A note on experiment design for accelerated life tests

Abstract Accelerated life tests which are designed to accelerate chemical or electrochemical failure modes are often designed so that no single device sees more than one level of stress (a constant stress accelerated life test). The purpose of this note is to show that this practice is potentially misleading, and to derive a class of experiment designs for accelerated life tests which performs much better for a broad class of models of chemical or electrochemical failure modes.