A Bayesian approach to construct bow tie diagrams for risk evaluation

Bow tie diagrams have become a popular method for risk analysis and safety management. This tool describes the whole scenario of a given risk graphically, and proposes preventive and protective barriers to reduce, respectively, its occurrence and its severity. The weakness of bow tie diagrams is that they are restricted to a graphical representation of different scenarios exclusively designed by experts that ignore the dynamic aspect of real systems. Thus, constructing bow tie diagrams in an automatic and dynamic way remains a real challenge. This paper proposes a new Bayesian approach to construct bow tie diagrams from real data and improve them by adding a new numerical that enables us to implement the appropriate preventive and protective barriers in a dynamic manner.

► In this paper we propose a new approach to construct a whole bow tie diagrams for risk analysis reflecting the real behavior of the existing systems.
► We propose an algorithm to construct the Skelton of bow tie diagrams from training set.
► A parameters learning phase relative to their numerical component is proposed.
► A dynamic implementation of protective and preventive barriers on the bases of the different criteria frequently used is done.
► An illustrative example in the petroleum field involving the construction of a bow tie diagram is proposed.
► In this study we conclude that the proposed algorithm can be considered as a decision support system to construct different bow tie diagrams reflecting the real behavior of existing system.