BOREALIS – Self-learning building controls for a greener and healthier society
Project leader: Sascha Hammes
Project manager overall project: Johannes Weninger, Bartenbach GmbH
- Bartenbach GmbH
- Zumtobel Lighting GmbH
- Hella Sonnen- und Wetterschutztechnik GmbH
- University of Innsbruck, Department of Computer Science
- University of Innsbruck, Unit of Energy Efficient Building
Funding agency: Federal Ministry for Climate Action, Environment, Energy, Mobility, Innovation and Technology (BMK) represented by the Austrian Research Promotion Agency (FFG)
Funding program: AI for Green
Total funding: 319.175 Euro (UIBK-EEB)
Project period: 01/07/2024 – 30/06/2027
Summary
Despite pioneering successes such as the introduction of LEDs and sensor-based control systems, there is an urgent need to increase efficiency in the lighting sector (15% to 20% of global electrical energy requirements in the building sector) to protect the climate and achieve sustainability. Improvements in integral artificial lighting and daylight control systems offer great potential for this, but this can only be fully exploited if application-specific characteristics such as individual room usage behaviour are taken into account. As this information is not normally available during the planning phase, artificial lighting and daylight control systems today are primarily based on generalized assumptions about user characteristics and behaviour without taking real individualities into account.
These erroneous assumptions made in the early planning and simulation phase often result in significant deviations from reality (so-called performance gaps) and prevent modern lighting systems from fully realizing their intended energy and health potential. In addition, the resulting discrepancy remains largely unrecognized due to a lack of extended commissioning or leads to resource-intensive and costly subsequent adjustments to control systems during operation to compensate for the incorrect assumptions made during the planning phase. However, improved mapping of application-specific properties in the planning process will continue to prove difficult in the future and in some cases (such as new buildings) is simply impossible due to a lack of user information. To ensure the optimal operation of integral control systems, it is therefore essential to adapt them to the respective application during operation. In addition, systems should also be able to efficiently map changes in utilization (e.g. seasonal effects or when users change) to ensure that the energy and health potential is utilized in the long term over the entire lifetime of the building. However, there are currently no applicable control concepts available for this.
The development of a control core based on reinforcement learning, which is the aim of the BOREALIS project, will therefore make a significant contribution to utilizing the overarching system potential and achieving both climate and health policy goals. The expertise required for this will be provided by the scientific partners. To ensure the broadest possible application and successful implementation of the results in the lighting sector, the consortium includes leading international companies that have their own control systems.