Why your 30 minute charge took an hour.

The short version

If you read nothing else:

Why you can't just divide

Say you drive a Lexus RZ 450e. It holds 64 kWh, you're going from 10% to 80%, so you need about 44.8 kWh. You pull into a 150 kW stall. That's roughly 18 minutes of charging, right?

It's 31 minutes on a mild day. If it's 30°F out, it's 56 minutes.

The problem is that 150 kW is a ceiling, not a speed. Your car touches that number briefly while the battery is nearly empty, then backs off, and keeps backing off for the rest of the session. By 80% the RZ is pulling around 28 kW. Dividing capacity by charger rating quietly assumes you got the peak the whole time, which is why every tool that does it tells you a number you never see.

The model integrates reciprocal power within each one-percent interval, splitting at curve breakpoints and charger limits. For a linear segment, time is proportional to ln(end power / start power) divided by the power change. Reported power is the energy/time average; battery health scales the energy in each interval.

Each vehicle carries its own list of measured breakpoints (at 10% it pulls this much, at 40% this much) and the model reads power straight off that list, interpolating in a straight line between the points. Real curves step like that, since a BMS derate is an on/off decision rather than a smooth slope, so a handful of honest breakpoints beats a polished curve fit that implies precision nobody measured.

Cold batteries charge slowly

Cold electrolyte is thicker and the cells fight the current harder. Push full power into a cold pack anyway and you start plating metallic lithium onto the anode, which permanently eats capacity and, in the bad cases, eventually shorts the cell. No manufacturer is willing to risk that, so the BMS limits current until the pack warms up.

The model treats that as a ceiling on power, set by the pack's own temperature: a share of the car's peak, from a floor at freezing to no limit at 68°F. With the default floor:

Two things follow that a flat multiplier missed. First, the ceiling only bites where the car would otherwise draw more: past 80% an RZ takes under 30 kW cold or warm, so the cold costs nothing there. Second, charging warms the pack. About 3.5% of the energy going in becomes heat in the cells, and a pack holds roughly 1 Wh per °F for each kWh of capacity, so a frozen RZ starting at 30°F ends a 10 to 80% session at about 53°F, and the ceiling lifts as it goes.

The RZ's floor is its own: 22% of peak, about 31 kW. That is what it takes for the model to match the published cold benchmark, 10 to 80% at 30°F in 50 to 60 minutes. It gives 55.8 minutes, 35 of them held down by the cold. The same session on a warmed pack takes 32.8.

The rate in each 1% slice is the lowest of four ceilings:

The second term is why "150 kW" means less than it says. A stall has a current limit as well as a power rating; the model takes it as the rating reached at 400 V unless you enter the amps on the label. A 400 V pack sits between about 320 and 400 V, so an RZ 300e on a 150 kW stall peaks at 132 kW, not its own 143. An 800 V Ioniq 5 on the same stall gets the full 150. The session chart colours every slice by which ceiling set it.

Selecting a warm pack starts the session at 77°F. That is an assumption, not a measurement: the model does not know how long your preconditioning ran.

When to walk away

Every minute at a charger buys range, but the later minutes buy less. The useful question on a road trip is not "is this still fast?" but "would the trip be quicker if I left now and stopped again later?" That has an exact answer, the one foragers and economists know as the marginal value theorem.

A stop costs a fixed overhead (pull off, park, plug in, pay) plus charging time, and returns miles. Over a long trip, total time is least when each stop ends where its average miles per minute, overhead included, peaks. Up to that point the next percent adds miles faster than the stop has averaged; after it, slower.

Here is the RZ 450e from 10% on a 150 kW stall on a mild day, with a 10 minute overhead, at its 70 mph efficiency of 2.35 miles per kWh:

The average peaks at 50%, the cliff itself: 12.6 minutes of charging against 32.8 to reach 80%. Until October 2026 this page used a rule of thumb (under a mile a minute, or 60% off peak) that told RZ drivers to sit through the cliff to 75%.

The overhead matters, and not smoothly. For the same session:

Past about 12 minutes of overhead the long plateau after the cliff becomes worth charging through, and the answer jumps to the second cliff at 75%. On a charger under 25 kW there is no answer at all: that's AC charging, where the charger sets the pace from start to finish.

The theorem assumes the next charger is within reach and no worse than this one. If it isn't, charge for the distance; the model does not plan a route.

Speed, cold, and the heater

The EPA test cycle averages under 50 mph with the climate control mostly idle, which is not what an interstate looks like. The model builds highway consumption from the forces involved, for each car.

Road load

Three things take energy at a steady speed on level ground: rolling the tyres (flat with speed), pushing air aside (rising with the square of speed), and the electrics that run whether you move or not (spread over fewer hours the faster you go).

Mass, drag coefficient and frontal area are each car's own. So is the share of its load that is drag, which is why speed costs some cars much more than others:

The one fitted number is k. A car with a published road test gets the k that reproduces it; every other car gets the fleet's typical k. How far that lands from the truth is measured below, not assumed.

Cold air, thin air

Air density ρ comes from the ideal gas law and the barometric formula: 0°F air is 15% denser than 70°F air, and Denver's is 17% thinner than the coast's. On top of that, cold cells, stiff tyres and thick lubricants cost up to 20% more, a separate penalty of 3% per 10°F below 65°F.

The heater

A resistance cabin heater pulls two to three kW, and unlike a gas car you don't get that heat for free off the engine. But it isn't a fixed load either: it runs flat out in a hard freeze, cycles at part duty on a cool day, and is off on a mild one. So the model scales the rated wattage by a duty fraction before spreading it over distance:

Air conditioning tops out at about a third of the heater's draw because a compressor moves heat rather than making it. Between 65°F and 75°F nothing runs. The heater costs less per mile the faster you drive, since it draws by the hour while you cover ground by the mile; drag wins that argument comfortably.

All of it at once

Same RZ 450e, same battery, four different days. The safe window is 80% down to 10%, since nobody road trips to zero:

141 miles down to 73 between the best and worst row, on a car that never changed. That spread is what strands people who planned around the window sticker.

Older batteries

You don't need an OBD-II dongle for a decent guess at pack health. Two things wear a battery down: time, as lithium gets locked into a layer that grows on the anode, and use, which the odometer stands in for. The layer grows ever more slowly, so calendar fade goes with the square root of age: a fast first year, then a slow decade. That is the shape the large fleet surveys show (Geotab: about 1.8% a year on average; Recurrent: most packs above 90% past 100,000 miles).

A worn pack gets hit twice. It holds less, and its internal resistance has grown, roughly twice as fast as its capacity has faded, so the BMS lets in less power to hold heat and cell voltage down:

Paying by the minute

Some networks bill by time rather than energy. If your car decides how fast it accepts power and the meter charges by the minute regardless, then a slow car is paying for the privilege of being slow. The rate-trap calculator bills the simulated session one percent at a time, so you can see where.

Here are real cars on the same plan at 40 cents a minute, with gas at $3.30 and a 30 mpg gas car to compare against:

The Bolt paid four times as much per kWh as the Ioniq 5 and went home with less energy. The last row works out to the cost of running a car that gets 12 miles per gallon.

The verdict compares like with like: what a mile costs on this charge against what it costs in the gas car. Under 60% of the gas car's cost per mile is fair, up to parity is elevated, and past parity driving on that charge costs more than driving on gasoline. (Until October 2026 it used fixed $/kWh bands, which called the same session fair in one state and a trap in the next whatever gas cost there.) Prices are per kWh that reached the battery: a DC stall meters about 5% more than the pack keeps.

What this gets wrong

All of the above is a model, and a model is a pile of decisions someone made. Here are the ones I'd push back on if I were reading this page instead of writing it.

How I check these numbers

These charging benchmarks are wired into the test suite. Change a curve or a formula in a way that pushes any of them outside its published range and the build fails.

The highway model is checked differently, because a car calibrated to its own road test will always match that test. Until October 2026 this page listed fifteen of them "within 1%", which proved nothing. Now each of the 16 published road tests is predicted from the car's mass, drag and frontal area with the fleet factor fitted to the other fifteen, never its own:

Median error 9%, mean 12%, 8 of 16 within 10%. That is the error to expect for the 26 cars with no road test, and their pages say so. The test suite pins these bounds and fails if the published file drifts from the engine.

The other 38 charge curves get a different safeguard. Each one records the date it was last checked against real world data, and the build fails if any of them goes 18 months without a look. Manufacturers push charging software updates that change these curves.