João Graça Gomes
Michel-Alexandre Cardin
Billy Wu
Dyson School of Design Engineering, Imperial College London

Despite our best efforts to forecast the long-term needs of the power sector, our predictions are invariably wrong. Trends change, policies shift, technology advances, and even the most advanced electricity grids can be vulnerable.

Last year’s widespread blackout across the Iberian Peninsula (Portugal, Spain, and Andorra) and parts of southern France is a clear reminder of this. And it wasn’t an isolated incident: Texas froze in 2021, South Australia went dark in 2016, and Tokyo’s recent heat waves pushed grids to the brink.

How the Iberian Grid Collapsed?

The blackout began on 28 April, just after noon, when an abrupt power failure at a substation in Granada, southern Spain, caused a chain reaction.

Two main explanations have emerged. One hints that a miscalculation in the electricity mix caused a voltage surge that destabilised the power grid. According to a Spanish government report, several thermal power plants failed to respond, worsening the situation and triggering a wave of shutdowns. This hypothesis attributes the fault to poor energy planning and management, arguing that there was inadequate backup, insufficient control, and insufficient power plants responding to the emergency.

The alternative explanation acknowledges that a voltage surge was the initial trigger but blames thermal power plants for failing to regulate voltage levels. It also points to anomalies in the disconnection of power plants and an unexpected spike in electricity demand from the transport network.

The question of what went wrong and who’s responsible remains up in the air.

How did they get the power back on?

Restoring power took several hours. Spain recovered faster than Portugal, partly because it has stronger connections to neighbouring countries. In Portugal, the process took around 12 hours, partly because only two power plants could restart the grid from scratch (black-start capacity). The recovery required careful coordination and was helped by the fact that hydroelectric dams were full due to a wet year.

What are the insights?

In the wake of the incident, experts broadly agreed that classic resilience approaches needed renewed attention:

• Building more connections between countries, especially between Iberia, France, and Morocco.
• Increasing the number of power plants capable of black-start operations.
• Upgrade the grid with storage solutions and smart demand management.

These are crucial priorities, but they are not sufficient.

Reevaluating our approach to resilience

The blackout should prompt a broader reassessment of how we think about power system reliability in the 21^st century, and beyond. Ironically, the very success of modern grids, so reliable they are rarely questioned, has lulled us into complacency. We design for what we expect, not for what might blindside us.

In recent decades, the power sector has embraced new approaches to address uncertainty in long-term design and planning. Sophisticated machine learning methods for forecasting energy markets and optimisation methods for generation expansion planning, such as stochastic programming and scenario-based modelling, are becoming part of planning processes in industry and government, especially with the rising integration of variable renewable energy sources.

Yet even with these breakthroughs, significant hurdles remain. Today’s methods may sharpen our view of the future, but they are still anchored to old assumptions and past patterns. When the unexpected strikes, whether in a matter of seconds, weeks, or years, these tools often falter. To build truly resilient power systems, we must go beyond conventional forecasting techniques and start incorporating the capacity to adapt dynamically as events unfold.

Strategic engineering

Picture an energy grid that self-adapts overnight to a hurricane or a market swing. This is where strategic engineering comes in (fig.1). Strategic engineering advocates that systems should be designed from early conceptual activities to be more adaptive and flexible, in the sense that they can change, evolve, and reconfigure to manage uncertainty and risks. It promotes embracing uncertainty early on in design and deploying an architecture enabling system operators to make value-enhancing decisions over time, as opposed to locking the system down to a particular design or technological path. This could mean designing assets that can be scaled up or down as demand shifts, systems that can switch between different fuel sources as technology evolves, or modular infrastructures that can be deployed incrementally or reconfigured as part of an emergency response plan. These “real” options are not just theoretical; they are already proving extremely valuable in many industry sectors like aerospace, transportation, and water management, and it’s time for the energy sector to do the same.

Figure 1: Strategic Engineering mindset.
Source: Michel-Alexandre Cardin, https://strategic-engineering.co/blog/concepts/engineering-for-uncertainty/, 2025.

Decentralised and flexible solutions

Strategic engineering is paving the way for a future powered by decentralised energy systems and markets. By embracing modularity, it uncovers economic and social benefits that traditional appraisal tools like net present value analysis often overlook. Decentralised solar PV, for instance, holds remarkable promise for boosting resilience, yet most systems remain grid-tied and shut down during outages. Empowering critical sites such as hospitals, military bases, and government facilities with off-grid capabilities, backed by batteries or a mix of wind, solar, and diesel power, can keep essential services running when the grid fails. While these adaptive investments may require a higher initial investment, their long-term rewards in resilience, security, and sustainability can far surpass the original costs.

How can technology help, and are we truly ready for the next grid shock?

These questions set the stage for exploring how immersive technologies like virtual reality and serious-gaming can play a critical role in supporting better design and planning in the future. Advanced control systems and real-time analytics can enable faster, smarter responses during grid disturbances by providing an environment to carefully train operators on dynamic adaptation plans. Integration with virtual replicas of physical systems can allow them to simulate and prepare for rare events, stress-testing scenarios that traditional planning might overlook, as is already being done in Denmark.

The path forward

Creating a truly resilient power system is about more than just reinforcing wires or speeding up repairs. It is about weaving networks that learn, shift, and react instantly to whatever surprises arise. Tomorrow calls for plans that bend without breaking, shaped by the wisdom of past crises and energised by the promise of fresh ideas. We can’t predict every challenge the future will bring, but by pooling global research and collaboration, we can design our energy systems to handle whatever comes next.