Nearly a year on, the verdict is in on Iberia’s blackout

Nearly a year after Spain and Portugal lost the grid, ENTSO-E’s final report has put voltage control back at the centre of the story.

Published on 20 March 2026, the report describes the event as the most severe blackout on the European power system in more than 20 years and the first of its kind, because the collapse was driven by overvoltage and cascading generator disconnections rather than by the more familiar pattern of simple generation shortfall or under-frequency decline. That recasts the outage from a culture war argument about renewables into a harder engineering question about how Europe is actually running a power system whose behaviour has changed faster than many of its operating assumptions.

ENTSO-E’s briefing identifies three major tripping events in south-west Spain — 355 MW in Granada at 12:32:57, 727 MW in Badajoz at 12:33:16, and 928 MW across Segovia, Huelva, Badajoz, Sevilla, and Cáceres at 12:33:17 — before the peninsula lost synchronism with the rest of Continental Europe at 12:33:19 and the AC links to France and Morocco disconnected by 12:33:21.

System defence plans were activated in Spain and Portugal, but they were built around a different class of emergency and could not arrest an event whose defining feature was fast voltage rise. As evidence, the sequence is decisive; as explanation, it is only the surface.

A faster grid than its rulebook

What the report really exposes is a control architecture that had become too static for the system it was supposed to govern. ENTSO-E says the blackout resulted from several interacting factors, among them oscillations, weak voltage and reactive-power control, differing voltage-regulation practices, rapid output reductions and generator disconnections in Spain, and uneven stabilisation capabilities.

The simplified root-cause tree offered up by ENTSO-E makes the same point more bluntly: a fast voltage increase sat at the centre of the collapse, fed by a low margin between operating and disconnection voltages, ineffective voltage support from conventional generation, no reactive-power response to voltage changes from renewable generators, lower line loading, manually operated shunt reactors, and a further loss of reactive-power absorption as the event progressed. Iberia did not black out because it lacked megawatts in aggregate; it blacked out because too many layers of the system were poor at controlling voltage dynamically, and several of them failed in the same direction.

That conclusion should not have come as a surprise. ENTSO-E’s own 2023 system needs study had already warned that voltage stability was becoming harder to manage as synchronous generation declined, converter-connected resources increased, and power had to travel further between production and demand centres. The same study said the system faces both a reduction in available synchronous units supporting voltage and an increase in voltage-support needs due to volatile transit situations and higher utilisation, with growing fluctuations in reactive-power demand and losses.

Viewed against the Iberian incident, those lines look less like general future-gazing than like an early draft of the failure itself.

Europe spent years discussing decarbonisation as a generation-and-transmission problem; the blackout is what happens when the voltage-control implications of that transition are left in the technical annex.

Several of the report’s findings sit exactly in that gap between energy policy ambition and dynamic system performance. On the renewable side, ENTSO-E says Spanish renewable generators provided no reactive-power reaction to voltage changes, and shows that units operating at fixed power factor changed Q in proportion to P rather than in response to system voltage. That is not an argument against inverter-based generation; it is an argument against pretending that connection alone is equivalent to control capability.

On the conventional side, the report is scarcely kinder, stating that voltage control support from conventional generators was not effective, that applicable Q-output requirements lacked specifications for dynamic behaviour, and that there was no economic consequence if actual reactive output diverged from the 75% rule.

The deeper problem is not technology preference but performance obligation: Europe has been better at specifying what plants may connect than what they must actually do when the grid moves quickly.

Reactive power as system security

The voltage margin issue makes the same point in a more operational register. ENTSO-E’s briefing shows Portugal operating with an upper 400 kV limit of 420 kV, while Spain’s upper limit was 435 kV, leaving less room before protective disconnections came into play as voltage variations intensified. That alone would have been uncomfortable enough, but the system was also short of fast absorption when it needed it most.

Shunt reactors and capacitors in Spain were being used for static voltage control and switched manually; at 12:32, only about 58% of Spanish shunt-reactor reactive capacity was connected, and the remainder could not be brought in quickly enough because manual switching required processing time.

The difference between a stable high-renewables grid and a brittle one is often presented as a matter of abstract system design. In practice, it can come down to whether the reactive assets exist, whether they are automated, and whether their response belongs to the timescale of the disturbance rather than to the timescale of operator workflow.

In addition, ENTSO-E records tripping of embedded generators below 1 MW, including household-scale PV reported by one inverter company, and links that loss of generation to increased flows from the transmission system to distribution grids. That in turn reduced line loading in parts of the grid and pushed voltage higher still.

ENTSO-E’s wider system study had already highlighted the need for a clear admissible range of reactive power exchange at the TSO-DSO interface and warned that distribution and transmission do not always suffer the same voltage problem at the same time — low voltage can exist at distribution level while transmission is already running high. Iberia therefore looks like a case study in a broader European weakness: distribution-connected resources are now system actors whether the institutional culture has caught up or not, and a transmission operator cannot manage voltage cleanly if distribution-level plant behaviour remains only partially visible, weakly coordinated, or locally optimised against the wrong objective.

Interconnection belongs in that discussion, but not as a miracle cure. Spain’s regulator, CNMC, has called for stronger links with France, harmonised Spanish and European voltage rules, and more rigorous inspections of protection systems, while also saying the event demonstrated the need to keep adapting technical, operational, and regulatory frameworks to a system in continuous transformation, with high renewable penetration and increasing voltage volatility.

All of that is fair, and stronger cross-border capacity does matter; ENTSO-E’s 2023 study likewise says strong interconnection is essential to exchange power flows from flexibility sources. Even so, interconnection cannot substitute for local voltage control any more than a larger motorway can compensate for a failed steering rack. Iberia’s outage was not fundamentally a story about insufficient opportunity to import support. It was a story about what happened when the support architecture on the system itself proved too slow, too fragmented, and too weakly enforced.

What Iberia changes for Europe

That has direct consequences for what grid operators and plant owners will now be pushed to buy, retrofit, and specify. ENTSO-E’s system-needs work says a good mix of network-based and generator-based solutions will be necessary, and points to sufficient dynamic reactive-power sources, including variable shunt reactors, STATCOMs, synchronous condensers, and energy storage systems that can act as dynamic reactive-power sources. Those are no longer optional refinements at the edge of the transition; they are part of the price of running a network where voltage support can no longer be assumed to arrive automatically from whichever large synchronous units happen to be online.

The standards picture is moving in the same direction, and the blackout is likely to accelerate it. In a 2025 report on grid-forming capability for power park modules, ENTSO-E set out proposed compliance tests for voltage source behaviour, including islanding tests in which the equipment under test must take over a local load, keep voltages stable at a new operating point, respond in under 5 ms, and settle in under 15 ms.

Those figures do not solve Iberia retrospectively, but they do show where European thinking is heading: away from coarse, largely static notions of compliance and towards demonstrable dynamic behaviour under stressed conditions. After a blackout in which fixed power factor, weak Q-discipline, and slow switching all featured prominently, it is hard to imagine regulators being content for much longer with rules that describe capability in theory but test it too softly in practice.

ENTSO-E’s own concluding slide says the power system is governed by physics, that the outage was triggered by no single cause, and that the solutions are already technically feasible. That is probably the sharpest summary available. The technical means to build a more resilient high-renewables grid already exist, but they demand a more serious view of reactive power, stronger plant-performance obligations, tighter TSO-DSO coordination, and a regulatory culture willing to treat voltage control as core infrastructure rather than supporting detail.

Europe has spent much of the past decade arguing over how quickly it can connect the next megawatt. Iberia suggests the harder question is whether it has been equally serious about who controls voltage once that megawatt arrives.