Confronted by an aging turbine fleet and determined to drive down operating costs, wind project owners are turning to software to optimize the performance of their assets. For owners whose assets encompass multiple wind farms and turbines from several manufacturers, third-party digital solutions simplify and streamline operations and maintenance logistics across projects.

A report released earlier this month by software analytics firm Uptake highlights the opportunity in the space. Uptake is among a growing cohort of companies employing data analytics to optimize wind farm operations. According to research conducted by advisory firm DNV GL that is cited in the report, the U.S. wind fleet’s current availability -- that is, the readiness of turbines to generate electricity -- sits at 94 percent.

“For each 1 percent gain in availability,” the report found, “an estimated additional 2.4 TWh of wind energy would be produced by the current fleet without new hardware.” Uptake believes it is possible to achieve 99 percent fleetwide availability by using predictive analytics software to proactively address maintenance issues.

According to Aaron Barr, a senior consultant with MAKE Consulting, “The wind energy market presents one of the most promising areas of industrial big-data application. There are now hundreds of thousands of wind turbines operating globally, each of them heavily instrumented and highly controlled, leaving many opportunities for data-based optimization.”

He added, “Improved prognostics through big-data applications can allow owners to get better pricing through advanced scheduling, catch component failures before they escalate, and ensure all turbines are operating at optimal conditions.”

Using analytics to predict problems, streamline maintenance

Sonny Garg, Uptake’s global energy solutions lead, said in an interview that his company’s software provides three types of insights: predictive, diagnostic and prescriptive. “We’re able to bring in data not only from the sensors in the turbine or from SCADA [supervisory control and data acquisition], but also from work management and asset management systems. We’re able to correlate failures with historic work orders.”

He went on" “We’re able to give diagnostics on why we think it’s failing and then give a prescription on what to do about it. What that does is not only increase the availability, [but] also reduce the time [needed] to fix it. So, when I send technicians up to the tower, they have a recommendation and a diagnostic that allows them to take the right tools and fix it the first time.”

In an interview, Chris Crosby, principal of global nuclear and renewable energy at OSIsoft, another player in the wind data space, said the industry is aiming for true prognostics: that is, “being able to calculate the remaining useful life [of turbine components] within some type of probability range.”

Delivering data-based optimization for multi-brand wind fleets

The pursuit of digital solutions to optimize wind farm O&M is not exclusive to software firms. Predix from GE, ClearSight from Vestas, and Winsight360 from Siemens are among the offerings from major turbine manufacturers.

“Probably the most sophisticated offerings in the market are coming from the manufacturers themselves,” conceded Uptake’s Garg. “The OEMs [original equipment manufacturers] were really positioned to win the space because they have the data science across multiple component systems.”

And yet both Garg and Crosby contend that software firms like theirs have an edge over the OEMs in providing optimization solutions, especially for asset owners with multi-brand wind fleets.

To start, they tout their independence and brand agnosticism. If a turbine manufacturer offers a recommendation to a project owner, “There is an inherent tension as to for whom the incentives are aligned,” said Garg. “If I am the OEM, and I give you a predictive insight that says, ‘You need to fix your gearbox,’ am I doing that because I’m trying to sell you after-market parts, or am I doing that because I’m really trying to help you be successful?”

“Whether it’s true or not,” he said, “there’s a perception that incentives aren’t aligned between the manufacturer and the owner.”

Crosby said one OSIsoft customer has turbines from eight manufacturers in its wind portfolio. He added that OEMs have not always made 100 percent of their data available to asset owners. It depends on the contract negotiated between the OEM and project owner.

“When you’re a customer and you have eight of these OEMs you’re dealing with,” he said, “you really want to have data infrastructure in place in which you’re owning as much of the data as possible and your own people are using it for multiple applications.”

Aaron Barr noted that asset owners often lack the technical or commercial expertise to develop digital infrastructure and machine-learning algorithms and are turning to software companies like Uptake, OSIsoft and Sentient Science, which is using its DigitalClone technology to predict gearbox failures at NextEra Energy wind farms, for solutions.

Uptake signed a 10-year deal in March of last year with Berkshire Hathaway Energy, the U.S.’ second-largest wind energy asset owner, to deploy its predictive analytics software across Berkshire’s wind fleet, which includes GE, Siemens and Mitsubishi turbines. “By August of this year,” said Garg, “all their sites, across their 7 gigawatts, will be on the software.”

Real-world wind farm optimization via big data

For asset owners, wind farm optimization tools offer a way to wring O&M savings from aging fleets. “Roughly 60 percent of a typical wind energy O&M budget is dedicated to unplanned maintenance -- this compares to less than 10 percent in a conventional power plant,” said Barr. “Using data to better forecast unplanned maintenance events can yield significant savings.”

In 2014, Dong Energy (now Ørsted) began deploying OSIsoft’s PI System with Esri’s ArcGIS mapping tool to reduce service calls to its offshore wind turbines. By feeding wind turbine data into the analytical software, Crosby said, “We were able to automatically detect an anomaly, automatically generate a work request in the work management system, and visualize all of that in a map.”

Maintenance at sea is staggeringly expensive. According to Ørsted, it is up to 15 times more expensive to service a turbine at sea than on land. Ørsted expects to save 20 million euros each year by 2020 by reducing unscheduled maintenance visits to each turbine from four to two annually.

Crosby cited another application for optimization tools: avoiding curtailment. Iberdrola is saving $3 million monthly, he said, by feeding wind turbine power generation data, across multiple wind farms, into the PI System to develop curtailment optimization plans.

The analytics enable Iberdrola to respond to production signals and curtailment requests from the transmission system operator. “What the application really is about,” said Crosby, “is the ability to get high-frequency data in, do the calculations very quickly, and then be able to respond back to the system operator.”

“The stories, in many ways, are the same,” Michael Kanellos, senior manager of corporate communications for OSIsoft, said in an email. “If you get enough data, you can get a far, far better view of what’s going on and fix problems before they happen or optimize your power production.”

“The challenge is getting that data. A turbine puts out a ton of information, and you have to act in real time.”