AI data analysis, combined with new information-collecting techniques, could help businesses more effectively maintain transformers and prevent failure, writes Emily Newton.

Transformer maintenance and management can be complex. Even well-maintained models can fail in various ways, and experts must analyze several data types to identify and solve problems.

AI data analysis, combined with new information-collecting techniques, could help businesses more effectively maintain transformers and prevent failure.

Manufacturers could also leverage AI to improve the quality of their transformers. Here are some things companies should keep in mind when it comes to the upkeep of these machines.

Right now, transformer tests and condition assessments are typically performed without significant amounts of automation. Technicians inspecting a transformer will use various techniques — like resistance and winding turns ratio testing — to ensure a transformer is functioning properly.

Transformers also require daily inspection in the form of oil level checks, leak inspections and the examination of silica gel, in the case of breather transformers.

Data gathered by transformer sensors, like magnetic oil gauges, usually needs to be obtained, transcribed and reported manually. Some systems may report important information automatically to a SCADA or asset management platform, but this isn’t the norm.

Daily and less-frequent tests can be labour-intensive and may require the skills of a technician with experience in assessing transformers.

Like most industries, the energy sector faces a significant labour crisis right now, and it’s not unusual for skilled experts to be in short supply. As a result, many businesses may struggle to properly maintain and inspect their transformers.

Have you read: AI is key in solving complex energy transition challenges Smart Grid AI Centre of Excellence launched in Canada Australia’s Telstra Energy turns to AI to optimise renewable assets

Transformer failure is typically expensive. It may cost as much as $14 million depending on the severity and amount of downtime it causes— and even brief outages can cost a company $1-$2 million.

Preventive maintenance can be costly, but it’s almost always less expensive than a major failure. Maintaining transformers isn’t always a straightforward process, however. Technicians can miss minor issues that may eventually develop into serious problems, even with a theoretically effective maintenance schedule.

As businesses struggle with a shortage of skilled technicians, these problems can become even more difficult to navigate.

Artificial intelligence-powered automation of transformer monitoring and testing is one possible solution to this issue. By combining smart remote monitoring technology and AI analysis of incoming data, businesses may be able to automatically predict failures and identify potential transformer issues.

New, smart sensors can continuously monitor essential transformer information — like oil levels — and report this information to the cloud. If necessary, the data these sensors collect can also be streamed to management platform dashboards and SCADA software.

Technicians can access this data no matter where they are, so long as they have access to the cloud network where the information will be stored. As a result, they can monitor transformer performance continuously.

Information that comes in can also be analyzed automatically, further reducing the workload placed upon a business’s transformer technicians.

Artificial intelligence algorithms have excellent pattern-finding abilities. They can detect subtle correlations between sensor information and failure conditions with enough data. As a result, an AI algorithm may be able to see failure coming long before a technician can — even if it’s looking at the same data that a business’s technicians are using.

These AI systems are trained using massive datasets of existing transformer testing and monitoring data. A comprehensive dataset will typically use information from new, in-use and failed transformers, exposing the AI algorithm to various operational states and failure conditions.

The greater variety of data available, the more effective the AI may be at teasing out correlations between operational variables and transformer maintenance needs.

Similar approaches to maintenance and testing — often called a “predictive maintenance” strategy — are becoming increasingly common in manufacturing and other heavy industry sectors.

The approach is a valuable cost-saving method that can help prevent downtime. This also extends the utility of existing IoT data-collection systems often used for remote monitoring, making smart sensors and similar technology a more valuable investment.

While still experimental, there’s evidence that predictive maintenance can be much more effective than preventive maintenance. This allows businesses that already invest in effective upkeep strategies to further reduce operational costs.

Data collected and analyzed by AI algorithms may have uses beyond maintenance and operation. AI analysis of factory and manufacturing process data can allow manufacturers to identify process bottlenecks, connect failure modes to manufacturing conditions or even develop new tools to accelerate transformer design.

These innovations could make it possible to design cheaper transformers that are more reliable, helping reduce the initial and operational costs of a new transformer.

Information from the field can also help manufacturers build better transformers. If one fails or begins to behave unusually, businesses may have access to months of sensor data leading up to the failure.

Manufacturers can use this information to pinpoint what may have caused the transformer to fail — allowing for new strategies and maintenance recommendations that may help make them easier to maintain.

As with predictive maintenance, this AI-informed approach to manufacturing is already being adopted in several sectors. The right technology can be used to apply AI’s predictive powers to identify design errors and manufacturing bottlenecks that may reduce productivity.

AI tools are also used for other quality-improvement processes. For example, machine vision AI can dynamically classify objects using information from video cameras. These products are often used to automate visual inspections during the quality control process, helping improve product quality without increasing employees’ workloads.

Businesses wanting to effectively maintain transformers face significant challenges. Testing and repair can often be a difficult, time-consuming process, and most companies face a considerable labour shortage right now.

Even with the best maintenance practices, transformer owners may sometimes miss problems that will eventually lead to failure.

New AI-powered solutions may help businesses keep transformers running for longer. The right technology allows companies to automatically gather and analyze important operational data, making maintenance easier and more effective.

Emily Newton is the Editor-in-Chief of Revolutionized Magazine. She enjoys writing articles in the energy industry as well as other industrial sectors.

© Synergy BV | Company number: 30198411 | Registered in the Netherlands at Bisonspoor 3002, C601, 3605 LT Maarssen