Principles for assessing the probability of failure of Risk Control Measures (RCMs)

Industrial risk management requires manufacturers to carry out risk analyses in order to identify the accident scenarios likely to occur at their facilities. Regardless of the analysis methodology used (bow-tie or other), the primary aim of these analyses is to ensure that the Risk Control Measures (RCMs) put in place enable the plant's safety level to reach the objective set by the operator. This objective may be set by :

• social acceptability (local residents, local authorities, etc.);

• regulatory compliance (hazard studies, Technological Risk Prevention Plans);

• a group's internal security policy.

The probabilistic approach used today in many risk analyses highlights the fact that the uncertainty surrounding the probabilities assigned to hazardous phenomena remains relatively high. This makes it difficult to assess the acceptability of risks by reference to probability "threshold values" alone. In order to take a transparent approach, risk analyses must justify the performance of all MMRs likely to reduce the occurrence and/or severity of the effects of hazardous phenomena that could impact surrounding targets.

Reliability" is one of the criteria used to judge the performance of an MMR; this criterion is often expressed in terms of the probability of failure. Justifying the failure probabilities allocated to MMRs can be more or less complex. Indeed, depending on the type of MMR (safety device, Safety Instrumented System, etc.), different methods are used to assess the probability of failure.

This article begins by explaining how it is possible to define an objective for the probability of failure of an MMR. Secondly, it discusses the methods used to assess the failure probabilities of active and passive safety devices (valves, drip trays, etc.). The main methodologies for assessing the probabilities of MMRs involving human intervention (organizational measures and manual safety action systems) will also be presented. For these different types of MMR, orders of magnitude of failure probability will be given.

The Safety Instrumented System (SIS) is the subject of article [SE 4 058] . This article presents the main concepts relating to SIL (Safety Integrity Level) and calculation formulas for assessing the probability of failure under stress.