Vertical Axis Wind Turbines (VAWTs)

• Vertical Axis Wind Turbines (VAWTs) are wind energy converters with vertical rotors that operate omni-directionally, making them suitable for urban and low-height deployments.
• Research shows that optimizing blade geometry and co-axial, multistage configurations can yield power coefficient improvements of up to 600% in low tip-speed regimes.
• Advanced computational and control strategies, including CFD, LES, and intracycle angular velocity control, enhance performance by reducing load transients and boosting power output by up to 79%.

Vertical Axis Wind Turbines (VAWTs) are a class of wind energy conversion devices in which the main rotor shaft is oriented perpendicular to the wind direction and to the ground. They differ fundamentally from the more common horizontal-axis wind turbines (HAWTs) in their ability to operate omnidirectionally without yaw mechanisms and their suitability for distributed, low-height, or urban-scale deployment. VAWTs encompass several geometric families—most notably Darrieus, Savonius, and H-type configurations—and display complex fluid-structure interactions including unsteady dynamic stall, asymmetric wake evolution, and multidimensional load transients. Advances in experimental techniques, computational modeling, and optimization strategies have clarified both performance trade-offs and design opportunities for VAWT systems, positioning them as complementary technology to HAWTs in urban, offshore, and power-dense array settings.

1. Core Aerodynamics and Dynamic Stall

The defining aerodynamic behavior of VAWTs is cyclically varying inflow caused by blade rotation through a fixed wind field, producing time-dependent effective angles of attack and velocities. At low tip-speed ratios ($\lambda<3$), the instantaneous angle of attack ($\alpha_\text{eff}$) commonly exceeds the static stall threshold ($\alpha_\text{ss}$), triggering dynamic stall: a sequence of coherent leading-edge vortex (LEV) formation, growth, and shedding. This produces significant transient lift and tangential force overshoots, followed by abrupt loss of lift and drag peaks when the vortex lifts off from the blade surface. Such unsteady events induce large torque fluctuations and pitching-moment excursions, with normalized moment standard deviations ($\sigma_M$) up to 0.90 at deep stall ($\lambda=1.2$). Experimental studies, e.g., with time-resolved strain-gauge measurements and high-speed particle image velocimetry (PIV), enable mapping of operating regimes—deep stall, light stall, and no stall—demonstrating that operation in a light stall regime (intermediate $\lambda \approx 3$) yields a 75% reduction in load transients compared to deep stall, with significant positive torque maintained over the rotation cycle (Fouest et al., 2022).

2. Multistage, Co-axial, and Array Interactions

Bio-inspired multistage co-axial VAWT configurations (e.g., nested rotors with phase-shifted blades) can exploit inter-blade and inter-rotor wake interactions analogous to vortex arrangements observed in fish schooling. Numerical simulations in such dual-stage architectures reveal up to a 600% increase in power coefficient at low $\mathrm{TSR}$ (tip-speed ratio) values (up to $\mathrm{TSR}=2.5$), though performance drops sharply at higher $\mathrm{TSR}$ due to unfavorable vortex interactions and increased drag. Key optimization variables include inner/outer rotor spacing (radius ratios), azimuthal phase angle, and kinematic matching to anticipated site wind regimes. Constructive wake–blade coupling in the low-$\mathrm{TSR}$ regime reduces unsteady separation on the outer rotor and smooths torque production. Optimal deployment requires careful geometric and kinematic parameterization, as excessive proximity or high $\mathrm{TSR}$ reverses the interaction benefit (Khalid et al., 2021).

3. Wake Structure, Modeling, and Array Design

VAWT wakes are fundamentally distinct from HAWT wakes due to dual governing length scales: rotor diameter ($D_0$) and height ($H_0$), yielding a rectangular wake cross-section and anisotropic mixing rates. Advanced analytical models (derived from integral momentum balance and matched against LES and field data) stipulate a top-hat or, preferably, a self-similar Gaussian velocity deficit profile, with the wake width parameter $\varepsilon$ given by $\varepsilon = \sqrt{\beta/(4\pi)}$ for initial mass-flux conservation ($\beta$ is the initial wake-to-frontal area ratio). This model accurately predicts near- and far-wake deficits, momentum thickness, and wake recovery rates, enabling optimization of array spacing for urban or offshore deployments. Critical findings include the feasibility of reducing in-line turbine spacing to 5–8 $D_0$ without undue power losses and potential farm-efficiency gains up to 15% in low-turbulence, tight-packed arrays (Ouro et al., 2020). In field tests of counter-rotating arrays, farm-level power density may reach 18 W/m$^2$—an order of magnitude higher than standard HAWT farms (2–3 W/m$^2$)—through wake harvesting and vertical turbulent kinetic-energy flux from the atmospheric boundary layer (Dabiri, 2010).

4. Modern Computational, Experimental, and Optimization Approaches

High-fidelity CFD using URANS or LES with blade-resolved or actuator-line methods has advanced understanding of both isolated and clustered VAWT performance. URANS models (with $k$–$\omega$ SST turbulence closure) capture integral quantities (power coefficient $C_P$, wake length) at low computational cost, while wall-adapting LES resolve fine-scale vortex structures, turbulence intensities, and weak dynamic-stall vortices critical at off-design $\lambda$ (Sheidani et al., 2022). Proper Orthogonal Decomposition (POD) of LES/RANS fields identifies dominant time and length scales in wake evolution. Advanced surrogate-assisted optimization, e.g., Kriging meta-models coupled to Grey Wolf metaheuristics or neural networks, accelerate multi-parameter design searches (geometry, deflector configuration) and yield measurable gains: up to 34% improvement in power coefficient for Savonius rotors with optimized cylindrical deflectors (Singh et al., 2023), and 30% increased power for hybrid Darrieus–Savonius designs tuned via neural nets (Liu et al., 27 Jan 2025).

3D printing and evolutionary algorithms allow direct physical evaluation of generated supershape geometries, bypassing the "reality gap" of numerical models. Iterative mutation/selection reliably discovers prototypes with higher rotational velocity and, presumably, enhanced aerodynamic efficiency (Preen et al., 2012).

5. Advanced Control Strategies and Unsteady Performance

Contra conventional fixed-speed or torque regulation, modern control strategies leverage real-time modulation of rotor angular velocity as a function of blade azimuthal position ("intracycle angular velocity control"). By aligning rotation-rate maxima with unsteady fluid torque peaks—especially those corresponding to dynamic stall vortex formation—cycle-averaged power output can be increased by up to 79% over constant-speed operation, without introducing pitching mechanisms. This principle has been validated experimentally via servo-controlled rotors with encoder feedback and load-cell measurements (Strom et al., 2016). Bayesian reinforcement learning, model-free algorithms (e.g., Markov Chain Monte Carlo–based policy search), and RBF neural controllers provide robust alternatives to classical maximum power point tracking (MPPT), proving superior for capturing energy during wind transients, accommodating aging or parameter drift, and maintaining high energy-capture efficiency under turbulent, urban wind profiles (Aghaei et al., 2022).

6. Challenges in Structural, Acoustic, and Extreme-Condition Design

VAWTs confront unique structural reliability challenges due to cyclic load transients, dynamic stall events, and blade–wake interactions. Light stall operation ($\lambda \approx 3$) offers a practical compromise—moderate maximum torque, low standard deviation in moment fluctuations, and high reliability. Blade shrouding and diffusive protective covers (e.g., dual-hemisphere patent with variable-section ducts) extend the operational wind-speed envelope and mitigate blade overload in extreme gusts, allowing rated power to be sustained for higher external wind speeds. The effective capping of internal wind speeds to design optima enables operation in hurricane or typhoon contexts with no loss of low-speed efficiency (Baca et al., 2024).

Aerodynamic noise, principally from blade–wake interaction and trailing-edge turbulence, is a critical factor in urban VAWT deployment. Modifying blade geometry (e.g., mid-span V-shaping) and adding trailing-edge serrations results in aerodynamic and acoustic gains, specifically a 28.3% increase in power coefficient and 2–3 dBA broad-band noise reduction compared to straight blades, with minimal rise in thrust loads (Su et al., 2023).

7. System-level Integration and Urban Applications

High-resolution meteorological modeling (e.g., WRF downscaling) combined with validated VAWT power curves demonstrates that urban VAWTs can achieve annual energy yields competitive with photovoltaic panels (e.g., 2,665 kWh/yr per turbine ≈ 16.5 m$^2$ PV equivalent), provided that the cut-in wind speed and power-curve inflection are matched to local microclimate statistics. Deployment in complex terrain and heterogeneous urban environments benefits from two-stage modeling workflows, blending mesoscale resource mapping with CFD micro-siting. The critical factors for maximizing urban output are turbine designs with low cut-in speeds (<1 m/s) and early curve inflection points, enabling generation in the prevalent 2–4 m/s wind regime (Brandi et al., 19 Jan 2025).

• Dynamic stall, regime mapping, and performance: (Fouest et al., 2022)
• Multistage co-axial VAWTs, inter-rotor interaction: (Khalid et al., 2021)
• Analytical wake models, Gaussian deficit, array design: (Ouro et al., 2020)
• Power-density, array layout, counter-rotation, field data: (Dabiri, 2010)
• Savonius optimization, Kriging-GWO methods: (Singh et al., 2023)
• Near-wake 3D structure and implications: (Wei et al., 2019)
• Control via Bayesian RL, urban applications: (Aghaei et al., 2022)
• Evolutionary supershape design: (Preen et al., 2012)
• Intracycle angular velocity control: (Strom et al., 2016)
• CFD, URANS/LES, POD structure: (Sheidani et al., 2022)
• Protective covers, extreme wind operation: (Baca et al., 2024)
• Machine learning optimization, hybrid designs: (Liu et al., 27 Jan 2025)
• Acoustic and aerodynamic performance via V-blades, serrations: (Su et al., 2023)
• Urban resource assessments, meteorology: (Brandi et al., 19 Jan 2025)
• Laboratory-scale design, parametric sweep: (CV et al., 2018)
• Unsteady loads and wake, CFD validation: (Bangga et al., 2017)
• HAWT–VAWT co-location: (Hansen et al., 2020)
• Fish-schooling inspired array geometry: (Whittlesey et al., 2010)
• VAWT clusters, cluster Cp, rotation effects: (G et al., 2023)
• Active flow control, LES/resolvent analysis: (Souza et al., 11 Jul 2025)

VAWT research at present focuses on resolving stall-induced efficiency losses, maximizing land-area power density via innovative array schemes, integrating advanced control and optimization strategies, and adapting turbine geometries for low-noise, robust performance in complex wind environments. The technology's unique multi-regime dynamics and wake topology provide fertile ground for further hybridization with machine learning and bio-inspired architectural paradigms.