AI-, DIGITAL-TWIN-, AND QUANTUM-ENABLED ROLLING-HORIZON OPTIMIZATION FOR YARD ALLOCATION AND GATE APPOINTMENT COMPLIANCE IN MODULAR CONTAINER TERMINALS

Abstract

Container terminals operate under increasingly tight capacity, energy, and service-level constraints, while facing growing uncertainty in arrival patterns, handling operations, and hinterland interactions. This paper proposes a hybrid decision-support framework that integrates advanced Artificial Intelligence techniques, Digital Twin modeling, and quantum-inspired optimization to address yard planning and operational optimization in container terminals. The core problem is formulated as a block-to-area assignment under soft and hard capacity constraints, priority rules, and rolling-horizon forecasts. The problem is mapped to a Quadratic Unconstrained Binary Optimization (QUBO) formulation and solved using both classical and quantum-inspired approaches, including the Quantum Approximate Optimization Algorithm (QAOA). Two alternative modeling strategies are investigated: a one-hot encoding with deterministic feasibility repair, and a feasibility-by-design binary encoding that reduces the number of decision variables and enforces assignment constraints structurally. A Digital Twin of the container yard acts as a truth model and constraint oracle, translating abstract optimization decisions into operational Key Performance Indicators such as reshuffling effort, crane interference, and truck waiting times. Computational experiments on a modular yard scenario highlight the trade-offs between encoding choice, feasibility, solution quality, and computational cost under time-bounded quantum-inspired optimization. The results demonstrate that quantum-inspired methods can effectively support combinatorial decision-making in port operations when combined with appropriate encodings and simulation-based validation. Rather than replacing existing planning tools, the proposed framework positions quantum optimization as a complementary accelerator within a scalable, explainable, and industry-ready Digital Twin-based decision-support architecture.