In the pursuit of optimizing wind farm performance, accurate and efficient prediction of wind flow, particularly the wake effects caused by upstream turbines, is paramount. Traditional methods such as physical measurements and numerical simulations, while providing valuable insights, are often time-consuming and resource-intensive. Physics-based models, on the other hand, can sometimes fall short in accuracy due to their simplified representations of complex physical phenomena. Data-driven models, leveraging the abundance of high-quality data, have emerged as a promising alternative.
A recent study introduces a novel, rapid, data-driven approach to modeling wind farm wake flow, drawing inspiration from video-frame interpolation and the principle of similarity. The method transforms field data into images, employing multi-scale feature recognition to identify, match, and interpolate wake structures. By utilizing Scale-Invariant Feature Transform (SIFT) and Dynamic Time Warping (DTW), the model generates intermediate flow fields with remarkable accuracy.
The study validated the approach using six representative mini wind-farm cases, encompassing variations in turbine spacing, turbine size, combined spacing-size variations, different turbine counts, and wind-direction misalignment. Impressively, the method achieved a mean absolute percentage error (MAPE) ranging from 0.68% to 2.28% across these diverse scenarios.
One of the standout advantages of this method is its computational efficiency. By flexibly computing both 2D and 3D wake fields, it offers substantial gains over large-eddy simulation (LES) and Meteodyn WT when 2D accuracy suffices for industrial needs. This efficiency makes it a practical alternative to measurements, high-fidelity simulations, and simplified physics-based models, enabling the efficient expansion of wake-flow databases for wind-farm design and power prediction.
The relevance of this research to the energy sector is significant. Accurate and efficient wake-flow modeling can lead to optimized wind-farm layouts, improved power forecasting, and ultimately, enhanced overall performance and profitability. By balancing speed and accuracy, this data-driven approach holds promise for advancing the field of wind energy and contributing to the broader transition towards sustainable energy sources.
This research was published on arXiv and can be read in full [here](http://arxiv.org/abs/2509.21005v1).