Parametric ROM using VMS for Wind Turbine Applications
Overview
This ongoing project addresses the limitations of standard global POD bases by developing a parametric ROM framework capable of handling variable operating conditions. By employing a control function (lifting) strategy to homogenize non-homogeneous Dirichlet boundary conditions, the method creates a parameter-independent basis valid across a range of inflow speeds. The approach is validated on canonical flows and extended to 3D ALM-VMS wind turbine wakes.
Results
Preliminary validation on a 2D laminar cylinder flow effectively captures shedding frequencies and aerodynamic forces across untrained Reynolds numbers ranging from 100 to 200.
For 3D wind turbine wakes, the parametric ROM is trained on inflow velocities of 8 m/s and 10 m/s, and successfully extrapolates to an unseen intermediate velocity of 9 m/s.
The surrogate model accurately predicts power output and near-wake statistics at the intermediate velocity.
Limitations & Future Work: The back-to-back turbine scenario indicates that current global parametric POD modes underestimate far-wake complexity due to the dominance of low-frequency dynamics. Ongoing work focuses on enriched sampling strategies, localized basis enrichment, and offline/online decomposition optimizations to provide robust surrogates for real-time wind farm control and forecasting.