Inference for Variable-Stress Experiments

• Activity Number: 165
• Type: Topic Contributed
• Date/Time: Monday, August 2, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Section on Quality and Productivity
• Author(s): Nandini Kannan*+
• Companies: The University of Texas at San Antonio
• Keywords: step-stress ; variable stress ; hazard function ; cumulative exposure model

Abstract

In this talk, we consider models for experiments in which the stress levels are altered at intermediate stages during the exposure. These variable-stress tests are part of the class of accelerated models used in reliability and life-testing applications. We consider a model in which the stress levels are applied for a fixed duration and then removed. After a period of rest, the stress is reapplied. We introduce a model that modifies the cumulative exposure model of Nelson to account for the lag effect that occurs before the effects of the change in stress are observed. We assume a continuous, piecewise constant hazard function. The model is fit to real data on decompression sickness, where subjects in an altitude chamber were exposed to altitudes in a repeated pattern. We compare the overall decompression rates for the variable stress model with the increasing stress model.