How long will my battery last? Checking runtime on a 12 V LiFePO4 off-grid system you already own
Disclosure: The case study is my own system (I run MiniGridMap). It has 1,780 W of solar panels, a 60 A PowMr charge controller, four 314 Ah LiFePO4 cells in series (12.8 V, 4.02 kWh), a BMS rated 150 A, a 3,000 W pure sine inverter, a 300 A Class T fuse and a Honda EU2200i backup generator. It runs two laptops, a monitor, Starlink and TP-Link networking. Nothing in it logs production or use, so every number on this page is arithmetic on manufacturer-published specs, and anything I had to assume is labeled as an assumption. I'm not an electrician. Follow your own equipment's manuals, and get a qualified electrician for anything you aren't sure about.
I've lived off grid for 22 years, in a treehouse in a remote part of Hawaii Island. Parts are always failing, and troubleshooting is part of life. One question comes back every time something changes, whether it's a new part, a cloudy week, or a battery getting older: how long will what I have actually last? This isn't about designing a new system. It's about checking the one already on the wall, using the manuals. These are the questions I asked and what the published numbers say. If you see a mistake, I'd like to hear about it.
Key findings (arithmetic, not measured)
Start from usable energy, not the label. Four 314 Ah cells at 12.8 V hold 4.02 kWh on paper. Planning on 80% of that leaves about 3.22 kWh. EVE rates these cells for 8,000 cycles to 70% state of health, so an older bank may give closer to 2.25 kWh.
The inverter costs energy even with nothing plugged in. Two published 3,000 W, 12 V inverters list 20 W (Victron) and under 2.5 A, up to about 32 W at 12.8 V (Renogy) with no load. Over 24 hours that's 0.48–0.77 kWh, or 15–24% of the usable bank, before any load runs.
Which Starlink kit you have changes the answer more than anything else on my list. Starlink publishes average draw of 25–40 W for the Mini and 75–100 W for the Standard kit. Run all day through a 90%-efficient inverter, that's 0.67–1.07 kWh for the Mini against 2.00–2.67 kWh for the Standard kit.
A no-sun day can take more than the whole bank. With a Standard kit, inverter idle, networking and eight hours of laptops and a monitor add up to 3.50–4.46 kWh a day, against 3.22 kWh usable. With a Mini it's 2.17–2.86 kWh. (The laptop, monitor and networking figures are illustrative assumptions.)
The inverter's low-battery alarm isn't a fuel gauge for LiFePO4. Renogy's datasheet puts its alarm at 11.0 V and shutdown at 10.5 V. On a four-cell bank that's 2.75 V and 2.63 V per cell, both below the 2.8 V EVE says to stay above in normal use.
How much energy is really in the bank?
Battery side (case-study bank: 4 x EVE MB31 314 Ah cells in series)
4,019 Wh x 0.7 x 0.8; EVE rates 8,000 cycles to 70% SOH at 25 °C
Everyday floor
2.8 V per cell (11.2 V)
EVE: "charge the cell regularly to keep the voltage above 2.8 V"
Absolute end of discharge
2.5 V per cell (10.0 V)
EVE, above 0 °C; over-discharge is "forbidden"
Self-discharge
about 3% a month
EVE, after the first month, at 25 °C
Why 80% and not 100%? EVE's nameplate capacity is measured on fresh cells at 25 °C, all the way from 3.65 V down to 2.5 V. Day to day, the BMS and charger settings stop short of both ends, and cells lose capacity as they age. 80% is a planning number, not a promise. Your BMS's settings decide where your real floor is.
What does each load cost per day?
Here's the part I could check against published specs and the part I couldn't. Starlink and the two inverters below publish their figures. My laptops, monitor and TP-Link gear don't come with a daily-use number, so I've used illustrative round figures (30 W per laptop, 25 W for the monitor, 10 W for networking). They aren't measurements. The label on a laptop charger is its maximum, not what it draws on average, so it will overstate.
Everything in my case runs as AC through the inverter, so each AC figure is divided by an illustrative 90% efficiency. Renogy lists ">90%" and Victron lists 93% maximum. Those are maximums, and at light loads efficiency may be lower.
Chart 1: estimated energy per day for each load, against the bank's usable energy. Dark bars are the low end and light bars run to the high end of the published range. Items marked * are illustrative assumptions.
Daily energy by load (from the battery)
Load
kWh per day
Share of 3.22 kWh usable
Inverter idle, 24 h
0.48–0.77
15–24%
Starlink Mini, 24 h
0.67–1.07
21–33%
Starlink Standard, 24 h
2.00–2.67
62–83%
Networking 10 W, 24 h*
0.27
8%
Laptops + monitor 85 W, 8 h*
0.76
23%
* Illustrative, not measured. Starlink figures are Starlink's published averages; real use varies. Inverter idle is the published no-load figure, added on top of the load. That's an approximation that may slightly overstate it.
So how many hours will it run?
Hours = usable watt-hours / watts drawn from the battery. Watts from the battery = AC load / 0.90 + inverter idle (20 W low end, 32 W high end).
Estimated hours from a full charge (arithmetic, not measured)
Both scenarios include networking at an illustrative 10 W. "Working" adds two laptops at 30 W and a monitor at 25 W, also illustrative. Ranges pair the low end of every figure together and the high end together.
For me the useful question is whether the bank gets through the night and a cloudy morning. So the overnight row matters more to me than the working one. The solar guide covers how much the panels put back, and the generator guide covers what to do when they don't.
How do I check the arithmetic against my own system?
The table above is a starting point. Your real number comes from your own system:
Read the current, not just the voltage. If your BMS or battery monitor shows battery current (many smart BMSs do, through an app), note it at night with your normal loads on and the panels idle. Amps x 12.8 V is roughly the watts you're drawing. That one reading tells you more about runtime than any table.
Read it once with everything off but the inverter. That's your inverter's real idle draw, and you can compare it with its manual.
Look up your actual model numbers. Starlink kit, inverter, BMS. The spread between the Mini and Standard kits, and between the two inverters, is wider than most of the assumptions on this page.
Remember the bank gets older. If runtime seems shorter than it used to be, a tired bank and a new load are both suspects. Your cell maker's datasheet has the cycle-life rating.
I don't log my system, so I can't give you my own nightly figure. If you've measured these loads on a similar setup, I'd like to learn from it.
Could some loads skip the inverter?
Maybe. Starlink lists the Mini's input as 12–48 V DC, 60 W, so on paper it can run from DC without an inverter. If nothing else needs AC overnight, the inverter's idle draw goes away too. But it's a wiring change, so the right place to work it out is Starlink's own documentation and accessories, your BMS and fuse ratings, or an electrician. I'm not describing how to do it here.
How much can I run at once?
Runtime is about energy. There's also a power limit, and the lowest rating in the chain sets it. In my case:
Continuous power limits in the case-study chain (arithmetic on ratings)
Part
Published or owner-provided rating
About
Inverter
3,000 W continuous
about 260 A from the battery at 12.8 V and 90% efficiency
EVE MB31 cells
0.5P max continuous discharge at 25 °C
502.4 W x 4 = about 2,010 W
BMS
150 A (owner-provided model rating)
150 A x 12.8 V = 1,920 W, less as voltage falls
Class T fuse
300 A
Protects the wiring against a fault. It isn't a load limit.
So on paper my inverter can ask for more than my BMS and cells are rated to give. My everyday loads are far below that, but if you have a big inverter on a smaller battery, it's worth asking your BMS maker what happens at its limit and keeping big loads well under it.
Safety first
Let the BMS and charger settings protect the cells, not the inverter alarm. EVE says over-discharge is forbidden and recommends keeping cells above 2.8 V in normal use. If the bank does hit its cutoff, EVE says to charge it as soon as possible.
Cold matters on the mainland. EVE allows charging only from 0 to 60 °C. A cabin bank that's below freezing at dawn shouldn't be charged until it warms up, whatever the runtime table says.
Stay inside every rating in the chain (inverter, BMS, cells, fuse and cable), and ask the maker when two of them disagree.
If you change wiring to save power, such as moving a load to DC, have it checked by a qualified electrician.
What I don't know
My real nightly draw, my inverter's measured idle draw, and what my laptops and networking actually use. Nothing logs them.
My bank's current state of health.
Real inverter efficiency at the light loads in these tables. Published efficiency is a maximum.
Methodology
Every figure is arithmetic on manufacturer-published specifications (sources below), accessed 2026-10-09, plus the illustrative assumptions labeled on the page. Nothing is measured.
Usable energy = 4,019.2 Wh x 0.8 = 3,215 Wh. Aged bank = 4,019.2 Wh x 0.7 x 0.8 = 2,251 Wh.
Battery draw = (AC load) / 0.90 + inverter idle. Idle low end is 20 W (Victron Phoenix 12/3000, 120 V). High end is 2.5 A x 12.8 V = 32 W (Renogy 3000 W 12 V, listed as "<2.5A").
Illustrative loads: laptops 30 W each, monitor 25 W, TP-Link networking 10 W.
Hours = usable Wh / battery draw. kWh per day = watts x hours of use / 1000.
Sources
EVE Energy, MB31 314 Ah LiFePO4 cell product specification (314 Ah, 1,004.8 Wh, 3.2 V nominal, 2.5 V end of discharge above 0 °C, 0.5P max continuous discharge at 25 °C, 8,000 cycles to 70% SOH, about 3%/month self-discharge, keep above 2.8 V in normal use, charging 0 to 60 °C; copy hosted by a distributor): https://www.solardirekt24.de/media/4b/bf/6c/1762180019/EVE_MB31_314Ah_Datasheet.pdf (accessed 2026-10-07)
Renogy, 12 V pure sine wave inverter datasheet (3000 W: no-load current draw <2.5 A; efficiency >90%; battery low alarm 11 V ±0.3 V; battery low shutdown 10.5 V ±0.3 V): https://www.renogy.com/cdn/shop/files/INVT-P2-Datasheet.pdf (accessed 2026-10-09). Cited as an example; it isn't the case-study inverter.
Cite this page: MiniGridMap, "How long will my battery last? Checking runtime on a 12 V LiFePO4 off-grid system you already own", https://minigridmap.com/guides/how-long-will-my-battery-last-12v-lifepo4/ (2026). Chart and tables may be reused with a link back.
If you'd like this arithmetic kept next to your own equipment list, GRID COMMAND's Overview estimates hours left from the loads you record. Every screen is free to try with no account, and saving your system is $5 a month. Open the app.
Case study is the site owner's own system. Every figure here is arithmetic on published specs or a labeled assumption, not measured.