Arc flash field data sheets, read into device tables for the engineer running the study.

A study starts with a walk through the electrical rooms. Technicians copy nameplates, trip unit dials and conductor runs onto sheets, sometimes with a photo, and someone back at the office types it all in. That person also has to notice that the switchboard on the sheet has another name on the one-line, or that a feeder's length never got written down.

In the test, AI reads copies of those sheets and drafts structured device tables: equipment names, conductors and lengths as written, protective device settings as recorded, and a list of every field left blank or in conflict with the one-line. Each entry points back to its sheet. The engineer running the study resolves the list.

The input list is long and specific. Western Michigan University's arc flash study section shows what a study takes in. For breakers and fuses: maker and model, type of trip, range of settings, short circuit rating, current rating and the settings themselves. For overcurrent relays: ampere tap, time delay and instantaneous ranges and the current transformer ratio. For low-voltage cable: size, length, number, conductor material and whether the conduit is magnetic. Busway adds impedance and length, and motors add horsepower and the NEMA code letter. The draft keeps one column per field, so an empty one shows instead of quietly going missing.

Names are the first thing to drift. The same section wants equipment tags consistent across the diagrams, protective device submittals, input data and recommended settings, data sheets cross-referenced to the diagram's tag numbers, and any gap in the one-line called to the architect's attention. In practice a technician writes what's on the door and the drawing says something else. The draft keeps both names side by side, never merges them, and lists the pair as a conflict.

Some fields can't be had on a walk. IAEI Magazine's 2021 look at commercial building studies describes a transformer nameplate with its percent impedance left empty, data that takes a shutdown to read, and cables too impractical to trace end to end, which some practices handle by assuming a length or counting floors. At a 2024 IEEE chapter meeting, one engineering group said field collection needs site personnel and covers removed, and listed what it falls back on when cable data is missing, including 19 feet for low voltage and 49 feet for medium voltage. Any assumption is the engineer's to make, so the draft shows only what the sheet says, and a blank stays blank.

IEEE 1584 asks for things older sheets never carried. Consulting-Specifying Engineer's account of the 2018 edition of IEEE 1584, the Guide for Performing Arc-Flash Hazard Calculations, notes it moved from two electrode configurations to five and that enclosure types and sizes will need identifying. An IEEE chapter talk adds that conductor gap usually isn't in submittals and can be dangerous to measure, and that an enclosure counts as shallow under 20 inches high and wide and 8 inches deep, below 600 volts. IEEE 1584.2-2025 now gives checklists for this data on systems of 1000 volts and below. The draft reports what the sheet shows. Choosing a configuration, a gap or a box belongs to the study engineer.

Buyers ask for the data itself. Onondaga County, New York, in its requirements for new installations, lists field data schedules, an input data report and field photos among the deliverables, with a registered engineer running the study to IEEE 1584 formulas. In a 2025 addendum, Jurupa Community Services District in California counted 402 panels and assets, called its data collection and verification task mandatory and told bidders to identify their assumptions. Your reviewer will check that each entry landed on the right device and tag, that blanks are real, and that every one-line conflict surfaced. We're summarizing public documents here, and this isn't legal or engineering advice.

After your reviewer's yes, sheets coming in from new surveys get the same reading for about 30 days, close enough to the walk that a technician can still go back.

The engineer running the study reviews each table before anything is keyed into a model, and resolves every open item.

A fixed price for those sheets comes before you sign. Anything further is its own job at its own price, with monthly support after.