What this simulator is actually showing you

The flat plateau is the whole problem with LiFePO4

Drag the state of charge slider and watch the pack voltage. Between roughly 20% and 90% a LiFePO4 cell moves about 60 millivolts — so a 16-cell pack moves about one volt across 70% of its capacity. This is why voltage-based fuel gauges lie about LFP, why your pack seems to sit at "half full" for hours and then falls off a cliff, and why a shunt-based coulomb counter is not optional equipment.

Your pack is a very large welder

A 16S pack of 280Ah prismatic cells has on the order of four milliohms of internal resistance. At 51 volts that is a prospective short-circuit current in the region of ten thousand amps — not for a moment, but until something in the path vaporises. NEC 110.9 requires every overcurrent device to be capable of interrupting the current available at its terminals. A Class T fuse is rated 20,000A. An ANL fuse is typically 2,000–6,000A. A great many DC breakers sold for this purpose publish no interrupting rating at all, which means they cannot be evaluated against 110.9 in either direction.

A device that fails to interrupt does not stop the fault. It becomes part of it.

One weak cell sets the capacity of the whole pack

Turn up the cell spread and charge the pack. The smallest-capacity cell arrives at 3.65V first, the BMS latches overvoltage, and charging stops — while every other cell is still short of full. Passive balancers bleed at around 60 milliamps, so recovering a few amp-hours of drift takes days, and no amount of top balancing fixes imbalance at the bottom of the charge. Pack state of charge in this tool is always the worst cell, never the average, because the worst cell is what the BMS acts on.

58.4 volts is not 48 volts

Sixteen cells at 3.65V is 58.4V at full charge. Plenty of inverters sold as "48V" accept 58V or 60V maximum. The mismatch never appears at nominal voltage — it appears the first time the pack completes an absorption charge, and exceeding a listed input rating is an NEC 110.3(B) problem as well as an expensive one.

Cold weather stops the charge, not the discharge

Below 0°C, charging LiFePO4 plates lithium onto the anode and the damage is permanent, so the BMS locks charging out. The pack will still run your loads; it simply will not accept a charge on the mornings you most want it to. Internal resistance also roughly doubles at freezing, so it sags harder under load at the same time.

And the part nobody wants to hear

A pack you assembled from individual cells is not a listed energy storage system. NEC 706.5 requires an ESS to be listed — UL 9540 covers the system as installed, UL 1973 covers the battery. Correct conductors, a Class T fuse and a beautiful build do not change that, and most authorities having jurisdiction will not permit an unlisted ESS on a grid-interactive dwelling. Unlisted batteries are also a common homeowner-insurance exclusion after a fire.

That does not make the pack useless — it makes it an off-grid, non-dwelling pack: a shop, a shed, a pump, a trailer. And if you want the same chemistry at the same voltage with the paperwork, a listed 48V rack battery is the identical 16 cells in a box someone else certified. Switch the cell type in the simulator to see the difference the listing makes to the findings, and none at all to the physics.

Design aid, not a licence

Everything here is a first-order model checked against the code sections it cites. A licensed electrician or professional engineer must review, connect and inspect before anything is energized.