Safety and Reliability Modeling and Its Applications Advances in Reliability Science 1St Edition Mangey Ram Editor Full Chapter PDF

Safety and Reliability Modeling and Its
Applications (Advances in Reliability

Science) 1st Edition Mangey Ram
(Editor)
Visit to download the full and correct content document:
https://ebookmass.com/product/safety-and-reliability-modeling-and-its-applications-ad
vances-in-reliability-science-1st-edition-mangey-ram-editor/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Total manufacturing assurance : controlling product

quality, reliability, and safety 2nd Edition John
Cesarone
https://ebookmass.com/product/total-manufacturing-assurance-
controlling-product-quality-reliability-and-safety-2nd-edition-
john-cesarone/
System Reliability Assessment and Optimization :

Methods and Applications 1st Edition Yan-Fu Li & Enrico
Zio
https://ebookmass.com/product/system-reliability-assessment-and-
optimization-methods-and-applications-1st-edition-yan-fu-li-
enrico-zio/
Advances in system reliability engineering Davim
https://ebookmass.com/product/advances-in-system-reliability-
engineering-davim/
Photovoltaic module reliability Wohlgemuth
https://ebookmass.com/product/photovoltaic-module-reliability-
wohlgemuth/
Engineering Reliability and Risk Assessment Harish Garg
https://ebookmass.com/product/engineering-reliability-and-risk-
assessment-harish-garg/
Food Safety and Human Health Ram Lakhan Singh
https://ebookmass.com/product/food-safety-and-human-health-ram-
lakhan-singh/
Reliability Investigation of LED Devices for Public

Light Applications 1st Edition Edition Raphael Baillot
And Yannick Deshayes (Auth.)
https://ebookmass.com/product/reliability-investigation-of-led-
devices-for-public-light-applications-1st-edition-edition-
raphael-baillot-and-yannick-deshayes-auth/
Lead-free Soldering Process Development and Reliability

Jasbir Bath
https://ebookmass.com/product/lead-free-soldering-process-
development-and-reliability-jasbir-bath/
Nature-inspired computing paradigms in systems :

reliability, availability, maintainability, safety and
cost (RAMS+C) and prognostics and health management
(PHM) Michael G. Pecht (Editor)
https://ebookmass.com/product/nature-inspired-computing-
paradigms-in-systems-reliability-availability-maintainability-
safety-and-cost-ramsc-and-prognostics-and-health-management-phm-
Safety and Reliability
Modeling and Its
Applications
Advances in Reliability Science covers traditional topics in reliability engineering (degra-
dation models, dynamic network and product reliability, maintenance and reliability
statistics) as well as important emerging topics such as multi-state systems reliability
and reliability decision-making. All of these areas have developed considerably in recent
years, with the rate of reliability research output climbing steeply. Books in this series
showcase the latest original research & development in reliability engineering science
from industry and academia, while exploring innovative research ideas for researchers
considering new projects and exploring the real-world utility of these concepts for
practitioners.
Series Editor: Mangey Ram, Professor at Graphic Era University, Dehradun, India
Safety and Reliability Modeling and Its Applications

Mangey Ram, Hoang Pham
978-0-12-823323-8
Reliability and Maintenance Optimization in Multi-indenture Systems
Won Young Yun
978-0-323-85054-4
Reliability Analysis and Asset Management of Engineering Systems
Gilberto Francisco Martha de Souza et al
978-0-12-823521-8
Engineering Reliability and Risk Assessment
Harish Garg, Mangey Ram
978-0-323-91943-2
Reliable and Resilient Logistics Systems
Agnieszka Tubis, Sylwia Werbińska-Wojciechowska
978-0-323-91752-0
Safety and Reliability
Modeling and Its
Applications
Edited by
Mangey Ram
Graphic Era (Deemed to be University), Dehradun, India
Hoang Pham
Rutgers University, New Jersey, United States
Elsevier
Radarweg 29, PO Box 211, 1000 AE Amsterdam, Netherlands
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom
50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States
Copyright © 2021 Elsevier Inc. All rights reserved.

No part of this publication may be reproduced or transmitted in any form or by any
means, electronic or mechanical, including photocopying, recording, or any information
storage and retrieval system, without permission in writing from the publisher. Details
on how to seek permission, further information about the Publisher’s permissions
policies and our arrangements with organizations such as the Copyright Clearance
Center and the Copyright Licensing Agency, can be found at our website:
www.elsevier.com/permissions.
This book and the individual contributions contained in it are protected under copyright
by the Publisher (other than as may be noted herein).
Notices
Knowledge and best practice in this field are constantly changing. As new research
and experience broaden our understanding, changes in research methods, professional
practices, or medical treatment may become necessary.
Practitioners and researchers must always rely on their own experience and knowledge in
evaluating and using any information, methods, compounds, or experiments described
herein. In using such information or methods they should be mindful of their own
safety and the safety of others, including parties for whom they have a professional
responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or
editors, assume any liability for any injury and/or damage to persons or property as a
matter of products liability, negligence or otherwise, or from any use or operation of any
methods, products, instructions, or ideas contained in the material herein.
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library
Library of Congress Cataloging-in-Publication Data
A catalog record for this book is available from the Library of Congress
ISBN: 978-0-12-823323-8
For Information on all Elsevier publications visit our website at

https://www.elsevier.com/books-and-journals
Publisher: Matthew Deans

Acquisitions Editor: Brian Guerin
Editorial Project Manager: Emily Thomson
Production Project Manager: Kamesh Ramajogi
Cover Designer: Mark Rogers
Typeset by Aptara, New Delhi, India
Contents
Preface xi
Acknowledgement xiii
About the Editors xv
List of Contributors xvii
1 Reliability analysis of asphalt pavements: concepts and

applications
Abhishek Mittal
1.1 Preamble 1
1.2 Concepts of reliability 2
1.3 Literature regarding the application of reliability concepts
for asphalt pavements 4
1.4 Issues with estimation of pavement reliability 9
1.5 Conclusions 13
Acknowledgements 13
Disclosure statement 13
References 13
2 Markov modeling of multi-state systems with

simultaneous component failures/repairs, using an
extended concept of component importance
Jacek Malinowski
2.1 Introduction 15
2.2 Basic assumptions, notation and definitions 17
2.3 Theoretical background 20
2.4 The illustrative model of an example system 23
2.5 Intensities of transitions between the system states 26
2.6 Obtaining useful reliability parameters from transition
intensities 27
2.7 Conclusion and future work 28
References 29
3 Reliability analysis of solar array drive assembly

by dynamic fault tree
Tudi Huang, Hong-Zhong Huang, Yan-Feng Li,
Lei Shi and Hua-Ming Qian
3.1 Introduction 31
v
vi Contents
3.2 DFT method 32

3.3 DFT Modeling for SADA 32
3.4 Reliability analysis of SADA 34
3.5 Conclusion 39
Acknowledgements 40
References 40
4 Reliability and maintainability of safety instrumented system

Rajesh S. Prabhu Gaonkar and Mahadev V. Verlekar
4.1 Introduction 43
4.2 Literature review 49
4.3 Problem formulation solution methodology 63
4.4 Reliability and maintainability 65
4.5 Case study on reliability and maintainability of SIS 70
4.6 Fault analysis 79
Conclusion 88
References 89
5 Application of Markovian models in reliability and

availability analysis: advanced topics
Danilo Colombo, Danilo T.M.P. Abreu and Marcelo Ramos Martins
5.1 Introduction 92
5.2 Markov chains theoretical foundation 96
5.3 Application of Markov chains to the reliability and
availability analysis of engineering systems 107
5.4 Importance measures using Markov chains 129
5.5 Uncertainty propagation in Markov chains 137
5.6 Multiphase Markov chains and their application to
availability studies 148
5.7 Final considerations 156
References 157
6 A method of vulnerability analysis based on deep learning

for open source software
Yoshinobu Tamur and Shigeru Yamada
6.1 Introduction 161
6.2 Deep learning approach to fault big data 162
6.3 Estimation of Vulnerability Based on Deep Learning 163
6.4 Numerical Examples for Estimation of Vulnerability 164
6.5 Concluding remarks 169
6.6 Acknowledgements 176
References 176
Contents vii
7 Mathematical and physical reality of reliability

Jezdimir Knezevic
Dedication 179
7.2 Mathematical reality of reliability 181
7.3 Voyage to the ice 185
7.4 Physical meanings of mathematical reality of reliability 189
7.5 Physical reality of reliability 193
7.6 Mathematical versus physical reality of reliability 224
7.7 Closing Question 224
Acknowledgement 224
References 226
8 Optimum staggered testing strategy for 1- and 2-out-of-3

redundant safety instrumented systems
Sun-Keun Seo and Won Young Yun
8.2 PFD of redundant safety systems 229
8.3 Staggered testing in 1-out-of-3 structure 231
8.4 Staggered testing in 2-out- of-3 structure 235
8.5 Conclusion 242
References 242
9 Modified failure modes and effects analysis model for

critical and complex repairable systems
Garima Sharma and Rajiv Nandan Rai
9.2 Repairable systems and imperfect repair 247
9.3 Fuzzy AHP 249
9.4 Estimation of RPN 252
9.5 Case study 254
9.6 Conclusion and future scope 258
Exercise 259
References 259
10 Methodology to select human reliability analysis technique

for repairable systems
Garima Sharma and Rajiv Nandan Rai
10.2 Selection of the best HRA technique for a particular case 266
10.3 Case study of space station 271
10.4 Conclusion and future scope 277
Exercise 277
viii Contents
Appendix 277
References 280
11 Operation risk assessment of the main-fan installations of

mines in gas and nongas conditions
G.I. Grozovskiy, G.D. Zadavinb and S.S. Parfenychevc
11.2 The ventilation system failures role in assessing the risk of
flammable gases explosion 288
11.3 Analysis of the occurrence and development of accidents 294
11.4 Analysis of the probability of explosion of flammable
gases/hydrogen sulfide at the mine from electrical equipment 299
11.5 The risk analysis results 300
11.6 Conclusion 303
References 303
12 Generalized renewal processes

Paulo R.A. Firmino, Cícero C.F. de Oliveira and Cláudio T. Cristino
12.2 The GRP models 307
12.3 The UGRP modeling 310
12.4 The WGRP modeling 311
12.5 The Gumbel GRP (GuGRP) modeling 337
12.6 Conclusion 349
Acknowledgement 350
References 350
13 Multiresponse maintenance modeling using desirability

function and Taguchi methods
Suraj Rane, Raghavendra Pai, Anusha Pai and Santosh B. Rane
13.2 Related works 356
13.3 Methodology 359
13.4 Case study 359
13.5 Result analysis 361
13.6 Conclusion and future research directions 367
References 369
14 Signature-based reliability study of

r-within-consecutive-k-out-of-n: F systems
Ioannis S. Triantafyllou
Contents ix
14.2 The signature vector of the r-within-consecutive-k-out-of-n:

F structure 375
14.3 Further reliability characteristics of the r-within-consecutive-k-
out-of-n: F structure 384
14.4 Signature-based comparisons among consecutive-type systems392
14.5 Discussion 393
References 395
15 Assessment of fuzzy reliability and signature of series–

parallel multistate system
Akshay Kumar, Meenakshi Garia, Mangey Ram and S.C. Dimri
15.2 Fuzzy Weibull distribution 399
15.3 Evolution of signature, tail signature, minimal signature,
and cost from structure function of the system 400
15.4 Algorithm for computing the system availability
(see Levitin, 2005) as 401
15.5 Example 401
15.6 Conclusion 406
References 406
Index 409
Preface
Safety and reliability analysis is definitely one of the most multidimensional

topics in system reliability engineering nowadays. This rapid development cre-
ates many opportunities and challenges for both industrialists and academics,
and has completely changed the global design and systems engineering envi-
ronment. More of the modeling tasks can now be undertaken within a computer
environment using simulation and virtual reality technologies. During the last
50 years, numerous research studies have been published that focus on safety
and reliability engineering. Supplementary experience has also been gathered
from industry. Therefore, safety and reliability engineering has emerged as one
of the main fields not only for scientists and researchers but also for engineers
and industrial managers. This book covers the recent developments in safety
and reliability modeling and its applications. It presents new theoretical issues
that were not previously presented in the literature, as well as the solutions
of important practical problems and case studies illustrating the applications
methodology.
The book Safety and Reliability Modeling and Its Applications is a combined
work of a number of leading scientists, analysts, mathematicians, statisticians,
and engineers who have been working on the front end of safety and reliabil-
ity science and engineering. All chapters in the book are written by leading
researchers and practitioners in their respective fields of expertise and present
various innovative methods, approaches, and solutions not covered before in the
literature.
Mangey Ram, Dehradun, India

Hoang Pham, New Jersey, USA
Acknowledgment
The editors acknowledge Elsevier and the editorial team for their adequate
and professional support during the preparation of this book. Also, we would like
to acknowledge all the chapter authors and the reviewers for their availability to
work on this book project.
Mangey Ram
Graphic Era (Deemed to be University), India
Hoang Pham
Rutgers University, USA
About the Editors
Prof. Dr. Mangey Ram received the Ph.D. degree major in Mathematics and
minor in Computer Science from G. B. Pant University of Agriculture and Tech-
nology, Pantnagar, Uttarakhand, India. He has been a faculty member for around
twelve years and has taught several core courses in pure and applied mathe-
matics at undergraduate, postgraduate, and doctorate levels. He is currently the
Research Professor at Graphic Era (Deemed to be University), Dehradun, India.
Before joining the Graphic Era, he was a deputy manager (probationary officer)
with Syndicate Bank for a short period. He is the editor-in-chief of International
Journal of Mathematical, Engineering and Management Sciences, Journal of
Reliability and Statistical Studies; the editor-in-chief of six Book Series with
Elsevier, CRC Press-A Taylor and Francis Group, Walter De Gruyter Publisher
Germany, River Publisher; and the guest editor and member of the editorial
board of various journals. He has published more than 250 research publications
(journal articles/books/book chapters/conference articles) in IEEE, Taylor &
Francis, Springer, Elsevier, Emerald, World Scientific, and many other national
and international journals and conferences. Also, he has authored/edited more
than 50 books for international publishers such as Elsevier, Springer Nature,
CRC Press-A Taylor and Francis Group, Walter De Gruyter Publisher Germany,
and River Publisher. His fields of research are reliability theory and applied
mathematics. Dr. Ram is a Senior Member of the IEEE, Senior Life Member
of Operational Research Society of India; Society for Reliability Engineering,
Quality and Operations Management in India; Indian Society of Industrial and
Applied Mathematics. He has been a member of the organizing committee of
a number of international and national conferences, seminars, and workshops.
He has been conferred with “Young Scientist Award” by the Uttarakhand State
Council for Science and Technology, Dehradun, in 2009. He has been awarded
the “Best Faculty Award” in 2011, “Research Excellence Award” in 2015, and
“Outstanding Researcher Award” in 2018 for his significant contributions in
academics and research at Graphic Era Deemed to be University, Dehradun,
India.
Dr. Hoang Pham is a Distinguished Professor and former Chairman (2007–

2013) of the Department of Industrial and Systems Engineering at Rutgers
University, New Jersey. Before joining Rutgers, he was a Senior Engineering
xvi About the Editors
Specialist with the Idaho National Engineering Laboratory and Boeing Com-
pany. He received his Ph.D. in Industrial Engineering from the State University
of New York at Buffalo. His research areas include reliability modeling of
systems with competing risks and random environments, software reliability,
and statistical inference. He is the editor-in-chief of the International Journal of
Reliability, Quality and Safety Engineering and an associate editor and editorial
board member of several journals, and the editor of Springer Series in Reliability
Engineering. His numerous awards include the 2009 IEEE Reliability Society
Engineer of the Year Award. Dr. Pham is the author/coauthor of 7 books and
has published his work in over 190 journal articles, 100 conference papers,
and edited 18 books including Springer Handbook in Engineering Statistics and
Handbook in Reliability Engineering. He has delivered over 40 invited keynote
and plenary speeches at many international conferences and institutions. He is a
Fellow of the IEEE and IIE.
List of Contributors
Danilo T.M.P. Abreu, Analysis, Evaluation and Risk Management Laboratory

(LabRisco), University of São Paulo, São Paulo, SP, Brazil
Danilo Colombo, Petrobras R&D Center (CENPES), Rio de Janeiro, RJ, Brazil
Cláudio T. Cristino, Department of Statistics & Informatics, Federal Rural University of
Pernambuco, Recife-PE, Brazil
S.C. Dimri, Department of Mathematics, Computer Sciences and Engineering, Graphic
Era (Deemed to be University), Uttarakhand, India
Paulo R.A. Firmino, Center for Science and Technology, Federal University of Cariri,
Juazeiro do Norte-CE, Brazil
Rajesh S. Prabhu Gaonkar, School of Mechanical Sciences, Indian Institute of Tech-
nology Goa (IIT Goa), Farmagudi, Ponda, Goa, India
Meenakshi Garia, Department of Mathematics, M.B.P.G. College, Haldwani, Nainital,
Uttarakhand, India
G.I. Grozovskiy, Deputy Director General on Science, Doctor of Engineering, Professor,
OJSC Scientific Technical Centre (STC) Industrial Safety, Moscow, Russia
Hong-Zhong Huang, Center of System Reliability and Safety, University of Electronic
Science and Technology of China, Chengdu 611731, China
Tudi Huang, Center of System Reliability and Safety, University of Electronic Science
and Technology of China, Chengdu 611731, China
Jezdimir Knezevic, MIRCE Akademy, Exeter, UK
Akshay Kumar, Department of Mathematics, Graphic Era Hill University, Uttarakhand,
India
Yan-Feng Li, Center of System Reliability and Safety, University of Electronic Science
and Technology of China, Chengdu 611731, China
Jacek Malinowski, Systems Research Institute, Polish Academy of Sciences, Newelska
6, 01-447 Warszawa, Poland
Marcelo Ramos Martins, Analysis, Evaluation and Risk Management Laboratory
(LabRisco), University of São Paulo, São Paulo, SP, Brazil
Abhishek Mittal, Principal Scientist, CSIR-Central Road Research Institute (CSIR-
CRRI), New Delhi, India
Cícero C.F. de Oliveira, Federal Institute of Education, Science and Technology of
Ceará, Crato-CE, Brazil
xviii Contributors
Anusha Pai, Associate Professor, Computer Engineering Department, Padre Conceicao

College of Engineering, Verna, Goa, India, 403722
Raghavendra Pai, Project Management Office Lead (Asia Pacific), Syngenta, Corlim,
Ilhas, Goa, India, 403110
S.S. Parfenychev, Researcher Junior, OJSC Scientific Technical Centre (STC) Industrial
Safety.Master’s degree student of the Moscow Aviation Institute, Faculty № 3 Control
Systems, Informatics and Power Engineering, Department 307 Digital Technologies
and Information Systems, Moscow, Russia
Hua-Ming Qian, Center of System Reliability and Safety, University of Electronic
Science and Technology of China, Chengdu 611731, China
Rajiv Nandan Rai, Subir Chowdhury School of Quality and Reliability, Indian Institute
of Technology Kharagpur, Kharagpur, West Bengal, India
Mangey Ram, Department of Mathematics, Computer Sciences and Engineering,
Graphic Era (Deemed to be University), Uttarakhand, India
Santosh B. Rane, Dean-Academics, Sardar Patel College of Engineering, Andheri, Mum-
bai, India, 400058
Suraj Rane, Professor, Mechanical Engineering Department, Goa College of Engineer-
ing, Farmagudi, Goa, India, 403401
Sun-Keun Seo, Department of Industrial and Management Systems Engineering, Dong-
A University, Busan, Korea
Garima Sharma, Subir Chowdhury School of Quality and Reliability, Indian Institute of
Technology Kharagpur, Kharagpur, West Bengal, India
Lei Shi, Center of System Reliability and Safety, University of Electronic Science and
Technology of China, Chengdu 611731, China
Yoshinobu Tamura, Tokyo City Univerity, Tamazutsumi 1-28-1, Setagaya-ku, Tokyo
158-8557, Japan
Ioannis S. Triantafyllou, Department of Computer Science & Biomedical Informatics,
University of Thessaly, Lamia, Greece
Mahadev V. Verlekar, Deccan Fine Chemicals (India) Pvt. Ltd., Santa Monica Works,
Corlim, Ilhas, Goa, India
Shigeru Yamada, Tottori Univerity, Minami 4-101, Koyama, Tottori-shi, 680-8552 Japan
Won Young Yun, Department of Industrial Engineering, Pusan National University,
Busan, Korea
G.D. Zadavin, Adviser to the Director General, Candidate of Engineering Science, OJSC
Scientific Technical Centre (STC) Industrial Safety, Moscow, Russia
Chapter 1
Reliability analysis of asphalt

pavements: concepts and
applications
Abhishek Mittal
Principal Scientist, CSIR-Central Road Research Institute (CSIR-CRRI), New Delhi, India
1.1 Preamble
The development of infrastructure, in particular the transportation sector, plays
a significant role in the economic growth of any country. The economic growth
demands a good road network with good connectivity all over the country.
With the reduced availability of funds, the highway agencies are placing more
emphasis on the design and construction of pavements that require minimum
maintenance during the service life. For this, it is necessary that pavements
should be designed such that a minimum design reliability (as specified in the
country’s national specifications) is achieved and the pavement construction
should be done with the latest machinery and under stringent quality control
requirements.
In India, majority of the roads (more than 90 %) are asphalt pavements,
popularly known as flexible pavements. This is due to their low construction
cost (in comparison to cement concrete/rigid pavements), ease of maintenance,
and relatively easier construction procedure. To ensure that the pavement has
adequate strength to cater to the expected traffic, it has to be designed properly
in accordance with the national specifications. For example, IRC:37 (2018) is
followed for the design of flexible pavements in India. A flexible pavement
is a multilayer structure consisting of many layers of materials starting from
subgrade at the bottom to bituminous wearing course at the top. A typical three
layer pavement structure is presented in Fig. 1.1.
The structural design of pavements deals with determining the thicknesses
of the various component layers keeping in consideration the material prop-
erties and the amount of traffic which is expected during the design life. The
current India pavement design procedure (IRC:37, 2018) is a deterministic one
wherein the various input variables like layer thicknesses, Poisson’s ratio, elastic
Safety and Reliability Modeling and Its Applications. DOI: 10.1016/B978-0-12-823323-8.00009-X
Copyright © 2021 Elsevier Inc. All rights reserved. 1
2 Safety and reliability modeling and its applications
FIG. 1.1 A typical three layer pavement structure (Dilip et al., 2013)
modulus, and design traffic are all considered as fixed. However, in reality none
of them are deterministic; they all are stochastic (probabilistic). So, in order
to develop reliable pavement designs, the uncertainty/variability of the input
variables need to be considered in the design process. This can be addressed
through the use of reliability concepts within the pavement design process. The
details of the reliability concepts and its applications to the pavement design
process, specifically in the context of flexible pavements, are discussed in the
following sections.
1.2 Concepts of reliability

Reliability is defined as “the probability that a component or system will perform
a required function for a given period of time when used under stated operating
conditions”. (Modarres et al. 1999).
In mathematical notations, the reliability can be expressed as:
R = P(T ≥ t|c1 , c2 , . . . . . .) (1.1)
Where,
t = the designated period of time or cycles for the system’s operation
T = time to failure or cycle to failure
R = reliability of the system
c1 , c2 ,…. = designated conditions, such as environmental conditions
Often, in practice, the designated operating conditions for a system c1 , c2 ,
….. are implicitly considered in the probabilistic reliability analysis and thus (1)
reduces to
R = P(T ≥ t ) (1.2)
Reliability analysis of asphalt pavements: concepts and applications Chapter | 1 3
Reliability is the probability of successful performance; thus, it is the converse

of the term “probability of failure.”
So, R = 1 − Pf (1.3)
The probability of failure is always associated with a particular performance

criterion.
Mathematically, the definition of probability of failure (Pf ) might be stated as
(Aguiar-Moya and Prozzi, 2011):

Pf = P(g(X ) ≤ 0) = . . . . . . fX (x)dx (1.4)
g(X )≤0
Where, X is the vector of basic random variables, g(X) is the limit state (or
failure) function for the failure mode considered and fX (x) is the joint probability
density function of the vector X. The expression g(X) < 0 indicates a failure
domain; g(X) > 0 indicates safe domain and g(X) = 0 denotes a failure surface.
Therefore, estimation of reliability requires the solution of a multidimen-
sional integral that can rarely be solved analytically. For this reason, other
methods such as numerical integration become essential. This might not be
practically feasible in probabilistic analysis because of the multi - dimensional
nature of the problem wherein a dimension is associated for each basic variable,
and the area of interest is usually in the tails of the distributions (Cronvall, 2011).
The uncertainty from all the sources which may affect that failure of the
component (or system) should be considered for a rigorous structural reliability
assessment.
This clearly involves taking into account all fundamental quantities entering
the problem, and also the uncertainties that arise from lack of knowledge
and idealized modeling. The structural reliability procedure is outlined by the
following steps (Cronvall, 2011):
(a) Identify all significant modes of failure of the structure or operation under
consideration, and define failure events.
(b) Formulate a failure criterion or failure function for each failure event.
(c) Identify the sources of uncertainty influencing the failure of the events,
model the basic variables and parameters in the failure functions and
specify their probability distributions.
(d) Calculate the probability of failure or reliability for each failure event,
and combine these probabilities where necessary to evaluate the failure
probability or reliability of the structural system.
(e) Consider the sensitivity of the reliability results to the input, such as basic
variables and parameters.
(f ) Assess whether the evaluated reliability is sufficient in comparison with a
target.
1.2.1 Levels of Reliability Methods

Structural reliability methods are divided into various levels which are charac-
terized by the extent of information about the structural problem that is used
and provided. The levels of reliability methods are given below (Madsen et al.,
1986):
(a) Level I methods : Reliability methods that employ only one ‘characteris-
tic’ value of each uncertain parameter are called level I methods. Examples
include load and resistance factor formats, including the allowable stress
formats.
(b) Level II methods : Reliability methods that employ two values of each
uncertain parameter (commonly mean and variance), supplemented with
a measure of the correlation between the parameters (usually covariance),
are called level II methods. Reliability index methods are examples of level
II methods.
(c) Level III methods : Reliability methods that employ probability of failure
as a measure, and which therefore require a knowledge of the joint distri-
bution of all uncertain parameters, are called level III methods.
(d) Level IV methods : Reliability method that compares a structural prospect
with a reference prospect according to the principles of engineering eco-
nomic analysis under uncertainty, considering costs and benefits, of con-
struction, maintenance, repair, consequences of failure, interest on capital,
etc., is called a level IV method. Such design methods are still in the process
of development.
1.3 Literature regarding the application of reliability concepts

for asphalt pavements
There is significant variability in the various input parameters involved in the
pavement design process. Many of these design inputs cannot be predicted
exactly due to lack of knowledge and information and uncertain future so-
cioeconomic conditions. Also, due to nonhomogeneous materials and variable
construction practices, there are inherent variations in the pavement strength.
This uncertainty in prediction and natural variation of input parameters results in
variable pavement system performance and early failures in pavements (Darter
and Hudson, 1973). Historically, variation or uncertainty (in design) has been
taken into consideration by the use of safety factors or arbitrary decisions based
on experience. However, the use of such safety factors or experience based
decisions without due consideration of input variables uncertainty has resulted
in few failures (Hudson, 1975).
Lemer and Moavenzadeh (1971) developed one of the first models dealing
with reliability of pavements. They pointed out that the factors affecting the
degree of variation in pavement system parameters have a significant effect on
system reliability.
The limit state function for the pavement reliability problem can be written
as:
D = log NF − log NA (1.5)
Where, NF = allowable number of axle load applications to failure
NA = number of actual axle load applications
The condition of the pavement is considered to have deteriorated below
acceptable limits when NA exceeds NF , or equivalently, D < 0. Assuming
lognormal distributions for NF and NA , the probability of failure is obtainable as
PF = φ(–β C ), where φ(.) is the cumulative distribution function of the standard
normal random variable and β C = E(D)/σ (D) is the reliability index, in which
E(D) and σ (D) are the mean and standard deviation of D (Darter and Hudson,
1973).
In the simulation model proposed by Alsherri and George (1988) for relia-
bility evaluation of pavements, the following equation based on present service-
ability index was used:

R = P p f ≥ pt (1.6)
where, pf = present serviceability index at time t, and pt = limiting (terminal)
serviceability index, generally set at 2.5 for AASHTO’s design and 3.0 for
premium design.
The following expression was used to estimate reliability under the assump-
tion that both pf and pt are normally distributed:
⎡ ⎤
⎢ μ p f − μ pt ⎥
R = ⎣ 1/2 ⎦
= (z0 ) (1.7)
σ p f + σ pt
2 2
where,
φ = standard normal distribution
μpf = mean value of pf
μpt = mean value of pt
σ pf = standard deviations of pf
σ pt = standard deviations of pt
z0 = standard normal deviate
In the AASHTO (1993) guide for the design of pavement structures, the
overall standard deviation of variation was considered by including the errors
in traffic predictions and in pavement performance prediction to analyze risk
and reliability in the design and reliability design factor was determined. The
reliability of design was defined as:
Reliability, R (percent ) = 100 × Probability (Actual pavement performance,

Nt ≥ Actual design period traffic in ESAL, NT )
(1.8)
The overall variance (S02 ) was defined as the sum of the variance in traffic
prediction (Sw2 ) and the variance in prediction of pavement performance (SN2 ).
S02 = Sw2 + SN2 (1.9)
The following equation for the reliability design factor (FR ) was derived:
FR = 10−ZR ×S0 (1.10)
Where, S0 is the overall standard deviation of variation and ZR is the standard

normal deviate.
2
Noureldin and his colleagues estimated the variance of traffic prediction (SW )
using the first order second moment approximation approach on the AASHTO’s
traffic prediction equation and the following was derived (Noureldin et al., 1996):

(COV. ADT∗ Dd )2 + (COV.P)2 + (COV.Ld )2 + (COV.TF)2
2
SW =
5.3
(1.11)
Where ADT∗ Dd represents average daily traffic in a heavier direction; P is the
percentage of trucks in the traffic mix; Ld is the lane distribution; TF is the truck
factor (number of ESALs per truck). The growth factor and the design period
were assumed to be constants.
Using AASHTO’s flexible pavement performance prediction model, the
variance of the pavement performance prediction S2N may be obtained as
(Noureldin et al., 1994) :
2 2 2
SN2 = COV(MR) + P2 SN .COV(SN) (1.12)
P2 = variance component of SN
To determine the COV(SN), the variance of SN was estimated in the following
way :
Var(SN) ∼
= ā21 Var(D1 ) + D̄21 Var(a1 ) + ā22 m̄22 Var(D2 ) + ā22 Var(m2 )D̄22
+ Var(a2 )m̄22 D̄22 + ā23 m̄23 Var(D3 ) + ā23 Var(m3 )D̄23 + Var(a3 )m̄23 D̄23
(1.13)
Kulkarni (1994) chose traffic as a design element for evaluating the reliability
of alternate pavement designs with different types of pavements. The reliability
R of a pavement design was defined as :

R = Probability (actual traffic load capacity,
(1.14)
N > actual cumulative traffic, n)
It was suggested that both ln N and ln n (‘ln’ indicates natural logarithm)
would follow normal distribution since N and n are log-normally distributed.
The safety margin (SM) of design was defined as:
SM = ln N − ln n (1.15)
The reliability index (β), is defined as the ratio of mean (E) and standard
deviation (SD) of safety margin (SM).
E[SM] E[ln N]−E[ln n]
β= =√ (1.16)
SD[SM] var[ln N]+var[ln n]
A mechanistic pavement model, WESLEA, and empirical transfer functions
were used to assess the effect of input variability on fatigue and rutting failure
models (Timm et al., 2000). Monte Carlo simulation technique was used to study
the uncertainty through a computer program called ROADENT. The reliability
computed through ROADENT was reported lesser than that computed through
AASHTO 1993 design guide.
Kim and Buch (2003) categorized the uncertainties affecting pavement per-
formance into the following four broad groups:
1. The difference in the basic properties of materials from one point to another
and fluctuation on material and cross-sectional properties due to construction
quality, termed as ‘spatial variability’
2. Random measurement error in determining the subgrade soil strength, traffic
volume estimation and other such factors, termed as ‘variability due to
imprecision in quantifying the parameters affecting pavement performance’
3. Assumption and idealization of a complex pavement analysis model with
simple mathematical expression, termed as ‘model bias (error)’
4. Lack of fit of regression models, termed as ‘statistical error’
The first two groups are called ‘uncertainties of design parameters’ and the
last two groups are called ‘systematic errors’. Uncertainties of design param-
eters cause the variation within the probability distribution of the performance
function, whereas systematic errors cause the variation in possible location of
the probability distribution of the performance function.
Kim and Buch (2003) proposed a load and resistance-factor-design (LRFD-
format-based practical reliability method. A total of 13 pavement sections were
designed using AASHTO method and were also redesigned using reliability-
based design procedure where the AC thickness was changed so that the revised
section would accommodate the design traffic and satisfy the threshold rut-depth
at a given target reliability. The target reliability of 90 % was assigned to both
design procedures. The reliability indices for the pavement sections determined
by both methods were computed using the first order reliability method (FORM).
It was indicated that the RBD procedure does successfully yield cross-sections
whose reliability indices are close to the target reliability index, while the
AASHTO method does not generally produce designs of uniform reliability for
actual mechanistic failure criterion.
To incorporate reliability in pavement design, Austroads used the laboratory
fatigue relationship published by Shell Petroleum (Shell, 1978), which was
further modified to include a reliability factor (RF) corresponding to the desired
project reliability.
5
6918(0.856VB + 1.08
N = RF (1.17)
mix με
S0.36
Where, N = allowable number of repetitions of the load

με = tensile strain produced by the load (in microstrain)
VB = percent by volume of bitumen in the asphalt (%)
Smix = asphalt modulus (in MPa), and
RF = reliability factor for asphalt fatigue
The value of reliability factor (RF) varies from 2.5 to 0.67 for desired
project reliability of 80% and 97.5%. The higher the desired project reliability,
the lower is the value of the reliability factor (RF). Permanent deformation
was not considered as a distress mode in the Austroads design model due to
the non-availability of an appropriate model which could reliably predict the
development of rutting with the passage of traffic/time, as mentioned in the guide
(Austroads, 2012).
The NCHRP (2004) guide for the mechanistic-empirical (M-E) design of
new and rehabilitated pavement structures analyzes the reliability of flexible
pavement design for individual pavement distresses, such as asphaltic concrete
fatigue (bottom up) cracking, longitudinal (top down) cracking, rutting or as-
phaltic concrete thermal cracking. The reliability (R) in general is defined as the
probability that the particular distress of a design project is less than the critical
level of distress over the life of the design.
Kim (2006) presented a practical probabilistic design format to incorporate
reliability in the M-E flexible pavement design procedure. It was suggested that
uncertainties due to spatial variation and imprecision in quantifying parameters
should be integrated as parameter uncertainties and quantified in terms of the
standard deviation (Sp ) of pavement performance. Similarly, it was suggested
that model bias and statistical error should be integrated as systematic error and
quantified in terms of the standard deviation (Sm ) of pavement performance. The
overall standard deviation (S0 ) was determined as follows:

S0 = Sp2 + Sm
2 (1.18)
The study suggested the following reliability-based pavement-design equa-

tion with a target reliability, R, using a rut prediction model:
RDmax = S0 × βtarget + RDpredicted (1.19)
Where, S0 = overall standard deviation, and β target = the target reliability
index
The depth of 12.7 mm was considered as a limit state (RDthreshold ). The value
of the difference between RDthreshold and RDmax was computed and compared
with the specified tolerance level. It was suggested that the design should be
changed until this criterion is satisfied.
1.4 Issues with estimation of pavement reliability

Though the available methods for reliability analysis can be applied for the
pavement reliability estimation, however there are still several concerns regard-
ing this. The following subsections briefly discuss the issues related with the
estimation of reliability for pavements and its incorporation in the pavement
design process.
1.4.1 Input Parameters Variability

For the reliability analysis, it is necessary to know all these inputs accurately
and without error. A number of parameters are required for the pavement design
process; all of these are uncertain and have some variability. Variability exists
in pavements due to material characteristics, traffic conditions, environmen-
tal conditions, construction practices and quality control (Darter and Hudson,
1973). Variability of the input variables may be described by statistical terms
(such as mean and variance) and the associated probability distribution. A useful
dimensionless parameter that indicates the variability of a material’s property is
the ratio of standard deviation over the mean, known as coefficient of variation
(COV). Knowledge of the COV of each design input is extremely important to
accurately estimate their influence on the predicted pavement life and, thereby,
reducing the chances of premature failure.
To simplify the calculation process, the input variables are considered to
be normally distributed which is not true always. So this consideration of
normal distribution for non-normal random variables introduces an error of
approximation in the reliability analysis. In addition to this, the interrelationship
and interaction between the input variables is not known and even if known, is
generally not used in the reliability analysis.
For the sake of simplicity, the input variables are considered to be independent
of each other, which is not true. This “blind’ approach doesn’t render accurate
predictions of reliability analysis.
A summary of the variability associated with the pavement input parameters
from the available literature is given in Table 1.1.
TABLE 1.1 Summary of pavement material variability from available literature
Parameter Description Range of Typical Distribution Reference

COV (%) COV (%) Type
Layer Bituminous 3 – 12 7 Normal Timm et. al., 1999;
thickness surface Noureldin et. al.
1994
3.2 – 18.4 7.2 Normal Aguiar-Moya and
Prozzi, 2011
Bituminous 11.7 – 16.0 13.8 Normal Aguiar-Moya and
binder Prozzi, 2011
course
5 – 15 10 Normal Noureldin et. al.
1994
Granular 10 – 15 12 Normal Timm et. al. 1999;
base Noureldin et. al.
1994
6.0 – 17.2 10.3 Normal Aguiar-Moya and
Prozzi, 2011
Granular 10 – 20 15 Normal Timm et. al., 1999;
sub-base Noureldin et. al.,
1994
Elastic Bituminous 10 – 20 15 Normal Noureldin et. al.,
modulus layer 1994
10 – 40 Lognormal Timm et. al., 1999
Granular 10 – 30 20 Normal Noureldin et. al.,
base 1994
5 – 60 Lognormal Timm et. al. 2000
Granular 10 – 30 20 Normal Noureldin et. al.,
sub-base 1994
Subgrade 10 – 30 20 Normal Noureldin et. al.,
1994
Traffic Extreme Timm et. al., 2000
Value Type I
1.4.2 Performance models

As mentioned previously, the current Indian pavement design guidelines
(IRC:37, 2018), which is a mechanistic-empirical approach, is based on two
performance criteria, viz., fatigue and rutting model, for a conventional three-
layered pavement structure. In simpler terms, the designed pavement must last
till the fatigue cracking of bituminous surface or rutting in the pavement reaches
its terminal values, whichever happens earlier.
The general form of these fatigue and rutting models is given below:
Fatigue Model
k2
1 1 k3
N f = k1 × C × × (26)
εt MR
C = 10M (27)

Vbe
M = 4.84 × − 0.69 (28)
Vbe + Va
Rutting Model
k5
1
Nr = k4 × (29)
εv
Where,
Nf = fatigue life of bituminous layer in terms of cumulative repetitions of
equivalent 80 kN standard axle load
Nr = subgrade rutting life in terms of cumulative repetitions of equivalent 80
kN standard axle load
εt = maximum horizontal tensile strain at the bottom of bituminous layer
εz = maximum vertical compressive strain at the top of the subgrade
MR = resilient modulus of the bituminous layer (MPa)
Vbe = percent volume of effective bitumen in the mix used in the bituminous
layer (varies between 3.5 to 4.5 percent)
Va = percent volume of air voids in the mix used in the bituminous layer
(varies between 10.5 to 11.5 percent)
ki = regression coefficients (i = 1 to 5)
The values of these regression coefficients are given as : k1 = 1.6064 × 10−4
and 0.5161 × 10−4 for design traffic ≥ 20 msa and < 20 msa respectively;
k2 = 3.89; k3 = 0.854; k4 = 4.1656 × 10−8 and 1.41 × 10−8 for design traffic
≥ 20 msa and < 20 msa respectively; and k5 = 4.5337.
These fatigue and rutting transfer functions were developed and calibrated
during the database collected through R-6 and R-19 research studies sponsored
by MORTH. A total of around 120 numbers of bituminous concrete (BC) and 160
numbers of bituminous macadam (BM) road sections from R-6 and R-19 studies
were considered for development of fatigue criterion and 86 number of BC road
sections from R-6 study were analyzed for the development of rutting criterion.
These pavement sections consisted of bituminous surfacing with granular bases
and subbases and they were assumed as three layered structure. The average
annual pavement temperature (AAPT) of all the sections were around 35°C
and the bitumen of 80/100 penetration grade was used for both BC and BM
surfacing. The maximum repetitions of equivalent single-axle load for the road
sections were 50 msa only. The thickness of BC layer was 40 mm on most of
the sections and BM was used as the bituminous binder course, just below the
BC layer. In the collected performance data, the scatter of data points was quite
large. This was attributed to the fact that the test pavements were located in
different parts of the country and they have been constructed in different climatic
conditions and probably quality control during the construction was not identical
resulting in wide variations. Because of the variabilities involved, wide scatter
was considered to be quite in order.
However, the present situation is totally different from what it was when these
transfer functions were developed and calibrated. The traffic on the pavements
has increased tremendously, both in terms of loading and number of repetitions.
The specifications of bituminous materials have changed. Flexible pavements
with thick bituminous layers are quite common these days. So, to what extent
the originally developed fatigue and rutting transfer functions are valid, is a
matter of concern and debate. There is a need for recalibration of the fatigue and
rutting transfer functions. Any changes to these transfer functions due to this
recalibration may affect the overall pavement thickness and ultimately affects
the reliability of the pavement. So, there is a need for recalibration of the fatigue
and rutting transfer functions. Different researchers have proposed different
coefficients for the fatigue and rutting equations.
The development of rutting and fatigue transfer functions is through field
calibration by ordinary least-square estimation (OLSE) technique/modelling of
the field data. The approach of OLSE in developing these equations itself violates
the basic assumptions of OLSE, due to the presence of measurement errors.
A better statistical method of functional linear-measurement error (FLME) has
been presented (Shukla and Das, 2008), which may be used for development of
fatigue and rutting equations.
1.4.3 Interaction between the failure modes

A pavement has many failure modes, the interrelationship between these failure
modes is quite complex. So it is difficult to consider such complexity in the
reliability model. Often, this complexity is simplified and for this reason the
analysis is not very certain/accurate.
For pavements, the two primary modes of failure are fatigue cracking and
rutting (IRC:37, 2018). Generally, these two are considered as independent and
in series. In simple words, this indicates that that pavement may fail by any of
the failure modes, viz. fatigue failure of rutting failure and the two failure modes
occur totally independent of each other.
However, very few studies have been done to establish the possible correlation
between the two failure modes (Dilip et al. 2013; Gogoi et al. 2013; Liu and Xu
2014; Peddinti et al. 2020). These studies have indicated that there is a possible
correlation between the two failure modes.
This needs to be taken into consideration during the reliability analysis.
However, the problem lies in the fact that the amount of correlation between
the two failure modes is not precisely known as it depends on many factors and
is still a matter of research.
1.4.4 Material strength degradation

A typical pavement structure consists of one or two layers of bituminous mate-
rials and two or more layers of unbound granular materials. Due to the action
of traffic and environmental conditions, a pavement deteriorates over a period
of time and its reliability decreases over time or load history. This decrease
in reliability may be a result of the degradation in strength of the constituent
pavement materials. However, the mechanical behavior of both the materials,
i.e. bituminous and granular materials, is quite different. Bituminous materials
are subjected to traffic induced fatigue cracking. The granular layer’s resistance
to traffic loading is commonly measured in terms of permanent deformation at
the base, which leads also to permanent deformation on the surface. To capture
the effect of material strength degradation, time dependent reliability for both
fatigue, and rutting failure modes need to be done.
1.5 Conclusions
The available methods for reliability analysis can be applied for the reliability
analysis of asphalt pavements. However, the issues indicated in the previous
section need to be taken care of during the application of such reliability analysis
for asphalt pavements.
Acknowledgements
The author would like to thank Prof. (Dr.) Satish Chandra, Director, CSIR-
Central Road Research Institute (CSIR-CRRI), New Delhi for his kind permis-
sion to publish this paper.
Disclosure statement
No potential conflict of interest was reported by the authors.
References
AASHTO, 1993. AASHTO Guide for Design of Pavement Structures. American Association of State
Highway and Transportation Officials, Washington, D.C.
Aguiar-Moya, J.P., Prozzi, J., 2011. Development of reliable pavement models. Texas Transportation
Institute, Texas A&M UNiversity System, College Station, Texas, USA.
Alsherri, A., George, K.P., 1988. Reliability model for pavement performance. J. Trans. Eng. 114
(2), 294–306.
Austroads, 2012. Guide to Pavement Technology Part 2 : Pavement Structural Design. AGPT 02-12.
Austroads Ltd., Sydney, Australia.
Cronvall, O., 2011. Structural Lifetime, Reliability and Risk Analysis Approaches for Power Plant
Components and Systems. In: VTT Publications, 775. VTT Technical Research Centre of
Finland, Vuorimiehentie.
Darter, M.I., Hudson, W.R., 1973. Probabilistic Design Concepts Applied to Flexible Pavement
System Design. Centre for Highway Research, University of Texas at, Austin, Texas, USA.
Dilip, D.M., Ravi, P., Babu, G.L.S., 2013. System reliability analysis of flexible pavements. J.
Transport. Eng. 139 (10), 1001–1009.
Gogoi, R., Das, A., Chakroborty, P., 2013. Are fatigue and rutting distress modes related? Int. J.
Pavement Res. Technol. 6 (4), 269–273.
Hudson, W.R., 1975. State of the Art in pedicting pavement reliability from input variability.
Department of Transportation, Federal Aviation Administration, Washington, D.C.
IRC:37, 2018. Guidelines for the design of flexible pavements. Indian Roads Congress, New Delhi,
India.
Kim, H.B., Buch, N., 2003. Reliability based pavement design model accounting for inherent
variation of design parameters. 82nd Transport. Res. Board Annu. Meet., Washington, D. C.
Kulkarni, R.B., 1994. Rational approach in applying reliability theory to pavement structural design.
Transportation Research Record, Transportation Research Board. National Research Council,
Washington D.C., USA, pp. 13–17 1449.
Lemer, A.C., Moavenzadeh, F., 1971. Reliability of highway pavements. Highway Res. Rec. 36, 1–8.
Liu, H., Xu, X., 2014. Reliability analysis of asphalt pavement considering two failure modes. In:
Mohammadian, K., Goulias, K.G., Cicek, E., Jieh-JiuhWang, Maraveas, C. (Eds.), Proceedings
of 3rd International Conference on Civil Engineering and Urban Planning III. CRC Press,
Balkema, The Netherlands, pp. 291–295.
Madsen, H.O., Krenk, S., Lind, N.C., 1986. Methods of Structural Safety. Prentice-Hall Inc.,
Englewood Cliffs, New Jersey, USA.
Modarres, M., Kaminskiy, M., Krivtsov, V., 1999. Reliability Engineering and Risk Analysis - A
Practical Guide. Marcel Dekker Inc., New York, USA.
NCHRP, 2004. Guide for mechanistic-empirical design of new and rehabilitated pavement struc-
tures. NCHRP Research Report 1-37A, Transportation Research Board, National Research
Council, Washington, D.C.
Noureldin, A.S., Sharaf, E., Arafah, A., Al-Sugair, F., 1996. Rational Selection of Factors of Safety
in Reliability-Based Design of Flexible Pavements in Saudi Arabia. Transport. Res. Rec.: J.
Transport. Res. Board, SAGE Publications 1540 (1), 39–47.
Noureldin, A.S., Sharaf, E., Arafah, A., Faisal, A.-S., 1994. Estimation of Standard Deviation of
Predicted Performance of Flexible Pavements Using AASHTO Model. Transport. Res. Rec.
1449, 46–56.
Peddinti, P.R.T., Munwar Basha, B., Saride, S., 2020. System Reliability Framework for Design
of Flexible Pavements. J.Transport. Eng. Part B: Pavements 146 (3) American Society of Civil
Engineers (ASCE).
Shell, 1978. Shell Pavement Design Manual - Asphalt Pavements and Overlays for Road Traffic.
Shell International Petroleum Limited, London, U.K.
Shukla, P.K., Das, A., 2008. A re-visit to the development of fatigue and rutting equations used for
asphalt pavement design. Int. J. Pavement Eng. 9 (5), 355–364.
Timm, D.H., Birgisson, B., Newcomb, D.E., Galambos, T.V., 1999. Incorporation of reliability
into the Minnesota mechanistic-empirical pavement design method. Final Report, Minnesota
Department of Transportation, Minnesota.
Timm, D., Newcomb, D., Galambos, T., 2000. Incorporation of reliability into mechanistic-empirical
pavement design. Transport. Res. Rec. 1730 (1), 73–80.
Non-Print Items
Abstract
The variability of the inputs parameters is not considered during the pavement
design process. As the input parameters are uncertain, this needs to be accounted
for during the design process through the reliability analysis approach. However,
the application of reliability concepts to pavements is not very straight forward and
many aspects need to looked into. The present paper attempts to summarize such
concepts and applications of reliability concepts for asphalt pavement design.
Keywords
Asphalt; Material strength degradation; Mechanical behavior; Pavement design;
Reliability; Flexible Pavement
Chapter 2
Markov modeling of multi-state

systems with simultaneous
component failures/repairs,
using an extended concept of
component importance
Jacek Malinowski
Systems Research Institute, Polish Academy of Sciences, Newelska 6, 01-447 Warszawa, Poland
2.1 Introduction
The concept of component importance is well-known among reliability en-
gineers and is mostly used in the context of two-state systems with two-
state components. There exist several different importance measures, e.g. Birn-
baum, Fussel-Vesely, Risk Achievement Worth, Risk Reduction Worth, etc. (see
Hoyland and Rausand [2009] for definitions and explanations). A comprehensive
synopsis of the topic is given in the recent survey [Kalpesh and Kirtee, 2017].
Several types of component importance are also discussed in the monograph
[Kuo and Zhu, 2012]. Finding the importance’s values may not be a simple task
for complex systems, but the effort can be worthwhile. For example, high values
can indicate critical locations in the system structure, where highly reliable
components should be placed in order to reduce or minimize the risk of system
failure. Also, as shown in this chapter, they can be essential in computing useful
reliability characteristics such as interstate transition intensities for multistate
systems with repairable components.
In this chapter, importance is attributed to a group of components rather than
to a single component alone. It is defined as the probability that simultaneous
failure or repair of all components in a set results in a system transition from
state a to state b, provided that the components are in “up” or “down” state
and a>b or a<b respectively. It will be denoted as Ia→b () and referred to as
the importance of the component set to a transition from state a to state b.
Let us stress that the considered system is assumed to be multistate and its
Safety and Reliability Modeling and Its Applications. DOI: 10.1016/B978-0-12-823323-8.00018-0
Copyright © 2021 Elsevier Inc. All rights reserved. 15
states to be partially ordered, where < is a strict-partial order relation. Thus

defined importance can be used in modeling dependent failures, e.g. cascading
or common-cause ones. The more detailed definition of Ia→b () will be given in
the next section. It is worth noting that some authors (see [Eryilmaz et al., 2016,
Lin et al., 2016], and [Miziula and Navarro, 2019]) also studied the importance
of mutually dependent components, although they defined their dependence
differently than herein. Also note that Ia→b () is an extension of the well-
established Birnbaum importance IB (c) equal to the probability that component
c is critical to the considered system, i.e. a failure or repair of c results in a failure
or repair of the whole system. However, IB (c) is only well-defined for two-state
systems with independent components, hence its applicability is limited.
The concept of multicomponent importance proves useful in reliability anal-
ysis of a system with certain dependencies between its components, whose
operation is modeled by a Markov chain on a finite state space. As will be shown,
this importance can be applied to conveniently express the intensity of transition
from state a to b, denoted as a→b (t) and defined as follows:
Pr[(Xt+t ) = b | (Xt ) = a]
a→b (t ) = lim (2.1)
t→0 t
Here, Xt is a vector of binary random variables representing the components’
states at time t, and is a function relating them to the system state. As will
be shown, a→b (t) is a useful key parameter for Markov modeled systems. A
comprehensive reminder of basic facts from the theory of Markov processes,
presented in the context of reliability models, can be found in [Hoyland and
Rausand, 2009].
Let us now briefly outline the paper’s contents. Sections 2.2 and 2.3 constitute
the theoretical background of the paper. Section 2.3 includes the author’s main
result, i.e. Theorem 2.1 stating how to express a→b in terms of Ia→b (). As
already mentioned, a→b is the key reliability parameter to be computed. In
Section 2.4 it is shown how a system operating according to the assumptions
given in Section 2.2 can be modeled by a Markov chain. First, transition
diagrams illustrating the system’s operation process are constructed, both on
the component-state and system-state level. Next, the Kolmogorov equations are
solved in order to find the state probabilities with which the system occupies
its individual states at a given time point. In Section 2.5 the state probabilities
are used to find all the nonzero importances Ia→b () which are subsequently
used to compute the intensities a→b with the formula given by Theorem 2.1. In
Section 2.6, it is demonstrated how the transition intensities a→b can be applied
to obtain several useful reliability parameters, e.g. the expected uninterrupted
residence time of the system in state a, or the expected number of transitions
from state a to state b in a time interval of a given length.
The presented concepts and results, including the computation of multicom-
ponent importances, transition intensities, and other reliability parameters, will
be illustrated by an explanatory example of a simple power supply system.
Markov modeling of multi-state systems with simultaneous Chapter | 2 17
2.2 Basic assumptions, notation and definitions

We will consider a Markov model of a technical system composed of n repairable
components with constant failure and repair rates. Each component can be either
in state 1 (operational) or 0 (failed). In turn, the system is a multistate one, i.e. its
individual states are the elements of a finite set S. The system state depends on
the component states, and that dependence is expressed by the structure function
mapping binary vectors of length n to the elements of S. It is assumed that
simultaneous failures or repairs of multiple components are possible (cascading
or common cause failures and parallel repairs), and components may have to
wait in a repair queue (limited maintenance personnel). Thus, the assumption of
components independence, frequently adopted in the literature, is relaxed.
The following notation will be used throughout the paper
{1,…,n} : set of the system components
: a nonempty subset of {1,…,n} xi : binary variable representing the state
of ei ; xi = 1/0 if ei is operational/failed x : vector of the component states,
x = (x1 ,…,xn )
{0,1}n : set of binary vectors of length n
D(x,y) : the set of coordinates in which x differs from y, where x,y ∈ {0,1}n
Xi (t) : state of ei at time t; Xi (t) is a random variable
Xt : vector of the component states at time t; Xt = [X1 (t),…,Xn (t)]
Px (t) : probability that at time t the component states are given by x, i.e.
Px (t) = Pr[Xi (t) = xi , I = 1,…,n]
λ / μ : rate of simultaneous failure/repair of all components in defined
as follows:
1
λ = lim Pr [Xi (t + t ) = 0, i ∈ |Xi (t ) = 1, i ∈ ] (2.2)
t→0 t
1
μ = lim Pr [Xi (t + t ) = 1, i ∈ |Xi (t ) = 0, i ∈ ] (2.3)
t→0 t
S : partially ordered set of the system states

(x) : order-preserving function relating the component states to the system
state
≤, < : nonstrict and strict partial order relation in {0,1}n or S
→ : direct transition relation between different elements of {0,1}n or S,
defined respectively as follows: x→y if x>y and λD(x,y) > 0, or x<y and μD(x,y)
> 0 a→b if a=b and there exist x,y∈{0,1}n such that x→y, (x)=a and (y)=b
The last four definitions require some clarification. The partial order intro-
duced in S allows to compare different levels of the system’s operating ability.
The order-preservation property is an extension of the monotonicity property of a
binary structure function defined in [Barlow and Proschan, 1975]. This property
says that failure/repair of a component cannot improve/deteriorate the system
state, which is intuitively obvious, but has to be expressed mathematically.
is an order-preserving function if x<y implies (x)≤(y), where ≤ (<) is a
nonstrict (strict) partial order relation in {0,1}n or S. We adopt the usual partial
order in {0,1}n , i.e. x ≤ y if xi ≤ yi for each i∈{1,…,n}, and x < y if x ≤ y and

xi < yi for at least one i∈{1,…,n}. Thus, the concept of a binary nondecreasing
function is extended to the nonbinary case.
As to the “→” relation, we can regard Xt as a series of events which occur as
failures or repairs of single components, or as simultaneous failures of repairs
of multiple ones, and the time-to-occurrence of the next event only depends on
the current state of Xt . Thus, Xt is a Markov chain with the state space {0,1}n ,
and the intensity of direct transition from state x to state y is equal to λD(x,y) or
μD(x,y) , depending on whether x>y or x<y. Such a transition is only possible if
λD(x,y) > 0 or μD(x,y) > 0 respectively, which explains the definition of the “→”
relation in {0,1}n . The definition of “→” in S is a natural consequence of the
previous one. According to it, a direct transition from a to b in S is realized by
each direct transition from x to y in {0,1}n , such that (x) = a, (y) = b. Also,
if no x and y exist such that x→y, (x) = a, (y) = b, then no direct transition
from a to b can take place.
Further, we assume that λ and μ defined by (2) and (3) are given data
obtained by statistical estimation or experts’ elicitation.
We now continue with the remaining notation used in the paper:
Zt : system state at time t, i.e. Zt = (Xt )
a→b (t) – transition intensity with which the system changes its state from a
to b, defined as follows:
1
a→b (t ) = lim Pr [Zt+t = b | Zt = a]; a, bS (2.4)
t→0 t
(x, 1 ) – binary vector obtained from x, such that xi =1, i∈

(x, 0 ) – binary vector obtained from x, such that xi =0, i∈
a→b crit () – set of binary vectors x such that xi =1 for i∈, (x)=a and
(x,0 )=b, where a>b; vectors in a→b crit () will be called critical to a direct
transition from a to b, caused by simultaneous failure of all components in
a→b crit () – set of binary vectors x such that xi = 0 for i∈, (x) = a and
(x,1 ) = b, where a<b; vectors in a→b crit () will be called critical to a direct
transition from a to b, caused by simultaneous repair of all components in
Ia→b (,t) – importance of to a transition from a to b, given that a>b, defined
as follows:

Ia→b (, t ) = Pr Xt ∈ crit
a→b () | Xi (t ) = 1, i ∈ (2.5)
i.e. Ia→b (,t) is the (conditional) probability that simultaneous failure of all
components in causes a transition from a to b, given that these components
are operable
Ia→b (,t) – importance of to a transition from a to b, given that a < b,
defined as follows:

Ia→b (, t ) = Pr Xt ∈ crit
a→b () | Xi (t ) = 0, i ∈ (2.6)
2 3
FIG. 2.1 The RBD of a mission-critical power supply system.
i.e. Ia→b (,t) is the (conditional) probability that simultaneous repair of all
components in causes a transition from a to b, given that these components
have failed.
If is a binary function (i.e. S={0,1}) and is a one-element set (i.e. ={i}),
then the quantities defined by (5) and (6) reduce to the Birnbaum importance of
component i.
For better understanding of the introduced concepts, let us consider a power
supply system for a critical device. It consists of mains (1), a standby generator
(2), and a UPS (3) that, in case of mains outage, supplies power from batteries
for the time necessary to start the generator. The system’s RBD is shown in
Fig. 2.1.
Let S={0,1,2}, i.e. the system is in state 0 if power is not supplied to the
device, in state 1 if supplied from UPS batteries or standby generator, and in
state 2 if supplied from mains. The RBD in Fig. 2.1 implies that
(0,0,0) = (0,0,1) = (0,1,0) = 0,
(0,1,1) = 1,
(1,0,0) = (1,0,1) = (1,1,0) = (1,1,1) = 2.
If we assume that the system is fully/partly operational in state 2/1, then
the inequalities 0 < 1 < 2 define the ensuing partial order in S (note that this
order is linear, because each two elements of S are comparable). It is easily
checked that is an order-preserving function from {0,1}3 to S. We also assume
for simplicity that the system components fail independently (no simultaneous
failures) and are nonrepairable. The transition diagrams for Xt and Zt are shown
in Fig. 2.2 and 2.3. An arrow between two states indicates that they are in the
transition relation. Analyzing these figures, we conclude that (1,1,1)→(0,1,1)
yields 2→1; (1,0,0)→(0,0,0), (1,0,1)→(0,0,1), or (1,1,0)→(0,1,0) yields 2→0;
and (0,1,1)→(0,0,1) or (0,1,1)→(0,1,0) yields 1→0.
Note that it is possible to define a nonorder-preserving function from {0,1}n
to S, but it may be difficult to interpret physically. For example, if n and S are as
in the above example, then defined below is such a function.
(0,0,0) = (0,0,1) = (0,1,0) = 0,
(0,1,1) = (1,0,1) = 1,
(1,0,0) = (1,1,0) = (1,1,1) = 2.
Indeed, (1,0,0) < (1,0,1), but (1,0,0) > (1,0,1), hence is not order-
preserving.
(1,1,1)
(0,1,1) (1,0,1) (1,1,0)
(0,0,1) (0,1,0) (1,0,0)
(0,0,0)
FIG. 2.2 Transition diagram of Xt .
(1,0,0), (1,0,1)
(1,1,0), (1,1,1)
φ=2
(0,1,1)
φ=1
(0,0,0), (0,0,1)
(0,1,0)
φ=0
FIG. 2.3 Transition diagram of Zt = (Xt ).
2.3 Theoretical background

We begin this section with an auxiliary lemma that will be used to prove the main
theorem.
Lemma 1
Let be an order-preserving function from {0,1}n to S, a,b∈S, and a→b.
Then either a>b or a<b.
Proof: Let a,b∈S and a→b. It then holds that a=b and there exist x∈–1 (a)
and y∈–1 (b) such that x→y. The definition of “→” in {0,1}n implies that either
x>y or x<y. Hence, as is order-preserving, either (x) ≥ (y) or (x) ≤
(y). In consequence, either a>b or a<b, because (x) = a, (y) = b and a =
b.
Lemma 1 justifies the definitions of critical vectors and importances, which
assume that a direct transition from a to b in S can only take place if a>b or a<b.
Theorem 1
Let Xt be the earlier defined Markov chain with states in {0,1}n , and let
Zt = (Xt ), where is an order-preserving function from {0,1}n to S. If
a→b, where a,b∈S, then the transition intensity a→b (t) is given by one of the
following formulas:
1

a→b (t ) = λ Ia→b (, t ), a > b (2.7)
Pr [Zt = a] ⊆{1,...,n}
1

a→b (t ) = μ Ia→b (, t ), a < b (2.8)
Pr [Zt = a] ⊆{1,...,n}
where λ in (7) or in (8) is the rate of simultaneous failure or repair of all

components in , and

Ia→b (, t) = Ia→b (, t) · Pr[Xk (t ) = 1, k] = Pr Xt ∈ crit
a→b () , a > b
(2.9)

Ia→b (, t) = Ia→b (, t) · Pr[Xk (t ) = 0, k] = Pr Xt ∈ b→a () , a < b
crit
(2.10)
It is more convenient to use I a→b () instead of Ia→b () in (7) and (8), because
I a→b () is usually easier to compute than Ia→b (). It should be noted that the
above formulas are generalizations of similar ones for a two-state system with
independent components, to be found in [Korczak, 2007].
Proof: We only consider the cases a>b and a<b, because Lemma 1 states that
if a→b then either a>b or a<b. Let us assume that a>b in order to prove (7).
The law of total probability yields:
Pr [Zt+t = b, Zt = a] =
= Pr [Xt+t = y, Xt = x]
x,y: (x)=a,
(y)=b
= Pr [Xt+t = y, Xt = x]+ (2.11)
x,y: x>y, (x)=a,
(y)=b
+ Pr [Xt+t = y, Xt = x]
x,y: x>y, (x)=a, (y)=b
If x≯y and (x)=a>b=(y) then a transition from a to b cannot be realized

as a single event in Xt . Indeed, each such event occurs as a simultaneous failure
or repair of one or multiple components, i.e. as a direct transition from x to y in
{0,1}n such that x>y or x<y. However, x>y contradicts x≯y, and x<y implies
a=(x)≤(y)=b which contradicts a>b. Hence, if x≯y and (x)=a>b=(y)
then a transition from a to b can only be realized as multiple direct transitions in
{0,1}n . Since Xt is a Markov chain, the probability that such transitions occur in
an infinitesimal time interval, divided by this interval’s length, is equal to zero.
In consequence we have:
1
lim Pr [Xt+t = y | Xt = x] = 0 (2.12)
t→0 t
x,y: x>y, (x)=a, (y)=b
i.e. the second sum on the right hand side of (11) is equal to zero. Expanding the
first sum we get:

Pr [Xt+t = y, Xt = x]
x,y: x>y, (x)=a,
(y)=b
= Pr [Xt+t = y, Xt = x]
⊆{1,...,n}, x>y, D(x,y)=, (x)=a, (y)=b
=∅ x,y:
= Pr [Xt+t = (x, 0 ), Xt = x]
⊆{1,...,n}, =∅ x∈crit

a→b ()
= Pr [Xt+t = (x, 0 ) | Xt = x] Pr [Xt = x]
⊆{1,...,n}, =∅ x∈crit
a→b ()
(2.13)
Let us note that for x∈a→b crit () it holds that
Pr [Xt+t = (x, 0 ) | Xt = x]
(2.14)
= Pr [Xk (t + t ) = 0, k ∈ | Xk (t ) = 1, k ∈ ]
According to the definition of λ , the right hand side of (14), divided by t,
converges to λ as t→0. Thus, for x∈a→b crit () we have:
1
limPr [Xt+t = (x, 0 ) | Xt = x] = λ (15)
t→0 t
In consequence, from (1), (11)-(13), and (15) we obtain:

a→b (t ) = lim 1
Pr [Zt+t = b | Zt = a]
t→0 t
= 1
lim 1 Pr [Zt+t = b, Zt = a]
Pr[Zt =a] t→0 t

= 1
Pr[Zt =a]
λ Pr [Xt = x] (2.16)
⊆{1,...,n}, =∅ x∈crit
a→b
()

= 1
Pr[Zt =a]
λ Pr Xt ∈ crit a→b ()
⊆{1,...,n}, =∅
This completes the proof of (7). The proof of (8) is analogous.

Important remark: the formulas (7) and (8) are only valid if λ and μ do not
depend on the states of components outside . Otherwise, according to the third
equality in (16), the expression under the sum in (7) or (8) must be changed to
either of the following ones:

λ (x) Pr [Xt = x], a > b (2.17)

x∈ crit ()
a→b
μ (x) Pr [Xt = x], a < b (2.18)
x∈crit
a→b ()
where λ (x) and μ (x) depend on xi , i∈. However, (7) or (8) still holds if λ (x)
or μ (x) are equal for x∈crit a→b () or x∈crit a→b (), because λ or μ can
then be written without the variable x. For better explanation, this issue will also
be addressed in the next two sections.
An important conclusion can be drawn from Theorem 1. If is an order-
preserving function from {0,1} to S, then Zt is a (nonhomogenous) Markov
process with the transition intensities given by (7) and (8). Indeed, Pr(Zt = a)
μS
111 110
λB(1− πS)
λBπS μE
μE λB
μB μB
011 101 010 100
λE λB λB
λE
μB μB
001 000
FIG. 2.4 Transition diagram of Xt .
and Ia→b (,t) are functions of the state probabilities Pr(Xt =x), x∈{0,1}n , which,
as solutions of the Kolmogorov equations, only depend on t and constant
failure/repair rates λ ,/μ , ⊆{1,…,n}. In view of (7) and (8), the same property
holds for transition intensities a→b (t), i.e. they do not depend on the history
of Zt =(Xt ) before time t. Moreover, a→b (t) converge to constant values as
t→∞, because the probabilities Pr(Xt =x) also do. Zt is thus asymptotically
homogenous.
2.4 The illustrative model of an example system

For further considerations we will use the model of a three-state power supply
system composed of basic (B) and emergency (E) power sources, and the switch
(S) automatically activating E when B fails and E is in operable condition.
We assume that emergency source is in cold standby while basic source is in
operation, each source can only fail during operation, and the switch can only
fail, with probability π S , when activating emergency source in case of main
source’s failure. Thus, the components are mutually dependent, because a failure
of S, if occurs, follows that of B, and failure of E can only occur if B is out of
operation. We also assume that no simultaneous repairs are possible, and the
order of repair priorities is B, E, and S. Let λB denote the failure rate of B, then
λB Å(1–π S ) is the failure rate of B alone, provided that S is operable, and λB Åπ S
is the rate of simultaneous failure of B and S. Failure rate of E and the repair
rates are denoted with λ and with the respective indices, i.e. λE , μB , μE , μS . As
stated in Section 2, all the failure and repair rates are assumed to be given data.
In Fig. 2.4 we can see the transition diagram of Xt =[X1 (t), X2 (t), X3 (t)], where
X1 (t), X2 (t), X3 (t) are respectively the states of B, E, S at time t.
The adopted assumptions yield that λ{1} depends on the state of S, i.e.
λ{1} (1,1,1)=λB Å(1 – π S ) and λ{1} (1,1,0) = λ{1} (1,0,1) = λ{1} (1,0,0) = λB .
Since simultaneous failure of B and S can only happen when E is in operable
condition, λ{1,3} depends on the state of E, i.e. λ{1,3} (1,1,1) = λB Åπ S and
λ{1,3} (1,0,1) = 0. A failure of E is only possible when B is out of operation,
hence λ{2} (0,1,1) = λ{2} (0,1,0) = λE , but λ{2} (1,1,1) = λ{2} (1,1,0) = 0. In turn,
the repair policy yields that μ{1} =μB does not depend on the states of E and S,
μ{2} (1,0,1) = μ{2} (1,0,0) = μE, μ{2} (0,0,1) = μ{2} (0,0,0) = 0, μ{3} (1,1,0) = μS ,
and μ{3} (1,0,0) = μ{3} (0,1,0) = μ{3} (0,0,0)=0.
The transition diagram of Xt is helpful in obtaining the Kolmogorov equations
from which the state probabilities for Xt can be computed. As shown further,
these probabilities are used to find the state probabilities Pr(Zt =a), a∈S, and
transition intensities a→b (t) for the process Zt , which, in turn, are necessary
to compute useful reliability characteristics of a considered system. Analyzing
Fig. 2.3, we obtain the following Kolmogorov equations for Xt :
dP1,1,1 (t )/dt = P0,1,1 (t )μB + P1,0,1 (t )μE + P1,1,0 (t )μS +
− P1,1,1 (t )[λB (1 − πS ) + λB πS ]
dP1,1,0 (t )/dt = P0,1,0 (t )μB + P1,0,0 (t )μE − P1,1,0 (t )(λB + μS )
dP0,1,1 (t )/dt = P1,1,1 (t )λB (1 − πS ) − P0,1,1 (t )(λE + μB )
dP1,0,1 (t )/dt = P0,0,1 (t )μB − P1,0,1 (t )(λB + μE )
dP0,1,0 (t )/dt = P1,1,1 (t )λB πS + P1,1,0 (t )λB − P0,1,0 (t )(λE + μB )
dP1,0,0 (t )/dt = P0,0,0 (t )μB − P1,0,0 (t )(λB + μE )
dP0,0,1 (t )/dt = P0,1,1 (t )λE + P1,0,1 (t )λB − P0,0,1 (t )μB
dP0,0,0 (t )/dt = P0,1,0 (t )λE + P1,0,0 (t )λB − P0,0,0 (t )μB (2.19)
For simplicity, we will compute only the asymptotic values of the system
parameters, i.e. the values of Px (t), Ia→b (t) and a→b (t) for t→∞. This is
sufficient for most practical purposes. Thus, from now on these parameters will
be written without the variable t. Equating dPx (t)/dt to 0 in (19) we obtain
the following equations for the steady state probabilities Px = lim t→∞ Px (t),
x∈{0,1}n :
P0,1,1 μB + P1,0,1 μE + P1,1,0 μS = P1,1,1 λB
P0,1,0 μB + P1,0,0 μE = P1,1,0 (λB + μS )
P1,1,1 λB (1 − πS ) = P0,1,1 (λE + μB )
P0,0,1 μB = P1,0,1 (λB + μE )
P1,1,1 λB πS + P1,1,0 λB = P0,1,0 (λE + μB )
P0,0,0 μB = P1,0,0 (λB + μE )
P0,1,1 λE + P1,0,1 λB = P0,0,1 μB
P0,1,0 λE + P1,0,0 λB = P0,0,0 μB
P1,1,1 + P1,1,0 + P0,1,1 + P1,0,1 + P0,1,0 + P1,0,0 + P0,0,1 + P0,0,0 = 1 (2.20)
The last equation expresses the obvious fact that the total probability of all
possible outcomes is equal to 1. However, we cannot solve Eq. (2.20) without it,
because the first eight equations are not algebraically independent. The solution
of Eq. (2.20) is given below.
μB μE μS (λE + μB )
P1,1,1 =
(λB πS + μS )[(λB + μB )(λB λE + λE μE + μB μE )]
λB πS
P1,1,0 = P1,1,1
μS
λB λE (1 − πS )
P1,0,1 = P1,1,1
μE (λE + μB )
λE (λB + μS )
P1,0,0 = P1,1,0
μE (λE + μB )
μE
P0,1,1 = P1,0,1
λE
(λB + μS )
P0,1,0 = P1,1,0
(λE + μB )
λB + μE
P0,0,1 = P1,0,1
μB
λB + μE
P0,0,0 = P1,0,0 (2.21)
μB
Let us note that P1,1,1 is computed first, and each subsequent state probability
of Xt is obtained from a previously computed one. Such approach allows to avoid
unnecessarily complicated formulas. Clearly, closed formulas for Px can only be
derived for a small system, in case of a more complex one a numerical method
of solving linear equations would have to be used.
An average user (electric power consumer) usually pays no attention to
the details of the system’s operation. For him/her it is essential that power be
supplied in sufficient quantity. Thus, a user perceives the considered system as a
three-state one with the state space {B,E,F}, where power is sufficiently supplied
from the mains in state B, supplied from emergency source in insufficient
quantity in state E, and not supplied in state F. Let Zt = (Xt ) be a power supply
process as perceived by the user, i.e.
F (1, 1, 1) = F (1, 1, 0) = F (1, 0, 1) = F (1, 0, 0) = B,
F (0, 1, 1) = F (0, 1, 0) = E, (2.22)
F (0, 0, 1) = F (0, 0, 0) = F.
As is usually the case, S has far less elements than {0,1}n , thus the states of
Zt =(Xt ) are obtained by merging the states of Xt . The transition diagram of Zt
is shown in Fig. 2.5.
B
ΛB→E
ΛE→B
ΛB→0 E
Λ0→B
ΛE→0
FIG. 2.5 Transition diagram of Zt = (Xt ).
2.5 Intensities of transitions between the system states

For computing the transition intensities of Zt we will use the extended Birnbaum
importances defined by (5) and (6) in Sections 2.2. First, we will calculate
I a→b () for each a,b∈S and ⊆{1,…,n} such that I a→b () > 0. For this purpose
we have to find the respective a→b crit () or a→b crit (). Analyzing Figures.
2.3 and 2.4 we conclude that
B→E ({1}) = (1, 1, 1), (1, 1, 0); B→E ({1, 3}) = (1, 1, 1)
crit crit
crit
B→F ({1}) = (1, 0, 1), (1, 0, 0)
crit
E→F ({2}) = (0, 1, 1), (0, 1, 0)
crit
E→B ({1}) = (0, 1, 1), (0, 1, 0)
crit
F→B ({1}) = (0, 0, 1), (0, 0, 0) (2.23)
which, in view of (9) and (10), yields:

IB→E ({1}) = P1,1,1 + P1,1,0 ; I B→E ({1, 3}) = P1,1,1
I B→F ({1}) = P1,0,1 + P1,0,0
I E→F ({2}) = P0,1,1 + P0,1,0

IE→B ({1}) = P0,1,1 + P0,1,0

IF→B ({1}) = P0,0,1 + P0,0,0 (2.24)
Let us note that the remark to Theorem 2.1 only pertains to ={1} and
crit B→E {1}, because λ (x) or μ (x) are constant within every other set of
Another random document with
no related content on Scribd:
he should proceed from Shanghai to Tientsin in a Russian vessel
which “he might borrow for the purpose.” Li Hung-chang’s reply,
telegraphed to Yüan Shih-k’ai for transmission to the Throne, while
outwardly respectful, clearly implies that Her Majesty has been to
blame for the disasters then occurring. “I am sincerely grateful,” he
says, “for Your Majesty’s gratifying confidence in me, but cannot help
recalling to mind the folly which has now suddenly destroyed that
structure of reformed administration which, during my twenty years’
term of office as Viceroy of Chihli, I was able to build up not
unsuccessfully. I fear it will not be possible for me to resume the
duties of this difficult post at a time of crisis like the present, destitute
as I am of all proper and material resources.” He proceeds even to
criticise Her Majesty’s suggestion as to his journey, observing that
“Russia possesses no vessel at Shanghai, and would certainly
refuse to lend if she had one, in view of the state of war now
existing.” Finally, he excuses himself for deferring his departure, on
the ground that the British Minister had requested him not to leave
until the foreign Ministers had been safely escorted from Peking to
Tientsin. “I do not know,” says he, “if any such arrangements for
safely escorting them can be made,” and therefore concludes by
asking Yüan to inform the Throne that he will start northwards,
journeying by land, “as soon as his health permits it.” To this plain-
spoken message from the great Viceroy, Tzŭ Hsi replied in two lines
of equally characteristic directness:—“Li Hung-chang is to obey our
earlier Decree, and to make all haste northwards. The crisis is
serious. Let him make no further excuses for delay.”
In spite of these peremptory orders, Li Hung-chang, who had a
very definite conception of his own predicament, remained at
Shanghai, ostensibly negotiating, but in reality waiting, to see what
would be the outcome of the siege of the Legations. He was
interviewed by The Times correspondent at Shanghai on the 23rd of
July, and then stated that he would not proceed to his post in the
north until convinced by clear proofs that the Empress Dowager had
seen the folly of her ways, and was prepared to adopt a conciliatory
policy towards the outraged foreign Powers. At the end of July, when
it became clear to him that the Court had determined on flight, he
forwarded by special courier a very remarkable Memorial, in which
he called the Throne to task in the plainest possible terms, and
urged an immediate change of policy. This Memorial reached the
Empress before her departure from Peking; certain extracts from it
are well worth reproduction, as showing Li Hung-chang at his best,
and displaying that quality of courageous intelligence which made
him for twenty years the foremost official in China and a world-wide
celebrity:—
“It is to be remembered that between this, our Empire of

China, and the outer barbarians, hostilities have frequently
occurred since the remotest antiquity, and our national history
teaches that the best way to meet them is to determine upon
our policy only after carefully ascertaining their strength as
compared with our own. Since the middle of the reign of Tao-
Kuang the pressure of the barbarians on our borders has
steadily increased, and to-day we are brought to desperate
straits indeed. In 1860 they invaded the Capital and burnt the
Summer Palace; His Majesty Hsien-Feng was forced to flee,
and thus came to his death. It is only natural that His
Majesty’s posterity should long to avenge him to the end of
time, and that your subjects should continue to cherish
undying hopes of revenge. But since that time, France has
taken from us Annam, the whole of that dependency being
irretrievably lost; Japan has fought us, and ousted us from
Korea. Even worse disasters and loss of territory were,
however, to follow: Germany seized Kiaochao; Russia
followed by annexing Port Arthur and Talienwan; England
demanded Wei-hei-wei and Kowloon, together with the
extension of the Shanghai Settlements, and the opening of
new treaty ports inland; and France made further demands for
Kuang-Chou wan. How could we possibly maintain silence
under such grievous and repeated acts of aggression?
Craven would be the man who would not seek to improve our
defences, and shameless would be he who did not long for
the day of reckoning. I myself have enjoyed no small favours
from the Throne, and much is expected of me by the nation.
Needless for me to say how greatly I would rejoice were it
possible for China to enter upon a glorious and triumphant
war; it would be the joy of my closing days to see the
barbarian nations subjugated at last in submissive allegiance,
respectfully making obeisance to the Dragon Throne.
Unfortunately, however, I cannot but recognise the
melancholy fact that China is unequal to any such enterprise,
and that our forces are in no way competent to undertake it.
Looking at the question as one affecting chiefly the integrity of
our Empire, who would be so foolish as to cast missiles at a
rat in the vicinity of a priceless piece of porcelain? It requires
no augur’s skill in divination to foresee that eggs are more
easily to be cracked than stones. Let us consider one recent
incident in proof of this conclusion. Recently, in the attack by
some tens of thousands of Boxers and Imperial troops upon
the foreign Settlements at Tientsin, there were some two or
three thousand foreign soldiers to defend them; yet, after ten
days of desperate fighting, only a few hundred foreigners had
been slain, while no less than twenty thousand Chinese were
killed and as many more wounded. Again, there are no real
defences or fortified positions in the Legations at Peking, nor
are the foreign Ministers and their Legation staffs trained in
the use of arms; nevertheless, Tung Fu-hsiang’s hordes have
been bombarding them for more than a month, and have lost
many thousands of men in the vain attempt to capture the
position.
“The fleets of the Allied Powers are now hurrying forward
vast bodies of their troops; the heaviest artillery is now being
brought swiftly to our shores. Has China the forces to meet
them? Does she possess a single leader capable of resisting
this invasion? If the foreign Powers send 100,000 men, they
will easily capture Peking, and Your Majesties will then find
escape impossible. You will no doubt endeavour once more to
flee to Jehol, but on this occasion you have no commander
like Sheng Pao to hold back the enemies’ forces from pursuit;
or, perhaps, you may decide to hold another Peace
Conference, like that at Shimonoseki, in 1895? But the
conditions to-day existing are in no way similar to those of
that time, when Marquis Ito was willing to meet me as your
Minister Plenipotentiary. When betrayed by the Boxers and
abandoned by all, where will your Majesties find a single
Prince, Councillor, or Statesman able to assist you
effectively? The fortunes of your house are being staked upon
a single throw; my blood runs cold at the thought of events to
come. Under any enlightened Sovereign these Boxers, with
their ridiculous claims of supernatural powers, would most
assuredly have been condemned to death long since. Is it not
on record that the Han Dynasty met its end because of its
belief in magicians, and in their power to confer invisibility?
Was not the Sung Dynasty destroyed because the Emperor
believed ridiculous stories about supernatural warriors clad in
miraculous coats of mail?
“I myself am nearly eighty years of age, and my death
cannot be far distant; I have received favours at the hands of
four Emperors. If now I hesitate to say the things that are in
my mind, how shall I face the spirits of the sacred ancestors
of this Dynasty when we meet in the halls of Hades? I am
compelled therefore to give utterance to this my solemn
prayer, and to beseech Your Majesties to put away from you
at once these vile magic workers, and to have them
summarily executed.
“You should take steps immediately to appoint a high official
who shall purge the land of this villainous rabble, and who
shall see to it that the foreign Ministers are safely escorted to
the headquarters of the Allied Armies. In spite of the great
heat, I have hurried northwards from Canton to Shanghai,
where your Majesties’ Decrees urging me to come to Peking
have duly reached me. Any physical weakness, however
serious, would not have deterred me from obeying this
summons, but perusal of your Decrees has led me to the
conclusion that Your Majesties have not yet adopted a policy
of reason, but are still in the hands of traitors, regarding these
Boxers as your dutiful subjects, with the result that unrest is
spreading and alarm universal. Moreover, I am here in
Shanghai without a single soldier under my command, and
even should I proceed with all haste in the endeavour to
present myself at your Palace gates, I should meet with
innumerable dangers by the way, and the end of my journey
would most probably be that I should provide your rebellious
and turbulent subjects with one more carcass to hack into
mincemeat. I shall therefore continue in residence here for the
present, considering ways and means for raising a military
force and for furnishing supplies, as well as availing myself of
the opportunity of ascertaining the enemies’ plans, and
making such diplomatic suggestions as occur to me to be
useful. As soon as my plans are complete, I shall proceed
northwards with all possible speed.”
The plain-spoken advice of Li Hung-chang was not without effect

on the Empress Dowager. The Decrees issued by her in the name of
the Emperor from Huai-lai on the 19th and 20th of August are the
first indications given to the outside world that she had definitely
decided on a policy of conciliation so as to render possible her
eventual return to the capital—an event which, as she foresaw,
would probably be facilitated by the inevitable differences and
jealousies already existing among the Allies.
In the Edict of the 19th of August, after explaining that the whole
Boxer crisis and the attack on the Legations was the result of
differences between Christian and non-Christian Chinese, she
querulously complains that the foreign Powers, although doubtless
well meaning in their efforts to “exterminate the rebels,” are behaving
in a manner which suggests aggressive designs towards China, and
which shows a lamentable disregard of proper procedure and
friendliness. She naively observes that the Chinese Government had
been at the greatest pains to protect the lives and property of
foreigners in Peking, in spite of many difficulties, and expresses
much surprise at such an evil return being made for her invariable
kindness and courtesy. If it were not for the unbounded capacity of
foreign diplomats, fully proved in the past, in the matter of credulity
where Chinese statecraft is concerned, it would be difficult to regard
utterances like these as the work of an intelligent ruler. But Tzŭ Hsi
was, as usual, justified, for at the very time when these Decrees
were issued, Russia was already using very similar arguments, and
making excuses for the Chinese government, in pursuance of her
own policy at Peking.
In the conclusion of the Decree above referred to, Her Majesty
orders Jung Lu, Hsü T’ung and Ch’ung Ch’i to remain in Peking to
act as peace negotiators, but she admits that, in dealing with
foreigners supported by troops and flushed with success, it may be
difficult for them at the outset to determine on a satisfactory line of
procedure. She leaves it to these plenipotentiaries, therefore, to
determine whether the best course would be to telegraph to the
respective Foreign Offices of the countries concerned, or to consult
with the Consuls-General at Shanghai (sic), with a view to obtaining
friendly intervention! It could not escape so shrewd a person as Tzŭ
Hsi that the atmosphere of Peking at this juncture was not likely to
be favourable to her purposes, and that it would be easier to
hoodwink the Foreign Offices and the Consuls at Shanghai than
those who had just been through the siege.
A Decree of the following day, also in the name of the Emperor, is
couched in a very different strain—a pathetic admission of the
Throne’s guilt, a plea for the sympathy of his people, and an
exhortation to return to ways of wisdom. “Cleanse your hearts, and
remove all doubt and suspicion from your minds, so as to assist us,
the Emperor, in our shortcomings. We have been utterly unworthy,
but the time is at hand when it shall be for us to prove that Heaven
has not left us without sense of our errors and deep remorse.” The
whole document reads with an unusual ring of sincerity, accepting, in
the name of the Emperor, full blame for all the disasters which had
overtaken the country, while reminding the official class that the first
cause of these calamities dates back to the time when they learned
and adopted habits of inveterate sloth and luxury. From depths of
contrition, the Edict admits fully the Throne’s responsibility, “We, the
Lord of this Empire, have failed utterly in warding off calamities from
our people, and we should not hesitate for one moment to commit
suicide, in order to placate our tutelary deities and the gods of the
soil, but we cannot forget that duty of filial piety and service which we
owe to our sacred and aged mother, the Empress Dowager.”
The policy of reform is now clearly enunciated and outlined as an
essential condition of the future government of the Empire. Provincial
and metropolitan officials are ordered to proceed at once to join the
Court, in order that the reform programme may be speedily initiated;
the Yangtsze Viceroys are thanked for preserving order in
accordance with “treaty stipulations,” and Chinese converts to
Christianity are once more assured of the Throne’s protection and
good-will.
These utterances of the Throne, which lost nothing in their
presentation to the respective Powers by Prince Ch’ing and his
colleagues, soon produced the desired effect, and reassured the
Throne and its advisers as to their personal safety. Accordingly, early
in September, we find all the Viceroys and high officials of the
Provinces uniting in a Memorial, whereby the Court is urged to return
at once to the Capital, advice which would never have been given
had there been any question of violent measures being taken by the
Allies against the Empress Dowager. At this time the question of the
future location of the Chinese Capital was being widely discussed at
Court, and there was much conflicting advice on the subject. The
Viceroys’ Memorial was drafted by Yüan Shih-k’ai and forwarded by
him to Liu K’un-yi, at Nanking, for transmission; it definitely blames
the Boxers and their leaders for the ruin which had come upon
China, and rejoices at the thought that “the perplexities which
embarrassed your Majesties in the past have now given place to a
clearer understanding of the situation.” Noting the possibility of the
Court’s leaving T’ai-yüan fu and making “a further progress”
westwards to Hsi-an, the Memorialists deplore the idea and proceed
to show that such a step would be unwise as well as inconvenient.
As an example of the way in which Chinese Ministers of State deal
with questions of high policy and strategy, the following extract from
this Memorial is not without interest:—
“It is true that, in times past, our Capital has been shifted on
more than one occasion of national danger, but in those days
our enemies were not able to push their armies far into the
interior of our country for indefinite periods, and were
compelled to withdraw after brief expeditions. The position of
affairs to-day, however, is very different, so that we can obtain
no reliable guidance from precedents of history. As regards
the province of Shensi, it has always been a centre of wars
and rebellions; its people are poverty stricken, and there is no
trade there. Seven centuries ago, Hsi-an was an Imperial city,
but is now anything but prosperous. Its vicinity to Kansu and
the New Dominion territories, infested with Mahomedan
rebels and adjoining the Russian Empire, renders it most
unsuitable as a site for your Majesties’ Capital. Supposing
that the Allies, flushed with success, should determine on an
advance westwards, what is there to prevent them from doing
so? If ten thousand miles of ocean have not stopped them,
are they likely to be turned back from a shorter expedition by
land?”
After referring to the fact that the cradle of the Dynasty and the
tombs of its ancestors are situated near Peking, and that it is
geographically best fitted to be the centre of Government, the
Memorialists remind the Throne that the foreign Powers have
promised to vacate Peking, and to refrain from annexing any territory
if the Court will return. These ends, they say, will not be attained
should the Court persist in its intention to proceed further westwards,
since it is now the desire of the foreign Ministers that China’s rulers
should return to Peking. In the event of a permanent occupation of
Peking by the Allies, the loss of Manchuria would be inevitable. The
Memorialists predict partition and many other disasters, including
financial distress, and the impossibility of furnishing the Throne with
supplies at Hsi-an or any other remote corner of the Empire. If the
Court’s decision to proceed to Hsi-an is irrevocable, at least a
Decree should now be issued, stating that its sojourn there will be a
brief one, and that the Court will return to Peking upon the complete
restoration of peaceful conditions. “The continued existence of the
Empire must depend upon the Throne’s decision upon this matter.”
The Memorial concludes by imploring their Majesties to authorise
Prince Ch’ing to inform the foreign Ministers that the withdrawal of
the allied armies will be followed by a definite announcement as to
the Court’s return.
In a further Memorial from the Viceroys and Governors, it is stated
that the Russian Minister for Foreign Affairs had suggested to the
Chinese Minister in St. Petersburg, that the location of the Capital at
Hsi-an would certainly prove undesirable, in view of the poverty-
stricken condition of the province, and that their Majesties would no
doubt, therefore, proceed to Lan-chou fu, in Kansu. Referring to this
interesting fact, the Memorialists observe:—
“Those who are in favour of establishing the Capital at Hsi-

an profess to claim that the Yellow River and the T’ung Kuan
Pass constitute natural and impassible frontiers against
attack. They forget, however, that foreign nations possess
artillery of very long range. At T’ung Kuan the Yellow River is
less than two miles wide, and their guns will easily carry twice
that distance. Your Majesties have nothing but the native
artillery, and a few inferior foreign guns, and would never be
able to hold the position. The foreigners would undoubtedly
penetrate far into the interior, and control all the waterways,
thus preventing transport and supplies. Even if one foreign
Power were to find it difficult, there is no doubt that it would be
easy for several of them acting together.
“Moreover, friendly Powers are entitled, by the law of
civilised nations, to send their diplomatic representatives to
our Capital. If peace be made, and the foreign Powers assent
to the proposed change of capital, they will surely insist upon
sending their envoys into Shensi. After their recent
experiences, they will require to have foreign troops to guard
their Legations, whose numbers must necessarily be large, in
proportion to the distance from the coast. Foreign garrisons
would thus have to be established at points in Honan, Shansi
and Chihli, in order to maintain their line of communications,
so that China would eventually be overrun by foreign troops. It
is, therefore, plainly out of the question that the Court should
leave Peking. In times of peace it might have been
suggested, but to think of it after a disastrous war is
impossible. The foreigners are acting in unison; China is
completely disorganised. They have ample resources and
reinforcements; China has none. If we have thoughts of
fighting any foreign Power we must first form alliances with
several others; in any case nothing can be done before an
ample supply of ordnance and munitions of war has been
accumulated. This is no time for considering such
possibilities. We, your Memorialists, venture to suggest that
Your Majesties have failed to take into consideration all these
facts, and in impressing them upon you, we earnestly beg that
you may now come to a wise decision.”
Before coming to a decision, however, Tzŭ Hsi required to be fully

assured that the foreign Powers would not insist on her abdicating
the supreme power as one of the conditions of peace. Convinced on
that point, the hesitation which she had previously shown in regard
to returning to Peking dropped from her like a garment. It had been
freely predicted by conservative officials and the literati that the Old
Buddha would never again wish to see her desecrated capital or to
visit the polluted shrines of her ancestors. In spite of her
superstitious nature, however, she was far too level-headed and far-
seeing a woman to attach supreme importance to sentimental
considerations, or to allow them to weigh heavily in the balance
when the question of her own rulership was at stake. The hesitation
which she had shown and the attention which she had paid to the
advice of those who, like Chang Chih-tung, desired her to establish a
new capital in Central China, were primarily a question of “face.” She
would only return to Peking if guaranteed the full dignity and power
of her former position. But as the peace negotiations proceeded, and
as it became clear to her that along the well-worn path of
international jealousies she might return unpunished, and even
welcomed, to Peking, she proceeded to make preparations for an
early return. Fully informed each day by Prince Ch’ing of the
progress which her plenipotentiaries were making towards the
completion of the Peace Protocol, and overjoyed at its terms, she
waited only until the condition of the roads, always more or less
impassable after the summer rains, had sufficiently improved to
permit of comfortable travelling. During the delay necessitated by the
collecting and packing of the enormous quantity of “tribute” collected
by Her Majesty and the Court during their stay at Hsi-an, she
received definite confirmation of the good news that her treasure
vaults in the capital had not been plundered by the foreign troops—
good news which increased her anxiety to return as quickly as
possible to superintend its removal before any pilfering by the
eunuchs should take place.
It was on the 24th day of the 8th Moon (20th October, 1901) that
the long procession started from Her Majesty’s temporary residence
in the Governor’s Yamên; followed by an enormous retinue, she
commenced her journey by sacrificing to the God of War, the
guardian spirit of her Dynasty (and, it may be added, patron of the
Boxers), at a small temple outside the city gates. From this onward
the Court advanced northward by easy stages of about twenty-five
miles a day, resting first at Ho-nan fu; thence on to K’ai-fêng, where
her sixty-sixth birthday was celebrated and where she remained for
some weeks. The travelling lodges and other arrangements for her
comfort and convenience along the whole line of her route were in
striking contrast to the squalor and privation which the Court had
endured in the flight from Peking.
It was during her stay at K’ai-fêng that the Peace Protocol was
signed at Peking. It was also before her departure from that city, at
the end of the 9th Moon, that Li Hung-chang died. His knowledge of
foreign affairs and remarkable ability in negotiations had been of the
greatest service to his Imperial mistress, and there is no doubt that
the liberal terms granted to China by the victorious Allies were very
largely due to his efforts. Her Majesty, while fully appreciating his
ability, had never treated him with marked favour, and had always
refused to appoint him to the Grand Council, giving as her excuse
that she could not understand his dialect. Upon his death, however,
she conferred upon him an honour which had never before been
granted to any Chinese subject under the Dynasty, namely, that of
having a shrine built to his memory at the capital itself, in addition to
those erected in the provinces where he had borne office.
It was significant of her impartial and intelligent rulership that,
although she had blamed him as originally responsible for the
Japanese War and its disastrous results, she had never approved of
the Emperor’s hasty and vindictive action in removing him from the
Viceroyalty of Chihli. Upon the signing of the Peace Protocol she
conferred additional posthumous honours upon him, taking occasion
at the same time, in an Imperial Decree, to congratulate and thank
Prince Ch’ing, Yüan Shih-k’ai and others, who assisted in bringing
about the settlement of peace terms. In particular she praised the
loyalty of Jung Lu, “who had earnestly advised the annihilation of the
Boxers, and who, in addition to other meritorious services on the
Grand Council, had been chiefly instrumental in protecting the
Legations.”
After a series of magnificent theatrical entertainments in honour of
her birthday, the Court left K’ai-fêng and continued its journey to the
capital. On the eve of her departure Her Majesty took occasion
sternly and publicly to rebuke the Manchu Prefect, Wen T’i,[121] who
had dared to advise her against returning to the capital, and to
predict that the treacherous foreigners would certainly seize her
sacred person—a useful piece of play to the gallery.
At the crossing of the Yellow River, which took place in beautiful
weather, she sacrificed to the River God, in expiation and
thanksgiving. The local officials had constructed a magnificent barge,
in the form of a dragon, upon which she and the ladies of the Court
crossed the stream. It was noticed from this point onwards that
wherever foreigners happened to be amongst the spectators of the
Imperial cortège, she made a point of showing them particular
attention and civility, and before her arrival in Peking she issued a
Decree commanding that Europeans should not be prevented from
watching the procession upon her arrival, and this in spite of the fact
that, in accordance with the usual custom, the Legations had issued
notices forbidding their nationals to appear in the streets during the
passage of the Imperial cortège. Everything indicated, in fact, that
Her Majesty now desired to conciliate the European Powers by all
possible means, and if it be borne in mind that it was part of her
deliberate policy thus to ingratiate herself with foreigners as a means
of furthering her own future policy, her actions lose nothing of
interest, while they gain something from the humorous point of view.
On crossing the borders of the Province of Chihli, Her Majesty
issued a Decree, couched in almost effusive terms of friendliness,
proclaiming that the Emperor would receive the foreign Ministers in
audience immediately upon his return to the Palace, and that the
reception would take place in the central Throne Hall of the sacred
enclosure. Chinese, reading this Decree, and ignorant of the terms
of the Peace Protocol which provided for this particular concession
to the barbarian, would naturally regard it as a spontaneous mark of
the Imperial clemency and goodwill. In the same Edict Her Majesty
proclaimed her intention of receiving the Ministers’ wives in person,
intimating that she cherished most pleasant memories of past
friendly intercourse with them. Here, again, we note fulfilment of a
plan, deliberately conceived and formed upon the best classical
models, “for dealing with strong and savage people.”
At noon on the 6th of January, 1902, the Imperial party arrived by
special train at the temporary station which had been erected close
to the Southern walls of Peking, and adjoining the old terminus at
Ma-chia pu. Large pavilions, handsomely decorated, had been
erected near the station, in which the Old Buddha and the Emperor
were to be received; they were furnished with a throne of gold
lacquer, cloisonné altar vessels and many valuable pieces of
porcelain. Several hundreds of the highest metropolitan officials
were in attendance, and a special place had been provided for
foreigners. As the long train of over thirty carriages drew up at the
station, the keen face of the Old Buddha was seen anxiously
scanning her surroundings from one of the windows of her car. With
her were the young Empress and the Princess Imperial, while the
chief eunuch, Li Lien-ying, was in attendance. Recognising Her
Majesty, every official fell upon his knees, whilst Chi Lu, chief officer
of the Household, officiously shouted to the foreigners to remove
their hats (which they had already done). The first to emerge from
the train was the chief eunuch, who proceeded forthwith to check the
long list of provincial tribute and treasure, mountainous loads of
baggage which had travelled with the Court from the start and under
Her Majesty’s close personal supervision. After the eunuch came the
Emperor, evidently extremely nervous, who, at a sign from Her
Majesty, hurried into his sedan-chair and was swiftly borne away,
without a word or a sign of recognition to any of the officials in
attendance. After his departure, the Empress came out and stood
upon the platform at the end of her carriage. “Quite a number of
foreigners are here, I see,” she was heard to observe. She saluted
them in accordance with the etiquette observed by Chinese women
—bowing and raising her crossed hands. Prince Ch’ing then
advanced to greet Her Majesty, and with him Wang Wen-shao (who
had succeeded Li Hung-chang as Peace Plenipotentiary). They
invited Her Majesty to enter her chair: “There is no hurry,” she
replied. She stood for some five minutes in full view of the crowd,
talking energetically with the bystanders, and looking extremely well
and youthful for her age, until the chief eunuch returned and handed
her the list of baggage and treasure, which she scanned with close
attention and then returned to him with an expression of satisfaction.
After this, at the request of the Viceroy of Chihli (Yüan Shih-k’ai),
the foreign manager and engineer of the railway were presented to
her, and received her thanks for the satisfactory arrangements made
throughout the journey. She then entered her chair, a larger and finer
conveyance than that supplied to the Emperor, and was borne away
towards the Palace; by her side ran one of her favourite eunuchs
repeatedly calling Her Majesty’s attention to objects of interest.
Whenever foreigners were in sight he would inform Her Majesty of
the fact, and by one he was heard distinctly to say: “Look! Old
Buddha, look quickly at that foreign devil,” whereupon the Empress
smiled and bowed most affably. Passing through the Southern gate
of the Chinese city, her bearers carried her straight to the large
enceinte of the Tartar city wall at the Ch’ienmen, where stands the
shrine dedicated to the tutelary God of the Manchus. Here crowds of
foreigners were in waiting on the wall. Looking down on the
courtyard towards the shrine, they saw the Old Buddha leave her
chair and fall upon her knees to burn incense before the image of the
God of War, whilst several Taoist priests chanted the ritual. Rising
she next looked up towards the foreigners, smiling and bowing,
before she was carried away through the gate into the precincts of
the Forbidden City. No sooner had she reached the inner palace (the
Ning Shou kung) at about 2 p.m., than she commanded the eunuchs
to commence digging up the treasure which had been buried there at
the time of her flight; she was gratified beyond measure to find that it
had indeed remained untouched.
Next, with an eye not only upon her future relations with foreigners
but also on public opinion throughout the Empire, she issued a
Decree conferring posthumous honours on the “Pearl concubine,”
who, as it will be remembered, was thrown down a well by her orders
on the morning of the Court’s flight from the Palace. In this Decree
Her Majesty praises the virtue and admirable courage of the dead
woman, which “led her virtuously to commit suicide when unable to
catch up the Court on its departure,” unwilling as she was to witness
the destruction and pollution of the ancestral shrines. Her trustworthy
conduct was therefore rewarded by the granting of a posthumous
title and by promotion of one step in rank in the Imperial harem. The
Decree was generally regarded as fulfilling all reasonable
requirements of atonement towards the deceased, for in China the
dead yet live and move in a shadowy, but none the less real,
hierarchy. Alive, a “Pearl concubine” more or less counted for little
when weighed against the needs of the Old Buddha’s policies; once
dead, however, her spirit must needs be conciliated and
compensated.
Many Europeans who had witnessed the arrival of the Empress
Dowager, remained at the railway station to see the unloading of her
long baggage train, a most interesting and instructive sight. First
were discharged the yellow chairs of the young Empress and the
Princess Imperial, and four green chairs with yellow borders for the
principal concubines; the other ladies of the Court followed in official
carts, two to each vehicle. There were about ninety of them
altogether, and the arrangements for their conveyance were
accompanied by no little noise and confusion, the loquacity of some
of the elder ladies being most noticeable. After their departure the
attention of the eunuchs and minor officials was directed to the huge
pile of the Empress Dowager’s personal baggage, which included
her cooking utensils and household articles in daily use. This
operation, as well as the removal of a very large quantity of bullion,
(every case of which was marked with the name of the province or
city that had sent it as tribute), was for a time superintended by the
Grand Council. But as the work was enough to last for several hours,
it was not long before, led by Jung Lu, they entered their chairs and
left for the City. It was noticed that Jung Lu seemed very infirm, and
was supported as he walked by two attendants of almost gigantic
stature.
From Cheng-ting fu to Pao-ting fu, and thence to Peking, the Court
travelled, for the first time in its history, by train. The following
description of the journey is reprinted, by kind permission of the
editor of The Times, from an article published in that paper in March,
1902. It shows an interesting side of the Empress Dowager’s
character, that of the thrifty mistress of her goods and chattels, and
gives a clear-cut impression of that vigorous personality which
devoted the same close attention to details of transport and domestic
economy as to niceties of Court ceremonial or historical precedents
on vital questions of State; characteristics which inevitably suggest a
marked resemblance between the Old Buddha and le petit Caporal.
“Early on December 31st the Court arrived at Cheng-ting fu,

escorted by a large body of cavalry and accompanied by an
enormous suite of officials, eunuchs and servants. The
baggage was carried by a train of carts, estimated by an eye-
witness at three thousand. The eunuchs numbered between
three and four hundred, and of cooks and other kitchen
servants there were almost as many. To provide
accommodation for such a mass of people was impossible,
especially as all the best quarters in the town had already
been occupied by the high officials who, with their retainers,
had come from the north to welcome the Empress Dowager
on her return. For three days the Court rested in Cheng-ting
fu, during which time the scene was one of indescribable
confusion; baggage, stacked haphazard, filled every available
corner, eunuchs and servants camping around and upon it,
stolidly enduring much physical discomfort with the apathy
peculiar to Asiatics. Yet, so great was the cold (on the night of
January 1st the thermometer stood at two degrees
(Fahrenheit) below zero) that many of these wayfarers gave
way to lamentations and tears. Officials of the lower and
middle grades, unable to obtain a lodging, were compelled to
pass these days in such makeshift shelter as they could find
in the vicinity of the railway station, where swarmed a mob of
undisciplined soldiery. On the second night a fire broke out in
the stables of the Imperial residence, which, though
eventually checked before much damage was done, added
greatly to the general disorder, and might well have had
serious results in the absence of all organisation and control.
The definite announcement of the Court’s intention to leave
for Pao-ting fu on the 3rd of January was received with
unmistakable relief by the hungry, motley crowd which
represented the pomp and pride of Asia’s greatest Empire.
“From the Yellow River to the railway terminus at Cheng-
ting fu—a distance of about two hundred and fifty miles—the
ever-growing Imperial procession had travelled almost
continuously in chairs, litters, carts, and on horse-back,
affording a spectacle which recalled in many of its chief
characteristics those of Europe’s mediæval pageantry as
described by Scott. Every Manchu Prince had a retinue of
horsemen varying from thirty to a hundred in number; along
the frost-bound, uneven tracks which serve for roads in
Northern China, an unending stream of laden waggons
creaked and groaned through the short winter’s day, and on,
guided by soldier torch-bearers, through bitter nights to the
appointed stopping places. But for the Empress Dowager and
the Emperor, with the Chief Eunuch and the ladies of the
Court, there was easy journeying and a way literally made
smooth. Throughout its entire distance the road over which
the Imperial palanquins were borne had been converted into a
smooth, even surface of shining clay, soft and noiseless
under foot; not only had every stone been removed, but as
the procession approached gangs of men were employed in
brushing the surface with feather brooms. At intervals of
about ten miles well-appointed rest-houses had been built,
where all manner of food was prepared. The cost of this
King’s highway, quite useless, of course, for the ordinary
traffic of the country, was stated by a native contractor to
amount roughly to fifty Mexican dollars for every eight yards—
say £1,000 a mile—the clay having to be carried in some
places from a great distance. As an example of the lavish
expenditure of the Court and its officials, in a land where
squalor is a pervading feature, this is typical.
“The hour for leaving Cheng-ting fu was fixed by the
Empress Dowager at 9.30 a.m. on January 3rd. It is significant
of the character of this remarkable woman, now in her sixty-
seventh year, that even in matters of detail she leaves nothing
to chance, nothing to others; the long arm of her
unquestioned authority reaches from the Throne literally to
the servants’ quarters. Without creating any impression of
fussiness, she makes a distinctly feminine personality felt,
and the master-mind which has guided the destinies of China
for the last forty years by no means disdains to concern itself
in minor questions of household commissariat and transport.
It is impossible not to reflect what such a woman might have
been, what she might have done for her people, had there
come into her life some accident or influence to show her, in
their true light, the corruption, dishonesty, and cold-blooded
cruelty of her reign.
“The departure of the Court by a special train, long since
prepared for its reception by the Belgian railway authorities
and Sheng Hsüan-huai, was fixed for 9.30 a.m. in accordance
with Her Majesty’s orders; that Imperial and imperious lady,
however, made her appearance at the station at seven
o’clock, accompanied by the young Empress, the Imperial
concubine, and the ladies-in-waiting. The Emperor had
preceded her, and upon her arrival knelt on the platform to
perform respectful obeisance, in the presence of an interested
crowd. The next two hours were spent by the Empress, who
showed no signs of fatigue, in supervision of the
arrangements for despatching the vast accumulation of her
personal baggage, and in holding informal audiences with
various high dignitaries, military and civil, on the platform.
Amongst others she sent for M. Jadot, and spent some time
in friendly conversation with him, expressing great satisfaction
at the excellent arrangements made for her comfort, and
pleasure at exchanging the sedan chair for her luxuriously-
appointed drawing-room car. She took pains to impress upon
the engineer-in-chief the importance which she attached to
keeping the Court’s baggage and effects within reach,
evincing on this subject much determination of a good-
humoured kind.
“Eventually, after the despatch of four freight trains, her
mind was relieved of this anxiety, but it was to be clearly
understood that the same personal supervision would be
exercised at Pao-ting fu, for in no circumstances could the
impedimenta be sent on in advance to Peking. There is a
touch of feminine nature in this incident which can hardly fail
to bring the Empress Dowager into some degree of kinship
with her fellow-women in other lands; there is also an implied
reflection on the honesty of persons in attendance on the
Court which is not without significance.
“The scene upon the platform was one of remarkable
interest. In utter subversion of all accepted ideas in regard to
the seclusion and privacy in which the Chinese Court is
supposed to live, move, and have its being, there was on this
occasion—and indeed throughout the journey—no sign of
either attempt or wish to guard Their Majesties from
observation and intrusion. The crowd, quietly inquisitive, but
showing no inclination to demonstration of any sort, came and
went at its pleasure; Yüan Shih-kai’s braves, who to the
number of about a thousand travelled to Peking as the
Empress Dowager’s bodyguard, crowded around the Imperial
party, invading even their railway carriages. While the ruler of
the Empire held audience with some of its highest officials,
none of their retainers were employed, as might have been
expected, in keeping the people at a respectful distance; the
scene, in fact, bore striking testimony to that democratic side
of the Chinese character which cannot but impress itself on
every foreign visitor to a Viceroy’s or magistrate’s yamên; in
the present instance, however, it must have been, for all
concerned, a new and remarkable experience.
“To the native spectators, the ladies of the Court with their
eunuch attendants were as much objects of interest as the
foreign railway officials; the Imperial concubine, ‘Chin’ (or
‘Lustrous’) Kuei fei, a lively young person of pleasing
appearance, attracting much attention. This lady, gaily clad
and with lavishly painted face, bestowed upon everything
connected with the train an amount of attention which augurs
well for the future of railway enterprise in China, running from
car to car and chatting volubly with the ladies-in-waiting. All
the ladies of the Court wore pearls in profusion—those of the
Empress being particularly fine—and all smoked cigarettes in
place of the time-honoured water-pipe. Herein again, for the
optimistically inclined, may be found a harbinger of progress.
During the Empress Dowager’s audiences, lasting sometimes
over a quarter of an hour at a time, the Emperor stood close
at her side; invariably silent, generally listless, though his
expression when animated is described as conveying an
impression of remarkable intelligence. The young Empress
has good features, marred, in European eyes, by excessive
use of paint; she, too, appeared to be melancholy, and
showed but little interest in her surroundings. The Emperor
and both Empresses were simply dressed in quiet coloured
silks.
“The special train in which, punctually at 9.30 a.m., the
rulers of China left for their capital consisted of a locomotive
and twenty-one carriages, arranged in the following order:—
Nine freight cars laden with servants, sedan chairs, carts,
mules, &c.; a guard’s van, for employés of the railway; two

Safety and Reliability Modeling and Its Applications Advances in Reliability Science 1St Edition Mangey Ram Editor Full Chapter PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Safety and Reliability Modeling and Its Applications Advances in Reliability Science 1St Edition Mangey Ram Editor Full Chapter PDF

Uploaded by

Copyright:

Available Formats

Safety and Reliability Modeling and Its

Applications (Advances in Reliability

Total manufacturing assurance : controlling product

System Reliability Assessment and Optimization :

Advances in system reliability engineering Davim

Photovoltaic module reliability Wohlgemuth

Food Safety and Human Health Ram Lakhan Singh

Reliability Investigation of LED Devices for Public

Lead-free Soldering Process Development and Reliability

Nature-inspired computing paradigms in systems :

Safety and Reliability Modeling and Its Applications

Copyright © 2021 Elsevier Inc. All rights reserved.

For Information on all Elsevier publications visit our website at

Publisher: Matthew Deans

1 Reliability analysis of asphalt pavements: concepts and

2 Markov modeling of multi-state systems with

3 Reliability analysis of solar array drive assembly

3.2 DFT method 32

4 Reliability and maintainability of safety instrumented system

5 Application of Markovian models in reliability and

6 A method of vulnerability analysis based on deep learning

7 Mathematical and physical reality of reliability

8 Optimum staggered testing strategy for 1- and 2-out-of-3

9 Modiﬁed failure modes and effects analysis model for

10 Methodology to select human reliability analysis technique

11 Operation risk assessment of the main-fan installations of

12 Generalized renewal processes

13 Multiresponse maintenance modeling using desirability

14 Signature-based reliability study of

14.2 The signature vector of the r-within-consecutive-k-out-of-n:

15 Assessment of fuzzy reliability and signature of series–

Safety and reliability analysis is definitely one of the most multidimensional

Mangey Ram, Dehradun, India

Dr. Hoang Pham is a Distinguished Professor and former Chairman (2007–

Danilo T.M.P. Abreu, Analysis, Evaluation and Risk Management Laboratory

Anusha Pai, Associate Professor, Computer Engineering Department, Padre Conceicao

Reliability analysis of asphalt

1.2 Concepts of reliability

Reliability is the probability of successful performance; thus, it is the converse

The probability of failure is always associated with a particular performance

1.2.1 Levels of Reliability Methods

1.3 Literature regarding the application of reliability concepts

Reliability, R (percent ) = 100 × Probability (Actual pavement performance,

S02 = Sw2 + SN2 (1.9)

FR = 10−ZR ×S0 (1.10)

Where, S0 is the overall standard deviation of variation and ZR is the standard

R of a pavement design was defined as :

Where, N = allowable number of repetitions of the load

The study suggested the following reliability-based pavement-design equa-

1.4 Issues with estimation of pavement reliability

1.4.1 Input Parameters Variability

TABLE 1.1 Summary of pavement material variability from available literature

Parameter Description Range of Typical Distribution Reference

1.4.2 Performance models

1.4.3 Interaction between the failure modes

1.4.4 Material strength degradation

Markov modeling of multi-state

states to be partially ordered, where < is a strict-partial order relation. Thus

2.2 Basic assumptions, notation and deﬁnitions

S : partially ordered set of the system states

order in {0,1}n , i.e. x ≤ y if xi ≤ yi for each i∈{1,…,n}, and x < y if x ≤ y and

(x, 1 ) – binary vector obtained from x, such that xi =1, i∈

FIG. 2.1 The RBD of a mission-critical power supply system.

FIG. 2.3 Transition diagram of Zt = (Xt ).

If x≯y and (x)=a>b=(y) then a transition from a to b cannot be realized

FIG. 2.5 Transition diagram of Zt = (Xt ).

I B→F ({1}) = P1,0,1 + P1,0,0

I E→F ({2}) = P0,1,1 + P0,1,0