Keeping warm in a winter outage: what a battery can and can't run

Every winter storm produces a wave of people asking which battery will run a space heater. The honest answer is that none of them will, for any length of time that matters, and the arithmetic is not close enough to argue about. That is worth printing plainly, because the follow-up question is the useful one: if not the heater, what should the battery be running? This guide works both halves from published figures. We have not tested any of these products, and the Amazon links above are affiliate links, disclosed at the top of this page.

The arithmetic nobody wants

Lasko publishes 1,500 watts on high and 900 watts on low for its 754200 ceramic heater, along with a fan-only mode. Those figures are typical of every plug-in resistive heater sold in the United States, because a 120V 15-amp circuit caps at 1,800 W and manufacturers design to sit under it.

Divide rated capacity by 1,500 W and here is what the units on this page deliver, at the ceiling:

• 1,024 Wh (DELTA 3 Plus): about 41 minutes.
• 2,042 Wh (Explorer 2000 v2): about 1.4 hours.
• 2,073.6 Wh (Elite 200 V2): about 1.4 hours.
• 3,840 Wh (SOLIX F3800 Plus): about 2.6 hours.
• 4,096 Wh (DELTA Pro 3): about 2.7 hours.

On the heater's 900 W low setting those roughly multiply by 1.7, so the DELTA Pro 3 reaches about 4.6 hours. Every one of those is a ceiling: it uses rated rather than usable capacity and ignores inverter losses, the low-voltage cutoff and the station's own standby draw. Real runtime is lower.

There is no product that fixes this and no setting that changes it. A resistive element converts electricity to heat at one to one, with no multiplication, which is the opposite of a heat pump and the reason space heaters are among the worst possible matches for stored energy in a house. A battery that carries a refrigerator for two days carries a space heater for two hours.

The corollary is that scaling up does not rescue the plan either. Even the largest units we track, at 6,144 Wh per battery on the EcoFlow DELTA Pro Ultra, are covering a single heater for one evening. Heating a house with stored electricity is an installed-system-plus-solar problem, not a suitcase problem.

Running the furnace instead

If your house is heated by gas, oil or propane, the fuel is doing the heating and the electricity is only running the fan, the draft inducer and the ignition. That is a completely different load, and it is the reason a modest battery can keep a house warm through a winter outage when it cannot keep one room warm with a heater.

Trane publishes the electrical data for its 80 percent AFUE gas furnaces in its product and submittal literature, and the figures are concrete:

• Blower motor full load amps at 120V: 4.1 on one 40,000 BTU configuration, 6.4 on another, and 9.2 on the 60,000 and 80,000 BTU models. All are constant-torque ECM motors with nine speed taps.
• Inducer motor full load amps: 0.30 on the smaller models, 1.33 to 1.40 on the larger ones.
• Ignition: a 120V silicon nitride igniter, drawing only during the ignition sequence.
• Total ampacity: 5.6 to 13.1 amps depending on model, with maximum overcurrent protection of 15 amps on every model in the table.

Read those carefully, because two of them are ceilings rather than draws. Full load amps is the motor's maximum, and on a nine-speed ECM the heating tap is normally not the maximum, which is set by the cooling airflow requirement. Trane does not publish a heating-speed draw, so we do not print one. Ampacity is a circuit-sizing figure for the electrician, not a consumption figure for you.

Take the largest published blower figure, 9.2 amps at 120V, as a worst case: roughly 1,104 W while the blower is turning. That is close to a space heater. The difference is duty cycle. A furnace runs to satisfy a thermostat and then stops, while a resistive heater runs continuously precisely because it cannot keep up. At a 30 percent duty cycle, a 1,104 W blower averages about 331 W, and a 4,096 Wh unit against that average is roughly 12 hours at the ceiling. At the 4.1 amp figure for a smaller furnace, about 492 W while running, the same battery goes much further.

The catch is connection. A furnace is a hardwired load, not a plug, so reaching it means an electrician-installed transfer switch or a backup panel. That is the same constraint covered in our critical loads panel guide, and it is the piece of this plan to arrange before winter rather than during a storm.

Heat pumps are a different problem

An air-source heat pump is efficient exactly where a resistive heater is not, and it is also the hardest thing on this list for a battery to start. The gate is the compressor's locked rotor current, and it is published: Goodman's GLXS3B specification sheet lists compressor locked rotor amps of 41.2 on the 1.5-ton model and 148.0 on the 5-ton, at 208/230V single phase. At 240V, that 148 amp figure is roughly 35.5 kVA for the starting instant, against a largest published surge of 14,400 W among the units we track.

Two further complications sit behind that. Heat pumps are 240V hardwired loads, so the connection problem applies before the surge problem does. And many air-source systems include electric resistance backup heat, sometimes 10 kW or more, which engages on the coldest nights. If your system has auxiliary strips, the load your battery would be asked to carry on the worst night of the year is the resistive one, at several times a space heater. Our air conditioner guide works the same locked-rotor arithmetic from the cooling side, and the conclusion is identical.

The loads that actually earn their capacity

Once you stop trying to make heat with the battery, the list of things worth backing up in a winter storm is short and cheap to run:

• The furnace blower and controls, as above, which is the whole heating system as far as the electricity is concerned.
• A refrigerator, which averages well under 100 W once the compressor's duty cycle is accounted for, per the arithmetic in our refrigerator sizing work.
• Network gear and phones, in the region of 15 W for a modem and router pair.
• A well pump, if your water is not municipal, which is a much larger and more demanding load covered in the well pump guide.
• Low-draw personal warmth. Heated blankets and mattress pads are the one heating product whose draw is small enough to matter, and here we have to flag a gap: of the brands we checked, including Sunbeam, none publishes a wattage figure on its own product pages. The number lives on the appliance's own label and in retailer listing spec fields, where figures in the region of 95 to 180 W appear for full-size blankets. That is a retailer-published range, not a manufacturer datasheet, and the figure to trust is the one printed on the blanket you own. At 100 W, a 1,024 Wh unit is roughly 10 hours at the ceiling and a 4,096 Wh unit roughly 41 hours, which is the only heating-adjacent arithmetic on this page that comes out in the battery's favor.

The thing the battery itself has to survive

A winter outage is also the condition under which the battery is least likely to perform to its datasheet, and every maker publishes an operating range that is narrower than most buyers expect, with charging limits usually tighter than discharging limits. A unit stored in an unheated garage may refuse to charge at all. That is a separate question from what it can run, and we cover it in home batteries in cold weather alongside the published temperature ranges.

The safety guidance, quoted rather than paraphrased

Two federal sources publish direct instructions about the alternatives people reach for when the heat goes out, and we quote rather than summarize them.

CDC states: "Never use a gas range or oven to heat a home," and "Never use a generator, pressure washer, or any gasoline-powered engine inside your home, basement, or garage or less than 20 feet from any window, door, or vent," adding that this applies "even if the doors or windows are open, unless the equipment is professionally installed and vented." It also advises using a battery-powered or battery-backup carbon monoxide detector when a generator is running. Source: CDC, What to Do to Protect Yourself During a Power Outage, read August 6, 2026.

Ready.gov carries the same instruction: generators and fuel "should always be used outdoors and at least 20 feet away from windows, doors and attached garages," with working carbon monoxide detectors installed on every level of the home, and it states directly that a gas stove or oven should not be used to heat a home. Source: Ready.gov, Power Outages, read August 6, 2026.

A battery has no exhaust, which is the one unambiguous advantage it holds over a generator in this scenario, and the honest comparison on every other axis is in our battery vs generator guide. For any heating appliance you do use, the manufacturer's own instructions govern.

Choosing between the picks

• A gas furnace and a storm-length outage: the EcoFlow DELTA 3 Plus at 1,024 Wh and 1,800 W, whose 3,600 W published surge clears the largest blower figure in Trane's table with wide margin.
• You want to cover a whole night: the BLUETTI Elite 200 V2 or the Jackery Explorer 2000 v2, at roughly double the capacity, with the Jackery publishing a switchover under 20 ms and the BLUETTI publishing pure sine wave output in its spec table.
• A furnace plus a refrigerator plus lights, for more than a day: the EcoFlow DELTA Pro 3 at 4,096 Wh, expandable to 12 kWh on one unit.
• A 240V circuit is in the plan: the Anker SOLIX F3800 Plus, whose 6,000 W is available through its L14-30R port only, with each 120V outlet group capped at 2,400 W.

Every figure above is a manufacturer datasheet or federal agency publication recorded with its stated basis: heater wattage from Lasko's own product page, furnace electrical data from Trane's published product and submittal literature, compressor locked rotor amps from Goodman's GLXS3B specification sheet, power station figures from each maker's own spec table, and safety guidance quoted from CDC and Ready.gov. The runtime numbers are arithmetic on those figures under the assumptions stated, not measurements we made.