What 100,000 simulations reveal about our power grid

On August 13, 2003, a single transmission line near Cleveland, Ohio, sags into an overgrown tree limb and short circuits. Within minutes, nearby lines overload and trip one after another, triggering cascading failures across grids already strained by an all-too-common convergence of high demand, inadequate maintenance and software bugs that leave operators blind.

Within hours, more than 50 million people across eight U.S. states and Ontario, Canada, lose power. The outage lasts up to 29 hours in some areas, which causes between $7 billion and $10 billion in economic damage, and contributes to almost 100 deaths. In major cities suddenly plunged into total darkness for the first time in decades, some residents say they can clearly see the Milky Way.

Twenty years later, infrastructure researcher Nasir Ahmad of the Department of Energy's Oak Ridge National Laboratory is working to ensure communities are never again unprepared to face the devastating aftermath of cascading infrastructure failures.

"We've always known these events could happen - but without data, they stayed theoretical. I wanted to help move our frame of mind from What if? to How likely?," said Ahmad.

Rising electrical demands - compounded by aging infrastructure, growing cyber security threats and unpredictable extreme weather events - make measuring these risks increasingly urgent. Today's power grids rely on interconnected networks, integrating advanced energy sources, smart-sensing technologies and cutting-edge digital communications. But this connectivity creates complex dependencies: A disruption in one component can rapidly propagate across interconnected systems, magnifying localized problems into widespread outages.

Predicting widespread outages

Using sophisticated modeling techniques and city-scale datasets, Ahmad and his collaborators at Arizona State University, The New School, and Texas A&M University recently ran more than 100,000 unique simulations to assess the likelihood of cascading grid failures in the infrastructure of modern U.S. cities. The team simulated interconnected water networks and electric grids in the city of Phoenix and found something startling. There's a roughly 4% chance that the failure of even a single critical substation could trigger widespread outages, impacting electricity and the water supply. This is an alarming probability for a scenario once considered too unlikely to warrant rigorous planning.

"We were quite surprised to see a figure this high," Ahmad said. "We were hoping for something less than 0.5%."

To arrive at these results, which were published in civil engineering journal ASCE OPEN, the team built detailed synthetic models of Phoenix's electrical and water distribution networks - realistic digital stand-ins constructed from public data.

"You can't always get the real infrastructure data," Ahmad added. "But we know things like population, demand and road layouts - and critical systems like power and water usually follow those roads."

Using these patterns, Ahmad designed digital networks that mimic the layout and behavior of real systems, down to the pressure in water pipes, the load on substations and the relationship between the two networks.

The models allowed the researchers to conduct extensive simulations, testing various failure scenarios and their potential impacts. By introducing disruptions and adjusting parameters such as the number, location and intensity of line overloads, the researchers could observe how failures might propagate through interconnected systems, providing valuable insights into potential vulnerabilities. The team's digitally linked networks allowed them to simulate whether outages stayed local or spread outward into the city.

"Ahmad was instrumental in developing our synthetic infrastructure models," acknowledged Mikhail Chester, a collaborator from Arizona State University. "His approach let us realistically simulate infrastructure vulnerabilities even when detailed data was scarce."

Across 120,000 simulations, almost 90% resolved without major incident. Another 7% caused localized or partial disruptions - such as water pressure drops in neighborhoods or isolated substation failures - without triggering widespread blackouts. But in 3.69% of cases, a single failure set off a chain reaction that swept through both systems, leaving wide swaths of the city without electricity or water. Some of these widespread failures represent what the researchers called "Black Swan" events, which are seemingly minor abnormalities that result in extreme societal disruptions. While scenarios this severe are uncommon, Ahmad emphasized that quantifying their likelihood is a critical first step to helping cities prepare.

Insights from synthetic infrastructure simulations like these will directly inform ongoing ORNL initiatives, translating theoretical understanding into practical tools that strengthen U.S. energy resilience and empower decision-makers with the tools to act before a crisis.

Mitigating Black Swans

The lab's Enhancing Distribution Grid Anticipatory Resilience, or EDGAR, project is building a regional early warning system for very small municipal and cooperative utilities, starting with the Tennessee Valley region. The system will automatically issue alerts up to 72 hours before a likely outage, highlighting not just the forecasted weather event, but its potential impacts on infrastructure and the communities it serves. That advance notice could shrink triage time, reduce service interruptions and improve outcomes for high-need customers including medical personnel and hospitals, first responders and elderly populations reliant on uninterrupted electricity.

Elsewhere, the Technical Assistance for States and Tribes Initiative: Grid Resilience Investment Decision-Making (TASTI-GRID) helps state energy offices pinpoint resilience upgrades that deliver the biggest impacts for the most people, guiding investment decisions with precise, data-driven insights.

Both projects leverage multidisciplinary expertise within ORNL's Critical Infrastructure Resilience group, which focuses on identifying vulnerabilities and enhancing infrastructure preparedness nationwide.

"The group does a lot of this kind of analysis to help states and utilities figure out where to invest," Ahmad said. "If we can show where the problems are before a failure happens, then we can help them act before things start to break down."

Ahmad and his colleagues view their work in Phoenix as a first step in a much larger effort. According to Chester, they hope to expand these robust modeling frameworks to test adaptation strategies across other cities and in defense environments - anywhere critical interdependent infrastructure may be at risk.

For Ahmad, the goal is to ensure that communities can anticipate and address vulnerabilities before they escalate into disasters. By better understanding complex infrastructure threats, decision-makers will be better equipped to safeguard the communities they serve and keep energy reliably flowing.

UT-Battelle manages ORNL for the Department of Energy's Office of Science, the single largest supporter of basic research in the physical sciences in the United States. The Office of Science is working to address some of the most pressing challenges of our time. For more information, visit energy.gov/science. - Galen Fader