The structure and exploration of reliability field data: What to look for and how to analyse it

It has for a long time been common practice to analyse reliability field data using standard techniques based upon certain implicit assumptions. Although there has been recent interest in testing for trend, the possibility of other abnormal structure in reliability time between failure data of repairable equipment has seldom been explored. Extensive investigation of the structure of diverse reliability data indicates that other structural relationships exist. Aspects such as serial correlation, non-monotonic trend, dependence between components, fault-free periods and non-exponential behaviour need to be investigated and eliminated before standard methods of analysis or databank storage, based on constant fault rates, can be employed. This paper describes common statistical structures of diverse reliability field data and how to explore and identify the structure in order to analyse the data appropriately.