On 28 April 2025, it took just 27 seconds for the Iberian peninsula’s electrical system to collapse, leaving millions without power. At 12:32:57 CET, generation trips totalling approximately 2200 MW began in southern Spain. By 12:33:24 CET, the system had completely collapsed. The Spanish government report[1] and ENTSO-E Expert Panel investigation[2] found that voltage instability had been building throughout the morning, and that when the critical moment arrived, some generator disconnections occurred before voltage thresholds set by regulations had been exceeded.
Whilst the official reports detail protection relay settings, frequency deviations, and system recovery timelines, they rarely capture a more fundamental challenge: the persistent gap between how power systems are expected to respond to disturbances and how they actually perform — and why this gap continues to surprise the industry.
In this article, we will explore the existing challenges and the practical steps we can take to address them.
When regulation lags technology
Three fundamental factors drive the widening gap between power system evolution and regulatory frameworks:
- System complexity is outpacing our analytical approaches. Twenty years ago, grid disturbance analysis focused on a limited number of large synchronous generators with well-understood dynamic characteristics. Contemporary systems comprise thousands of distributed inverter-based resources, each with distinct control algorithms and response characteristics. The wide range of meteorological conditions, evolving load patterns (electric vehicles, data centres), and diverse market strategies deployed by participants significantly increase the complexities of operating bulk power systems in a stable and secure manner.
- Generator behaviour is becoming increasingly software defined. Modern power systems operate with reduced margins for error, requiring every generator to perform precisely to specification. However, generator behaviour is now increasingly determined by software and firmware rather than only by physical characteristics. Software-defined generation can be reconfigured with a firmware update, potentially changing grid support capabilities in a flash. Many grid codes still lack provisions for verifying that these critical settings remain compliant months or years after commissioning.
- The pace of deployment exceeds the pace of learning. Each generation of technology arrives before the previous generation’s performance in real grid events can be fully assessed and incorporated into updated standards. This pattern repeats globally: deploy technology; experience grid events; analyse performance; update grid codes; repeat. Each cycle takes years, whilst technology deployment accelerates, creating an ever-widening gap between what grid codes require and what modern power systems actually need.
The disconnect between planning and reality
In planning studies, grid events are modelled with clinical precision. Generator trips are instantaneous. Protection operates exactly as specified. Operational experience consistently reveals a messier reality.
Consider the wind farm response shown in Figure 1. During a grid disturbance involving circuit breaker recloser operation — two consecutive faults within seconds — the facility’s actual behaviour demonstrates a complexity absent from planning models. Active power fluctuates throughout the event sequence in response to changing wind speed. Reactive power exhibits sharp transient responses during each fault, with magnitudes and recovery patterns that no simplified dynamic model would predict and would be challenging even for a detailed EMT (electromagnetic transient) model.
Real grid events involve multiple overlapping transients, equipment and neighbouring generators responding to conditions that exist for fractions of a second, control systems reacting to rapidly changing voltage and frequency signals.
Planning studies model single, clean disturbances. Reality delivers a complex, sometimes unpredictable, sequence of events in which each system response influences the next.
This disconnect represents a fundamental risk in modern power systems. When grid codes are written based on simplified models, and compliance is verified through mathematical modelling and one-time commissioning tests, the gap between assumed and actual performance during real grid stress events remains unknown — until a major disturbance reveals it.
The transition from centralised power stations to more dispersed renewable generation has also dispersed technical expertise across multiple organisations, contractors, and frequently rotating personnel (Figure 2).
A few decades ago, asset knowledge was concentrated in specialist teams that understood equipment throughout its multi-decade operational life. Contemporary arrangements have made institutional knowledge more fragile despite advances in data collection and processing capabilities.
The Australian experience
Australia confronted its own grid reliability challenges following the 2016 South Australia blackout.[3] The subsequent response has been systematic, including: a significant uplift in generator performance standards and assessment methodologies; the widespread deployment of grid-scale battery energy storage systems (Figure 3) and grid-forming inverters; and the redesign of operational frameworks for technical requirements such as primary frequency response.
There are also requirements for comprehensive monitoring and testing programmes demonstrating ongoing compliance with technical standards.[4] High-resolution data from actual grid events reveals patterns that planning studies miss. Continuous monitoring can identify power plants operating in incorrect control modes or with inappropriate setpoints — appearing compliant on paper but configured in ways that could result in an unstable grid.
Similarly, periodic testing (eg, on plant change or every 3-5 years) provides confidence that generators can operate under stressed conditions.
This approach acknowledges that assumptions about generator performance cannot remain unchallenged for years following commissioning; there’s simply too much at stake.
Implications for grid codes and operations
Many grid codes need to be updated to manage contemporary grid realities, where inverter-based resources constitute substantial and growing proportions of generation capacity and where generator behaviour is increasingly defined by software rather than purely by equipment characteristics. Addressing these challenges requires co-ordinated improvements:
- Grid codes must be modernised to reflect the technical capabilities and limitations of inverter-based resources, moving beyond frameworks designed for synchronous machines.
- Planning studies should acknowledge that we have an integrated grid with many complex equipment types interacting and responding to each other, not isolated components behaving according to textbook models.
- The commissioning paradigm must extend from one-time testing to continuous compliance verification. Continuous monitoring enables identification of configuration drift and performance degradation — problems that develop gradually over months or years but manifest catastrophically during disturbances.
- Equally critical is the preservation of institutional knowledge. Design decisions, operational constraints, and lessons learned from grid events must be systematically documented to ensure critical information survives personnel transitions.
- Finally, enhanced co-ordination between asset owners, network operators, and system operators is essential to ensure grid support capabilities are not merely specified but actively maintained throughout asset lifecycles.
Mind the gap
The gap between modelled and actual grid behaviour is not narrowing as power systems evolve. It is widening. The proliferation of inverter-based resources, increasingly complex control systems, and fragmented institutional knowledge create conditions in which the next major grid event will almost certainly reveal additional blind spots in current planning and operational practices.
The question is not whether these vulnerabilities exist — recent experiences have demonstrated that they do. The question is whether the industry will implement systematic approaches to identifying and addressing them before the next 27-second cascade demonstrates the cost of complacency.
* Aditya Upadhye is managing director of VeriConneX and a director at GridWise Energy Solutions. He has over 20 years of power systems engineering experience across Australia and the United States, with specific expertise in grid modelling, analysis and generator compliance. At VeriConneX, Aditya leads the development of the COMET platform, Australia’s leading generator compliance monitoring solution, with over 2.5 GW of assets under management.
References:
[1] Spanish government, Report from the committee for the analysis of the electricity crisis of April 28th 2025, 17 June 2025 [2] ENTSO-E Expert Panel, Grid incident in Spain and Portugal on 28 April 2025, 3 October 2025 [3] AEMO, Black system South Australia 28 September 2016 – final report, March 2017 [4] National Electricity Rules, Clause 4.15
