Voltage, Not Frequency, Led the 2025 Iberian Blackout Cascade, Analysis Finds

At approximately 12:33 Central European Summer Time on 28 April 2025, the mainland power systems of Spain and Portugal collapsed while the wider Continental European synchronous area remained largely intact. A new open-access analysis in the journal Carbon Neutral Systems examines the blackout from the perspective of converter technology and argues that the available public records point to a voltage-led propagation sequence, in which voltage excursions and generation disconnections preceded the steepest fall in frequency. The study, led by Hans-Peter Beck with colleagues including Michael Zhengmeng Hou and Jianfeng Liu, links converter behaviour, published laboratory experiments, reported event ordering and engineering options for grid reinforcement.

The pre-event operating state was distinctive: low demand, high photovoltaic output, a large share of non-synchronous generation and relatively few synchronous units online, according to the final report of the European Network of Transmission System Operators for Electricity and an analysis by Compass Lexecon and INESC TEC. The ENTSO-E report identifies interacting oscillations, deficiencies in voltage and reactive-power control, generation disconnections and a rapid voltage rise during the blackout. The Compass Lexecon and INESC TEC work examines the withdrawal of combined-cycle gas turbine units, limited reactive-power absorption, overvoltage warnings and protection-related generation loss. Together these findings raise a locational question that the authors place at the centre of their analysis: was sufficient voltage control and dynamic support available to keep generation connected?

The event records show rising voltages in the Spanish 400 kV transmission network in the minutes before the collapse. At low load, line charging can inject reactive power, and voltage rises when available absorption and voltage-control resources are insufficient. High-voltage warnings increased relative to the previous Monday and appeared across the control environment, rising by approximately 3.1 times on the blackout day. The authors stress that estimating short-circuit ratios or reactive-power margins would require network and plant-level data beyond these warning counts, but the warnings document widespread voltage stress before the final collapse.

The temporal sequence is central to the argument. Frequency traces for Málaga, Belfort and Porto were initially close to 50 Hz before the Spanish and Portuguese traces declined rapidly. The annotations identify generation losses in Granada at 12:32:57, Badajoz at 12:33:16 and Seville at 12:33:17, followed by loss-of-synchronisation protection and a cascade of generation losses. When voltage and frequency are placed on the same timeline, voltage excursions and generation losses appear before the most severe frequency decline. The authors emphasise that this ordering is consistent with voltage-led propagation but does not establish a unique initiating cause or reconstruct individual plant responses; individual trip causes require plant-specific evidence that remains unavailable.

The mechanism the authors examine is a coupling between voltage stress and generation loss. A disconnected unit removes active power and, depending on its operating state, reactive-power absorption or other voltage-control support. Disconnection of a unit absorbing reactive power can aggravate voltage rise, whereas losing a unit injecting reactive power can have a different effect. The response depends on pre-trip set points, network flows and location. The authors formalise this as a conditional feedback hypothesis using a linearised reactive-power-to-voltage sensitivity, a phenomenological disconnection equation and the classical swing equation, while explicitly noting that no event-specific protection threshold, voltage sensitivity or converter parameter is estimated and that the equations provide a conceptual framework rather than a calibrated model of the Iberian system.

Device-level evidence comes from published laboratory experiments on virtual synchronous machine control, known as VISMA, which embeds a synchronous-machine model in converter control. In one test, a conventional grid-following photovoltaic inverter supplying a resistive load continued operating for approximately 150 milliseconds after separation from the grid emulator, before voltage and current decreased and the local grid collapsed about 170 milliseconds after separation. In a run-down experiment with a VISMA unit, the switching transient lasted approximately 5 milliseconds, with active- and reactive-power settling over roughly 20 milliseconds. A separate controlled-islanding study using two identically parameterised VISMA systems reached a new steady state after about 40 milliseconds, with the virtual-rotor frequency near 49.91 Hz and voltage below its initial nominal value because voltage regulation was disabled. These experiments show why adequate active-power supply alone does not establish that a converter can maintain a local voltage reference, though they do not reproduce the Iberian overvoltage cascade or identify the controls of the affected plants.

The engineering assessment distinguishes three transmission-scale technologies. Static synchronous compensators, or STATCOMs, provide dynamic reactive-power exchange and voltage regulation. Extended STATCOMs with energy storage, or E-STATCOMs, add supercapacitors for deliberate short-duration absorption and injection of active power. High-voltage direct-current links obtain active energy through the DC connection to another AC system. All three can use grid-forming control, and their energy and power limits determine the magnitude and duration of frequency support. The authors highlight the Mehrum Nord installation near Hanover in northern Germany, dated by Siemens Energy to 2025 and described as the first E-STATCOM in operation, comprising a 400 kV connection, an SVC PLUS platform rated at ±300 Mvar and a 200 MW supercapacitor unit with 400 megajoules of available energy. Dividing the energy by the power rating gives an idealised scale estimate of about 2 seconds, without establishing a guaranteed discharge duration. The authors are explicit that these manufacturer-reported ratings do not establish that the installation would have prevented the Iberian blackout.

Spain’s 2025 system statistics frame the continuing challenge. Year-end installed power capacity reached approximately 142.6 gigawatts, including storage but excluding self-consumption facilities, and renewables supplied 55.5 percent of generation excluding estimated self-consumption, or 56.6 percent including it. National renewable generation excluding estimated self-consumption reached 150,988 gigawatt-hours, including 58,801 gigawatt-hours from wind, 50,188 from solar photovoltaics, 33,685 from hydropower and 3,682 from solar thermal. Spain added roughly 10 gigawatts of wind and solar capacity in 2025, with 8.8 gigawatts of photovoltaics and 1.2 gigawatts of wind. Demand rose 2.8 percent to 256,086 gigawatt-hours. Storage systems integrated 9,213 gigawatt-hours, 6.2 percent more than in 2024, and installed storage power reached 3,427 megawatts. Spain was a net electricity exporter for the fourth consecutive year with a net export balance of 12,794 gigawatt-hours. The authors caution that annual totals cannot establish blackout causality or dispatch-level stability margins.

Operational and regulatory changes followed the event. The average number of online combined-cycle gas turbine and nuclear units between 10:00 and 18:00 increased from 14 before the blackout to 24 afterwards, although the count alone does not distinguish voltage-control requirements from scheduling or market conditions. The revised Operational Procedure 7.4, approved in June 2025, expanded participation in voltage control; Red Electrica authorised the first renewable installations capable of following dynamic voltage set points in October 2025, and a joint industry statement reported full implementation on 17 March 2026. During 2025 Spain also added 212 substation bays and 486 kilometres of line circuit, bringing the transmission network to 46,155 kilometres, and increased transformer capacity by 1,455 megavolt-amperes to 99,071 megavolt-amperes.

The study’s conclusion is that renewable-rich systems require coordinated voltage-control margins, dynamic reactive-power resources, electrical reference formation, short-duration active-power support and protection, alongside adequate generation and operational flexibility. Local voltage support emerges as a priority because absorbing line-charging reactive power during low-load conditions requires nearby resources; active-power reserves elsewhere cannot provide the same service. The authors note the limits of their analysis: it uses public reports and system-level data rather than a full dynamic simulation of Spain and Portugal, plant protection settings and converter parameters remain partly observable, and manufacturer material on Mehrum Nord lacks complete control parameters and independent disturbance measurements. Future work, they suggest, should reconstruct regional power flows and voltages before 12:33 on 28 April 2025 and test whether current limits and protection settings create correlated disconnection risk.

Subject of Research: Converter-technology analysis of the 28 April 2025 Iberian blackout and grid reinforcement needs

Article Title: The 28 April 2025 blackout in the Spanish interconnected power system from a converter-technology perspective and the need for grid reinforcement

Article References: Beck, H.-P., Hou, M. Z., Wang, Q., Zhang, T., Guo, Y., Huang, L., Shi, T., Wang, J., & Liu, J. (2026). The 28 April 2025 blackout in the Spanish interconnected power system from a converter-technology perspective and the need for grid reinforcement. Carbon Neutral Systems, 2(1), Article 20. https://doi.org/10.1007/s44438-026-00043-1

Image Credits: AI Generated

DOI: 10.1007/s44438-026-00043-1

Keywords: Iberian blackout, Spain, power system stability, grid-forming converters, STATCOM, reactive power, voltage control, ENTSO-E, VISMA, E-STATCOM, renewable energy, HVDC