The increasing frequency of extreme weather events presents a growing challenge for the nation’s electrical grid and energy infrastructure. With over a quarter of U.S. homes fully electrified and solar installations projected to triple in the next five years, understanding how homes respond to power outages, winter storms and heat waves has never been more critical.

At Stevens Institute of Technology, Hoboken, N.J., assistant professor Philip Odonkor, an expert in energy technologies and artificial intelligence, is at the forefront of this effort. His latest research leverages A.I.-driven models to identify the hidden vulnerabilities of solar-powered and electrified homes, providing invaluable insights that could revolutionize emergency preparedness, urban planning and sustainable infrastructure development.

The push toward electrification is essential for reducing carbon emissions, but it also raises important questions about resilience during extreme weather events. As power grids experience greater strain due to climate change, how well do fully electrified homes withstand disruptions?

The study revealed a critical insight: solar-powered homes exhibit strong resilience during summer heat waves, thanks to abundant sunlight for energy production. However, these same homes proved to be highly vulnerable during winter storms when solar generation is low and heating demands surge.

Fully electrified homes were found to be nearly three times more vulnerable to winter outages compared to those relying on a mix of energy sources, according to Stevens Institute of Technology.

While solar panels provide critical support during hot summer months, they struggle to meet the intense heating needs of winter blackouts. Without proper backup generators, energy storage or alternative heat sources, homeowners could face significant risks when extreme cold disrupts power supplies.

Traditionally, identifying electrified homes required manual inspections or door-to-door surveys—a time-consuming and invasive process. Odonkor’s team eliminated this challenge by developing machine-learning models that can determine a home’s energy system and potential vulnerabilities with over 95% accuracy, simply by analyzing its energy consumption data.

The implications of this research extend far beyond individual homes. As aging power grids are increasingly tested by severe weather events, A.I.-powered planning tools will become essential for building climate-resilient communities. Emergency response teams can prioritize vulnerable households during power failures. Energy providers can develop more reliable backup systems for electrified neighborhoods. City planners can design long-term infrastructure that balances sustainability and resilience.