On 4 June 2022, in Odessa, Texas, a surge arrester failed and caused a single phase to ground fault on the 345 kV system. Protection cleared it in three cycles, and the system still lost 2,555 MW: the grid did its job and the generation walked off on its own. That event produced the ride-through standard that took effect in North America on 1 October 2026, and its checklist is useful in Colombia today.

What ride-through is and what changed on 1 October

Ride-through means staying connected and continuing to operate through a disturbance instead of disconnecting to protect itself. NERC PRC-029-1, approved by FERC in Order No. 909 of 24 July 2025, made it mandatory from 1 October 2026 for inverter-based resources on the Bulk Electric System, and from 1 January 2027 for the rest. The obligation falls on the Generator Owner, not on the inverter manufacturer, and it reaches resources outside the BES that aggregate 20 MVA or more at a common point of 60 kV or above.

The deeper change is one of philosophy. A synchronous generator supports the grid through a fault by construction. An inverter is power electronics whose behavior during a voltage sag is a firmware decision, and for two decades that decision was to protect the asset. With inverters at 2 percent of the mix, that cost the system nothing. With inverters at half of it, every prudent disconnection is a concurrent loss of thousands of units.

The three ride-through regions: what the curve requires

Attachment 1 divides the voltage-time plane into zones, and only one of them allows a trip. This is the table for solar and storage plants.

Inside the mandatory region, simply hanging on is not enough: the inverter has to inject current up to its maximum capability to support the voltage, with reactive priority. Blocking is accepted in the permissive region, but current exchange has to resume within 5 cycles (about 83 ms at 60 Hz) and active power has to be restored within 1.0 s. Type 3 and Type 4 wind on the ac side gets shorter times for deep sags.

Two rules change the settings you will actually write: the times are cumulative over 10 s windows, so two sags in a row add up, and instantaneous voltage trips with filtering shorter than one cycle (16.6 ms) are prohibited.

The failure modes are not theoretical: they have MW attached

In the ERCOT preliminary analysis of the roughly 1,710 MW of solar lost at Odessa, every mode carries a number: overcurrent 445 MW, phase angle jump 385 MW, overvoltage 295 MW, dc bus unbalance 198 MW, slow ramp 147 MW and momentary cessation 131 MW.

PLL loss of synchronism is the emblematic case: a grid-following inverter needs a phase-locked loop to know what reference it is injecting against, and in a deep sag it loses that reference. The historical reaction was to trip, and the standard expressly closes that door. Momentary cessation is the inverter that stops injecting but never opens: it stays connected and silent. And transient overvoltage is the counterintuitive mode, because the plant does not fall over on the sag but on the rebound, when the fault clears, the voltage rises and the plant is still injecting reactive current. That is why the standard requires the inverter not to cause that overvoltage itself.

The link nobody audits

This is not our claim. The NERC State of Reliability 2026, published on 24 June 2026, lists inverter and plant controls, protection systems and model quality among the causes. And the standard measures at the high side of the main power transformer, not at the inverter terminals. Between the two there is a transformer, kilometers of collector system and the voltage drop produced by the very reactive current the inverter is required to inject. The ride-through margin is consumed inside the plant.

Seven checks before the next voltage sag

Turn the most demanding band of your ride-through curve into numerical requirements. One: specify the sag profile (0.50 p.u. for 1.20 s and 0.25 p.u. for 0.32 s) as an immunity requirement for station service, testable to IEC 61000-4-11, -4-34 and -4-29. Two: take ac contactor coils out of the critical path. Three: size the dc system and the UPS against that profile, not against backup autonomy. Four: plot the curve and every setting capable of opening the circuit on one graph. Five: review the unbalance protections. Six: require the sag profile as an explicit case in the coordination study. Seven: verify that ride-through is autonomous with the link to the plant controller down.

Specification lesson. Specify ride-through as a plant requirement, not as an inverter feature, and translate it into three lines of the specification: immunity of the station service, control and protection chain; settings, so that no threshold sits inside the continuous, mandatory or permissive regions; and evidence, with the curve and the settings overlaid on one signed graph and the sag tested at FAT. IEC 61439 switchboards and IEC 62271-200 switchgear panels are the vehicle for that requirement, but neither standard contains it. If it is not in the specification with a number and an acceptance criterion, it will not be in the plant.

And in Colombia?

The framework has already moved. CREG Resolution 101 104 of 18 April 2026 updated the technical requirements for solar and wind plants connected to the national transmission system, the regional systems and local distribution. The scheme has two tiers: CREG imposes the obligation and the Consejo Nacional de Operación (CNO) sets the numbers. Plants have to operate inside the low voltage ride-through (LVRT) and high voltage ride-through (HVRT) curves, but those limits arrive through a CNO Agreement, and for local distribution between 1 MW and 5 MW, CNO Agreement 1603 has already fixed them.

For transient overvoltages the reference is an international standard: plants on the national and regional transmission systems have to meet the curves recommended in IEEE Std 2800-2022, which therefore becomes mandatory in Colombia even though it remains voluntary in North America. On frequency, plants have to operate between 57.5 Hz and 63 Hz and have no instantaneous trip on rate of change of frequency (RoCoF). And the RMS and EMT models have to be certified by the manufacturers and updated two months before operation, together with the factory test reports.

Which means the EMT model and the FAT are already being asked of you. A model that represents neither the station service switchboard nor the real switchgear panel settings is not a model of the plant: it is a model of the inverter with the plant name on it. The ride-through checklist North America wrote after losing 2,555 MW does not apply to you, and it is still useful to you today, because a voltage sag does not check the jurisdiction first.

One boundary, to be clear about scope: compliance belongs to the Generator Owner and covers the plant as a whole. What sits inside our scope is making sure ride-through is not the thing that gets lost in the auxiliaries. Industrias Ectricol S.A.S. manufactures switchboards to IEC 61439 and switchgear panels to IEC 62271-200 in Funza, and integrates transformer substations, shelters and skids where the complete chain is verified before it leaves the factory, for utilities and renewable energy, mining and Oil & Gas. The conceptual groundwork is in what a transformer substation is.

Frequently asked questions

What is ride-through in a solar plant?

It is the ability to stay connected and keep operating through a voltage sag or a frequency excursion instead of disconnecting. It has been mandatory in North America since 1 October 2026 under NERC PRC-029-1, which also requires the plant to inject current to support the voltage during the disturbance.

Where is ride-through compliance measured?

At the high side of the main power transformer, the one that steps up from collector system voltage to interconnection voltage, not at the inverter terminals. That is why a certified inverter can see a condition at its own terminals that is more severe than the one on its certificate, and it is worth confirming with an electromagnetic transient study of the complete plant.

Why does a plant trip when the inverter meets the curve?

Because ride-through is frequently lost in links that never make it into the audit: the station service that feeds the relays and the controls, the medium voltage protection settings that sit inside the zone where the plant is required to stay connected, and the missing coordination between switchgear panel protection and the inverter curve.

Does Colombia require ride-through from solar and wind plants?

Yes. CREG Resolution 101 104 of 18 April 2026 requires operation inside the LVRT and HVRT curves that the CNO defines by Agreement, compliance with the transient overvoltage curves recommended in IEEE Std 2800-2022, operation between 57.5 Hz and 63 Hz and no instantaneous trip on rate of change of frequency, with RMS and EMT models certified by the manufacturers.

What happened at Odessa in 2022 and why does it matter?

A surge arrester failure caused a single phase to ground fault on the 345 kV system that protection cleared normally in three cycles. Even so, 2,555 MW were lost: 844 MW of synchronous generation and 1,711 MW of solar that disconnected without needing to, and frequency fell to 59.70 Hz. That event is the direct origin of the ride-through standard.