Volta – Locational Real Time Prediction (LRTP)
Project summary
The project will apply state-of-the-art forecasting at high spatial and temporal granularity to support scheduling and decision-making within control room balancing teams. It will assess the current data landscape, develop and test scalable forecasting approaches across hundreds of existing, modified and new network nodes, and explore model explainability, including the use of AI agents.
| Name | Status | Project reference number | Start date | Proposed End date | Expenditure |
|---|---|---|---|---|---|
| Volta – Locational Real Time Prediction (LRTP) | Live | NIA3_NESO011 | Jan 2027 | Jan 2028 | £640,000 |
The project will apply state-of-the-art forecasting at high spatial and temporal granularity to support scheduling and decision-making within control room balancing teams. It will assess the current data landscape, develop and test scalable forecasting approaches across hundreds of existing, modified and new network nodes, and explore model explainability, including the use of AI agents. The project has three phases:
1) assess data availability, quality and user requirements (4 months);
2) build a proof-of-concept model and test scalability across hundreds of network nodes (5 months); and
3) evaluate model explainability options, including AI agents (3 months). Key risks include the availability and quality of metered data at GSP (Grid Supply Point) level, ongoing network reconfiguration, and availability of computational resources during testing.
Benefits
1. What is the expected benefit delivered directly by THIS PROJECT?
1a.Short outcome statement (clear and specific)
Volta LRTP project will improve locational demand forecasting for Control Room decision-making by building on the Volta Real Time Predictor. It aims to improve operational decision-making, support costs efficiency, strengthen future readiness, and build confidence in AI-based forecasts, ultimately delivering a scalable, explainable forecasting approach that supports more efficient reserve allocation, lower balancing costs, and better whole-system coordination.
1b. Who will benefit from this project and in what way?
Volta Programme: The adaptive model outputs will directly enable Volta’s core optimisation in scheduling and dispatch timeframes.
NESO ENCC (Electricity National Control Centre) Control room operators: They will benefit from more accurate, explainable real-time forecasts, improving trust, situational awareness, model judgement and informed decision-making.
NESO Balancing and dispatch teams: High-resolution probabilistic forecasting will support better reserve allocation, more efficient dispatch decisions and reduced need for expensive short-term corrective actions.
Consumers / end users: They may benefit indirectly through reduced balancing and thermal constraint costs, which can support more efficient system operation.
Distribution System Operators: Greater visibility at Grid Supply Points will create a stronger coordination point with DSOs, helping optimise flexibility use across the wider system.
NESO Data and AI readiness teams: The project will evaluate data quality, coverage and requirements at more regionalised network levels, improving the control room’s readiness for AI-enabled tools.
NESO Network planning / operational teams: Improved forecasting for new or similar nodes with little historical data will support better decision-making where data is limited.
1c. How confident are you this project will deliver this benefit?
Low / Medium / High
There is a high level of confidence that the project can deliver the proposed benefits, because it builds on the previously successfully delivered Volta Real Time Predictor `and addresses a clearly defined operational need. Confidence is strongest around improving forecasting insight, explainability, and decision support. The level of benefit delivered will depend on data availability, model performance, scalability, and successful integration into Control Room processes.
1d. Can this benefit be quantified? (For development and demonstration projects provide quantified estimates where reasonably possible. If quantification is not currently feasible, explain why and what evidence supports the expected benefit)
Yes - provide metric, baseline, expected improvement
No - explain why not at this stage
Yes: benefits can be partly quantified, although some benefits will need to be evidenced qualitatively or through proxy measures. The benefit can be quantified through measurable improvements in forecast accuracy, improved visibility of constraint risks, and potential reductions in inefficient redispatch or balancing actions. Current annual balancing costs attributed to redispatch between network zones is estimated at £1.7bn. Benefits such as improved explainability, operator confidence, and whole-system coordination are likely to require qualitative assessment alongside operational performance metrics.
Future / Full-Scale Benefit (if applicable - think beyond direct project outputs, e.g. operational rollout, industry adoption, policy influence, future cost reduction, resilience improvement)
Apply relevant category (Knowledge, Social, Environmental, Financial)
At full scale, the project has the potential to deliver high-resolution, explainable and continuously updated forecasts across network nodes, enabling more efficient scheduling, reserve allocation and dispatch decisions. This would support reduced balancing and thermal constraint costs, improved coordination with DSOs, better utilisation of flexibility across the whole system, and greater control room readiness for AI-enabled decision support. Ultimately, this could improve operational efficiency, system reliability and cost outcomes for consumers.
For the wider industry, successful full-scale deployment could support a more coordinated, efficient, and data-driven electricity system. Improved locational forecasting would help network operators, DSOs, flexibility providers, and market participants make better-informed decisions, improve utilisation of distributed flexibility, reduce whole-system balancing and constraint costs, and support the transition to a more decentralised, low-carbon network. Over time, this could contribute to improved system resilience, lower consumer costs, and greater confidence in AI-enabled operational decision support across the energy sector.
2. What is the expected benefit if this idea is implemented at full scale?
2a. High-level outcome statement
At full scale, the project could deliver a high-resolution, explainable forecasting capability that supports more efficient scheduling, reserve allocation and dispatch across the electricity network. This could help reduce whole-system balancing and constraint costs, improve coordination with DSOs and flexibility providers, strengthen system resilience, and support better cost outcomes for consumers.
2b. Who will benefit from this project in the future and in what way?
Link benefit - stakeholder clearly
NESO: better real-time visibility of locational demand, constraints and network risks to support faster, more informed decisions. Improved scheduling, reserve allocation and dispatch, helping reduce inefficient balancing actions and operational costs.
Distribution System Operators: clearer whole-system visibility to support better coordination of flexibility and local network constraints.
Flexibility providers and market participants: improved signals and forecasting insight, enabling more efficient participation in balancing and flexibility markets.
Consumers: potential long-term benefit through reduced whole-system costs, improved reliability and more efficient use of network assets.
2c. What needs to happen for this future benefit to be realised?
e.g. rollout, policy change, integration, further development
For the future benefit to be realised, the project would need to successfully move from modelling and evaluation into practical operational use, undergoing a phased rollout, and integrating into Control Room processes and systems whilst supporting policy or process changes. This would require:
> Reliable data access and quality to support accurate, high-resolution locational forecasting.
> Strong model performance demonstrated through measurable improvements in forecast accuracy and constraint visibility.
> Scalable architecture that can operate across relevant network nodes and changing system conditions.
> Explainability and trust so Control Room users can understand and confidently use the forecasts.
> Operational integration into scheduling, reserve allocation and dispatch processes within Volta and/or OBP
> User adoption and training so outputs are used consistently in decision-making.
> Collaboration with DSOs and flexibility providers to support whole-system coordination.
> Ongoing monitoring and refinement to ensure the model remains accurate, useful and resilient over time.
2d. How confident are you this project will deliver this benefit?
Low / Medium / High
Confidence is high, as the project builds on an established Volta capability and targets a clear operational need; however, full benefit realisation will depend on data quality, model performance, successful integration, phased rollout, supporting process or policy change, user adoption and further development at scale.
The scale of benefit will depend on whether the project can prove:
> reliable data availability and quality;
> strong forecasting performance;
> scalable architecture;
> explainable outputs that Control Room users trust;
> pathway to successful operational integration, rollout and adoption.
2e. Can this benefit be quantified? (For development and demonstration projects provide quantified estimates where reasonably possible. If quantification is not currently feasible, explain why and what evidence supports the expected benefit)
Yes - provide metric, baseline, expected improvement
No - explain why not at this stage
Yes, the benefit can be quantified where the project links to measurable operational performance and cost outcomes including forecast accuracy improvements, earlier identification of constraint risks, more efficient reserve allocation, and reductions in inefficient redispatch or balancing actions. Comparison will be made between current methodologies for predicting regional electricity demand to quantify improvements in forecast accuracy. Annual balancing costs attributed to zonal redispatch are currently estimated at £1.7bn (£1.7bn figure taken from NESO’s 2025 Annual Balancing Costs Report, where NESO reported that thermal constraints were the main driver of rising balancing costs in FY2024/25. A secondary article summarising the report states that balancing GB’s grid cost £2.7bn in 2024/25, with thermal constraints accounting for £1.7bn of that total. It also notes that thermal constraint volumes rose to 13.5 TWh, increasing costs by 64% year on year.
Improved regional forecasts will contribute to bringing down these costs. It may be possible to assess case studies where improved forecasts would have led to early identification of constraint risks and more efficient use of balancing actions as a result. Benefits such as explainability, operator confidence, and whole-system coordination may need to be assessed through qualitative evidence and operational feedback.
| Name | Published |
|---|---|
| NIA Project Registration and PEA Document | 03 Nov 2026 |