Bow tie: a risk quantification method

In many sectors, such as aeronautics, chemicals, oil and nuclear power, it is necessary to assess risks in order to decide on their acceptability. The concepts of hazard and risk are often confused, as risk is always linked to the existence of a danger or hazardous situation. To differentiate between them, we can consider a hazard to be "real" and a risk to be "potential".

The activities of some industrial facilities present a number of hazards. One example is the storage or synthesis of flammable and/or toxic products. In such facilities, one of the most feared events is loss of containment, which can lead to dangerous phenomena such as fire, flaming jet or explosion in the case of a flammable product, and atmospheric dispersion in the case of a toxic product.

Risk assessment involves evaluating the two components of the probability/severity pair. The severity of hazardous phenomena is usually estimated by modeling the intensity of effects, using tools or software. Today, estimating the probability of occurrence of industrial hazards requires recourse to methodologies that have been used for many years in other fields, such as nuclear power or aeronautics. Indeed, in France, prior to the 2000s and in contrast to Anglo-Saxon countries, risk assessment was based on a deterministic approach. The vast majority of accident probabilities were estimated by expert opinion. The tragic accident in Toulouse on September 21, 2001 triggered a profound overhaul of French regulations, which now advocate a probabilistic approach.

The aim of this article is to present the "bow-tie" risk analysis method, based on the combination of a fault tree and an event tree, centered on a single feared event. After outlining the foundations of this method, an overview will be given of the various databases that can be used in the quantification phase. Finally, the limitations of the method will be discussed.