Project Summary
As DNOs develop their distribution system operator functions, the current annual process used to forecast load at extra-high-voltage/high-voltage needs to become increasingly granular, at the monthly, weekly, daily and hourly level, to support flexibility dispatch and defer or avoid reinforcement. Moreover, the increasing prevalence of Low-voltage monitoring data enables new use cases to support network planning and the extension of flexibility markets at ED3. The Artificial Forecasting project will address these unmet needs by building innovative AI solutions to expand load forecasting capability at primary (EHV-HV) and secondary (HV-LV) substations, enabling the development of DSO functions across the sector.
Innovation Justification
Challenge: preparing for a net zero power system: supporting the safe and reliable operation of a net zero power system by 2035.
The project directly addresses the theme area of accessing grid or system support from novel supply and demand side sources, by developing AI forecasting solutions to inform flexibility procurement. It satisfies the eligibility criteria on funding, partner contribution, start date and duration, including partners who are aggregators with a portfolio of demand-side resources including electric vehicles charging points and demand side response. Faculty, as a disruptor in data science, will lead development of the technical solution.
Lessons for Alpha: Discovery provided strong evidence to continue into Alpha, reflecting stakeholder feedback. Consideration of the innovation landscape, user research and exploratory data analysis confirmed the user needs, deliverability and risks associated with developing these forecasting capabilities. The project scope has been refined to reflect these findings.
Stakeholder engagement: NPg's planning and operation functions depend on accurate/timely data, and internal users from forecasting, digitalisation and flexibility were consulted in Discovery to validate and refine use cases. External stakeholders are increasingly concerned with the information NPg's forecasting data can provide, flexibility service providers have been consulted in Discovery and two partners are joining the Alpha Phase as a result.
Innovation: This project rigorously implements state-of-the-art AI methods for load forecasting. DNOs have carried out innovation work in load forecasting (NPg NIA_NPG_012: Improving Demand Forecasting, ENW NIA_ENWL_020 Artificial Intelligence and Machine Learning, UKPN NIA_UKPN0070 Envision, and SSEN SSEN_0 Transition).
This project tests a wider set of AI methods relative to previous work, reflecting state-of-the-art machine-learning forecasting techniques. It also undertakes a novel approach to forecasting HV net demand, by separately estimating gross demand and distributed generation. This overcomes a significant shortfall in existing studies which treat net demand as a single time series; methods that will become inaccurate as distributed generation assets increase in scale and volume. It also develops forecasts using LV monitoring data, a nascent area of work across the sector. The appendix sets out these methods considered in further detail.
TRL, IRL, & CRL
TRL 4; by testing the technical/commercial application of forecasting models.
IRL 2; Alpha will define how forecasts will be integrated into NPg systems
CRL4/5; Alpha will develop the Value Proposition and refinement of the product hypothesis
Size and Scale: AI has already impacted ways of working; however, much can still be learnt. Network operators have identified where AI can increase capabilities and begun to test applications. Innovation is needed to de-risk future applications of AI in utilities, by building up the evidence base of successes and failures. AI adoption in the medium-term will benefit from due consideration at this point. The scale of the project meets SIF requirements and is proportional to the requirements to rigorously test AI solutions.
Funding/BAU: Along with other licensees, NPg will incorporate AI methods into their processes over the coming years alongside implementation of the Digitalisation Strategy and Action Plan. The benefits from introducing AI methods will be magnified by the foundational work undertaken within this SIF project, which develops capability at a speed and rigour over and above current plans, and potentially unlocks further related innovations (e.g., flexibility digital marketplace).
Counterfactuals: To enable DSO functions, existing annual load forecasting methods must become increasingly granular to support flexibility dispatch. Without further innovation, existing methods could be broadened to create greater spatial and temporal coverage, however it would require a significant increase in workforce to run similar tools to produce the greater volumes of information required. This is not a feasible solution at scale.
Impacts and Benefits
Pre-innovation baseline: The innovative AI forecasting solutions within this project represent a critical enabler of NPg's DSO functions, providing new capability over and above business-as-usual (BAU). The baseline therefore reflects BAU expenditure line items set out in NPg's ED2 DSO strategy from 2023-2028. BAU expenditure is phased using the rate of asset investment as a proxy.
To estimate BAU expenditures over ED3, a conservative uplift of 5% was used to estimate aggregate expenditure relative to ED2, reflecting the additional costs associated with managing an increasingly complex network.
The baseline therefore covers the period 2023-2032, the period for which reasonable foresight can be provided. Where possible, assumptions will be amended at Alpha as new information becomes available (e.g., ED3 planning).
The baseline scenario therefore includes costs/savings line items associated with:
Network interventions, predominantly reinforcement costs
DSO strategy investment in skills and systems
LV monitoring rollout expenditure
DNO-contracted flexibility (savings/expenditures)
Flexibility market stimulation investment
Expenditures associated with asset outages
Tracking: Baseline expenditures/savings through ED2/ED3 can act as an aggregate metric to compare the scale of realised benefits. For network reinforcement decisions and in procuring flexibility, NPg also assesses the relevant opportunity costs (cost of the next-best option), which can track costs/savings enabled by AI forecasting tools relative to alternatives, per intervention. This can provide a bottom-up estimate of the benefits enabled through innovative forecasting solutions.
The core technical performance of AI load forecasts will be benchmarked to a baseline metric per model (e.g., versus linear extrapolations or assumptions-based methods) to demonstrate proof-of-value relative to existing approaches. A tangible success criteria would then represent their adoption within decision-making, including HV flexibility procurement and LV operational planning.
Quantified benefits to date include:
i) Reduced costs of operating the network
Reinforcement: Accurate HV/LV load forecasts enable more efficient flexibility procurement and targeting of reinforcement spend, estimated at 3% of HV/LV reinforcement capex. (£27m benefit total 2023-32, pre-discounting and inflation)
Outages: accurate forecasts enable more efficient procurement of market-based solutions and targeted asset maintenance, avoiding asset outages (£11m).
Flexibility: Operational forecasting enables dynamic procurement closer to the time of use, and the development of a digital marketplace for such services, delivering both better outcomes and efficiencies relative to tender-based methods. (£8m)
Low Voltage: (£9m) resulting from improved targeting of LV monitoring rollout and the implementation of forecasting models to support new connections, promoting further cost deferral specific to LV.
ii) Cost savings
Resources: Expanding current manual approaches in the absence of integrated AI forecasting solutions would require considerable labour costs, estimated at £0.5m annually conservatively. In practice, the resources required to expand existing methods on a substation-by-substation basis at HV/LV could be considerably higher.
Further net benefits, for example through solutions facilitating 'green' asset connections (e.g. EV/PV, driving CO2e contributions) and new revenues for flex providers will be quantified at Alpha, leveraging flexibility partner input. Flexibility partners can benefit through the development of dynamic procurement models closer to the time of use (enabled by operational forecasts), and the development of a digital marketplace for these services, which can stimulate market growth and new revenue opportunities.
The 10-year NPV (NPg-level) is estimated at £31.82m, whole-life NPV £43.84m. The additional investment required to facilitate these benefits relates only to the innovation funding through to the end of Beta.
Benefits realised to date include a centralized understanding of HV/LV monitoring data, and how AI solutions can be applied to network decision-making. Socializing learnings across partners has highlighted related projects and activities at other DNOs, and identified a clear innovation gap for this project to address.
