A new artificial intelligence system has designed a wind turbine for the first time in history, according to its developer. The British company EvoPhase claims that its AI has thrown all the established conventions for engineering such devices out the window and that, based on their tests, its invention is seven times more efficient than existing designs. The Birmingham Blade—as the company calls it—consists of six arms parallel to the ground connected to a central vertical axis. Each arm has a vertical blade, a surface with two waves that change their angle of attack across its height and length.
Traditional wind turbines are optimized for wind speeds of about 33 feet per second. In contrast, the Birmingham Blade is designed for the lower average wind speeds typical of urban areas like Birmingham, which are around 12 feet per second. That’s just a bit over eight miles per hour. The design is optimized to work between tall buildings, which produce turbulence patterns that affect the effectiveness of other urban turbine designs. If this proves true, the design could open the door to unlimited electricity production in office and residential buildings at virtually no cost.
“Using AI was essential for breaking free from the long-standing biases that have influenced turbine designs for the past century. AI allowed us to explore design possibilities beyond the scope of traditional human experimentation,” says Leonard Nicusan, EvoPhase’s Chief Technology Officer, in a press release. According to Nicusan, they were able to “generate, test, and refine over 2,000 wind turbine designs in just a few weeks, significantly accelerating our development process and achieving what would have taken years and millions of pounds through conventional methods.”
The magic of evolutionary design
AI-guided evolutionary design is a methodology based on the same idea as natural selection. The process begins by generating thousands of design variants that are evaluated according to a “fitness to survive” function, which determines how well each variant meets the project goals. The AI selects the best alternatives to use as a basis for new iterations, recombining and varying features to develop improved versions. These steps are repeated until a AI reaches a solution that achieves the optimization of all the parameters, such as aerodynamic efficiency, structural stability, weight, or compactness.
The company says its process avoids the human biases present in traditional engineering. By nature, traditional engineering is limited by preconceived ideas and prior knowledge. AI, on the other hand, explores a wide range of possibilities without the limitations in the human mind. When you combine the AI generation with evolutionary iteration, this can yield innovative results that often defy common sense but still work.
This evolutionary design approach is not totally new. The aerospace industry has used software with this capability for years. Airbus, in collaboration with Autodesk, used a similar process to design an ultralightweight cabin partition for its A320 aircraft. The result appeared inspired by natural bone structures, achieving a 45% reduction in weight compared to similar structures designed with conventional methods.
General Electric also applied evolutionary algorithms in redesigning a new jet engine bracket, achieving an 80% reduction in the weight of the part. NASA has also been using this technique for years. In 2006, the agency used an evolutionary algorithm to design an “evolved antenna” for the Space Technology 5 mission.
And more recently, Ryan McClelland, a research engineer at NASA’s Goddard Space Flight Center, spoke to me about designed “alien” spacecraft parts designed with evolutionary AI. He was emphatic that even with the power of AI, the design process needs an engineer’s expert eye to determine if the parts will work. “You still have to apply human intuition,” he said, noting that generative AI tools tend to make parts too thin. “We can look at a tree branch and think, I’m not going to hang from that tree branch because it doesn’t look strong enough,” McClelland describes. That’s not something AI can do.
The Birmingham Blade success
The EvoPhase team, led by Dr. Kit Windows-Yule of the University of Birmingham, applied this evolutionary process to solve a problem that many turbine designs struggle to address: how to operate efficiently in urban environments, where winds are slower and more turbulent due to buildings. “We needed a turbine that could capture Birmingham’s relatively low wind speeds while managing turbulence caused by surrounding buildings,” says Nicusan. “The design also had to be compact and lightweight to suit rooftop installations.”
The resulting wind turbine rotates around a central axis with curved blade designs to optimize kinetic energy capture from the wind in cities. This design can generate power at significantly lower wind speeds compared to conventional turbines, representing a fundamental shift from traditional models, which are typically designed for optimal conditions in rural, open exposed areas. “Our evolutionary simulations have confirmed the Birmingham Blade is up to seven times more efficient than existing designs in Birmingham’s wind speeds and urban environment,” Nicusan says. “The final design is not just a prototype—it is a predictive solution that is ready for real-world use.”
Kwik Fab Ltd, an English company specializing in the precision metal fabrication of high-accuracy components, produced the initial version to validate the technical feasibility of the design and to test its mass production capability, using aluminum for its lightness and corrosion resistance. The turbine will now be installed on a rooftop in Birmingham for evaluation testing to verify lab tests in a real environment. If the data is confirmed, the final product will be available by late 2025, the developers say.
