Surrogate-based global sensitivity analysis of a floating fish cage array with shared mooring

The deployment of large steel-framed cages for salmon farming has yielded successful trial harvests offshore. Following this technologic trend, an innovative shared-mooring solution is proposed in this study to accommodate multiple semi-submersible cages within a single array for cost-effective harvesting. However, directional ocean currents during harvesting seasons can induce excessive horizontal displacements of the fish cages, posing risks to critical infrastructure, such as import power cables or refill pipes. This study introduces an interpretable surrogate-based Sobol global sensitivity analysis (GSA) framework to identify both critical mooring design variables and worst-case current headings. The surrogate GSA framework is built upon a simplified numerical scheme, employing catenary equations to efficiently calculate the mooring forces at 23 mooring lines. Drag forces at both the hull and nettings of each fish cage structure are also accounted for. Using the training data generated from the numerical model, Kriging surrogates are constructed for fish-cage offset prediction. Approximately 95 % accuracy and an 80-fold improvement in efficiency are achieved compared to direct full simulations. The computed Sobol indices identify the current directions that lead to significant displacements across all possible mooring configurations. Further interpretation of the indices reveals that increasing mooring line diameters offers the most effective stiffness gains. These insights inform design recommendations for both array orientation and local reinforcements. The proposed surrogate based Sobol GSA framework thus enables fast and interpretable assessment for design of offshore shared-mooring systems. • A novel concept of shared-mooring fish cage array is presented. • Joint influence of directional environmental loads and mooring configurations on cage displacements is investigated. • Global sensitivity analysis (GSA) using Sobol indices is performed • Surrogate modelling technique is used to improve GSA efficiency. • Impactful load directions are identified, following the requirements in NS-9415 guideline. • Recommendations on layout headings and system upgrading are provided.