Simulation-based optimal design for accelerated degradation tests

Abstract

Accelerated degradation test (ADT) is widely used to evaluate the reliability of highly reliable products with long life if there exists a product quality characteristic whose degradation over time can be related to reliability. It is still a main challenge that how to design optimal plan, including sample size, measurement frequency, and termination time, for ADT which can result in the most accurate estimation within predetermined budget. This paper presents a novel approach of optimal design for ADT based on Monte Carlo simulation. There are three key steps in the approach: simulating; modeling; optimizing, by which it is convenient to set up a framework for applications. In this paper, we first apply a widely-used mix-effect model to describe a typical ADT problem. Next, the optimal test plans are obtained by minimizing the estimate of the mean-squared error of the estimated 100pth percentile of the product's lifetime distribution by statistical analysis of simulating degradation data of test units under the constraint that the total experimental cost does not exceed a predetermined budget. Finally, an example is provided to illustrate the proposed method and evaluations are made to assess the sensitivity of optimal plans to misspecifications of values of the parameters.