Fréquence des évènements initiateurs d’accident

Safety cases for industrial plants are often based on a semi-probabilistic risk assessment method, such as LOPA (Layer Of Protection Analysis). These methods require the ability to estimate in quantitative terms both the frequency of initiating events (such as leaks or pump failures) and the availability of safety barriers (preventive barriers such as pressure relief valves as well as protective barriers such as blastproof control rooms). This failure and availability data is also used for preventive maintenance.

This data concerns rare events, and observations on a single industrial facility are generally insufficient to provide adequate estimations. Therefore, operators collect information on equipment failures and barrier dependability at an industry-wide level, over a large population of installed equipment. However, failure rates are known to depend on numerous factors, such as the quality of the installed equipment (dependent on a plant's purchasing policy), the inspection and maintenance activities, and the age of equipment.

The present document is the result of a working group on initiating event frequencies and barrier availability in the petrochemical, chemical and non-nuclear energy sectors. Participants from a number of industrial firms operating in Europe exchanged information on the quantitative frequency ranges used in their firm for different types of events, on appropriate data sources and on factors used to select a low or high value within the suggested frequency range. The frequencies and probabilities given in this document are averages over these industries, that working group participants consider to be reasonable generic values.

This document will be useful for people wishing to carry out a risk assessment of an industrial facility or wishing to conduct a critical review of an existing risk assessement. When using values from this document, it is important to take into account the specific characteristics of the plant being studied, including specific technological characteristics and organizational issues.