Underground residential distribution (URD) transformers are vital to urban grid reliability but often operate in confined, hard-to-access environments that make condition monitoring difficult. As these assets age, unexpected failures can lead to costly outages and safety concerns.

This session explores how survival analysis—a technique traditionally used in medical and engineering fields—is being applied to predict probability of failure and remaining useful life (RUL) for underground transformers. Using historical failure data, operational metrics (e.g., load, temperature), and environmental conditions, advanced survival models like Kaplan-Meier estimators and machine learning-based survival regressors can pinpoint high-risk assets.

Attendees will learn how this predictive framework enables proactive maintenance, reduces downtime, and informs capital planning decisions. The methodology is scalable across other underground assets, making it a valuable tool for utilities pursuing smart grid modernization and resilience.

Takeaways:

• Understand how survival analysis techniques can be applied to underground transformer failure prediction.
• Identify key data inputs and modeling techniques to estimate probability of failure and RUL.
• Transition from reactive maintenance to predictive asset strategies to extend transformer life.
• Apply this framework to optimize capital investment and grid reliability for underground assets.
• Scale the approach across other critical infrastructure in support of smart grid initiatives.