The Iberian Blackout: A Black Swan or a Gray Rhino? A Protection-Aware Dynamic Voltage Security Assessment
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Robotics Radio. Generated commentary on the latest robotics and control papers.
Rosa: We're looking at the paper: "The Iberian Blackout".
Dev: Here's the gist of the paper. On 28 April 2025, the Iberian mainland power system collapsed after a rapid voltage rise, widespread generation disconnections, and loss of synchronism The paper develops a protection-aware dynamic voltage security assessment for this question This method is designed to screen, rank, and mitigate the vulnerability class exposed by the blackout from quantities that operators can know or bound at run time The Core Problem Addressed This paper uses the Iberian blackout as a motivating case for a broader operational voltage security problem: given the present grid state, can the next plausible trip, ramp, topology action, or shunt action push protected downstream voltages above relay thresholds before available voltage controls can respond The objective is to screen, rank, and mitigate the vulnerability class exposed by the blackout from quantities that operators can know or bound at run time The difference between this problem and classical voltage security is that conventional voltage stability studies focus on QV curves, reduced Jacobians, continuation power flow, modal analysis, and long term load and on-load tap-changing transformer (OLTC) dynamics The objective should instead answer a time critical question: how much of each protected device’s overvoltage margin is consumed by the next plausible event, and can available controls restore the margin before the relay acts Modeling and Mechanism Identification The paper starts from a nonlinear hybrid differential-algebraic equation (DAE) model to derive mode wise finite window voltage maps that include automatic voltage regulators (AVRs), inverter-based resources (IBRs), static synchronous compensators (STATCOMs), high-voltage direct-current (HVDC) links, loads, shunts, transformers, protection functions, and limiter behavior whenever the corresponding models are available The mechanism of interest is a high voltage operating condition in which plausible events raise the voltage seen by downstream or collector side protection The vulnerability class addressed is broader than the Iberian event and narrower than voltage stability in general, specifically focusing on "protection-driven overvoltage cascades with reactive balance positive feedback" characterized by three properties Screen Development and Operational Workflow The proposed method develops a protection-aware dynamic voltage security assessment for this mechanism The screening algorithm is structured around a mode wise linear model where the system is described by x˙ = f m(x, y, u, d; θ) and 0 = gm(x, y, u, d; θ) It computes the normalized pickup and trip erosion matrices K¯ m,pk and K¯ m,tr using an additive threshold map Φm(S) The mitigation optimization solves a linear program to compute preventive fast reactive actions that restore margin Validation and Operational Utility Case studies on a 2000-bus mechanism replica and multiple dynamic benchmark systems show that the screen predicts nonlinear cascade propagation, ranks risky operator actions, identifies missing-data limits, and computes useful fast-control actions before nonlinear simulation is run The method has four limitations including a finite event library and the conservative nature of the additive cascade screen The operational output provides a ranked list of protected locations with low or uncertain margin, seed events that can cause relay pickup or delayed trip, controls with positive lower bound, residual slack η⋆, and data limited flags The realized trajectory had black swan features: two oscillation episodes, operator countermeasures, fixed power factor ramps near a schedule boundary, local protection behavior, missing data, and fast generation disconnections interacted in a way that was difficult to infer from any single indicator The vulnerability class was a gray rhino A system is exposed to this mechanism when protected side voltage margins are small, plausible next events can consume those margins through loss of reactive absorption or topology dependent voltage rise, and available controls cannot reduce the protected voltage within the relay window The contribution of the proposed framework is to turn that distinction into an operational screen References
Paper summary: Dev: So wrapping up this look at "The Iberian Blackout: A Black Swan or a Gray Rhino? A Protection-Aware Dynamic Voltage Security Assessment" and the authors Abdallah Alalem Albustamia and Ahmad F Tahaa <ref:2511.17433#pg1>, they’re asking if this system is exposed to that specific protection-driven cascade mechanism <ref:2511.17433#pg2>.
Rosa: The title itself sets up the tension between a sudden, unexpected crash, the black swan event, and this more insidious, slow erosion of safety that looks like a gray rhino <ref:2511.17433#pg2>.
Taro: What they’re saying is that when protected side voltages are small and plausible next events can consume those margins through reactive loss or topology-dependent voltage rise, and your available controls can't stop it within the relay window, you have a problem <ref:2511.17433#pg2>.
Dev: The authors’ main contribution is making that distinction operational. They developed this protection-aware dynamic voltage security assessment to answer the question of whether a real-time or planning tool could have detected that the system was becoming protection-fragile even if conventional static checks still looked fine <ref:2511.17433#pg2>.
Rosa: It’s about moving beyond just looking at QV curves or modal analysis for long term load dynamics, and focusing instead on how much of each protected device’s overvoltage margin is consumed by the next plausible trip or ramp action <ref:2511.17433#pg2>.
Taro: The implications are that we need tools that look at this immediate, time-critical question: can available controls restore the margin before the relay actually triggers <ref:2511.17433#pg2>.
Dev: They give us a ranked list of risky locations and seed events, plus data-limited flags and residual slack, which is what operators need to know before things go south <ref:2511.17433#pg2>.
Rosa: It turns a complex stability concept into something you can screen for risk at run time <ref:2511.17433#pg2>. That’s what this paper is about.
Conclusion: Rosa: So we're wrapping up this look at "The Iberian Blackout: A Black Swan or a Gray Rhino? A Protection-Aware Dynamic Voltage Security Assessment" and its authors Abdallah Alalem Albustamia and Ahmad F Tahaa.
Dev: It’s about whether that sudden blackout was just a one-off accident, or if it reveals a deeper, persistent vulnerability in the grid's safety margin.
Rosa: Exactly. The paper is trying to figure out if this event is a black swan—something totally unpredictable—or something more like a gray rhino, which means something we can predict and plan for if we know where to look.
Dev: So they’re using that Iberian blackout as the test case for a bigger question: can the next plausible trip or load change push our protected equipment past its safety limit before our controls can react?
Rosa: Right. They aren't just looking at long-term stability curves or how transformers age; they're asking a time-critical question about what happens in the next few milliseconds.
Dev: The methodology involves building a model that includes everything—inverters, static compensators, protection functions—and then creating a screening algorithm that tells us which parts of the system are most likely to fail under these conditions.
Rosa: And they develop an optimization part where they figure out what fast actions operators can take right now to bring those protected voltages back into safe territory.
Dev: The results show that this approach can actually predict those nonlinear cascade effects, rank the most risky operator actions, and even find data limits where we just don't have enough information to make a call.
Rosa: It gives us a way to turn that complex stability concept into something operational—a real tool for planning.
Dev: The authors flag some limitations, like using a finite library of possible events and the fact that their screening approach is quite conservative, but the output they give you is useful right now.
Rosa: So it’s not a magic fix, but it gives operators a ranked list of places with low or uncertain safety margins and tells them exactly what kind of seed event could cause a relay to trip prematurely.
Dev: It turns that distinction into something we can actually screen for risk in real time, which is pretty significant for grid management.
Rosa: It’s about moving from just understanding the physics of stability to having a practical way to manage the immediate dangers exposed by events like this one.
Vanderbilt University
eess.SY, cs.SY
Submitted: 2025-11-21
Updated: 2026-10-31
License: http://creativecommons.org/licenses/by/4.0/
Importance score: 77/100
The gist: On 28 April 2025, the Iberian mainland power system collapsed after a rapid voltage rise, widespread generation disconnections…
Key concepts
- Protection-Aware Dynamic Voltage Security Assessment
- A method designed to screen, rank, and mitigate voltage vulnerabilities by predicting how potential next grid actions (like trips or ramps) will affect protected downstream voltages relative to relay thresholds.
- Protection-Driven Overvoltage Cascades with Reactive Balance Positive Feedback
- The specific vulnerability class studied. It describes a dangerous chain reaction where plausible events cause voltage rises that trigger protection systems, which in turn causes further voltage changes that reinforce the initial problem.
- Mode-wise Linear Model
- A mathematical framework used for screening. The system is simplified into different operational modes, allowing the researchers to compute matrices (like pickup and trip erosion) that show how much margin each protected device loses under various conditions.
- Gray Rhino
- A risk classification suggesting a threat that is not immediately obvious but has the potential to cause significant damage. In this context, it means a system is exposed if small margins are consumed by plausible events because available controls cannot restore the voltage within the relay's time window.
Terminology
Summary
On 28 April 2025, the Iberian mainland power system collapsed after a rapid voltage rise, widespread generation disconnections, and loss of synchronism The paper develops a protection-aware dynamic voltage security assessment for this question
Motivation and Problem Statement
The paper uses the Iberian blackout as a motivating case for a broader operational voltage security problem: given the present grid state, can the next plausible trip, ramp, topology action, or shunt action push protected downstream voltages above relay thresholds before available voltage controls can respond The objective is to screen, rank, and mitigate the vulnerability class exposed by the blackout from quantities that operators can know or bound at run time This paper develops a protection-aware dynamic voltage security assessment to answer that question
Mechanism of Vulnerability
The mechanism of interest is a high voltage operating condition in which plausible events (generation trips, fixed power factor ramps, shunt actions, topology changes, exchange reductions, load or pump disconnections, or local protection operations) raise the voltage seen by downstream or collector side protection The vulnerability class addressed in the rest of the paper is broader than the Iberian event and narrower than voltage stability in general It is the class of protection driven overvoltage cascades with reactive balance positive feedback, characterized by three properties (i) cascade propagation through loss of reactive absorption after each trip, (ii) binding voltage measurement on the protection side (collector, downstream, or transformer LV), and (iii) a relay relevant time window short enough that fast dynamic controls can mitigate but slow controls cannot
Modeling Framework
The paper starts from a nonlinear hybrid differential-algebraic equation (DAE) model to derive mode wise finite window voltage maps The model is hybrid at the parent level because relay trips, shunt operations, tap changes, control mode changes, and limiter activations change the network and device equations For a given assessment window, the mode fixes the discrete statuses that cannot change inside that window The linearized DAE gives a reduced ODE for dynamic states and protected voltages
Screening and Mitigation
The proposed method is a screening layer between the energy management system (EMS), real time contingency analysis (RTCA), and full nonlinear RMS simulation It ranks plausible next events by their protected margin impact, predicts protection driven propagation through a threshold cascade model, issues robust no trip or data limited certificates, and computes preventive fast reactive control actions when sufficient control authority exists The mitigation optimization solves a problem to compute the minimum fast reactive action needed to keep protected voltages below relay thresholds
Validation and Operational Value
The case studies show the operational value of this structure, including reproducing the key feedback identified in the final report on a 2000 bus mechanism replica The screen produced no unsafe false negatives across 140 validated seed events spanning all four systems The triage scaling is the operational complement of the safety result, focusing validation effort on the head of the risk distribution where detailed simulation has the highest operational value The method is an advisory screening layer between EMS/RTCA and full nonlinear dynamic simulation
Limitations
The method has four limitations, including a finite event library, being mode wise over one assessment window, the additive cascade screen being conservative, and dependence on protection and controller data The method is not a replacement for nonlinear RMS simulation but focuses validation effort where it is most valuable
Conclusion
The paper developed a protection-aware dynamic voltage security assessment for overvoltage cascades driven by loss of reactive absorption and protection side voltage margin erosion The case studies show the operational value of this structure The Iberian blackout was used as the motivating event but the method targets a broader operating condition: a high voltage state in which plausible trips, ramps, topology actions, shunt actions, load or pump disconnections, or local protection operations can push relay measured voltages above pickup before available controls respond The contribution of the proposed framework is to turn that distinction into an operational screen
References
[2] ENTSO-E Expert Panel, Iberian blackout of 28 april 2025: incident report, Tech. rep., ENTSO-E (2025)
[1] ENTSO-E Expert Panel, Final report on the grid incident in spain and portugal on 28 april 2025, Tech. rep., ENTSO-E (Mar. p. 0 Tahaa
aVanderbilt University, Civil and Environmental Engineering Department, Nashville, 37235, TN, USA
Abstract
On 28 April 2025, the Iberian mainland power system collapsed after a rapid voltage rise, widespread generation disconnections,
and loss of synchronism. The ENTSO-E Expert Panel final report attributes the blackout to multiple interacting factors including ineffective voltage control, fixed power factor reactive behavior, fast generation ramps, protection settings not aligned with
requirements, slow or unavailable reactive absorption, and limited observability outside the transmission system.Where the paper says it
[3] Ü. Çetinkaya, M. Ye¸sil, R. Bayındır, E. Irmak, Technical analysis and strategic insights from the 2025 spain blackout, in: 2025 13th Interna20 tional Conference on Smart Grid (icSmartGrid), IEEE, 2025
[4] L. Rouco, F. Echavarren, E. Lobato, The overvoltage-driven blackout of the iberian peninsula on 28th april 2025, Sustainable Energy, Grids and Networks (2026)
[5] H. Morão, The 2025 iberian peninsula blackout: Lessons for modern power systems and policy implications, Utilities Policy 101 (2026)
[6] P. Kundur, et al., Power system stability, Power system stability and control 10 (1) (2007)
[7] C. W. Taylor, Power system voltage stability, McGraw-Hil, 1994
[8] B. Gao, G. K. Morison, P. Kundur, Voltage stability evaluation using modal analysis, IEEE transactions on power systems 7 (4) (2002)
[9] T. Van Cutsem, C. Vournas, Voltage stability of electric power systems, Springer Science & Business Media, 2007
[10] P. Kundur, J. Paserba, V. Ajjarapu, G. Andersson, A. Bose, C. Canizares, N. Hatziargyriou, D. Hill, A. Stankovic, C. Taylor et al., Definition and classification of power system stability ieee/cigre joint task force on stability terms and definitions, IEEE transactions on Power Systems 19 (3) (2004)
[11] N. Hatziargyriou, J. Milanovic, C. Rahmann, V. Ajjarapu, C. Canizares, I. Erlich, D. Hill, I. Hiskens, I. Kamwa et al., Definition and classification of power system stability–revisited & extended, IEEE Transactions on Power Systems 36 (4) (2020)
[12] M. Glavic, T. Van Cutsem, Wide-area detection of voltage instability from synchronized phasor measurements. p. 1 The difference between this problem and classical voltage
security is that conventional voltage stability studies focus on
QV curves, reduced Jacobians, continuation power flow, modal
analysis, and long term load and on-load tap-changing transformer (OLTC) dynamics [6, 7, 8, 9]. More recent work extends these tools to wide area detection of voltage instability
and to a broader stability classification that explicitly accounts
for converter-driven dynamics [10, 11, 12].Where the paper says it
[13] I. Dobson, Estimating the propagation and extent of cascading line outages from utility data with a branching process, IEEE Transactions on Power Systems 27 (4) (2012)
[14] P. Hines, J. Apt, S. Talukdar, Large blackouts in north america: Historical trends and policy implications, Energy Policy 37 (12) (2009)
[15] B. A. Carreras, V. E. Lynch, I. Dobson, D. E. Newman, Complex dynamics of blackouts in power transmission systems, Chaos: An Interdisciplinary Journal of Nonlinear Science 14 (3) (2004)
[16] Vaiman, Chen, Chowdhury, Dobson, Hines, Papic, Zhang, Risk assessment of cascading outages: Methodologies and challenges, IEEE Transactions on Power Systems 27 (2) (2011)
[17] J. Bialek et al., Benchmarking and validation of cascading failure analysis tools, IEEE Transactions on Power Systems 31 (6) (2016)
[18] F. Milano, F. Dörfler, G. Hug, D. J. Hill, G. Verbic, Foundations and ˇ challenges of low-inertia systems, in: 2018 power systems computation conference (PSCC), IEEE, 2018
[19] P. Tielens, D. Van Hertem, The relevance of inertia in power systems, Renewable and sustainable energy reviews 55 (2016)
[20] L. Badesa et al., Ancillary services in great britain during the covid-19 lockdown: A glimpse of the carbon-free future, Applied Energy 285 (2021)
[21] A. A. Albustami, Iberian blackout: Protection-Aware Dynamic Voltage Security Assessment, https://github. p. 0 The Iberian Blackout: A Black Swan or a Gray Rhino? ⋆
A Protection-Aware Dynamic Voltage Security Assessment
Abdallah Alalem Albustamia
, Ahmad F.Where the paper says it
[22] R. H. Lasseter, Z. Chen, D. Pattabiraman, Grid-forming inverters: A critical asset for the power grid, IEEE Journal of Emerging and Selected Topics in Power Electronics 8 (2) (2019)
[23] J. Matevosyan et al., Grid-forming inverters: Are they the key for high renewable penetration? p. 1 These structural changes have been
identified across the low-inertia and energy-transition literature
as drivers of new dynamic stability challenges [18, 19], and
operational records from systems with very high instantaneous
non-synchronous penetration document the resulting strain on
ancillary service procurement and system stability [20]. These
conditions do not make renewable generation the direct cause
of the Iberian blackout.Where the paper says it
[24] H. Cui, F. Li, K. Tomsovic, Hybrid symbolic-numeric framework for power system modeling and analysis, IEEE Transactions on Power Systems 36 (2) (2020)
[25] A. B. Birchfield et al., Grid structural characteristics as validation criteria for synthetic networks, IEEE Transactions on power systems 32 (4) (2016) p. 1 synchronous units online in some areas, lighter loading on extra high voltage lines, faster active power ramps, more fixed
power factor operation, and less direct visibility of plant side or
distribution side voltages. These structural changes have been
identified across the low-inertia and energy-transition literature
as drivers of new dynamic stability challenges [18, 19], and
operational records from systems with very high instantaneous
non-synchronous penetration document the resulting strain on
ancillary service procurement and system stability [20].Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Protection-Aware Dynamic Voltage Security Assessment System: An AI system can take an EMS snapshot, protection data, candidate events, and control actions to compute robust no trip certificates or identify high risk seeds by using a mode wise linear model and finite window maps.
-
Cascade Prediction Module: The system can predict the propagation of overvoltage cascades by iteratively applying the additive threshold map Φm(S) to rank seed events based on their predicted fixed points, which are
the least fixed point containing S 0.
-
Mitigation Optimization Engine: The AI can solve a constrained optimization problem to compute preventive fast reactive actions, minimizing control effort and residual slack η, by using time resolved control coefficients Hmik(t) derived from the finite window maps.
-
Uncertainty Quantification Layer: The system can quantify data uncertainty by calculating robust pickup and trip erosion matrices Kbar pk and Kbar tr using suprema over an uncertainty set Theta, flagging assets as data limited if the margin hmi cannot be certified positive.
-
Risk Triage Prioritization: The AI can rank candidate events based on their protected margin impact, using the
worst normalized margin erosion Kbar pk per action family
to focus subsequent expensive nonlinear validation simulations on the head of the risk distribution.
Abstract
On 28 April 2025, the Iberian power system collapsed after a rapid voltage rise that led to widespread generation disconnections and loss of synchronism. The final report of the ENTSO-E Expert Panel attributes the blackout to several interacting factors including ineffective voltage control, fixed power factor operation, misaligned protection settings, manual shunt reactor switching, and limited observability. These factors raise a general question: given the current state, can the next trip, ramp, topology action, or shunt switching raise the voltages measured by overvoltage relays above their settings before voltage controls respond? We answer it with a protection-aware dynamic voltage security assessment. For each candidate event, the method computes the voltage at every overvoltage relay over the relay time window from a dynamic model of the network, its controls, and its relays. It predicts which relays trip after a first trip, identifies assets whose relay or measurement data are insufficient for a safety conclusion, and, when one exists, computes a minimum-cost fast reactive power action that keeps all relay voltages below their settings. Case studies on four benchmark systems and a 2000-bus system that reproduces the mechanism of the Iberian blackout show that the method flagged every event that caused a secondary trip within the relay window, never declared safe a plant that tripped within that window, and, on the benchmarks, found fast reactive power actions that prevented the trips within the relay window whenever it judged such actions feasible. These results were validated with nonlinear time-domain simulations. The code is publicly available.
Related papers
- Industrial Load Modeling and Optimization for Market-Based Interaction with Power Systems
- Exact and Approximate Convex Reformulation of Linear Stochastic Optimal Control with Chance Constraints
- Certified Set Convergence for Piecewise Affine Systems
- Symbolic Control for Stochastic Systems via Finite Parity Games
- Multiobjective optimization-based design and dispatch of islanded, hybrid microgrids for remote, off-grid communities in sub-Saharan Africa
- Physiology-informed layered sensing for intelligent human-exoskeleton interaction