Comparative Probabilistic Risk Assessment of Alternative Marine Fuels in Traffic Separation Schemes Using Monte Carlo Simulation

The transition towards low-carbon shipping has accelerated the adoption of alternative marine fuels such as ammonia, hydrogen, methanol and liquefied natural gas (LNG). Although these fuels reduce greenhouse gas emissions, they introduce new operational hazards related to toxicity, flammability, cryogenic storage and explosion risk. Existing comparative assessments are predominantly deterministic, representing accident probability and consequence severity by single values while neglecting operational uncertainty. This study proposes a probabilistic consequence-weighted risk assessment framework for alternative marine fuels operating within high-density Traffic Separation Schemes (TSSs). The methodology integrates historical traffic data, a location-specific contextual vulnerability multiplier and fuel-specific accident characteristics, propagating uncertainty through Monte Carlo simulation using Beta distributions for conditional loss-of-containment probability and triangular distributions for consequence severity. One million simulations per fuel-location combination were adopted after numerical convergence checks. The framework was applied to three Spanish TSSs (Tarifa, Cabo de Gata and Finisterre) and four candidate fuels (ammonia, hydrogen, methanol and LNG). Unlike deterministic approaches, the methodology provides complete risk distributions, P5–P95 uncertainty intervals and probabilistic fuel rankings. Results show that geographical location materially modifies operational risk and must be considered jointly with fuel-specific hazard; within the adopted scenario matrix, Tarifa remains the most critical location. Ammonia consistently exhibits the highest consequence-weighted risk. The framework is intended as a transparent comparative screening and decision-support tool rather than as a substitute for an absolute quantitative risk assessment based on physical consequence endpoints and observed accident frequencies.