Dynamic Adaptive Hybrid Censoring Improves Partially Accelerated Life Testing Under Step Stress: Sustainability Application by Precipitation and Carbon Fibers Data

ABSTRACT Accelerated life testing is crucial for assessing the durability of products under various stresses like pressure, temperature, and voltage. The modified Kies exponential distribution, known for its flexible probability density and hazard rate functions among others, finds wide applications in real‐world scenarios including modeling accelerated life testing scenarios under step stress. In this study, we employ an adaptive Type II progressively hybrid censoring scheme assuming that the samples are drawn from the modified Kies exponential distribution in the analysis of step stress partially accelerated life testing models. Under the classical approach, maximum likelihood estimates of model parameters are computed, and estimated confidence intervals using three distinct methods which are approximate confidence, bootstrap, and pivotal quantity confidence intervals. Subsequently, various well‐known loss functions are utilized to derive Bayes estimates of unknown parameters. Furthermore, we explore three potential optimal test strategies to identify the optimal censoring scheme based on different optimality criteria. Through simulation experiments based on finite samples, we evaluate the effectiveness of the proposed techniques. Finally, we demonstrate the applicability of the suggested approach using an actual dataset in the context of accelerated life testing under the step stress.