Dataset for the AMR-Wind digital-twin simulations of the three-turbine experimental facility

This dataset contains the numerical setup and sampled outputs of a digital twin of a three-turbine wind-tunnel facility implemented in AMR-Wind v3.8.0, an open-source, performance-portable flow solver based on AMReX. AMR-Wind is designed for high-fidelity wind farm and atmospheric boundary-layer simulations, including LES, actuator-based turbine models, and adaptive mesh refinement.

The simulations solve the incompressible Navier–Stokes equations with Boussinesq buoyancy, Coriolis forcing, and actuator forcing. The computational domain spansx ∈ [-10.0, 14.2] m, y ∈ [-6.92, 6.92] m, and z ∈ [0.0, 3.84] m.A base mesh of 304 × 176 × 48 cells is used, with up to three AMR levels, yielding a local mesh spacing of approximately 0.01 m at the turbine rotor disk.

Boundary conditions are prescribed as follows: no-slip walls on y_lo, y_hi, z_lo, z_hi, mass inflow at x_lo, and pressure outflow at x_hi. Realistic inflow is imposed through time-resolved velocity fields ingested from a precursor large-eddy simulation (see https://zenodo.org/records/18860646 and https://zenodo.org/records/19004026) using the AMR-Wind boundary-plane inflow capability. A minimal source-code modification was introduced to allow scaling of the ingested velocity fields by a constant factor in order to match the experimental wind conditions.

The flow is advanced with a fixed time step of Δt = 2.5 × 10^-4 s, target CFL = 0.95, and total simulated time of 30 s. Subgrid-scale closure is provided by the AMD model, and the atmospheric boundary-layer setup uses a surface roughness length of z0 = 10^-3 m.

The three-turbine cluster is modeled with the Actuator Line Model (ALM) corrected by the vortex-based tip/smearing correction of Meyer Forsting (A vortex-based tip/smearing correction for the actuator line, Meyer Forsting et al. , 2019, https://doi.org/10.5194/wes-4-369-2019) coupled to OpenFAST, in order to represent the aero-servo-elastic response of the G1 turbines. Each rotor is discretized with 60 actuator points per blade and 12 points along the tower. The Gaussian smearing width is ε = 0.013 m.

This repository contains the numerical setup, auxiliary scripts, and representative outputs of AMR-Wind simulations reproducing selected cases of the experimental dataset Multiple wakes measurement with lidar (https://zenodo.org/records/18731994)

To support validation against experiments, planar sampling is performed during runtime on:

one x–y plane located slightly above hub height,

multiple y–z planes spanning the test-section height.

Velocity fields are sampled every 200 time steps.

This repository is intended to support reproducibility and reuse of the AMR-Wind numerical setup associated with the corresponding publication. It includes the simulation input files, documentation of the computational setup, information on the software environment, and sampled outputs required to reproduce the reported analyses.