612 – Product Reliability and Life Testing
Lower Tolerance Bounds in Accelerated Life Testing for the Weibull Distribution
Ananda Jayawardhana
Pittsburg State University
V.A. Samaranayake
University of Missouri-Rolla
In this paper we investigate the problem of obtaining lower tolerance bound for a future observation from a Weibull population at field use (design) stress level, using Type II censored accelerated life test data from two levels of stress higher than design stress levels. The scale parameter of the life distribution is assumed to have an inverse power relationship with the stress level. We use the maximum Likelihood Predictive Density method to derive a predictive density for a future observation as described by Jayawardhana and Samaranayake (2003). The use of a lower percentile point of the predictive density as a lower tolerance bound is investigated using Monte Carlo simulation. The results show that reasonable tolerance bounds can be provided using the predictive density for different levels of tolerance, tolerance content, sample sizes, and acceleration factor. An example using the data from Zhang et al. (2012) is demonstrated.