Localized Generative Design of Ship Hulls with Diffusion Models and High-Fidelity Assessment
This paper presents a reward-directed diffusion framework for localized ship hull optimization but demonstrates through high-fidelity CFD validation that current surrogate-based screening and reward formulations can miss critical performance trade-offs and fail to capture complex geometric features like faired bulbous bows, highlighting the necessity of integrating detailed physical assessments to refine generative design objectives.