The adoption of Machine Learning (ML) and Artificial Intelligence (AI) in engineering design has significantly advanced numerous industries. However, the maritime sector still faces challenges in leveraging these technologies, due to the limited availability of structured and high-quality datasets. This paper presents a methodology for building a flexible dataset of ship hull geometries using the open-source Python library PyGeM. By applying Free Form Deformation (FFD) techniques, we generate several hull variants suitable for training data-driven models. The approach is generalizable to a broad range of hull forms, we demonstrate it through two representative case studies: the KCS and ITTCA1 container ships. The resulting dataset aims to support ML applications for the prediction of hydrodynamic performance and design optimization. This accelerate innovation in the ship design process, helping to advance sustainable ship design and supporting compliance with IMO’s revised greenhouse gas reduction targets.
Giovannini et al. (Thu,) studied this question.