go to top scroll for more


Projects: Projects for Investigator
Reference Number NIA2_NGESO022
Title BC Forecasting
Status Completed
Energy Categories Other Power and Storage Technologies(Electricity transmission and distribution) 100%;
Research Types Applied Research and Development 100%
Science and Technology Fields PHYSICAL SCIENCES AND MATHEMATICS (Computer Science and Informatics) 20%;
ENGINEERING AND TECHNOLOGY (Electrical and Electronic Engineering) 80%;
UKERC Cross Cutting Characterisation Not Cross-cutting 100%
Principal Investigator Project Contact
No email address given
National Grid plc
Award Type Network Innovation Allowance
Funding Source Ofgem
Start Date 01 November 2022
End Date 30 April 2024
Duration ENA months
Total Grant Value £350,000
Industrial Sectors Power
Region London
Programme Network Innovation Allowance
Investigators Principal Investigator Project Contact , National Grid plc (100.000%)
  Industrial Collaborator Project Contact , National Grid plc (0.000%)
Web Site https://smarter.energynetworks.org/projects/NIA2_NGESO022
Objectives "This project will consist of four work packages which will develop and test solutions to improve the forecast accuracy and output resolution, and if necessary, optimise the code to meet NGESO requirements.  The work will focus on four areas: Improving upon the existing modelling techniqueMoving from a monthly resolution towards a daily resolution outputExploring alternative modelling techniques such as machine learning methodsIf necessary, profiling and optimising the model code for the deployment environment.In line with the ENAs ENIP document, the risk rating is scored Low. TRL Steps = 2 (4 TRL steps)  Cost = 1 (£350k)  Suppliers = 1 (1 supplier)  Data Assumptions = 2  Total = 6 (Low) " "Four main work packages and one optional work package will form the basis for the project plan. These are as follows: WP1 – Knowledge exchange and exploratory data analysis. Exploratory data analysis will be performed, looking at the current and proposed datasets in depth to determine what may be useful, and any limitations of the data, or additional processing needed. NGESO will explain the models they have developed and make them available to Hartree as code to ensure they can run these as a baseline for subsequent workWP2 – Improve existing time series models. Ways of improving existing ARIMA models will be explored. The exact areas explored will depend to some extent on the findings of WP1, but it is likely to include; (1) Systematically exploring the choice of parameters trends and (2) the use of additional datasets as regressors. If successful the model can be run in parallel to the existing model, demonstrating the improved forecast.WP3 – Improve temporal resolution of models. Adapt the models from WP2 to run at daily resolution, using similar approaches to WP2. Initially use the same models as in WP2, then adapt them for higher spatial and temporal resolution. This will likely entail a step up in computing power to allow the models to run in a reasonable time frame (although optimisation is included as a later work package). If successful the model can be run in parallel to the existing model, demonstrating the improved forecast.WP4 – Exploration of alternative modelling approaches. This work package will focus on application of machine learning techniques such as Convolutional Neural Networks, Deep auto-encoders, and Recurrent Neural networks, to model and make predictions of balancing costs. Depending on the volume and type of data for each variable a suitable technique for each dataset will be selected for the modelling and prediction processes according to the literature. The performance of these models will be assessed and if they are not satisfactory alternative modelling techniques will be implemented to improve the results. If the model output is satisfactory techniques such as Monte Carlo sampling will be explored, to generate a probabilistic outcome for the trained models.WP5 (Optional) – Code Optimisation. If runtime optimisation of the developed model is required, a code review and profiling pass will be carried out before preparing a detailed work plan. " "The objectives for the project are as follows: Develop a model to forecast balancing costs for 1-12 months ahead at a monthly resolution which uses more advanced statistical techniques than the current NGESO model and/or additional datasets. Produce a balancing cost forecast model with better temporal resolution (ideally daily) than current NGESO model."
Abstract "BSUoS is the Balancing Services Use of System charge, paid by transmission connected generation and demand to cover the cost of balancing the electricity system. To set the tariff, an accurate forecast of the costs and the variability is required, but these balancing costs are highly volatile and difficult to forecast accurately. This project is looking to improve existing short term (<12 month) forecasts by applying machine learning and cutting-edge forecasting methods. Additionally, the project seeks to increase the temporal granularity to weekly or daily. Accurate BSUoS forecasting benefits all consumers by enabling better business planning and risk management by NGESO and its customers. It may also bring opportunities for the control room and planners to reduce spend by taking more cost-efficient actions. "
Publications (none)
Final Report (none)
Added to Database 01/11/23