Optimization of real-world manual container loading with practical constraints: A metaheuristic approach

• Defined a realistic container loading problem with manual loading constraints. • Proposed a genetic algorithm with domain-specific solution encoding. • Developed a realistic evaluation function simulating human loading process limits. • Designed repair and improvement methods to enhance solution feasibility and profit. • Validated the algorithm on realistic tests; provided open-source Python implementation. In the context of international trade, the transportation of goods is contingent upon the utilization of containers, a practice that underscores the imperative of optimizing efficiency. This imperative stems from the need to curtail expenditures and mitigate carbon emissions. This paper addresses the combinatorial optimization problem of container loading with packages of varying sizes and shapes, a well-known NP-hard challenge, employing a metaheuristic approach based on genetic algorithms. The study's primary focus is on manual loading by operators, incorporating realistic constraints such as package rotation, contiguity, and others related to space accessibility for a person. The algorithm has been meticulously designed, incorporating solution improvement mechanisms that have led to a substantial enhancement in its performance, enabling it to obtain high-quality solutions within reduced computation times. The effectiveness of the proposed method is validated through a series of computational experiments, demonstrating its capacity to address complex scenarios and providing a robust framework for addressing the logistical challenges associated with container loading.