Healthy transformers do not experience catastrophic failures overnight. As such, electric utilities need to look for early warning signs of an ailing transformer. By predicting a failure, utilities can curb power outages and protect their field workforce from the dangers of an explosion.
The New York Power Authority (NYPA) experienced significant issues when transformers failed in 2012 at its Blenheim-Gilboa Power Project and again in 2014 at the Niagara Power Project. Based on these events, NYPA identified that it needed a better way to protect transformer health. At this time, NYPA also started taking steps toward forging a relationship with the Israeli Smart Energy Association, which was looking for a partner in the U.S.
The association invited NYPA to share a presentation about the challenges it was facing with its transformers at one of its conferences. Several vendors and companies approached NYPA with suggestions, but one firm — mPrest, a division of Rafael Industries — dove into the main issues confronted by NYPA. The company, an Israeli software development firm that had developed the Iron Dome anti-missile system for Israel based upon a smart algorithm, proposed a solution. By applying its defense industry algorithm to a utility asset, mPrest was able to work with NYPA to develop a way to resolve its transformer issues.
Creating a Partnership
To obtain funding for the research project, mPrest and NYPA filed an application with the Israel-U.S. Binational Industrial Research and Development (BIRD) Foundation, which was established in 1977 to promote collaboration and award grants to U.S.-based companies and Israeli startups. In the summer of 2015, mPrest and NYPA were awarded a $900,000 funding opportunity to develop a transformer asset health system.
Using this opportunity, the two companies worked together on a new program to gauge the health of transformers and identify problems before they could lead to a transformer issue or even a loss of power. In addition to providing funding, NYPA research and development, operations, maintenance and transformer experts provided insights, lessons learned, and operational data and history to ensure that the system accurately depicted real-life transformer performance.
Currently, the electric utility industry has about six different methods to predict the failure of a transformer, which can happen in a variety of ways and for many reasons. For example, through dissolved gas analysis (DGA), labs can analyze the gases accumulated in a transformer’s oil sample. About every six months, utilities typically send samples to a lab to predict whether a transformer is healthy, needs to be taken down for maintenance, or will experience a catastrophic failure in the near- to mid-term.
Developing New Software
Unlike other systems on the market, the new software from mPrest and NYPA combines historical transformer data as well as information from several different on-line sensors. Because the project was vendor-agnostic, the two companies took all known methodologies and combined them into a single asset health indicator for transformers.
By taking the best of what was available on the market, the partners were able to build an algorithm based upon several different sensor types. For example, one sensor takes continuous readings of gases in the transformer oil, and another monitors the oil and winding temperature to detect any heating inside the transformer. In addition, an infrared camera on top of the transformer finds hot spots, much in the way a fire department uses this technology to find victims inside of a burning building.
The result is a system that can look ahead two and a half months to see if a transformer failure may be upcoming, but is not in the immediate future. For example, NYPA entered its previous transformer data into the system, and based upon this data and its known failures and issues, the system was able to accurately predict transformer health. In addition, NYPA worked with the Electric Power Research Institute (EPRI), which has a test transformer in Lenox, Massachusetts, to successfully test its Transformer Asset Health System. Following the testing, NYPA installed the system at the Robert Moses Niagara Power Project, the largest power plant in New York state. The system is now commercially available.
The new technology, which prevents power plant operation disruptions and loss of power, uses data from a power plant’s transformer as well as sensors, advanced algorithms, historical performance data and lab reports to assess the real-time health of a transformer. As such, engineers will have enough lead time to address issues with transformers before they turn into costly failures.
“By finding efficiencies at New York’s Niagara Power Plant, which is one of the largest renewable energy sources in the nation, this new partnership will increase efficiency and reduce costs, while ensuring that New York remains at the forefront of technological innovation,” stated New York Gov. Andrew M. Cuomo.
Tracking Data
One way that the new system differs from some others on the market is that it can be tied into an asset management system. For example, NYPA ties all of the information from the transformer sensors and Maximo asset health software into the mPreset system. If a sensor anomaly pops up, NYPA can immediately find out if it is caused by someone who had worked on the transformer and thus introduced a false positive indication.
The work orders and data from the transformer sensors all run on a computer server. The data is displayed in NYPA’s asset health center in White Plains, New York, and by reviewing the incoming information, NYPA can detect anomalies. In addition, the utility can also use various tools in the system to drill down and analyze different pieces of information, and get a better understanding of what is happening inside the transformer.
Improving Productivity
Currently, NYPA has installed the sensors on one of its transformers, but the field workforce is planning to roll out the program to about 50 of its transformers across the state. While some of the transformers are already equipped with monitors, they are not all connected to the network. As a result, the technicians and linemen currently must personally visit each transformer to read the DGA data.
By connecting all of the sensors to the network, however, all of the data will be fed back into the system. In turn, the site and maintenance personnel can log on to a computer to view the data in the office or the field. The research department worked closely with the field workforce to determine what they wanted to see in the system, how they wanted the screens to look, and how they wanted the information to be displayed on the screen. NYPA field staff also voiced their opinion about which values were more important than others.
The Transformer Asset Health System is improving the efficiency of the field workforce and enhancing safety. For example, it allows NYPA to maintain its transformers in a timely manner and warn its field employees about any assets that are due to fail. As a result, the company is also able to protect its assets and ensure that the transformers live longer — reducing the cost to NYPA and its customers. ♦
Alan Ettlinger is the director of research, technology development, and innovation for the New York Power Authority (NYPA), where he has worked for more than 20 years. He is responsible for all research and development projects at NYPA, which generates 30% of all electricity in New York state.