Reliability estimation for one-shot devices under cyclic accelerated life-testing

Xiaojun Zhu, Kai Liu*, Mu He, N. Balakrishnan

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

12 Citations (Scopus)

Abstract

A one-shot device, like an automobile airbag, is a product or an equipment that can be used only once. Better quality and longer lifetime of one-shot devices nowadays increase the cost of life test experiment under normal operating condition. Cyclic stress test, adopted in life-testing experiments by increasing stress levels to induce more failures, has been used to investigate the reliability analysis of one-shot devices based on Coffin–Manson principle. However, the Coffin–Manson model only considers temperature change in each cycle. Moreover, it assumes that the reliability is independent of the cycling frequency, which may not be a realistic assumption in practice. Birnbaum–Saunders distribution, originally developed to model fatigue failure under cyclic loading, has been used widely to model lifetime data. As the Norris–Landzberg model is proposed for modeling fatigue life due to cyclic temperature fluctuation, it is used in this work together with Birnbaum–Saunders distribution for modeling lifetimes of one-shot devices under accelerated life-tests with different cyclic temperature fluctuations. It contains the Coffin–Manson model as a special case. Inferential methods for model parameters, reliability and mean lifetime are developed in this paper. Simulation study and model discrimination are carried out to evaluate the performance of the proposed model and inferential methods. Finally, an example is presented to illustrate the model and the inferential results developed here.

Original languageEnglish
Article number107595
JournalReliability Engineering and System Safety
Volume212
DOIs
Publication statusPublished - Aug 2021

Keywords

  • Accelerated life-test
  • Birnbaum–Saunders distribution
  • Coffin–Manson model
  • Cyclic tests
  • Norris–Landzberg model
  • One-shot device testing
  • Stochastic EM-algorithm

Cite this