By Ryan Blitstein, Vice President of Renewable Energy
The current projection of the wind industry is an optimistic one. U.S. wind-power capacity increased from 60 gigawatts (GW) five years ago to more than 90 GW in 2017, and is expected to top 220 GW by 2030. Last year alone, industry growth came from $11 billion in private investment.
However, despite its growth, wind power is still a bit player in the highly competitive and diversified U.S. energy market. Other clean energy sources such as solar are catching up on the price curve. To grow faster and stay competitive, the sector must continue to innovate and further decrease its levelized cost of energy, or LCOE.
A promising way to accomplish this is by improving how machines and humans work together, through the digital transformation of renewables. This means engineers, wind techs, and machines collaborating on wind-turbine inspections, maintenance, and troubleshooting to fully maximize wind-farm production.
The World Economic Forum predicts that $17.8 trillion in value will be created during the next decade from digital transformation in some of the largest industrial sectors. The power sector is expected to gain around $3.1 trillion in value. Artificial intelligence (AI) and data are the keys to this growth and development.
While enormous potential exists with the application of AI, the wind industry has only experienced minor value from it to date. One reason is because wind owners and operators are typically focused only on data, instead of the business value that can be derived from data.
There are two recurring problems with current data-collection and analysis methods:
- Companies are looking at the wrong data
- Companies are typically analyzing it with rudimentary methods
When applying big data to challenging problems, it is critical to first determine where the most value exists — as opposed to where the most data is available. Railroads are an example of an industry that’s maximizing the use of AI. By applying AI effectively, some railroads are capable of generating $160,000 in value per locomotive each year. With a typical Class I railroad operating more than 5,000 locomotives, the numbers quickly add up.
Railroads are a collection of many systems — track, signals, locomotives, and cars — that all have worth. It is, therefore, important to first determine the asset that has the greatest value and then whether to optimize for reliability or productivity. For railroads, the greatest opportunity is in the locomotive and its reliability. For example, questions a dispatcher might ask include: Is the locomotive ready when there’s a shipment of goods? And if so, will it arrive at its destination without delay or breaking down?
In the wind industry, this is called time-based or energy availability, and it works in much the same way as in the rail industry. An operator may ask: Is my turbine available to generate power when wind is blowing? And what’s the best way to schedule work to optimize wind-farm operation and costs? This entails an analysis of many confounding variables such as air density, grid conditions, technician availability, market prices, and more.
Thanks to effective AI use, the millions of dollars saved across a fleet of locomotives is only one part of the rail industry’s success story. The main part is the people — because even in a digital, AI-driven business, data only goes so far. Owners, operators, manufacturers, and engineers are still needed to dissect the data and make appropriate value judgments. This is why leading industrial AI companies have invested heavily in researching user experience, so they gain insight to the most useful and effective information.
Effective AI is also working in the wind industry. In fact, this lesson is why Berkshire Hathaway Energy Renewables and MidAmerican Energy Company’s wind fleet is producing more energy from the same conditions on their existing 2,400 wind turbines than previous to AI’s application. According to one company, AI software recently prevented a turbine’s main-bearing failure. A find (or “save”) like this can mean up to $250,000 to an owner, including lost project revenue from the downtime.
There is little reason why the wind industry cannot produce a much greater share of U.S. energy generation. The industry has the required tools and data. Now it just needs to use them.
Five ways to maximize artificial intelligence at wind farms
1. Focus on where the value is, rather than where the data is
The area where you might have the most project data may not be where you can drive value in your operations. Start with where your main challenges are and work from there.
2. Ask how AI can drive smarter fleet-wide turbine decisions
Software should identify individual component failures letting wind-farm owners and operators make more informed, data-driven decisions across their wind fleet.
3. Make sure your AI system can be used across all your turbines
Typically, installing software that only works on one brand of turbine will result in more complexity and higher costs.
4. Even with machine help, people still matter
Even in today’s digital age, wind techs are needed to make repairs. Involve them early in the process of your site’s digital transformation.
5. Digital transformation is challenging, but worth it
The World Economic Forum estimates there is $3.1 trillion of value that can be unlocked in the power markets alone during the next decade.