Abstract:
Product reliability is a function of effective usage and applied stress (both operational and environmental) conditions. In real life, both usage and stress levels are not fixed, but random variables; i.e., they are statistically distributed. During engineering life test design, people often pick the high percentiles for population usage and stress level as the field reference for the sake of conservativeness and safety margin, such as 90th percentile, 95th percentile, or even 99th percentile. This often yields very ambitious and even unrealistic test sample size and/or test duration requirements because the majority (say 90% or higher) of users in the field are assumed to be operating under extreme usage and stress conditions. The question often comes up as to which usage and stress percentile is the appropriate choice so that the overall population reliability is assured. Is it mean, median, or some other percentile (such as 60%, 70%, etc.)? This talk is trying to answer that question by introducing the so-called Reliability Equivalence Principle, and then presenting the analytical expression of reliability-equivalent reference usage and stress value so that the same reliability target can be achieved at the population level. Numerical example is given to illustrate the advantage of the method for reliability life test design over the traditional practice, especially for a high-reliability product.
Co-sponsored by: SRE Society of Reliability Engineers
Speaker(s): Frank Sun
Room: Laural Room, Bldg: Senior Center, 550 E Remington Drive, Sunnyvale Community Center, Sunnyvale, California, United States, 94087
Reliability-Equivalent Field Reference Usage and Stress Level when Both are Random
