EV Charger Load Management Retrofit

Purpose

The retrofit call arrives after the house changed. A second vehicle, a heat pump, a storage system, or a customer who wants the charger to run at the rate the dealership quoted. The charger that passed its calculation three years ago now sits inside a different building, and the honest answer is not a setting, it is a re-derivation.

The trap here is the demand data. Fresh interval data already contains the existing charger, because it was charging during the month you measured. Add that charger back in on top and you have charged the service twice for the same load, in the direction that sells a service upgrade nobody needed.

Scope

Covers retrofitting load management onto a premises with an existing, working EV charging installation whose governing calculation no longer holds, at a dwelling or small commercial building, 120/240 V single phase.

Does not cover a first EV charger installation and its original calculation, owned by the EV charger install and load calculation SOP; the mechanics of installing and verifying the management device itself, owned by the load management device SOP, which step 4 calls rather than repeats; or a service upgrade, which is its own job when this procedure concludes one is needed.

Roles and responsibilities

Procedure

1. Retrieve the original calculation and establish what actually changed. Read the original load basis, its date, the EVSE current it assumed and the headroom it left, then list every load added or removed since. Acceptance: original calculation in hand with its headroom in amps, a dated list of changes, and the adopted NEC edition recorded, since the 2023 edition added 220.57 setting the EVSE load at 7,200 VA or nameplate whichever is larger, while earlier editions carry no EVSE demand factor and are worked at 125 percent of nameplate. Wrong looks like starting a fresh calculation without reading the old one, which loses the reason a past technician made a choice; stop rule, no original calculation on file means one gets rebuilt from scratch and said so plainly on the new sheet. Hazard: none, a desk step.

2. Pull fresh demand data and separate the existing charger's contribution from it before any arithmetic. Where a year of maximum demand is unavailable, NEC 220.87 permits 30 consecutive days of 15-minute interval demand; then use the charger's own energy log, time-aligned to the interval peak, to establish what it was drawing during that interval. Acceptance: the interval peak in amps, the charger's measured contribution during that same interval in amps, and the non-charger demand shown as the subtraction of the second from the first. Wrong looks like adding the existing charger to a peak that already contains it, which double-charges the service and manufactures a shortfall; stop rule, where the charger's contribution during the peak interval cannot be established, the unmodified interval peak is used and the EVSE is not added twice, which is conservative and is said to be conservative on the sheet. Hazard: none in the arithmetic, but a spot ammeter reading on one afternoon is not demand data under 220.87 and does not substitute for it.

3. Confirm the existing charger can be controlled the way the calculation will rely on. There are two different things called load management: reducing the current the charger offers the vehicle through the pilot signal, and opening the charger's circuit with a contactor. Acceptance: the existing unit's listing and its documented control capability recorded, along with whether the vehicle resumes a session unattended after a circuit interruption. Wrong looks like a calculation resting on an owner-adjustable app setting, which is behavior rather than an engineering control; stop rule, a charger that can only be interrupted, on a vehicle that will not auto-resume, is not a managed load, and the routes are a communicating unit, a permanently reduced fixed set current, or a service upgrade. Hazard: none while reading documentation, though verifying resume behavior means running the vehicle, which belongs in step 6.

4. Install and verify the management device to its own procedure, then set the charger's maximum. Follow the load management device SOP for current transformer placement, configuration and the shed demonstration, and separately set the EVSE's maximum current to the figure the re-run calculation will use, as NEC 625.42 permits where an energy management system per Article 750 limits the load. Acceptance: the device verification log complete per that SOP, and the EVSE's set current confirmed by reading it back at the unit rather than at the app. Wrong looks like a set current entered in an installer menu and never read back; stop rule, a value that will not persist through a power cycle is not a control, and the unit is power-cycled once to prove it. Hazard: any work inside the panel or at the charger's supply means open the main, lock and tag, and prove every conductor dead on a known live source immediately before and after per NFPA 70E-2021, 120.5, with work practices at 29 CFR 1910.333(b)(2), which governs because 1910.147 excludes electrical utilization work at (a)(1)(ii)(C).

5. Re-run the whole calculation with the control as the input, line by line. Take the non-charger demand from step 2, apply 125 percent, add the managed EVSE maximum at 125 percent, and compare the total to the service overcurrent device rating. Acceptance: every input on its own line with its multiplier, the total, and the headroom in amps, with the result stated as pass or fail against the service rating. Wrong looks like adjusting the total to make it fit rather than re-running it with the new input; stop rule, a total within a couple of amps of the service rating is a fail, not a pass, because the inputs carry more uncertainty than that. Hazard: none in the arithmetic, but a service pushed past its rating trips on a January evening, not during your test.

6. Prove the throttle on site with the vehicle actually charging and the house actually loaded. Bring the vehicle to a low state of charge so it pulls full rate, then add real premises load until the device acts. Acceptance: EVSE current measured before and after the device acts, total service current recorded at both points, the reduced EVSE current at or below the managed maximum, and the session confirmed to continue or to resume without anyone touching the vehicle. Wrong looks like a handover with the vehicle near full, drawing a fraction of rated current and proving nothing; stop rule, a session that stops and does not resume sends the job back to step 3, whatever the settings screen says. Hazard: this step drives a live service toward its limit with the family home, so no value is forced into the device and no sensing is jumpered, current is read with a category-rated clamp on the load side rather than with a hand in an open enclosure, and terminations are checked by infrared after at least 30 minutes at the reduced rate.

7. Label the dependency and hand over what a future occupant needs. Label the panel with the managed EVSE circuits, the set current and the sentence that the load calculation depends on the control remaining in service, and give the customer the new headroom figure and what a larger charger would require. Acceptance: labels installed and photographed, the customer's signed acknowledgment, and the superseded calculation marked superseded rather than discarded. Wrong looks like a new sheet filed beside the old one with nothing saying which governs; stop rule, no signature, no close-out. Hazard: none, and the label is what survives the sale of the house.

The record this produces

One superseding calculation sheet plus one verification log.

The sheet: the original calculation referenced by date with its headroom, the dated change list, the adopted NEC edition, the interval data range and peak, the charger's contribution during that interval and the non-charger demand as a shown subtraction, then every line of the re-run with its multiplier, the total and the headroom in amps. The old sheet is stamped superseded.

The log: EVSE current and total service current before and after the device acts, the reduced EVSE current against its managed maximum, the resume observation, and the infrared check.

The next technician reads the headroom line and the dependency label together, because the headroom is only real while the control is in place.

Worked pass: 200 A service, second charger requested

An existing 32 A charger on a 40 A circuit, installed three years ago. The customer has bought a second vehicle and wants a second charger of the same size.

Step 1: the original calculation left 32.5 A of headroom. Since then the house gained a heat pump water heater. The jurisdiction's adopted edition predates 220.57, so EVSE load is worked at 125 percent of nameplate.

Step 2: the fresh 30-day, 15-minute interval peak is 128 A. The charger's own energy log shows it drawing 31.6 A through that same interval, so the non-charger demand is 128 minus 31.6, or 96.4 A. Skipping that subtraction and adding a 40 A charger to 128 A would have produced 160 plus 40, or 200 A against a 200 A service, and sold a service upgrade on a number that counted the same charger twice.

Step 3 fails. The existing unit is an older model with no communication input, so the only control available to it is a contactor that opens its circuit, and the vehicle's own behavior after an interruption is a plug-out and plug-in to resume. Overnight charging would silently fail. The stop rule fires and the three routes go to the customer: replace the older unit with a communicating one, fix both chargers at a permanently reduced set current, or upgrade the service. They choose replacement, because the reduced fixed current would have limited both vehicles every night rather than only during the rare peak.

Step 5, re-run with the control as the input: non-charger demand 96.4 A at 125 percent is 120.5 A. The energy management system is set to a combined EVSE maximum of 48 A across both units, which at 125 percent is 60 A. The total is 120.5 plus 60, or 180.5 A against the 200 A service, leaving 19.5 A of headroom, which is 9.8 percent of the service rating. Pass.

Step 6: vehicle brought to a low state of charge. With one vehicle charging, EVSE current measures 31.4 A and total service current 108 A. The second vehicle is plugged in and the oven and dryer are switched on; total service current climbs and the management system reduces the combined charging current, measured as 15.8 A and 15.7 A at the two units, or 31.5 A combined, which is below the 48 A managed maximum. Both sessions continue with nobody touching either vehicle. After 35 minutes at the reduced rate, infrared at both units' terminations shows no rise above neighbouring terminations.

Step 7: panel labeled with both charger circuits, the 48 A combined maximum and the dependency sentence. The customer signs, and the old calculation is stamped superseded.

References

• NEC (NFPA 70) Article 625, including 625.41 for the continuous-load overcurrent rating and 625.42 for the rating where an energy management system limits the load, with Article 750 for the system itself; confirm the adopted edition
• NEC (NFPA 70) 220.87 for existing loads from maximum demand data, and 220.57 in the 2023 edition for the EVSE load; the two produce different totals, so record which edition governs
• 29 CFR 1910.333(b)(2) for de-energizing and verifying at step 4, with NFPA 70E-2021, 120.5 for live-dead-live
• Manufacturer's documentation for both chargers: control capability, set-current persistence, and the vehicle's resume behavior after an interruption
• See related: the EV charger install and load calculation SOP, and the load management device installation SOP, which owns step 4