaic systems. IET Renewable Power Generation 17 (7):
1635–1653.
Schweizer, M. and Kolar, J.W. (2011). High efficiency drive system with 3-level T-type inverter. In:
Proceedings of the 2011 14th European Conference on Power Electronics and Applications, 1–10. IEEE.
Schweizer, M. and Kolar, J.W. (2013). Design and implementation of a highly efficient three-level
T-type converter for low-voltage applications. IEEE Transactions on Power Electronics 28 (2):
899–907. https://doi.org/10.1109/TPEL.2012.2203151.
Singh, S., Saket, R.K., and Khan, B. (2023). A comprehensive state-of-the-art review on reliability
assessment and charging methodologies of grid-integrated electric vehicles. IET Electrical Systems in
Transportation 13 (1): e12073.
Surana, K. and Jordaan, S. (2019). The climate mitigation opportunity behind global power
transmission and distribution. Nature Climate Change 9: 1–6. https://doi.org/10.1038/s41558-019-
0544-3.
Tayyab, M., Sarwar, A., Murshid, S. et al. (2022). Grid-connected operation and control of single-phase
asymmetrical multilevel inverter for distributed power generation. IET Renewable Power Generation
16 (16): 3629–3642. https://doi.org/10.1049/rpg2.12581.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
466 23 Reliability Assessment of Multilevel Inverter for Modern Power System Applications

Villanueva, E., Correa, P., Rodriguez, J.É., and Pacas, M. (2009). Control of a single-phase cascaded
H-bridge multilevel inverter for grid-connected photovoltaic systems. IEEE Transactions on
Industrial Electronics 56 (11): 4399–4406. https://doi.org/10.1109/TIE.2009.2029579.
Wang, D., Hemming, S., Yang, Y. et al. (2023). Multilevel inverters for electric aircraft applications:
current status and future trends. IEEE Transactions on Transportation Electrification 1. https://doi
.org/10.1109/TTE.2023.3296284.
Xiao, B. and Tolbert, L.M. (2014). Efficiency improved and current balanced three-phase modular
cascaded H-bridge multilevel PV inverter for grid-connected applications. In: 2014 IEEE Energy
Conversion Congress and Exposition (ECCE), 4661–4669. IEEE https://doi.org/10.1109/ECCE.2014
.6954039.
Yao, Z. and Xiao, L. (2012). Two-switch dual-buck grid-connected inverter with hysteresis current
control. IEEE Transactions on Power Electronics 27 (7): 3310–3318. https://doi.org/10.1109/TPEL
.2011.2179318.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
467

24

Reliability Aspects in Snubber Circuit for Industrial Power Applications

Dhawal Dwivedi 1 , Saumya Singh 1 , Kumari Sarita 2 , R. K. Saket 1 , and P. Sanjeevikumar 3
1
Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
2
Department of Electrical Engineering, Government Engineering College, Aurangabad, Aurangabad, Bihar, India
3
Department of Electrical Engineering, IT and Cybernetic, University of South-Eastern Norway, Notodden, Norway

24.1 Introduction
The early workers quickly noticed the inherent issue of triggering a switch with an inductive load
due to their repetitive switching. When the switch opened at first, an arc was seen across the main
switch contacts. Two things happened as a result of this arc. The connections were quickly cor-
roded. Second, the primary switch, which frequently heated quickly, used a portion or possibly the
majority of the energy available intended for the secondary spark. There was an excessive switch-
ing loss! The connection between switch behavior and loss in an inductive circuit was discovered
more than 160 years ago (Kothari et al. 2022; Sarita et al. 2023). Armand Fizeau devised a solution
in 1853 by placing a capacitor across the contacts. When the switch is turned off, the current of
switch is commutated to the capacitor (Severns 2008). Still, because the switch has discharged it,
the voltage across the capacitor is very low and only slowly increases as the integral of the current.
The primary arc is reduced. As a result, he is allowing the switch contacts to open with relatively
little voltage applied across them. The action is the same as that found in modern semiconduc-
tor capacitive turn-off snubbers. Fizeau’s paper shows that he saw the issue clearly and devised a
solution. Of course, the ringing voltage waveform that frequently associated with modern snubbers
was caused by a capacitor was placed across the primary while the secondary was being discharged.
Snubbers can serve various extremely valuable purposes, but they almost always come with “side
effects” that must be dealt with. Research efforts were redirected toward the development of res-
onant converters in the 1980s. The idea was to use resonant tanks in the converters to produce
oscillating voltage and/or current waveforms, which are typically sinusoidal, so that the power
switches may operate in a zero voltage switching (ZVS) or zero current switching (ZCS) condition.
The switching frequency of resonant converters can now reach hundreds of kilo-Hertz (usually
100 to 500 kHz) thanks to the reduction of switching loss and ongoing advancements in power
switches (Rashid et al. 2018). Power electronics would only be complete with snubbers. The purpose
of snubbers, compact network components in power-switching circuits, is to regulate the effects
of circuit reactances. Snubbers improve switching circuit performance and produce smaller size,
lighter size, more efficient, and more reliable switching circuits with greater switching frequen-
cies. A snubber’s main purpose is to remove energy from the circuit’s reactive components. Circuit
damping, regulating the voltage or current change rate, or clamping voltage overshoot are possible
benefits. A snubber lowers the stress and hence increases the reliability (Todd 1993). Snubbers are
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
468 24 Reliability Aspects in Snubber Circuit for Industrial Power Applications

often divided into two categories: passive networks and active networks. The only passive snubbers
covered in this chapter are resistors, capacitors, inductors, and diodes. Transistors and other active
switches are examples of active snubbers, frequently suggesting additional circuitry. The way of
energy is transferred to and from snubbers may influence their classification. The snubber is cat-
egorized as dissipative if its energy goes to a resistor for dissipation but as nondissipative if the
energy is returned to the input or output (Todd 1993; García-Caraveo et al. 2010). Another way
to categorize snubber circuits is based on their mode of operation, which allows us to modify the
load line to maintain it inside the safe operating area (SOA) for overload snubbers and minimize
or eliminate voltage or current spikes (García-Caraveo et al. 2010). The structure of this chapter is
organized as follows. Section 24.2 deals with the theory of passive snubber circuits and the con-
ventional design methods, Section 24.3 discusses the selection of turn-OFF snubber, Section 24.4
discusses the design of a discharge-suppressing resistor-capacitor-diode (RCD) snubber, the sim-
ulation results of RCD snubber are given in Sections 24.5 and 24.6 discusses reliability aspects in
snubber design for industrial power applications, and Section 24.7 concludes the chapter.

24.2 Passive Snubber Circuit

During turn-off, transient voltages are protected by snubber circuits. A series combination of resis-
tor R with a capacitor C is typically used in a simple RC snubber. A power MOSFET and an RC
snubber are connected in parallel. A circuit’s voltage rises significantly when the current source
is eliminated because of stray inductances. The snubber dampens this surge voltage to save the
power of MOSFET and other components nearby. Snubbers are capable of a wide range of actions,
including (Severns and EMI Reduce 2006):
1. Reduce or eliminate voltage or current spikes.
2. Limit dI/dt or dV/dt.
3. Shape the load line to keep it within the SOA.
4. Transfer power dissipation from the switch to a resistor or a useful load.
5. Reduce total losses due to switching.
6. Reduce EMI by damping voltage and current ringing.
The RCD turn-off snubber and the resistor-capacitor (RC) damping network are the two most
used types of snubbers.

24.2.1 Surge Across Switch

The stray inductance of the wire on the PCB layout stores energy when a MOSFET turns on. Surge
current occurs when the stored energy resonates with the MOSFET’s parasitic capacitance. Surge
voltage develops at the drain-source across the switch when it is turn-off due to a resonant phe-
nomenon between Parasitic/stray inductance Lstray and the parasitic capacitance of the MOSFET
Coss (Cds +Cgd ). The path of the “ringing current” in a half-bridge topology with high-side and
low-side switches is shown in Figure 24.1. Current IL flows through the stray inductance Lstray when
either the low-side or high-side is turned on. From ROHM application note (Rohm Semiconductor
2020), the frequency of surge (fsurge ) is given as:
√
√ {√ }2
√
1 √ LS ∕Coss
fsurge = √ √1− (24.1)
Ls Coss 2ROFF
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
24.3 Selection of Turn-OFF Snubber 469

Figure 24.1 Surge current path when turn-off Lstray/2

occurs. +Ve

(High side)
Cdc-link
I0

Vdc
(Low side)
I0
–Ve
Lstray/2

GND

24.3 Selection of Turn-OFF Snubber

A circuit’s stray inductances generate surge voltage. To absorb the surge voltage, a turn-OFF snub-
ber should be connected in parallel with a switching device. Snubbers can be connected across
individual switching devices or collectively across a power bus for all switching devices. The snub-
ber should ideally be positioned as close to the MOSFET as feasible to enhance its effectiveness.

24.3.1 Snubbers Across Each Switching Device

24.3.1.1 RC Snubber Circuit
It can be put close to the MOSFET, but every time the MOSFET switches on, Rs (snubber resistance)
must release the energy held in Cs (snubber capacitance). The size of Cs is constrained by the signif-
icant amount of energy (several watts) that the Rs would dissipate if the switching frequency were
high enough so, it is not suitable for high-switching applications. As a result, the snubber’s ability
to inhibit surges is diminished. A small-value resistor is required for an RC snubber to prevent the
drain current from increasing during turn-on in a high-capacitance switching device. It is ideal for
chopper circuits. The snubber resistor’s power dissipation is calculated as follows:
2
P = Cs Vdc fsw (24.2)
where Vdc = DC supply voltage and fsw = Switching frequency.

24.3.1.2 RCD Charge-Discharge Snubber

The RC snubber’s resistance can be increased by including a diode. This makes it possible to elim-
inate a current that switching devices have to share while turning on. During switch ON, Rs loses
energy at a similar rate to (a), but Cs ’s capacity for surge absorption is superior to (a) because surge
current passes through the diode. High di/dt in the snubber circuit can occur during the switching
transient. Hence, it is essential to take the diode’s recovery characteristics into account. To prevent
overvoltages, stray inductances should be reduced to a minimum. The outcome is the same if Rs and
Cs are connected in parallel. An RDC snubber is inappropriate for high-frequency applications due
to the significant power lost through the snubber resistor (TOSHIBA 2020). The snubber resistor’s
power dissipation is calculated as follows:
Lstray I0 fsw 2
Cs Vdc fsw
P= + (24.3)
2 2
where Lstray = Stray inductance of the main circuit and I0 = Drain current during device turn-off.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
470 24 Reliability Aspects in Snubber Circuit for Industrial Power Applications

24.3.1.3 Discharge-Suppressing RCD Snubber

In this turn-off surge voltage is suppressed, and it does not discharge all of the energy held in Cs
during each switching transient because Rs only dissipates the energy absorbed by Cs produced
during the overvoltage and hence the dissipation of power by a snubber is small. Therefore, even
at high switching frequencies, the energy consumption at Rs does not significantly rise and makes
it ideal for high-frequency switching applications As a result, a huge Cs can be used, realizing a
very efficient snubber circuit. However, it should be emphasized that this method demands an
extremely complicated wire layout that can be accomplished by a PCB with more than four layers.
The snubber resistor’s power dissipation is calculated as follows:
Lstray I02 fsw
P= (24.4)
2
Figure 24.2 shows different snubbers across each switching device examples: (a) RC snubber,
(b) RCD charge-discharge snubber, and (c) Discharge-suppressing RCD snubber.

24.3.2 Lumped Snubbers Between Power Buses

24.3.2.1 C Snubber
The simplest is C snubbers, but the main circuit’s stray inductances and the snubber capacitor’s LC
resonance make them vulnerable to voltage oscillation.

+Ve +Ve +Ve

Cs Ds Rs Ds
Rs
Rs Cdc-link Cs Cdc-link Cs Cdc-link
Vdc

Vdc

Vdc
Rs
Cs Ds Rs
Ds
Rs Cs Cs
–Ve –Ve –Ve
(a) (b) (c)

Figure 24.2 Snubber circuits: (a) RC snubber, (b) RCD charge-discharge snubber, and (c)
discharge-suppressing RCD snubber.

+Ve +Ve

Ds Rs
Cdc-link Cdc-link
Cs
Vdc

Vdc

Cs

–Ve –Ve
(a) (b)

Figure 24.3 Lumped snubbers between power buses: (a) C snubber and (b) RCD snubber.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
24.4 Design of a Discharge-Suppressing RCD Snubber 471

24.3.2.2 RCD Snubber

Snubber diodes should be chosen carefully since they could result in voltage oscillation and a sig-
nificant voltage spike during reverse recovery. Figure 24.3 shows the lumped snubber (a) C snubber
circuit and (b) RCD snubber.

24.4 Design of a Discharge-Suppressing RCD Snubber

Figure 24.4 shows a detailed circuit of RCD. The following equations can be used to compute the
voltages and current constants shown in the waveform.

(a) Figure 24.4a VLs is the voltage generated by the snubber’s inductance Ls , and it can be calculated
as follows:
di
VLs = Vdc + Vfr + Ls × (24.5)
dt
The forward voltage Vfr of the diode Ds should be kept as low as possible and decrease Ls that
can result in a voltage surge in order to achieve this.
(b) VLs2 and Cs When the snubber capacitor Cs is overcharged with energy from the main circuit’s
stray inductance Lstray , a peak voltage known as VLs2 is observed across the capacitor Cs . Their
energies are equal since the Lstray energy is transferred to Cs . As a result, the following equation
is valid:
1 2 1
× Lstray × IOFF = × Cs × (VLs2 − Vdc )2 (24.6)
2 2
This equation yields the following result for the value of Cs :
2
Lstray × IOFF
Cs = (24.7)
(VLs2 − Vdc )2
where Ls = inductance of the snubber circuit, Vfr = forward voltage of Ds , Cs = capacitor of the
snubber circuit, VLs1 = peak surge voltage of Ls , VLs2 = peak forward voltage of Cs and IOFF = turn
OFF current. It is necessary to determine VLs2 , considering the MOSFET’s sustained voltage.

Lstray
+Ve
Ls VDS
ID
Ds
Rs
Cs Cdc-link
Vdc

VLs1
Rs IOFF
VLs2 VDD
Ds di/dt
Cs
Ls
–Ve
(a) (b)

Figure 24.4 Discharge-suppressing RCD (a) snubber circuit and (b) waveform.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
472 24 Reliability Aspects in Snubber Circuit for Industrial Power Applications

V Figure 24.5 Time constant versus the amount of

discharge.

37%

10%
t
τ 2.3 τ

(c) Selection of the value of snubber resistor Rs : The snubber resistor Rs aims to release the elec-
tric charge stored in the snubber capacitor before a MOSFET starts its further turn-on action.
Suppose that is the discharge time constant 𝜏, then:
𝜏 =R⋅C (24.8)
Where, 𝜏 represents the time it takes for the voltage to drop to 37% of the stored voltage, as
illustrated in Figure 24.5, the voltage drops to 10% in 2.3𝜏, or 90% of the capacitor’s stored
charge is discharged.
Before performing the next turn-off, the capacitor needs to be discharged. As a result, the follow-
ing equation needs to be satisfied. The following formula can be used to compute Rs :
1
2.3𝜏 ≤
fsw
1
2.3 × Rs ⋅ Cs ≤
fsw
1
Rs ≤ (24.9)
2.3 ⋅ Cs ⋅ fsw
The snubber may display current oscillation if the resistor’s value is too low. Use a resistor with a
resistance that is as high as possible.

24.5 Simulation Results of RCD Snubber

Consider a two-level half-bridge inverter with a stray or parasitic induction of around 2.9 nH. The
input Vdc voltage is 100 V, the load resistance is 20 Ω, and the switching frequency is 10 kHz. The
waveform of surge/spike is shown in Figure 24.6. Figure 24.6a For snubber across each switching
device: Figure 24.7 displays the waveform of the drain to source voltage after inserting a C snubber
of 560 pf across the switching devices. Typically, the range of snubber capacitors varies from a few
pf to 1𝜇f depending upon the application and is made up of MLCC (multi-layer ceramic capacitor),
film capacitor, etc. But for a higher current rating, the snubber capacitor goes up to 10𝜇f. Then
analyze the waveform of one cycle, Tr (ringing time period)=80 𝜇s. For optimum snubber design:
1
fr = √ (24.10)
2𝜋 Lstray Cs
For the underdamped condition:
√
L
Rs < (24.11)
C
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
24.5 Simulation Results of RCD Snubber 473

200
VDS (V)

100

1
Gating pulse

0.8
0.6
0.4
0.2
0
0.0688 0.069 0.0692 0.0694 0.0696 0.0698 0.07
Times (s)

Figure 24.6 Surge across the drain to source voltage (VDS ) with gating pulse of MOSFET.

200
150
VDS (V)

100
50
0

1
Gating pulse

0.8
0.6
0.4
0.2
0
0.0842 0.0843 0.0844 0.0845 0.0846 0.0847 0.0848 0.0849 0.085
Times (s)

Figure 24.7 The drain to source voltage (VDS ) with the gating pulse of MOSFET after inserting C snubber.

200 Tr
VDS ( V)

100

0
0.084326 0.08432605 0.0843261 0.08432615 0.0843262 0.08432625
Times (s)

Figure 24.8 The oscillatory view of drain to source voltage (VDS ) after inserting C snubber.

Rs lesser than 10 Ω. This Rs is for RC or RCD snubber circuits. Figure 24.9 shows the drain to source
voltage with gating pulse of MOSFET after inserting RC snubber.

Example 24.1
Calculate the required parameters for a snubber circuit to provide protection to a MOSFET used
in a single-phase bridge converter. The MOSFET has a maximum capability of 60 V/𝜇s. The input
line to line voltage has a peak value of 425 V and the stray inductance is 0.2 mH.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
474 24 Reliability Aspects in Snubber Circuit for Industrial Power Applications

200
150
VDS (V)

100
50
0

1
Gating pulse

0.8
0.6
0.4
0.2
0
0.0374 0.0375 0.0376 0.0377 0.0378 0.0379 0.038
Times (s)

Figure 24.9 The drain to source voltage (VDS ) with the gating pulse of MOSFET after inserting RC snubber.

Solution: Given:

dv
= 60 V∕𝜇s, L = 0.2mH, Vm = 425V
dt
Then,
2
⎛ ⎞
( )2
1 ⎜ 0.564Vm ⎟ 1 0.564 × 425
C= ⎜ ⎟ = × 10−6
2L ⎜ dv ⎟ 2 × 0.2 × 10 −3 60
⎝ dt ⎠

C = 0.04𝜇F
In the above equation observe that 10−6 have been multiplied in numerator inside the brackets. It
60
comes from 60 V/𝜇s, i.e. 1×10 −6 to be substituted in equation for C. Let the damping factor 𝜁= 0.65.

Then,
√
R C
𝜁=
2 L
√ √
L 0.2 × 10−3
R = 2𝜁 =2×
C 0.04 × 10−6
R = 92Ω.

Example 24.2
dv
For the circuit shown in Figure 24.10, the IGBT is operated at 2 kHz. The dt
required is 100 V/𝜇s.
The discharge current is to be limited to 100 A. Determine
i) Values of Rs , and Cs
ii) Snubber loss
iii) Power rating of Rs
Load and stray inductances are negligible.
Solution: Given:
dv
Load resistance, R = 5Ω, frequency= 2 kHz, Vs = 200 V, dt
= 100 V/𝜇s, ITD = 100 A, L = 0
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
24.5 Simulation Results of RCD Snubber 475

Figure 24.10 Circuit for Example 24.2. S

Rs
S1 Cs

Vs = 200 V
R=5Ω
Load
L

Figure 24.11 Path of ITD .

Rs

ITD Cs
VCs = 200 V

(i) To obtain the value of Rs and Cs

Rs limits the discharging current through MOSFET S1 . S1 − Rs − Cs forms a loop when S1 turns
ON. Prior to turning ON of S1 , Cs charges to 200 V. From this Figure 24.11,
VCs = Rs ITD

VCs Vs 200
Rs = = = = 2Ω
ITD ITD 100
KVL can express the charging current of Cs to Vs − Rs − Cs − R − L loop as,
1
Vs = Rs iC (t) + i (t) dt + VC (t − 0) + RiC (t)
C∫ C
on solving the above equation,
Vs t
iC (t) = exp− 𝜏 , Here 𝜏 = Rs + R Cs and Vc (t = 0) = 0.
Rs + R
Hence, the voltage across the MOSFET can be expressed as,
Vs t
iC (t) = exp − , Here 𝜏 = (Rs + R)Cs and Cs and vC (t = 0) = 0. (24.12)
Rs + R 𝜏
Hence, the voltage across the MOSFET can be expressed as,
RVs − t
vS1 (t) = Vs − Ric (t) = Vs − e 𝜏
R + Rs

RVs Rs Vs
At t = 0, vS1 (0) = Vs −
Rs + R R s + R

RVs 0.9678RVs
and at t = 𝜏, vS1 (0) = Vs − exp−1 = Vs −
Rs + R Rs + R
Rs Vs + 0.6321RVs
=
Rs + R
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
476 24 Reliability Aspects in Snubber Circuit for Industrial Power Applications

dv
Now dt
can be expressed as,
Rs Vs + 0.6321RVs RV
− s s
dv vT1 (𝜏) − vT1 (0) Rs + R Rs + R
= =
dt 𝜏 (Rs + R)Cs
0.632RVs
=
(Rs + R)2 Cs
0.632RVs 0.632 × 5 × 200
Cs = = = 0.1289𝜇F (24.13)
dv 100
(Rs + R)2 −6
(2 + 5)2
dt 1 × 10
(ii) To obtain snubber loss
The power stored in Cs is dissipated in Rs . Hence, it is snubber loss
1
Ps = C V 2f
2 s s s
1
= × 0.1289 × 10−6 × 2002 × 2 × 103
2
= 5.2W
(iii) To obtain power rating of Rs
The power stored in Cs is dissipated in Rs . It is 5.2 W. Hence power rating of Rs will be 5.2 W.

24.6 Reliability Aspects in Snubber Design for Industrial Power

Applications
Before diving into the reliability aspects, it was essential to understand the basic structure and
functionality of snubber circuits that were covered in Sections 24.3–24.5. This section will discuss
the reliability aspects of the snubber circuit.
In the realm of industrial power applications, snubber circuits play a pivotal role. Their primary
function is to suppress voltage transients, limit voltage spikes, and protect power semiconductor
devices. Ensuring the reliability of snubber circuits is of paramount importance to maintain the
overall efficiency, safety, and longevity of power systems. In industrial applications, these are the
main reasons behind the need for reliable operation of snubber circuit.
● Protection of Expensive Equipment: Voltage spikes can cause damage to semiconductor
devices, leading to costly replacements.
● Operational Continuity: An unreliable snubber can cause outages, affecting production
timelines.
● Safety: High-voltage transients can cause accidents if not managed appropriately.
The following are the factors affecting system reliability in industrial power electronics
applications.
● Component Selection: The quality and specifications of resistors, capacitors, and diodes deter-
mine their performance under stress.
● Thermal Management: Excessive heat can reduce the life of snubber components.
● Circuit Design: Proper placement and interconnections can play a role in enhancing or under-
mining reliability.
● Environmental Factors: Moisture, dust, and corrosive agents can degrade components over
time.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
24.6 Reliability Aspects in Snubber Design for Industrial Power Applications 477

System reliability is typically expressed in terms of the MTTF, representing the average lifespan
of a system until it encounters its first failure as discussed by Singh et al. (2023). This parameter
can be determined by examining the system’s failure rate, especially when components showcase
exponential behaviors during their useful operational periods, as discussed by the authors Sarita
et al. (2023).
For streamlining reliability analysis, certain assumptions are taken into account:
● The snubber’s failure would result in an open circuit.
● Each individual component’s failure within the snubber circuit would render the entire circuit
nonfunctional.
● The combined failure rate of the snubber is the aggregate of its individual components’ failure
rates:
𝜆 S = 𝜆R + 𝜆 C + 𝜆 D (24.14)

This supposition suggests that a malfunction in the snubber circuit doesn’t lead to the entire
system’s breakdown. Put differently, even if the snubber circuit encounters an issue, the pri-
mary circuit remains operational. However, the entire system would only collapse if the switch
malfunctions.
The switch’s likelihood of breaking down is affected by the condition of the snubber circuit. As
long as the snubber circuit is working, the switch’s failure chance, represented by 𝜆W , depends
on the snubber’s capacitor size. But if the snubber breaks, the switch uses more power and has a
greater risk of failure, represented by 𝜆W1 .
The system’s reliability can be determined by combining the reliability of separate, nonover-
lapping events that result in the system working correctly, as illustrated in Figure 24.12. These
successful outcomes are: (i) both the switch and snubber remain functional for a time period from
0 to t, or (ii) the snubber breaks between 0 to t1 , but the switch continues to work from 0 to t. We
can represent the reliability of these events with functions R1 and R2 , which are detailed in the the
following equations:
R1 (t) = exp(−𝜆W (t)) exp(−𝜆S (t)) (24.15)
t
R2 (t) = (exp(−𝜆W t1 ))(𝜆S exp(−𝜆S t1 dt1 ))(exp(−𝜆W1 (t − t1 ))) (24.16)
∫0
The terms on the right side of R2 (t) represent the following: the first term denotes the switch
working successfully up to time t1 with a failure rate of 𝜆W ; the second term indicates the likelihood
of the snubber failing between times t1 and t1 + dt1 ; and the third term describes the switch’s

Successful
SW-Success operation SW-Success
SN-Success SN-Failure
λS

λW λW1

SW-Failure Failure SW-Failure

SN-Success operation SN-Failure

Figure 24.12 Reliability of switch and snubber represented in state space diagram.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
478 24 Reliability Aspects in Snubber Circuit for Industrial Power Applications

successful operation from time t1 to t with a failure rate of 𝜆W1 , assuming the snubber isn’t
functioning. By combining R1 and R2 , we can determine the overall system reliability and then
compute the MTTF:
𝜆W1 + 𝜆S
MTTF = (24.17)
(𝜆W + 𝜆S )𝜆W1
The average failure time for a circuit without a snubber is given by 1∕𝜆W1 . Thus, the MTTF ratio
comparing a system with a snubber to one without it can be defined as:
𝜆W1 + 𝜆S
MTTFratio = (24.18)
(𝜆W + 𝜆S )
A compelling observation emerges when examining MTTF as a function of snubber capacitance,
as explained by Parvari et al. (2018). The Authors concluded that the initial increments in capaci-
tance dramatically enhance MTTF due to a sharp decline in the switch’s failure rate. However, as
capacitance continues to rise, the benefits plateau. This behavior stems from the balance between
reducing switch power loss and increasing snubber-set energy storage. This trade-off suggests an
optimum capacitance value that strikes a balance between efficiency and reliability. To achieve
reliable operation of the system, the following methods can be implemented.

● Component Derating: Using components at levels below their maximum rated values ensures
longer lifespans.
● Heat Sinks and Cooling: Prevents thermal overload and ensures steady performance.
● Quality Checks: Regular inspection and testing can identify issues before they escalate.
● Protective Coatings: Can protect against environmental hazards.

24.7 Conclusion

In the vast landscape of industrial power applications, snubber circuits might appear to be minute
details. Yet, their reliability ensures the smooth operation, safety, and efficiency of much larger
and more intricate systems. By understanding and prioritizing their reliability aspects, industries
can ensure operational excellence and reduce downtimes. This chapter dealt with different aspects
of the reliability assessment of snubber circuits. The choice of optimal capacitor values does not
ensure reliable operation. The chosen values of R, C, and D must be such that reliability is enhanced
as discussed in this chapter.

References

García-Caraveo, A., Soto, Á., González, R., and Bañuelos-Sánchez, P. (2010). Brief review on snubber
circuits. In: 2010 20th International Conference on Electronics Communications and Computers
(CONIELECOMP), 271–275. https://doi.org/10.1109/CONIELECOMP.2010.5440754.
Kothari, D.P., Nagrath, I.J., and Saket, R.K. (2022). Modern Power System Analysis. Tata McGraw-Hill
Publishing Company.
Parvari, R., Zarghani, M., and Kaboli, S. (2018). RCD snubber design based on reliability consideration:
a case study for thermal balancing in power electronic converters. Microelectronics Reliability 88:
1311–1315.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 479

Rashid, M.H., Hui, S.Y.R., and Chung, H.S.-H. (2018). Chapter 12 - Resonant and soft-switching
converters. In: Power Electronics Handbook, 4e (ed. M.H. Rashid), 339–383. Butterworth-
Heinemann. ISBN 978-0-12-811407-0. https://doi.org/10.1016/B978-0-12-811407-0.00012-X. https://
www.sciencedirect.com/science/article/pii/B978012811407000012X.
Rohm Semiconductor (2020). Snubber circuit design methods. Application Note. https://fscdn.rohm
.com/en/products/databook/applinote/discrete/sic/mosfet/sic-mos_snubber_circuit_design_an-e
.pdf (accessed 9 February 2024).
Sarita, K., Saket, R.K., and Khan, B. (2023). Reliability, availability, and condition monitoring of
inverters of grid-connected solar photovoltaic systems. IET Renewable Power Generation 17 (7):
1635–1653.
Severns, R.P. (2008). Snubber Circuits for Power Electronics. Rudy Severns. https://books.google.co.in/
books?id=SmVFPwAACAAJ.
Severns, R. and EMI Reduce (2006). Design of Snubbers for Power Circuits. International Rectifier
Corporation.
Singh, S., Saket, R.K., and Khan, B. (2023). A comprehensive state-of-the-art review on reliability
assessment and charging methodologies of grid-integrated electric vehicles. IET Electrical Systems in
Transportation 13 (1): e12073. https://doi.org/10.1049/els2.12073.
Todd, P.C. (1993). Snubber circuits: theory, design and application. In: Unitrode-Power Supply Design
Seminar. Texas Instruments Incorporated.
TOSHIBA (2020). Power MOSFET selecting MOSFFETs and consideration for circuit design.
Application Note. https://toshiba.semicon-storage.com/info/application_note_en_20180726_
AKX00064.pdf?did=13416 (accessed 9 February 2024).
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
481

25

Reliability Assessment of Power Electronics Devices and Systems for

Modern Power Applications
Jyoti Maurya 1 , Saumya Singh 1 , Sachin Kumar 2 , P. Sanjeevikumar 3 , and R. K. Saket 1
1
Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
2
Department of Electrical Engineering, G.B. Pant Institute of Engineering and Technology, Pauri-Garhwal, Uttarakhand, India
3
Department of Electrical Engineering, IT and Cybernetics, University of South-Eastern Norway, Porsgrunn, Norway

25.1 Introduction

Power electronic applications involve the management and control of power flow. In power elec-
tronics (PE), the predominant focus is efficiently controlling power flow by employing various
switching techniques (Peyghami et al. 2020). Different power semiconductor devices, highlighting
their usage as switches in power electronic applications. Some commonly used power semiconduc-
tor switches include bipolar junction transistors (BJTs), diodes, thyristors, gate turn-off thyristors
(GTOs), insulated gate bipolar transistors (IGBTs), metal oxide semiconductor field effect tran-
sistors (MOSFETs), and metal oxide semiconductor-controlled thyristor switches (MCTs). Power
MOSFETs and IGBTs significantly impact the control of modern power equipment, so their study
is emphasized.
The stability of these systems is today a primary concern and most challenging difficulty in mod-
ern power applications, where these semiconductor switches have been integrated with power
converters for grid-connected power systems. Power system modernization is required for reliable
and secure power delivery with a zero carbon footprint. It necessitates the deployment of new tech-
nology and infrastructure and the deregulation of the electrical sector.
Electronic transmission, renewable energy resources (RER), distribution system storage, and
e-mobility are well-established technologies essential to power system modernization (Kothari
et al. 2022). Additionally, PE are a vital component in the energy conversion processes of the
above-mentioned technologies. Power systems are heading for 100% renewable energies in the
future, which increases the necessity of PE. However, as power converters are a common source
of failure in many applications (Peyghami et al. 2021), substantial downtime and maintenance
expenses are required. Furthermore, industrial experience and field data show that power con-
verters are susceptible to wear-out and growing defeats resting on operation conditions (Peyghami
et al. 2019).
Power converters consist of various elements, which include gate drivers, power modules, capac-
itors, cooling systems, and control units. Among these, electrolytic power modules and capacitors
are the most delicate parts, susceptible to wear-out failures. Various elements, such as the durabil-
ity of the device’s structure, the amount of electrical load it handles, the surrounding environment,
Sometimes this is also known as static reliability.
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
482 25 Reliability Assessment of Power Electronics Devices and Systems for Modern Power Applications

parts. These factors collectively contribute to the gradual deterioration of component materials over
the converter’s extended operation, ultimately activating potential failure modes.
Utilizing principles of failure physics within stress-strength analysis (SSA) allows for anticipa-
tion of component reliability and improvement of wear-out performance across different failure
modes. Electrothermal modeling is crucial in SSA and operates on three levels: device, converter,
and system.
System modeling identifies the load on the converter by analyzing its utilization within the power
system. Methods like power-sharing are employed to gauge the load on converters that are con-
nected in parallel. Additionally, converter-level examinations utilize electrical models and simula-
tions to evaluate strain on individual components due to control and switching systems in place.
Electrothermal device modeling becomes indispensable for capturing crucial thermal parameters
such as hot-spot temperature, which, in turn, help predict failure modes within a designated mis-
sion profile. Consequently, SSA analysis encompasses both fast dynamics at the device level in the
switching frequency range and slow dynamics at a system level in the hourly load fluctuation zone.
As a result, forecasting wear-out failures over a yearly mission profile takes time. It will place a sub-
stantial computational load on system-level analysis, especially in enormous-scale power systems.
In the context of power electronic devices and systems (PEDS), the technique of SSA has been
harnessed to model and extend the lifespan of both capacitors and semiconductors. Focusing the
converter level, active thermal management strategies, including adaptive switching frequency,
suitable modulation methods, and controlling reactive power, are applied to bolster the converter’s
overall lifespan.
Converter topologies and solar array parameters are two further factors that influence converter
lifespan. Furthermore, capacitors’ lifespan expansion is examined by interleaving converters.
Again, by correctly altering converter shifting and loading device damages from a high-stress
converter to a low-stress converter, system level dependability in multi-converter systems is
increased (Singh et al. 2019).
The reliability assessments mentioned earlier focus solely on forecasting and improving the
lifespan of converters, even within broader system-level investigations. Nevertheless, making the
best decisions when designing, planning, operating, and maintaining power system converters
requires comprehensively exploring their impact on the reliability of the power system. This calls
for the fusion of concepts from power electronic reliability with evaluation methods for reliability
in power systems. Power system reliability refers to the capacity of a power system to fulfill
consumer demands (Fischer et al. 2019).
The measurement of this capacity is primarily assessed using adequacy indicators like loss of load
expectation (LOLE) and anticipated energy deficit Expected Energy Not Supplied. Moreover, apart
from conventional evaluation of power system dependability, there has been significant research
into the reliability of power systems based on PE, such as wind farms (WFs) and high-voltage direct
current (HVDC) transmission systems.
Furthermore, Carroll et al. (2014) investigate microgrid systems’ dependability by considering
conditions affecting power electronic converters. In power system analysis, historical data from
advanced research on analogous situations is used to establish the failure rate of converters. Addi-
tionally, the assessment of power system reliability has yet to incorporate wear-out failures of con-
verters.
In real-world applications, converter and component wear-out may occur sooner than expected
(Enslin et al. 2019). Consequently, an operational failure rate of converter will rise, and its effective
lifespan will become restricted. This will lead to the deterioration of converter components over
time, undermining the reliability of the entire system and elevating the potential for downtime
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
25.2 Concept of PEDS Reliability in Modern Power System 483

and maintenance expenses in Maurya and Saket (2023). This scenario is particularly pertinent in
extensive power electronic-based power systems (PEPS). To circumvent these challenges, imple-
menting appropriate protocols for achieving the best decisions is crucial for designing, running,
and up-keeping contemporary systems reliant on PE.
At system level, this requires an in-depth analysis of reliability, involving the modeling of con-
verter reliability within the evaluation of power system reliability. However, this method is highly
time-intensive and almost unfeasible for extensive, large-scale power plants. This is because when
utilizing electrothermal modeling relying on SSA, it requires a time-domain examination covering
time spans from microseconds, which are associated with the converter switching frequency, all
the way to hourly adjustments in load. Consequently, there is a need for streamlined approaches
to electrothermal modeling across various time scales to effectively model the reliability of power
electronic systems at a system level. This will easily tackle the mentioned challenges stemming
from converting converters into power systems.

(1) Due to necessity of making decisions about converter operation, planning, and mainte-
nance system-wide basis, significance of system-level dependability modeling in power
electronic-based systems cannot be overlooked. This chapter strives to connect models of
converter reliability with power system reliability principles, intending to assess the reliability
of PE-based power systems.
(2) Due to growing utilization of converters within power systems, the likelihood of their malfunc-
tions becomes intertwined with stochastic factors, while deteriorative breakdowns influence
the power system’s effectiveness. This chapter aims to demonstrate interplay between converter
malfunctions, aging, and their repercussions on power system dependability.
(3) Due to intricate nature of reliability modeling in SSA-based converters, it is imperative to
introduce more straightforward approaches for analyzing systems at macro level. This piece
elaborates on the required precision when it comes to modeling and reliability of converters,
particularly in the context of power system analysis across different applications.
(4) Due to effects of component aging, there will be a rise in the failure likelihood of converters.
Consequently, it becomes necessary to replace them in accordance with an appropriate main-
tenance strategy.
As followed by an introduction, this chapter is structured in the following manner. An intro-
duction, considering the recent literature available, is discussed in Section 25.1. Section 25.2
includes concept of PEDS reliability in modern power systems, Section 25.3 delves into V-shape
model-based reliability assessment in PEDS, and converter reliability modeling has been elab-
orated in Section 25.4. At last, Section 25.5 serves as the chapter’s conclusion and future chal-
lenges.

25.2 Concept of PEDS Reliability in Modern Power System

The schematic of evaluating the modern power system reliability is depicted in Figure 25.1. Reli-
ability plays a vital role in PEDS utilized within contemporary modern power systems. Reliability
pertains to the capacity of an entity or system to execute its designated operations under predefined
circumstances and across a specified duration. As defined, system’s effectiveness must remain con-
sistent within a defined range over a specific time frame, as depicted in Figure 25.2.
The specific system application dictates evaluation of performance metrics. The measure of
dependability performance in scenarios involving missions, such as spacecraft, relates to the
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
484 25 Reliability Assessment of Power Electronics Devices and Systems for Modern Power Applications

Mission profile

WT Wind power
Converter Acon reliability uncertainty
Reliability data reliability model model
model

WT
generation
Reliability data Line
model
converter
reliability
Mission profile
model
Acon AWT

Reliability data Converter WF EENP

reliability generation model Atime
Mission profile model Aprod

Acon
AWF
HVDC system AHVDC
Reliability data
model
Generation
system
Generators Generators AGen model
reliability data model

Load profile
Convolving
LOLE
Power

generation and
LDC EENS
load models

Time

Figure 25.1 Modern power systems reliability evaluation.

possibility of lasting for the entire specified period. Similarly, in systems that can be maintained
or repaired, like automobiles, ensuring continuous functionality irrespective of past malfunctions
is paramount. This is gauged through availability metrics, which indicate the probability of
the system is operational at any point in time (Alhmoud 2018). This description asserts that the
system’s up-time should be upheld at a desirable level by employing either proactive maintenance
methods before potential failures or reactive maintenance approaches following malfunctions
available in Figure 25.3.
Moreover, performance metrics could represent a tangible characteristic of an item or system, like
capacitance of a capacitor, which falls below 80% of its nominal value, indicating a malfunction. A
power system constitutes an intricate amalgamation of numerous subsystems and parts that must
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
25.2 Concept of PEDS Reliability in Modern Power System 485

Target performance
Acceptable
Performance

Error deviation
Actual performance

Remedial
action

Target Time
lifetime

Figure 25.2 Characterization of reliability as a metric for evaluating performance within a system.

Power system
reliability

Adequacy Security

Long term Short term Static Dynamic

Figure 25.3 Notion of reliability within electric power systems encompasses both adequacy and security
considerations.

function uninterrupted over a prolonged period of time. Following this declaration, it is imperative
to guarantee the power system’s capability to satisfy demand consistently. The operational phase
of power systems is intertwined with their immediate effectiveness, delineated as their ability to
endure unforeseen circumstances and disruptions. This capacity is characterized as a security con-
nection related to stability of a power system (Jiang et al. 2021). Moreover, sustained effectiveness
of a power system is connected to its capacity to provide power to demands despite fluctuations in
generator accessibility and load requirements. This requires appropriate infrastructure for genera-
tion, transmission, and distribution. This capacity is referred to as power system adequacy, a metric
for both design and planning (facility and operational planning) aspects.
Consequently, power system reliability is characterized by power system’s capability to address
sudden events in short run by reacting to unexpected situations and, in long run, by establishing
and utilizing suitable infrastructure. To guarantee the system’s effectiveness, these infrastructures
must exhibit a fair degree of accessibility as per these benchmarks.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
486 25 Reliability Assessment of Power Electronics Devices and Systems for Modern Power Applications

25.3 V-Shape Model-Based Reliability Assessment in PEDS

PEPS are structured into three tiers: the power system, subsystems, and individual components,
as depicted in Figure 25.4 (Yang et al. 2011). While subsystems discussed here primarily pertain
to power converters, the scope can also cover other subsystems like generators and transformers.
Within PEPS, each converter has a distinct function in facilitating power and energy distribu-
tion, contingent on its designated purpose. These converters interlink renewable energy setups,
electronic transmission systems, e-transportation charging points, and energy storage units. Con-
sequently, their influence on PEPS performance varies. Moreover, each converter’s reliability level
is dictated by its specific capabilities within the PEPS framework.
Moreover, various elements exist within each converter, including power devices, capacitors,
cooling mechanisms, and control and protection modules. The specifications for converter’s relia-
bility within the PEPS dictate design and dimensions of these elements. Hence, performance of
the converter in PEPS is determined by its level of dependability, and its constituent parts are
crafted according to this measure of reliability. Consequently, modeling, evaluating dependability,
and enhancing the PEPS must be approached hierarchically.

25.3.1 Hierarchical Reliability Modeling and Assessment

Power converters represent one of the most precarious components within contemporary power
grids, frequently susceptible to deterioration-related malfunctions. This segment will specifically
delve into power converter breakdowns while maintaining relevance to other subsystems. Fail-
ures in power converters often stem from capacitors and power modules. Operational specifics,
surrounding conditions, and task requirements dictate the duration of their operational efficiency

System reliability Figure 25.4 Correlation between PE and

Power system level (LOLE, EENS,...) power system reliability concept. Source:
transmission and Adapted from Peyghami et al. (2021).
distribution systems
Maintanance

Availability
Subsystem reliability
power electronics converters
and other subsystems
Failure probability and
failure rate (F, λ,...)

Wear-out
Component level
modeling
resistor, inductor diode and
other components

Lifetime model
(Nf, Lo,....)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
25.3 V-Shape Model-Based Reliability Assessment in PEDS 487

and reliability. The longevity of power modules is directly linked to their susceptibility to failure
across numerous operational cycles (Nf ), an aspect that is notably influenced by fluctuations in
junction operating temperature. These temperature variations, in turn, are governed by variables
such as converter workloads and external environmental factors, including ambient temperature
and fluctuations.
Different manufacturers and technologies offer a range of lifespan models for power equipment.
The choice of an appropriate model for assessing reliability depends on specific device and avail-
ability of relevant data. For instance, in realm of power devices, one can calculate the cycles until
failure (Nf ) using subsequent equation:
( )
𝛼 𝛽
Nf = A ⋅ ΔTj ⋅ exp (25.1)
Tj

where ΔT and T are the junction temperature swing and its average value. The constants of A, 𝛼,
and 𝛽 are curve-fitting constants obtained from aging tests.
( )−n2
Tr −To
Vo
Lo = Lr ⋅ 2 nf
(25.2)
Vr
Lr signifies rated lifespan under rated voltage Vr and upper category temperature Tr , while Lo
denotes operational life of capacitor under operating voltage Vo , and temperature To . As stated
in Eq. (25.2), longevity and dependability of a capacitor are influenced by its voltage and temper-
ature. These factors are intertwined with operational conditions, encompassing capacitor’s load
and ambient temperature. It’s worth highlighting that more sophisticated models for a lifespan of
capacitors and power devices will encompass additional factors n2 such as vibration, humidity, and
other stresses, particularly pertinent in specific applications.
Due to the influence of converter loading on lifespan of capacitors and power devices, it becomes
essential to comprehensively examine the converter’s operational circumstances at the system
level. Hence, the hierarchical approach’s initial phase to ensure reliability involves analyzing
power distribution. This analysis is grounded in energy management strategy employed within
broader power system framework.
Once operational profiles for individual converters are established, next task involves translating
these profiles into thermal and mechanical domains. This is essential for gathering information
about thermal factors that impact the longevity of converter components, including metrics like
junction temperature (i) and capacitor voltage (ii). Consequently, subsequent stage entails execut-
ing an electrothermal modeling exercise at the subsystem level. This modeling effort aims to gauge
thermal stress experienced by each device, thereby contributing to a comprehensive understanding
of their performance and durability.
Thermal stresses are linked to distinctive and predominant ways individual devices fail at the
device level. It’s worth noting that each device exhibits distinct failure modes and mechanisms,
necessitating an appropriate lifetime model for accurate modeling. In this research, a general
lifetime model is considered for each device. Nevertheless, this approach can be expanded to
encompass diverse failure scenarios, enhancing precision of the dependability model. The thermal
variables collected at sub-system level can compute the expended lifetime based on a specified
mission profile and failure mode. With a particular lifetime model in a component, an uncertainty
analysis becomes necessary to determine time until failure for a specific failure mode associated
with it.
One way to achieve this is by conducting a SSA, which assesses the probability of failure
associated with a specific failure mechanism. This analysis also helps quantify the manufacturing
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
488 25 Reliability Assessment of Power Electronics Devices and Systems for Modern Power Applications

variations and modeling uncertainties affecting the reliability of a component when it operates
according to a specified mission profile. Consequently, analysis at device level will establish a
connection between thermal stresses and functions that describe reliability and failure rates for
each specific failure mechanism.
After estimating device’s reliability, converter’s likelihood of failure and operational availability is
established using failure rates of its parts. The last step in evaluating PEPS reliability involves merg-
ing converter’s availability with the energy accessibility of the primary generation mechanisms and
load patterns. This calculation determines extent of energy or load loss due to the absence of differ-
ent components. This can be achieved by concurrently refining load pattern and generation system
model. A pertinent dependability measure like LOLE, EENS, and others can represent resulting
outcome.
The initial phase entails the dissection of PEPS into its subsystems, constituents, and potential
points of failure. This process results in segmentation of overall system into sub-systems. Within
this phase, a power flow analysis is conducted to establish load characteristics of each converter.
Subsequently, converter is deconstructed down to its component levels. During this stage, an elec-
trothermal mapping technique is utilized to assess the stress experienced by components. Lastly,
analysis progresses to categorize failure mechanisms of components and various failure scenarios
to ascertain their respective lifetimes under applied stress. The second stage of proposed V-shape
model involves amalgamating reliability models derived from distinct failure causes for each com-
ponent, extending up to PEPS.
Consequently, a cumulative failure rate of various failure modes is aggregated to calculate the
overall failure rate for each device. This device failure rate is subsequently employed in modeling
the availability of converters. Ultimately, converter’s availability is integrated with load model to
assess the performance of PEPS using metrics like LOLE, EENS, and similar indices.

25.3.2 Model-Based Reliability Enhancement

Reliability evaluations usually occur two times: before system construction to facilitate planning
and design goals and during operation to improve reliability, plan for expansion, and schedule
maintenance. The primary objective in both instances is to configure and run PEPS with optimal
cost efficiency. After assessing system performance in these dual scenarios, the subsequent task
involves ascertaining:

(1) Acceptability of reliability levels

(2) Economic viability and appeal of achieved reliability

If these criteria aren’t met, appropriate actions must be taken to tackle these challenges, which
will be further elaborated below. In outlining and creating the PEPS, it’s vital to ensure system’s
effective functioning, such as maintaining a Low Occurrence of Loss of Energy (LOE) that is below
a specified threshold, like 7.5 hours per year. This assessment of LOE will be conducted by consid-
ering the chosen equipment and their individual reliability and failure rates. If LOE surpasses the
target level, two approaches are available to maintain it.

● The primary objective is to enhance vulnerabilities within the system. The proposed assessment
tool, which relies on a model-based approach incorporating the principles of failure mecha-
nisms, assists in identifying the least resilient subsystems, their most fragile components, and
corresponding failure modes. Consequently, planners of power systems can opt for subsystems
with superior dependability, such as converters, to guarantee the reliability of PEPS. Moreover,
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
25.4 Converter Reliability Modeling 489

developers of power converters can create converters tailored to specific applications, while
upholding the reliability standards sought by PEPS planners. In this scenario, manufacturers
can personalize their products, enabling them to formulate more economical and efficient
converters. Collaboration between power system planners and power converter manufacturers
has potential to enhance overall efficiency of the supply chain.
● Next approach involves maintaining system according to the initial plan and getting ready to
replace converters at suitable intervals. Once more, since this recommended technique relies on
models, PEPS planner can identify least resilient components in the system and their impact on
power system changes over time. Consequently, planner can strategize for having an adequate
number of backup units during appropriate time frames, ensuring swift replacement of these
units to uphold overall reliability of the PEPS in alignment with intended performance targets.

Significantly, initial choice is notably more appropriate for systems with a restricted period
of operation, like a vessel, perhaps around 20 years. Nevertheless, all elements must be regu-
larly upgraded since the electrical grid is anticipated to run indefinitely. Consequently, second
alternative is better suited for distribution power systems like the electric network. Establishing
cost-effective replacement schedules and designing converters for that duration is imperative.
Moreover, during the operation, it is imperative to assess the system’s reliability due to many
uncertainties that can impact its performance, such as unexpected breakdowns and aging of indi-
vidual components. Consequently, an annual evaluation of LOLE will be conducted based on the
reliability model and past system performance. If LOLE exceeds the acceptable limit, a suitable
maintenance approach must be implemented to maintain the system’s efficiency. Maintenance
plans can be categorized into two primary types: corrective and preventive maintenance. Correc-
tive maintenance will be applied in instances where impact of a converter on system performance
is minimal. In such cases, converter will be replaced as soon as it malfunctions, following this
approach. Conversely, if failure of the converter significantly diminishes overall system perfor-
mance; specifically the LOLE, preventive maintenance techniques must be employed.
Various preventive maintenance methods are available, including calendar-based, age-based,
and condition-based approaches. These methods may entail differing maintenance costs, contin-
gent on factors like the system’s size, the characteristics of failures, the role of the converter in Elec-
tronic Power Systems (PEPS), and more. Depending on the extent of system impairment resulting
from unit failure, one of these maintenance protocols will be chosen to replace or repair converter
before any failure occurs.

25.4 Converter Reliability Modeling

The converter’s failure patterns align with failure zone depicted by bathtub curve, similar to other
systems. These patterns encompass three distinct phases: an initial stage of early failures, a period
of operational effectiveness, and a later phase marked by wear-related issues. Issues during the
initial phase often stem from debugging and production processes. Consequently, converter is sus-
ceptible to random failures and wear-related deterioration during its operational lifespan. External
factors like overcurrent and overvoltage commonly trigger these random failure events (Yang et al.
2011). Consequently, within bathtub curve, failures follow an exponential distribution throughout
the device’s usable lifespan. Usually, the suitable rate of loss is estimated by considering historical
reliability information and operational knowledge. Moreover, wear-out failure rate is connected to
deterioration of device materials as they age, and this is depicted by Stress-Strength-Aging (SSA)
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
490 25 Reliability Assessment of Power Electronics Devices and Systems for Modern Power Applications

model for components prone to aging:

( ) ( )
−0.3
𝛼 𝛽 t
Nf = A ⋅ ΔTj ⋅ exp ⋅ on (25.3)
Tj 1.5
where ΔT and T denote the swing in junction temperature and its mean value, while ton rep-
resents the rise time of the temperature cycle. A, 𝛼, and 𝛽 are constants obtained through aging
experiments. It’s important to note that the junction temperature is influenced by the thermal dissi-
pation of the device, its thermal impedance, and the surrounding temperature. Moreover, thermal
dissipation is split into conduction loss and switching loss. The conduction loss is linked to the
device’s current, voltage, and ON-state resistance. Additionally, the switching loss is impacted by
the switching frequency and ON-OFF energy loss. To assess the device’s lifespan, the yearly mis-
sion profile needs to be converted into thermal parameters in Eqs. (25.1) and (25.4), specifically
temperature and voltage, using electrothermal analysis.
Consequently, a comprehensive electrical and thermal representation of converter elements must
be employed, requiring time-domain examination. The approach for predicting reliability is bifur-
cated into two phases. The initial phase is accountable for electrothermal simulation, while the
subsequent step pertains to reliability simulation. The primary stage is further segmented into two
domains: electrical and thermal. Within these domains, yearly mission profile (potentially encom-
passing factors like solar irradiance, wind speed, and ambient temperature) is transformed into
electrical parameters such as device voltage and current patterns and subsequently into thermal
parameters like junction and hot-spot temperatures.
This procedure necessitates electrical modeling that takes into account evolving frequency char-
acteristics of switching, which can vary from a few kilohertz to several tens of kilohertz, depending
on the converter’s capacity, across the entire mission profile that covers various power levels. Fur-
thermore, the durability of power components is affected by changes in junction temperature,
which is controlled by alterations in the converter’s power load and the thermal impedance of the
device.
Thermal impedance is commonly represented using a higher-order transfer function with three
to five orders. To accurately translate variations in the mission profile into fluctuations in junc-
tion temperature, which notably impact operational lifespan of switch as outlined in Eq. (25.2), it
becomes essential to carry out a comprehensive electrothermal mapping. The alterations in temper-
ature will be determined by solving differential equation of a thermal model designed for extended
mission profiles. This endeavor will introduce greater intricacy and computational load. Evaluating
the thermo-mechanical degradation of devices necessitates converting the mission profile into ther-
mal parameters to ascertain longevity of converter. The accrued damage to capacitors, referred to as
accumulated capacitor damage (ADCap), for a specific mission profile is computed subsequently:
∑ tk
ADCap = (25.4)
k
Lo-k
In a context where tk represents the operating period of a capacitor within parameters (Vo , To ),
and Lo-k is the resultant lifetime derived; process continues by categorizing the junction temper-
ature profile to calculate accumulated damage in semiconductor devices (ADS). This involves
defining different classes for temperature profile, with class “h” characterized by a specific set
of variables (Th , tonh , ncycle-h ). Subsequently, the accumulated damages (ADs) are computed
according to following procedure:
∑ ncycle-h
ADS = (25.5)
h
Nf-h
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
25.4 Converter Reliability Modeling 491

In Eqs. (25.4) and (25.5), device’s longevity corresponds to the inverse of AD, where ncycle-h rep-
resents the cycle count in category h, and Nf-h signifies the cycle count in category h. These AD
values, derived from individual devices, are combined with distinct values from the lifespan model
alongside temperature models specific to each device. Practical implementation involves account-
ing for diverse fluctuations in lifetime models and device thermal characteristics. Consequently,
it becomes essential to establish the distribution of AD considering uncertainties stemming from
the model and manufacturing variations. Monte Carlo simulations offer a means to ascertain the
distribution function of the annual downtime (AD). It should be recognized that employing Monte
Carlo simulations for dependability modeling can be an uphill task. The function representing the
unreliability of device, denoted as F(t), is essentially the inverse of the AD distribution function.
Once unreliability function has been derived, it becomes possible to estimate failure rate due to
wear-out of components through the following procedure:
1 d
𝜆w-X (t) = F(t) (25.6)
1 − F(t) dt
If we denote the differential operator as d/dt and represent the wear-out failure of device X as
𝜆w-X , it indicates the decline in device X’s functionality over time. Degradation in performance is
often described by a Weibull distribution function, which incorporates a hazard function denoted
as h(t). This structure is illustrated below:

h(t) = 𝜆w-X (t) + 𝜆c-X (25.7)

𝜆c-X denotes a consistent failure rate that can be expected based on historical failure data and
operational observations over the valid lifetime.
This chapter has examined reliability within traditional power systems by elucidating its core
components. Subsequently, it has depicted the configuration of contemporary power systems,
encompassing emerging technologies like intelligent microgrids, renewable sources, and energy
reservoir systems. The difficulties confronting modern power systems have been acknowledged
and discussed. Lastly, a novel structure for evaluating reliability in modern power systems has
been introduced. Effective and secure management of modern power systems necessitates novel
regulations for intelligent microgrids and the establishment of consistent standards for gauging
reliability in cyber-physical distributed power systems stated in Figure 25.5.

Challenges in future
power system

Micro-grid operation Renewable energies Power electronics

induced challenges proliferation challenges induced challenges

Design of distribution Adequacy issues Converter interaction

systems Security issues issues
Operational planning Reliability issues
Control and operation Protection issues
Cyber-security

Figure 25.5 Challenges in upcoming power systems reliant on power electronics.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
492 25 Reliability Assessment of Power Electronics Devices and Systems for Modern Power Applications

25.5 Conclusion and Future Challenges

Furthermore, conducting reliability analyses across various tiers demands efficient tools and
strategies for the planning and operation of cyber-physical power systems containing a multitude
of components and elements, particularly in light of the variable failure rates associated with novel
technologies. Consequently, decentralized, self-organizing methods can facilitate optimal and
dependable planning and operation of forthcoming intelligent grids. In the future, research will
be directed toward appraising the reliability of contemporary power systems using the proposed
framework while accounting for diverse aspects of reliability.

References

Alhmoud, L. (2018). Reliability improvement for a high-power IGBT in wind energy applications. IEEE
Transactions on Industrial Electronics 65(9): 7129–7137.
Carroll, J., McDonald, A., and McMillan, D. (2014). Reliability comparison of DFIG drive train
configuration with PMG drive train configuration in the first 5 years of operation. In: 3rd Renewable
Power Generation Conference (RPG 2014), 1–6. IET.
Enslin, J.H., Whisenant, S.G., and Hadidi, R. (2019). Third eGrid workshop maps the grid of the future:
attendees engage to examine the role of power electronic applications in modern electric power
systems. IEEE Power Electronics Magazine 6(1): 48–55.
Fischer, K., Pelka, K., Bartschat, A. et al. (2019). Reliability of power converters in wind turbines:
exploratory analysis of failure and operating data from a worldwide turbine fleet. IEEE Transactions
on Power Electronics 34(7): 6332–6344.
Jiang, J., Peyghami, S., Coates, C., and Blaabjerg, F. (2021). A decentralized reliability-enhanced power
sharing strategy for PV-based microgrids. IEEE Transactions on Power Electronics 36(6): 7281–7293.
Kothari, D.P., Nagrath, I.J., and Saket, R.K. (2022). Modern Power System Analysis. New Delhi: Tata
McGraw-Hill Publishing Company.
Maurya, J. and Saket, R.K. (2023). Performance analysis of single-stage and two-stage VSI-fed IM drive
for solar pump irrigation systems. In: 2023 IEEE IAS Global Conference on Renewable Energy and
Hydrogen Technologies (GlobConHT), 1–6. IEEE.
Peyghami, S., Davari, P., and Blaabjerg, F. (2019). System-level reliability-oriented power sharing
strategy for DC power systems. IEEE Transactions on Industry Applications 55(5): 4865–4875.
Peyghami, S., Wang, Z., and Blaabjerg, F. (2020). A guideline for reliability prediction in power
electronic converters. IEEE Transactions on Power Electronics 35(10): 10958–10968.
Peyghami, S., Blaabjerg, F., and Palensky, P. (2021). Incorporating power electronic converters
reliability into modern power system reliability analysis. IEEE Journal of Emerging and Selected
Topics in Power Electronics 9(2): 1668–1681.
Singh, C., Jirutitijaroen, P., and Mitra, J. (2019). Introduction to Reliability, 1–9. Wiley-IEEE Press.
Yang, S., Bryant, A., Mawby, P. et al. (2011). An industry-based survey of reliability in power electronic
converters. IEEE Transactions on Industry Applications 47(3): 1441–1451.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
493

26

Reliability Aspects in the Design and Development of Microgrids

Amit Kumar 1 , Sachin Kumar 2 , Almoataz Y. Abdelaziz 3 , R. K. Saket 4 , and D. P. Kothari 5
1 Flameproof and Equipment Safety Department, CSIR-Central Institute of Mining and Fuel Research, Dhanbad, Jharkhand, India
2
Department of Electrical Engineering, G.B. Pant Institute of Engineering and Technology, Pauri-Garhwal, Uttarakhand, India
3
Faculty of Engineering & Technology, Future University in Egypt, Cairo, Egypt
4
Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
5 Department of Electrical Engineering, Indian Institute of Technology, Delhi, India

26.1 Introduction
The electric power system is the most extensively manufactured system that significantly improves
humankind’s quality of life. A power system consists of power generation, distribution, and utiliza-
tion. The high-voltage AC transmission system is famous for transmitting power from the genera-
tion end to the distribution area because of low transmission losses (Kothari et al. 2022). Due to the
increasing energy demand, modern grids have certain constraints, like continuous power supply,
quality of power, and reliability. Due to specific reasons like energy shortages and environmen-
tal effects, fossil fuels are becoming unpopular in the modern power system. Renewable energy
sources are being integrated into the grid swiftly (Shahgholian 2020).
Microgrids are becoming popular because they operate independently in an islanded mode with-
out external power grid interference. The microgrid concept provides a solution for demand and
location problems. Local generation, utilization, and renewable energy integration make the micro-
grid popular. Using integrated renewable energy generation and storage technologies, microgrids
are small-scale energy systems that can deliver enough electricity to meet local demand (Kumar
et al. 2021).
Microgrids may be designed as AC or DC according to their distribution system. Power electron-
ics converters are an intrinsic part of microgrid technology since efficient power conversion and
control are necessary at various points. Due to the increased use of power electronics converters,
control and utilization flexibility have increased, but reliability has become a more significant con-
cern. Components of the converters, like semiconductor switches, are prone to failure, which makes
the whole system highly vulnerable. The chapter covers the microgrid’s reliability, considering its
technology in detail.
An introduction that includes a literature survey has already been described in Section 26.1.
The rest of the chapter has been organized as follows: A detailed description of the architecture
and operation of the microgrid is explained in Section 26.2. Section 26.3 discusses microgrid con-
trol, which includes multiple modes to ensure stable and secure operations. Reliability Aspects of
microgrid planning and design, which include reliability indices, optimization, layer-wise reliabil-
ity, etc., are discussed in Section 26.4. Finally, the conclusion, challenges, and future directions are
discussed in Section 26.5.
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
494 26 Reliability Aspects in the Design and Development of Microgrids

26.2 Architecture and Operation of Microgrid

Microgrid systems have several distributed energy resources and use a low-voltage distribution.
Microgrids may or may not be connected to any external grid. A microgrid mainly consists of dis-
tributed generators, variable loads, energy storage devices, control systems, and points of standard
coupling (PCC). A typical microgrid is shown in Figure 26.1, and most of the components are
denoted.

26.2.1 Types of Microgrid

Microgrid classification is based on various design parameters such as size, application, operational
distribution system, architecture, distribution configuration, scenario, and type of sources. Micro-
grids are broadly classified into three categories based on their operation mode.
1. grid connected,
2. transited, and
3. standalone.

AC
DC

PV plant
AC loads

AC
DC

Energy storage
devices

DFIG

Wind energry

AC bus

Figure 26.1 Architecture of a microgrid.

Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
495
26.2 Architecture and Operation of Microgrid

Different classifications of microgrid. (a) DC microgrid and (b) AC microgrid.

AC loads
AC loads
AC
AC

AC

DC
DC

DC
AC

DC bus
(a)
DC

AC bus
(b)
DC

DC
DC
DC

AC
DC

DFIG
AC

AC
Energy storage

DC

DC
Wind energy
devices
PV plant

Wind energy
Energy storage

Figure 26.2
PV plant

devices
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
496 26 Reliability Aspects in the Design and Development of Microgrids

In the standalone microgrid, the reactive power balance shall be maintained independently due
to the absence of an infinite bus (Kaushal and Basak 2019). The challenges for standalone micro-
grids are as follows.
1. Stable magnitude and frequency of the voltage
2. Stable power balance
In grid-connected microgrids, the control of microgrids becomes much more complex since the
grid is connected to the microgrid through a static transfer switch. Based on the size of the micro-
grid, it is classified as small scale (<10 kW), medium scale (10 kW–1 MW), and large scale (>1 MW).
Microgrids are also classified as AC and DC microgrids based on their distribution systems. AC
and DC microgrids are based on their distribution systems (Shahparasti et al. 2015). Both AC and
DC have their own merits and demerits. These microgrids are designed to fulfill local demands
according to their size. Similarly, suitability, stability, and reliability are considered on every basis;
different classifications are done, as shown in Figure 26.2a, and 26.2b.

26.2.2 Microgrid Technology

As previously discussed, the microgrid system comprises generation units, converters, energy stor-
age systems, control stations, and interconnected switches. Technology supports all of these micro-
grid components. As day-by-day research is going on in this area, the latest technologies in every
section are used to improve the grid’s efficiency, reliability, and safety. The distributed generation
unit technology utilizes solar, wind, and fuel cells. In modern-day microgrids, fuel gas with com-
bined heat and power technology as a microturbine helps improve efficiency significantly.
Along with batteries, supercapacitors and flywheels are also used in the microgrid. Due to this
new energy storage system, several benefits are observed, such as power stabilization (less fluctu-
ations), reduced load spikes, and enhanced reliability. Similarly, microgrid switching technology
uses a digital signal processor to improve the response time. Due to advancements in switches,
high-frequency switches are being used. Microgrid control technologies can be grouped into island-
ing mode and network mode. Control technologies introduced in power converters regulate active
and reactive power supply and frequency control.

26.3 Microgrid Control Strategies

From the consumer’s perspective, a microgrid is a grid system that provides reliable, indepen-
dent, and high-quality electric power (Hosseini and Shahgholian 2017). According to Hu et al.
(2015), it is challenging to coordinate various micropower grid types to create a stable frequency
and voltage-controlling microgrid system. Its primary goals are to control active and reactive power
separately, fix voltage sag and system imbalances, and meet the grid’s load dynamics needs. Micro-
grids have a control system unlike traditional power systems for the following reasons:
1. Frequent unbalanced loading due to one phase.
2. Energy storage plays a huge role, for which a control mechanism is required.
3. Different static and dynamic characteristics than other conventional grid.
Power systems need the right control strategies to ensure proper operation. Microsource and load
controllers, a central controller for the microgrid system, and a distribution management system
comprise the microgrid control.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
26.3 Microgrid Control Strategies 497

Controller

System

Unit 1 Unit 2 Unit 3 Unit 4

Power
connection

Figure 26.3 Schematic of centralized control of microgrid.

1. Centralized Control: As shown in Figure 26.3, in this type of control strategy, the central con-
trol unit determines the output of each local control unit based on the input production capacity
of each unit. Various parameters, such as energy demand, market price, and production, deter-
mine the control strategy.
2. Decentralized Control: In a decentralized control strategy, each local control determines the
operation command to fulfill the energy production demand, as shown in Figure 26.4. This type
of control strategy focuses on overall performance rather than financial income.
3. Multi-agent System: Multi-agent system (MAS) is a development of the traditional distribu-
tion control strategy that can be used to manage very large and complex entities. The primary
characteristic of MAS is its capacity to integrate intelligence components into each local control.
Both centralized and decentralized systems need trustworthy data communication infrastruc-
ture. A data communication network carrier can be a power line, radio, or telephone. As shown in
Figure 26.5, these communication networks connect the main control system or applications for
the energy management system to each distributed energy unit. They also run the microgrid.

26.3.1 Advantages and Disadvantages of Microgrid

Despite the advancements in microgrid technology, its implementation is a complicated process.
Poor understanding and unsuitable government policies are the main obstacles to applying
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
498 26 Reliability Aspects in the Design and Development of Microgrids

Controller 1 Controller 2 Controller 3 Controller 4

System

Unit 1 Unit 2 Unit 3 Unit 4

Power
connection

Figure 26.4 Schematic of decentralized control of microgrid.

Microgrid

MCC

LC
LC LC
LC

Local
Grid level generation
DNO MO agent

Renewable Storage
DG energy
management
source
Microgrid Microgrid agent agent agent

MCC

Storage
Filed WT
Storage agent 2
agent PV
level agent agent 1

Figure 26.5 Schematic of multi-agent control of microgrid.

 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
26.4 Reliability Aspects in Microgrid Planning and Design 499

Table 26.1 Advantages and disadvantages of microgrid

Microgrid advantages Microgrid disadvantages

1. Microgrids benefit from operating separately 1. Energy storage units require significant space
or with the outside grid and regular maintenance
2. During peak load periods, energy may be 2. Resynchronization with the outer grid is a
supplied using the generation or energy storage difficult job
units to provide uninterrupted power 3. Maintaining the power quality up to the
3. Microgrids with more renewable generation applicable standard requires complete control
units provide low emissions over the grid
4. The problem of safety in microgrids

microgrids. Microgrid has drawbacks in addition to its many benefits, and the only way to
overcome them is through ongoing research into the problems. Table 26.1 covers the advantages
and disadvantages of microgrids.

26.4 Reliability Aspects in Microgrid Planning and Design

Reliability is a critical aspect in the planning and design of microgrids. Microgrids are localized
energy systems operating independently or in conjunction with the main grid. Ensuring their reli-
ability is essential to providing uninterrupted power to the connected loads. Here are some key
reliability aspects to consider in microgrid planning and design:
1. Redundancy and Resilience: Microgrids should be designed with redundancy to ensure they
can continue operating despite equipment failures or external disruptions. Redundant compo-
nents such as generators, energy storage systems, and inverters can help maintain reliability.
The microgrid should also be designed to be resilient against extreme weather events, cyberat-
tacks, and other unforeseen circumstances.
2. Component Selection: The choice of components, such as generators, energy storage sys-
tems, and renewable energy sources, plays a crucial role in microgrid reliability. High-quality
and reliable equipment should be selected to minimize the chances of failures (Kumar et al.
2020).
3. Control and Management Systems: Robust control and management systems are essen-
tial for reliability. These systems should monitor the state of the microgrid, predict potential
issues, and take corrective actions. Automation and advanced control algorithms can optimize
microgrid performance and enhance reliability.
4. Black Start Capability: A microgrid should be able to “black start,” meaning it can restart
and re-establish its operations independently in case of a complete grid outage. This often
involves having dedicated generators or energy storage systems that can initiate the microgrid’s
recovery.
5. Islanding Capability: Microgrids should be designed to transition between grid-connected
and islanded modes seamlessly. The islanding capability allows the microgrid to operate
autonomously when the main grid is down, ensuring power continuity.
6. Predictive Maintenance: Implementing predictive maintenance strategies can help iden-
tify potential issues in advance and schedule maintenance activities to prevent unplanned
downtime.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
500 26 Reliability Aspects in the Design and Development of Microgrids

7. Cybersecurity: Microgrids are vulnerable to cyberattacks, which can compromise their reli-
ability. Robust cybersecurity measures, including firewalls, intrusion detection systems, and
regular security assessments, should be part of the microgrid design.
8. Energy Storage: Energy storage systems can enhance reliability by providing backup power
during grid outages and smoothing the integration of intermittent renewable resources.
9. Load Management: Efficient load management strategies can help balance supply and
demand within the microgrid, ensuring a reliable power supply to critical loads.
10. Grid Interaction: Microgrids can be designed to interact with the main grid to import or
export power as needed. This interaction can enhance reliability by accessing additional
resources during peak demand or emergencies.
11. Testing and Simulation: Before deployment, microgrids should undergo rigorous testing and
simulation to validate their reliability in various scenarios.
12. Training and Maintenance: Ensuring that personnel are adequately trained in operating and
maintaining the microgrid is essential for long-term reliability.

Reliability in microgrid planning and design is a multifaceted challenge that requires a holistic
approach. By addressing these aspects, microgrid operators can provide a stable and continuous
power supply to meet critical infrastructure needs and enhance energy resilience.

26.4.1 Reliability Evaluation

While designing the microgrid, it is necessary to include economic generation, utilization, and
adequate reliability. Reliability evaluation of the designed microgrid helps determine the factors
responsible for the power interruption or low power quality. In IEEE Standard 2030.9-2019
(Hosseini and Shahgholian 2019), reliability indices are elaborated, as shown in the flow chart
in Figure 26.6. Load point and system indices are discussed in IEEE Standard 2030.9-2019. Load
point indices indicate the service continuity of an individual load. On the other hand, system
indices indicate the service continuity of a group of loads (Allan 2013).

1. Load Point Indices

The average failure rate denotes the interruption rate (number of interruptions per year)
∑
n
𝜆s = 𝜆i (26.1)
i=1

𝜆i = average failure rate at ith load point.

Average Annual Outage Time (Us ): The average annual outage time denotes the overall
downtime in one year. It is denoted in hours. This index represents the unavailability of the
supply.
∑
n
Us = 𝜆i ∗ r i (26.2)
i=1

ri = duration of interruption per failure at ith load point.

Average Outage Time per Failure rs : The average outage time per failure denotes the average
duration of interruption. It is generally denoted in hour per interruption.
∑n
U i=1 𝜆i ∗ ri
rs = s = ∑ (26.3)
𝜆s n
i=1 𝜆i
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
26.4 Reliability Aspects in Microgrid Planning and Design 501

Start

Design and planning

objectives

Electrical system Automation system

design design

Both
objectives
fulfilled

Selection, evaluation
and comparison of
scheme

Selected scheme is
proposed

End

Figure 26.6 Microgrid planning procedure as per IEEE 2030.9-2019. Source: Adapted from (Hosseini and
Shahgholian 2019).

2. System Indices
System Average Interruption Frequency Index (SAIFI): SAIFI is the ratio of the total num-
ber of interrupted customers to the total number of served customers
∑n
i=1𝜆i ∗ N i
SAIFI = (26.4)
NT

where
NT = Number of customers served
Ni = number of customers interrupted for ith interruption.
Average Service Availability Index (ASAI): ASAI is the ratio of actual served hours (hours
for which service was available) to the total demanded customer service hours.
∑n
N − i=1 Ci ∗ Li
ASAI = (26.5)
N
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
502 26 Reliability Aspects in the Design and Development of Microgrids

where
Ci = Available generation capacity
Li = Maximum load of the ith day
N = Number of days.
System Average Interruption Duration Index (SAIDI): SAIDI is the ratio of the total dura-
tion of customer interruption to the total number of customers for which service was available.
∑n
r i ∗ Ni
SAIDI = i=1 (26.6)
NT
where
ri = Restoration time for ith interruption.
Loss of Load Probability (LOLP): LOLP represents the probability that the load demand
exceeds the generating capacity.
∑n
P ∗ (Ci ∗ Li )
LOLP = i=1 (26.7)
N
Analytical methods like Markov’s and Monte Carlo simulations are used to calculate the sys-
tem’s reliability using these indices. The flowchart shown in Figure 26.6 describes the procedure
of microgrid planning and design (IEEE 2019).

26.4.2 Energy Scheduling, Forecasting, and Optimization Techniques

A microgrid’s short-term planning and energy management are easily implemented due to dis-
tributed energy resources. Since it is easier to predict loads in large conventional power systems
than in small-load microgrids, this unpredictability is the main reason for implementing a proper
energy management and scheduling scheme to supply uninterrupted power. This scheduling time
scale may vary from minutes to hours on several-day scales. Optimization techniques are used in
scheduling with the help of parameters like energy unit capacity, minimized cost, and environ-
mental effects. High reliability can be achieved with these parameters and a proper optimization
method. For good optimization, adequate sizing of distributed energy units is required (Yuan et al.
2017; Swaminathan et al. 2020).
Corrective scheduling is a crucial method of energy management in microgrids. The concept mit-
igates errors due to uncertainty by adjusting the scheduling based on new information about load
demand and environmental conditions. Optimization techniques are selected based on the com-
plexity of the grid. Various optimization techniques used are linear programming, convex optimiza-
tion, Genetic algorithms, particle swarm optimization techniques, etc. A similar study provided a
unified framework for optimal and reliable microgrid design with the help of two different opti-
mization techniques, integer linear programming, and particle swarm technique (Machado et al.
2018). A practical planning procedure is provided for microgrids on the robust optimization method
within the operational constraints, and the total cost of planning is minimized (Khodaei 2014).

26.4.3 Reliability Aspects of Power Electronics in Microgrids

Determination of failure rate plays an essential role in evaluating the reliability of the power elec-
tronics converters. Failure rate follows a bathtub curve with time, which includes three stages:
infant mortality, helpful life, and wear-out phase. Since the installed power electronics converter
passes its infant mortality stage, the failure rate during the useful life and wear-out phase is con-
sidered for reliability evaluation.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
26.4 Reliability Aspects in Microgrid Planning and Design 503

λPETotal = λuseful + λwear out

λPETotal

Failure
rate
λwear out

λuseful

Figure 26.7 Failure rate versus time curve.

The failure rate during useful life is nearly constant and significantly increases during the
wear-out period due to the aging effect of components. Both the failure rates are considered for
accurate and precise conversion reliability. However, due to the uncertainty of wear-out, most of
the study ignores the wear-out failure rate and relies on the constant failure rate only, as shown
in Figure 26.7. Analysis of field failure data of similar components provided by supply gives an
accurate failure rate.
Reliability handbooks provide failure rates based on historical field data. Some popular reliability
handbooks are military handbook (MIL-HDBK-217), Handbook of Reliability Data (HRD5), Inter-
national Electro-technical Commission standards IEC-61709, and IEC-62380-TR. These handbooks
provide an empirical model. The main disadvantage of these handbooks is that they are based on
historical data on the technologies that have yet to be practiced.
To determine the Constant failure rate (constant), FIDES reliability prediction method is used
due to its higher accuracy. This method uses the component’s sensitivity concerning stress for a
longer period. The input mission profiles are disintegrated into different phases. Since these phases
represent different stress parameters. Physical failure rates are determined after this decomposition
of mission profiles into phases. With the help of this physical failure rate in each phase, constant
failure rates are determined (Carton et al. 2017).
Modeling based on the physics of failure and stress analysis is used to obtain the wear-out
failure rate. For Power electronics converters, electrolytic capacitors and switches are considered
critical components. Both components are prone to failure due to thermal and electrical stress.
Electro-thermal modeling is developed to obtain the stress profile of these too-critical components
from the mission profile. With the help of loading decomposition, the impact of stress level can
be evaluated. Lifetime consumption estimation is done by plotting conception with time. This
lifetime consumption concerning time gives a corresponding distribution function. A probabilistic
assessment uses this probability distribution function to determine the wear-out failure rate.

26.4.4 Component to System: Layer-Wise Reliability

Reliability is calculated at different levels, such as system, subsystem, and component levels, as
shown in Figure 26.8. The component level reliability helps determine the reliability of the system.
In the same way, while evaluating the reliability of microgrids’ reliability, the smallest components’
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
504 26 Reliability Aspects in the Design and Development of Microgrids

Utility
grid

Breaker

DC AC System
AC DC level

DFIG

AC loads

Microgrid

Sub-system
AC/DC level

DC/AC

Component
level

Figure 26.8 Layer-wise reliability evaluation of microgrid.

reliability must be known. Microgrid reliability can be evaluated in three levels: component level,
converter level, and microgrid level (Peyghami et al. 2020). The first level includes switches, capac-
itors, resistors, inductors, etc. The physics of failure is studied in this layer, and possible failure
reasons are found. The expected lifetime determination of components in this layer helps in eval-
uating reliability. In the second level, the reliability of converters is evaluated with the help of
component reliability. The healthiness of this layer shows that healthy inverters are available to
use. In the third level, overall microgrid reliability is evaluated with the help of the other two layers
and their corresponding components’ connection. The topology and control methodology become
essential in this layer since their interconnection differs.

26.5 Conclusion and Future Challenges

In the long-term planning of microgrids, adopting new methods for forecasting and scheduling
energy demand is necessary. Distributed energy resource sizing with suitable optimization
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 505

techniques can be used to develop new standards for future microgrid design. Incorporating
different stressors in reliability studies will enhance the reliability-oriented design of microgrids.
A better optimization technique can help to meet the cost minimization and climate requirements.
Transformation of existing microgrids into intelligent, robust, reliable, and safe microgrids is a big
challenge.
Microgrid is seen as an alternative to conventional grids, which is becoming increasingly popular.
In this chapter, microgrid type, control techniques, and the importance of reliability are discussed,
and some popular reliability evaluation methods are provided. Since uninterrupted and long dura-
tion operation of grid is a demand of the time. It is vital to develop more reliable methods and
techniques for energy scheduling. Proper inclusion of external stress in the planning and study is
required.

References

Allan, R.N. (2013). Reliability Evaluation of Power Systems. Springer Science & Business Media.
Carton, P., Giraudeau, M., and Davenel, F. (2017). New fides models for emerging technologies. In:
2017 Annual Reliability and Maintainability Symposium (RAMS), 1–6. IEEE.
Hosseini, E. and Shahgholian, G. (2017). Partial-or full-power production in WECS: a survey of control
and structural strategies. EPE Journal 27 (3): 125–142.
Hosseini, E. and Shahgholian, G. (2019). Recommended practice for the planning and design of the
microgrid. In: IEEE Std 2030. 9-2019. BOG/SC21 - Distributed Generation, Energy Storage and
Interoperability Standards Committee, 1–6. IEEE.
Hu, J., Zhang, T., Du, S., and Zhao, Y. (2015). An overview on analysis and control of micro-grid
system. International Journal of Control and Automation 8 (6): 65–76.
Kaushal, J. and Basak, P. (2019). A decision making methodology to assess power quality monitoring
index of an AC microgrid using fuzzy inference systems. Electric Power Components and Systems 47
(14-15): 1349–1361.
Khodaei, A. (2014). Provisional microgrids. IEEE Transactions on Smart Grid 6 (3): 1107–1115.
Kothari, D.P., Nagrath, I.J., and Saket, R.K. (2022). Modern Power System Analysis. Tata McGraw-Hill.
Kumar, S., Saket, R.K., Dheer, D.K. et al. (2020). Reliability enhancement of electrical power system
including impacts of renewable energy sources: a comprehensive review. IET Generation,
Transmission & Distribution 14 (10): 1799–1815.
Kumar, S., Sarita, K., Saket, R.K. et al. (2021). Reliability assessment for DFIG-based WECS
considering the impact of 3-phase fault and lightning impulse voltage. International Transactions on
Electrical Energy Systems 31 (8): e12952.
Machado, P., de Souza, L.E., and Netto, R.S. (2018). Framework to support grid-connected microgrid
optimal planning. In: 2018 IEEE International Conference on Environment and Electrical Engineering
and 2018 IEEE Industrial and Commercial Power Systems Europe (EEEIC/I&CPS Europe), 1–6. IEEE.
Peyghami, S., Blaabjerg, F., and Palensky, P. (2020). Incorporating power electronic converters
reliability into modern power system reliability analysis. IEEE Journal of Emerging and Selected
Topics in Power Electronics 9 (2): 1668–1681.
Shahgholian, G. (2020). Coordinated design of power system stabilizer and variable impedance devices
to increase damping of inter-area modes using genetic algorithm. Nashriyyah-i Muhandisi-i Barq va
Muhandisi-i Kampyutar-i Iran 75 (4): 271.
Shahparasti, M., Mohamadian, M., Baboli, P.T., and Yazdianp, A. (2015). Toward power quality
management in hybrid AC–DC microgrid using LTC-L utility interactive inverter: load voltage–grid
current tradeoff. IEEE Transactions on Smart Grid 8 (2): 857–867.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
506 26 Reliability Aspects in the Design and Development of Microgrids

Swaminathan, S., Pavlak, G.S., and Freihaut, J. (2020). Sizing and dispatch of an islanded microgrid
with energy flexible buildings. Applied Energy 276: 115355.
Yuan, C., Liu, G., Wang, Z. et al. (2017). Economic power capacity design of distributed energy
resources for reliable community microgrids. Energy Procedia 142: 2561–2567.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
507

Abbreviations

2TBN Two-slice temporal Bayesian network

3LT2 C Three-level T-type converter
ABAO As-bad-as-old
AC Alternating current
ac Air cooling with extended fin
ADS Accumulated damage in semiconductor devices
AELC Average extra length for charging
AENS Average energy not supplied
AETC Average extra time for charging
AGAN As-good-as-new
AIDI Average interruption duration of EV charging index
AIFI Average interruption frequency of EV charging index
ALT Accelerated life testing
AM Asset management
AMI Advanced metering infrastructure
ANN Artificial neural network
APM Application of probability methods
ARA Arithmetic reduction of age
ARI Arithmetic reduction of intensity
ASAI Average service availability index
ASAI Average system availability index
ASUI Average system unavailability index
ATHEANA A technique for human event analysis
BITE Built-in test equipment
BJT Bipolar junction transistors
BLDC Brushless DC motor
BMS Battery management systems
BN Bayesian network
BTO Break-through technologies
CAIDI Customer average interruption duration index
CBM Condition based maintenance
CC Capital costs
CCS Combined charging system
CDF Cumulative distribution function
CDS Composite distribution system
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
508 Abbreviations

CFR Constant failure rate

CHB Cascaded H-bridge
CHP Combined heat and power
CIC Customer interruption cost
CIL Critical items list
CLOE Complete loss of excitation
CM Corrective maintenance
cm Centimeter
CMF Common mode failure
COS Change-over switch
CPT Conditional probability table
CREAM Cognitive reliability and error analysis method
CRT Cathode ray tube
CS Charging stations
dB Decibel
DBN Dynamic Bayesian network
DC Direct current
DE Differential evolution
DFIG Doubly fed induction generator
DfR Design for reliability
DG Diesel generator
DG Distributed generation
DIgSILENT Digital simulation and electrical network
DN Distribution network
DOE Design of experiment
DPLVC Daily peak load variation curve
DSP Digital signal processor
DSR Distribution system reliability
E Erying
EENS Expected energy not supplied
EENS Expected energy not served
EM Electromagnetic
EMI Electromagnetic interference
EMS Energy management systems
ENSPI Energy not served per interruption
EPDN Electrical power distribution network
EPNS Expected power not supplied
EPS Electric power system
ESS Energy storage systems
ETA Event tree analysis
EV Electric vehicle
FC Flying capacitor
FC Failure costs
FFTA Fuzzy fault tree analysis
FLISR Fault location, isolation, and service restoration
FMEA Failure modes and effects analysis
FMECA Failure modes effects and criticality analysis
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Abbreviations 509

FORM First-order reliability method

FP Flower pollination
FTA Fault tree analysis
G2V Grid to vehicle
GA(s) Genetic algorithm(s)
GDA Gaussian distribution approach
GRP Generalized renewal process
GTO Gate turn-off thyristors
H, M, and R Three-state model (healthy state, marginal state, and risky state) for reliability
analysis
HAWT Horizontal-axis wind turbines
HL Hierarchical level
HL-1 Hierarchical level 1
HL-2 Hierarchical level 2
HL-3 Hierarchical level 3
HMM Horizon mission methodology
HMM Hidden Markov models
HPP Homogeneous Poisson process
HRD Handbook of reliability data
HRES Hybrid renewable energy systems
HVAC Heating ventilation and air conditioning
HVDC High-voltage direct current
I, O Denotes input and output nodes
ICA International Commission on Accreditation
ICs Integrated circuits
IEC International Electrotechnical Commission
IEEE Institute of Electrical and Electronics Engineers (USA)
IGBT Insulated gate bipolar transistor
IIoT Industrial Internet of Things
IM Induction motor
LCCA Life cycle cost analysis
LCC Life cycle cost
LCM Life cycle management
LDC Load duration curve
LHS Latin hypercube sampling
LOE Loss of energy
LOEE Loss of energy expectation
LOLE Loss of load expectation
LOLF Loss of load frequency
LOLP Loss of load probability
LRU Line replaceable unit
LSAR Logistics support analysis record
LSPWM Level-shifted pulse-width modulation
MAS Multi-agent system
MC Monte Carlo
MC Maintenance costs
MCS Monte Carlo simulation
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
510 Abbreviations

MDT Mean down time

MHT Mean healthy time
MIL-HDBK Military-handbook
MK Markov models
ML Machine learning
MLEs Maximum likelihood estimators
MLI Multilevel inverter
MMC Modular multilevel converter
MMF Magneto-motive force
MMT Mean marginal time
MOSFET Metal oxide semiconductor field effect transistors
MPP Most probable point
MPP Maximum power point
MPPT Maximum power point tracking
MRO Maintain, repair and overhaul
MRT Mean risky time
MTBF Mean time between failures
MTTF Mean time to failure
MTTR Mean time to repair
MUT Mean up time
MVar Mega volt ampere reactive
MW Mega watt
mW Milli watt
NASA National Aeronautics and Space Administration (USA)
NCPI Novel collapse prediction index
NHPP Nonhomogeneous Poisson process
NOCT Normal operating conditions temperature
NPC Neutral-point clamped
OC Operating costs
OLS Ordinary least squares
ORU Orbital replaceable unit
P Peck
pc Passive cooling with heat pipe
PCB Printed circuit board
PDF Probability density function
PDF Probability distribution function
PDU Power distribution unit
PE Power electronic
PEDS Power electronic devices and systems
PEPS Power electronic-based power system
PEVs Plug-in electric vehicles
PF Probability of failure
PI Performance indices
PI Performance indicators
PII Probabilistic insecurity index
PLC Peak load consideration
PLOE Partial loss of excitation
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Abbreviations 511

PM Preventive maintenance
PMSM Permanent magnet synchronous motor
PoF Physics of failure
Pof Probability of failure
PRE Power reliability engineer
PSCAD Power systems computer-aided design
PSF Probabilistic sufficiency factor
PSPWM Phase-shifted pulse-width modulation
PSR Power system reliability
PSS/E Power system simulation for engineering
PV Photovoltaic
RAT Reliability assessment techniques
RBD Reliability block diagram
RBDO Reliability-based design optimization
RBTS Roy Billinton test system
RC Resistor–capacitor
RCD Resistor–capacitor–diode
RCM Reliability-centered maintenance
RDB Reliability block diagrams
RE Renewable energy
RER Renewable energy resources
RES Renewable energy sources
RHPVES Renewable hybrid PV energy system
RNN Recurrent neural network
RP Renewal process
RPN Risk priority number
RPS Reward and penalty system
RSM Response surface method
RTS Reliability test system
RUL Remaining useful life
RV Residual value
SAIDI System average interruption duration index
SAIFI System average interruption frequency index
SBX Simulated binary crossover
SCC Short circuit current
SCR Silicon-controlled rectifiers
SE Shielding effectiveness
SEIG Self-exited induction generator
SFC Safety factor concept
SHE Selective harmonic elimination
SLDC Stepped load duration curve
SOA Safe operating area
SoC State of Charge
SoH State of Health
SORM Second-order reliability method
SoT State of temperature
SPAR-H Standardized plant analysis risk human reliability
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
512 Abbreviations

SPE Solar power Europe

SPWM Sinusoidal pulse-width modulation
SRM Switched reluctance motor
SSA Stress–strength analysis
STC Standard temperature condition
SUT Switch under test
SVM Space vector modulation
TBF Time between failures
TDEP Temperature-dependent electrical parameters
TE Transverse electric
THD Total harmonic distortion
TL Transmission line
TLBO Teaching-learning-based optimization
TM Transverse magnetic
TOU Time-of-use
TTL Total technical life
TVSM Thevenin-based voltage stability margin
UPS Uninterruptible power supply
V2D Vehicle to distribution
V2G Vehicle-to-grid
V2H Vehicle to Home
VAWT Vertical-axis wind turbines
VVO Volt-VAR optimization
“W” and “F” Working (up) and failed (down) state, respectively
W and F Success set of event and fail set of event
WAMC Wide-area monitoring and control
WC Water cooling
WECS Wind energy conversion system
WF wind farms
WS Wick structure
WT Wind turbine
YoY Year-on-Year
ZCS Zero current switching
ZVS Zero voltage switching
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
513

Notations

α Failure mode ratio (for an item, α = 1)

𝛼 Re-installation rate of component
𝛼 Distribution parameter
𝛼I Relative frequency of occurrences
𝛼 K, 𝛽 K Coefficients of cost for rate of failure and time of repair,
respectively
𝛼 vceo Temperature coefficient of V ce
𝛽 Distribution parameter
𝛽 Ratio of common cause failure to total failure rate
𝛽 The likelihood of function or mission loss given specific
conditions
𝛽 Safety index
𝛽 ro Temperature coefficient
Λ(t) Expected number of failures through time
𝛿j Slope of error
𝜉i The right eigenvector corresponding to 𝜆i
𝛾 ge Temperature coefficient of ΔV ge
𝜂 battery Battery efficiency
𝜂c Distribution index
𝜂 inverter Inverter efficiency
𝜆 Failure rate
𝜆 and 𝜇 Single component failure and repair rate, respectively
𝜆base Basic failure rate of components
(capacitors/diodes/switches)
𝜆c Failure rate due to common cause
𝜆C Capacitance failure rates
𝜆c-X Consistent failure rate
𝜆cap Capacitor failure rate
𝜆constant Content failure rate
𝜆C_converter Failure rate in a converter
𝜆C_useful Useful life failure rate of converter
𝜆C_wearout Wear-out failure rate of converter
𝜆D Diode failure rates
𝜆DC−DC Converter DC–DC converters failure rates
𝜆diode Diode failure rate
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
514 Notations

𝜆e and 𝜇 e Equivalent failure and repair rate, respectively

𝜆f Expected failure rate
𝜆Filter Filter failure rate
𝜆i The ith eigenvalue of the Jacobian for load flow
𝜆i Failure rate of ith phase
𝜆i Average failure rate at ith load point
𝜆inverter Inverter failure rates
𝜆j Transmission line failure
𝜆k , r k The kth distributor segment’s rate of failure and average
repair time, correspondingly
𝜆k, min and r k, min Attainable minimums for the kth distributor segment’s
rate of failure and repair time
𝜆k, max and r k, max The kth distributor segment’s maximum permitted rate of
failure and time to repair
𝜆max Maximum load flow eigenvalue Jacobian
𝜆min Minimum eigenvalue of [H] or [J]
𝜆p Item failure or hazard rate
𝜆PETotal Power electronics converters failure rate
𝜆phy_cap Failure rate of capacitor
𝜆phy_scd Failure rate of semiconductor
𝜆R Resistance failure rates
𝜆s Failure rate of series system
𝜆s Switch failure rate
𝜆s Snubber failure rates
𝜆Switch Switch failure rate
𝜆sys Failure rate of the system
𝜆sys, i Failure rate of the system at the ith load point
𝜆r Failure rates of rth component
𝜆useful Failure rate during useful
𝜆W Switch’s failure chance
𝜆w-X Wear-out failure of device X
𝜆wearout Wear-out failure rate
𝜆XY Total transition rate from state X to Y
𝜆+1 and 𝜆+2 Failure rates of components while only one of them is
working for load sharing system
𝜆1 , 𝜆2 Failure rates of each component while both02 are in
operation
𝜆 1 , 𝜆 2 , … , 𝜆n Component failure rates
𝜆0k Base failure rate
𝜇 Repair rate
𝜇Bj Mean of vector B, the jth constraint
𝜇Cl Mean of vector C, the lth constraint
𝜇 i and 𝜆i Repair and failure rates of ith DG unit
𝜇j Repair rates
μr Expected repair rate
𝜔r Rotor speed
𝜔s Synchronous speed
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Notations 515

πCF Capacitance factor

πCCF Contact construction factor
πES Effect of environment stress
πQF Quality factor
πTF Temperature factor
Πi Factor for electrical mechanical and thermal overstress in
ith phase
Πpm Factor for manufacturing quality impact
Πprocess Factor for all processes such as operation and maintenance
𝜎c Standard deviation of capacity
σd Standard deviation of demand
𝜎 x , 𝜇 x , 𝜎 z , and 𝜇 z Pertinent shape parameters for the PDFs
𝜏 Discharge time constant
A Availability
A Hypothetical event A
A Transition intensity matrix
A Area of cylinder bore
A, α, and 𝛽 Curve-fitting constants
Ac Peak amplitude of carrier waveform
Aconverter (t) Availability of the converter
Aconstant (t) Availability associated with time-constant failure rate
ADCap Accumulated capacitor damage
ADCOST Extra money put toward distributed generations (DGs) to
buy power
AH , AM , and AR Availability in H, M, and R states
ai The network’s output for the ith training dataset
aij The ijth load flow element [J ′ ] for Jacobian
Ai Series of exhaustive
Ai and Ai Availability and unavailability of ith DG unit
Aj Long-term availability
Aj Transmission line’s unavailability
AOC Availability of other components of converter
ADC Availability of DC transmission line
Am Peak amplitude of reference waveform
AREC (t) Availability of receiving end converter
ASEC (t) Availability of sending end converter
Aswitch (t) Availability of semiconductor switches
Asys System availability
Asys System unavailability
Atotal (t) Availability of total system
ATi Unavailability of the transmission line
ATi Availability of the transmission line
Awearout (t) Availability associated with time-varying failure rate
B Hypothetical event B
B Event
Bc Susceptance to lines charging
BcT Total susceptance of the line charging
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
516 Notations

bh and by Bias vectors that let every node to learn and offset
C The total number of training data samples
C Generation capacity
C Capacitor value in micro Farad
C Capacitor
C Mean generation capacity
CC Capital costs
CdgUS Capacity available due to distribution substation DG units
ci Indicates the cumulative count of instances the ith EV
engaged in V2G activities during the simulation duration
Ci ith component
Ci Lith load level mean generation capacity
Ci Available capacity in interval i
Ci Available generation capacity
Cm Criticality number
Cnetwork Network’s overall reliability costs
Coss = Cds + Cgd Parasitic capacitance of the MOSFET
Cox Oxide layer capacitance
CRP The utility’s cost of the reward or penalty
Cpk Interruption costs in rupees per kW
Cs Snubber capacitance
CS, S Capacity available due to distribution substation
Cs1 , Cs2 , … , Csn , and Csm Minimal cut-sets
CT Capacity of a composite distribution system
CT Tth minimal cut-set
society
Ctotal Total reliability cost of society
utility
Ctotal The utility’s overall reliability costs
d Duty cycle
D Detection
DA Battery autonomy
dI C /dt Rate of change of collector current
DL Repair rate in hour
e Electron charge
Eavg Average demand energy
Eact Thermal activation energy of the degradation process (eV)
Ecap Energy capacity of the battery
Ecur , Ecel Current energy level of the battery
EENSO Annual estimated or expected energy supply not provided
when distributed generations are not linked
EENSD Annual estimated or expected energy supply not provided
when distributed generations are linked
Ei Curtailed energy in interval i
Ei and Ej Events
Em Maximum energy capacity of the battery
f Failure tests
F Failure probability
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Notations 517

F Objective function to decide the optimized values of

various parameters
FaCen(t) Cumulative distribution function of central topology
Fa(t) Cumulative distribution function
FaStr(t) Cumulative distribution function of string topology
FC Failure costs
f cu (t) Cumulative distribution function
f f (t) Failure density function
f H , f M , and f R Frequency of the system in H, M, and R states
f ii Denotes the count of power interruptions experienced by
the ith user throughout the simulation period
fm Frequency of reference signal
f max Objective function’s maximum value
FP Failure probability
fpCen(t) Probability density function of central topology
fpStr(t) Probability density function of string topology
fp(t) Probability density function
f surge Frequency of surge
f sw Switching frequency
f (t) Failure density function
f (t) Probability density function
F (⋅) Probability of failure
F(t) Cumulative distribution function
F (t) Unreliability of device
( )
F x⃗ The fitness function
( )
f x⃗ The objective function
fi Frequency of transitions from “W” to “F” state
f r (tr ) Repair density function
f pos Posterior distribution
f por Prior distribution
f 1 (t), f 2 (t),…, f n (t) Failure density function of components
f𝜇B ,𝜇C The joint pdf of all random design variables and
parameters
fz(y) PDF for the daily mileage
g(t) Distribution function
g(t) Failure density function of parallel system
G(t) Cumulative distribution function
Gmax Maximum generation number
gRi The nominal resistance, i.e. capacity or strength
gSi The nominal load, i.e. demand or stress
h Characterized by a specific set of variables (T h , tonh ,
ncycle-h )
h(t) Hazard function of the system or device
h(t) The hidden layer’s neuronal output
h(t) Hazard function
ℋt History of failure process through time t
Hz(t) Hazard rate function
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
518 Notations

I Solar cell output current

IC Collector current
Ic EV charging current
ID Diode current
I dis Discharging power
I g peak Peak gate current
IL Light-generated current
IL Current through stray inductance
I OFF Turn off current
IRk Reliability importance of kth component
I0 Drain current during device turn-off
J An objective function to decide the locations of DGs
J jth cut-set or tie-set
J Possible outcomes
J Load flow Jacobian
J 𝜆min
′
The load flow Jacobian’s minimum eigenvalue
k Boltzmanns constant
k/n k-out of n configuration for partially redundant system
KB Boltzmann constant
Kf Probability of failure
Ks Probability of success
L Likelihood function
L Step in FP for L’evy flight distribution
L Length of stroke
Lavg Average load
Lcap Capacitor lifetime
LCC Life cycle cost
LCCA Life cycle cost analysis
ld and 𝜎 ld Mean of normally distributed load and standard deviation
Li Mean of Lith load level
Li Connected average load at the ith load point
Li Peak load forecast in interval i
Li Maximum load of the ith day
li Denotes the distance covered by the ith EV
Lik Likelihood function
Lo Operational life of capacitor
Lo-k Resultant lifetime derived
Lr Rated lifespan
Ls Snubber’ s inductance
Ls /Lstray Parasitic/stray inductance
L0 Lifetime under voltage V 0
m Standard operational time of the system
M Total number of households
Ma Amplitude modulating index
MC Maintenance costs
mf Frequency modulating ratio
MWh Megawatt hours
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Notations 519

n Total study duration

n Number of components
(capacitors/diodes/switches/filters)
N Combined count of simulation trials
N Cut-set quantity
N Connection matrix
N Argument an interval N (a, b]
NA Acceptor concentration
NB Set of buses
Nc A total of all distributor segments
ncycle-h Cycle count in category h
ND Donor concentration
Nf Failure across numerous operational cycles
NG Set of PV buses
Ni Time duration for occurrence of Lith load level
Ni Customer count at load location i
Ni Number of customers interrupted for ith interruption
Nf Number cycle to failure
NL Set of transmission lines
NL Number of customers at Lth load
NS Number of samples
NT Number of customers served
N(t) Value of counting random variable, i.e. failure numbers
occurred in the interval [0, t]
nr (tr , Δtr ) Number of repairs completed between tr to tr + Δtr
nj Quantity of blocks in the jth cut set or tie set
Nr Total number of repair carried over in a specified period of
study
nXY Number of transitions from state X to Y
N(0,1) Normal distribution with mean = 0 and standard
deviation = 1
O Network’s output value
O Occurrence
P Power output of a solar cell
P Transition probability matrix
P Operative pressure
P Probability matrix
P Snubber resistor’s power dissipation
p̂ Security index without including the unavailability of a
transmission line
p, q Reliability and unreliability of component
pc Crossover probability
Pc EV charging power
Pc(t) Indicates the system’s load decrease at time t and is set to
zero when the network operates normally
Pd Load demand
Pd Average load demand
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
520 Notations

Pdis Discharging voltage

pH , pM , and pR Probability of H, M, and R states
pm Mutation probability
Pf Probability of failure
p fi The pof for the solution with respect to the ith constraint
p fi The approximated pof for the ith performance criteria
Pfi Probability of failure of Lith load level
pTf The true probability of the failure
Pi Probability of loss of load in interval i
PLoad Load power
pn , pO, pS Probabilities of normal, open circuit, and short circuit for
diode
p0 (t + dt) Probability of system being in “0th” state at t + dt
Ppvg PV generator power
Pr Rotor active power
Pr(t) Probability function
ps Probability of successful change-over for standby system
Ps Stator active power
PS Success probability
ptfi The target pof
p1 (t) Availability
p2 (t) Unavailability
p1ss Steady-state availability
p2ss Steady-state unavailability
Pa (X i ) Parent nodes
( )
Pa Xti Parents of Xti and Xti is the ith node at time t
k
Pg,min , Pg,max
k Lower and higher limits to the active power output at the
kth bus
P(Ai ) Mutually exclusive events
PIIo Current insecurity index obtained from the ANN
PIIth Threshold value of the indicator
P(A, B) Joint probability
P(A/B) Conditional probability
P(B) Probability of event B
P(B/A) Conditional probability
P(Ei ) Probability of occurrence of ith event
pij Transition probabilities from state i to state j
pij (t, Δt) Transitional probability
P(S/X)
( ) Probability of system success provided “X” is working.( )
P S∕X Probability of system success given that “X” has failed X
P(X) Reliability of “X” component
P(xi /Pa (xi )) Conditional probability
P(X t ∣ X t − 1 ) Transition probability
P(X 1 : T ) Joint probability distribution of 2 TBN network
PW Wind Turbine power output
PW,r Rated wind turbine output power
q Total rate of transition
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Notations 521

Qf Fixed oxide charge

Qfb The cumulative number of simulations resulting in
failures
Qkg,min , Qkg,max Lower and higher limits to the reactive power output at
the kth bus
qij Transition rate from state i to j
qjk Transition rates from jth to kth state
QL Reactive load
Qm Mobile ion charge
Qot Intrinsic charge of oxide
QP Unreliability of parallel system
“r” Number of standby component
r Interruption duration
r Signifies the duration of failure
R Resistance to lines
R Reliability
rand Arbitrary digit between [0, 1]
Re(t) Reliability function
RD (%/h) Inverse to the meantime to failure (MTTF)
RD Erying Rate of degradation by Erying
RD Peck Rate of degradation by Peck
Re(t) Reliability function
Rep(t) Parallel equivalent reliability function
Res(t) Series equivalent reliability function
ReStr(t) Reliability of string topology
RGtotal Total gate resistance
Rg ext External gate resistance
rh Relative humidity (%)
ri Duration of interruption per failure at ith load point
Ri Reliability of block/component i
Ri (t) Reliability of the ith component
Roff Resistance when the MOSFET turns-off
RP Reliability of parallel system
Rs Reliability of series system
Rs Reliability of system
Rs Snubber resistance
Rs. Indian currency rupees
RT Total resistance of the line
R (⋅) Reliability of a component at the respective time
R(t) Reliability
R(t) Reliability function
R(t) Reliability of the system
RV Residual value
r 1 , r 2 ,…,r n Time independent reliability
S Sample space
S Sensitivity matrix
S Success tests
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
522 Notations

S Severity
SAIFIt , SAIDIt , CAIDIt and AENSt The desired/target values of the corresponding indices
Sd System load overall (real and reactive load)
SDBj Standard deviation of vector B, the jth constraint
SDCl Standard deviation of vector C, the lth constraint
SI Security index with including the unavailability of a
transmission line
S𝜆max Higher limit on the highest eigenvalue of [S]
SPIIi Sensitivity of PII with respect to U i
SPIIr Sensitivity of indicator with respect to the “r” control
variable
SPIIDi Sensitivity of indicator (PII) with respect to load shedding
at ith bus
t Sequence of time
t Operating or at-risk time of item
tdon Turn-on delay time
t1 , t2 … Failure times in success
T Network’s target value
T Tie set quantity
T Average time between instances of failure
T Average value of temperature
ΔT Junction temperature swing
T amb Ambient atmospheric temperature
Tc Standby system the cycle time
Tc Battery charging time
T cel Cell temperature
T dis Discharging time
Tf Life length of component
ti The target output for the ith training dataset
ti Stands for the cumulative charging disruption time during
the simulation period for the ith EV
Tj Junction temperature in degree Celsius
T jest Estimated junction temperature
T jn Junction temperature
T jo Base junction temperature
tk Operating period
Tk kth system state duration
Tn Tournament size
To Operating temperature
Tr Time to repair of component
Tr Upper category temperature
Tr rth minimal tie-set
T 1 ,T 2 ,…, T n Life lengths of components
u Random number [0, 1]
U Unavailability
(k)
Ubest The best choice discovered in FP among all solutions in
the generation that currently exists
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Notations 523

(k)
Ubest The best choice discovered in Rao optimization method
among all solutions in the generation that currently exists
(k)
Uworst The current worst solution discovered using Rao
optimization techniques, out of all the solutions at this
generation
Ui Input variable at ith node
Ui Component of the input vector [U]
U(k)
i
At kth generation, the resultant solution vector
Ui(k+1) Vector modified at the kth generation
Uij0 jth parameter of U i vector
U j, max and U j, min Upper and lower limits for the variable U j
UL Hour/year
Us Average annual outage time
U sys, i Duration of the system’s yearly outage at the ith load point
u(t) Intensity function of a point process
V Voltage across the solar cell terminals
Vc EV charging voltage
V ceo Collector-emitter voltage at T jo
V ceon(I high) Collector-emitter voltage under high current
Vc,in Cut-in speed
Vc,out Cut-out speed
V dc DC-Link voltage
V dis Discharging current
V fr Forward voltage of the diode
ΔV ge Gate emitter voltage variation
′
VLs Voltage across snubber s inductance
VLs2 Peak voltage across snubber capacitor
ΔV m, i The change in the load bus voltage vector as a result of
modal variation
i
Vmin , Vmax
i ith load bus voltage’s lower and higher limits
Vn Virtual age after nth repair of the system
Vo Operating voltage
Vr Rated wind speed
Vr Rated voltage
V Sr Voltage stress factor
VT Thermal voltage
V thr Threshold voltage
V0 Nominal voltage
W hh The weight matrix that recurs between neighboring time
steps’ hidden layers
W hx The matrix of weights between the i/p and the hidden
layers
W ij Weight that connects the ith input node to the jth hidden
neuron
ΔW jk Weight change W jk
X Key-stone component in decomposition method
X An element of the input vector
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
The time between failures

Total reactance of the line

Instantaneous failure rate

Stator circuit reactance
Rotor circuit reactance

The network’s output

Reactance to lines
Notations

Z(t)
XT
Xn

Xs
Xr
X

Y
524
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
525

Index

a Average interruption duration of EV charging

Accelerated life testing (ALT) 104 index (EAIDI ) 447
Accumulated capacitor damage 490 Average interruption frequency of EV charging
AC microgrids 495, 496 index (EAIFI ) 446
Active snubbers 468 Average outage time per failure 500
Adequacy 4, 6, 485 Average service availability index (ASAI) 306,
Adequacy assessment 304 501–502
Advanced control algorithms 321
Advanced metering infrastructure (AMI) 443 b
Advanced simulation methods 125 Backpropagation algorithm 202, 204
AIAG scoring system 107 Bahir Dar’s power distribution test system
Analytical physics 105, 108 319–320
Analytical probabilistic models 444 Bathtub curve 101, 10, 489
Annual downtime (AD) 491 burn-in screening 123–12
Apportionment 104 infant-mortality period 10–123
Approximate reliability techniques 125 random failures 10, 123
Arithmetic reduction of age (ARA) 156–161 wear-out 10, 123
Arithmetic reduction of intensity (ARI) 156 Battery electric vehicle (BEV). See also Electric
Artificial neural networks (ANNs) 126–127, vehicle (EV)
131–132, 134–137, 202–205 battery system 431–432, 434
Audits 109 drive motors 432
Availability 114, 326, 420–421, 454 motor control unit 432
HVDC system 422, 423 PDU 432
nonrepairable system 121–122 power-train system 433
power converter 422 Battery management systems (BMS) 442
repairable system 121 Bayesian approach 7
steady-state 120 Bayesian network (BN)
Average annual outage time 500 advantages 27–28
Average Energy Not Supplied (AENS) 249, 259, dynamic 26–27
260, 263, 267, 306 example 21–22
Average extra length for charging (EAELC ) 447, limitations of 28
448 parent and child nodes 20
Average extra time for charging (EAETC ) 447 Bayesian reliability evaluation
Average frequency of EV participating in V2G flowchart 24
(EAFPV ) 447 formula 22
Reliability Analysis of Modern Power Systems, First Edition. Edited by R. K. Saket and P. Sanjeevikumar.
© 2024 The Institute of Electrical and Electronics Engineers, Inc. Published 2024 by John Wiley & Sons, Inc.
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
526 Index

Bayesian reliability evaluation (contd.) filter 388, 390, 394, 397

inference 25–26 switch 388, 390, 393, 393
parameter modeling 25 Component level reliability 503
series-parallel system 24 Composite distribution system (CDS) 285
steps in 22–23 components modeling 286
structure modeling 23–25 reliability indices of 290
verification and validation 26 sample 289
Beta distribution 25 Conditional probability 20
Binary particle swarm optimization 248 Condition-based maintenance (CBM) 146
Binomial distribution 220 Contingency
Block diagram analysis 94–95 evaluation 169
Bootstrapping method 248 planning 314
Bridge reliability network 66, 69–70 selection 169
Brushless DC (BLDC) motor 432–433 Contingency enumeration method 437–438
Buck–boost converter 332 Converter 332
Built-in test equipment (BITE) 120 Converter reliability modeling 489–491
Burn-in screening 123–12 Corrective maintenance (CM) 145
Cost(s)
c customer interruption cost 251, 269, 271
Capacity modeling 286 maintenance activity adjustment 271
Capital cost (CC) 455 network 269
Cascaded H-bridge (CHB) multilevel inverters reduction 5
452–454, 456–458, 463 vs. reliability 269, 270
Catastrophic failure 412, 413 Cost–benefit analysis 252–253, 256, 263
Cat swarm optimization 248 Cumulative distribution function (CDF) 12,
Cell temperature 356–357 305, 371–377, 382, 384, 402–403
Central-inverter PV system Cumulative present value (CPV) technique
reliability block diagram 389–392 252–253
reliability modeling 385 Customer Average Interruption Duration Index
Charge controller 332 (CAIDI) 249, 259, 260, 263, 306
Charging infrastructure 429, 431, 434 Customer interruption cost (CIC) 251, 269, 271
Chronological probabilistic models 444–445 Customer satisfaction 5
Common mode failure (CMF) 55 Cut-set
Communication and control networks 315 block diagram analysis 94–95
Competitive benchmarking 106 method, of reliability evaluation 62–63, 71
Complex reliability networks 57 minimal 176, 201
cut-set method 62–63 obtaining 65–66
decomposition method 59–61 PII computation 201–202
event space method 57–59 Cybersecurity 315
examples 69–72
tie-set method 61–62 d
Complex system 330–331 Data analytics 314
Component failure 453 Data envelopment analysis (DEA) 268
Component failure rate, in PV system DC-link capacitor failure rate 387, 389, 393, 396
388, 411 DC microgrids 495, 496
DC-link capacitor 387, 389, 393, 396 Decomposition method 59–61, 67–68, 70, 72
diode 387–388, 390, 393, 397 Demand response 430, 443
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Index 527

EVs 311 Distribution systems reliability with integrated

and flexibility 321 EVs 443
optimization 314 bidirectional power flow 443
Depletion capacitance 418 infrastructure overload 443
Design for reliability (DfR) 105–107, 109 load variability 443
metrics 107 reliability assessment 443
tools 106–107 reliability enhancement 443–444
Design studies 107 reliability parameters 446–448
Deterioration 143 Dynamic Bayesian networks (DBN) 26–28
DG only system 334–335 Dynamic object-oriented Bayesian networks
Diesel engine 333 (DOOBNs) 26
Diesel with battery backup system 336 Dynamic voltage and reactive power optimization
Diode-clamped multilevel inverter (DCMLI). 314
See Neutral-point clamped (NPC)
multilevel inverter e
Diode equation 307 Effect analysis 101. See also Failure modes and
Diode failure rate 387–388, 390, 393, 397 effects analysis (FMEA)
Discharge-suppressing RCD snubber 470–472 Electrical distribution reliability indices 249
Discrete Markov process 34, 36 Electric motors 434
Discretization 35 Electric power systems 4, 6
Distributed generation (DG) 247 Electric vehicle (EV)
advantages 285 battery management systems 442
cost–benefit analysis 252–253, 256, 263 challenges 313
locations 250, 255 charging
reliability 247 battery charging time 312
reliability data of 290 bidirectional 313
in RHPVES 333 power 312
Roy Billinton test system bus-4 251, 256, state of charge 312
258–263 structure 311
scheduling 248 demand response 311
as standby units 250–252 discharging
Distribution network (DN) reliability 444 battery discharging time 312
Distribution system integrated with renewable power 312
energy systems 303 energy storage 311
Bahir Dar’s power distribution test system grid ancillary services 311
319–320 islanded operation 312
challenges 320 peak shaving 311
future directions 320–321 reliability assessment 435
IEEE RTS-79 system 315–318 considerations in 439–442
optimization techniques 313–315 contingency enumeration method 437–438
renewable energy sources 303 Markov models 435–436
advantages 313 Monte Carlo simulation 436–437
challenges 313 state space method 438
EV charging and discharging 310–313 trend 439
solar energy 307–309 voltage regulation 312
wind energy 309–310 Electric vehicle grid integration 427
Roy Billinton Test System 318–319 benefits and challenges 430–431
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
528 Index

Electric vehicle grid integration (contd.) constant 384

charging infrastructure 429, 431 density function 52
data and communication systems 428 effects of 102
demand response programs 430 frequency 41, 44
distribution system inquiries 115
planning 428 modes 101, 102
upgrades 430 most probable point of 125
energy storage systems 430 probability density function of 454
fleet charging optimization 430 random 10, 123
grid stability 429–430 repeated 108
load management 429–430 wear-out 10, 123, 413–414, 421
modernization 428 Failure cost (FC) 455
renewable energy sources 428 Failure density function 305
smart charging 430 Failure modes and effects analysis (FMEA) 100,
TOU pricing 430 454
V2G technology 428 with Advanced Product Quality Planning
voltage and frequency control 430 101, 103
Electric vehicle sub-components 431 application of 101
battery system 431–432 benefits of 104
drive motors 432 DfR 106
motor control unit 432 modified 104
PDU 432 worksheet 101, 102
reliability studies 433 Failure probability 13, 63, 125, 175, 199, 201,
importance of 434 238–241
methodologies 434–435 Failure rate 388, 454
scope of 434 constant and variable 421–422
Electromagnetic interference (EMI) shielding DC-link capacitor 387, 389, 390, 391
design 126, 131, 135–137 diode 387–388, 390, 393, 397
effectiveness 138–139 filter 388, 390, 394, 397
Enabling events 95 physical 503
Energy distribution companies (EDCs) 268 PV inverters 411
Energy Not Served per Interruption (ENSPI) switch 388, 390, 393, 393
445 vs. time 502–503
Energy storage system (ESS) 444, 500 Fault analysis 304, 443
Event space method 57–59, 66–67 Fault location, isolation, and service restoration
Event tree analysis (ETA) 445 (FLISR) 314
Expected Energy Not Supplied (EENS) 48, 306, Fault tree analysis (FTA) 445
327, 331 applications of 98
Exponential distribution 221, 228–230, 232–233 fuzzy 98
Exponential failure law 53 purpose of 96
steps 98
f symbols 96, 97
Failure(s) 327 Fault tree, and Petri nets 99, 101
analysis 5 Filter failure rate 388, 390, 394, 397
catastrophic 412, 413 First-order reliability method (FORM) 125
causes of 102 Flower pollination (FP) algorithm 253–256, 261,
common mode failure 95 274–280
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Index 529

Flying capacitor multilevel inverter (FCMLI) rotor core 342

454, 458–461, 463, 464 as SEIG 341–343
Forced outage 306 Inferences, Bayesian 23, 25–26
Fossil fuels 381, 451 Intensity function
Fourteen-bus, twenty-line IEEE test system bathtub-shaped 150
179–186, 192–193, 206–210 complete 151
Frequency-duration (FD) 286–287 Inverter 332
Frequency of combined events 42–44
Functional life period 10 j
Fuzzy fault tree analysis (FFTA) 98 Joint probability density function (pdf) 125
Joint probability distribution 20
g
Junction temperature measurement
Gamma distribution 227
direct method 414
Gaussian distribution approach (GDA)
indirect method 414
233–235
short circuit current 416, 417
Gaussian distribution techniques 108
Generalized renewal process (GRP) 148 thermo-sensitive internal gate resistance
ARA models 156–161 418–420
arithmetic reduction of intensity 156 threshold voltage (V thr ) 415–417
examples 161–164 turn-on/off delay time 417–419
Genetic algorithm, real-coded 133–135 V ceon(Ihigh) 415, 416
Geometric distribution 223–224 Junction tree algorithm 26
Grid-forming inverters 321
Grid integration strategies 303 k
Grid modernization 428 Kijima-I model 157–159
Grid stability 320 Kijima-II model 159–161
Grid system reliability parameters 447–448
Grid upgrades 444 l
Laplace transform method 35
h Latin hyper cube design 131
Hardware reliability 19 Latin hypercube sampling (LHS) 131
Hazard function 10, 123 Layer-wise reliability 503–504
Hazard rate function 305, 382, 384 LC filter 388
Homogeneous Poisson process (HPP) 147, 151
Life cycle cost analysis (LCCA) 455
Horizon mission methodology (HMM) 115
Life cycle management (LCM) 6
HVDC availability model 422, 423
Life distribution of power components 220
binomial distribution 220
i
exponential distribution 221
IEEE 1624 scoring system 107
IEEE RTS-79 system 315–318 gamma distribution 227
IEEE standard test system geometric distribution 223–224
6-bus 7-line 178–179, 191 normal distribution 225–227
14-bus 20-line 179–186, 192–193, 206–210 Poisson distribution 221–223
25-bus 35-line 186–190, 194–196, 210–214 Weibull distribution 224–225
Imperfect repair 147 Life length 41
Induction motor (IM) 432 Likelihood function 152
magnetization graph without load 344 Linear programming model 200
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
530 Index

Load Markov chain analysis 455

analysis 304 Markov models 435–436, 444
balancing 314 Markov process 31–32
flow analysis 443 applications 328
management 429–430 assumption in 328
modeling 286 calculating reliability indices 329
shedding 206, 210, 211, 214, 314 discrete 34, 36
Load duration curve (LDC) 236 general state transition diagram 32–33
Load flow’s minimal eigenvalue Jacobian homogeneity 328
174–175 homogeneous/stationary process 32
Load point indices 500 irreducibility 328
Load-sharing system reliability evaluation memoryless 32
81–83 states 327
Logistics Support Analysis Record (LSAR) 117 stationarity 328
Loss of energy expectation (LOEE) 445 transition probabilities 327
Loss of energy (LOE) 488 two-state continuous 34–36
Loss of load expectation (LOLE) 423, 445, 489 Markov property 328
Loss of load frequency (LOLF) 445 Maximum likelihood estimation (MLE) 152
Loss of load probability (LOLP) 236–241, 327, Mean down time (MDT) 46–48
331, 502 Mean squared error (MSE) 132
Lumped snubbers 470–471 Mean time between failure (MTBF) 22, 41, 106,
120, 121, 232–233, 438, 454
m Mean time to failure (MTTF) 14, 53, 106,
Machine learning (ML) 146 121–122, 231, 438, 454, 477, 478
Maintainability 113 Mean time to repair (MTTR) 120–122, 438
elements of 114–120 Mean up time (MUT) 46–48
enhancements 114 Meta-models 126
and system engineering 115–118 Microgrids 493
Maintainability analysis 113 AC 495, 496
factors 118 advantages of 497, 499
mathematics 119–120 architecture 494
process 118–119 black start capability 499
Maintained systems challenges and future directions 504–505
maintenance actions on 145–146 classification 494–496
parametric reliability estimation models component selection 499
149–153 control and management systems 499
Maintain, repair, and overhaul (MRO) 143 control strategies 496
Maintenance 143 centralized 497
actions on maintained systems 145–146 decentralized 497, 498
condition-based 146 multi-agent system 497, 498
corrective 145 cybersecurity 500
predictive 146 DC 495, 496
preventive 143, 146 and decentralization 321
proactive 146 disadvantages of 497, 499
Maintenance activity adjustment costs 271 energy scheduling 502
Maintenance cost (MC) 455 energy storage systems 500
Markov analysis 98, 99 grid-connected 496
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Index 531

grid interaction 500 comparative 463–464

islanding capability 499 techniques 453–455
layer-wise reliability 503–504 reliability factors 453
load management 500 switched-capacitor-based 452
optimization 314, 502 switches 452
planning and design 499–501 T-type 452, 460–462
predictive maintenance 499 types of 456–462
redundancy 499 Multistate model 66–68
reliability evaluation 500–502
standalone 496 n
technology 496 N-1 contingency selection 170
testing and simulation 500 Network cost 269
training and maintenance 500 Neutral-point clamped (NPC) multilevel inverter
Micro-inverter PV system 452, 454, 458, 459, 463
reliability block diagram 396–399 Node elimination method 64–65
reliability modeling 386 Noise, numerical 126
Minimal cut set 201 Nominal operating conditions temperature
Minimal repair 147 (NOCT) 356
Model-based reliability enhancement Nonhomogeneous Poisson process (NHPP)
488–489 147–148, 151–156
Modern power system (MPS) reliability Nonlinear programming methods 170
423–424, 483–485 Normal distribution 225–227, 232
Monte Carlo simulation (MCS) 126–127, 130, Novel collapse prediction index (NCPI) 200
171, 248, 303, 436–437, 443, 444
reliability indices evaluation of CDS 288–289 o
security index evaluation 172–174 Operating conditions 233
SEIG failure and success 345, 349–350 Operating cost (OC) 455
steps 172–174 Operational efficiency 5
Most probable point (MPP) Optimal placement of distributed energy
of failure 125 resources 314
multiple 126 Optimal switching and reconfiguration 314
Multi-agent system (MAS) 497, 498 Optimization techniques 313–315
Multilayer perceptrons 132 Overhaul 144
Multilevel inverter (MLI) Over-voltage 442
benefits 452
cascaded H-bridge 452–454, 456–458 p
comparison of 463–464 Parallel components system 330
configurations 451 Parallel reliability network 53–56
conventional 452 Parallel system 233, 454
diode-clamped 454 Parameter uncertainty 172
flying capacitor 454, 458–461 Parametric reliability estimation models
FMEA 454 149–153
modular multilevel converter 452 Partially redundant system 56–57
modulation strategies 456 Passive snubbers 468–469
NPC 452, 454, 458, 459 Path tracing method 61–62
output voltage waveforms 452 Peak load consideration (PLC) 235–236
reliability assessment Perfect repair 146–147
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
532 Index

Performance-based reliability 127–128 in modern power system 483–485

Performance indicators (PI) 169 stress-strength analysis 482
Performance indices (PI) 127–128 V-shape model-based assessment 486–489
Permanent magnet synchronous motor (PMSM) Power electronics 434
432 Power electronics converters
Petri nets catastrophic failure 412, 413
applications 99–100 wear-out failure 413–414
fault tree and 99, 101 Power law model 151
logical relations for 99, 100 Power Law Process (PLP) 144–145
stochastic 99 Power reliability engineer (PRE) 4
Photovoltaic (PV) system 307–309, 325–326, Power-sharing 482
332, 381 Power system reliability (PSR) assessment 304
cumulative distribution plots 402–403 Predictive maintenance 146, 314
failure rate 387, 411 Preventive maintenance (PM) 143, 146
DC-link capacitor 387 Proactive maintenance 146
diodes 387–388 Probabilistic insecurity index (PII) 200
filter 388 computation
switch 388 ANN 202–205
probability density plots 403–404 cut-set method 201–202
RBD 384–385 flowchart 203
central-inverter 389–392 IEEE standard test system 206–214
micro-inverter 396–399 sensitivity of 202–206
string-inverter 391, 393–396 Probabilistic insecurity index (PISI) 171, 176
reliability modeling 385–386 Probabilistic modeling 5–6
central-inverter 386 Probabilistic sufficiency factor (PSF) 127, 129,
micro-inverter 386 134–137
string-inverter 386 Probability density function (PDF) 369–371,
reliability plots 400–401 382, 403–404, 437, 444, 454
topologies 382, 383 Probability distribution
Physical reliability methods 93 binomial 220
accelerated life testing 104 exponential 221, 228–230, 232–233
block diagram analysis 94–95 gamma 227
fault tree analysis 96–98 geometric 223–224
FMEA 100–104 normal 225–227, 232
Petri nets 99–101 Poisson 221–223
reliability apportionment 104 for reliability evaluation 227–232
state space analysis 98, 99 Weibull 224–225, 230–232
Physics of failure (PoF) 106 Probability of failure (pof) 126
Plug-in electric vehicles (PEVs) 427 Probability of load shedding 199
Point process 149 Probability theory 7–9
intensity function of 150 Productivity 5
mean function of 149 Product reliability 326
Poisson distribution 221–223 PV/DG/battery system 337–338
Power converter availability model 422 PV/DG system 336–337
Power electronic devices and systems (PEDS) PV-only system 334
reliability PV with battery backup system 335
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Index 533

q regulatory compliance 5
Quality of service 220 role of 4–5
safety improvement 5
r Reliability functions 305–306
Random failures 10, 123 Reliability handbooks 503
Rao algorithms 273–280 Reliability importance 53
Rate of occurrence of failure (ROCOF) 150 Reliability indices 326–327
RCD charge-discharge snubber 469, 470 Reliability indices assessment 285
RCD snubber 470–473 frequency-duration concept 286–287
RC snubber 469, 470, 473, 474 MCS-based 288–289
Reactive power control variable 200, 205 Reliability life cycle 108–109
Reactive power rescheduling 207, 211 Reliability maintenance 109
Real-coded genetic algorithm 132–134 Reliability metrics 303–304
Recurrent neural network (RNN) 171 Reliability network 51
architecture 177, 178 complex 57–63
formulation of 178 examples 68–72
inputs of 177 parallel 53–56
repeated module in 177, 178 series 51–53
Regulatory compliance 5 Reliability predictive modeling 106
Reliability analysis techniques 148 Reliability team 105
Reliability apportionment 104 Reliability verification 105, 108, 109
Reliability-based design optimization (RBDO) Renewable energy sources (RESs) 303, 325, 451
125, 126 advantages 313
MCS-based 126–127 challenges 313
problem 128–129 EV charging and discharging 310–313
using PSF and ANNs 134–137 EVs 428
Reliability block diagram (RBD) 94–95, 384, 445 solar energy 307–309
DG/battery system 336 wind energy 309–310
DG only system 334 Renewable hybrid PV energy system (RHPVES)
parallel configuration 384, 385 326, 331
PV/DG/battery energy system 338 charge controller 332
PV/DG system 337 converter 332
PV system 334 DG set 333
with battery 335 inverter 332
central-inverter 389–392 PV arrays 332
micro-inverter 396–399 solar batteries 332–333
string-inverter 391, 393–396 Renewal process (RP) 147, 151
series and parallel combination 384 Repair 144
series configuration 384 general 147
Reliability-centered maintenance (RCM) 313 imperfect 147
Reliability development 219 minimal 147
Reliability engineering 4, 453 perfect 146–147
cost reduction 5 Repairability 327
customer satisfaction 5 Repair density function
features of 5–6 defined 31
goals of 93 plot 32
productivity 5 Repair effectiveness (RE) 156
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
534 Index

Residual magnetism in SEIG issues 341

losses 342–343 magnetization graph 343, 344
restoration 343 Monte Carlo modeling 345, 349–350
Residual value (RV) 455 reliability assessment 344–348
Resiliency planning 304 residual magnetism
Resonant converters 467 losses 342–343
Response surface method (RSM) 126 restoration 343
Rewards and penalties scheme (RPS) 267–269 survival function 347
configurations 270 Vg/F variation 343, 344
FP and Rao algorithms 273–280 voltage change 343, 345
problem identification 271–273 Semiconductor failure 412
RBTS-2 274–280 Sensitivity analysis 304, 443
theory 269–270 Sensitivity matrix 175
Risk assessment 5, 107, 304 Series components system 329–330
Risk assessment and mitigation 314 Series–parallel system 330–331
Risk priority number (RPN) 103
Series reliability network 51–53
Roy Billinton Test System Bus-2 (RBTS-2) 251,
Series system 232–233, 454
256
Short circuit current (SCC) 416, 417
distributed generation
Silicon-controlled rectifiers (SCRs) 332
customer information 258
Simpson 1/3rd rule 236–241
FP, TLBO, and DE control parameters 258,
Single components
261
availability and unavailability 37–38
parameter values 259
state transition diagram 37
reliability indices 263
Six-bus, seven-line IEEE test system 178–179,
repair time and failure rate 256, 258, 262
191
system information 257
Smart grid 319, 321
rewards and penalties scheme 274–280
Roy Billinton Test System (RTBS) 318–319 Snubbers
actions 468
s active 468
Safety factor concept (SFC) 235–236 C snubber 470
Safety factors (SF) 129 discharge-suppressing RCD snubber 470–472
Safety improvement 5 examples 474–476
Scenario analyses 314 industrial power applications 476–478
Scheduled maintenance. See Preventive lumped 470–471
maintenance (PM) passive 468–469
Schur’s inequality 175 purpose of 467
Second-order reliability method (SORM) 125 RCD charge-discharge snubber 469, 470
Security 4, 6 RCD snubber 470–473
assessments 169 RC snubber 469, 470, 473, 474
index evaluation 172–174 reliability aspects in 476–478
Self-energized induction generator (SEIG) turn-OFF 469–471
cumulative failure distribution 347 Socioeconomic optimum reliability level 269
excitation failure 343 Solar batteries 332–333
failed density function 346 Solar cells 332
failure and success 345, 348–350 Solar energy 307–309, 325
hazard graph 348 Solar panel reliability and mean life assessment
IM as 341–343 353
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Index 535

atmospheric temperature 354, 355 Stochastic Petri nets (SPN) 99

CDF calculations 371–374 Stress-strength analysis (SSA) 482, 487
cooling up to 25 ∘ C 371, 370–371 String-inverter PV system
cooling up to 35 ∘ C 371, 372 reliability block diagram 391, 393–396
cooling up to 45 ∘ C 371, 373 reliability modeling 386
without cooling 374 Supportability 113, 114
cell temperature 356, 357 Switched reluctance motor (SRM) 432
cooling methods 353, 354 Switch failure rate 388, 390, 393, 393
degradation equations 357, 378 Synchronous reluctance (SynRel) machines 432
mean life after cooling 359, 378 System Average Interruption Duration Index
increment in 378 (SAIDI) 248, 249, 259, 260, 263, 267, 306,
up to 25 ∘ C 359–361 445, 447, 502
up to 35 ∘ C 361–365 System Average Interruption Frequency Index
up to 45 ∘ C 359–360 (SAIFI) 248, 249, 259, 260, 263, 267, 306,
mean life without cooling 358–359 445, 447, 501
PDF calculations System engineering, maintainability and
cooling up to 25 ∘ C 369–371 115–118
cooling up to 35 ∘ C 371, 372 System Expected Energy Not Supplied (SEENS )
cooling up to 45 ∘ C 371, 373 447
without cooling 370 System reliability 101, 19
relative humidity 353, 355 bathtub curve 101–12
reliability with cooling 365 mean time to failure 14
up to 25 ∘ C 366, 367 reliability function 12–13
up to 35 ∘ C 367
up to 45 ∘ C 367–368 t
statistical techniques 354 Thermal impedance 490
workflow 355 Thermal management 453
Solution of state equations 34–36 Thermal stress 487
Standard temperature conditions (STC) 356 Thermos-sensitive internal gate resistance
Standby system reliability evaluation 418–420
assumptions 75 Thevenin-based voltage stability margin (TVSM)
examples 83–88 200
ideal conditions 75–78 Three-component state transition diagram
nonideal conditions 40–41
component failure during idle mode 80–81 Threshold voltage (V thr ) calculation 415–417
imperfect switching 79–80 Tie-set
perfect switching 78–79 block diagram analysis 94–95
State enumeration technique 44–48 method, of reliability evaluation 61–62, 71
State of charge (SoC) 442 using node elimination 64–65
State of health (SoH) 442 Time between failures (TBF) 147
State of temperature (SoT) 442 Time-of-use (TOU) pricing 430
State space analysis 98, 99 Time to repair 41
State space method 438 Time value of money strategy 252
Static reliability 51 Transitional probabilities 32
Stepped load duration curve (SLDC) 236–237, Transition intensity matrix 33
241 Transition probabilities 327
Stochastic optimization 314 Transmission coefficient 138
 Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Universita Di Firenze Sistema , Wiley Online Library on [29/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
536 Index

Transmission line v
availability and unavailability 173, 202 Variable elimination algorithm 26
down state 173, 202 V ceon(Ihigh) measurement 415, 416
MCS for security index evaluation 172–174 Vehicle-to-grid (V2G) technology 428
parameter uncertainty 171 Verification 105, 108, 109, 145
two-state model 173, 202 Virtual age-based reliability metrics 161, 162
up state 173, 202 Voltage security 199, 200
T-type inverter MLI 452, 460–464 Voltage security enhancement 205–206
Turn-on/off delay time 417–419 Volt-VAR optimization (VVO) 314
Twenty five-bus, thirty five-line IEEE test system V-shape model-based reliability assessment
186–190, 194–196, 210–214 486–489
Two-component state model 38–40
Two-slice temporal Bayesian network (2TBN) w
26–27 Wear-out failure 10, 123, 413–414, 421
Two-state continuous Markov process 34–36 Weibull distribution 224–225, 230–232, 444
Wide-area monitoring and control (WAMC)
u 314
Uncertainty Wind energy 309–310, 325
parameter 171
quantification 28 y
Under-voltage 442 Yardstick theory 268