Introduction
We are entering the smart grid revolution, moving to a new paradigm of electricity forecasting and trading.
It is intended that the smart grid revolution will empower domestic consumers, enabling them to participate in the market on both the consumption and production sides. In particular, they will be rewarded for any flexibility they have, in terms of being able to adjust their behaviour according to the needs of their local grid.
This marks a profound change from the past, where consumers were seen as passive points of load, and flexibility was the domain of the larger industrial consumers.
The revolution will entail important changes to forecasting, so where does Matrica think that the electricity industry is headed?
Changes to electricity market rules - empowering the ‘prosumer’
Revisions to market rules is another facet of the new paradigm, especially where empowerment of the consumer creates the need for more forecasting.
The UK electricity market is dominated by the ‘Supplier Hub’ model. Under the UK Balancing & Settlement Code (BSC), this model makes the Licensed Supplier the sole intermediary between the retail customer and the wholesale electricity market. The BSC makes that sole Supplier responsible for the administration of the total metered volume at the household or premises boundary point.
In response to political and technological pressure, Elexon is working on implementing modifications to the BSC to give retail customers the freedom to trade electricity with multiple suppliers, fundamentally breaking up the traditional Supplier Hub monopoly. Under the proposed BSC ‘P379’ modification, consumers could appoint additional or ‘Secondary’ Suppliers for different components of their load and generation. One Supplier would be designated the ‘Primary Supplier’, and be responsible for key BSC administrative tasks such as metering and data aggregation.
As of August 2019, it was believed that each Supplier would be responsible for the imbalance settlement of its share of the customer’s load or generation.
Matrica believes that BSC P379 and its related Modifications will have profound implications for electricity forecasting. Elexon itself states in its P379 FAQ that suppliers participating in a P379 meter-splitting arrangement,"... will have a much more variable and unpredictable energy demand than suppliers currently experience. This may result in higher risks for the supplier.
The devil will be in the detail.
As a member of the P379 Workgroup, Matrica will be commenting further on the detail of the forecasting implications of P379 and related modifications in later Blogs.
Matrica’s views on the future of demand and renewables forecasting under ‘Big Data’
Matrica believes that the next generation of smart meter systems will take forecasting a stage further by allowing the decomposition of domestic and commercial load into a series of individually forecastable ‘channels’, each specific to measuring a particular component of consumption such as EV, lighting, heating, machinery, entertainment, batteries, and cold storage. These new data streams will enable forecasters to calibrate their models on all aspects of domestic and industrial consumption, using a wide range of variables.
Matrica also expects that smart meters will offer at least one export channel, reflecting the consumer’s freedom to sell or donate its generation output to a 3rd party as an alternative to self-consumption or storage.
Regarding domestic consumers, in the short-term Matrica believes it is most unlikely that energy forecasters will focus on modelling for each individual household. Instead, smart meter data will be used to improve the accuracy of deemed profiles used within aggregated forecasts. However, we have yet to see the forecasting impact of P379 in decomposing the household boundary meter into separate import and export channels, each subject to competitive supply, trading and balancing.
The focus of forecasting will shift to identifying sources of load flexibility arising from specific channels within the smart meter, themselves centrally controllable by suppliers or the distribution network operator. This is the first step on the road to the truly ‘smart’ grid, where AI systems will forecast consumption, generation, storage, and flexibility, and optimise power flows around and between local grids, responding to market price signals.
On the generation side, Matrica has seen capacity forecasts predicting that by 2030, 50% of EU power generation will come from renewable sources. The intermittent nature of renewable generation coupled with the move to intra-day trading now means a move to 15-minute or even 5-minute data granularity. Intra-day trading around 5-minute settlement periods places forecasting systems under increased pressure because forecasting wind and solar output every five minutes burns available model processing time, even with the most efficient hardware configurations for relational databases.
Energy forecasters will have little choice in moving to ‘big data’ solutions if their models are to keep pace.
Concluding Remarks - we live in ‘interesting times’
Given the rise of renewables, smart meters, intra-day balancing, big data, and new market rules, this is no time for complacency in the forecasting process and supporting systems. This is particularly true for those relying on relational database solutions that will soon be hitting their performance limit.There will be many more consumption data streams to be forecast, as boundary point volumes are decomposed into separate export and import volumes, each with its own supplier responsible for balancing.
The growth in renewables as a proportion of the generation mix looks set to continue for the foreseeable future, leading to more sophisticated forecasting of generation. Advances in weather data mean new weather variables at finer resolutions at greater frequencies.
Generally, forecasting input data granularities will tend towards real-time, as market participants become increasing focused on intra-day trading, balancing, and optimisation.
New market rules will recognise the ‘prosumer’, reward local flexibility, enable P2P trading, and facilitate the growth of community energy schemes.
In the UK, Matrica expects to see Elexon implement the P379 and related BSC modifications in late 2020, or by 2021 at the latest. EU Member States are unlikely to be far behind with implementing their changes, so for forecasters at multinational European suppliers, this is a broad geographical problem.
In particular, Matrica expects the first impact of P379 to be the splitting of EV supply from that of the rest of the household, enabling consumers to initially choose two suppliers. This will likely be followed by the splitting of household generation from supply, enabling export volumes to be quantified such that the household may elect to supply itself, or sell or donate its exports to third parties, including aggregators, suppliers, charities, neighbours, and community energy schemes.
In a forecasting version of the Chinese curse, the expansion into a true multi-supplier environment is when electricity forecasting will enter the most ‘interesting times’!