Buildings account for 40% of energy consumption in the EU, and the bulk of this demand goes towards heating, ventilation, and air conditioning (HVAC). With increasing electrification and digitalization, maintenance and control of HVAC systems in buildings is fast transforming into a challenge in the data domain that requires new ICT tools. The ADRENALIN project, funded by ERA-Net Smart Energy Systems through the joint call on Digital Transformation, seeks to develop and test machine learning (ML) and smart control algorithms to enable or improve data services for buildings. The project consortium consists of research organizations, ICT companies and building owners specializing in data services for buildings.
The main energy meter is the only data source in most existing buildings. However, the development of data services for buildings depends on additional information, such as the energy used for heating and cooling, indoor temperature, etc. This requires the installation of additional sensors and sub-metering systems that are either intrusive or costly, or impractical. ADRENALIN addresses this challenge by inferring as much information as possible from the existing energy meter data by employing advanced ICT techniques such as ML and energy informatics.
Three core data-related topics have been identified in ADRENALIN – i) to disaggregate or separate HVAC loads from total main meter data, ii) to automatically label the energy class of a building based on HVAC loads and iii) to schedule and control HVAC loads in a smart, flexible manner in response to price signals. In this context, ADRENALIN project will collect large amounts of data from buildings in a sandbox and crowdsource ML solutions by organizing competitions for each data challenge. The best solutions will be implemented under real-world conditions on partner companies' digital platforms to test their overall validity and scalability, thus bringing them closer to market readiness.
Buildings represent a high share of peak electricity demand, but thanks to their slow thermal inertia they also offer the potential to be one of the lowest-cost opportunities for providing the flexible demand needed to support increasing levels of variable renewable energy resources in electricity grids. To activate and scale this latent flexible demand opportunity, new data-driven software services are needed. ADRENALIN aims at facilitating large scale roll-out of data services and smart controls in the existing building stock. By collecting a large and varied pool of measurement data from real buildings (data sandbox), ADRENALIN will crowdsource to data challenge competitions the development of new algorithms. The best-performing solutions will be implemented in real-life conditions on the digital platforms of the partner companies to test their general validity and replicability, and to demonstrate real-life performance.