Psst… We received a tip…
The ICU’s Most Dangerous Moment
It's 2:47 a.m. in Tampa General's cardiac ICU. A hurricane is three hours from landfall. The grid is already wobbling.
A ventilator hums beside bed 12, where a 67-year-old woman is recovering from a triple bypass. Down the hall, a premature infant lies in the NICU. In the basement, diesel generators wait for the grid to drop so they can take over.
Except the most dangerous moment can come before the generators ever matter.
Hospital electricians have a name for it: "the blink". The gap between utility power dying and backup power stabilizing. Sometimes it's half a second. Sometimes it's long enough for voltage to sag and sensitive gear to hiccup. Monitors reboot. Pumps reset. Screens go black, then back.
Most of the time, nothing bad happens. But “most of the time” isn’t good enough when you have hundreds of patients on life support.
In 2017, Hurricane Irma cut power to a nursing home in Hollywood, Florida. The generator couldn't carry the load. Twelve residents died from heat exposure in the days that followed. They had a backup plan. It still failed in the gap.
Researchers at Florida International University have studied what breaks during hurricanes and why. The sad takeaway is that our backup strategy is reactive. We wait for failure, then scramble to respond.
This week we're uncovering a patent that gives batteries the ability to predict blackouts and act first.
Here’s the inside scoop
The patent originates from Florida International University (FIU), specifically the lab of Arif Islam Sarwat, who brings two decades of Siemens-honed grid expertise to the project, alongside co-inventor Alexander Joseph Stevenson.
The project ties in with how we operate BESSs.
In case you didn’t know
A battery energy storage system (BESS) is a battery plus power electronics and control software that stores electricity and releases it later when needed. It’s used to shift energy across time (charge off-peak or from surplus generation, discharge during peaks), stabilize power quality and the grid, and provide backup power during outages.
In practice it’s often paired with on-site generation like solar panels (and sometimes a diesel or gas generator in a hybrid microgrid), which is a big reason BESS is getting popular as wind and solar expand and grids need flexibility to handle variable supply. Read more here.
The problem
A big on-site battery ends up living a double life.
Most days, it is a money tool. You charge when power is cheap, discharge when it is expensive, and shave demand charges. Plenty of commercial platforms already do this well.
Then hurricane season rolls around and that same battery is supposed to be your insurance policy.
If the grid drops, you want it as close to full as possible. The nightmare scenario is that you saved a bit on Tuesday, and then on Wednesday the lights go out and you are sitting on a half-empty battery.
That tension is the "battery paradox." Somebody has to decide when to stop chasing savings and start hoarding charge for a likely outage.
What already exists
Most serious facilities already have parts of the stack.
Economic dispatch software that charges when power is cheap, and discharge when it is expensive.
Microgrid controllers and protection equipment that can "island" a site (disconnect from the utility and run locally).
Storm playbooks, often manual. A watch gets issued, someone tops up to 100%, and you hope you timed it right.
The real question is whether you can switch into storm posture early enough to matter, without constantly overreacting.
How the patent solves the problem
The core idea is to keep doing the normal, cost-reducing dispatch, but run an outage-risk pipeline in the background using a microcontroller. When risk crosses a threshold, the controller tightens its rules and starts protecting the state of charge.
The claim summary describes a controller that
Pulls a local weather forecast and market price data via the web.
Reads the battery's current state of charge directly from the BESS.
Predicts two things it needs for planning: how much solar the site will generate and how much load the facility will draw.
Predicts a third thing that drives the "resilience" decision: the likelihood of a regional outage in the surrounding distribution system.
Converts that outage outlook into a charge requirement (how much energy you should hold for a likely outage duration).
Solves for a dispatch schedule using an optimizer (it explicitly mentions mixed-integer linear programming for the economic schedule), then pushes setpoints through the control hierarchy for execution.
So the patent is really a way to stop the battery paradox from becoming a human timing problem. It keeps the everyday money-saving dispatch running, but adds a simple trigger in the background that watches weather and outage risk and then starts “saving the battery for tomorrow” early enough to matter.
Publishing the future
Extreme Weather Is Turning the Grid Into a Health Risk
Climate change is stress-testing the grid in ways that expose how fragile it actually is. When Hurricane Irma hit Florida in 2017, Centers for Disease Control and Prevention mortality surveillance later linked 47 deaths to power outages, including 17 heat-related deaths tied to the loss of air conditioning and three deaths among people whose medical care depended on electricity, like supplemental oxygen. Fourteen of the heat-related deaths were concentrated in a single assisted-living facility that went days without power. (CDC)
The whole power system was built and designed in one climatic era and now is being asked to work in a different climatic era.
Market Opportunity for Behind-the-Meter Resilience Solutions
This creates an opening. Utilities are conservative institutions and experimentation with critical infrastructure can kill people. Their conservatism creates space for private solutions. If a hospital can’t trust the grid to stay up during a hurricane, and can’t trust its utility to invest in resilience fast enough, it can buy its own battery system with its own predictive intelligence.
The regulatory environment is slowly shifting to accommodate this. Federal Energy Regulatory Commission Order 2222, finalized in 2020, allows distributed energy resources like batteries to participate in wholesale electricity markets, creating revenue streams that help pay for resilience investments. (Federal Energy Regulatory Commission)
Traditional backup generator manufacturers are about to learn something uncomfortable. Diesel generators are cheap and proven, but they can’t predict anything. They sit inert until power fails, then take 10–30 seconds to spin up and reach conditions where transfer typically occurs. (Cummins Inc.) A battery system paired with fast switching can provide effectively seamless continuity, including “zero transfer time” in some UPS configurations. (CyberPower) For facilities where seconds matter, that’s worth paying a premium.
The risk of it all
The tradeoff is false positives. If the AI predicts an outage that doesn’t materialize, the facility has just paid peak electricity rates to charge a battery it didn’t need and potentially foregone revenue from grid services. The patent addresses this through confidence thresholds, but tuning those thresholds involves real tradeoffs between safety margins and operating costs.
The patent press travels far and wide…
Extra! Extra! Read All About It!
Battery storage is scaling into a real grid asset class, and AI is becoming the control layer that decides when batteries chase profit versus when they hold charge for resilience.
BloombergNEF says global storage additions hit 97 GWh in 2023, which is roughly enough stored electricity to run about 110,000 average U.S. homes for a month on a single full discharge. By 2030, it projects 442 GWh a year, enough for roughly 500,000 homes for a month. (BloombergNEF)
From a smaller base, MarketsandMarkets estimates “AI in energy” at $8.91B in 2024, rising to $58.66B by 2030, driven by forecasting, grid optimization, and storage optimization use cases. (MarketsandMarkets)
Tesla deployed 46.7 GWh of storage in 2025 and posted a record $1.1B gross profit in energy generation and storage in Q4. Autobidder is the value multiplier, a real-time trading and dispatch software that can turn each new battery deployment into recurring, software-like economics. (Tesla)
Fluence sells grid-scale battery storage systems and related software. It reported IPO proceeds of around $1bn in 2021, a clear sign that “storage plus optimization” is being financed at platform scale. As of late January 2026, Fluence Energy (FLNC) has a market capitalization of approximately $3.54B to $4.93B, driven by strong growth sentiment and a record backlog. (Fluence Energy)
Stem, Inc. provides software to manage and optimize clean energy assets. It reported contracted storage assets under management of 5.6 GWh at end of Q4 2024. (Stem Inc.)
Resilience is becoming a funded category, too.
California Public Utilities Commission says it authorized more than $1B in SGIP funding through 2024, including eligibility for “critical facilities” tied to community resilience. (California Public Utilities Commission)
Florida is a stress test because storms make “battery paradox” decisions unavoidable
Florida Power & Light Company is the state’s largest electric utility. Its planning updates have pointed to expanding battery deployments, reflecting how quickly storage is becoming part of resilience and reliability planning. (Utility Dive)
The paper boy always delivers
An American University’s hurricane corridor has patented an AI system that watches weather radars and tells batteries to prepare for blackouts before they happen, turning "backup power" from a reactive scramble into a proactive handoff.
If your facility's backup plan involves waiting for the lights to go out before doing anything, how much is that gap really costing you?
Read the source: US 12,470,072 B1, System and Methods for the Resilient AI-Based Data-Driven Dispatch of a Battery Energy Storage System • Granted Nov. 11, 2025 • Assignee: Florida International University Board of Trustees
For the nerds
In emergencies, hospitals are relying on microgrids with Tech Brew: A readable tour of why hospitals are building “their own grid” and what actually changes when they do.
Hurricane Helene underscores need for more solar-battery microgrids with Canary Media: On-the-ground reporting about communities using solar plus storage when flood damage strands them from the grid.
How Google’s DeepMind tool is ‘more quickly’ forecasting hurricane behavior with The Guardian.
Tomorrow’s Power Grid Will Be Autonomous with IEEE Spectrum: A big-picture “software eats the grid” piece that pairs well with your predictive-dispatch thesis.
Is energy trading Elon Musk’s next big plan? with Trellis: Fun read on battery dispatch software as a money machine, which sets up the “profit vs preparedness” tension.
The research paper behind the patent: If you really wanted to get deep in the technicals.