Taking the politics out of power generation strategy
How would a rational AI algorithm choose to optimise the path to a net zero electricity grid?
Following the open data publication of the Digest of UK Energy Statistics 2024 (DUKES), we ran an experiment to gain some new insights into GB’s power generation options, using a systems thinking approach and AI modelling techniques.
The future demand for electricity in the UK is highly uncertain. Not knowing the future demand makes it harder to make robust policy decisions on the best balance of power generation technologies that should supply our future grid.
Any future power mix needs to ensure security of supply, i.e. it needs to be capable of meeting demand at all times of the day, whatever the weather, or plant maintenance requirements. Alongside this uncertainty sits the imperative to decarbonise electricity generation, which currently accounts for around 15% of UK carbon emissions. Finally, given the above, we want to ensure that the solution provides the best value for money to consumers. This challenge is present internationally and is known throughout the world as the Energy Trilemma.
"The Energy Trilemma – how to balance security of supply, emissions, and cost."
Andy Moore, Group Leader for Strategic Modelling, explains:
"We used a systems modelling approach to explore how the UK power industry could securely decarbonise and at what cost. Applying AI techniques to the problem helped us to quickly explore a range of decarbonising strategies and optimise the balance between investment cost, security of supply and overall emissions for potential future demand scenarios."
Our aims were to:
- Test the usefulness of AI on a complex policy challenge.
- Calculate some “no regret” options to inform power generation policy.
In other words, what generation technologies should we encourage in the next 5‑10 years that we won't regret 15‑20 years later.
Here’s what we found:
- AI, and probabilistic modelling techniques, can be applied to complex and uncertain problems like the Energy Trilemma to generate robust sets of strategic options. The AI technique applied here generates new solutions quickly, which could allow more scenarios to be explored, new technologies tested, or policy options analysed faster.
- The AI model is successful at reducing the carbon emissions of the grid mix, whilst the costs of electricity show a slight reduction in current generation costs. It achieves this whilst matching demand for each year modelled.
- The best options over the period to 2050 focus on increasing the capacity of nuclear, wind and solar power. Reducing the generating capacity of fossil fuels is feasible while still meeting demand. The AI determined that building more nuclear capacity than current plans can mitigate the risk of uncertain future demand.
You can view and download a copy of our report below.
