Systemic study of the analysis of risks - Presentation of a global approach

Major industrial accidents such as those at Seveso, Bhopal, Challenger, Chernobyl or AZF often have various root causes of different origins, which can be identified by post-hoc risk analyses. The aim of these analyses is to understand the chain of events that led to the accident. They enable us to capitalize on experience feedback to set new prevention and/or protection policies for specific sources of danger, so as to prevent such an accident from happening again.

These accidents are the outcome of scenarios that are much more difficult to perceive a priori. This is all the more true now that our society is subject to increasingly diverse risks, the occurrence of which may prove improbable, as in the case of the September 11, 2001 attacks in New York, or the explosion at the AZF plant on September 21, 2001. The past can therefore no longer simply serve as a knowledge base for risk analysis. We need to adopt approaches that enable us to anticipate phenomena that have never yet occurred.

In addition to a posteriori risk analyses, decision-makers have at their disposal a priori risk analysis methods. By studying systems, they can identify the main accident scenarios that are likely to occur. The results of these studies enable risks to be ranked in order of importance, and make it easier to implement the protection and/or prevention measures needed to control risks.

Today, however, experts are questioning the usual a priori risk analysis methods. In view of the conclusions of a posteriori risk analyses, they are no longer able to carry out in-depth, sufficiently detailed or complete analyses of the complex industrial systems for which they are responsible. Despite a range of methods and tools available in the field of safety, we still deplore numerous industrial accidents, as Barpi's Aria database testifies.

Beyond a purely theoretical vision of the intrinsic ability of complex systems to overcome the barriers set up to stop them, it is clear that the "preventionist" cannot resign himself to the fact that accidents happen. This is all the more true as society seems increasingly "averse" to the risks involved, and this is reflected in the high expectations placed on corporate social responsibility. In this issue, we seek to understand why, with the current tools at their disposal, risk managers are finding it difficult to implement efficient risk management policies that deliver relevant results. These shortcomings give rise to dissatisfaction, with stakeholders both inside and outside the company often criticizing the vision of in-house experts and questioning the validity of the results obtained. But, beyond this external questioning, it is regrettable that the methods implemented reinforce the compartmentalization...