The LDE UTC microgrid will feature the first rooftop wind installation in the city of London, Chris Griffin, CEO and founder of Hover Energy, told Microgrid Knowledge.

Rather than leverage the horizontal axis turbines common to land- and off-shore-based wind farms, Hover developed a vertical axis turbine better suited for urban settings.

Mounted at the edge of the building’s roofline, the turbines harness wind as it flows up and over the roof. A rotating shroud surrounding the turbine compresses the airflow through a smaller opening.

The aerodynamics are complicated but squeezing the same volume of air through a smaller opening speeds it up, much like covering part of a hose’s opening with your thumb makes the water shoot out faster. 

“The shroud is designed to turn into the wind direction, but it does not weathervane into it,” Griffin explained. In other words, the shroud does not rotate to directly face the oncoming wind.

A lot of prior vertical axis wind turbines with shrouds made that mistake because engineers “thought of the building as a tower, which is the wrong way to think about aerodynamics,” he added.

Unlike traditional wind farms, which require significant amounts of land (or ocean floor), Hover’s rooftop wind turbines can be placed relatively close together. Through years of testing and experimentation, the firm found that properly sizing and placing the machines can increase production capacity by more than 50%.

Equally important, the systems are community friendly. “We found a solution that is vibration free, noise free, and therefore very unobtrusive to a building installation, and of course, those things matter to the people who live and work in buildings,” Griffin said.

The turbine reportedly generates roughly 64 dB of noise, comparable to a normal conversation.

While this will be the first commercial deployment of Hover Energy’s vertical-axis turbines, Griffin is quick to point out that the LDE UTC microgrid is more than just a proving ground for rooftop wind.

“We need as much production as we can possibly get on a site, and so we love working with wind and solar and biofuels and batteries and bringing all those technologies together…but we're not going to make the mistake that just because we built a cool wind turbine, we think that the wind's always going to blow because it isn't,” he said. “This is not about wind turbines. This is about microgrids.”

The power of agentic AI in optimizing microgrids

Hover co-developed its intelligent microgrid management system with IBM and Watson X. The system operates on-site through a small data center — a Qualcomm-based server rack preloaded with Hover's software — using both historical and predictive data. Because it runs locally rather than on the cloud, it continues to operate even if the main power grid fails or the connection to Hover's cloud platform is lost.

For Griffin, the real elegance in a microgrid’s design comes by appropriately sizing it to the load. For Hover, that process begins with a detailed analysis of the building's consumption profile, examining every asset in the building, what will change once the installation goes in, and what future upgrades might affect demand.

The goal is to avoid oversizing the system, which would leave excess unused power, or undersizing it, which could leave the building out of compliance with its resilience target.

“Through the installation we'll get the site to an 81% offset number, but that's not really the target,” Griffin said. “The real target here is 100% because the goal is to have full resilience 100% of the time.”

That’s where agentic AI comes in.

Hover's agentic AI continuously optimizes the microgrid's performance, and Griffin expects it will add another 15% to 20% in production beyond the system’s initial sizing, with gains leveling off around the second or third year of operation as the system approaches the point of diminishing returns.

The AI targets each asset separately. On the wind side, he expects optimization to add roughly 8% to 10% to production. Solar optimization is expected to contribute another 3% to 5%. The battery storage system, meanwhile, works to balance output from both wind and solar, storing excess energy and releasing it strategically, which Griffin says accounts for the remaining few percentage points of improvement.

Power at the edge

While Hover’s microgrid systems can be installed during new building construction, Griffin envisions it as a retrofit solution to turn existing infrastructure into a power source.

Generating power at the edge — where buildings consume energy — frees up grid capacity for other users. That means things like data center development could continue without having to build a new substation.

While Griffin said Hover has no desire to operate as a utility, and the economics of selling power to the U.K. grid don’t pencil out, connecting multiple microgrid systems together does present a number of interesting possibilities.

“If we put five or six of these together and then we optimize those, we're going to end up with a lot of excess power,” he said. One option could be offering that power to a nearby low-income housing development.

To that end, Hover has other projects on deck in the Royal Docks. The firm is building a similar microgrid at the University of East London’s Docklands location and is partnering with the university’s London Renewable Energy Laboratory to further develop intelligent renewable microgrids.

"Our ambition is to develop next-generation renewable energy systems that combine cutting-edge research with industrial application,” Hasan Baig, head of the London Renewable Energy Lab said in a statement announcing the partnership. “Working closely with Hover Energy allows us to bring together expertise in artificial intelligence, computational modelling, renewable energy systems and digital engineering to create smarter, more resilient microgrids.”

Now that the LDE UTC microgrid has gained full planning approval, which encompasses the design, layout, size and materials, the final hurdle before construction can begin is to get approval from the Department for Education, which Griffin expects to come in the next two months.

Assuming there are no speedbumps with that process, the microgrid should be operational in around nine months.

And then the real work begins.

“There are all these other benefits that come when you generate power at the edge, and showing what they are, I think, is what's needed right now,” Griffin said. “We've got to educate the planet on what can be done when you develop at the edge, as opposed to centralizing everything and building huge more power plants.”