Optimising bulk carriers' fuel efficiency: PCA and Shapley value analysis under varying wind conditions

Ensuring the operational reliability of ships under variable environmental conditions is vital for regulatory compliance, emissions control, and cost-effective performance. This study presents a probabilistic framework combining Principal Component Analysis (PCA) and Shapley Value interpretation to assess main engine fuel consumption in bulk carriers affected by wind variability. Using operational data from seven vessels, key parameters were identified, including ship speed, wind force, engine revolutions per minute (RPM), and engine slip. Fuel consumption per ton was evaluated under rare conditions with no current or swell, and Shapley Values enabled a comparative assessment of fuel performance along a specific voyage route from Rotterdam to New Orleans. The analysis revealed considerable variation in fuel consumption, ranging from 0.2470 to 0.4865 t/GRT across the fleet. Eco-design vessels exhibited the lowest but distinct Shapley Values (0.0189 and 0.0170), suggesting reduced sensitivity to wind-related fuel consumption. Similar trends were observed in smaller ships, even among vessels with identical designs. This probabilistic analysis effectively identifies vessels more sensitive to operational and environmental conditions, supporting reliability-oriented decision-making and voyage optimisation. The approach also enhances dynamic performance monitoring and offers a methodology for improving energy resilience and environmental compliance in modern maritime operations.