Getting PRC-028, PRC-029 and PRC-030 Right

An Engineering Guide for IBR Owners: recording, ride-through, event response and SCADA detection that holds up to audit

Why this matters now

As of October 1, 2026, inverter-based resource (IBR) owners in North America are carrying three new obligations at once: record disturbances (PRC-028-1), stay connected through them (PRC-029-1), and detect, analyse and fix unexpected output losses (PRC-030-1). PRC-029-1 and PRC-030-1 both took effect on that date, and the PRC-028-1 recording deadlines for the existing fleet run through January 1, 2030.

These standards were written because inverter-based plants have tripped, curtailed or cut output in large numbers during ordinary grid faults. Losses of over 1,000 MW from a single fault have been recorded. NERC's response was to stop treating ride-through, recording and event follow-up as good practice and to make each one auditable.

The hard part is not reading the standards. It is proving them on a real plant, where the evidence sits in relay settings, recorder configuration files, PLC logic, historian archives and the clock that time-stamps all of it. A compliance claim is only as good as the weakest link in that chain.

This guide explains what the three standards require in engineering terms, where real plants fail them, how SCADA-based PRC-030 detection should be built and tested, and what to expect from an engineering partner. It closes with three anonymised case studies and a detailed FAQ.

The case studies in this guide are composite and illustrative. They are drawn from recurring patterns in field and design reviews and do not describe any specific client, site, project, manufacturer, integrator or utility.

The three standards in engineering terms

Each standard answers one question: PRC-028-1 asks whether you recorded the disturbance, PRC-029-1 asks whether you stayed on through it, and PRC-030-1 asks whether you noticed when you didn't and fixed the cause. They depend on each other: PRC-030 analysis is impossible without PRC-028 data, and PRC-029 performance is judged from the same records.

PRC-028-1: disturbance monitoring and reporting for IBRs

PRC-028-1 sets what an IBR facility must record, where, at what resolution, how accurately it is time-stamped, and how long it is kept.

Implementation is phased. New BES IBRs must comply with R1–R7 within 15 months of the effective date or by commercial operation, whichever is later. Existing BES IBRs must reach 50% compliance within three years of the effective date and 100% by January 1, 2030. R8 applied nine months after the effective date.

PRC-029-1: frequency and voltage ride-through

PRC-029-1 requires IBRs to stay connected and keep injecting current through defined voltage and frequency excursions. FERC approved it on July 24, 2025 (Order No. 909), and it took effect on October 1, 2026.

• Voltage ride-through: continuous operation between 0.90 and 1.05 per unit, with mandatory operation for defined cumulative durations below 0.90 pu and permissive operation only in the deepest band below 0.10 pu.
• Frequency ride-through: continuous operation from 58.8 to 61.2 Hz, ride-through of rate of change of frequency up to 5 Hz/s, and only narrow allowances for tripping on large phase-angle jumps.
• Recovery: real power restored promptly after voltage recovers, typically within 1.0 second.
• Exemptions: documented hardware limitations for in-service equipment, with documentation due within 12 months of the effective date. NERC filed PRC-029-2 on August 28, 2026 to widen exemptions for long-lead projects and HVDC-connected IBRs.

The practical point: a plant can have compliant inverters and still fail PRC-029 because a relay, a plant controller limit or a utility-side element trips first.

PRC-030-1: unexpected IBR event mitigation

PRC-030-1 turns event follow-up into a four-step, time-boxed process. It took effect on October 1, 2026.

The R1 test is conjunctive: both 20 MW and 10% must be met. On a 90 MW plant the 20 MW figure governs; on a 300 MW plant 10% (30 MW) governs. The exclusions cover resource variability, dispatch and ramping, planned outages and testing, transmission or collector losses, and protection misoperations handled under PRC-004.

ERCOT overlays

In ERCOT, NOGRR255 adds requirements on top of PRC-028-1 for disturbance monitoring, and NOGRR245 adds ride-through requirements alongside PRC-029-1. Under NOGRR255 as we apply it, fault recording for newer equipment is held to 128 samples per cycle, total record length to at least 5 seconds (we design to 6), rolling retention to 30 days and data delivery to 7 calendar days. ERCOT values should always be confirmed against the current Nodal Operating Guides before design.

Where compliance actually breaks

Most IBR plants that fail these standards have the right equipment and the wrong configuration, data path or evidence. Below are the failure modes we find most often, grouped by where they sit.

Two patterns run through the list. First, capability is not compliance: a PMU-capable relay, a networked recorder or a fast meter proves nothing until its configuration and an actual record show it working. Second, the chain fails at the handoffs: between the meter and the RTAC, the RTAC and the PLC, the PLC and the historian, the clock and the recorder. Each device can be correct while the system is not.

We also separate three kinds of evidence, because confusing them causes both false deficiencies and false confidence:



• Installed capability: drawings, device models, as-left settings and configuration files.
• Ongoing operating evidence: historian archives, recorder and SER files, alarm histories, screening records, retention logs.
• Event-driven evidence: PRC-030 R2 analyses, Corrective Action Plans and their implementation. If no qualifying event has occurred, the absence of an R2 analysis is not a deficiency.
  

SCADA implementation done right

A defensible PRC-030 R1 implementation is a measured data path, a small piece of deterministic logic, an archive and an alarm route, each verified end to end. The logic itself is rarely more than 30 function blocks. The work is in making every link provable.

1. Start from the measurement, not the logic

• Pick the source deliberately. Use the POI revenue or check meter's real power (net or gross, documented), not a computed sum of inverter outputs, which drops out when inverter communications fail.
• Measure the update rate where the logic runs. A meter may update its MW register every 200 ms, but by the time the value has crossed a substation RTAC, a SCADA RTAC and a serial DNP3 link to a PLC, it can arrive every 2–3 seconds. Our acceptance target is at least 1 Hz at the PLC or RTAC tag.
• Check every hop for deadband and event-only reporting. DNP3 analog deadbands, class assignments and event buffers are the usual culprits. Integrity or continuous polling with unsolicited responses, and a deadband near zero on the MW point, fix most cases.
  

2. Build the detection logic to the standard's geometry

R1 says "within a 4-second period", which means any 4-second window, not fixed 4-second blocks. A single sample-and-hold register every 4 seconds misses a 22 MW drop split 12/10 across the boundary.



• Rolling window: a 1-second self-resetting clock shifts the current value into a four-stage register (1, 2, 3 and 4 seconds ago). The register must execute oldest-first; if the newest copy runs first, all four stages hold the same value and the window collapses to about 1 second.
• Four comparisons: subtract each stored value from the current value, take the absolute value, and compare each against the threshold. The standard says "changes", so detection should be direction-neutral unless a documented basis says otherwise.
• Conjunctive threshold: set the threshold to the larger of 20 MW and 10% of verified gross nameplate. If the nameplate is not verified, do not hard-code the number.
• Hold the trigger: a pulse timer of about 5 seconds keeps the trigger visible to a 1-second historian and to the alarm system.
  

3. Supervise the data, not just the device

• Communications permissive: AND the trigger with a "communications healthy" signal built from every stage of the path. At least one status must be generated by the device running the logic itself (its own driver or heartbeat), because a status polled over the link it supervises freezes "healthy" when that link dies.
• Restore delay: after communications return, the stored samples are stale for up to 4 seconds. A restore delay of about 5 seconds before re-arming prevents a false trigger on the first fresh sample.
• Separate communication alarm: loss of supervision must alarm on its own, distinct from the event trigger, with its polarity documented.
• Optional total-loss flag: a supplemental flag for complete loss at low output (below the R1 threshold) is useful operationally. It should use the same 4-second window and the same communications permissive, and stay separate from the R1 trigger.
  

4. Archive and alarm for evidence

• Archive at 1 second with deadband off: the MW source, the computed changes, the trigger, the communication alarm and the communication statuses. Retain at least 90 days of 1-second data.
• Export raw values: time-averaged or interpolated exports are not evidence of what the plant did. Raw-value exports for a ±10-minute window should be a documented, tested procedure.
• Route the alarm beyond the HMI: a remote operations desk and a named notification list, with an alarm on bad data quality as well as on value.
• Time stamps that trace to UTC: GPS clock to RTAC by IRIG-B, RTAC to PLC and historian by SNTP or better. One second matters here; one millisecond does not.
  

5. Test it like protection

Every implementation should have a test mode that substitutes a test MW value, plus a witnessed or evidence-backed test plan:

Two procedural rules prevent most test-day problems. Set the test value equal to live output before enabling test mode, so entering and leaving test mode does not itself trigger. And record the time test mode was switched off, with data proving the logic is back on the live meter.

Case study 1: Four logic revisions to a PRC-030 trigger that works

Composite and illustrative. Not a description of any specific client, site, integrator or project.

Situation. A utility-scale solar plant of under 100 MW needed PRC-030-1 R1 detection in place by the effective date, with about three weeks to go. The plant's SCADA integrator was to program the logic in the plant controller (a PLC). The engineering firm's role was to set the design basis, review each logic revision and verify the deployment evidence.

Approach. Every revision was reviewed from the compiled project file itself: wiring, negation flags and the controller's execution (solve) order were read out block by block. Screenshots were used only as a cross-check.

What the reviews found.

Deployment verification. The test results were checked second by second against the historian export. Every intended test passed, including a direction-neutral increase and suppression during a simulated communication failure. The review also found that four of eight trigger alarms in the test window came from entering and leaving test mode, and that the export ended with test mode still enabled. A time-stamped screenshot showing the controller logic equal to the reviewed file, and the time test mode was switched off, closed both points. Live output during the test-mode period never changed by more than about 2 MW in 4 seconds, so no real event was masked.

Result. Detection was in service before the effective date, with one documented corrective action and an evidence package an auditor can follow.



Lesson. The two most serious defects would have passed a casual functional test because they cancelled each other. Reviewing the compiled logic, not the screen capture, is what found them.

Case study 2: "We record everything through the RTAC"

Composite and illustrative. Not a description of any specific client, site, manufacturer or project.

Situation. A large solar facility with no dedicated disturbance recorder reported that its PRC-028 approach was "RTAC event-based collection of relay event records and SER." The owner wanted to know whether that claim would hold up, and what would be needed if it did not.



Approach. Rather than argue from the architecture drawing, the review asked for the records themselves. The owner provided about 30 relay event files from line, bus, transformer and feeder relays, plus the as-built protection and communications drawings and the RTAC project exports.

What the records showed.

What the gap assessment concluded. Relay-native records could support SER and time-synchronisation evidence, but could not be the fault-recording basis at any site, and certainly not under ERCOT's tighter rules. The facility needed either dedicated fault and dynamic recording at the main transformers and feeders, or a demonstrated relay-plus-PMU design meeting the rates, lengths and continuous recording the standard requires. The RTAC collection configuration and historian retention moved to the top of the information request, because the whole recording claim rested on them.



A contrasting case. At a second facility, two dedicated recorders were already on the station network and their failure alarms were mapped to SCADA. The gap there was narrower: no retrieval software was installed, so 20- or 30-day retention depended on the recorders' own storage. The fix was software and firewall configuration, not new hardware.

Lesson. "We record everything" is a hypothesis. Thirty real event files answered in a day what an architecture debate would not have settled in a month, and they also produced positive evidence (time quality) the owner had not known it had.

Case study 3: Compliant inverters, a gen-tie that trips first

Composite and illustrative. Not a description of any specific client, site, manufacturer, utility or project.

Situation. A large battery storage plant was preparing its PRC-029-1 ride-through evaluation. The inverter manufacturer had supplied standard settings, and the owner wanted them checked before the as-built settings were finalised.

Approach. The review followed the fault current path from the inverter to the point of interconnection and listed every element that could disconnect the plant during a voltage or frequency excursion: inverter protective functions, plant controller limits, collector and main transformer relays, the plant-end gen-tie relay, and the utility's line relays at the interconnection substation.

What the review found.

1. Inverter settings. Most manufacturer defaults sat outside the ride-through envelope, but several needed site-specific values. Each field on the as-built settings form got a default, a recommended value, an action and a reason, so the owner could sign off line by line.
2. Utility line relay. The utility's gen-tie relays carried a positive-sequence undervoltage element at 0.90 pu with a 1.0 s definite time. PRC-029 requires continuous operation down to 0.90 pu and mandatory operation below it for cumulative periods measured in seconds. A sustained sag to 0.85 pu would therefore trip the gen-tie, and the whole plant with it, inside the zone where the plant is required to stay on.
3. Ownership disagreement. The construction contractor proposed a much lower, slower setting. The utility declined, on the basis that ride-through and protection-coordination obligations apply to the generator's protection, not to transmission-owned line relays.

Why it matters even when the relay is not the owner's. In a real event the plant would disconnect, the output loss could cross the PRC-030 threshold, and the owner would be the one documenting the event, the root cause and the response. A gen-tie trip may fall under the PRC-030 transmission-loss exclusion, but the owner still has to evidence that classification. The owner is also the one whose PRC-029 performance record shows a trip.

Path forward.



• Put the conflict in writing: the element, the ride-through curve it intersects, and the voltage-time zone affected.
• Ask the utility for a coordination study or a revised delay that keeps the element outside the mandatory-operation region while still meeting its own protection objectives.
• Set every plant-owned element (inverter, plant controller, plant-end relays) clear of the envelope with margin, and document it.
• Keep the technical record ready, so that if a trip occurs, the PRC-030 analysis can show the cause in days, not months.

Lesson. Ride-through compliance is a property of the whole connection, not of the inverter datasheet. The review must reach the last relay before the grid, including the ones the owner does not set.

Why Keentel Engineering

Keentel Engineering is a P.E.-led power systems firm that works at the exact junction where these three standards are won or lost: protection settings, disturbance recorders, station automation, SCADA logic and the evidence that ties them together. Most firms cover one side of that junction. We cover all of it, from the relay trip equation to the historian export.

What sets the work apart

Firm credentials

• Founded and led by Sandip R. Patel, P.E., IEEE Senior Member.
• Power system studies, protection and control, and substation work from 4 kV to 765 kV.
• Four service lines under one roof: electrical design, power system studies, NERC compliance, and QA/QC.
• Offices in Tampa (HQ), Austin, Sacramento and Baltimore; NSPE member firm, D-U-N-S registered, BBB A+ accredited.
• Utility-scale solar, wind and BESS engineering, EMT modelling, owner's engineer services and interconnection engineering, so findings connect to the models and agreements that govern the plant.

What you receive

• PRC-028-1 / NOGRR255: hardware verification report, software and configuration assessment, recorder and DDR gap analysis, time-synchronisation review, retrieval and retention design, disturbance monitoring SOP, R8 response procedure.
• PRC-029-1 / NOGRR245: inverter, plant controller and relay settings review against the ride-through curves; field-by-field settings recommendations; exemption documentation support; coordination findings for elements owned by others.
• PRC-030-1: R1 design basis and threshold determination; detection logic design or independent QA/QC of an integrator's logic; data-path and update-rate verification; test plan and readiness memo; R2 analysis framework, R3 CAP and technical-justification templates, R4 tracking; O&M evidence and retention matrix.
• SCADA implementation support: RTAC, PLC and historian requirements, alarm and notification design, communications supervision, and witnessed or evidence-based acceptance testing.

Frequently asked questions

PRC-028-1: disturbance monitoring

PRC-029-1: ride-through

PRC-030-1: unexpected event mitigation

Working with Keentel

Getting started

The fastest route to a defensible position is a short, file-based review: send what you have, and we will tell you within days which parts of PRC-028, PRC-029 and PRC-030 your plant already proves, which it only claims, and what it would take to close the gap. Where you have an incumbent SCADA integrator, we will work alongside them.

Keentel Engineering · 400 N Ashley Dr, Suite 2600, Tampa, FL 33602 · (813) 389-7871 · contact@keentelengineering.com · Offices in Tampa, Austin, Sacramento and Baltimore

Sources

• NERC PRC-028-1, Disturbance Monitoring and Reporting Requirements for Inverter-Based Resources
• NERC Project 2021-04, PRC-028-1 Implementation Plan
• NERC PRC-030-1, Unexpected Inverter-Based Resource Event Mitigation
• NERC petition for approval of PRC-029-2 (August 28, 2026), including PRC-029-1 status and dates
  

Disclaimer

This article is general technical information, not legal advice or a compliance determination for any facility. Requirements are summarised; the governing text is the current NERC Reliability Standard, its implementation plan, and for ERCOT facilities the current Nodal Operating Guides. Applicability depends on registration and facility characteristics. The case studies are composite and illustrative: they do not describe any specific client, site, project, manufacturer, integrator or utility. Product names are trademarks of their owners, used for identification only, and imply no affiliation or endorsement.