Time-of-use electricity and charging-site margins, hour by hour

A charging site's electricity bill has two halves: a demand half billed on peak kilowatts, and an energy half billed on kilowatt-hours. The demand half has its own explainer. This post is about the energy half — where the price of a kWh is not a number but a schedule, and the schedule moves by a factor of two to three over the course of an ordinary day.

What the filed spreads look like#

Three California EV rates in the tariff library show the pattern. Each prices energy in three windows — peak, off-peak, and super-off-peak — and the model carries each rate as a month-weighted blend of the utility's filed seasonal tables:

The shape is consistent even where the levels differ: a 4–9 p.m. peak window priced far above everything else, and a midday window priced below the overnight hours.

That midday dip is deliberate. California's grid is solar-heavy, so its cheapest energy now arrives in daylight — SCE's winter super-off-peak window runs 8 a.m. to 4 p.m., not overnight. For a charging site this is a favorable accident: the cheap window overlaps the hours a retail location has traffic anyway.

What the mix does to the blended rate#

The model reduces the schedule to one input: the share of the site's energy that falls in each window. At the default 30% peak / 45% off-peak / 25% super-off-peak mix, the PG&E calibration blends to about $0.215 per purchased kWh.

Because the windows are so far apart, small changes in that mix move real money. Shifting ten percentage points of load from peak to super-off-peak — 30/45/25 to 20/45/35 — drops the blend to about $0.191, a saving of 2.4 cents per purchased kWh. Run through the 12% energy loss factor and a million-dispensed-kWh year, that ten-point shift is worth about $27,000 a year — from the same cars charging at the same site, arriving at different hours.

The bounds are wider than the mix slider suggests. A kWh delivered at 6 p.m. under SCE TOU-EV-9 costs 3.2 times what the identical kWh costs at noon, through the same meter, on the same day. A site whose traffic skews evening and a site whose traffic skews midday can sit on the same tariff and see energy bills a third apart.

Where the mix comes from#

A host does not set the mix; drivers do. What the assumption stands in for is the site's traffic shape:

• Highway-corridor sites take demand when trips happen — which includes the evening peak window, like it or not.
• Retail and workplace sites skew daytime, which under a solar-shaped schedule is the cheap side of the day.
• Pricing feedback exists but is bounded: charging retail prices can vary by hour, and some traffic shifts with them, but a road-trip car at 20% battery at 6 p.m. is not waiting for the window to change.

The mix slider is therefore a statement about site selection as much as operations — the same lens the economics map post applies to territory-level costs, one level down.

Where TOU doesn't apply at all#

The three-window pattern is a filed-EV-rate feature, not a universal one. A large share of the 86 modeled schedules price energy flat — the TVA-standard EV rates in Tennessee, for example, bill one energy price at any hour, and the mix input does nothing there. On those rates the energy half of the bill is genuinely simple, and the action is all in the demand structure. The model carries both shapes so a flat-rate site and a three-window site can be compared on the same screen — and the comparison often turns less on the headline rates than on which hours a site's traffic actually keeps.

What a TOU schedule rewards is knowing the site's day before committing to the rate. The blend is arithmetic; the traffic shape is the input that decides it — and it is set by where the site is, long before the first bill arrives. The cost-stack pillar places the energy line among the rest of the year's costs.