The energy loss factor: why the meter bills more kWh than cars receive

A charging site has two kWh numbers, and they are never equal. The first is what the cars receive — the number on the customer's screen, the number revenue is billed on. The second is what the utility meters at the service entrance — the number the electricity bill is computed from. Between the two sit conversion losses in the power electronics, resistive losses in cabling, and everything the site draws that never reaches a vehicle.

The model calls the gap the energy loss factor, and it defaults to 12%: for every 100 kWh dispensed, the meter records about 114. That one input touches more of the cost stack than any other single assumption short of the tariff itself, and it works in one direction only — against the site.

The arithmetic: a divisor, not a subtraction#

The loss factor converts dispensed energy to purchased energy as a divisor:

kWh purchased = kWh dispensed / (1 − loss factor)

At the 12% default, the multiplier is 1 / 0.88 = 1.136 — the meter bills 13.6% more energy than the cars received, not 12%. The distinction matters because the factor is defined against the purchased side (that is how utility loss studies state it), and the divisor form compounds slightly against the site. A site that dispenses 1,000,000 kWh in a year purchases about 1,136,000.

Where it lands on the bill#

Every energy charge in the model is computed on purchased kWh. Under the model's PG&E BEV-2-S calibration, the time-of-use mix at the default 30/45/25 peak/off-peak/super-off-peak split blends to about $0.215 per purchased kWh. Grossed up by the loss multiplier, that is about $0.244 per dispensed kWh — the loss factor alone adds roughly 2.9 cents to every kWh the site actually sells.

At the million-dispensed-kWh year, the ~136,000 kWh gap costs about $29,000 — a line item that appears nowhere on the tariff sheet and on no quote, because it is not a rate. It is the difference between the quantity two different meters see.

Revenue, meanwhile, is billed on the dispensed side. So is the per-kWh site fee and any per-kWh charging excise a state levies on the sale. The asymmetry is the point: costs scale with the big number, revenue with the small one.

The quieter places it shows up#

The loss factor would be simple if it stopped at the energy line. It doesn't — two demand-charge structures in the tariff library consume it too:

• Load-factor-tiered rates (the Massachusetts EV-rate pattern) pick a demand rate from the site's annual load factor, and the load factor is computed from purchased kWh. A higher loss factor pushes the load factor up, which can move the site into a different tier.
• Capped-demand rates (the Florida GSLD-1EV pattern) bill demand as monthly kWh divided by a cap-hours figure once the cap binds — and the kWh in that division is again the purchased quantity, month by month.

None of this requires a decision from the reader; it is what the filed rate designs do when the meter reads 13.6% high relative to sales. The model applies the same purchased-kWh convention in all three places so the structures interact the way the bills would.

What moves the factor itself#

Twelve percent is a deliberately conservative default, not a measured constant. The realized figure at an operating site depends on:

• Conversion efficiency of the rectification hardware at the load points the site actually operates at — efficiency curves sag at low partial load, and early-ramp sites spend a lot of hours there.
• Cable runs between switchgear and stalls; resistive loss scales with distance and current.
• Auxiliary loads — cabinet cooling, lighting, network equipment — which the meter records regardless of whether a single car shows up. Because these are roughly fixed, the effective loss factor is worst at low utilization and improves as the site ramps.

The input is a slider, not a verdict. A site with short runs, efficient hardware, and strong utilization can defensibly model below 12%; a sprawling layout billed through a long service run may sit above it. The model's role is to show what each assumption does to the year — at the default calibration, each percentage point of loss factor moves the energy line by roughly 1.1%, compounding through margin, payback, NPV, and IRR from there.

The check worth doing#

The one-line takeaway for reading any charging pro-forma: find out which kWh column the energy costs are computed on. A model that prices electricity on dispensed kWh is quietly assuming a loss factor of zero — and at the volumes a multi-stall site runs, that assumption is worth about $29,000 a year at the default calibration. The cost-stack pillar walks the rest of the lines the same way: what the model computes, which quantity it computes it on, and what the default does to the total.