The propagation of fire-induced domino effects in chemical plants largely depends on the primary fire scenario, on separation distances between the units, and on the presence of fire protection barriers. Passive and active safety barriers are widely employed to prevent or delay the initiation or propagation of domino effects. In the present study, a methodology has been developed based on Bayesian network to account for the impact of such safety barriers on the propagation of fire domino scenarios. The Bayesian network has been extended to a limited memory influence diagram in order to identify a cost-effective allocation of additional safety barriers to further mitigate the fire propagation. The application of the methodology has been demonstrated using a chemical tank farm. The results are in good agreement with the results of a graph theoretic approach developed in a previous study, proving the reliability of the developed methodology in cost-effective protection of process plants.@en