Nowadays industries have to place their products in a highly competitive market, and there have been increasing pressures on efficiency test programs and product development cycles. In this context, the classical statistical methods (based on sampling theory), which are used to analyze product reliability, are negatively affected, while the bayesian methods are favorably regarded, and their use is becoming a necessary alternative. In order to evaluate the performance of an item with a very low failure rate, the classical statistical techniques require a large sample size and a long test time. This is particularly true when the product technology limits the acceleration factor, as with electronic products, for example. The methodology proposed in this paper combines test results, which are routinely performed during the product development cycle, with additional relevant information that is useful in the assessment of its reliability. In order to illustrate the methodology, it was applied to an electronic equipment, assessing its reliability during the design phase. The computations were performed considering component reliabilities, attribute test data, and also judgement of the product development team.
reliability; bayesian statistics; quality; failure rate; expert opinion
