Runtime evidence · sourced load
A window AC draws 390 W while running, and 1,800 W at startup. 8 stations clear the capacity, inverter, and startup checks at that load.
By the BatteryRank Engineering Team
01 · Runtime ladder
Nominal capacity × depth of discharge × inverter efficiency ÷ sourced average draw.
| Capacity class | Estimated runtime | Evidence |
|---|---|---|
| 300 Whstep 1 | 0.6 hr | Window AC (8,000 BTU) load · 2026-07-06 |
| 500 Whstep 2 | 1 hr | Window AC (8,000 BTU) load · 2026-07-06 |
| 1,000 Whstep 3 | 2 hr | Window AC (8,000 BTU) load · 2026-07-06 |
| 2,000 Whstep 4 | 3.9 hr | Window AC (8,000 BTU) load · 2026-07-06 |
These are comparison projections, not a promise. Temperature, inverter behavior, cycling, and the individual station's measured delivered capacity can change real runtime.
02 · Reading the numbers
An 8,000 BTU window unit runs at roughly 650 W once the compressor is turning, cycles about 60% of the time to hold a set temperature, and spikes to about 1,800 W at each compressor start. That spike is the whole story: it is nearly three times the running draw, and it repeats every time the thermostat closes.
Reading "650 W" and buying a 1,000 W station. It will not start the unit — not once, not on a good day. The other common error is assuming an inverter-driven or "soft start" model behaves the same way: those ramp the compressor and can cut the surge dramatically, which is a genuinely different sizing problem from the conventional unit modelled here.
This models a conventional single-speed 8,000 BTU unit. Larger units scale up roughly with capacity, and inverter-compressor models break the pattern entirely. If your unit is labelled inverter or variable-speed, treat the surge figure here as a ceiling, not a prediction.
03 · Passing stations
Some links on this page are affiliate links. BatteryRank may earn a commission at no extra cost to you. Commission never affects whether a station passes these checks; when two stations tie on every measured criterion, the one with a live retail offer is shown.
Critical-load fit
Why it fits
6,550 W above the plan's running load (7,200 W continuous output)
Main compromise
4,800 W below the highest output reserve across 8 eligible models
Best all-round fit
Why it fits
75.6 BatteryRank score — the highest of any station that fits this plan
Main compromise
4 points below the highest-scoring eligible model across 8 eligible models
Maximum useful reserve
Why it fits
80.3 projected runtime hours
Main compromise
2,560 Wh below the largest sourced expanded reserve across 8 eligible models
This ordered data view is separate from the three recommended buying strategies above.
Runtime uses each linked model's sourced capacity and output data; open its dossier for provenance.
Solar effect
Computed with this model's sourced solar-input limit and BatteryRank's published harvest assumptions. Check the solar method → · Model sources →
04 · Evidence record