Data collection on human factors at Electricite de France (EDF)

This paper is concerned with the collection of data on human errors that occur at operating power plants. Three collection methods are used, each relating to a different level of analysis. (1) Simplified statistical analysis of the causes of human errors: Events which have occurred at operating power plants and which are attributable to human errors are selected. The errors thus identified are analysed briefly and are described by a simplified classification, statistical analysis then being applied to find the principal factors underlying these errors. By way of example, an analysis is given of data on emergency shut-downs involving a human error component that occurred at 900 MW(e) PWR plants during 1982. (2) In-depth statistical analysis of the causes of certain human errors: The errors selected are analysed and described by means of a detailed classification. By way of example, the collection and evaluation of data on human errors occurring during periodic tests at a 900 MW(e) power plant over a period of six months are described. (3) In-depth analysis of certain events due to human errors: The events selected are analysed by means of a method which reconstitutes the multicausal aspect of the event and of each human error. By way of example, a description is given of an emergency core cooling required at 900 MW(e) PWR plant. In conclusion, it is explained how these three methods of collection play complementary roles. (author)