Multifidelity data fusion for the estimation of static stiffness of suction caisson foundations in layered soil

The static stiffness of suction caisson foundations is an important engineering factor for offshore wind foundation design. However, existing simplified design models are mainly developed for nonlayered soil conditions, and their accuracy for layered soil conditions is uncertain. This creates a challenge for designing these foundations in offshore wind farm sites, where layered soil conditions are commonplace. To address this, this paper proposes a multifidelity data fusion approach that combines information from different physics-based models of varying accuracy, data sparsity, and computational costs in order to improve the accuracy of stiffness estimations for layered soil conditions. The results indicate that the proposed approach is more accurate than both the simplified design model and a single-fidelity machine learning model, even with limited training data. The proposed method offers a promising data-efficient solution for fast and robust stiffness estimations, which could lead to more cost-effective offshore foundation designs.