PRC-028-1: human versus the Dashboard

A utility-scale solar plant in Texas had its PRC-028-1 assessment done twice: once by a protection engineer by hand, once by GridStrong's Dashboard from the same documents. By hand: 31 days, 234 of the 644 checklist items answered, only the fault recorder and the inverter graded and the plant's 15 relays left blank. A senior protection engineer reviewed both studies.

PRC-028-1 requires the owner of an inverter-based plant to record sequence of events, fault and dynamic disturbance data at the plant, synchronize it to UTC, keep it, and provide it on request in standard formats. For this plant that is 644 questions about its fault recorder, its relays and its inverters, and each answer needs the evidence behind it.

The Dashboard read the plant's drawings, settings files, relay test reports and manufacturer manuals, answered all 644 checklist items, and attached the evidence to every answer: about 1,300 attachments from 45 documents and about 2,700 citations, each a quoted line and a page number. Every citation was compared with its page before delivery, and about 150 source pages were cropped and attached as images.

Where one requirement can be met by more than one device, the Dashboard names the device to rely on and grades the others against it. Where only the owner can close a question, it says what to send.

Five findings the manual study missed:

• The owner's signed procedure commits to retrieving recorder data for 10 days; the standard requires 20. The manual study graded it met. The updated procedure, already drafted, only needs a signature.
• The transformer high-side currents are recorded by the primary differential relays. The manual study said low side only, which the reviewer flagged as wrong.
• The fault recorder has no feeder current channel, so the feeder relays are the only record of a feeder fault, and they hold 1.0 second per event against the 2.0 seconds required. The manual study graded the feeders on the fault recorder.
• The neutral overcurrent trigger exists: the transformer relays trigger on their restricted earth fault elements and the fault recorder on sequence current. The manual study graded it not met.
• Nothing on file proves the clocks are accurate, and the fault recorder's time-sync-loss alarm is disabled. The manual study graded clock accuracy met at every location.

The Dashboard's assessment was reviewed and delivered 9 hours after the first look at the plant's files, with seven recommendations, starting with which device serves which requirement at each location, and a 16-item request list tied to the requirements it closes. The senior protection engineer's review found one error in its 644 answers; the review of the manual study raised 21 points, 12 of them errors. The Dashboard's assessment sits in the owner's record beside the manual study, so the two stay comparable, and it re-runs when a setting or a document changes.