M.Ing. / Ever since man had invented the wheel, or for that matter any of his first hand tools, he had a definite need to maintain that object. Man did not have any means to predict failures and he had no idea of how to prevent them. He could not plan for these failures and would not know what material and resources to have on hand. He was obviously restricted to a very primitive and elementary form of breakdown maintenance or a sort of run to break strategy as we would call that in modern maintenance terms. His success could only be measured in terms of the time to the next failure. Due to the nonexistence of engineering standards, the large standard deviation in mean time between maintenance (MTBF) values would render these same values rather useless for predictive purposes. This situation restricted our forefathers to a very reactive approach to maintenance. The astonishing present levels of sophistication in man's skills, engineering design and manufacturing came about through revolution and evolution. This indicates a huge amount of change that has taken place over the years and is still taking place today. As the world is in a constant state of change no business organization can escape the effects of operating m a continually evolving landscape. The very forces of change come about typically through industrial globalization, technological advance, political upheaval, the opening up of new markets and the changing expectations of customers that become more knowledgeable and demanding. For any business to respond to customer needs in a satisfactory manner, that business will have to be highly dependent on logistics. As logistics is a major contributor to life cycle cost (Blanchard, 1992: 70-84) and therefor impacts directly on profit margins, there is a growing need for more effective and efficient management of an organization's resources. Logistic support for any organization or plant is a major consideration in the early design stages of any system or organization or plant. From a logistic support perspective it becomes therefor necessary to assure the effective and economical support of a system, organization or plant throughout its programmed life cycle. One of the most important logistic support elements is maintenance. A lot has been said and written about maintenance. A number of well known philosophies and procedures (Blanchard, 1992: 9-25) such as 'planned preventive maintenance' (PPM), `condition based monitoring' (CBM), 'reliability centered maintenance' (RCM), 'total productive maintenance' (TPM), 'just in time' (JIT), 'life cycle costing' (LCC), etc. have been developed over many years and are applied throughout the world today. Techniques such as 'failure mode effect and criticality analysis' (FMECA), 'fault diagnostics', 'quality circles', and others have also been introduced to the industry. However various studies (Willmott, 1990: 17) have shown that present levels of maintenance management effectiveness are still unsatisfactory. One of the most important reasons for this state of affairs is that maintenance is still viewed by many organizations as a technical activity rather than an integrated management discipline. This misconception is then reinforced by several factors such as that the maintenance department is at its most visible when an emergency malfunction occurs. As soon as the malfunction has been restored, production carries on with its activities and no one asks the question why the breakdown was not anticipated before it occurred and disrupted production. A second reason is that maintenance managers view themselves as fire fighters and not as managers. A third reason is that the maintenance department is viewed as an organizational function with unpredictable response times and erratic priorities. All of these prejudices and misconceptions are costing industry dearly. This study will indicate that much can be done to transform the maintenance function of any organization to such an extent that costs are minimized and plant availability is improved that will ultimately lead to higher profit margins.

M.Ing. / This dissertation covers a literature study of the reliability engineering, and this is then applied to the hospital oxygen supply system in order to determine the reliability of the system. The hospital oxygen supply system must comply with international and local legislation, which insists that the reliability of the system must be very high, since it supports life in the hospital. Since there were no previous studies conducted in terms of the oxygen supply system to the knowledge of the author, it definitely opens a new study field for the application of reliability engineering concepts. In the research it was found that no records were kept by the company on the failures occurring with the oxygen supply system. This increased the difficulty to calculate the actual reliability of the supply system. A reliability prediction was done, based on the failure rate data from a database. The reliability prediction of the .system was very low, and possibly not a very accurate prediction of the actual reliability of the system. The author therefore created a reliability calculation program, which calculates the reliability of the system and also keeps, an accurate failure data record on each component of the system. The main conclusion reached with this dissertation is that failure data feedback, and accurate records are a very important factor of reliability engineering. This may influence the company's ability to rectify design changes in their systems, as there is no idea where the failure occurred and how much money value is linked to the failures occurring.

In this dissertation the estimation of reliability for a developmental process generating attribute type data is examined. It is assumed that the process consists of m stages, and the probability of failure is constant or decreasing from stage to stage. Several models for estimating the reliability at each stage of the developmental process are examined. In the classical area, Barlow and Scheuer's model, Lloyd and Lipow's model and a cumulative maximum likelihood estimation model are investigated. In the Bayesian area A.F.M. Smith's model, an empirical Bayes model and a cumulative beta Bayes model are investigated. These models are analyzed both theoretically and by computer simulation. The strengths and weaknesses of each are pointed out, and modifications are made in an attempt to improve their accuracy. The constrained maximum likelihood estimation model of Barlow and Scheuer is shown to be inaccurate when no failures occur at the final stage. Smith's model is shown to be incorrect and a corrected algorithm is presented. The simulation results of these models with the same data indicate that with the exception of the Barlow and Scheuer's model they are all conservative estimators. When reliability estimation with growth is considered, it is reasonable to emphasize data obtained at recent stages and de-emphasize data from the earlier stages. A methodology is developed using geometric weights to improve the estimates. This modification is applied to the cumulative MLE model, Lloyd and Lipow's model, Barlow and Scheuer's model and cumulative beta Bayes model. The simulation results of these modified models show considerable improvement is obtained in the cumulative MLE model and the cumulative beta Bayes model. For Bayesian models, in the absence of prior knowledge, the uniform prior is usually used. A prior with maximum variance is examined theoretically and through simulation experiments for use with the cumulative beta Bayes model. These results show that the maximum variance prior results in faster convergence of the posterior distribution than the uniform prior. The revised Smith's model is shown to provide good estimates of the unknown parameter during the developmental process, particularly for the later stages. The beta Bayes model with maximum variance prior and geometric weights also provides good estimates.

Digitized by Kansas Correctional Industries

Call number: LD2668 .T4 1965 D227 / Master of Science

published_or_final_version / abstract / Mechanical Engineering / Master / Master of Philosophy

Throughout this work, conditional failure rates for discrete positive integer-valued random variables and some of their applications are considered in some detail. Conditional failure rates are of fundamental importance in the study of lifetime distributions and many of their properties. All the notions introduced and the results derived here can be used in reliability theory, operations research, inventory theory, biometry, etc. Chapter 1 begins with the concept of conditional failure rate of a discrete random variable. Then, it is shown how to obtain explicit expressions for probability densities and survival distributions in terms of this notion. Next, extensions of the univariate results are discussed for bivariate discrete random vectors. Finally, some multivariate concepts and results are outlined. One of the fundamental applications of conditional failure rates is the mathematical representation of ageing. In Chapter 2, several univariate notions of ageing are given for discrete random variables. Such notions constitute the starting point for the classification and study of lifetime distributions that have significant importance in reliability theory, biometry, and several other areas. In Chapter 3, three important ordering relations, and a chain of implications among them, are discussed; the likelihood ratio ordering, the failure rate ordering, and the stochastic ordering. These orderings are useful in applied probability, stochastic processes, statistics, etc. In particular, they are an essential tool in the study and analysis of systems with dependent components, specially when the components are associated. No attempt is made, however, to consider specific applications of these orderings here. Finally, Chapter 4 contains an application of conditional failure rates in the analysis of repairable systems. A random mechanism of repair of failed units, called imperfect repair, is introduced and some simplified models are considered in some extent. These models can be used in the analysis and design of maintenance policies.

The finite mixed Weibull distribution is an appropriate distribution in modeling the lifetime of the units having more than one possible failure cause. Due to the lack of a systematic statistical procedure of fitting the distribution to a data set, it has not been widely used in lifetime data analyses. Many areas on this subject have been studied in this research. The following are the findings and contributions. Through a change of variable, 5 parameters in a two Weibull mixture can be reduced to 3. A parameter'vector (p₁, η, β) defines a family of two-Weibull mixtures which have common characteristics. Numerous probability plots are investigated on Weibull probability paper (WPP). For a given p₁ the η-β plane is partitioned into seven regions which are labeled by A through F and S. The Region S represents the two Weibull mixtures whose cdf curves are very close to a straight line. The Regions A through F represent six typical shapes of the cdf curves on WPP, respectively. The two-Weibull mixtures in one region have similar characteristics. Three important features of the two-Weibull mixture with well separated subpopulations are proved. Two existing methods for the graphical estimation of the parameters are discussed, and one is recommended over the other. The EM algorithm is successfully applied to solve the MLE for mixed Weibull distributions when m, the number of subpopulations in a mixture is known. The algorithms for complete, censored, grouped and suspended samples with non-postmortem and postmortem failures are developed accordingly. The developed algorithms are powerful, efficient and they are insensitive to the initial guesses. Extensive Monte Carlo simulations are performed. The distributions of the MLE of the parameters and of the reliability of a two Weibull mixture are studied. The MLEs of the parameters are sensitive to the degree of separation of the two subpopulation pdfs, but the MLE of the reliability is not. The generalized likelihood ratio (GLR) test is used to determine m. Under H₀: m=1 and H₁: m=m₁>1, ζ, the GLR is independent of the parameters in the distribution of H₀. The distributions of ζ or -21n(ζ) with n=50, 100 and 150 are obtained through Monte Carlo simulations. Compared with the chi-square distribution, they fall in between x²(4) and x²(6), and they are very close to x²(5). A FORTRAN computer program is developed to conduct simulation of the GLR test for 1 ≤ m₀ < m₁ ≤ 5.

D. Ing. / This dissertation deals with the important issue of reliability management for 3kV DC Traction Substations used by the national railway company in South Africa. Maintenance is one of the critical and most costly phases in the lifecycle of any plant. It looks at the total life cycle of the equipment in a typical substation, but the focus in the latter chapters is on the maintenance. Through improved maintenance management, the reliability of the system can be improved. The approach to maintenance is addressed as a predictive strategy, avoiding even more costly nonproductive time due to downtime caused by failure or induced by maintenance. Condition monitoring and assessment is described as one of the effective tools in the maintenance engineer’s armoury to apply a predictive approach. A direct link between predictable maintenance and reliability is explored. In the definition of reliability, concepts such as time and expected performance can be linked to a predictable delivery of the designed function. In other words, if down time is expected and can be prepared for, it is more acceptable than the unexpected. In essence, the system is still reliable as it performs according to expectation. The concept of predictable maintenance can be applied wider than just the 3kV traction substation. The process of identifying critical equipment, to measure the condition and to take decisions based on the rate of change in the condition can be used in any maintenance environment, even outside electrical. The crucial ingredient to this is to understand that condition monitoring is not based on fixed values, but the rate at which these values change. This is called Fuzzy logic. Can we predict the future? If yes, how accurate will the predictions be?

by Siu-Keung Wan. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1996. / Includes bibliographical references (leaves 69-70). / Chapter Chapter 1 --- Introduction and Review / Chapter 1.1 --- Introduction --- p.1 / Chapter 1.2 --- Review --- p.3 / Chapter Chapter 2 --- A general IRR model for a system with unobservable states / Chapter 2.1 --- Model and assumptions --- p.8 / Chapter 2.2 --- Long-run average cost per unit time incurred in a replacement cycle --- p.14 / Chapter Chapter 3 --- The Algorithm / Chapter 3.1 --- The key point and the local turning point of the average cost --- p.16 / Chapter 3.2 --- A finite algorithm --- p.23 / Chapter Chapter 4 --- Numerical examples and sensitivity analysis with discussion / Chapter 4.1 --- Weibull Distribution Case --- p.25 / Chapter 4.2 --- Gamma Distribution Case --- p.28 / Chapter 4.3 --- Exponential Distribution Case --- p.30 / Chapter 4.4 --- Sensitivity analysis --- p.32 / Chapter 4.5 --- Conclusion and further development --- p.38 / Figures --- p.40 / Tables --- p.46 / References --- p.69