Wind farms are stealing each other’s wind. When a cluster of turbines generates a wake—a turbulent, slow-moving stream of air—that wake can travel tens of kilometres downstream and slash the power output of the next farm in line, while also battering its turbines with extra loads. This problem is already visible in the North Sea and the US Great Plains, where new farms are often built upstream of older ones. Yet the physics of how these wakes form, merge, and recover remains poorly understood. Without that knowledge, planners cannot set optimal farm spacing or design layouts that minimise wake deficits. This project will combine large- and small-scale laboratory experiments, high-fidelity numerical simulations, and real-world field data from industrial partners. Four research teams across three countries will model how momentum deficits are produced and how they interact with downstream farms, accounting for ground roughness, the Coriolis force, and thermal stratification. If successful, the work will produce computationally cheap models that run on a laptop—tools that wind farm developers can use to plan new projects without inadvertently crippling existing ones. The result could be more reliable power generation and longer turbine lifespans across the clustered wind farms that increasingly dot the world’s seas and plains.
View original technical description
The rapid expansion of the wind energy sector has led to increasingly large and clustered wind farms, sitting at close proximity to each other. Often new wind farms are built upstream of already existing ones, reducing the power production of the farms that were installed first, together with the turbine lifespan due to increased loads. Examples of that are the clustered farms in the North Sea, and the US great planes. The wind turbines in a farm generate wakes which coalesce, forming a massive turbulent air flow at the outlet of the farm, with significant mean wind speed deficits and enhanced turbulence. This "combined" flow propagates downstream for tens of kilometres, reaching subsequent farms. Therefore, farm-to-farm aerodynamic interactions are expected to become a major challenge in the wind industry, with wind farms "stealing the wind" from each other, lowering the overall power production. To prevent this, proper planning guidelines need to be developed. These should ensure that the wind farm spacing is adequate, so that the reduced kinetic energy of the wind at the wake of a farm has had sufficient time to recover, before reaching subsequent farms, and new wind turbine and wind farm arrangements are developed that minimize the production of very large wind wake deficits. However, at present, the wind farm-to-farm wake production and interaction is poorly understood, and thus there is a critical need for developing new knowledge. To address this issue, the proposed research will investigate the problem of successive wind farm wake interactions and develop reduced-order representations of use to industry and national laboratories. Specifically, we will quantify and develop models for (i) the production and recovery of momentum deficit that a single farm generates at its outflow, and (ii) its interaction with downstream farms. A combination of four research teams in three countries will perform detailed large- and small-scale laboratory experiments and high-fidelity numerical simulations that will be used to consider a multitude of real-world conditions, including changes in ground surface roughness, the Coriolis force, thermal stratification, global blockage and entrainment between the atmospheric boundary layer and the wind farm wake. The results will be complemented by wind farm field data, provided from industrial partners. The datasets will be used to develop and test new wind farm flow models that are able to capture the important flow physics, yet are computationally inexpensive, i.e. designed to run in laptop computers of use to industry when developing preliminary analyses in future wind farm projects.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know