Introduction to Reliability Engineering

When something or someone is reliable they are dependable and trustworthy. They can be counted on to get the job done, and they do not fail in the task they are expected to do, Reliability Engineering is so named because it is concerned with the dependability and expected performance of products and systems. Engineers in every field apply the sciences of physics and mathematics to find suitable solutions to problems or to make improvements. What distinguishes reliability engineering is that it has a time-based concept of quality. It is concerned not only with whether or not a product does fail, but also the time when the failure of that product occurs.

A formal definition would be that Reliability Engineering is a field that deals with the ability of a system or a component to perform its required functions under stated conditions for a specified period of time. Predicting if a product will or will not fail and when this failure might happen involves uncertainty. These are questions that reliability engineers answer with mathematical probabilities and statistical methods.

The main objectives of reliability engineering are to prevent or reduce failures, to identify and correct the causes when failures do occur, to find ways of coping with failures if their causes have not been corrected yet, and to estimate the reliability of new designs and analyze reliability data. These tasks are managed by a reliability engineer, who will have an accredited engineering degree and have additional reliability-specific education and training. Reliability engineering as a separate discipline originated in the United States during the 1950's and has since continued to grow in its effectiveness and influence.

"What Makes Reliability Analysis Different?

* The central role played by the Weibull and the lognormal distributions, rather than the normal distribution, to represent the statistical distribution of product lifetime.

* The interest in the distribution tails (for example, determining the time at which 1% of the product will fail), rather than mean life and its standard deviation.

* The prevalence of censored data (for example, on unfailed units for which the current running times, but not the eventual failure times, of some units are known).

* The use of accelerated testing to help measure and improve reliability.

* The frequent need to evaluate the reliability of systems made up of often replaceable parts, each with their own lifetime distributions.

* The frequent need to extrapolate beyond the range of the data (for example, extrapolation in time to predict three-year warranty returns from one-year data or extrapolation in temperature to assess device reliability from a high-temperature accelerated test)."

For more great references and materials see ASQ's web site at http://www.asq.org/.

Hahn, Gerald J., Doganaksoy, Necip, "Statistics, By the Numbers." Six Sigma Forum Magazine, Aug. 2009