Reliability growth

´╗┐Reliability is a key product attribute as failure of products to function can be costly and lead to unsafe operating conditions. Improving reliability is referred to as reliability growth. This is achieved through a programme of analysis, testing and re-design. Such programmes are costly and modelling plays a vital role in decision support eg decisions to allocate resources in an attempt to realise improvements in system reliability, or decisions to terminate testing when there is evidence that reliability has matured.

In the department of Management Science, we've been involved in various projects concerning modelling reliability growth. Some highlights of our work include the development of:

  • processes for the elicitation of expert judgment to construct prior probability distributions  which support Bayesian inference on test data
  • inference methods for assessing reliability metrics with few data
  • cost models for decision support on optimal time to terminate testing
  • methods to support the management of reliability programmes

Our research in this area is high on impact. We've developed these methods in close collaboration with industry. Our work has informed the content of international standards on reliability growth modelling (IEC61164). These are adopted globally by manufacturing organisations or agencies procuring engineering systems.

Publications

PhD opportunities

Professor Tim Bedford’s research interest concern risk analysis and decision making under uncertainty.  Topics of particular interest include:

  • Bayesian methods
  • Bayes Linear
  • Maximum Entropy
  • Vines
  • Elicitation of Expert Judgement

Find out more about how to apply for a PhD in Management Science.