Can smart-meter data support an electrical load calculation?

Choose the measured-demand path before collecting records

A saved Reddit question compares a home monitor’s 15-minute peak with a utility’s lower one-hour peak and asks whether a 100 A service can support charging. The disagreement is a useful starting point: different averaging intervals and measurement periods can describe the same house differently. The forum answers do not decide what the permitting authority accepts.

For an existing installation, section 220.87 of the 2023 NEC, adopted with amendments in California’s 2025 Electrical Code, provides a measured-demand method. It starts with actual maximum-demand data for one year. The maximum existing demand is taken at 125%, then the new load is included; the result must remain within the applicable feeder ampacity or service rating, with the required overcurrent and overload protection.

That is a defined method, not a rule to multiply any number on a bill by 1.25. Have the responsible professional establish whether the records and installation fit it. Identify the feeder or service being assessed, the relevant rating, the data period, and loads added or removed since that period.

Source: Question on 100A load calc · California Building Standards Commission: code editions and errata · LBNL: Characterizing electrical panel capacity, breaker space, and loads in U.S. single-family homes · Daly City: building handouts, including EMS plan review and inspection guidelines

A 30-day record has conditions that an annual record does not

A short recording is a conditional route when the necessary annual record is unavailable. AEMC’s 2016 manufacturer guidance describes the section 220.87 recording approach using 15-minute demand over at least 30 days on the highest-loaded phase, with occupancy and heating/cooling considered. That document explains the measurement technique; the designer must verify all conditions against the code edition accepted for the current permit.

Before commissioning a study, confirm the permitted duration, interval and seasonal or periodic-load adjustments. A 30-day recording option may be unavailable altogether where solar, wind or peak-load shaving is present; verify the adopted code text and permit requirements first. UC Berkeley’s campus guidance also describes 30-day studies, but its campus-engineer exception must not be carried over to residential permits. A quiet month or net-import chart does not automatically satisfy the method.

Source: AEMC: 2016 manufacturer Customer Support Tip on load recording; historical guidance, not adopted California code · UC Berkeley Facilities: load studies under CEC section 220.87; campus-specific guidance

Read the interval and channel definitions

Monthly kilowatt-hours describe energy over a billing period. Interval kilowatt-hours divided by the interval duration yield average kilowatts, not an instantaneous peak or automatically a valid design input. For example, 4 kWh within one hour averages 4 kW; the same energy used entirely in half an hour averages 8 kW during that half hour. The coarse record conceals that difference.

LBNL’s 2026 study notes that section 220.87 does not explicitly prescribe the annual record’s averaging interval; its research used 15-minute data to avoid smoothing. The absence of an explicit annual metering interval does not automatically disqualify an hourly export. It still needs acceptance for the actual assessment. Provide the finest available interval, units, time zone, missing-data flags and a definition of import, export or gross consumption channels.

Source: LBNL: Characterizing electrical panel capacity, breaker space, and loads in U.S. single-family homes

Understand a conditional calculation without treating it as approval

Illustrative comparison: assume a professional has accepted a complete annual record and established 60 A as the relevant existing maximum demand for the assessed supply. The section 220.87 existing-load contribution is 60 × 1.25 = 75 A. If the separately determined addition contributes 20 A under the applicable rules, the comparison becomes 95 A against the assessed rating.

The example assumes the correct phase/loading basis and that the new-load value already reflects any requirements applicable to it. It does not establish that a 20 A appliance, charger or circuit always contributes 20 A. Nor does a result below a 100 A rating prove conductor suitability, equipment condition, spare breaker space or acceptance of the original data.

Deliver the original records with the interpretation

Save the unedited export, the channel definitions and the calculation identifying the maximum accepted demand. Document vacancies, missing intervals, unusual outages, solar or storage operation, and later equipment changes. A screenshot of the lowest month cannot establish an annual maximum.

If the data cannot support the method, use an accepted inventory calculation or a properly scoped recording study rather than guessing a correction factor. The professional and permitting authority determine which evidence is sufficient for the actual project. Keep that decision with the report so a later reviewer can understand why the records were used.

Put it into practice

A decision table to keep handy