Risk-informed multi-objective decision-making of emergency schemes optimization

When an accident occurs, there is a surging demand for methods to generate efficient and effective emergency schemes through optimizing the resource allocation of emergency activities. The trade-off of multiple objectives (e.g., risk, time, and cost) in emergency scenarios can be beneficial to improve the effectiveness of emergency schemes. In this paper, we propose a hybrid methodology integrating dynamic Bayesian network (DBN) and graphical evaluation and review technique (GERT) for evaluating and optimizing emergency schemes. In the proposed methodology, DBN is applied to parameterize the dynamic risk of the emergency response process. Based on the logical relationships between activities, a mapping mechanism from DBN to GERT is established to construct risk-influencing scenarios. Subsequently, a risk-informed multi-objective optimization model is constructed by the fast elitist non-dominated sorting genetic algorithm (NSGA-Ⅱ). Eventually, we discuss the impact of resource investment on the evaluation indicators. The installation of a capping stack, a recognized emergency technique for deepwater blowout accidents, is used to demonstrate the applicability of the methodology. The results show that the proposed model can determine effective emergency schemes through the trade-off of multiple objectives during accidents.