In this paper, we formulate a physics-based surrogate wake model in the framework of online wind farm control. A flow sensing module is coupled with a wake model in order to predict the behavior of the wake downstream of a wind turbine based on its loads, wind probe data and operating settings. Information about the incoming flow is recovered using flow sensing techniques and then fed to the wake model, which reconstructs the wake based on this limited set of information. Special focus is laid on limiting the number of input parameters while keeping the computational cost low in order to facilitate the tuning procedure. Once calibrated, comparison with high-fidelity numerical results retrieved from Large Eddy Simulation (LES) of a wind farm confirms the good potential of the approach for online wake prediction within farms. The two approaches are further compared in terms of their wake center and time-averaged speed deficit predictions demonstrating good agreement in the process.
Individual pitch control has shown great capability of alleviating the oscillating loads experienced by wind turbine blades due to wind shear, atmospheric turbulence, yaw misalignement or wake impingement. This work presents a bio-inspired structure for individual pitch control where neural oscillators produce basic rhythmic patterns of the pitch angles, while a deep neural network modulates them according to the environmental conditions. This mimics, respectively, the central patterns generators present in the spinal chord of animals and their cortex. The mimicry further applies to the neural network as it is trained with reinforcement learning, a method inspired by the trial and error way of animal learning. Large eddy simulations of the reference NREL 5MW wind turbine using this biomimetic controller show that the neural network learns how to reduce fatigue loads by producing smooth pitching commands.
The development of new wake models is currently one of the key approaches envisioned to further improve the levelized cost of energy of wind power. While the wind energy literature abounds with operational wake models capable of predicting in fast-time the behavior of a wind turbine wake based on the measurements available (e.g., SCADA), only few account for dynamic wake effects. The present work capitalizes on the success gathered by the Dynamic Wake Meandering formulation and introduces a new operational dynamic wake modeling framework aimed at capturing the wake dynamic signature at a low computational cost while relying only on information gathered at the wind turbine location. In order to do so, the framework brings together flow sensing and Lagrangian flow modeling into a unified framework. The features of the inflow are first inferred from the turbine loads and operating settings: a Kalman filter coupled to a Blade Element Momentum theory solver is used to determine the rotor-normal flow velocity while a Multi-Layer Perceptron trained on high-fidelity numerical data estimates of the transverse wind velocity component. The information recovered is in turn fed to a Lagrangian flow model as a source condition and is propagated in a physics-informed fashion across the domain. The ensuing framework is presented and then deployed within a numerical wind farm where its performances are assessed. The computational affordability of the proposed model is first confirmed: 7 × 10−4 wall-clock seconds per simulation second are required to simulate a small 12 turbines wind farm. Large Eddy Simulations of wind farm using advanced actuator disks are then used as a baseline and a strong focus is laid on the study of the wake meandering features. Comparison against the Large Eddy Simulation baseline reveals that the proposed model achieves good estimates of the flow state in both low and high Turbulence Intensity configurations. The model distinctly provides additional insight into the wake physics when compared to the traditional steady state approach: the wake recovery is consistently accounted for and the wake meandering signature is captured as far as 12D downstream with a correlation score ranging from 0.50 to 0.85.
This work extends the capabilities of an operational dynamic wake model to yawed cases. The proposed framework brings together flow sensing and Lagrangian flow modeling into a unified framework: both the freestream flow field and the wake one are discretized as series of information-carrying particles. A source condition for these particles is thus obtained from the wind turbine measurements through flow sensing techniques. The estimated flow field state across the wind farm is finally reconstructed by propagating these particles downstream at their own characteristic velocity. The resulting framework is first presented and its extension to yawed turbine is then discussed. Comparison against high-fidelity Large Eddy Simulations of yawed wind turbines confirms the good potential of the approach: different yaw angles are considered and the performance of the model are evaluated. This study indicates that the proposed framework captures the relevant large scale wake features caused by the combined effect of yawing and wake meandering at a low computational cost thereby making it suitable for online model-based control.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.