Rui Miguel Rodrigues

Abstract
Onshore power supply (OPS) is a key enabler for decarbonizing port operations and meeting upcoming regulatory targets such as the EU AFIR Regulation 2023/1805 and Portugal's PNEC 2030. This paper presents a simulation-based framework for estimating the OPS demand of container ships at berth, integrating ship hoteling loads, reefer thermal dynamics with flexible control, and OPS/auxiliary engine (AE) dispatch under port grid constraints. A case study at Terminal XXI of the Port of Sines demonstrates the approach using high-resolution (1-minute) simulations. Results show that reefer flexibility enables peak shaving, OPS demand can be enforced within available grid capacity without violating thermal limits, and AE provides reliable backup. Complementary machine learning modules based on Gradient Boosting, Random Forest, and XGBoost enable accurate imputation of missing ship descriptors and OPS demand forecasting (R2 > 0.95). The framework provides an AI-ready decision-support tool for OPS infrastructure planning and port energy management. © 2025 IEEE.