Why Your Power Station Runs Out Sooner Than the Math Says

By Hicham M. · Updated October 17, 2026

Most runtime figures, ours included, take the battery’s watt-hours, knock off about 15% for the inverter and divide by the device’s watts. That works well for a kettle. For a router or a CPAP left on overnight, the real runtime can be a fraction of that, and the main reason is a power draw that never shows up on the spec sheet.

1. The spec-sheet math

Runtime (hours) = Capacity (Wh) × 0.85 ÷ device watts

It’s what our runtime calculator uses by default (see how we calculate). The 85% stands for the inverter’s conversion loss. It treats that loss as a fixed share of the load, and that’s the weak spot.

2. The inverter’s idle draw

An inverter turns the battery’s DC into 120 V AC. Once it’s switched on, it uses power to keep its own electronics running, whether anything is plugged in or not. Standalone inverter makers publish this as no-load or zero-load power. Victron’s datasheet for its Phoenix inverters (230 V models) lists 8 W for the 1,200 VA model on a 12 V battery and 20 W for the 3,000 VA model, falling to 5 W and 15 W in its power-saving AES mode and to 2 W and 8 W in search mode.

Power station makers rarely publish it. BLUETTI is the exception we found: several of its US manuals give a “self-consumption” figure and build it into their runtime formula.

Operation time = Battery capacity × DoD × η ÷ (load + self-consumption)

DoD is the share of the battery the station lets you use and η is inverter efficiency. The manuals use the formula for small loads, and the AC180 manual works an example: a 40 W fridge on its 1,152 Wh battery, at 90% DoD and 85% efficiency, with about 15 W of self-consumption, runs “about 17 hours”. (The formula as printed actually works out to about 16.) Leave out the 15 W and the same formula gives 22 hours.

BLUETTI published self-consumption and what it does to a router’s runtime
Model	Self-consumption (BLUETTI)	DoD, η (BLUETTI)	Router, formula without it	Router, formula with it
BLUETTI AC180	about 15 W	90%, 85%	2.4 days	29 hours
BLUETTI AC200L	about 12 W	95%, 93%	5 days	2.8 days
BLUETTI Elite 100 V2	about 10 W	90%, 90%	2.3 days	33 hours
BLUETTI Elite 200 V2	about 10 W	90%, 85%	4.4 days	2.6 days
BLUETTI Apex 300	about 20 W	95%, 90%	6.6 days	2.8 days

Self-consumption, DoD and efficiency as stated in each user manual. Where a manual says “over 90%” or “typically over 93%”, we used that figure. Router = 15 W Wi-Fi router plus modem. Calculated with BLUETTI’s formula, not measured.

With a 15 W router, a self-consumption of 10 to 20 W cuts runtime by 40% to more than half. That’s the whole story of small loads: a fixed overhead is a big share of a small number.

3. Why small loads lose the most

Here is the share of the battery’s output that goes to the inverter itself, assuming 15 W of idle draw, roughly what BLUETTI states for the AC180.

Idle draw as a share of total battery draw
Load	Watts	Share lost to a 15 W idle draw
Wi-Fi router + modem	15 W	50%
CPAP (no humidifier)	30 W	33%
Fridge, averaged over a day	58 W	21%
Desktop PC + monitor	150 W	9%
Microwave	1,500 W	1%

A fridge sits in the middle for a second reason. Its compressor is off much of the time, but the inverter isn’t. Through every off-cycle the station keeps paying the idle draw for nothing. See the refrigerator page for how we average a fridge’s draw.

4. What our simulation shows

Because the flat 85% hides this, every model page also carries a simulated runtime, and the real-world runtime simulator runs the same model for any device. It steps the battery down every 2 minutes and repeats 200 times with the uncertain inputs varied:

An idle draw of 8 W plus 6 W per 1,000 W of rated output, varied from 60% to 150% of that, unless the maker publishes a figure.
A conversion loss of 4–8% of the load, plus a loss that rises with the square of the load (4% at full output).
92–98% of rated capacity usable before the protection cutoff.

These are assumptions about a typical unit, not measurements. Here’s what they give for three common loads on four models, using the exact numbers shown on the model pages.

Spec-sheet runtime vs simulated runtime
Model	Device	Spec-sheet math	Simulated, typical	Likely range
EcoFlow RIVER 3 Plus 286 Wh, 600 W	Wi-Fi router + modem	16 hours	9.8 hours	8.5 hours to 11 hours
	CPAP (no humidifier)	8.1 hours	6.2 hours	5.7 hours to 6.8 hours
	Refrigerator (full-size, ENERGY STAR)	4.2 hours	3.7 hours	3.4 hours to 4 hours
Jackery Explorer 1000 v2 1,070 Wh, 1,500 W	Wi-Fi router + modem	2.5 days	29 hours	25 hours to 37 hours
	CPAP (no humidifier)	30 hours	20 hours	18 hours to 24 hours
	Refrigerator (full-size, ENERGY STAR)	16 hours	13 hours	12 hours to 14 hours
BLUETTI AC180 1,152 Wh, 1,800 W	Wi-Fi router + modem	2.7 days	35 hours	35 hours to 36 hours
	CPAP (no humidifier)	33 hours	23 hours	23 hours to 24 hours
	Refrigerator (full-size, ENERGY STAR)	17 hours	14 hours	14 hours to 15 hours
Anker SOLIX F3800 3,840 Wh, 6,000 W	Wi-Fi router + modem	9.1 days	2.5 days	47 hours to 3.2 days
	CPAP (no humidifier)	4.5 days	46 hours	39 hours to 2.5 days
	Refrigerator (full-size, ENERGY STAR)	2.3 days	33 hours	30 hours to 40 hours

Simulated, not measured. Nonstop runtime from full to cutoff at room temperature. Likely range = middle 80% of 200 runs.

Across all 54 models that can run it, the simulated router runtime is a median 53% shorter than the spec-sheet figure, from 34% on the Jackery Explorer 240 to 76% on the Jackery Explorer 5000 Plus. For the CPAP the median gap is 35%, and for the fridge 22% (on the 44 models that can start it). The biggest stations lose the most on small loads, because their bigger inverters are assumed to idle higher.

How good is the assumption? The model pages for these BLUETTI models now use BLUETTI’s published figures instead of our guess. Our generic formula would have assumed about 19 W for the AC180, against BLUETTI’s stated 15 W. For the five BLUETTI models in the table above, the stated self-consumption is below our generic assumption in 5 of 5 cases, so the generic formula probably errs on the pessimistic side for big inverters. We don’t know the test conditions behind BLUETTI’s figures, and we found no such figure in the other makers’ manuals we read.

5. The other losses

Reserve. Stations stop before the cells are truly empty. BLUETTI’s manuals say the AC180 and Elite 100 V2 work at 90% depth of discharge and the AC200L and Apex 300 at 95%.
Conversion loss. BLUETTI states inverter efficiency from 85% (AC180) to over 93% (AC200L). Victron lists a maximum of 92–95% for its Phoenix range. Those are best cases; at tiny loads the idle draw dominates instead.
Cold. Usable capacity drops in a cold garage. The model pages show a simulated 32°F (0°C) runtime next to the room-temperature one.

6. How to win runtime back

Use DC outputs for DC devices. A router with a 12 V barrel plug, a laptop on USB-C or a CPAP with the maker’s DC converter (ResMed lists one for the AirSense 11) can skip the inverter entirely. We count DC at 90% instead of 85%, and with the AC output off there’s no AC idle draw. See the Wi-Fi router and CPAP pages.
Turn the AC output off when nothing needs it. Charging phones from USB with AC switched on wastes the idle draw for hours.
Know your eco mode, both ways. These modes switch outputs off at low load to save power, which helps when you forget the AC button and can bite when a small load is the point. BLUETTI’s EB3A manual turns AC off after 4 hours at 10 W or less; its AC200MAX manual uses 50 W or less over 4 hours. Jackery’s Explorer 1000 v2 manual says its Energy Saving Mode is on by default and turns all outputs off after 12 hours when the AC load is 25 W or less, which a 15 W router meets. EcoFlow’s DELTA 3 manual warns that its AC timeout can trip on a fridge resting between cycles, and says to set it to “never” when the fridge must keep running. Anker’s C300 manual says the unit powers off after 2 hours with outputs off and no load, and the C2000 Gen 2 manual says USB outputs shut off when they draw under 1 W “for a long time”.
Add the idle draw to your own estimate. The runtime calculator has an inverter idle setting; 10–20 W is the range BLUETTI’s manuals give for its own models.

None of this changes the big picture for heavy loads. A microwave or a space heater swamps the idle draw. It’s the overnight, low-watt essentials where the gap between the label and reality is widest, and those are usually the ones you care about in an outage. For sizing a whole setup, see how to size a power station for an outage.

Questions

Why did my power station last half as long as the calculator said on a router?

The usual reason is the inverter’s own draw. On a 15 W load, 10–20 W of self-consumption, the range BLUETTI states for its models, roughly halves the runtime. Use a DC output if your router has one.

Do power station makers publish idle or no-load draw?

Rarely. Of the manuals we checked, BLUETTI’s state a self-consumption figure (about 10–20 W, by model). We didn’t find one in the EcoFlow, Jackery, Anker or Goal Zero manuals we read.

Is the simulated runtime on model pages a test?

No. It’s a computer simulation using assumed idle draw and losses. It’s there to show the direction and rough size of the gap, not to replace a measurement.

Why Your Power Station Runs Out Sooner Than the Math Says

1. The spec-sheet math

2. The inverter’s idle draw

3. Why small loads lose the most

4. What our simulation shows

5. The other losses

6. How to win runtime back

Questions

Sources

Keep reading

Calculators

Runtime calculator

What size do I need?

Solar charging calculator

Real-world runtime simulator

Generator vs power station

Compare models