A Stochastic Network Approach to Modelling Ship Evacuation

This paper presents a development of a novel, efficient, and robust tool to conduct evacuation from maritime ships using multi-commodity network theory and graph search techniques. MSC/1533 (IMO, 2016) presents guidelines to conduct simplified/quasi-advanced analysis methods for the evacuation analysis of ships.  The IMO techniques describe a prescriptive approach and don’t consider path variabilities or uncertainties associated with the dynamic nature of the evacuation environment. To enhance the work presented by the IMO, a linear programming multi-commodity graphing solution has been developed and implemented in Excel.  This model is based on network and graph theory and specifies basic passenger characteristics, simulates ships geometry and specifies path constraints.  The simulation determines the shortest route to the nearest safe point/muster station. Monte Carlo is used to characterise model uncertainties in the simulation. Initial experiments indicate this technique is feasible.  As a Systems Engineering tool, it can be used in trade-off studies to finalise ship layout design and identify path delay/choke points and can be used to rank evacuation performance of existing ship configurations.