SIF R3 Discovery - Probabilistic Pathways for Energy System Planning

Planning for an inherently uncertain future is complex and time consuming, significantly benefiting from the ability to quantify risk within planning decisions, and analyse more pathways.

This project will develop an enhanced end-to-end network planning methodology for the whole energy system. We will explore applying advanced computational techniques, such as artificial intelligence and probabilistic modelling, to capture risk and uncertainty within future energy pathways, enable rapid iterative network needs analyses, risk-based network options assessments, and deliver optimised planning decisions.

Innovation Justification

Innovative Aspects and Activities

Supply and Demand Modelling: Current network planning methodologies are underpinned by Future Energy Scenarios (FES), developed via complex analyses to determine credible pathways of supply and demand to 2050. FES analysis currently generates deterministic outputs, with limited quantification of uncertainties inherent in inputs and assumptions used.
This project will revolutionise FES methodology, developing enhanced risk-based network planning processes. It will implement probabilistic techniques to capture uncertainty within the parameter space and account for how it propagates through the process; enabling increased confidence in outputs, based on lowest balance of cost and risk.

Network Needs: Currently, FES outputs are used to determine network requirements and capability over a 10-year period, identifying the need for network reinforcement. The detailed energy system analysis required to establish reinforcement needs is computationally and resource intensive, so only the most onerous pathway is currently analysed in detail.
The project will explore how reduced-order surrogate models can lower computation timescales, thereby maximising the number of pathways analysed, enabling an iterative analysis process.

Network Design: Currently, Transmission Owners (TOs) propose network design options with the potential to address the network needs identified by NGESO. These are assessed by NGESO for efficiency and cost-effectiveness which are based on fluctuant future costs, adding further uncertainty. The project will develop probabilistic cost-benefit analysis for long-term strategic plans for network development, with reinforcement learning and AI to analyse the greatest opportunity for strategic prioritisation of initiatives that will accelerate whole energy system development and operability.
Project Scale

It will consider the end-to-end planning process to develop AI-augmented flexible tools and probabilistic methodologies to de-risk the range of potential pathways. This will reduce resource requirements and increase the depth of analysis, meeting scales needed for future whole energy system planning. TRL, CRL and IRL for this project is 4 but anticipated progression to 5 during Discovery.

SIF Funding

This project aims to address limitations in deterministic models and enhancing them with probabilistic modelling and state-of-the-art AI techniques. The high level of innovation, whole-energy system benefit and natural progression through Discovery to Beta makes SIF the most appropriate funding mechanism for this project.

Counterfactuals

The project will assess several advanced modelling techniques, applied across the entire future whole energy system planning process, opposed to a single technique and application. The counterfactual of assessing one technique provides fewer benefits, compared to the ambitious and innovative simultaneous integration of multiple complementary techniques.

Impacts and Benefits

1. New processes: The pre-innovation baseline relies on mostly deterministic approaches, from FES to cost-benefit analysis, with only a single, most arduous pathway explored in greatest detail.
Integration of probabilistic and advanced AI modelling with new holistic planning processes will improve confidence in network design decisions, by embedding uncertainty analysis through the end-to-end process, and by increasing the quantity and range of pathways analysed in detail.

Probabilistic modelling will give insight to the likelihood of each pathway being realised, highlighting policy gaps. This will support the FSO advisory role, and subsequently produce clearer recommendations for efficient implementation by policy makers.

2.Operation Costs: Pre-innovation baseline uses deterministic outputs to inform network reinforcement options. Instead, our approach allows risk quantification, associated with investment and planning decisions. This enables adoption of long-term plans, which optimally balance cost, risk and performance across all credible pathways, whilst simultaneously empowering true least-regrets investment decisions. For example, the initial Accelerated Strategic Transmission Investment framework encompassed 26 projects, worth roughly £20bn. This equates to £200m benefit for every 1% reduction unlocked through risk-based approaches.

Also, increasing analysis speed, as we are proposing, allows for the range of policy options to be better understood and compared, supporting cost-effective policy implementation.

3.Consumer Costs: Facilitation of faster low-cost and low-carbon generation connections, risk-based decision-making and more efficient network operation will reduce consumer bills. Reducing curtailment through strategic grid capacity improvements and/or storage connection would also reduce bills, as about £800m relating to curtailment was added to bills in 2022.

Running different pathways more quickly helps to support long-term development planning based on a balance of multiple factors, including societal costs such as consumer bills.

4.Network Service Costs: Lower risk planning decisions will allow better coordination of network services and give greater confidence to supply chains.

Enhanced long-term development planning will permit more strategically located flexibility, improving cost-effectiveness of network services. In 2021, Carbon Trust reported flexibility as a no-regrets decision, providing potential net savings up to £16.7bn/annum.

Reduced-order models which allow decreasing analysis times, will result in better informed policy decisions and market arrangements, such as targeted incentivisation and highlighting the existing gaps in market requirements.

5. Indirect Carbon Reduction: Holistic planning processes will improve connection times for low carbon energy resources, whilst a risk-based planning approach will help to optimise the strategic siting locations for distributed storage and flexibility providers, reducing curtailment of low carbon energy generation.

ENA smarter networks portal

Impacts and benefits

The innovation would have allowed the efficient development of multiple probabilistic future energy pathways. This would support long-term network investment decision-making that could accommodate multiple factors in a consistent manner and promote a more optimal allocation of National Grid’s planned £118bn pot for network investment (over the 10 years to 2035). It would allow for risk quantification, improved risk-based investment and planning decisions, and promote the balancing of costs and performance across all credible pathways that are identified. This would facilitate faster, low-cost and low-carbon generation connection, improving the efficiency of network operation, cutting consumer bills and lowering grid carbon content. 
 
For the Discovery Phase, the projected benefits were assessed through consideration of a case study that considers the benefits that would have arisen if the transmission system upgrades had been brought forward by 6 years, balanced against the implementation costs of the proposed advanced modelling techniques. Two variations were examined: 

• The first assumed a 100% reduction in Scottish wind curtailment, assuming that all curtailment results from capacity constraints on B6. The net present value (NPV) of this is £3.8bn over 45 years, with an associated carbon saving of 11.8m tonne CO2e.
• The second assumed a 50% reduction in Scottish wind curtailment. This has an NPV of £1.9bn, avoiding curtailment of at least 28 TWh of Scottish wind between 2023 and 2028, which equates to a reduction in emissions of at least 5.9m tonnes. 

 
Bringing forward upgrades to B6 is the type of network decision that might be informed by an enhanced planning process, which this project aimed to address. 
 
Operationally, the proposed modelling approaches would facilitate the rapid production of future energy system scenarios. This element of the benefits would have been reported in terms of the number of pathways NESO would be able to analyse with the same resources, relative to the pre-innovation baseline of three.