Вы находитесь на странице: 1из 9

Pioneers Of The Reliability Theories Of Past 50 Years

Alice Rueda, TRLabs/University of Manitoba, Winnipeg


Mirek Pawlak, Ph. D., TRLabs/University of Manitoba, Winnipeg
Key Words: History, Survey, Renewal Theory, Distribution, Analytical techniques, Reliability models, Coherent structure.

SUMMARY & CONCLUSIONS These theories include renewal theory, Markovian models,
component importance measures, Monte Carlo simulation,
This paper is dedicated to all the researchers for their Bayesian approach, and optimization techniques. The most
contributions in reliability theories in the past 50 years. The well known methods currently used in reliability calculation,
paper provides a summary on the pioneers of reliability such as parts-count, block diagram analysis, combinatorial
theories, and how their works placed a great influence on our methods, and Mean-Time-Between-Failure (MTBF), are
reliability analysis today. This is also a survey paper on based on these theories. Applications that use the same
reliability theories and methods. The information provided in theories include accelerated life-test, maintenance, software
this paper is mostly based on literatures found first hand to reliability, network reliability, and system reliability. Similar
provide as much a neutral view as possible. However, some of survey literatures on history of reliability have been published
the information is adopted from Refs. 1-4. Area of interest in in the past (Refs. 1-4). However, they are tailored for the
the reliability analysis included representation of reliability scientific community and needed to be updated.
parameters, renewal theory, coherent structure, diagram-based Before we start, we shall define reliability. According to
models, theoretical methods, and other miscellaneous ISO 8402 (1986, 3.18), reliability is defined as the ability of
techniques. Diagram based models included block diagrams, an item to perform a required function, under given
Fault Tree Analysis (FTA), Event Tree Analysis, and environmental and operational conditions and for a stated
flowgraphs. Theoretical methods included queueing theory, period of time. The US Military Standard 785 defines
asymptotic analysis, Boolean algebra, Bayesian method, reliability as the duration or probability of failure free
Monte Carlo simulation, optimization techniques. performance under stated conditions. Ushakov (Ref. 1)
Miscellaneous methods that cannot be classified in any of the classified modern reliability theories into 6 groups; “pure”
categories are also provided. reliability analysis, effectiveness, survivability, safety,
Looking back in the last century, a lot of the contributions security, and software reliability. Most of the theories covered
to reliability research were done in the last 50 years. Weibull, in this paper are defined as pure reliability analysis defined by
Epstein and Sobel had made a significant influence on the Ushakov.
distribution functions we used today. Lotka, Campbell, Feller,
Cox, Smith, Barlow, Proschan, Hunter, Marshall, Esary, 2. PARAMETER REPRESENTATION
Gnedenko, Belyaev, and Solov'yev had advanced the theories
for reliability. Takács’ paper in sojourn time provided an Parameters used in reliability theories are used to represent
initiative to the asymptotic studies. Birnbaum started a whole the characteristics of the subject under study such that mean
family on component importance measure for coherent lifetime or number of renewal (N), its variance and
structure. confidence interval can be estimated. Typically, these
parameters include representation of the failure rate, failure
1. INTRODUCTION probability distribution function, representation of the
component variation, and/or the subjected stress environment.
The theories behind reliability analysis have been around
as early as 1773 when Pierre-Simon Laplace developed the 2.1 Distribution Functions
Laplace Transform and published the Théorie analytique des
probabilités in 1812. Also, Andrei Andreevich Markov Distribution functions are used to represent characteristics
developed Markov Chains in 1880. However, little attention of the time that a system is operating or failed. The most
was paid to reliability research until the late 50s. Late 50s and influential distribution functions are; Weibull distribution for
early 60s were the golden time of reliability. A lot of the describing material fatigue failure (Ref. 5), and exponential
theories were established then. It was the time when modern distribution introduced by Epstein and Sobel for life testing
technology took off. There was a need for reliable products in (Ref. 6).
both the commercial and military sectors. Reliability analysis Weibull was studying fatigue life in materials and the
was used in the auto industry, aircraft industry, related subject of extreme value theory. Weibull proposed a
telecommunication systems, study of the missile system, as distribution as an appropriate distribution to describe the life
well as rockets. A lot of the same analysis techniques were length of materials (Ref. 5) in 1939. The distribution was later
still being used today. namely after Weibull. Later in 1945, Daniel used normal
distribution in the study of the strength of thread (Ref.7). In
1948, Epstein used log-normal distribution function to study elegant proof for n iid random variables for the problem
the extent of progressive damage (Ref. 8) and the related addressed in Barlow et al 1963.
subject of extreme value theory. In 1951, Weibull applied a
slightly different form of Weibull distribution for different 2.3 Part Variation
applications (Ref. 9). In 1952, Davis analyzed failure data and
goodness-of-fit tests for failure distributions. This paper gave In 1956, Meltzer used analysis of variance to obtain the
a distinct edge to the exponential distribution (Ref. 10). In expressions for the module reliability in terms of random
1953, Epstein and Sobel started using an exponential variables to represent the part characteristics (Ref. 26). In
distribution to represent the life-testing for electronic 1958, Blanton estimated the possible variations in equipment
components (Ref. 6). Epstein and Sobel’s paper made reliability due to reliability variations in parts by using partial
exponential distributed a popular distribution in reliability derivatives of the reliability formula and shown that the
analysis. In papers published in 1956 and 1958, Kao brought module reliabilities can cause great change in equipment
attention to Weibull distribution (Refs. 11, 12). In 1957, reliability (Ref. 27). In 1959 and 1960, Dreste expressed the
Buehler applied binomial distribution to calculate the derivation of part tolerance relating to output tolerance (Refs.
reliability of missile products (Ref. 13).In 1958, Birnbaum 28, 29).
and Saunders (Ref. 14) presented an ingenious statistical
model for life lengths of structures under dynamic loading by 2.4 Confidence Interval
representing the failure rate as a function of deterioration and
wear. A special case of a constant load led them to suggest the In 1957, Buehler applied confidence intervals for products
gamma distribution for life strengths in certain situations. In of binomial parameters to calculate the reliability of missile
1959, Herd summarize the application of various distributions, (Ref. 13). Later, Steck (1958, Ref. 30), Madansky (1958, Ref.
their density functions and variances, means, failure rates, etc 31) and Rosenblatt (1963, Ref. 32) also did similar studies. In
(Ref. 15). In 1959, motivated by the study in Ref. 14, Tate 1966, John and Lieberman determined the confidence bound
obtained the minimum variance unbiased estimation (ref. 16) of reliability for Weibull distribution (Ref. 33). In 1970,
that a system or structure would survive a specified time based Thoman et al estimated the maximum likelihood, confidence
on n observed times to failure for gamma, Weibull and two interval and tolerance of the reliability for Weibull
parameter exponential distributions. In 1960, Cox has shown distribution (Ref. 34). In 1985, Martz and Duran applied
that if a gamma distributed random variable T represents the Bootstrap method to evaluate the confidence estimation of
life time of the whole system, then the number of component system reliability based on component test data (Ref. 35).
replacement can be determined without the inversion of
Laplace Transforms (Ref. 17). In 1961, Zelen and 3. RENEWAL THEORY
Dannemiller pointed out that many life test procedures based
on the exponential distribution are not robust (Ref. 18). In Lotka, Campbell, Feller, Cox and Smith have tremendous
1962, Nylander provided a summary on life test distribution contributed in the development of renewal theory for
functions (Ref. 19). Blanton also provided a list in 1962 (Ref. reliability. Renewal theory is used to solve replacement
4). In 2001, Shortle and Mendel showed that the lifetime problem. The assumption is that a failed component will be
spaces are not physical Euclidean spaces, but is an equivalent replaced by a new component immediately at the time of
of collection of fibre bundles in differential geometry (Ref. failure. The aggregated lifetime of the components contributes
20). to the length of the operation. To estimate the lifetime of (n-1)
replacements, we can estimate the total of the expected
2.2 Failure Rate lifetime of all n components.
In 1939, Lotka applied renewal theory to industrial
In 1961, Watson and Wells found that the number of replacement problem (Ref. 36). In 1941, Campbell applied
articles that can work more than a specified time can be renewal theory to study the cost effectiveness of replacing at a
increased by giving all articles a suitable "burn-in" period or constant interval replacement and replacing whenever a unit
lifetime (Ref. 21). In 1963, Barlow et al used total positivity has failed (Ref. 37). In 1941, Feller developed renewal theory
to prove the Increasing Failure Rate (IFR) convolution as a mathematical discipline (Ref. 38). In 1948, Blackwell
theorem of n independent identically distributed (iid) random determined the expected N (number of renewal events) in a
variables (Ref. 22). In 1964, Barlow and Marshall published given time in Ref. 39. In 1948, Doob discussed renewal
one of the papers on inequalities and bounds for IFR theory in terms of probability (Ref. 40). In 1949, Feller
distributions (Ref. 23). In 1965, Barlow and Proschan (Ref. 2) published a paper on recurrent events. This paper provided a
published a book on reliability. This book introduced the new formulation for the expected N within a given time and its
concepts of monotone systems, distributions with monotone variance for discrete renewal process (Ref. 41). In 1953, Cox
increasing and decreasing failure rates. In 1972, Ross and Smith provided a direct proof for renewal theory (Ref.
provided a simplified proof of the Increasing Failure Rate on 42). In 1954 Smith defined the theory of regenerative
the Average (IFRA) closure theorem for coherent system stochastic processes (Ref. 43) by determining the expect N
(Ref. 24). In 1976, Block and Savits (Ref. 25) provided an and its variance for the given time for continuous renewal
process. Smith showed that as t goes to infinity the number
approaches a linear function. In 1957, Feller talked about the discussed and studied the conditions of incoherency for
problem of discrete time renewal process in one of the coherent structure (Ref. 60). In 1989, Huseby provided a more
chapters in Ref. 44. In 1958, Smith provided an elegant abstract unified theory of domination and signed domination
summary of the known mathematical results in renewal for coherent system with application to exact reliability
theory, especially on N (Ref. 45). Also in 1959, Smith proved calculation (Ref. 61). In 1995, Fu and Koutras (Ref. 62) found
the theorem for the expected number of N for pooled the bound for coherent structure with independent
processes, and showed that N and its variance approach a components. Tsitmidelis et al found a procedure to find these
linear function as t as to infinity under the assumed condition bounds using genetic algorithm in 2002 (Ref. 63). In 2000,
(Ref. 46). In 1959 and 1960, Barlow and Hunter published Boutsikas and Koutras determined the generalized bounds for
two articles on “Mathematical models for system reliability” coherent structure (Ref. 64). Inagaki and Henley (1980, Ref.
(Refs. 47, 48). In 1962, Cox published a book on renewal 65) and Jackson (1983, Ref. 66) had studied the important
theory (Ref. 49). His results in stochastic point processes implications in a noncoherent system. In 2003 Andrews and
theory and the joined work with Isham made an outstanding Beeson extended and generalized Birnbaum's component
impact on the reliability theory development. It opened new importance measures to include importance measure for
ways to apply point stochastic processes in reliability noncoherent systems (Ref. 67).
modeling. In 1964 Watson and Leadbetter published a paper
on analyzing the hazard function (Ref. 50). In 1965 and 1969, 5. DIAGRAM BASED MODELS
Gnedenko et al published Ref. 51 in Russian and in English
respectively. This book contained many results on repairable Diagram based models provide a visual representation of
redundant systems, asymptotic analysis, and reliability data the system. These models allow engineers and mathematician
analysis. to have a better understanding of the target system before
applying mathematical analyses. Besides block diagrams,
4. COHERENT AND NONCOHERENT STRUCTURES FTA, event tree analysis, and state diagram, flowgraphs (Refs.
68-70) can also be used for reliability analysis.
Birnbaum, Esary, and Saunders published a paper on
coherent structure in 1961 to determine the reliability of 5.1 Block Diagrams
coherent systems. The idea is to calculate the probability of
system failure due to a given component failure. This In 1954, Kleinerman used "reliability block diagram" to
component importance measure uses a simple binary show the reliability interrelations and proposed its
indication to indicate the failed and operating states of the mathematical representation (Ref. 71). In 1957, Blanton used
system. engineering block diagram with additional symbols to
In 1955, Birnbaum used random variable to represent represent power supplies and similar units and proposed its
structural components failure (Ref. 52). In 1961, Birnbaum et mathematical representation (Ref. 72).
al significantly extended the earlier work of Moore and
Shannon (1956) on relay network reliability and expanded the 5.2 Fault Tree Analysis (FTA)
study of coherent structure. They applied vector analysis
techniques to study the reliability of multi-component There are a lot of studies had been performed on FTA
structures in series and parallel configurations (Ref. 53). In (Refs. 72-88). In the 1960s, Fault Tree Analysis (FTA) was
1963, Esary and Proschan obtained the upper and lower born as the result of a study on the possibility of an accident
bounds of an arbitrary two-pole network with known structure that can occur on a Minuteman missile. Watson participated in
(Ref. 54). These bounds were expressed through complete sets the analysis project (Ref. 73). FTA is one of the most
of networks path and cuts. In 1965, Birnbaum and Esary commonly used techniques in risk assessment. In 1960,
defined and studied the modules of the binary coherent Holtzman and Marshall used block labeling such as
systems (Ref. 55). In 1966, Birnbaum et al defined the IFRA representing serial blocks with "and if" and parallel blocks
(Increasing Failure Rate on the Average) closure theorem for with "or" (Ref. 74). In 1980, Rosenthal used a divide and
coherent system (Ref. 56). In 1969, Birnbaum (Ref. 57) conquer approach to reduce the cost of analysis for FTA (Ref.
started a new stream of complex system analysis using 76). In 1992, Dugan et al started developing a dynamic FTA
BINARY failure type representation, "up" and "down" that methodology called DITree (Refs. 77). Dugan et al and Gulati
was introduced by Kolmogorovs 1945. In 1972, Ross continued the research as described in Ref. 78 and Ref. 79. In
provided a simplified proof of the Increasing Failure Rate on 1993, Coudert and Madre used Binary Decision Diagrams
the Average (IFRA) closure theorem for coherent system (BDD) to solve large fault trees with 2^10 basic events (Ref.
(Ref. 24). In 1978, Satyanarayana and Prabhakar described a 80). In 1995, Doyle and Dugan incorporated coverage
more efficiency probability calculation for coherent structures modeling into a BDD-based fault tree solution and found that
(Ref. 58). They introduced the idea of domination into the static fault trees can be used for no sequence-dependent
reliability literature for coherent system. In 1983, failure behaviour (Ref. 81). In 1996, Dutuit and Rauzy applied
Satyanarayana and Chang (Ref. 59) proved that the Factoring divide and conquer approach and achieved a linear-time
Algorithm for coherent system is the most efficient algorithm algorithm which identifies independent subtrees (Ref. 82). In
for unidirected networks. In 1984, Bendell and Ansell 1998, Anand and Somani applied decomposition method to
analysis fault trees with dependencies (Ref. 84). In 2001, reward model (MRM) for computing component importance
Andrews found that Not Logic is also essential for fault tree measure (Ref. 105). In the same year, Lefebvre used (Non-
analysis for multitasking systems (Ref. 87). homogeneous Poisson Process) NHPP model to represent the
aging effect on failure rate to determine unavailability as an
5.3 Event Tree Analysis alternative to Monte-Carlo simulation (Ref. 106).

In 1998, Papazoglou developed a mathematical 6.2 Asymptotic Analysis


representation for event trees building on basic concepts of set
theory and probability theory (Ref. 89). In 2000, Andrews and In 1954, Smith discovered the asymptotic property of N
Dunnett pointed out that Not logic is essential for proper event and its variance (Ref. 43). In 1956, Moore and Shannon
tree analysis on systems (Ref. 90, 91). published a paper that performed asymptotic analysis on relay
network reliability (Ref. 107). In 1959, Takacs determined the
5.4 State Diagrams asymptotic distribution of a two-state process (Ref. 94). In
1964, Gnedenko use asymptotic methods to analysis highly
State diagram is a very popular representation for available systems (Refs. 108, 109). In 1964, Pogozhev
queueing theory. In 1945, Kolmogorov used binary states to generalized one of the two key asymptotic theorems (Ref.
analyze the probability of plane destruction by anti-aircraft 100).
fire (Ref. 92). In 1954, Frantik uses three states to represent
components state (Ref. 93). In 1959, Takács used two states to 6.3 Boolean Algebra
represent the states of a system (Ref. 94). In 1959, Barlow and
Hunter applied the concept of on-off model to represent the In 1952, Gates used Boolean algebra to determine the
state of the unit in operating or failed state (Ref. 47). reliability of redundant systems (Ref. 110). In 1954, Frantik
Nowadays, Markov model with two states, operating and applied Boolean algebra to study components with three
failed states, is being widely used in reliability analysis. In distinct states (Ref. 93). In 1959, Mine used Boolean
1979, Elsayed and Zebib analyzed multistate devices (Ref. functions to represent complex systems (Ref. 111).
95).
6.4 Bayesian Approach
6. THEORETICAL METHODS
In 1961, books by Mosteller et al and Bazovsky proposed
A lot of the theoretical methods used in other fields are to use Baye's theorem to reliability (Ref. 112, 113). In 1982,
also applicable for reliability analysis. Some of the most Martz and Waller published a book on reliability analysis
popular methods include Queueing theory, asymptotic using Bayesian approach (Ref. 114), and more papers later
analysis, Boolean algebra, Bayesian approach, Monte Carlo (Refs. 115, 116). In 1994, Hulting and Robinson used the
Simulation and optimization. Others include genetic algorithm Bayesian approach to study a repairable series system (Ref.
(Ref. 63) and modular decomposition (Ref. 96). 117). In 1998, Kerscher III et al (Ref. 118) developed a new
methodology using Bayesian updates to characterize the
6.1 Queueing Theory reliability of new products under development.

In 1959, Barlow and Hunter said in Ref. 47, “If the future 6.5 Monte Carlo Simulation
probability states of equipment depend only on the immediate
past history, then the process is Markovian.” In this paper, In most cases, an exact analytical solution is not feasible
Barlow and Hunter applied Markov model to study system because of the difficulty in the equation solving. Monte Carlo
reliability. In 1960, Bharucha-Reid (Refs. 97), Keilson and simulation is a popular alternative to solve these problems. In
Kooharian (Ref. 98) used queueing theory to study system 1958, Firstman published a paper using Monte Carlo models
availability and maintainability, as well as Bosinoff et al in to estimate reliability (Ref. 119). In 1962, Blanton (Ref. 4)
1961 (Ref. 72). Also in 1961, Brender and Tainiter indicated provided a summary techniques being used to predict
that Markovian techniques can be used to consider the effects reliability, including Monte Carlo simulation. In 1967,
of both component drift and catastrophic failure (Ref. 99). In Belyaev obtained some results on confidence estimation using
1964, Pogozhev used Poisson process to study the failure Monte Carlo simulations (Ref. 120).
(Ref. 100). In 1965, Gnedenko et al (Ref. 51) applied
queueing theory ideas to reliability problems, especially for 6.6 Optimization
solving maintenance and replacement problems. In 1970,
Solov’yev applied queueing theory to study the reliability of In 1956, Moskowitz and McLean published the first paper
systems with small renewal time (Ref. 101, 102). In 1993, on optimal redundancy (Ref. 121). In 1957, Gordon studied
Dugan et al used Markov models to study reliability for fault optimum component redundancy (Ref. 122). In 1958, Bellman
tolerant systems (Ref. 78). In 1999, Ouhbi and Limnios used and Dreyfus applied dynamic algorithm in finding optimal
semi-Markov to study system availability and reliability (Refs. redundancy (Ref. 123). In 1959, Black and Proschan
103, 104). In 2003, Fricks and Trivedi introduced Markov determined optimal reliability using total positive techniques
(Ref. 124). In 1962, Kettelle proposed an effective practical 7. H.E. Daniels, "The statistical theory of the strength of
modification of dynamic programming algorithm for finding bundles of threads", Proc. Roy. Soc. London, vol. 183, 1945,
optimal problem (Ref. 125). In 1963, Barlow et al determined pp. 405-435.
the optimal redundancy for components with two possible 8. B. Epstein, "Application of the theory of extreme
kinds of failure (Ref. 126). In 1965, Barlow and Proschan’s values in fracture problems", J. Amer. Stat. Assoc., vol. 43,
book (Ref. 2) gave a deep presentation of optimal 1948, pp. 403-412.
maintenance and optimal redundancy problems. In 1965, 9. W. Weibull, "A statistical distribution function of wide
Proschan and Bray further developed the optimization applicability", J. Applied Mech., vol. 18, 1951, pp. 293-297.
methods for reliability under multiple constraints (Ref. 127). 10. D.J. Davis, "An analysis of some failure data", J.
In 1980, Tillman et al published a book on optimizing system Amer. Stat. Assoc., vol. 47, 1952, pp. 113-150.
reliability with further development on optimization methods 11. J.H.K. Kao, "A new life-quality measure for electron
(Ref. 128). tube", IRE Trans. Rel. and Quality Ctrl., PGRQC- 7, 1956,
pp. 1-11.
7. MISCELLANEOUS 12. J.H.K. Kao, "Computer methods for estimating
Weibull parameters in reliability studies", IRE Trans. Rel. and
This section contains techniques, theories or methods that Quality Ctrl., PGRQC-13, 1958, pp. 15-22.
do not belong to the above sections. In 1957, Portz and Smith 13. R.J. Buehler, "Confidence intervals for the product of
calculated the reliability as a product of the design reliability two binomial parameters", J. Amer, Stat. Assoc., vol. 52,
times the component reliability times the reliability of 1957, pp. 482-493.
fabrication (Ref. 129). In 1957, Lipp derived mathematical 14. Z.W. Birnbaum, S.C. Saunders, "A statistical model
representation for complex multi-element series-parallel for life-length of materials", J. Amer. Stat. Assoc., vol. 53, no.
networks (Ref. 130). In 1958, Flehinger proposed to improve 281, 1958, pp. 151-160.
reliability by adding redundancy at different level of the 15. G.R. Herd, "Some statistical concepts and techniques
system (Ref. 131). In 1959, Philipson demonstrated that the for reliability analysis and prediction", Proc. Natl. Symp. Rel.
overall reliability figure may be obtained from a combination and Quality Ctrl., Jan 1959, pp. 126-136.
of the reliabilities for the separate intervals (Ref. 132). In 16. R.F. Tate, "Unbiased estimation: Functions of
1960, Kaufmanns (Ref. 133) described three prediction location and scale parameters", Ann. Math. Stat., vol. 30, no.
techniques included using a computation of a "K" factor for 2, 1959, pp. 341-366.
each device. In 2000, Cramer and Kamps (Ref. 134) studied 17. D.R. Cox, "On the number of Renewals in a Random
the mathematical presentation of the life-time of sequential k- Interval," Biometrika, vol. 47, issue 3/4, Dec. 1960, pp. 449-
out-of-n systems using a scaling factor to adjust the hazard 452.
rates upon each failure. In 2002, Tsitmidelis et al used genetic 18. M. Zelen, M.C. Dannemiller, "The robustness of life
algorithm to find reliability bounds (Ref. 63). In 2003, testing procedures derived from the exponential distribution",
Kerscher III et al applied Fuzzy logic in PREDICT Technometrics, vol. 3, no. 1, 1961, pp. 29-49.
(Performance and Reliability Evaluation with Diverse 19. J.E. Nylander, "Statistical distributions in reliability",
Information Combination and Tracking) (Ref. 135). In 2003, IRE Trans. Rel. and Quality Ctrl., RQC-11, July 1962, pp. 43-
Marseguerra et al used neural network to predict reliability 53.
(Ref. 136). 20. J.F. Shortle, M.B. Mendel, "Physical foundations for
lifetime distributions", Proc. System and Bayesian Rel., 2001,
REFERENCES pp. 257-266.
21. G.S. Watson and W.T. Wells, "On the possibility of
1. I. Ushakov, “Reliability: past, present, future,” Recent improving the mean useful life of items by eliminating those
Advances in Reliability Theory: Methodology, Practice, and with short lives", Technometrics, vol. 3, 1961, pp. 281-198.
Interface, 2000; Birkhäuser, Boston, pp. 3-14. 22. R.E. Barlow, A.W. Marshall, F. Proschan, "Properties
2. R. Barlow and F. Proschan, Mathematical Theory of of probability distributions with monotone hazard rate", Ann.
Reliability, 1965; Wiley, New York. Math. Stat., vol. 34, 1963, pp.375-389.
3. R. Barlow, “Mathematical reliability theory from the 23. R.E. Barlow, A.W. Marshall, "Bounds for
beginning to the present time,” Proc. Intl. Conf. Math. distributions with monotone hazard rate, I and II", Ann. Math.
Methods in Rel., 2002. Stat., vol. 35, 1964, pp. 1234-1274.
4. H.E. Blanton, R.M. Jacobs, "A survey of techniques 24. S.M. Ross, Introduction to Probability Models with
for analysis and prediction of equipment and reliability", IEEE Optimization Applications, 1972; Academic Press, New York.
Trans. Rel., July 1962, pp.18-35. 25. H. Block, T.H. Savits, "The IRFA closure problem",
5. W. Weibull, "A statistical theory of the strength of Ann. Prob., 4, 1976, pp. 1030-1032.
materials", Ing. VetenskapsAkad. Handl., no. 151, 1939, 26. S.A. Meltzer, "Designing for reliability", IRE Trans.
pp.293-297. Rel. and Quality Ctrl., PGRQC-8, Sept 1956, pp. 36-43.
6. B. Epstein, M. Sobel, "Life testing", J. Amer. Stat. 27. H.E. Blanton, "Reliability-sensitivity- function
Assoc., vol. 48, no. 263, 1953, pp. 486-502. analysis", Electronic Design, vol. 6, Feb 1958.
28. F.E. Dreste, "Statistics: Key to reliable military 48. R.E. Barlow, L.C. Hunter, "Mathematical models for
electronic design", Military Electronics, vol. 6, Mar. 1959, pp. system reliability, part II", The Sylvania Technologist, vol. 13,
4, 6, 8. 1960, pp. 55-65.
29. F.E. Dreste, "Circuit design concepts for high 49. D.R. Cox, Renewal Theory, 1962; Wiley, New York.
reliability", Proc. Natl. Symp. Rel. and Quality Ctrl., Jan 50. G.S. Watson and M.R. Leadbetter, "Hazard Analysis.
1960, pp. 121-133. I," Biometrika, vol. 51, June 1964, pp. 175-184.
30. G.P. Steck, "Upper confidence limits for the failure 51. B.V. Gnedenko, Y.K. Belyaev, A.D. Solovyev, Math.
probability of complex networks", Sandia Corporation Methods in Reliability Theory, 1969; Academic Press, New
Research Report, SC-4133(TR), 1958. York. (Please note that this book was published in 1965 in
31. A. Madansky, "Uses of tolerance limits in missile Russia)
evaluation", Proc. Stat. Techniques in Missile Evaluation 52. Z.W. Birnbaum, "On a use of the Mann-Whitney
Symp., Aug 1958. Statistic", Proc. Berkeley Symp. Math. Stat. and Prob.,
32. J.R. Rosenblatt, "Confidence limits for the reliability University of California Press, Los Angeles, vol. 1, 1955, pp.
of complex systems", Stat. Theory of Rel., University of 13-17.
Wisconsin Press, Madison, Wisconsin, 1963, pp. 115-148. 53. Z.W. Birnbaum, J.D. Esary, S.C. Saunders, "Multi-
33. M.V. Johns, G.J. Lieberman, "An exact component systems and structures and their reliability",
asymptotically efficient confidence bound for reliability in the Technometrics, vol. 3, Feb 1961, pp. 55-77.
case of the Weibull Distribution", Technometrics, vol. 8, 54. J.D. Esary, F. Proschan, "Coherent structure of non-
1966, pp. 135-175. identical components", Technometrics, no. 5, 1963, pp. 191-
34. D.R. Thoman, L.J. Bain, C.E. Antle, "Maximum 209.
Likelihood Estimation, Exact Confidence Intervals for 55. Z.W. Birnbaum, J.D. Esary, "Modules of coherent
Reliability, and Tolerance Limits in the Weibull Distribution", binary systems", J. SIAM, vol. 13, 1965, pp. 444-462.
Technometrics, vol. 12, 1970, pp. 363-371. 56. Z.W. Birnbaum, J.D. Esary, A.W. Marshall,
35. H.F. Martz, B.S. Duran, "Comparison of three "Stochastic characterization of wear out for components and
methods for calculating lower confidence limits on system systems", Ann. Math. Stat., 37, 1966, pp. 816-825.
reliability using component data", IEEE Trans. Rel., vol. R- 57. Z.W. Birnbaum, "On the importance of different
34, no. 2, 1985. components in multicomponent system," Multivariate
36. A.J. Lotka, "A contribution to the theory of self- Analysis, II, 1969; Academic Press, New York, pp.581-592.
renewing aggregates with special reference to industrial 58. A. Satyanarayana, A. Prabhakar, "New topological
replacement", Ann. Math. Stat., vol. 10, 1939, pp. 1-25. formula and rapid algorithm for reliability analysis of complex
37. N.R. Campbell, "The replacement of perishable networks", IEEE Trans. Rel., R-27, 1978, pp. 82-100.
members of a continually operating system", J. Roy. Stat. 59. A. Satyanarayana, M.K. Chang, "Network reliability
Soc., vol. 7, 1941, pp. 110-130. and the factoring theorem", Networks, 13, 1983, pp. 107-120.
38. W. Feller, "On the integral equation of renewal 60. A. Bendell, J. Ansell, "The incoherency of multistate
theory", Ann. Math. Stat., vol. 12, 1941, pp. 243-267. coherent system", Rel. Eng., vol. 8, 1984, pp. 165-178.
39. D. Blackwell, "A renewal theorem", Duke Math. J., 61. A.B. Huseby, "Domination Theory and the crapo
vol. 15, 1948, pp. 145-150. beta-invariant", Networks, 19, 1989, pp. 135-149.
40. J.L. Doob, "Renewal theory from the point of view of 62. J.C. Fu, M.V. Koutras, "Reliability bounds for
the theory of probability", Trans. Amer. Math. Soc., 63, 1948, coherent structures with independent components", Stat. and
pp. 422-438. Prob. Letters, vol. 22, 1995, pp. 137-148.
41. W. Feller, "Fluctuation theory of recurrent events", 63. S. Tsitmidelis, M.V. Koutras, V. Zissimopoulos,
Trans. Amer. Math. Soc., vol. 67, 1949, pp. 98-119. "Evaluation of reliability bounds by genetic algorithm", Proc.
42. D.R. Cox, W.L. Smith, "A direct proof of a Math. Methods in Rel., 2002.
fundamental theorem of renewal theory", Skand. Aktuar., 36, 64. M.V. Boutsikas, M.V. Koutras, "Generalized
1953, pp. 139-150. reliability bounds for coherent structures", J. of Applied Prob.,
43. W.L. Smith, "Asymptotic renewal theorems", Proc. vol. 37, 2000, pp. 778-794.
Roy. Soc. Edin., A, 64, 1954, pp. 9-48. 65. T. Inagaki, E.J. Henley, "Probabilistic evaluations of
44. W. Feller, Introduction to Probability Theory and its prime implicants and top events for noncoherent systems",
Applications, 2nd ed., 1957; Wiley, New York. IEEE Trans. Rel., vol. R-29, no. 5, Dec. 1980, pp. 361-367.
45. W.L. Smith, "Renewal theory and its ramifications", 66. P.S. Jackson, "On the s-importance of elements and
J. Roy. Stat. Soc., Ser. B, vol. 20, no. 2, 1958, pp. 243-302. prime implicants of noncoherent systems", IEEE Trans. Rel.,
46. W.L. Smith, "On the cumulants of renewal vol. R-32, no. 1, Apr. 1983, pp. 21-25.
processes," Biometrika, vol. 46, issue 1/2, June 1959, pp. 1- 67. J.D. Andrews, S. Beeson, "Birnbaum's measure of
29. component importance for noncoherent systems", IEEE
47. R.E. Barlow, L.C. Hunter, "Mathematical models for Trans. Rel., vol. 52, no. 2, June 2003, pp. 213-219.
system reliability", Sylvanica Electronic Defense Labs., Rept. 68. T.F. Oltorik, "Reliability analysis using flowgraphs",
No. EDL-E35, Aug 1959. Air Force Institute of Technology, USGRR Document, no. AD
420-047, Aug 1963.
69. W.W. Haap, "Application of flowgraph techniques to 89. I.A. Papazoglou, "Mathematical foundations of event
the solution of reliability problems", J. Physics of failure in trees", Rel. Eng. and System Safety, vol. 61, 1998, pp. 169-
Electronics, US Department of Commerce Office of Technical 183.
Services AD-434/329, 1964, pp. 375-423. 90. J.D. Andrews, S. Dunnett, "Improving accuracy in
70. E. Dolozza, "System states analysis and flowgraph event tree analysis", Proc. Foresight and Precaution, Proc.
diagrams in reliability", IEEE Trans. Rel., R-15, 1966, pp. ESREL 2000, SARS and SRA-Europe Annual Conf., May
383-394. 2000.
71. M.M. Kleinerman, G.H. Weiss, "On the reliability of 91. J.D. Andrews, S. Dunnett, "Event tree analysis using
networks", Proc. Natl. Electronics Conf., vol. 10, 1954, pp. binary decision diagrams", IEEE Trans. Rel., vol. R-49, no.2,
128-136. June 2000, pp. 230-239.
72. E. Blanton, "Reliability-Prediction technique for use 92. A.N. Kolmogorov, "A number of target hits by
in design of complex systems", IRE Natl. Convention Record, several shots and general principles of effectiveness of
pt. 10, 1957, pp. 68-79. gunfire", Proc. Moscow Inst. Math., issues 12, 1945.
73. http://www.newhousenews.com/archive/macpherson0 93. R.O. Frantik, "The determination application and
21103.html limitations of circuit reliability equations", Sandia Corp.,
74. C.W. Holtzman, Jr., W.E. Marshall, "A new method report no. SC-3288(TR), April 1954.
of communication between engineering and mathematician 94. L. Takács, “On a sojourn time problem in the theory
aids system reliability prediction", Proc. Natl. Symp. Rel. and of stochastic processes,” Trans. Amer. Math. Soc., vol. 93,
Quality Ctrl., Jan 1960, pp. 403-408. issue 3, Dec. 1959.
75. P. Chatterjee, "Modularization of fault trees: A 95. E.A. Elsayed, A. Zebib, "A repairable multistate
method to reduce cost of analysis", SIAM Rel. and Fault Tree device", IEEE Trans. Rel., R-28, 1979, p. 81-82.
Analysis, 1975, pp. 101-137. 96. L.D. Bodin, "Approximation of system reliability
76. A. Rosenthal, "Decomposition methods for fault tree using a modular decomposition", Technometrics, vol. 12, no.
analysis", IEEE Trans. Rel., vol. 43, June 1980, pp. 136-138. 2, 1970, pp. 335-344.
77. J.B. Dugan, S.J. Bavuso, M.A. Boyd, " Dynamic fault 97. A.T. Bharucha-Reid, Elements of the Theory of
tree models for fault tolerant computer systems", IEEE Trans. Markov Processes and their Applications, 1960; McGraw-
Rel., vol. 41, no. 3, Sept 1992, pp. 363-377. Hill, New York.
78. J.B. Dugan, S.J. Bavuso, M. Boyd, "Fault trees and 98. J. Keilson, A. Kooharian, "On time dependent
Markov models for reliability analysis of fault tolerant queuing processes", Ann. Math. Stat., vol. 31, 1960.
systems", Rel. Eng. and System Safety, vol. 39, 1993, pp. 291- 99. D.M. Brender, M. Tainiter, "A Markovian model for
301. predicting the reliability of an electronic circuit from data on
79. R. Gulati, J.B. Dugan, "A modular approach for component drift and failure", IRE Intl. Convention Record,
analyzing static and dynamic fault trees", Proc. RAMS, 1997. 1961, pt. 6, pp. 230-241.
80. O. Coudert, J.C. Madre, "Fault tree analysis: 10^20 100. I.B. Pogozhev, "Estimation of deviation of failure
prime implicants and beyond", Proc. RAMS, 1993, pp. 240- flow in multi-use equipment from Poisson process",
245. Cybernetics in Service for Communism, vol. 2, Energiya,
81. S.A. Doyle, J.B. Dugan, "Dependability assessment Moscow, 1964.
using binary decision diagrams", Proc. IEEE Int. Symp. Fault- 101. A.D. Solov'yev, "Reliability and queueing theory:
Tolerant Computing, FTCS-25, June 1995. standby with rapid renewal," Eng. Cybernetics, no.1, 1970,
82. Y. Dutuit, A. Rauzy, "A linear-time algorithm to find pp. 49-64.
modules in fault trees", IEEE Trans. Rel., Sept. 1996. 102. A.D. Solov'vey, "Redundancy with fast repair", Eng.
83. H.F. Martz, R.G. Almond, "Using higher-level failure Cybernetics, no. 1, 1970.
data in fault tree quantification", Rel. Eng. and System 103. B. Ouhbi, N. Limnios, "Estimation of kernel,
Society, 1997, pp. 29-42. availability and reliability of semi-Markov systems", Proc.
84. A. Anand, A.K. Somani, "Hierarchical analysis of Stat. and Prob. Methods in Rel., 1999.
fault trees with dependencies, using decomposition", Proc. 104. B. Ouhbi, N. Limnios, "Non-parametric estimation
RAMS, Jan 1998, pp. 69-75. for semi-Markov processes based on its hazard rate", Stat. Inf.
85. R. Manian, D. Coppit, K.J. Sullivan, J.B. Dugan, Stoc. Proc., vol. 2, 1999, pp. 151-173.
"Bridging the gap between systems and dynamic fault tree 105. R.M. Fricks, K.S. Trivedi, "Importance analysis with
models", Proc. RAMS, Jan 1999, pp. 105-111. Markov Chains", Proc. Rel. and Main. Symp., 2003, pp. 89-
86. C.W. Griffin, "Introducing the fault tree as a tool for 95.
nuclear safety analysis", Trans. Amer. Nuclear Soc., vol. 9, 106. Y. Lefebvre, "Using equivalent failure rates to
no. 1, 1966, pp. 157. assess the unavailability of an ageing system", Proc. Rel. and
87. J.D. Andrews, "The use of Not logic in fault tree Main. Symp., 2003, pp. 82-88.
analysis", Qual. Rel. Eng., vol. 17, 2001, pp. 143-150. 107. E.F. Moore, C.E. Shannon, "Reliable circuits using
88. S.V. Tsitmidelis, S.V. Zissimopoulos, M.V. Koutras, less reliable relays", J. of the Franklin Institute, vol. 262,
"Fault tree analysis via set cover model", EUROGEN 2001, 1956, Pt. I, pp. 191-208, and Pt II, pp. 281-297.
Athens, Greece.
108. B.V. Gnedenko, "On spare duplication", Eng. 129. K.E. Portz, H.R. Smith, "Methods for the
Cybernetics, no. 4, 1964. determination of reliability", IRE Trans. Rel. and Quality
109. B.V. Gnedenko, "On duplication with renewal", Ctrl., vol. 11, Aug 1957, pp. 65-73.
Eng. Cybernetics, no. 5, 1964. 130. J.P. Lipp, "Topology of switching elements vs.
110. C.R. Gates, "The reliability of redundant systems", reliability", IRE Trans. Rel. and Quality Ctrl., vol. 10, June
JPL-California Inst. Of Tech. Pasadena, CA, Memo 20-76, 1957, pp. 21-33.
Aug 1952. 131. B.J. Flehinger, "Reliability improvement through
111. H. Mine, "Reliability of physical system", IRE redundancy at various system levels", IBM J. of RnD, issue 2,
Trans. Information Theory, IT-S (Special Supplement), 1959, 1958, pp. 148-158.
pp. 138-150. 132. J.L.Philipson, "Operational reliability model for a
112. F. Mosteller, R.E. Rourke, G.B. Thomas, Jr., reconnaissance system”, IRE Natl. Convention Record, pt. 6,
Probability with statistical applications, 1961; Addison- 1959, pp. 79-88.
Wesley, Reading. 133. M.I. Kaufmann, R.A. Kaufmann, "Predicting
113. I.I. Bazovsky, Relability Theory and Practice, 1961; reliability", Machine Design, Aug 1960, pp. 178-184.
Prentice Hall, Englewood Cliffs. 134. E. Cramer, U. Kamps, "Sequential k-out-of-n
114. H.F. Martz, R.A. Waller, Bayesian Reliability systems", http://www.math.uni-oldenburg.de/preprints/get/
Analysis, 1982; Wiley, New York. source/Hb.pdf, Feb 2000.
115. H.F. Martz, R.A. Waller, E.T. Fickas, "Bayesian 135. W.J. Kerscher III, J.M. Booker, M.A. Meyer, R.E.
reliability analysis of series systems of binomial subsystems Smith, "PREDICT: A case study, using fuzzy logic", Proc.
and components", Technometrics, vol. 30, 1988, pp. 143-154. RAMS, 2003, pp. 188-195.
116. H.F. Martz, R.A. Waller, "Bayesian reliability 136. M. Marseguerra, E. Zio, M. Ammaturo, V. Fontana,
analysis of complex series/parallel systems of binomial "Predicting reliability via neural networks", Proc. RAMS,
subsystems and components", Technometrics, vol. 32, 1990, 2003, Tampa, Florida, pp. 196-201.
pp. 407-416.
117. F.L. Hulting, J.A. Robinson, "The reliability of a BIOGRAPHIES
series system of repairable subsystems: a Bayesian approach",
Naval Research Logistics, vol. 41, 1994, pp. 483-506. Alice Rueda
118. W.J. Kerscher III, J.M. Booker, T.R. Bement, M.A. Telecommunications Research Laboratories (TRLabs)
Meyer, "Characterizing reliability in a product/process design #100-135 Innovation Drive
-assurance program", Proc. RAMS, 1998, pp. 105-112. Winnipeg, Manitoba Canada R3T 6A8
119. S.L. Firstman, "Monte Carlo models for estimating
reliability: An exploratory analysis", The Rand Corp., Report e-mail: alice@trlabs.ca
no. RM2149, ASTIA Doc. No. AD213036, June 1958.
120. Y.K. Belyaev, T.N. Dugina, E.V.Chepurin, Alice Rueda is currently working as researcher at TRLabs.
"Computation of lower confidence limit for the complex She is also pursuing doctoral degree in Electrical and
system reliability", Eng. Cybernetics, no's 2 & 3, 1967. Computer Engineering (ECE) at the University of Manitoba.
121. F. Moskowitz, J. McLean, "Some reliability aspects She has obtained M.Sc. (1999) in ECE at the University of
of system design", IRE Trans. Rel. and Quality Ctrl., vol. 8, Manitoba, Canada. She worked as software designer, software
1956, pp. 7-35. architect, system architect, product requirement specifications
122. R. Gordon, "Optimum component redundancy for and product line manager at Nortel Networks (1996-2000).
maximum system reliability", Operations Res., vol. 5, 1957, She was a senior network engineer and as the assistant of the
pp. 229-243. director in marketing and product management for Norsat
123. R. Bellman and S. Dreyfus, "Dynamic programming International (2000-2001).
and reliability of multi-component devices", Operations Res.,
vol. 6, 1958, pp. 200-206. Miroslaw Pawlak, PhD
124. G. Black, F. Proschan, "On optimal redundancy", Department of Electrical & Computer Engineering
Operations Res., vol. 7, no. 5, 1959, pp. 581-588. The University of Manitoba
125. J.D. Kettelle, "Least-cost allocation of reliability Winnipeg, Manitoba Canada R3T 5V6
investment”, Operations Res., vol. 10, no. 2, 1962.
126. R.E. Barlow, L.C. Hunter, F. Proschan, "Optimum e-mail: pawlak@ee.umanitoba.ca
redundancy when components are subject to two kinds of
failure", J. SIAM, vol. 11, 1963, pp.64-73. Miroslaw Pawlak is a research scientist at TRLabs. He is also
127. F. Proschan, T. Bray, "Optimum redundancy under a professor at the University of Manitoba in the Electrical and
multiple constraints", Operations Res., vol. 13, no. 5, 1965. Computer Engineering department. He obtained his M.Sc.
128. F.A. Tillman, H. Ching-Lai, W. Kuo, Optimization (1979) and Ph.D. (1982) degrees in Technical University of
of System Reliability, 1980; Marcel Dekker, New York. Wroclaw, Poland, Department of Engineering Cybernetics.
His Ph.D. title is "Pattern recognition by orthogonal
expansions" under the supervision of Professor W. Greblicki.
He had worked as research Associate in the Department of
Computer Science with Dr. A. Krzyzak (1985) at Concordia
University, Montreal. He was a visiting professor at the
University of Ulm, Department of Mathematics and
Information Science and at the University of Frankfurt,
Department of Statistics.

Вам также может понравиться