Credible Reliability Prediction Monograph
Credible Reliability Prediction Monograph
by Laurence George
Engineers want to know how things work. Reliability engineers want to know how things fail. My goal has been to help engineers and service providers learn and use age-specific field reliabilitythat is, what really happens to products in the hands of customers as a function of agebecause the data and methods are available, the needs are great, and we’ll get fired if we don’t produce credible predictions!
Meanwhile, people want MTBF (Mean Time Between Failures) and age-specific reliability predictions. They want these predictions for systems made complex by redundancy, new designs, repair, dependence, and maintenance. They want these predictions for more than comparison; they also want these predictions for
- Evaluating alternative designs and allocating reliability
- Estimating warranty risk and setting warranty reserves
- Providing early warning and statistical process control of excessive infant mortality or premature wear out
- Forecasting service and spares requirements
- Verifying improvement and evaluating and verifying fixes
Reliability engineers recognize that MTBF prediction methods don’t accurately predict age-specific field reliability. These engineers are paid for accurate, precise, and defendable predictions that give just cause for design, process, and service actions, so I try to help them make the best predictions possible, objectively, accurately, and precisely, using all available data and information.
Part of the accuracy problem is resistance to change. Parts-count MTBF predictions require only looking up part FITs (Failures In Time, also Failures In Thousands of millions of hours), multiplying by -factors, adding the products, and inverting the sum. This simplicity has such strong appeal that the parts-count method persists despite out-of-date FITs data, invalid model assumptions, irrelevance of MTBF during useful life, and lack of credibility. The methods in this monograph create the need for learning and using age-specific field reliability while improving accuracy of MTBF predictions.
This monograph describes methods that extend MTBF prediction to complex, redundant, dependent, standby, and life-limited systems. It also describes a credible method for predicting age-specific field reliability, using observed reliability of comparable products or parts. Finally, this monograph adapts insurance credibility to update predictions as user data becomes available.
- People need statistically correct and credible MTBF and age-specific reliability predictions.
- The methods are credible, because they are based on the observed reliability of products and parts.
- Generally accepted accounting principles require statistically sufficient data for making nonparametric estimates of age-specific reliability and failure rate functions.
This monograph is intended for practicing reliability engineers. Those who might become practicing reliability engineers and others who may be interested in what practicing reliability engineers can do are also welcome to read the monograph. I hope that teachers will also be interested, because the monograph introduces new methods. I’ve omitted theorems and proofs, so there’s a lot of room for academic work. I hope managers will also be interested, because the new methods provide more accuracy and precision leading to cost savings, profit, and reduction in uncertainty, for management benefit.
It is not necessary to read earlier chapters to understand later chapters. Some may regard chapters 1 through 3 and 5 as a diatribe but they are intended to be motivation. If you’re already motivated, read chapters 4, 6, and 7; they contain the new methods for MTBF and reliability prediction. Perhaps these new methods will satisfy some needs I hadn’t thought of.
Chapters 2 and 3 summarize the MTBF prediction state of the art and criticize it. Chapter 5 describes the best of all possible worlds, in which people acquire old product user data, incorporate it in reliability predictions, use them, modify them as new product user data accumulates, plan for obsolescence, and feed reliability information back to repeat the process. Chapters 4, 6, and 7 provide extensions to MTBF and reliability prediction, and to incorporating field data as it becomes available. Chapter 4 extends MTBF prediction methods and corrects a common error. Chapter 5 explains that statistically sufficient data is available to support age-specific reliability predictions for all products and service parts. Chapter 6 proposes a method for making age-specific reliability predictions by using observed reliability and MTBF predictions for comparison products. Chapter 7 adapts insurance credibility to updating predictions as user data accumulates. Chapter 8 describes a do-it-yourself MTBF prediction workbook and spreadsheet implementations of the MTBF and reliability prediction extensions in chapters 4, 6, and 7.
Download link: Credible Reliability Prediction