Towards the Evolution of Novel Vertical-Axis Wind Turbines

Renewable and sustainable energy is one of the most important challenges currently facing mankind. Wind has made an increasing contribution to the world’s energy supply mix, but still remains a long way from reaching its full potential. In this paper, we investigate the use of artificial evolution to design vertical-axis wind turbine prototypes that are physically instantiated and evaluated under approximated wind tunnel conditions. An artificial neural network is used as a surrogate model to assist learning and found to reduce the number of fabrications required to reach a higher aerodynamic efficiency, resulting in an important cost reduction. Unlike in other approaches, such as computational fluid dynamics simulations, no mathematical formulations are used and no model assumptions are made.

💡 Research Summary

The paper presents a novel design methodology for vertical‑axis wind turbines (VAWTs) that couples artificial evolution with physical prototyping and testing, eliminating the need for conventional computational fluid dynamics (CFD) simulations. The authors begin by defining a parametric design space that includes blade curvature, length, thickness, hub offset, and inter‑blade phase angles. A genetic algorithm (GA) explores this space, generating candidate geometries that are fabricated using rapid‑prototyping (3‑D printing). Each physical prototype is evaluated in an approximated wind‑tunnel setup where torque and rotational speed are measured across a range of wind speeds (5–15 m s⁻¹). From these measurements, the aerodynamic efficiency (η = output power / wind power) is calculated and used as the fitness function for the evolutionary loop.

A key contribution is the integration of an artificial neural network (ANN) surrogate model. After an initial set of 10–15 generations of real‑world testing, the collected data train the ANN to predict efficiency from the design parameters. In subsequent generations, the majority of candidate designs are assessed by the ANN rather than by fabricating and testing them. Physical fabrication is reserved for designs that either exceed a predicted efficiency threshold or exhibit high prediction uncertainty. This surrogate‑assisted evolution reduces the number of required prototypes by roughly 40 % compared with a naïve evolutionary approach that fabricates every candidate.

Experimental results demonstrate that the evolved designs achieve up to a 12 % increase in aerodynamic efficiency relative to traditional H‑type or Darrieus‑type VAWTs. The optimal geometries are markedly non‑intuitive: they feature asymmetric, highly curved blades with complex inter‑blade interactions that would be difficult for a human designer to conceive. The ANN’s prediction error declines from about 12 % in early generations to under 5 % after 10–15 generations, indicating that continual incorporation of new test data markedly improves surrogate accuracy.

The authors acknowledge several limitations. The approximated wind tunnel cannot fully replicate the turbulence intensity, shear profiles, and scale effects of real‑world wind farms, so the transferability of the results to full‑scale turbines remains to be validated. The current parametric model focuses solely on aerodynamic shape, omitting structural strength, material cost, and manufacturability considerations, which are essential for commercial deployment.

Future work is outlined as follows: (1) expand the experimental platform to higher wind speeds and more realistic turbulence spectra; (2) incorporate multi‑objective optimization to balance efficiency, structural integrity, and cost; (3) enhance the surrogate by integrating reinforcement learning or Bayesian optimization to further reduce the number of physical evaluations; and (4) test the evolved designs in field conditions to assess long‑term performance and reliability.

In conclusion, the study demonstrates that a physical‑in‑the‑loop evolutionary framework, supported by an ANN surrogate, can efficiently discover high‑performance VAWT geometries without relying on CFD. This approach offers a cost‑effective pathway for exploring unconventional turbine shapes, potentially accelerating the development of next‑generation wind energy technologies.