Land-use planning (LUP) has widely been employed as a protective safety measure in risk management of major hazard installations such as chemical plants. In the European Union countries, a majority of relevant work over the past years has been inspired by the Seveso II Directive. The inclusion of LUP in the Seveso II Directive has been with the aim of mitigating off-site damage of major accidents on public via setting criteria for (i) the identification of the location and layout of new installations, (ii) the development of existing installations, and (iii) the land developments in the vicinity of existing installations. We, in the present study, have proposed a methodology based on Bayesian network (BN) for costeffective allocation of safety measures in chemical plants so that both internal and external risks could effectively be mitigated, particularly in compliance with the requirements of LUP. We first employed BN to calculate risks, and then extended the BN to a limited memory influence diagram using additional decision and utility nodes so that it can be used for multi-attribute decision analysis. The development and application of the methodology have been illustrated via fireproofing of a hypothetical fuel storage plant.
- Limited memory influence diagram
- Bayesian network
- Multi-attribute decision analysis
- Land-use planning
- Domino effect