Airtime Blog

How Artificial Intelligence is Enhancing P&WC Engine Maintenance

By PRATT & WHITNEY CUSTOMER SERVICE
March 10, 2022 | Customers, Diagnostics, Performance | 4 min read
Pratt & Whitney Canada is using machine learning, a type of artificial intelligence, to step up the capabilities of its oil analysis technology. Here’s how it works—and what it means for customers.

From three dimensions to a hundred

Pratt & Whitney Canada’s innovative oil analysis technology, available on a wide range of engines,  is part of the company’s expanding portfolio of digital engine health solutions that are reducing maintenance costs and helping customers move toward 100% planned maintenance. By analyzing microscopic traces of metal debris from oil-wetted components inside the engine, oil-analysis technology can detect specific engine conditions with high level of detail.

Now, the oil analysis team is combining the solution with cutting-edge AI applications to make it even more sensitive and enhance its early-detection capabilities.

Previously, data collected from oil samples was carefully analyzed by P&WC experts, who looked for trends and outliers that indicated possible future issues and recommended maintenance actions to mitigate them. Now, as the size and complexity of the datasets continue to expand and the solution continues to evolve, additional tools and technologies are being used. 

That’s where AI and machine learning come in. AI’s ability to find patterns in vast arrays of data is exponentially greater than that of humans and other analytics tools, explains Julien de la Bruère-Terreault, applied data science engineer at Pratt & Whitney Canada. And, as the name suggests, with machine learning, the computer i.e., “the machine”, can learn complex decisions rules from training examples provided by P&WC experts.

“Oil analysis technology allows us to classify each particle of metal in the engine oil as a specific material from a certain engine component and track the levels of it. With AI, we can identify and quantify more abstract patterns in the data, which are difficult to extract and analyze using simple numerical values in a chart, plot, or diagram,” he said.

Before we introduced AI, it was like we were looking at oil analysis data on the engine in three dimensions. Now we can look at it in a hundred dimensions or more. It helps us to better interpret the data and deliver more precise and accurate maintenance recommendations to customers.
Julien de la Bruère-Terreault, applied data science engineer, Pratt & Whitney Canada

AI leads to faster analysis, earlier detection

AI and machine learning makes the task of comparing oil samples with existing data simpler, more efficient, and easily scalable.

Through AI, the computer can, for instance, compare a new sample with 500 others very quickly – work that’s much more labour-intensive when performed by people. Automating the process frees up personnel to focus on more value-added activities, such as finding new insights from the data and developing new machine learning models to help customers reduce maintenance costs and avoid unplanned events.

The greater precision also leads to earlier detection of engine-wear patterns that may require proactive intervention.

“With our oil analysis technology, the biggest thing we’re trying to bring to the customer is proactive maintenance planning and cost savings,” said Andy Kim, OilAT technical specialist at Pratt & Whitney Canada. “The aim is to inform them about possible issues before they even realize anything is happening in the engine so the issues can be minimized or avoided altogether.” 

He estimates that with the AI-enhanced oil analysis technology, customers will be informed about potential issues as far as 1,000 hours in advance, compared to several hundred hours with the legacy version. The results of the enhanced solution are already having a significant impact.

For example, a major government customer is now using P&WC’s oil analysis technology with AI capabilities as its primary tool for mitigating engine issues across its fleet of PW210-powered helicopters. The enhanced detection capabilities have led to cost savings on potential secondary damage and an overall improved customer experience.

Our oil analysis technology has always been much more sensitive than traditional methods. With AI, we can give customers an even longer lead time for scheduling preventive maintenance actions such as removing or repairing a part and avoiding secondary damage. For our customers, who are in the business of flying without delay, this provides them with a significant advantage in terms of staying on schedule.
Andy Kim, OilAT technical specialist, Pratt & Whitney Canada.

More comprehensive and consistent analysis

AI’s capacity to go beyond the limitations of human analysis leads to improved and more consistent results and eliminates the element of subjectivity.
Thanks to AI, we can also more accurately determine when we need oil samples and reduce the number of interventions required of customers.
Andy Kim, OilAT technical specialist, Pratt & Whitney Canada

“Our oil analysis technology is a tool that’s always evolving. We’re continuously investing in it and collecting more data to better support operators. What we’re doing with AI is taking place behind the scenes, so it’s not obvious to customers, but it is enhancing the value of oil analysis for all of them,” said Andy. 

Oil analysis technology backed with trained AI models is already being used for PT6A-66D, PW210, PW600, PW150A, PW306 and PW308 engines, with more to be added in the future.

For more details on the way this technology works and how it is designed to support customers’ needs, check out an earlier Airtime article: 3 Key Benefits of Our Oil Analysis Technology.