NEC 220.87 Load Study: Using Measured Demand to Size an Existing Service

NEC 220.87 lets you size new load on an existing feeder or service from its measured maximum demand instead of a calculated load. The maximum demand comes from a year of utility demand data or, when that isn’t available, a continuous recording of at least 30 days. If 125 percent of that maximum demand plus the new load does not exceed the feeder’s ampacity or the service rating, and the feeder and service have the overcurrent and overload protection the code requires, the existing equipment can take the new load.

That is often the difference between adding EV chargers, a rooftop unit or a production line on the service you have and paying for a service upgrade. Below we cover what the rule requires, how a 30-day recording is set up and read, a worked example, and the mistakes that get load studies rejected.

NEC 220.87 at a Glance#

The section number and wording above follow the 2020 and 2023 editions. Your authority having jurisdiction (AHJ) enforces a specific edition, sometimes with local amendments, and that edition governs. The 2026 NEC moved the load calculation rules from Article 220 to a new Article 120, so in jurisdictions that adopt the 2026 edition this provision is 120.87; check the edition and section your AHJ cites.

Why Measured Demand Is Often Lower Than a Calculated Load#

A standard Article 220 calculation adds up lighting, receptacle, motor and equipment loads from nameplates and applies the demand factors the code allows. It has to be conservative, because it is written for buildings that don’t exist yet. In an existing building, not every load runs at nameplate, and not every load runs at once. The measured maximum demand captures that diversity directly.

A service that looks fully loaded on paper often shows unused capacity when its demand is measured. It can go the other way too: a building whose use has intensified since it was designed may already draw more than its drawings suggest. Either way, the measurement replaces a guess about the existing load with a number.

Only the existing load comes from the measurement. The new load is still calculated under Article 220, for example at 7,200 watts or the nameplate rating, whichever is larger, for each EV charger under 220.57 (2020 edition and later).

Option 1: A Year of Utility Demand Data#

If the utility has metered the service with a demand meter for at least a year, the highest demand in that year can be the existing load. Ask the utility for interval data or 12 months of billing demand for the specific meter, and confirm:

• The meter serves only this service. A meter that totals several services, or a service fed from more than one meter, needs sorting out first.
• The units. Utilities usually bill demand in kilowatts. Converting kilowatts to amperes takes the voltage and the power factor, and assuming a power factor of 1 understates the current.
• Actual demand, not billed demand. Some tariffs bill a ratcheted or minimum demand that isn’t what the meter recorded. Use the metered maximum demand.
• The year was representative. A year with a vacant floor, a shutdown or a lost tenant isn’t.
• There is no on-site generation behind the meter. A utility meter records net demand. With PV or battery storage on the service, the net number can understate the load the service conductors actually carry when the PV isn’t producing.

Option 2: A 30-Day Load Recording#

When a year of data isn’t available, the NEC accepts a recording of at least 30 days. The rule is short, and every phrase in it matters.

What the recording has to capture

• Continuous recording for at least 30 days. A gap from a dead battery, a full memory card or a tripped meter supply breaks the continuity the code asks for. Plan on restarting the recording.
• The highest 15-minute average. The demand is the highest average kilowatts reached and maintained for a 15-minute interval. Instantaneous peaks such as motor starting current don’t count, and neither does a longer average that smooths out the peak.
• The highest loaded phase. The rule names it because one phase is usually heavier than the others. Recording every phase shows which one it is.
• An occupied building. The recording has to be taken while the building or space is occupied and operating normally.
• Seasonal load. The recording has to include, by measurement or calculation, the larger of the heating or cooling equipment load and any other periodic or seasonal loads. A recording made in a mild month needs the cooling or heating load added by calculation, and the report should show how.

Setting up the meter

The recording uses a recording power meter on the service or feeder conductors: a current sensor on each phase and on the neutral, and voltage connections to each phase, as Figure 1 shows. Recording voltage, kilowatts and kVA as well as current lets the engineer convert between them with the measured power factor instead of an assumed one.

Installing current sensors in an energized switchboard is energized work under NFPA 70E. It is done by qualified persons, after shock and arc flash risk assessments, wearing the PPE the arc flash label calls for. Where a short outage is available, installing the sensors with the equipment de-energized is the better choice.

Reading the result

Record at 1-minute intervals or shorter. The 15-minute average can then be computed for every 15-minute window, not just the windows that start on the quarter hour. The highest rolling 15-minute average is the same as or higher than the highest fixed-block value, so it is the conservative figure to report. Figure 2 shows how much an instantaneous peak can overstate the demand, and how a long averaging interval can understate it.

When the 30-Day Option Isn’t Allowed#

The NEC doesn’t allow the 30-day option on a feeder or service with a renewable energy system, such as solar PV or wind, or one that uses any form of peak load shaving. On-site generation and peak shaving lower the demand the meter sees, so a 30-day window could miss the demand the conductors carry when the PV isn’t producing or the peak shaving system is offline.

Peak load shaving can take several forms: battery storage or generators dispatched to cap demand, or controls that shed load when demand climbs. If any of these are on the service, ask the AHJ how it reads the rule, and plan for a year of interval data or a full Article 220 calculation.

The 125 Percent Test: A Worked Example#

A hypothetical building has a 1,200 A, 208Y/120 V three-phase service. The owner wants to add six Level 2 EV chargers, each rated 32 A at 208 V single-phase (about 6.7 kVA), two on each pair of phases.

1. Measured maximum demand. The 30-day recording, taken in July with the building fully occupied, shows a highest 15-minute average of 610 A on phase B.
2. 125 percent of maximum demand. 1.25 × 610 A = 762.5 A.
3. New load. Under 220.57, each charger counts at 7,200 VA, the larger of 7,200 VA and its nameplate, for 43,200 VA in total. Article 625 treats EV charging as a continuous load, and this example carries that into the service check at 125 percent, which is the conservative choice: 54,000 VA. Balanced across the three phases, that is 54,000 VA ÷ (1.732 × 208 V) ≈ 150 A per phase.
4. Total. 762.5 A + 150 A = 912.5 A, under the 1,200 A service rating.
5. Protection. The service has overload protection in accordance with 230.90, so the condition is met.

The service passes with 287.5 A to spare. An energy management system that caps the chargers’ combined current can reduce the new load figure further, where the edition your AHJ enforces allows it.

What 220.87 Doesn’t Check#

A passing 220.87 calculation shows that the feeder or service has the capacity for the new load. It doesn’t check everything else the new load affects:

• The utility transformer. The utility owns it and decides whether it can carry more load. Tell them early.
• Panel and bus ratings, and space. The panel feeding the new load needs the capacity and the spaces, not just the service.
• Voltage drop on the new branch circuits and feeders, especially the long runs common in parking lots. Our voltage drop calculator gives a first check.
• Neutral loading where the new load is line-to-neutral or nonlinear.
• Short circuit and arc flash. Adding load alone doesn’t change the available fault current, but a new transformer, a new feeder or changed breaker settings do. When the system changes, the short circuit and arc flash studies and the labels need updating.

Common Mistakes in 220.87 Load Studies#

1. Recording for less than 30 days, or with gaps. “About a month” with a two-day gap isn’t a continuous 30-day recording.
2. Reporting the instantaneous peak. Motor starting current inflates the number and can kill a project that would have passed.
3. Reporting a daily or hourly average. It hides the 15-minute peak the code asks for and makes a service look better than it is.
4. Recording one phase that isn’t the heaviest. Without all three phases, there is no way to show the measured phase was the highest loaded.
5. Recording a half-empty building. Holidays, a shutdown, a vacant suite or reduced shifts all lower the number.
6. Ignoring the season. A spring recording in Texas misses much of the cooling load unless it’s added by calculation.
7. Using the 30-day option on a service with PV or peak shaving. The code doesn’t allow it.
8. Applying the 125 percent factor to the wrong number. It applies to the measured maximum demand. The new load is calculated separately under Article 220.
9. Converting kilowatts to amperes at unity power factor. Most facilities run below unity, so the converted current comes out too low.
10. Leaving out how the study was done. Without the meter, the connection points, the dates and the interval, the AHJ can’t check the result and nobody can repeat it.

What a Load Study Report Should Include#

A report an AHJ can review without a phone call shows:

• The service or feeder measured, its rating or ampacity, and its overcurrent and overload protection
• The meter make and model, the current sensor ratings, and where they were connected
• The start and end dates and times, the logging interval and how the 15-minute demand was computed
• Occupancy and operating conditions during the recording, and the weather where heating or cooling matters
• The maximum 15-minute demand on each phase, with its date and time, and plots of the full recording
• Any seasonal load added by calculation, and the method
• The new load calculation under Article 220
• The 125 percent test and a clear conclusion
• A PE seal where the AHJ or the owner requires one

Frequently Asked Questions#

Can I use utility bills instead of a 30-day recording?

Yes, if they cover a full year of maximum demand for that service. Twelve monthly billing demands from a single meter that serves only that service can work. Ask the utility for interval data where it’s available, and convert kilowatts to amperes with a realistic power factor.

How long does an NEC 220.87 load study take?

The recording itself runs for at least 30 continuous days. Allow time before it to plan the meter connection and, where possible, schedule an outage, and time after it to analyze the data and write the report.

Does a 220.87 load study need a PE stamp?

The NEC doesn’t require one. Many AHJs and owners do, especially on commercial services, so check before the recording starts.

Can a load study avoid a service upgrade for EV chargers?

Often, yes. EV charging is a common reason for a 220.87 study, because the calculated load of an older building often leaves no room for chargers when its measured demand does. Our post on integrating EV charging stations into commercial electrical design covers the rest of the design.

How Dreiym Engineering Can Help#

We perform NEC 220.87 load recordings and capacity calculations as part of our electrical load analysis service, from setting up the meter to a report your AHJ can review. Our electrical power studies are led by Brian J. Brackensick, M.S., P.E., with field data collection by Ruben A. Rodriguez, and reports are PE-stamped where required. When the new load changes the system, we also update short circuit analysis and arc flash studies, and our electrical design services can carry the project through to construction drawings. For background, see our posts on when to get a residential load analysis and the tools used in electrical load analysis. Contact us or call (866) 621-6920 for a free quote. We typically respond within one business day.