NEC 220.87 for EV Chargers: The 12-Month Utility Demand Method That Beats a Panel Upgrade Quote (2026)

Most homeowners who get a panel-upgrade quote alongside an EV charger install never hear NEC 220.87 mentioned. The quote runs the 220.82 Optional Method (or, in some cases, no calc at all and a rule-of-thumb estimate), lands close to the service rating, and the homeowner gets a $4,000 service upgrade attached to a $1,500 charger install. The 220.87 path is the existing-loads rule the NEC wrote for exactly this situation: when a dwelling has been lived in long enough to have real utility demand data, the calc gets to use the real data instead of the conservative estimate.

The Rule in Plain English: NEC 220.87 Verbatim

NEC 220.87 (2017 edition) reads: “The calculation of a feeder or service load for existing installations shall be permitted to use actual maximum demand to determine the existing load under all of the following conditions: (1) The maximum demand data is available for a 1-year period.”

The exception, verbatim: “If the maximum demand data for a 1-year period is not available, the calculated load shall be permitted to be based on the maximum demand (the highest average kilowatts reached and maintained for a 15-minute interval) continuously recorded over a minimum 30-day period using a recording ammeter or power meter connected to the highest loaded phase of the feeder or service, based on the initial loading at the start of the recording. The recording shall reflect the maximum demand of the feeder or service by being taken when the building or space is occupied and shall include by measurement or calculation the larger of the heating or cooling equipment load, and other loads that may be periodic in nature due to seasonal or similar conditions.”

The second condition, paraphrased from the NEC 2017 text continuing past the cutoff, requires that the existing maximum demand at 125 percent plus the new load not exceed the ampacity of the feeder or the rating of the service. That is the math line. Existing demand times 1.25, plus new EV load, compared to the service rating.

In one sentence, NEC 220.87 says: if there is real demand data for an occupied dwelling, use it, multiply it by 1.25, add the EV, and check it against the service. The 125 percent multiplier is the continuous- load margin the code already uses everywhere else; applying it to the historical peak gives the same conservative cushion that a freshly calculated 220.82 bakes into the appliance demand factors.

The Math: A Worked Example

A 1,900 sqft 1995-built ranch on a 200A service, gas heat, central AC, electric range, electric dryer, electric water heater. Two adults, one teenager, a Ford Mustang Mach-E about to land in the driveway with an 11.5 kW / 48A continuous Ford Connected Charge Station. The Qmerit-network electrician ran the 220.82 calc and landed at 186A (close to the 160A safe-capacity ceiling on a 200A service at 80 percent), then quoted the charger plus a $4,200 service upgrade.

Pulled the homeowner's 12-month utility history. Peak demand over the prior twelve months was 14.8 kW, recorded in July during the air-conditioning peak with the dryer and electric range running.

• Existing peak demand: 14.8 kW = 14,800 VA. At 240V single phase, 14,800 ÷ 240 = 61.7A actual recorded peak.
• Existing demand at 125 percent (NEC 220.87 condition 2): 14,800 × 1.25 = 18,500 VA = 77.1A.
• New EV load (NEC 625.41 continuous, 125 percent): 48A × 240V = 11,520 VA; 11,520 × 1.25 = 14,400 VA = 60A.
• Total under NEC 220.87: 77.1A + 60A = 137.1A.
• Service rating: 200A.
• Headroom: 62.9A available.

The same house that landed at 186A under the appliance-by-appliance 220.82 Optional Method lands at 137.1A under the 220.87 actual-demand method. The $4,200 service upgrade was a number from a conservative calc on an over-estimated load profile, not from the real load profile the meter has been recording for a year. The Mach-E install is a $1,500 clean job under 220.87.

Run the same calc on your own panel for $12.99. The assessment returns the NEC 220.82 number first, then adds the NEC 220.87 cross-check when the utility-demand field is filled in.

Why 220.87 Almost Always Shows More Headroom Than 220.82

The 220.82 Optional Method is a conservative estimator. It assumes the electric range is operating, the electric water heater is heating, the dryer is running, and the air conditioner is at peak, then applies the first-10-kVA-at-100-percent and remainder-at-40-percent demand factor. The math is intentionally pessimistic because the calc has no idea when the homeowner actually runs those loads.

The 220.87 actual-demand method does know. The utility meter has been recording 15-minute interval averages for a year, and the worst 15-minute interval is the peak. In almost every occupied dwelling, the real peak is lower than the calculated estimate for three reasons:

• Diversity. No occupied house runs every appliance simultaneously. The dryer, the range, the AC compressor, and the water heater do not all hit their peak in the same 15-minute window. 220.82 partially accounts for diversity with the 40 percent remainder factor; 220.87 measures the diversity directly.
• Behavior. Households schedule heavy loads to avoid each other without thinking about it. The dishwasher runs after dinner, the dryer runs in the morning, the AC is at its peak in the afternoon. The utility meter sees the actual schedule; the calc assumes simultaneous worst-case.
• Sub-nameplate operation. A 4-ton AC compressor rated at 5,000 watts continuous rarely draws that nameplate in a steady state; modern variable-speed compressors spend most of the cycle at 40 to 60 percent of nameplate. The calc uses the nameplate; the meter sees the actual.

The typical result on a 200A service: a 220.82 calc that lands at 160A to 190A often shows 60A to 90A of real recorded peak under 220.87. That is the gap that creates the $4,000 panel-upgrade quote on a house that does not need one.

The 30-Day Recording Exception: When There Is No 12-Month History

New construction, recent move-ins, and any dwelling that does not have a full year of utility data is not shut out of 220.87. The exception lets a licensed electrician install a recording ammeter or power meter on the service for a minimum 30-day period and use the highest 15-minute average as the existing peak demand. Three conditions apply.

• Occupied. The recording has to be taken when the building is occupied, which keeps an empty-house reading from understating the demand.
• Heating or cooling included. The larger of the heating or cooling equipment load has to be included by measurement or calculation. A 30-day recording in May does not capture the July AC peak; the electrician calculates that load and adds it to the recorded number.
• Seasonal loads included. Pool pumps, electric heat strips, and other loads that are periodic in nature have to be included by measurement or calculation, same rule as the heating/cooling adder.

A 30-day recording costs $150 to $400 in most markets, including the meter rental, the install, the disconnect, and the data summary. For a borderline house that would otherwise face a $4,000 service upgrade, the recording is the cheapest possible tiebreaker.

When 220.87 Does Not Work: The Three Cases for 220.82

Honest framing on when the Optional Method is still the right answer.

• New construction or recent move-in with no recording option. 220.87 requires real data, either 12 months of utility history or a 30-day recording. A house that has been lived in for three months with no recording available cannot meet either condition. The 220.82 calc is the path. The NEC 220.82 walkthrough walks the calc line by line.
• Major load broken or disconnected during the recording window. If the heat pump compressor failed in March and the air handler ran on emergency heat strips for a month, the utility data understates the real load. Same problem if the electric water heater was off for two months during a remodel. 220.87 can pass an EV that 220.82 would fail, and a defensible quote runs both numbers and notes the disconnect.
• AHJ requires 220.82 or the standard method. Some jurisdictions default to a specific calc method on residential permits. NEC 220.87 is permitted but not mandatory. A defensible package includes both calcs and lets the inspector pick.

What the Honest Quote Looks Like

A 2026 EV charger quote that respects NEC 220.87 will include four details that the cheap bids tend to skip.

• Both calc methods on the page. The 220.82 Optional Method number and the 220.87 actual-demand number, with the source data for each. If only 220.82 appears and the house has 12 months of utility history, the electrician left the more accurate method on the table.
• The utility demand summary attached. The 12-month peak kW or peak kVA reading from the utility, with the month it occurred. A 220.87 calc with no attached demand summary is not a 220.87 calc; it is a number on a sheet.
• The 125 percent math shown. Existing demand at 125 percent, plus new EV load, equals total. The condition (2) line of 220.87 is the formula and the bid should show it.
• The fallback path if the calc fails. When 220.87 lands above the service rating, the honest quote names the next step, NEC 625.42 automatic load management on the EV side, a sub-panel with arbitrated loads, or the actual service upgrade with the math behind it. The smart panel and load management post walks the EVEMS scope; the sub-panel vs service upgrade post walks the sub-panel scope.

The 30C Deadline Angle for 220.87 Households

The federal Section 30C EV charger tax credit expires June 30, 2026, twenty-one days from the publish date of this post. The credit covers 30 percent of qualified expenditure up to $1,000 for the EV charger and the electrical work directly required to install it. A clean charger install (no panel work, no service upgrade) lands at $500 to $1,500 and captures the full credit. A service-upgrade install lands at $3,500 to $8,000 and captures the same credit ceiling, which means the $4,000 panel-upgrade portion is paid by the homeowner with no offset.

The 220.87 method is the calc that flips quote-with-upgrade into quote-without-upgrade for borderline houses. A homeowner who is about to sign a $7,000 bid because the 220.82 calc landed close to the service rating should ask the electrician to pull the utility demand history first. If the 220.87 number clears the service, the quote drops to the EV-only scope and the 30C credit covers most of it. The 30C 26-day install timeline post walks the deadline math.

Where 220.87 Sits in the ChargeRight Workflow

The $12.99 assessment runs the NEC 220.82 Optional Method first because the Optional Method works without any utility data. The questionnaire collects square footage, appliance inventory, HVAC system, and the EV charger amperage. The AI panel-photo analysis verifies the main breaker rating and the panel brand. The report returns the 220.82 calculated load against the safe-capacity ceiling.

The 220.87 cross-check fires when the homeowner supplies a peak monthly kWh or peak kW demand reading from the utility. The gap-detector flags the missing field as a high-impact upgrade path with the headline: “Want to use your real utility data instead of a calculated estimate?” When the data is added, the report shows both numbers side by side, and the borderline-house cases that 220.82 would flag for an upgrade are re-evaluated against the actual recorded peak.

The right framing: 220.82 is the default because it works without any input the homeowner does not already have. 220.87 is the upgrade path when the data is available. For an occupied dwelling with 12 months of utility history, 220.87 is the more accurate number and the one an honest quote should lead with.

When 220.87 Triggers a Real Upgrade Conversation

The 220.87 method is not a guaranteed pass. There are houses where the actual recorded peak plus the EV charger exceeds the service rating, and the calc returns an honest no.

• All-electric house on a 100A service with a recorded peak above 65A. 100A × 0.8 = 80A safe-capacity ceiling. Existing demand at 125 percent already exceeds 80A before the EV charger is added. The 220.87 calc returns the same answer as the 220.82 calc on this house: the EV does not fit without load management or a service upgrade. The heat pump on a 100A panel post walks the worked example.
• Two-EV household. The 220.87 calc on an existing one-EV house does not include the second EV in the recorded demand because the second EV has not been installed yet. The math has to add both EVs at 125 percent on top of the recorded peak, and the result frequently exceeds a 200A service. The fix is NEC 625.42 automatic load management on the EV side; the two-EV power-sharing post walks the 625.42 scope.
• FPE or Zinsco panel. 220.87 is a load calculation, not a panel-safety check. A Federal Pacific Stab-Lok or Zinsco panel needs to be replaced regardless of the calc result; the documented breaker-failure history is the trigger, not the load number. The FPE and Zinsco post walks the replace-do-not-add rule.

The Bottom Line

NEC 220.87 lets the electrician use real utility data instead of a calculated estimate to size an existing service for a new load. For an occupied dwelling with 12 months of utility history (or a 30-day recording under the exception), the math is existing demand at 125 percent, plus the new EV load, compared to the service rating. The result is almost always more headroom than the NEC 220.82 Optional Method shows, because real-world peak demand is almost always lower than the conservative appliance-by-appliance estimate.

The right workflow for an EV charger quote on a fully occupied house: run 220.82 first (the calc that works without utility data), then run 220.87 (the calc that works better when the data is available). A defensible quote shows both numbers and uses the lower one with the method noted. A quote that shows only 220.82 on a house with 12 months of history left the more accurate method on the table.

Run the $12.99 NEC 220.82 calc first, then add the utility peak demand field for the 220.87 cross-check. The report returns both methods so the electrician's quote has both numbers to work from, and the 30C deadline rush does not push a panel upgrade the real demand data does not support.