Optimum Accelerated Life Test Plans for Log-Location-Scale Distributions with Multiple Objectives

Most of the previous work on planning accelerated life tests (ALTs) is focused on a sole estimating objective, such as some specified 100p-th quantitle lifetime, the reliability of the product over some specified period of time, and accelerating factor. However, it is impossible to estimate only a single objective parameter after conducting such costly tests. In this paper, we consider optimum constant stress ALT plans with multiple estimating objectives. We suggest two kinds of design criteria: one is minimizing the absolute difference between the variance of each objective and its minimum variance, and the other is minimizing the relative difference. A Weibull model is used to compare the proposed plans with other test plans in the literature for various combinations of planned value of the model parameters. The results show that the proposed plans provide acceptable precision of estimation, and the precision is much higher than the precision of estimation based on D-optimal plans.