AI is Not a Magic Bullet for Renewables

Print Friendly, PDF & Email

In this special guest feature, Blair Heavey, Chief Executive Officer of WindESCo, suggests that we hear so much about how AI is the next big thing in the energy industry, but in reality, it’s really over-hyped, and is nowhere near the magic bullet that some thought it might be. Blair brings with him a long history of success in scaling up companies across a broad range of industries, including software, healthtech, fintech and digital marketing. An undergraduate of Boston College with an MBA from Babson College’s Olin Graduate School of Business, he is an experienced software, SaaS, digital marketing, fintech, media and healthtech executive, with proven success in leading both startup and Fortune 100 teams to achieve outstanding results, including multiple acquisitions and IPOs valued in excess of two billion dollars.

As governments look to transition away from fossil fuels and expand renewables, clean energy sources such as wind power are on the rise. The International Energy Agency (IEA) predicts that the world’s renewable power supply will grow 50% between 2019 and 2024. Also, the clean energy provisions included in the recently signed Covid-19 stimulus relief bill will further the deployment of renewable energies across the US.

Wind the Number One Source of Renewable Energy

Smart Energy International reported that wind is the number one source of renewable energy.   According to the American Wind Energy Associates (AWEA), in 2019, $14 billion was invested in new wind projects, and wind energy accounted for 39% of all new US generation capacity. However, most wind plants are not producing energy at their full potential, and certainly not to the level of their original investor projections. Most do not know the extent of their hidden value. With downward pressure on energy prices, wind farm operators are looking for solutions to ensure every asset achieves its optimum energy production and reliability.  

Artificial Intelligence Hyped as Game Changer for Renewables

Artificial Intelligence (AI) alone has been touted as a trailblazing tool to reduce operating costs, boost efficiencies, and maximize AEP.  But while AI has been a boon to many industries, including healthcare, retail, and marketing, it has not yet been the predicted game-changer for the renewable energy industry. 

One of AI’s strengths is its ability to work with vast quantities of standardized data.  However, the algorithms are ultra-sensitive to poor data. There are no industry standards for supervisory control and data acquisition (SCADA) information, performance operating data, or other data on the numerous turbine models today in the wind industry. This lack of standardization or consistency across the data patterns creates a challenge for algorithm accuracy. Furthermore, wind power has multiple variables, including different turbine models, gearboxes, turbine locations, climate change, wind patterns, etc. again, which impact the algorithms’ output.

There is a misunderstanding in the market as to what AI actually is.  Often it is confused with machine learning (ML) or deep learning.  Although each describes software that behaves intelligently, there are marked distinctions between all of them.

Human Oversight Necessary to Maximize Wind Farms’ Profit and Performance

Machine learning has proven beneficial in monitoring turbines’ performance for preventive maintenance, which has reduced turbine pole climbs to check on blades.  But while AI and ML are useful tools in optimizing wind farms—they have specific limits due to the many variables explained above. It is critical to have human expertise oversee what decisions the AI models recommend on shifts in yaw, pitch, blades, wake, and other specific turbine performance factors to optimize wind plant output.

For example, an algorithm could interpret all of the data correctly for one type of turbine. But because there are so many turbine models and no industry standard data for each, the turbine model or location at that specific wind plant could be misinterpreted by the algorithm.  This error could result in a proposal to adjust the pitch or blades in a manner that damages the turbine, or the anticipated energy value is not delivered, causing a decrease in revenue from the farm. Additionally, an adjustment to one turbine can impact all downstream turbines, causing further loss. To ensure that recommendations to address inefficiencies and optimize performance are accurate, it is vital to have expert human oversight with wind farm domain expertise to validate proposed solutions. 

OEM Stranglehold on Turbine Data

Original Equipment Manufacturers (OEMs) typically withhold turbine performance data for a variety of reasons. They are very defensive about their products’ performance and often do not allow their customers or third-party partners to view complete performance data.  Without this data, wind farm operators cannot make adjustments that could dramatically improve their farms’ efficiency. 

Rather than relying solely on AI to solve performance issues, the wind industry would be better served by taking steps to support industry data standardization and releasing performance data by OEMs.  These initiatives and ensuring people with considerable wind farm domain expertise oversee optimization proposals would help boost the annual energy production (AEP) of wind farms and make the most of complex wind data.

Sign up for the free insideBIGDATA newsletter.

Join us on Twitter: @InsideBigData1 –

Speak Your Mind