Advanced Pattern Recognition Transforms Electric Power Operations

Every month, Kristin C., a Predictive Maintenance Manager at a U.S.-based electric utility logs more than 900 miles of road time on the job, in her pursuit of operational excellence. At each of the seven power generation sites under her supervision, Kristin methodically takes vibration readings of the equipment that serves the utility’s coal, natural gas, and wind turbine operations.

Kristin has driven this route each month for the past three years, collecting and analyzing the data as part of the time-based management system the company relies upon to reduce costs, minimize risk, and maximize productivity.

Nevertheless, there was a growing consensus at the company that more needed to be done. With the many thousands of assets her utility employed – from furnaces to cooling towers, from turbines to turnbuckles – the company was awash in sensor data. Yet the data languished, unused.

“We were data rich but action poor,” says Kristin.

TAMING THE DATA OF HUNDREDS OF THOUSANDS OF ASSETS

From its headquarters, the utility manages hundreds of thousands of asset tags. Paradoxically, the enormous volume of data this creates had led to both information overload – alarm rates of fifty to sixty per hour were typical under normal conditions – and poor visibility of emerging issues.

In early 2016, a company reliability team set out to define the company’s requirements in a data-driven Monitoring and Diagnostics solution, and solicited bids from several companies. The management team ultimately selected a solution that bundled Atonix Digital Monitoring & Diagnostics (M&D) with Black & Veatch’s Remote Monitoring and Diagnostics services.

Atonix M&D collects sensor, device, and subsystem data from across the ecosystem, applies machine learning to benchmark healthy operational behavior patterns, and alerts personnel of anomalies, while streamlining problem isolation and diagnosis. M&D is powered by AtonixOI, a SaaS platfom that powers a portfolio of SaaS products that Atonix Digital customers can use throughout the infrastructure life cycle – including design, construction, and operation phases – to improve ROI, reduce risk, and improve decision making. Black & Veatch leverages AtonixOI capabilities as part of its Remote Monitoring and Diagnostics services, providing an additional layer of surveillance and analysis for busy operations specialists in lean organizations.

A CRISIS AVERTED DURING PILOT

While analyzing historical asset data in preparation for the pilot, AtonixOI algorithms uncovered an unusual pattern. Data pulled from the utility’s data historian indicated an increase in vibration, too low to trigger an alarm, yet troubling enough to warrant further investigation. Indeed, closer inspection by company engineers found a dangerous development: early-stage delamination of a cooling tower fan shaft.

“We had just taken vibration data on that equipment,” says Kristin, “and found no issues. This was about a month before a twelve-week outage we planned for Spring.”

Had Black & Veatch not detected the issue using Atonix Digital AtonixOI, the fan assembly would have failed during the peak Summer season. In other words, there would have been an unscheduled outage of four to six weeks at a time when the utility’s customers needed it most.

Further, the unscheduled outage would have cost the company the materials and labor to replace the fan, shroud, and motor, as well as to repair certain damage to the cooling tower. The company would undoubtedly have faced safety risks and the very real prospect of a plant derating. “Since implementing AtonixOI we’ve discovered three more issues, nearly identical,” says Kristin. “We estimate that all those failure scenarios combined would have cost us around $120,000.”

LOSSES HALTED FROM WIND TURBINE DEFECTS

Another significant issue arose in 2018, this one at one of the utility’s wind farms. After connecting to each of the site’s 33 turbines, the Black & Veatch M&D team found that two were underperforming. Upon investigation, company reliability analysts found that defective anemometers had been overreporting wind speeds, causing the turbine blades to overpitch.

Kristin recounts, “This equated to $20,000 a year in lost revenue, starting in 2012 for one turbine and 2014 for the other.”

A DATA-DRIVEN TIME-BASED MANAGEMENT STRATEGY

Company engineers have since woven Atonix Digital’s Monitoring & Diagnostics deep into the fabric of their operations.

“We can email directly from AtonixOI,” explains Kristin, “and attach any graphs or reports we want. Then recipients can click ‘Reply’ and respond to everyone while posting back to the AtonixOI thread.”

The newly enhanced time-based management model has already begun to influence the utility’s plans. Company managers have begun to evaluate more advanced sensors, capable of providing even richer performance data and enabling alarm banding.

This will enable us to see when an issue’s actually happening,” says Kristin. “When we get that, we’ll want to feed it all into AtonixOI as well.”