Resumen de Novel realistic approaches to Container Premarshalling
Celia Jiménez Piqueras
In recent decades, the rapid growth of containerized freight has posed significant challenges to port efficiency. Optimizing operations has become crucial for minimizing costs for ports and shipping companies and reducing environmental impact.
The organization and management of the port yard are essential to ensure port efficiency, as it is used for container storage and connects all port areas and activities. This dynamic environment demands effective strategies for placing and retrieving containers. Since containers are stacked, retrieving one may require moving containers above it to other stacks, which slows down the retrieval. These movements can be avoided by arranging the containers beforehand through a process known as container premarshalling.
The classical optimization problem associated with premarshalling, which aims to find the minimum number of relocations necessary to arrange the containers, relies on several unrealistic assumptions. This thesis aims to facilitate the practical application of premarshalling by reformulating some of these assumptions.
Containers in the yard are organized into groups called bays, and the premarshalling problem focuses on arranging one bay of containers at a time. The classical formulation only allows relocations within the bay being arranged, but this is not a real limitation in practice. We propose a novel version of the problem where containers can be moved to an adjacent bay to facilitate the arrangement process.
Another unrealistic aspect of the classical formulation is its objective. Previous literature shows that minimizing the total crane time during premarshalling leads to more efficient solutions than minimizing the number of relocations. This thesis goes beyond incorporating crane times and acknowledges limited crane availability. While the original formulation fails to provide a solution when the required time for completing the arrangement exceeds crane availability, we define a novel formulation that yields a complete premarshalling when time allows and a partial arrangement when it does not. Defining good-quality partial arrangements is challenging, and we explore three alternative strategies.
Studying these problem variants required a simple solution method that could be easily adapted to varying assumptions and provide optimal solutions for problem validation. To meet these requirements, we propose a solution method that combines the use of a constraint programming solver with simple algorithms that are easy to implement. Constraint programming has received very little attention in the premarshalling literature despite showing strong performance in solving combinatorial problems in various fields. This thesis reveals the effectiveness of this technique for the premarshalling problem by providing constraint programming models that outperform the state-of-the-art mathematical programming approaches.
Overall, this thesis aims to bridge the gap between the mathematical formulation of the premarshalling problem and real-world application. To achieve this, it addresses both theoretical and technical aspects. On the theoretical side, it introduces several novel variants of the problem that represent significant progress toward a more realistic formulation. On the technical side, it presents constraint programming solution methods that demonstrate the efficiency of this technique in addressing the premarshalling problem.
