Growth of predictive analytics bolsters wind power’s efficiency

Sumant Kawale runs Business Development at SparkCognition and his team is responsible for Customer Project Delivery and Channel Partnerships, along with supporting sales and marketing activities. Sumant is an electrical engineer by training, and has worked in Semiconductors and Telecom. He has an MBA from the Tuck School of Business at Dartmouth, and prior to joining SparkCognition he was a Project Leader at the Boston Consulting Group.

With America’s rapidly increasing reliance on renewable energy, wind farm operators are quickly adopting predictive technologies to help improve efficiency, safety and cost reductions.

Predictive maintenance was one of the hottest topics at the AWEA Wind Project O&M Safety Conference, earlier this month in San Diego. A panel of experts discussed advances in technology, best practices and applications in a discussion titled, “Predictive Futures.”

The Predictive Futures panel featured executives from Vestas, GE Wind, Genpact, Sentient Science, Element Analytics and SparkCognition. The discussion covered how advances in technology can help wind operators run their assets in ways that are safer, more reliable and more efficient.

While opinions differed on how exactly new technology would be adopted, all the panelists agreed that change is coming. Peter Wells, Vice President of Vestas Services and the panel moderator, succinctly summarized the discussion by saying, “Do something.”

predictive analytics
Discussion at AWEA’s 2016 O&M Safety Conference.

One of the key points centered on how U.S. wind can adopt cognitive technologies.

Andy Holt, Vice President of GE Renewables Services, compared building an efficient site manager to IBM’s Watson system building a virtual cancer specialist that can identify early signs of cancer 80 percent more accurately than human doctors. SparkCognition, through our partnership with IBM Watson, has long worked on cutting-edge use cases in cyber security and asset management. So it was encouraging to see that senior executives in wind technologies are now also thinking about step changes, not just incremental ones.

When it came to predictive analytics, different paths discussed by various panel members included a physics based approach and a cognitive analytics approach, which SparkCognition has employed successfully with several wind clients, including Invenergy.

I feel that each of these approaches has its own merits, and each can be a tool best suited for particular problems. A physics-based approach can be successful when determining “what-if” scenarios. The data driven (or cognitive) approach is real-time and learns in an automated manner, thus readily identifying patterns that otherwise would be very difficult to find, even with expert human effort. Cognitive analytics can shoulder a significant portion of the analytics burden and allow precious human expertise to chase the most difficult problems. Additionally, recent advances in machine learning, which is a subset of cognitive analytics, allow us to determine unknown failures.

From the panel and in my work with many wind clients, in my opinion, the key takeaway was that changes in the software that shape the U.S. wind industry will come not with home-run efforts – similar to a giant ERP implementation – but in singles and doubles. The key will be to find small projects that have positive returns on investment, then scale them in a manner that doesn’t overwhelm an organization.

Stay informed

Take Action

Subscribe to the American Clean Power blog and receive the latest renewable energy news, policy updates, and opportunities to get involved.