Human Evacuation planning for Passenger Ships-Uncertainty Modelling

Abstract

The burgeoning cruise industry has emphasized the importance of maritime safety and efficient evacuation protocols. The tangible risks and past incidents involving passenger ships spotlight the critical necessity for enhanced protection and effective evacuation strategies. This pressing demand has propelled initiatives by organizations such as the International Maritime Organization (IMO) and the Maritime Safety Committee (MSC) to fortify safety protocols and evacuation plans to safeguard various stakeholders, including passengers, crew members, and emergency response teams.

This research pivots on the critical analysis and development of human evacuation models, in the context of passenger ships. While traditional models have been divided into simplified and advanced analyses, this study endeavors to address the complexities and uncertainties inherent in human evacuation models rendering it more advanced than simple.

A systematic literature review was conducted to assess the existing state of human evacuation modeling for passenger ships and subsequently identify gaps. Based on the identified gaps and critical parameters and objectives in this research area, three distinct human evacuation optimization models (HEM 1, HEM 2, and HEM 3) for passenger ships under varying uncertainties were developed. The first model optimizes total evacuation time considering the uncertainty in passenger walking speed by utilizing robust optimization (RO) by introducing uncertainty sets. The second model, addressing the hybrid uncertainty of passenger walking speed and travel distance, employs a risk-neutral, two-stage, scenario-based stochastic optimization technique (RSSP). The third model optimizes the total evacuation time under mixed uncertainty involving walking speed and door capacity disruptions by applying HRSSRP (hybrid risk-neutral, two-stage, scenario-based stochastic ρ-robust programming). Further, the generated scenarios for passenger walking speed are updated throughout evacuation to follow the real-time circumstances. Moreover, the modeling process incorporated considerations for various starting locations, situational awareness (i.e., alert or non-alert), and the passenger ship’s general arrangement (e.g., number of exit doors and corridor’s width).

Another contribution of this research is the inclusion of families as separate entities in the models to capture the unique dynamics of group evacuations. Also, the models maintain proximity between crew and passengers by allocating optimal crew-to-passenger ratios during the evacuation process. The models were validated using a single deck of a real-life passenger ship.

The developed models serve dually on macroscopic and microscopic levels to facilitate decision-making in overall evacuation organization and devising individualized evacuation plans under uncertainties. This research carves out a pathway for practical applications in various domains, such as ship design, simulation software development, digital twins, and machine learning algorithms, all within the human evacuation process from passenger ships. The developed models were built using real-world data from the IMO and then underwent partial validation via a case study focused on a single deck of a passenger ship. Nonetheless, it is crucial to acknowledge that this validation is not yet complete. Further investigation and real-time testing are necessary to fully confirm their effectiveness and accuracy for broader applications, including multi-deck environments, in future studies.

While this study provides a robust foundation, limitations such as its focus on single-deck evacuation scenarios and reliance on fixed parameters for initial passenger location and awareness levels open avenues for future research; future improvements could integrate multi-deck scenarios with real-time data through sensor technology. Additionally, exploring other uncertainty modeling techniques, such as Bayesian network-based approaches, could offer additional insights by leveraging the expertise of maritime specialists and mitigating data scarcity in this research area.