Context
Safety studies of control systems and equipment are often established from predictive reliability studies. It is necessary to compare these studies with an analysis of operational reliability resulting from failures observed on sites. There are several databases of feedback from the fleet of reactors in operation. We are currently developing a tool to analyze this data from site observations.
Technical objective of the internship
As part of the creation of the Return of Experience exploitation tool, data pre-processing work (from the different databases) must be carried out to standardize the data and refine the data already entered.
Educational objectives
Confronting field data and making it usable,
Understand the methods for calculating operational failure rates and the limits imposed by the data (obtain consistent and accurate results),
Become familiar with the Weibull distribution.
Internship content
Based on data from the Qualex, Base Composants, Bapplication databases and the T-REX Tool specification, the work consists of requalifying certain data in order to reclassify/requalify them (eg add the distinction between Human Common Cause Failures (CCF), Design CCF, Manufacturing CCF, induced failures; identify CCF failures, and identify the nature of the CCF),
Establish formulas to fill in missing or tool-specific fields (e.g. find the number of the reactor concerned, from the failure dates and start dates of the units),
Retrieve and import all card nomenclatures, in all their versions,
Finalize the regression with a 3-parameter Weibull distribution.