faNN: a Multi-Geometry RANS Dataset of turbomachinery Fan Rotors

We present faNN, a dataset of Reynolds-Averaged Navier–Stokes (RANS) simulations of turbomachinery fan rotor stages designed to support the development and benchmarking of deep learning surrogate models. The dataset contains three-dimensional RANS computations distributed across fan rotor geometries, derived from six primary rotors via parametric blending with the Parablade library. Computations were performed using the FINE/Turbo v18 solver with the k–ω SST turbulence model on structured meshes of approximately 10.6 million points. Available outputs include global performance metrics (isentropic efficiency ηis and total pressure ratio Π) and 3D volumetric fields of velocity, pressure, and turbulence quantities. The faNN dataset is also publicly available on HuggingFace and addresses the scarcity of realistic, industrial-scale fluid mechanics/turbomachinery CFD data for deep learning surrogate development and benchmarking.