Skip to content
RtRunTimeFit

Methodology

How RunTimeFit estimates runtime

Hours = rated Wh × (1 − reserve) ÷ ((average AC watts + idle draw) ÷ inverter efficiency) calculated. When a maker publishes no efficiency we use 0.85; only 1 of 43 stations publish both efficiency and idle draw maker spec. Solar recharge counts 75% of panel rating. Data updated .

Each page lists the assumptions it used, so you can check them against your own device label.

Formulas

  • Battery draw (W) = (average AC watts + inverter idle watts) ÷ AC efficiency + average DC watts ÷ 0.95.
  • Runtime (h) = rated Wh × usable share ÷ battery draw. The usable share is 100% unless the maker states a reserve or you set a discharge limit.
  • AC efficiency: the maker's published figure; otherwise 0.85, a default. Idle draw is counted only where the maker publishes it.
  • Runs: the total running watts must not exceed the station's continuous rating.
  • Starts: the other loads' running watts plus the largest starting surge must not exceed the surge rating. If no surge rating is published, we compare to continuous and say so.
  • Boost modes (such as X-Boost or Power Lifting) lower the voltage for resistive loads. They are never counted.
  • Cycling loads such as fridges use the ENERGY STAR median annual energy ÷ 365 per day; otherwise running watts × duty cycle × hours, or running watts × hours.
  • Solar recharge (h) = rated Wh ÷ (the lowest of panel W, max solar input W and any combined input cap × 75%).
  • Outside the maker's published temperature range, we show "not supported by maker" instead of hours.

Small overnight loads

When the AC load averages under 100 W, the 0.85 default and any unpublished idle draw can move the answer by a lot. A CPAP or a router alone is in this range. The calculator shows a warning for these plans, and the hours should be read as an upper estimate for stations that don't publish idle draw.

Stock status

Each station row records whether the maker's own store showed the base unit in stock, out of stock, or didn't say, with the date checked. Picks and shortlists skip out-of-stock stations by default; the calculator has a switch to include them.

How the verdict box picks

  • Only current stations with a maker list price that the maker store showed in stock are eligible. Stations shown out of stock, or whose store shows no availability ("availability not shown"), are left out of verdict boxes and best-of picks; the calculator can include them with its checkbox.
  • Best fit: the cheapest eligible station that runs, starts and lasts at least 1.2 times the plan's hours, among stations with a published warranty of at least 2 years and a published surge rating. If none qualify, we first drop the warranty and surge preference; if no station has the spare runtime, we take the cheapest station with the warranty and surge preference that lasts the plan, then the cheapest that lasts it at all. The box says which fallback applied. On a station page, that station takes this slot.
  • Budget: the cheapest eligible station that works at all, shown with its trade-offs. Left out when it is the best fit.
  • More headroom: the cheapest eligible station that works in the next larger capacity class that has one, other than the Budget pick (class bounds at 600, 1,000, 1,500, 2,000, 3,000, 5,000 Wh).
  • Best-of pages pick the same way as Best fit. Comparisons pick the station with the lower list price per watt-hour of usable output that one reviewer measured for both, with the same method.
  • The calculator's three picks are cheapest that fits (the Budget rule), best value (price per usable Wh, adjusted for chemistry and warranty) and best with solar (fastest recharge from your panels), all from the same eligible stations.

What each label means

  • Maker spec: a value the maker publishes, with the source and the date we read it.
  • Calculated: our formula applied to spec values, with the assumptions listed.
  • Third-party test: a result published by a reviewer or lab, quoted as published, with a link and its method.

RunTimeFit does not run its own runtime experiments. Sources and the dataset are described on the data page. Mistakes are fixed under the corrections policy.