← All posts

Spain's Rules Split Voltage Control Four Ways, and the Government's and the Grid Operator's Blackout Reports Disagree Over Whether 28 April's Morning Was a Warning

Published November 2026 · 30 min read

Written 21 November 2026. The facts are as of that date. The ENTSO-E expert panel's final report of 20 March 2026 was not read in full for this post. Everything said here about that report comes from ENTSO-E's own 25-slide briefing on it, unless a line says otherwise.

At 12:33 on 28 April 2025 (all times here are Central European Summer Time), the grid of mainland Spain and Portugal collapsed. Three public accounts of why are free to read. The first is the Spanish government committee's report, a 182-page "non-confidential version" published on 17 June 2025. The second is Red Eléctrica's 21-page report, dated 18 June 2025. Red Eléctrica (REE) is the system operator, the company that runs the Spanish high-voltage grid in real time. The third comes from the expert panel of ENTSO-E, the association of Europe's grid operators, which published a factual report on 3 October 2025 and a final report on 20 March 2026.

All three agree that there was no single cause. They disagree over two questions: whether the morning's swings in voltage were a warning, and whether REE had scheduled enough power plants to hold voltage down. Read side by side, the documents also show a detail on timing. A smaller overvoltage on 22 April tripped equipment. The Ministry asked REE for a report on it, and the government report says that report arrived "el 28 de abril poco antes de las 11h". That is shortly before 11:00 on the morning of the blackout, about an hour and a half before the collapse. The request itself was public earlier: El Debate (23 October 2025) reported testimony from the Ministry's director general of energy policy. The press coverage this post found did not give the delivery time.

This post rebuilds what each party could see and what each was obliged to do, quoting each side's best case before testing it. Spanish quotations are exact. The English glosses that follow them are this post's translations, not the sources' words.

A few terms first

Who was supposed to control voltage

The rules split the job four ways. The government report and REE's report describe the split in the same terms.

  1. Conventional plants scheduled for voltage control had the dynamic duty under P.O. 7.4: regulate to a setpoint (government report, p.66). They are paid for it. Units scheduled this way "perciben una retribución por ello", they receive payment for it (p.110).
  2. Renewables, cogeneration and waste plants run under Royal Decree 413/2014 at a fixed power factor, "entre 0,98 capacitivo y 0,98 inductivo de forma horaria" (p.112), set hour by hour. The committee adds that renewables have long had the technology to regulate to a setpoint, "aunque la normativa aún no se lo exige ni se lo permite": the rules neither require it nor allow it (p.66). It estimates that "19 GW de potencia instalada fotovoltaica (casi la mitad del total) y 5 GW de eólica (un 16% del total)" could already do so (p.149, footnote 16). That is 19 GW of installed solar, almost half the total, and 5 GW of wind, 16% of the total.
  3. Distribution networks and large consumers also run at a fixed power factor under P.O. 7.4 (pp.66, 173).
  4. The operator has the all-or-nothing reactors and capacitors, and it decides which synchronous units to schedule through the "restricciones técnicas" process, its technical-constraints scheduling (pp.65–66).

Three gaps in the rules matter later in this post. First, the obligations apply at the boundary point with the grid: "la regulación es clara al establecer que las obligaciones y requisitos técnicos de operación y conexión son exigibles en el punto frontera con la red" (p.114). Second, nothing limits how fast voltage may change: "los procedimientos de operación no establecen actualmente limitaciones al ritmo de variación de tensión" (p.116). Third, two pending rule changes had not been approved. REE writes that "al no haberse aprobado la propuesta de nuevo P.O. 7.4 no toda la generación regula tensión de forma dinámica" (because the proposed new P.O. 7.4 had not been approved, not all generation regulates voltage dynamically), and that "La actualización del P.O.1.4 está pendiente de aprobación desde el año 2021", meaning the update to P.O. 1.4 had been awaiting approval since 2021 (REE report, p.15).

ENTSO-E adds a point about enforcement: "The requirements applicable to the conventional generators' Q output do not include specifications for dynamic behaviour and there is no economic consequence if Q output is not aligned with the 75% rule" (final briefing, slide 15). Here "Q" is reactive power.

Each side's best case, in its own words

Red Eléctrica. About the state of the grid at noon, REE wrote that "nada hacía presagiar ni predecir remotamente los sucesos que acontecieron": nothing remotely foreshadowed what happened (REE report, p.3). It concluded that "no ocurrió nada relevante en el sistema entre las 00:00 h y las 12:00 h digno de mención", that nothing worth mentioning happened between midnight and noon (p.3). Its English press release of 18 June 2025 puts the central claim this way: "The SO's analysis concluded that the generation currently subject to P.O. 7.4 (which requires dynamic voltage regulation) did not comply with the established obligations. So, on 28 April, it did not absorb the reactive power as required. It is essential to note that on 28 April, like any other day, the SO made the requisite calculations and the relevant decisions regarding the programming of technical restrictions, always assuming that all groups satisfied the technical provisions outlined in the regulations." It also called the first disconnections "incorrect, as plants were outside the voltage range determined for these actions under current regulations". After the final report, REE said the panel's conclusions "evidencian que no hay incumplimiento ni conducta atribuible a Red Eléctrica que pudiera ser la causa del cero eléctrico del 28 de abril": they show no breach or conduct by Red Eléctrica that could have caused the blackout (press note, 20 March 2026, p.1).

The government committee. It found the collapse was due "con toda probabilidad, a una combinación de condiciones … no habiéndose identificado un 'fallo único' que pueda explicar por sí solo la caída del sistema": in all probability a combination of conditions, with no single failure that explains the fall on its own (p.64). Its first conclusion is aimed at the operator: "El número de grupos acoplados con capacidad de control de tensión era inferior a los programados por el operador del sistema en semanas y meses previos, e inferior al programado el día antes al haber fallado la tarde del 27 de abril uno de los grupos previstos y no haber sido sustituido por el operador del sistema." In English: fewer voltage-controlling units were connected than the operator had scheduled in previous weeks and months, and fewer than it had scheduled the day before, because one planned unit failed on the evening of 27 April and the operator did not replace it (p.131). Its next conclusion is aimed at the generators, and it is hedged. The plants paid to control voltage "podría no haberse ajustado a los parámetros fijados", they might not have kept to the set parameters (p.131). Some disconnections "pudieron no estar ajustadas", may not have been correct (p.133).

The generators. Aelec is the association of Iberdrola, Endesa and EDP. According to the newspaper Ara (23 June 2025), Aelec presented a report that day, prepared by Inesc Tec and Compass Lexecon as a reply to REE's report. Ara says it held REE solely responsible, blamed a serious failure of voltage control, and said "Las fluctuaciones del 28 de abril no fueron normales" (the fluctuations of 28 April were not normal). This post did not read that report; it is known here only through Ara. A separate Aelec document dated 28 July 2025 could not be retrieved: the address returned an error. Inside the government report, the generators' case appears in one line: "Ninguno de los agentes ha reportado haber detectado o programado el 28 de abril un comportamiento de los grupos térmicos distinto a días o épocas anteriores", meaning no party reported having seen or scheduled any different behaviour from the thermal units on 28 April than on earlier days (p.111). The owner of the first installation to trip, at Granada, "manifiesta que la protección estaba correctamente ajustada y que funcionó adecuadamente": it says its protection was correctly set and worked properly (p.90).

ENTSO-E's expert panel. "Not a single root cause, but a combination of several key factors resulted in the system failure" (final briefing, slide 12). The panel lists six root factors, all labelled "Spain". They are a "Low margin between voltage limit and disconnection voltage on 400kV-level"; "No reactive power reaction on voltage changes by Renewable generators"; "Voltage control support by conventional generators not effective"; lower line loading that changed how much reactive power the lines needed; "Shunt reactors operated manually requiring processing time and not re-connected"; and "Loss of reactive power-absorption of generators" (slide 12). The panel also states that the report "is not intended to allocate or imply any liability or responsibility to any party" (slide 4).

The timeline, from the documents

22 April, 19:00. An overvoltage episode on the transmission grid "provocó la desconexión de más de una [redacted] algunas subestaciones de evacuación de generación [redacted] y grandes consumos industriales". It disconnected some substations that carry generators' output to the grid, along with large industrial consumers (government report, p.118). The report's chart (p.119, Gráfico 50) shows 400 kV reference points peaking a little above 440 kV around 19:02. That is this post's reading of the chart; the text gives no figure. Parties to the inquiry also cited earlier episodes, on "31 de enero, el 19 de marzo y el 22 y 24 de abril" (pp.24, 118).

The report on 22 April. "Tras advertir los impactos de este evento, desde el Ministerio para la Transición Ecológica y el Reto Demográfico solicitó al Operador del Sistema informe de lo ocurrido. Dicho informe se recibe el 28 de abril poco antes de las 11h." (p.119.) In English: after noting the event's impact, the Ministry asked the operator for a report, and the report was received on 28 April shortly before 11:00. The committee summarises what REE explained. The causes were common on their own but coincided that day; they included "La reducida generación convencional, térmica e hidráulica, acoplada en el sistema", the small amount of conventional thermal and hydro generation connected (pp.119–120). The committee also records that "REE también manifiesta que la situación del día 22 hubiera sido distinta si esta generación renovable (fotovoltaica y eólica) controlase la tensión": REE also said the 22nd would have gone differently if solar and wind had controlled voltage (p.120). In public, according to the committee, REE said 22 April could not be linked to 28 April (p.120). In fairness to REE, the Ministry's director general of energy policy, Manuel García Hernández, testified (as reported by El Debate, Unai Mezcua, 23 October 2025) "Solicité información de lo ocurrido al operador del sistema", and that the operator "En ningún momento manifestó que hubiera un riesgo inminente para el suministro": at no point did it say there was an imminent risk to supply. That piece did not report the delivery time.

27 April, evening. A voltage-control unit in the southwest broke down and was declared unavailable at 19:47, until midnight on 30 April. At 20:36 "el Operador del Sistema decide mantener acoplada una [redacted]", the operator decided to keep a unit connected that was already running. That unit was scheduled "de 0:00 h a 2:00 del 28 de abril por control de tensión en Andalucía (y entre las 8:00 y 9:00, por reserva)", from midnight to 2:00 for voltage control and from 8:00 to 9:00 as reserve (p.22). ENTSO-E's account: "In the morning of 28 April, the Spanish TSO performed security assessments and concluded that it was not necessary to replace it for the rest of the day" (factual briefing, slide 14). TSO stands for transmission system operator, which here is REE. The day-ahead plan had "10 grupos térmicos (3 grupos nucleares y 7 ciclos combinados de gas)", ten thermal units of which three were nuclear and seven gas-fired combined-cycle plants, scheduled for voltage control, two of them in the southwest (government report, p.21).

28 April, morning. Voltage was steady to 06:00, varying by less than 5 kV (p.24). "Es a partir de las 10:30 h cuando se aprecia una mayor excursión", from 10:30 the swings grew larger (p.25). ENTSO-E's factual briefing gives the same time, "First significant variations of the voltage from 10:30 CEST", on a chart that also carries the label "Within the normal range of operations" (slide 12). Three small 0.2 Hz oscillations produced voltage swings of 4 kV at 10:32, 7 kV at 11:03 and 6 kV at 11:23, measured from peak to trough (government report, p.29). Those are the amplitudes of three small oscillations, not the size of the morning's excursions. REE's own report says voltages fell to about 10 kV below 400 kV around 11:00 and then "se recuperó subiendo unos 30 kV" (recovered, rising about 30 kV) between 11:04 and 11:09 (REE report, p.2). At 11:00, demand jumped by "unos 1.200MW sin causa conocida", about 1,200 MW with no known cause (p.101). Between about 11:04 and 11:09, REE reports, rail (ADIF) transformers tripped at the Terrer and Rueda de Jalón 400 kV substations. REE calls this "un preludio de lo que pudo ocurrir posteriormente en otras instalaciones", a prelude to what may have happened later elsewhere (REE report, p.2). The committee notes that the owner says those trips were for undervoltage (government report, p.67). The government report redacts the Zaragoza installations (p.67), so equating them with the two substations REE names is this post's inference.

The committee records that several parties called REE's control centre "solicitando información sobre las variaciones de tensión observadas", asking about the voltage swings they were seeing (p.26). The next ten or so lines of the report are redacted.

12:03. An oscillation at 0.6 Hz (cycles per second) began and took "4 minutos 42 segundos" to die out (p.30). Voltage at Almaraz swung 31.2 kV and dipped "ligeramente por debajo del umbral de 375 kV", slightly below the 375 kV threshold (pp.31–32). The operator cut exports to France, switched the HVDC link to France to a fixed-power mode, connected circuits and disconnected four reactors because voltage had dropped (pp.33, 37). HVDC is a high-voltage direct-current line.

12:19. A second oscillation, at 0.2 Hz, in the East-Centre-West mode, in which Iberia and Turkey swing against central Europe, died out in 3 minutes 20 seconds (pp.34–35). ENTSO-E's final briefing says the mitigations were "successful in damping the oscillations" and that they were "naturally leading to voltage increase (still within operational range)" (slide 16).

12:18 to 12:26. The operator ordered a new synchronous unit, and it could not arrive in time. "REE dio orden de acoplar generación térmica con capacidad de control de tensión por consigna, buscando el grupo que acoplaría más rápido en la zona sur. Resultó ser el [redacted] el que dio el tiempo más corto, 1 hora y 30 minutos … Sin embargo, nunca llegó a acoplarse pues el cero de tensión se produjo antes." (p.35.) In English: REE ordered the fastest-connecting thermal unit in the south. The shortest time on offer was an hour and a half, and the unit never connected because the blackout came first. In REE's own account, another gas plant in Andalusia needed two to two and a half hours (REE report, p.7).

12:30. Generation was "un 82% de generación renovable, un 10% nuclear", 82% renewable and 10% nuclear, and "había 11 centrales térmicas acopladas con obligación de regular tensión por consigna", 11 thermal plants connected with a duty to regulate to a setpoint (government report, p.38). ENTSO-E: "At 12:32 about 58% of reactive power capacity of Spanish shunt reactors were connected"; "Remaining capacities could not be connected in time due to needed processing time for manual switching" (final briefing, slide 19).

12:32:00 to 12:32:57. Voltage climbed "de manera casi lineal", almost in a straight line: from 413 to 428 kV at Olmedilla and from 411 to 424 kV at Arroyo de San Serván, "en 57 segundos" (government report, p.39). That is (428 − 413) ÷ 57 = 0.26 kV per second at Olmedilla, and (424 − 411) ÷ 57 = 0.23 kV per second at Arroyo de San Serván. Small generation tripped in the same minute. The committee reports that "el operador del sistema ha identificado pérdidas … por 525 MW", "de los cuales 317 MW provendrían de generación distribuida menor de 1 MW", of which 317 MW would have come from distributed generation below 1 MW (p.41). So 525 MW and 317 MW are the operator's figures as the committee reports them. The report's 18-row table identifies the other 525 − 317 = 208 MW. The rows add up to 2.6 + 21.6 + 5.9 + 4.8 + 2.9 + 19.2 + 3.4 + 22.4 + 55.6 + 2.3 + 0.6 + 13.3 + 1.1 + 11.9 + 20.0 + 4.8 + 10.0 + 6.0 = 208.4 MW (pp.41–42), so the table and the text agree.

12:32:57.140, Granada. The first large trip: 355 MW (p.42). REE puts the transmission voltage at the time "inferior a 418 kV, es decir, dentro de rango", below 418 kV and therefore within range (REE report, p.9). ENTSO-E gives "417.9 kV in the 400 kV transmission side" (factual briefing, slide 19).

12:33:16 to 12:33:18. Badajoz lost about 727 MW, then Sevilla and other sites lost more, in a window ENTSO-E puts at "approximately 930 MW (or more than 1,100 MW, as implied by the variation of frequency)" (factual briefing, slide 19). Adding figures: about 525 MW of small losses (factual briefing, slide 18) + 355 (government report, p.42) + 727 + 928 (final briefing, slide 8) = 2,535 MW. That matches the factual briefing's "Loss of 2.5 GW generation in Spain" up to 12:33:18 (slide 18). At 12:33:18.380 one plant tripped at "443,8 kV, superando por tanto la tensión que están obligadas a soportar las instalaciones con conexión a 400kV", above the voltage that 400 kV installations must withstand (government report, p.47).

12:33:19.620. Spain lost synchronism with continental Europe (p.48): its grid stopped turning in step with the rest of the continent. That was 12:33:19.620 − 12:32:57.140 = 22.48 seconds after the Granada trip. REE gives "22,5 segundos" (REE report, p.12). Load shedding followed, then the last AC line to France opened at 12:33:21.535 (government report, p.55).

The zero. The sources disagree by a few seconds. The government report gives "12:33:30" (p.3). REE writes "A las 12:33:24 h el sistema eléctrico colapsa" and puts the 400 kV voltage below 1 kV at 12:33:27.300 (REE report, p.13). ENTSO-E says "at 12:33 CEST" (final briefing, slide 6).

Testing the claims

Was the morning a warning? REE says nothing worth mentioning happened before noon. The government committee says "múltiples agentes han calificado como atípicas y extraordinarias" the morning's voltage swings: several parties called them atypical and extraordinary (p.66). It reads them as "evidencias compatibles con una ausencia, de suficiente capacidad de control dinámico de tensión", evidence consistent with too little dynamic voltage control (pp.66–67). The two national reports agree on one fact. The government report says that, "si bien en las primeras horas de la mañana del evento se vieron variaciones relevantes de tensión" (although relevant voltage variations were seen in the early hours of the morning of the event), 400 kV voltages "parecen mantenerse dentro de los niveles marcados por los procedimientos de operación 1.1 y 1.3 … en todo el periodo hasta las 12:30h" (they appear to stay within the levels set by operating procedures 1.1 and 1.3 throughout the period up to 12:30; p.27, see also p.37). ENTSO-E's factual briefing reports significant variation from 10:30 on a chart that also carries the label "Within the normal range of operations". So the dispute is over what the swings meant, not whether voltage stayed within limits. The documents cannot settle what the operator made of the swings at the time, because the record of the calls to its control centre is redacted. They also do not say who read the 22 April report that reached the Ministry shortly before 11:00, or whether anyone acted on it. The Ministry's director general of energy policy has said the operator never told the Ministry there was an imminent risk to supply (El Debate, 23 October 2025); what the report itself said is not public in the documents read.

Did REE schedule too few voltage-controlling units? The government's figure of 11 thermal units at 12:30 is the strongest point against REE. The committee's paragraph runs in this order. First a comparison limited to one redacted zone: "Atendiendo a situaciones similares a la del día 28 de abril en cuanto a grupos acoplados en [redacted] … no parecen identificarse situaciones problemáticas para la operación y con correctos niveles de tensión (días 31/03/2025, 01/04/2025 al 07/04/2025 o 27/04/2025)". In English: on days similar to 28 April in the units connected in that redacted zone, no operating problems appear and voltage levels were correct, the days including 27 April. Then the national count: "En cualquier caso, como resultado de las circunstancias descritas y de las decisiones adoptadas por el operador del sistema, el número final de grupos térmicos acoplados a las 12:30, justo antes del evento, fue de 11, cifra que no se ha registrado en ningún otro día este año. Desde el 1 de enero de 2025, hubo 12 grupos térmicos acoplados en 13 días [redacted], superando esa cifra el resto de los días." In English: in any case, as a result of the circumstances described and of the decisions taken by the system operator, the final number of thermal units connected at 12:30 was 11, a figure seen on no other day that year; there were 12 on 13 days and more on all the rest (pp.109–110). The zonal comparison is therefore not a comparison of the national count. The committee's count also leaves out hydro: "En este análisis no se incluye la generación hidráulica acoplada" (p.110).

REE, citing the factual report, says that between 12:00 and 13:00 "más de 40 grupos convencionales (térmicos e hidráulicos) de más de 30 MW" had a duty to control voltage dynamically: more than 40 conventional thermal and hydro units above 30 MW (REE statement, 3 October 2025, p.2). The two figures do not contradict each other. One counts thermal plants only, the other counts thermal and hydro units. Neither settles whether 11 was too few.

How many hours the kept-connected unit covered is not clear either. The government report gives the scheduled hours as midnight to 2:00 and 8:00 to 9:00 (p.22), which adds up to 2 + 1 = 3 hours. Elsewhere the same report says the replacement "solo cubrió las referidas 4 horas (hasta las 9:00)", only covered those four hours, up to 9:00 (p.109). The documents read do not say how the two figures fit.

Did the plants that were connected do their job? REE says no. ENTSO-E says "Some generators diverged from the applicable requirements for 75% alignment of actual Q with Q reference, without prejudice to any potential derogations from these requirements" (final briefing, slide 15). The final clause matters: the panel leaves open that exemptions may apply. The committee found one southern unit behaving "visiblemente distinta", visibly differently (p.111). On 22 April that unit absorbed "más de 3 veces", more than three times, its maximum at 12:32 on 28 April (p.120). None of the documents read for this post explains why.

Were the first trips "incorrect"? REE says the Granada disconnection was wrong because transmission voltage was within the legal range (REE report, p.10). The owner says its protection was set correctly (government report, p.90). The committee's figures show why the two can both believe they are right. It works out the protection's trip point as 70 V × 2,000 × √3. With √3 ≈ 1.7321, that is 242,494 V, about 242.5 kV. (The report prints √3 as 1.73, which gives 242.2 kV, but it states 242.5.) The report says this "se corresponde con más del 110% de la tensión nominal de 220 kV" (corresponds to more than 110% of the 220 kV nominal voltage; p.91). The post's own check: 242.5 ÷ 220 = 1.102, or 110.2%. The report says the "al menos 242,5 kV" would correspond to "440 kV (16 kV más que los reportados por REE)", 440 kV, 16 kV more than REE reported (p.91). The post's own calculation at the nominal 220/400 ratio of 0.55 gives 242.5 ÷ 0.55 = 440.9 kV. The committee sets its 440 kV against a 424 kV figure from REE's data. That 424 kV is a post-trip value. REE writes, in the same sentence that calls the trip incorrect, "Prueba de ello es que tras el disparo las tensiones pasan a estar en 424 kV" (REE report, p.9), and the government report gives 418 kV "en el instante previo al disparo, y en 423,9 kV en el instante posterior" (pp.89–90). So the committee's comparison sets REE's post-trip voltage against the protection's trip reading. Run the other way, 424 × 0.55 = 233.2 kV, about "10 kV menos" than the protection read (p.91). Using the pre-trip 418 kV instead, 418 × 0.55 = 229.9 kV, which is 242.5 − 229.9 = 12.6 kV below the 242.5 kV trip point. The committee calls these "datos divergentes" that "podrían ser compatibles con un disparo 'temprano'", divergent data that could be consistent with an early trip (pp.91, 93). It also offers another reading: "podría ser compatible con una regulación o actualización de las tomas de transformador menos rápida de lo necesario". In English, the transformer's tap changer (its device for adjusting the voltage ratio) may have moved more slowly than needed (p.91). The owner could not check this. On 10 June it reported "no disponer de información precisa de los cambios de tomas durante el día 28 de abril", that it had no precise record of the tap changes on 28 April (p.92). ENTSO-E's general finding is "Many protection settings diverged from applicable requirements or were not aligned with system needs" (final briefing, slide 20).

How much room was there? Spain's operating ceiling on the 400 kV grid is 435 kV, and generators must ride through 440 kV for 60 minutes (government report, pp.64, 173). That is a gap of 440 − 435 = 5 kV, or 5 ÷ 400 = 1.25% of nominal. This post infers that this is the margin ENTSO-E calls "Low margin between voltage limit and disconnection voltage on 400kV-level". The panel's own figure for the margin is in the final report, which was not read. The briefing notes on the 435 kV figure: "These values may occur on occasion at the 400 kV level (confirmed by CNMC, Spanish regulator)" (slide 13). The CNMC is Spain's energy and competition regulator.

Inertia, renewables and cyberattack. Inertia is the energy stored in spinning machines, which slows how fast frequency can change. The committee puts it at "2,3 s" against a 2-second ENTSO-E reference (pp.107–108) and judges it "altamente probable", highly likely, that the cascade would have happened even with more (p.107). REE: "El incidente NO se ha producido por un problema de inercia", the incident was not caused by an inertia problem (REE report, p.16). No account says renewables caused the blackout. What the rules show is that renewables were required to run at a fixed power factor and were not allowed to regulate to a setpoint (government report, p.66). On cyberattack, the committee found "no se han encontrado indicios o evidencias que apunten a un ciberataque o ciberincidente, en las entidades analizadas" (p.129). The last clause limits the finding to the organisations it examined.

Who could see what

The operator could not see small generation. Plants below 1 MW and self-consumption are "no observable directamente desde la red de transporte", not directly visible from the transmission grid (government report, p.84). REE estimates "casi 700 MW" of such losses around 12:22, invisible to it (REE report, p.6). That estimate covers a different time window from the operator's 317 MW, as the committee reports it. Both are inferred from net load at the transmission boundary, not metered, so they cannot be added together. The committee's check with one inverter manufacturer (inverters are the electronics that connect solar panels to the grid) covered the losses after the 12:03 and 12:19 oscillations, around 12:10, 12:16 and 12:27. It does not test the 12:32 minute or the 317 MW estimate. For those phases the losses were "en todo caso inferior al 10%", under 10% of that maker's output. The committee concluded "no puede afirmarse que se produjese la desconexión de una proporción relevante de la generación … al menos en el caso de los inversores de [redacted] reportados": it cannot be said that a relevant share disconnected, at least for that maker's inverters (p.86). Only some of the five inverter groups it asked replied (p.85).

Generators were not always able to see the lines that carried their own output. The committee says downstream generators "no siempre tienen visibilidad ni capacidad de gestión" (do not always have visibility or the ability to manage) the evacuation infrastructure, and that "distintos generadores han reportado no conocer los datos relativos a las medidas y gestión de las infraestructuras de evacuación a las que se conectan" (p.113). In English, several generators reported that they did not know the measurements or operation of the lines they connect through. REE, for its part, says it "lacks sufficient information to assess those [load sheddings] carried out within the distribution network" (press release, 18 June 2025, p.3).

ENTSO-E's full root-cause tree names a gap that links these: "Absence of real time TSO monitoring of difference between reactive power output and Q-reference - no risk identified when voltage approaches critical voltage level" (final briefing, slide 11). The slide is an image, so this is a transcription of it.

Who held the records afterwards

On 26 May 2025, ENTSO-E wrote to REE that the panel needed data "including data owned by third parties (e.g. DSOs and generators)" (letter to REE). On 28 May 2025, ENTSO-E wrote to Spain's energy minister: "Red Eléctrica recently informed us of the reluctance of the third parties to share the relevant data, which could lead to potential delays in the investigation of the Expert Panel. The third parties cite confidentiality requirements to which they would be legally bound." In October it reported "Significant difficulties arose in obtaining high-quality data from several DSOs and generation companies from Spain" (factual briefing, slide 9). DSOs are distribution system operators, the companies that run local networks. The government committee depended on voluntary cooperation and says it still lacked some information when it finished (pp.8, 12). Of the 67 parties it dealt with, 6 asked for confidentiality, 3 required prior approval, 57 gave no express authorisation to publish and 1 allowed publication: 6 + 3 + 57 + 1 = 67 (p.10). In the public version, names are redacted throughout.

On 17 April 2026 the CNMC announced that it had opened sanctioning proceedings after finding signs of non-compliance. It added that the proceedings do not by themselves attribute the blackout's cause to the companies involved (CNMC press release). That release was read only through a summary, so it is paraphrased here.

How to check this yourself

All three main documents are free; their addresses are listed under Sources. Page and slide numbers in this post were taken from text-layer readings of the PDFs, and two such readings disagreed by one. For the two ENTSO-E decks, the post's slide numbers may be one lower than the printed slide number or the PDF page; for the government report, some cited passages may sit one page later than the number given (for example "16 kV más" and "10 kV menos" may be on p.92, not p.91). If a cited item is not on the page named, check the page after it. This post did not open the PDFs in a viewer to settle the offset. Its timeline runs from p.18 to p.62, and its conclusions are at pp.131–133. REE's report is 18 numbered pages, and its account of the morning is on pp.2–7. The 208.4 MW sum, the Granada conversion and the 5 kV margin can each be redone with a calculator from the pages cited.

What could not be established

Sources

All read on 21 November 2026.

  1. Gobierno de España, Consejo de Seguridad Nacional, Versión no confidencial del informe del comité para el análisis de las circunstancias que concurrieron en la crisis de electricidad del 28 de abril de 2025, approved 16 June 2025, published 17 June 2025, 182 pp. https://www.lamoncloa.gob.es/consejodeministros/resumenes/Documents/2025/Informe-no-confidencial-Comite-de-analisis-28A.pdf
  2. Red Eléctrica, Dirección General de Operación, Incidente en el Sistema Eléctrico Peninsular Español el 28 de abril de 2025, 18 June 2025. https://d1n1o4zeyfu21r.cloudfront.net/WEB_Incidente_SistemaElectricoPeninsularEspanol_18junio2025.pdf
  3. Red Eléctrica, press release (English), 18 June 2025. https://www.ree.es/sites/default/files/2025-06/EN_NP%20Red%20El%C3%A9ctrica%20Informe%20OS%20180625_VF.pdf
  4. Red Eléctrica, statement on the factual report, 3 October 2025. https://www.ree.es/sites/default/files/2025-10/Comunicado_Expert_Panel_europeo_confirma__informe_operador_sistema.pdf
  5. Red Eléctrica, press note on the final report, 20 March 2026. https://www.ree.es/sites/default/files/2026-03/NP%20-%20FinalReport%20Expert%20Panel%2020%20marzo.pdf
  6. ENTSO-E, Expert Panel Final Report … 20 March 2026 – Media Briefing, 25 slides. https://eepublicdownloads.blob.core.windows.net/public-cdn-container/clean-documents/Publications/2025/iberian-blackout/260318_Final%20report_press%20briefing%20PPT_Final.pdf
  7. ENTSO-E, Expert Panel Factual Report, 3 October 2025 Press Briefing, 28 slides. https://eepublicdownloads.blob.core.windows.net/public-cdn-container/clean-documents/Publications/2025/iberian-blackout/251003_ENTSO-E-Media%20briefing%20on%20Factual%20Report.pdf
  8. ENTSO-E, letter to Red Eléctrica, 26 May 2025. https://eepublicdownloads.blob.core.windows.net/public-cdn-container/clean-documents/Publications/2025/iberian-blackout/250526_ENTSO-E%20letter%20to%20RE%20for%2028%20April%20blackout%20investigation_final.pdf
  9. ENTSO-E, letter to the Spanish Ministry, 28 May 2025. https://eepublicdownloads.blob.core.windows.net/public-cdn-container/clean-documents/Publications/2025/iberian-blackout/250528_ENTSO-E%20letter%20to%20Spanish%20Ministry_28%20April%20blackout%20investigation.pdf
  10. ENTSO-E, Final Report on the Grid Incident in Spain and Portugal on 28 April 2025, 20 March 2026 (not read in full; listed for readers to check). https://eepublicdownloads.blob.core.windows.net/public-cdn-container/clean-documents/Publications/2025/iberian-blackout/Final%20Report%20on%20the%20Grid%20Incident%20in%20Spain%20and%20Portugal%20on%2028%20April%202025.pdf
  11. CNMC, press release, 17 April 2026 (read through a summary). https://www.cnmc.es/prensa/incoacion-incidente-electrico-28A-20260417
  12. Ara, Carlota Serra Llagostera, 23 June 2025, on the Aelec report prepared by Inesc Tec and Compass Lexecon (secondary). https://es.ara.cat/economia/energia/electricas-atribuyen-apagon-mala-gestion-red-electrica_1_5421074.html
  13. POWER magazine, Aaron Larson, Anatomy of a Blackout: Findings from the Spain-Portugal Grid Collapse Final Report, 23 March 2026 (secondary). https://www.powermag.com/anatomy-of-a-blackout-findings-from-the-spain-portugal-grid-collapse-final-report/
  14. Science Media Centre Spain, expert reactions, 20 March 2026 (secondary). https://sciencemediacentre.es/en/final-blackout-report-european-operators-confirms-event-was-caused-multiple-factors-recommends
  15. El Debate, Unai Mezcua, 23 October 2025, on testimony by Manuel García Hernández, the Ministry's director general of energy policy (secondary). https://www.eldebate.com/economia/20251023/gobierno-cuestiono-red-electrica-problemas-red-dias-antes-apagon-esta-descarto-riesgo-inminente_347706.html