Reliability is the study of the failure of systems. These systems can involve mechanical or electrical machinery, computer software, weapons or materials. Reliability theory is concerned with modelling the failure mechanisms of systems. Statistics enters reliability theory when failure is an inherently unpredictable phenomenon in which case statements about reliability can only be made in terms of probabilities. Reliability is not a physically measurable quantity like electrical resistance or elastic modulus and can only be assessed via a statistical analysis of data collected from past experience or experimentation. This book describes statistical techniques used for the assessment of reliability. Aimed at readers who have a first course in statistical methods, it develops the specific techniques used in reliability data analysis from a modern computer-orientated viewpoint. Techniques covered include probability plotting, maximum likelihood and Bayesian methods, proportional hazards modelling and the analysis of repairable systems. Some of these techniques are already familiar to those working with survival data in a medical context, but this book describes the differences as well as the similarities when they are applied in reliability. Several data sets are included in full. The final chapter includes the subject of load-sharing systems, highlighting a class of models that have not previously been covered in book form. The book is designed for industrial statisticians, students and teachers of applied statistics, engineers and computer scientists wishing to extend their knowledge of statistics as it is applied to their disciplines.

Table of Contents:
Part 1 Statistical concepts in reliability: reliability data; repairable and non-repairable systems; component reliability and system reliability; the binomial and hypergeometric distributions; the Poisson process; the reliability literature. Part 2 Probability distributions in reliability: preliminaries on life distributions; the exponential distribution; the Weibull and Gumbel distributions; the normal and lognormal distributions; the gamma distribution; some other lifetime distributions; censored data; simple data analytic methods - no censoring; data analytic methods - type II censoring, general censoring. Part 3 Statistical methods for single samples: maximum likelihood estimation - generalities, illustrations; tests and confidence regions based on likelihood; remarks on likelihood-based methods; goodness-of-fit. Part 4 Regression models for reliability data: accelerated life models; proportional hazards models; proportional odds models; generalizations; an argument from fracture mechanics; models based on the Weibull distribution; an example - breaking strengths of carbon fibres and bundles; other examples of comparing several samples; Weibull ANOVA; Buffon's beams - an historical example of reliability data. Part 5 Proportional hazards modelling: analysis of the semiparametric PH model; estimation of the survivor and hazard functions; model checking; numerical examples. Part 6 The Bayesian approach: a review of the Bayesian approach to statistics; elements of Bayesian statistics; further topics in Bayesian inference; decision analysis; Bayesian analysis of reliability data. Part 7 Multivariate models: some multivariate failure time distributions; complete observation of "T"; competing risks. Part 8 Repairable systems: framework; ROCOF; simple statistical methods; non-homogeneous Poisson process models; NHPP with log-linear ROCOF; NHPP with ROCOF v2; choice of NHPP model. Part 9 Models for system reliability: coherent systems; estimation of reliability for coherent systems; multi-state reliability theory; load-sharing systems - the Daniels model; extensions of the Daniels model; time to failure; a more general model; local load-sharing; exact calculations; approximations for local load-sharing systems; statistical applications of load-sharing models. Appendix: The delta method.