Application of Bayesian network and multi-criteria decision analysis to risk-based design of chemical plants

Fires and explosions in chemical plants are still among the major accidents threatening human lives and causing huge asset losses. Although might not completely be eliminated, the risks of such accidents can be reduced by allocating safety measures, applying inherently safer design (ISD) methods, observing land use planning (LUP) regulations, and practicing emergency management. Compared to other risk reduction measures, applications of ISD and, in particular, LUP in chemical plants are new. In most of previous work, however, ISD and LUP have been considered as individual safety elements rather than parts of a coherent safety policy. This, to some extent, arises from contradictory guidelines inferred from the principles of ISD and the requirements of LUP. The present study aims to employ the principles of ISD and LUP, altogether, in risk-based design of chemical plants so that the levels of both on-site and off-site risks can be reduced as low as reasonably practicable. For this purpose, a Bayesian network (BN) methodology is employed to estimate both on-site and off-site risks posed by potential major accidents in chemical plants. The results of the BN modelling are then used as input data in Analytic Hierarchical Process (AHP), a multi-criteria decision analysis technique, to find an optimal layout for chemical plants of interest. While BN facilitates the incorporation of complicated interdependencies and conditional probabilities encountered in accident analysis and risk assessment, AHP allows considering incommensurate and conflicting decision parameters inevitable in most decision analyses. The outcome of the proposed methodology is an optimal layout for the chemical plant under consideration by taking ISD and LUP principles into account.