ECONOMIC OF RELIABILITY TESTING

(The Five Rules for Effective Reliability Program)

Statistical Bulletin - Volume 11 - Bulletin 2

(Steps to Reliability)

Statistical Bulletin - Volume 16 - Bulletin 8

(A Typical Scenario on the Steps to Reliability)

Statistical Bulletin - Volume 17 - Bulletin 1

 Anyone responsible for the reliability of the consumer product which his or her industries produces must have an adequate testing program for every component and assembly which is finally sold to customers.  For last fifty years the Weibull Method of treating durability test data has been employed by American and foreign manufacturers in their testing programs.  Now there is an innumerable number of so-called experts in Weibull Analysis, who are advertising their seminars and computer software in the Weibull approach to reliability testing.  However, these Johnny Come Lately are not up to- date with the most powerful Principles of Weibull Reliability Method based on the techniques which have been formulated by  DRI's Leonard G. Johnson, American Originator of Weibull Analysis. Furthermore, many of the new "experts" are not yet familiar with the latest powerful techniques such as:

• The Concept of Entropy as applied to Reliability
• The Concept of Evidence as related to Product Compliance
• The Universal Law of Odds
• The Use of Semi-Parametric Decision Methods (In order to generate more reasonable Confidence Bands)
• Design Life Tests with sample sizes and confidence levels dictated by possible gains due to compliance when compared to possible losses due to non-compliance to desired goals.
• Employing the Dollar Basis of Confidence, instead of using confidence indices taken off the top of somebody's head.  For example, some experts are so ignorant that they think that 90% Confidence is always adequate, when the truth is that it is totally inadequate whenever a non-compliance case losses more than nine times what a complying sale to a customer gains.  This is because 90% Confidence represents 9:1 Odds, since the 90% in favor of compliance also means 10% probability against compliance, and the Ratio of these two probabilities is 9:1.

In today's economy, consisting of domestic and foreign enterprises in competition with one another, the business manager must concern himself or herself with profits and losses.  For this simple reason, product reliability and the confidence of attaining it must take profits and losses into account.  Insufficient reliability causes excessive losses when warranty promises are violated too frequently, thus putting a business in jeopardy.  On the other hand, aiming for absolutely perfect reliability can be so expensive that the price we must charge for our product to make money becomes so high that competitors with lower prices and acceptable none-perfect reliability will take away our customers.  What all this amounts to is the fact that there are optimum reliability and confidence levels, which dictated by possible gains and losses.  This, in turn, means that the test sample size for reliability are also determined by the economic factors of profits and losses.

Detroit Research Institute provides clients with software tool for implementing corporate reliability strategies and improving financial results. Our software system can assist you in determining the role of profits and losses in product reliability developmental programs needed in qualifying a product before it is finally released for production and sold to the consumer to generate an acceptable profit margin.  Thus, it actually turns out that the degree of profitability, as defined by the ratio of (Gains/Losses), should be the fundamental criterion for the reliability required, together with the proper confidence level, as the determining factor for the sample size required in a life test.