Wind turbines leave a swirling “shadow” in the air that engineers have never fully seen until now
Every spinning wind turbine leaves a wake — a churning column of disturbed air that can stretch for miles and cut into the energy output of turbines downstream. Engineers have spent decades measuring that turbulence, but the tools they rely on flatten its true structure into a single dimension, erasing the directional patterns that shift and reorganize depending on the scale at which you look.
A new study suggests that picture may be missing something fundamental — and that what’s missing could matter for how wind farms are modeled, designed, and optimized.
The hidden geometry of wind-turbine wakes
When a wind turbine spins, it doesn’t just extract energy from the air — it fundamentally reorganizes it. The wake that forms downstream is a region of disturbed, turbulent flow that can stretch for several rotor diameters and reduce the power output of turbines further along a farm. That much engineers have known for years.
Reference: Sadeghi, S., Ghane, M., and Ali, N.: Scale-Dependent Anisotropy in Two-Dimensional Spectra of Wind-Turbine Wake Boundary-Layer Flows, Wind Energ.
What’s less understood is the internal geometry of that turbulence. Energy inside a wake isn’t distributed equally in all directions. This property — called anisotropy — changes depending on the scale at which you examine the flow.
Large swirling structures behave differently from small ones, and the degree of directionality shifts accordingly. Standard spectral models used in engineering don’t capture any of this, describing turbulence along a single dimension and collapsing the full directional picture into a curve that’s mathematically convenient but physically incomplete.
Why one dimension is not enough
One-dimensional spectra have dominated turbulence analysis for practical reasons. They’re relatively easy to measure in the field, straightforward to compute, and well-supported by decades of theoretical work. For many applications, they’re good enough.
There’s a cost to that convenience, though. A one-dimensional spectrum integrates across the entire wavenumber space, averaging over all directions in the process. The result tells you how much energy exists at a given scale — but not where that energy is pointing. In a wind-turbine wake, that distinction is consequential: large, energy-containing eddies tend to be strongly elongated in one direction, while smaller eddies are closer to isotropic.
A one-dimensional model treats both identically, making it impossible to distinguish between these two very different physical regimes.
A new framework built around two dimensions
To address this gap, researchers developed a two-dimensional spectral model that maps the full planar wavenumber domain — preserving directional information across all scales rather than collapsing it.
The approach centers on a bidirectional remapping of wavenumber space. Two directional stretch factors, applied jointly to streamwise and spanwise wavenumbers, allow the model to represent the strong anisotropy seen at large, energy-containing scales. As scale decreases and turbulence becomes more isotropic, those stretch factors relax.
The shape of that relaxation is encoded using constrained Bernstein polynomial surfaces, a mathematical tool ensuring that directional stretching varies smoothly across the spectrum. A Jacobian correction is applied throughout, keeping spectral energy physically consistent as the wavenumber space is remapped.
Tested against real and simulated turbulence
The framework was put to work on large-eddy simulation datasets — high-resolution computational fluid dynamics runs that model turbulence at fine scales. Three scenarios were covered: wind-turbine wakes under high-turbulence inflow, wakes under low-turbulence inflow, and a reference atmospheric boundary-layer case with no turbine present.
Fitting the model required estimating parameters for the streamwise and wall-normal velocity components. The researchers used a staged, coarse-to-fine optimization strategy, progressively refining the fit while enforcing constraints on component energy, Jacobian admissibility, and surface smoothness.
The fitted spectra were then used to generate synthetic turbulent flow fields — artificial turbulence that statistically mirrors the simulated wakes — and those fields were validated against spatial correlations, integral scales, structure functions, and one-point statistics. Across all tests, the agreement was strong.
What this means for wind farm science
The practical value of this framework extends well beyond academic simulation. By preserving directional, scale-dependent structure, the model gives engineers a more complete picture of how a turbine wake — and the ambient turbulence intensity of the incoming flow — reshapes the surrounding air.
That information could feed directly into load predictions for downstream turbines, which face fatigue from complex, directionally structured turbulent forces that simpler models may underestimate. Smarter farm layouts and sharper wind resource assessments are also within reach. Critically, the approach is computationally feasible — it doesn’t require the kind of resources that confine high-fidelity methods to research environments, making it a realistic candidate for actual engineering workflows rather than just journal pages.
As wind farms grow larger and turbines are packed more densely, wake interactions will only become more consequential. Frameworks that capture what standard models miss may prove essential to getting those designs right.
Reference: Sadeghi, S., Ghane, M., and Ali, N.: Scale-Dependent Anisotropy in Two-Dimensional Spectra of Wind-Turbine Wake Boundary-Layer Flows, Wind Energ. Sci. Discuss. [preprint], https://doi.org/10.5194/wes-2026-152, in review, 2026.
Daniel García is an Editor-in-Chief with strong expertise in structural work and engineering principles. He combines this technical foundation with deep knowledge of energy, spatial design, and emerging technologies, bringing a forward-thinking and analytical approach to editorial leadership.
