Simulating Reserve Power Systems in Future Climates: Bias Adjustment Approaches for Regional Climate Projections

Critical infrastructure, such as telecommunications networks and hospitals, are in many cases required to have reserve power systems in place, mitigating transmission network failures and protecting against national power grid outage. A previous case study of Great Britain (GB) telecommunications assets implemented a temperature-driven model of infrastructure electricity demand (Fallon et al., 2023), used to plan reserve capacity installation sufficient to meet the highest anticipated 5-day periods of energy consumption (or other regulatory targets). Extending this work with climate models (UKCP18), we demonstrate that the capacity planning framework reliant upon reanalysis observations underestimates capacity installation appropriate to meet historic weather risk, while assessments are improved using historic period climate model outputs. Additionally, climate projections simulating future periods support further upgrading the installed reserve capacity beyond historic requirements.

Quantile-correcting bias adjustments of climate model outputs can address significant discrepancy between the model world and observations temperature distributions across the historic period (model timespan where global climate matches recent observations). Uncorrected, this climate model error leads to an exaggerated frequency of extreme temperature events, hence overestimating the reserve capacity requirement. But under a quantile-correcting approach, assuming a consistent underlying representation of the weather dynamics, the temperature distribution is adjusted to match the reanalysis distribution.

Temperature delta-shifts are calculated to represent the GB historic period climate variability observed across model ensemble members. The resulting infrastructure electricity demand timeseries are compared against timeseries produced from historic period temperature data adjusted by quantile delta mapping, demonstrating that reanalysis data alone is insufficient to capture the greater reserve capacity requirements predicted by quantile delta-mapping of climate model outputs in the historic time period.

Using future period climate model outputs, we compare three alternative treatments of model temperature timeseries simulating future climate: a delta-shift adjustment of reanalysis data, a regional trend-preserving mean bias adjustment, and quantile delta mapping. In each case, reserve capacity requirements increase (5% to 10% increase in a world 2.0°C above pre-industrial temperatures). There is significant variability across different model ensemble members, and sensitivity to individual weather years.

Reserve system operators can use the approaches outlined to make an informed assessment of the need for upgrading or installing new reserve systems, ensuring the stability and resilience of critical infrastructure assets. The consistent trajectories across different approaches and model ensemble members may improve confidence in results, whilst individual model ensemble members can be investigated to identify potential ‘worst case’ outcomes.