Abstract
As conventional reliability demonstration test (RDT) can be time and cost prohibitive, reliability accelerated demonstration test (RADT) with harsher-than-normal stress levels is widely adopted for system-level products with high reliability and long life to verify whether they can meet a specified reliability requirement. However, the multi-source uncertainty due to the discrepancy of lifetime distribution and variances of field stress and usage rate could lead to an inaccurate estimation. This paper proposes a mathematical approach on the true demonstrated reliability of the zero-failure system-level RADT and discusses multi-source uncertainty effects. First, we present the RDT model with a Weibull lifetime distribution. Then, the comprehensive acceleration model is derived by synthesizing voltage and temperature stress, as well as usage rate in a time scale. Further, a system-level RADT model is established by connecting system target with unit reliability based on the risk priority number (RPN) method. The analysis method is illustrated using a numerical example on the solid-state drive (SSD), and sensitivity analysis is provided to evaluate the acceleration model and lifetime distribution uncertainty over the field product population.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call