Hybrid STPA-HMM-C Framework for Dynamic Risk Quantification in LNGFV Port Entry Scenario

Abstract Liquified natural gas (LNG)-fueled vessels (LNGFV) are seen as a pivotal solution for the maritime sector’s energy transition. However, the inherent risks associated with LNG and the increasing complexity of LNGFV systems create significant safety challenges. This study proposes a novel hybrid framework that integrates system-theoretic process analysis (STPA) and hidden Markov model with the cloud model (HMM-C) for dynamic risk assessment in LNGFV navigation. Initially, the 3L risk control structure is employed, combining strategic, tactical, operational, and physical layers to systematically identify and manage potential unsafe interactions and risk influencing factors (RIFs) under three control modes: leakage prevention (LP); leakage stoppage (LS); and leakage remediation (LR). Then, the HMM-C model is introduced to address uncertainties and qualitative data, enabling real-time, dynamic risk quantification. Using the LNGFV port entry scenario as a case study, real-time sensor data are used to simulate and assess risk fluctuations across different control modes. The simulation results reveal that, while risks are ranked as LP > LS > LR mode, they all remain at relatively low levels, confirming the effectiveness of the hybrid framework in minimizing risks. Key RIFs such as wind and wave conditions, traffic density, and emergency shutdown device (ESD) status are identified as critical factors influencing LNGFV navigation safety. This integrated methodology provides a robust, data-driven approach to proactive risk management, offering valuable insights for LNGFV operations and safety optimization.