What one battery actually runs during a Triangle ice storm

Ice storms are the outage event that matters here. Not hurricanes, which arrive with warning and mostly weaken inland, and not summer thunderstorms, which take down a line for two hours. Ice loads limbs across overhead distribution under heavy canopy, and restoration in Durham and Orange counties can take a day or more.

In February 2026 a freezing rain event left parts of our service area without power for between nine and thirty four hours. Eleven of our monitored storage systems islanded. We pulled the telemetry afterwards, with each owner's permission, because runtime claims should come from data rather than from a brochure.

The load schedules people actually chose

Across the eleven sites the backed up daily energy ranged from 6.1 kWh to 19.4 kWh, with a median near 10.2 kWh. Every single schedule included the refrigerator, lighting and networking. Nine included at least one mini split zone. Four included a well pump. Two included a medical device. None included an electric water heater, and none backed up a conventional central air conditioner.

That last point is worth dwelling on. The single most requested backup load in our sales conversations is air conditioning, and the single most commonly dropped load once people see the arithmetic is air conditioning. A 3 ton condenser consumes more energy in six hours than most of these households used in a day.

Runtime results

Single module sites, 13.5 kWh usable, ran a median 21 hours before reaching the 20 percent reserve floor. The shortest was 11 hours at a site carrying a well pump and two mini split zones. The longest single module result was 31 hours at a site with a 6.1 kWh daily schedule and, critically, a clear afternoon on day two.

Two module sites, 27 kWh usable, all survived the event with reserve to spare. The lowest state of charge recorded across those sites was 38 percent, at the Durham home that had storage installed before its array.

Solar recharge during the outage was the difference between one day of backup and continuous backup. Six of eleven sites recharged above 70 percent on day two.

The recharge effect

Six of the eleven sites had an array alongside storage and saw useful sun on the second day. Those systems recharged the battery while running the house, with export blocked, and all six finished the outage above 70 percent state of charge.

The February sun angle is low and the days are short, so production during the event ran between 30 and 55 percent of the same array's June output. That was still enough. A 7 kW array producing even 15 kWh on a cold clear February day covers a 10 kWh backup schedule with 5 kWh left to put back in the battery.

This is the argument for pairing rather than buying storage alone. A battery on its own is a fixed quantity of energy. A battery with an array is a quantity that refills every morning.

What went wrong

Two observations worth publishing. At one site the owner had added a space heater to a backed up circuit after commissioning. A 1,500 W resistive load runs a 13.5 kWh module flat in nine hours by itself, and it very nearly did. We now include a short written note in the handover pack about what not to plug into backed up outlets.

At a second site the reserve had been set to 10 percent rather than our default 20 percent, at the owner's request, to maximise day to day arbitrage. The system rode the outage down to the floor overnight and the household woke to no refrigerator. Reserve settings are a resilience decision, not an efficiency one.

What we changed

• Storm reserve now defaults to 30 percent between December and March, with a simple in app override.
• The handover pack carries a one page list of what the backed up circuits will and will not tolerate.
• We run the outage drill at commissioning with the household present, not as a remote test.
• Where a well pump is present we now size the inverter for the pump surge before anything else on the list.

The honest summary

One 13.5 kWh module carrying a realistic essential load schedule covers a typical Triangle outage. Two modules cover almost anything short of a multi day event without sun. Neither will run your whole house the way the grid does, and any installer who implies otherwise has not done the arithmetic with you.