How long will it run? Runtime, explained and calculated
Divide the station's usable watt hours by the load's average watts. Usable means the rated capacity times 0.85, and average means the rated watts times the duty cycle, because most appliances cycle on and off. A fridge rated at 150 watts typically averages about 60 watts across a day, so a battery with 3,264 usable watt hours runs it for roughly 54 hours, not the 21 hours the naive math suggests.
The two numbers everyone gets wrong
Rated watts are not average watts. A refrigerator compressor runs in bursts. Across 24 hours, a typical home fridge averages 35 to 45 percent of its rated draw. A chest freezer is similar. A sump pump might average 10 to 20 percent in normal rain. A TV, a laptop, and a router are the opposite: they draw close to their rating the whole time they are on.
Rated capacity is not usable capacity. The battery's watt hours pass through an inverter to become AC power, and the conversion costs roughly 15 percent on this class of equipment. Planning on 85 percent of the printed capacity is the standard honest adjustment, and it is why two stations with the same sticker size can feel different in a real outage.
Worked examples, on one mid-size battery
Using a battery with 3,264 usable watt hours (an Anker SOLIX F3800 Plus at the 85 percent factor), here is the arithmetic this site uses everywhere:
| Load | Rated watts | Duty cycle | Average watts | Runtime alone |
|---|---|---|---|---|
| Refrigerator | 150 | 40% | 60 | about 54 hours |
| Chest freezer | 120 | 40% | 48 | about 68 hours |
| Fridge and freezer together | 270 | 40% | 108 | about 30 hours |
| Sump pump, 1/3 HP, steady rain | 800 | 15% | 120 | about 27 hours |
| TV and laptop setup | 210 | 100% | 210 | about 15 hours |
Combine loads and the averages simply add. The fridge, freezer, and sump together average about 228 watts in these assumptions, which is about 14 hours on that battery with nothing left for lights or phones. That is why the order you care about matters, and why solar recharge changes the answer completely on day two.
Day two is a different problem
Everything above assumes one charge. The moment panels enter the picture, the question changes from how long the battery lasts to whether the daily harvest covers the daily load. Using the load list from our sizing guide (about 3,200 watt hours a day for a fridge, a freezer, lights, phones, and a router) and a battery with 3,264 usable watt hours:
| Setup | Daily harvest | Daily net | What it means |
|---|---|---|---|
| No panels | 0 Wh | -3,200 Wh | Empty in about a day |
| 800W of panels (about 2,700 Wh on a decent day) | about 2,700 Wh | about -500 Wh | The same battery now stretches past six days |
| 1,600W of panels (about 5,400 Wh on a decent day) | about 5,400 Wh | Positive | Holds even in good sun; the battery becomes a buffer, not a countdown |
Two honest caveats ride on that table. Decent days are not every day: a grey December day can return a third of those harvests, which is why winter plans are sized on winter sun. And the station's solar input cap can quietly veto the third row; a station capped at 800W of input lives in the second row no matter how many panels lean against the garage. Both caveats are checkable before purchase, in the compatibility explainer and on the station's spec sheet.
Run your own numbers
The runtime calculator does this math with editable presets for every load above, plus lights, a router, and a CPAP. Change the duty cycles to match your house: a fridge in a hot garage cycles far more than one in a cool kitchen, and a sump pump in a spring thaw is not the same load as one in a drizzle. If your combined runtime comes out under your typical outage length, the fix is a bigger battery, an expansion battery, or solar input, in that order of cost.
Who should skip this
Skip the fine math if your outages last under two hours. At that scale, almost any station in the 1,000Wh class covers the fridge and the phones, and your time is better spent on the buying mistakes page so you do not overpay.
Your next step: run your own loads and get hours, not guesses.
Open the runtime calculator