The challenge of maintaining grid reliability has intensified in recent years due to a variety of factors.

Weather patterns have become increasingly volatile and unpredictable, compelling communities and their leaders to prepare for dangerous storms with little notice.

Coupled with rising customer expectations and heightened regulatory scrutiny, utilities face mounting pressure to perform effectively when it matters most. In this context, innovative solutions are essential for navigating challenges.

Currently, utilities look to traditional means to improve reliability such as moving circuits underground, sectionalizing circuits, increasing budgets for tree trimming and installing smart sensors, automation and grid-edge technologies. While these approaches are effective, they often require significant time to implement. The key missing elements are a reliability data hub, a proven artificial intelligence/machine learning (AI/ML) framework and innovative ways of working.

But what if technology could do more than just assist utilities in confronting these storms? What if it could anticipate storms and mitigate power outages before they occur?

A recent collaboration between Ernst & Young LLP (EY US) and Eversource Energy has led to the development of a patent pending framework, methods and algorithms capable of doing just that. This was made possible with the integration of diverse data sources such as weather patterns, geographical nuances, supervisory control and data acquisition (SCADA), geographic information system (GIS), vegetation management insights and other data sets. As a result, 40,000 customer outages have been avoided in just two months.

While 73% of utility executives and employees report using AI, only 18% feel the technology has met expectations, according to the EY Future of Energy Survey. However, initiatives like the work that EY US and Eversource Energy have collaborated on aim to bridge that gap. By synthesizing data from various silos, the approach facilitates efficient root cause analysis and remediation. Focusing on “AI-ready data” enhances predictive accuracy, enabling utilities to shift from reactive to proactive grid management. This transition improves operational efficiency and enhances customer experience by minimizing unexpected outages.

“The shift towards AI-driven outage prediction is reshaping the utilities sector,” Rockie Solomon, automation and analytics leader at Eversource, said. “By leveraging the right AI/ML frameworks, methods and algorithms with a reliability data hub and new ways of working, utilities can truly improve reliability and improve customer satisfaction, positioning themselves as forward-thinking organizations ready to meet the demands of a dynamic energy future.”

As utilities face the dual challenge of aging infrastructure and the integration of renewable energy sources, the effectiveness of their approach will be critical in shaping the sector’s future.

The role of AI and data integration in outage prediction

Utilities must enhance service reliability as customers demand uninterrupted power and quick outage responses, alongside regulatory pressures for justifying investments. Outages disrupt daily life and can cause significant financial losses for businesses. Traditional reactive management increases costs and decreases customer satisfaction, while the evolving grid complicates responses.

Developing a patent pending framework, methods and algorithm to predict sustained outages marks a significant leap in outage management. This was made possible with the integration of diverse data sources including weather patterns and vegetation management insights, to enhance predictive accuracy and operational efficiency.

At the core of this approach is a modern data platform that synthesizes information from systems like SCADA, advanced metering infrastructure, outage management systems and GIS. By breaking down data silos, utilities can create a comprehensive view of operations, enabling them to identify potential outage risks before they escalate.

The algorithm employs sophisticated machine learning techniques to analyze historical outage data and predict the likelihood of sustained outages. The model establishes correlations that inform maintenance strategies. For example, if it detects a pattern of voltage dips, it can trigger targeted inspections to prevent future outages.

Collaboration with field teams enriches the model’s predictive capabilities. Patrol teams provide insights into maintenance activities that impact grid reliability, creating a feedback loop that is essential for refining the algorithm.

Implementing AI-driven outage prediction models enhances service reliability and operational efficiency by identifying potential outage locations in advance. For instance, if the algorithm predicts a high likelihood of sustained outages in an area due to vegetation interference, utility teams can prioritize tree trimming and maintenance activities there. This strategy not only prevents outages, but also reduces costs associated with emergency response.

Transforming utility operations and customer experience

The benefits extend to an improved customer experience. Customers are increasingly aware of the implications of power outages, not just in terms of lost electricity, but also regarding disruptions to their connected lives. By minimizing unexpected outages, utilities can enhance customer satisfaction and build trust through effective communication about potential outages and preventive measures.

Insights from the outage prediction model enable utilities to make data-driven decisions regarding capital investments and resource allocation. As regulators demand evidence of capital expenditure effectiveness, demonstrating improved reliability metrics through predictive analytics becomes a powerful tool for securing necessary approvals.

As the energy landscape evolves with renewable resources and electric vehicles, the predictive outage model equips utilities with the agility needed to prioritize maintenance based on real-time data.