19.5. Pertti Järventausta
Achievements in 2015
In 2015 a new hybrid method was developed for forecasting the dynamic responses of groups of electrically heated full storage houses. The method combines machine learning (neural network) and a physically based model structure. In comparison simulations the new hybrid forecasting method performed better than its component methods, for example. Thus the research on this segment was successfully completed and the research focus shifted to the response forecasting and optimisation of other important segments, such as groups of partial storage heated house, heat pump heated houses, houses with solar power, etc.
The possibilities of open data in load forecasting were under way. Along with the Europe 2020 strategy EU directives such as INSPIRE and PSI as well as national initiatives such as Finnish government’s agenda for open information are increasing the availability of public sector data, creating a significant economic gain in terms of new data-based products and services. The opening public data together with massive smart metering data will open possibilities for testing and developing new data-driven analytics and modelling concepts, which can be used in the planning and management of future flexible energy systems.
Demand response forecasting is necessary for enabling a sustainable energy system
Here we explain how accurate forecasting and optimal control of dynamic demand responses enables the high penetrations of demand response that are necessary for sustainably integrating high shares of power generation from renewable energy sources. We also explain how the project RESPONSE contributes to this.
The importance of demand side responses for the sustainability of energy infrastructures has been increasingly understood. For example, in the international electricity grid workshop CIRED 2016 conference more than one third of the papers are related to demand side flexibilities. In the provisioning of inertia and balancing power, demand side flexibilities are often a cost efficient, environmentally friendly and fast responding alternative to fossil fuel based power plants and network investments. In the future electricity systems the power generation is increasingly based on renewables that are either variable according to the weather (wind), distributed (many biofuels) or both (solar and CHP). Possibilities to build new hydro power and long distance transmission lines are limited and both construction and getting the permissions takes time. In Finland the legislation and regulation aiming to weatherproof distribution grids with traditional expensive grid investments such as underground cabling are causing in many rural areas so high distribution tariff increases that severe economic and political consequences to the worst affected areas may not be avoided, if the legislative and regulatory framework is not improved to treat alternative solutions, such as utilising demand side flexibilities, in a level and unbiased way. Batteries are improving and are in some cases part of the solution but still remain too costly to be the main solution. Demand side flexibilities are now widely seen as an essential or even the most important part of the solution. Static Time of Use tariffs and controls are since 1964 applied in large scale in Finland, but now and in the future dynamic demand flexibilities are needed to respond to the weather dependent variations in generation and demand and short duration overloading of grid bottlenecks.
Demand responses typically do not cause any losses nor emissions. They are faster and more reliable than big power plants. But they have their relative weaknesses, such as limited duration of the responses and poor fit with the existing electricity market structures, legislation and regulation that are designed based on traditional power plants and grid technologies. A further barrier for large scale adoption of dynamic demand responses is the fact that much of the demand responses are still poorly predicted and thus tend to increase the stochastic variations in electricity system balancing and grid operation. The project RESPONSE is providing solutions to this predictability challenge. It develops methodologies, methods and models for the forecasting and optimal control of demand responses. Aggregated responses of homogenous consumer groups are often better predictable and more relevant for the electricity system and grid management than the responses of individual consumers. Individual customer models are often needed in the development of aggregated models. The project has completed models for certain customer segments and is modelling further segments. Publication of the results continues. The models are developed based on 1) the hourly interval consumption data that is now in Finland available from every relevant customer, 2) minute level power measurements from the substations of the electricity grids, 3) measured and forecast weather data.