Design-Space Dimensionality Reduction Benchmark Dataset – Bio-Inspired Underwater Glider

This dataset provides a benchmark database for the evaluation and comparison of design-space dimensionality reduction methods in shape optimization problems. The dataset consists of 16,385 configurations of a bio-inspired autonomous underwater glider geometry generated through Sobol sampling of a CAD-based parametrization with 32 design variables. The geometry is defined through four airfoil sections parameterized using NACA 4-digit profiles. Each configuration includes: • discretized geometry coordinates (3 × 784 points)• design variables defining the CAD parametrization• pressure coefficient distribution (Cp) evaluated on 784 panels• integrated hydrodynamic quantities (Drag and Lift) Hydrodynamic quantities were computed using the PUFFIn potential-flow solver developed at ENSTA, with viscous correction based on a flat-plate approximation. Simulation conditions: freestream velocity: 0.25 m/s angle of attack: 8 degrees fluid density: 1030 kg/m³ The dataset includes: database.mat (3170 × 16385 matrix) range_design.mat (design variable bounds) metadata.txt (matrix structure) README.md (dataset documentation) Dataset statistics: total configurations: 16385 valid simulations: 7467 solver failures: 7967 parametrization failures: 951 The dataset was generated within the BIODRONES project and used in: Serani, A., Palma, G., Wackers, J., Diez, M. "A Machine Learning Enabled MDO for Bio-Inspired Autonomous Underwater Gliders." arXiv:2602.08508 (2026). Serani, A., Palma, G., Wackers, J., Quagliarella, D., Gaggero, S., Diez., M. "Extending parametric model embedding with physical information for design-space dimensionality reduction in shape optimization." Engineering with Computers (2025): 1-21.