Designing weather-informed stress tests scenarios for net-zero Energy systems

Energy systems across the world are rapidly evolving to meet climate mitigation targets. This requires lower reliance on fossil fuels and more weather-dependent renewable generation (such as wind power, solar power, and hydropower). This increased weather dependence adds a new set of challenges for balancing supply and demand due to the inherent variability of weather, increasing the need for investment in storage and flexible technologies. Both in terms of security of supply risks from system level challenges (e.g., energy shortfall events, where existing generation is insufficient to meet demand) or from smaller-scale infrastructure challenges (e.g., extreme weather impacting the operability of energy system components) there is the need for stress testing of new power system configurations.  

A challenge for this stress testing is existing power system models are often limited to running single-year simulations, and there is therefore a need to be able to subset years of different challenge levels (e.g. different return period levels of weather-driven stress) that may cause weather-driven stress. Existing methodologies to explore weather-driven stress translate large volumes of gridded meteorological data into demand and renewable generation timeseries which are analysed, often in terms of demand-net-renewables. However, this involves significant interdisciplinary training in energy and climate impacts modelling and large volumes of storage space to convert many decades of data into demand-net-renewables for a robust stress test selection.  

In this study we extend previous work where weather-driven risks to the European energy sector in both a present and future climate have been explored, with a particular focus on the timing, duration, and severity of energy shortfall events [1]. We consider three methods of choosing a stressful year based on demand-net-renewables. These are: 

1. A year with a short duration extreme event. 

2. A challenging year. 

3. A year of challenging large-scale weather. 

The first two types of stress test are defined using weather-driven demand, wind power and solar power timeseries, whereas the final type of stressful year involves matching the most commonly occurring weather-patterns [2] from a historically challenging year to those that are occurring in a climate model, therefore bypassing the need to convert all of the climate model data into energy system variables. 

We demonstrate results from recent stress testing exercises using state-of-the-art outputs from the UK climate projections (UKCP18) 2.2km convection permitting model. We show how from an energy-systems perspective the most challenging short duration extremes are often not contained within the most challenging year, and that this distinction between types of stress needs to be driven by user needs.  

We also show that using traditional large-scale weather to subset the stress test event does not lead to the highest impact energy-stress events contained within a large sample of climate data. It does save a significant amount of processing time for users wishing to stress test a system for a ‘reasonably challenging’ event. 

 

[1] Bloomfield, H. (2025). Reasonable worst-case stress-test scenarios for the UK energy sector in the context of the changing climate. Available at: https://www.theccc.org.uk/publication/reasonable-worst-case-stress-test-scenarios-for-the-uk-energy-sector-in-the-context-of-the-changing-climate/ 

[2] Pope, J. O., Brown, K., Fung, F., Hanlon, H. M., Neal, R., Palin, E. J., & Reid, A. (2022). Investigation of future climate change over the British Isles using weather patterns. Climate Dynamics, 58(9), 2405-2419.