Distributed Agent-Based Scheduling of Smart Port Operations under Cost and Emissions Constraints

Abstract This paper presents a distributed, emission-aware scheduling framework for smart port energy management, designed to coordinate heterogeneous flexible electric loads under operational, economic, and environmental constraints. The proposed approach addresses the coordinated scheduling of refrigerated containers (reefers), plug-in electric vehicles (PEVs), and shore-to-ship power supply systems, with the objective of reducing total operational cost while satisfying asset-level constraints and ship-level and port-level emission limits. A key feature of the framework is the modeling of berthed ships as controllable prosumers, capable of operating in parallel with shore power systems and participating in cost-aware energy exchange with the port electrical network. To address the distributed and heterogeneous nature of the problem, a Multi-Agent System (MAS) architecture implemented in the Java Agent DEvelopment Framework (JADE) is employed. In the proposed architecture, each asset-level agent formulates and solves its own local optimization problem using its private technical parameters and information received through MAS communication, such as electricity prices and emission constraints. Manager agents aggregate the resulting local schedules and perform coordination tasks, including the verification and enforcement of port-level emission constraints when required. IBM CPLEX Optimization Studio is used as the local optimization engine within the relevant agents, rather than as a single centralized optimizer for the entire port scheduling problem. The proposed framework is evaluated through a one-day simulation scenario involving 500 reefers, 300 PEVs, and two shore-to-ship power supply stations. Three MAS-controlled cases are examined, corresponding to different levels of emission-constraint enforcement. The simulation results show that the proposed framework reduces ship-generator fuel consumption — and hence all six pollutant emissions — by 48–52%, while total port operating cost falls by 28.3%, 28.0% and 25.0% for the three cases compared with the baseline scenario, while operating within the imposed operational constraints and at the imposed emission limits, subject to the small implementation discrepancy quantified in Section “Comparison with a Centralized Formulation”. These results demonstrate the practical applicability of distributed agent-based coordination for cost-efficient and emission-conscious smart port energy management.