Behind New York Power Authority’s ‘Digital Avatar’ Project With GE
How a $1 billion project will yield $2 billion in savings through predictive analytics and integrated operations
The New York Power Authority is already a major user of smart grid technologies. With one of the country’s biggest smart energy R&D labs, big investments to ease the flow of wind power, and its New York Energy Manager platform that’s crunching data from more than 1,000 public buildings, the oldest public power organization is embracing the future.
Now it’s turning to General Electric to sensor and analyze its 16 power plants and 1,400 miles of transmission lines as part of a $1 billion investment in asset performance analytics that’s expected to yield $2 billion in savings over the next decade.
It’s the biggest system-wide deployment yet for GE’s Predix analytics platform, the software brains behind the company’s “Industrial Internet” of sensored and networked equipment and devices. It will start at four of NYPA’s natural gas-fired peaker plants, with asset performance management (APM) software that will predict failures, conduct preventative maintenance and other efficiency-boosting tasks.
But eventually, it will grow to include the integrated smart operations center (ISOC) that NYPA is building, CEO Gil Quiniones said. This central command center for its power plant and grid operations, customer-side operations, and its emergency response coordination will use GE’s Predix platform to tie together the various operations software platforms in use, like the energy management system (EMS) from Alstom Grid, which was acquired by GE.
“Think of it as a decision support system in terms of how we run our assets and how we invest in our assets,” Quiniones said. “Largely, we will suck information from our existing EMS and SCADA systems. But we’re also deploying sensors strategically, wherever there’s a data point we need to correlate.”
Using that data, GE’s software will create “digital twins, or avatars, of each of our pieces of equipment — our power plants, our substations, our power lines,” he said. These virtual models, built from real-time and historical data, can predict turbine failures days before they occur, or model the likely life remaining in individual parts within a machine to inform maintenance schedules — items that add significant cost to running power plants.
For a typical gas turbine, “our solutions can deliver up to $50 million in net present value,” said Niloy Sanyal, chief marketing officer for GE’s digital power team. GE is already doing a lot of asset analytics at the four power plants where NYPA is starting its project. “Right off the bat, we have about 65 percent coverage to all of their assets,” said Sanyal. Future release cycles will bring that coverage up to 80 percent by year’s end, all without new sensors.
The “digital twin” is GE’s way of describing the models it builds from “data that’s coming from across the enterprise, not just one source at a time.” The Predix platform, which GE built itself after trying out other big data platforms from Pivotal and other companies, uses physics-based and statistical algorithms to model and predict how assets will perform in real time.
GE competitors like Siemens and ABB offer similar asset management systems (AMS), with projects to match the scale of what GE is doing at NYPA. But GE’s software is also “at the heart of the NYPA operations center,” Sanyal said, opening the potential to use the Predix platform’s analytics capabilities for grid assets as well.
While that’s not part of the project announced last week, Quiniones said that NYPA plans to roll out the software to all of its 16 power plants, as well as to some transmission lines and substations, over the coming years. That could give it better information on how much capacity it has left on a critical transmission lines when upstate wind power is peaking, or the ability to optimize different power plants based on their relative cost to produce across the course of the day.
NYPA is also applying data analytics to the 1,000-and-counting public buildings it’s collecting data on through its Energy Manager platform, Quiniones noted. “I’d say by 2018 or so, we will have developed digital twins, or avatars, of every device in our customers’ premises.”
