Optimal firefighting to prevent domino effects: Methodologies based on dynamic influence diagram and mathematical programming

Fire is one of the most costly accidents in process plants due to the inflicted damage and the required firefighting resources. If the firefighting resources are sufficient, firefighting will include the suppression and cooling of all the burning units and exposed units, respectively. However, when the resources are inadequate, optimal firefighting strategies to answer "which burning units to suppress first and which exposed units to cool first?" would be essential to delay the fire spread until more resources become available. The present study demonstrates the application of two decision support techniques to optimal firefighting under uncertainty and limited resources: (i) Dynamic influence diagram, as an extension of dynamic Bayesian network, and (ii) mathematical programming. Both techniques are illustrated to be effective in identifying optimal firefighting strategies. However, unlike the dynamic influence diagram, the mathematical programming is demonstrated not to suffer from an exponential growth of decision alternatives, making it a more efficient technique in the case of large process plants and complicated fire spread scenarios.