Book Review

Our Review

First, let me declare my bias - I am not a strong believer in the validity of Weibull analysis and other statistical techniques at a practical level, in a Maintenance environment. This book comes strongly from the statistical analysis "camp" of engineers. So it was with some scepticism that I started to read this book. However, by the end, I can honestly say that this book is excellent, and that all Maintenance engineers who want a good, practical text which covers the principles behind Weibull, and other reliability modelling techniques, should read it.

The author, David Smith, is a past chairman of the UK Safety and Reliability Society, and is clearly eminently qualified and experienced in the field. This shows throughout the book. The writing style is very readable.

The book starts by giving a brief outline of the history of safety and reliability technology, and outlines definitions for some key terms. It then moves on to a section on interpreting failure rates (which includes a good explanation of Weibull, its applicability, and its weaknesses). The next section focuses on Reliability prediction and modelling techniques, and covers all of the major modelling techniques commonly in use, including Markov Analysis, Reliability Block Diagrams, FMEA, QRA and others. Specific mention is made of issues relating to human reliability, and software reliability. Finally, it includes a section relating to Legal, Management and Safety considerations, written from the perspective of a supplier who is being asked to guarantee performance of an equipment item or system after it has been installed.

Mr Smith clearly has an instrumentation and software background, and it is in this area that statistical modelling techniques perhaps have the greatest likelihood of impact from a Maintenance perspective. RCM techniques, in these areas, tend to state that instrumentation or software failures are not predictable, nor (generally) does their probability increase as the item becomes older, and so, if the risk of the failure is not tolerable, then the RCM decision diagram tends to default to a "Redesign" - either mandatory or optional. In this situation, then Reliability modelling techniques can be useful in assisting to determine which "redesign" is most likely to lead to an acceptable reduction in the level of risk.

Nevertheless, the author is quite clear about the limitations of the technique. He points out the dangers of using generic failure data, drawing on publically available data from 4 sources to illustrate that failure rates for the same electronic component, operating at the same temperature, can vary by as much as 20,000%, or 20 orders of magnitude. Equally, he states that, when generating your own failure data and applying Weibull that "a satisfactory result will not be obtained with less than at least six (failure) points. On the other hand, as long as you are aware of the limitations of the approach, then there may be situations that making an educated guess is better than making no guess at all.

Later in the book (afficionados of the SAE standard on RCM JA1011 take note) he introduces a technique called "Quantified RCM", which involves using statistical modelling and calculation to balance the cost of unavailability with that of excessive maintenance. We won't go into the debate here, except to say that this approach is not "RCM" as defined by JA1011, and, given the above admissions regarding the accuracy, and availability, of failure data, this (albeit short) chapter is probably best left unread.

Nevertheless, there is a lot of valuable information in this book. Those involved in Reliability Modelling would find it a useful desk reference which they would keep coming back to again and again.

Copyright 1996-2009, The Plant Maintenance Resource Center . All Rights Reserved.
Revised: Thursday, 08-Oct-2015 12:08:05 AEDT
Privacy Policy