When Grids Go Dark, Neighbors Keep the Lights On

When a storm knocks out the grid, most communities can only wait for repairs.  A new publication from Internet Think Tank’s Energy Sustainability Lab shows how linked neighbors can do better: see the outage coming and line up shared power before the lights go out.

The journal paper, A Resiliency Assessment of Predictive Power Outage Integration in Energy-Transactive Networked Microgrids Architectures, was published September 15, 2026, in the Journal of Power and Energy Engineering. It was led by Dr. Mohamed Abaas, Director of Internet Think Tank’s Energy Sustainability Lab, with Drs. Pritpal Singh and Ross A. Lee of Villanova University’s College of Engineering. The paper tests the autonomous microgrid design for which Dr. Abaas was recently awarded a U.S. patent (US-12388256-B2).

The headline finding: in a simulated community with solar panels and battery storage, adding AI-based outage prediction to a blockchain-based transactive energy platform made outages 36% shorter on average for each affected customer.

The U.S. power grid is reliable, but it is also centralized, which leaves it exposed to severe weather and cyberattacks. When a major line or plant fails, the outage can spread quickly, and restoring power often depends on slow manual work. The 2003 Northeast blackout cut power across eight states and cost an estimated $6 billion. In 2017, Hurricane Maria plunged Puerto Rico into a total blackout, and the island’s grid took two years to reach 98% recovery.

Networked microgrids offer a different path: clusters of local grids that can run on their own or support one another when the main grid fails. Until now, research on these systems has looked either at predicting outages or at trading energy, rarely both. This study is the first to show the advantages of combining the two in a single platform and measure the effect on resilience.

Imagine two residents, Bob and Sally, are neighbors in an autonomous microgrid. Bob usually has solar power to spare, and Sally buys it from him automatically over a secure blockchain network.

Now a storm is approaching. Normally, Sally’s home gets half its power from the utility and half from neighbors. But when the AI model predicts she is about to lose grid power, it shifts all of her demand to the neighborhood market, so her purchases from Bob are lined up before the grid fails.

That shift, from reacting to an outage to preparing for it, is exactly what the new study set out to measure.

The researchers simulated three linked microgrids in the Pittsburgh area serving nine facilities, from two hospitals and a police department to homes, a school, and local businesses. Facilities used solar panels, batteries, and, at critical sites, backup diesel generators.

The simulation used real historical weather data from western Pennsylvania, ran trading on a blockchain test network, and forecast outages hourly with an AI model that had shown 98% agreement with recorded grid reliability data.

Each facility has its own blockchain account, and a smart contract handles every trade automatically: it collects offers to buy and sell, checks balances, executes the trade, and records the payment. Prices are tied to Pennsylvania’s solar sell-back rate, which rises during an outage, to encourage neighbors with surplus power to sell. Critical facilities such as hospitals run backup generators sized to federal emergency requirements and trade for part of their remaining needs. The team then compared the same community with and without outage prediction.

The study’s central result is a 36% improvement in resilience, and battery storage is what made it possible. In the community equipped with both solar panels (PV) and batteries, outage prediction made outages 36% shorter on average for each affected customer and cut total outage time across the community by 6%.

To measure resilience, the team used standard IEEE reliability indicators, the same kind utilities use to report how long customers go without power. The 36% figure reflects how much shorter outages were for the customers who experienced them.

The difference comes down to timing. Solar panels produce power only while the sun shines, but batteries let neighbors bank that energy and keep sharing it through the night and the length of an outage. Prediction tells the community when to prepare; batteries give it something to prepare with.

• Homes and small businesses gained the most.  With batteries, outage time fell 22% and 14% for the two homes and 12% for the retail store.
• Outages got shorter, not less frequent.  Prediction helps a community ride through an outage faster; it does not prevent the outage itself.
• Costs held steady. Overall costs were nearly unchanged, and some homes and the retail store saved 3% to 6% of their annual electricity bills.

The study also points to a refinement. Because the test market was on a first come – first serve basis, some large critical facilities received slightly less power during outages. Future trading rules should give priority to hospitals, police, and other critical services.

The researchers call these results preliminary and plan to test more facility types, build a smarter trading optimizer, and model how communities actually trade.

Meanwhile, Internet Think Tank continues to mature the patented technology, including testing autonomous microgrids in storm-prone regions such as Puerto Rico.

As severe weather grows more frequent, the lesson of this research is that resilience does not have to come only from bigger power plants and longer transmission lines. It can also come from neighbors who see trouble coming and are ready to share.

About Dr. Mohamed Abaas. Dr. Abaas earned his PhD in Sustainable Engineering at Villanova University, where his research examined how blockchain and AI can make microgrids more resilient through a cyber-secure energy trading platform. He is a Fulbright alumnus and holds U.S. Patent US-12388256-B2 for autonomous energy microgrid design. See the INTTK team page.

Learn more

• Full paper (open access): Journal of Power and Energy Engineering
• Background: Internet Think Tank patents new microgrid technology
• The lab: Energy Sustainability Lab
• Follow INTTK on LinkedIn

 Contact. Energy Sustainability Lab, Lab Director Dr. Mohamed Abaas: info@inttk.org, (Tel) 267-322-8352. Utilities, municipalities, and researchers interested in pilots or collaboration are welcome to reach out.