SmartCorners - Design report on thermal and cabin comfort control (D3.1)

The market share for battery-electric vehicles (BEVs) is constantly increasing and displacing vehicles powered by a combustion engine. Tough competition in the new market segment is promoting innovation. The propulsion systems are constantly being optimized. Since the energy density of batteries remains the limiting factor for both cost and range, it is of particular importance to improve the performance of the main energy consumers such as the propulsion system, the thermal system (powertrain) and the heating, ventilation, and air conditioning (HVAC) system (cabin). At the same time customer demands for safety, cost, and comfort remain consistently high and continue to drive technological progress forward.

The SmartCorners project aims to significantly improve the holistic thermal management performance. This is done by the integration of artificial intelligence (AI) and machine learning (ML) based control functions to increase energy-efficiency, user comfort and circularity.
The design process of the architecture of a thermal and HVAC system for BEVs is evaluated in in this document. Operating ranges and functional requirements of modern thermal systems are defined. A definition of a state-of-the-art Thermal and HVAC system architecture is done which meets the requirements of modern BEVs’ demands. The requirements are compared with the thermal and HVAC system of a demo vehicle (thermal management carrier) and the suitability of the thermal system for use in this project is checked.
Reinforcement learning (RL) models are used to optimize control algorithms based on operating data (vehicle driving characteristics and data on the ambient conditions) to increase comfort and efficiency. The RL models are trained using so-called digital twins. Digital twins are numerical simulations of physical systems, in this case the thermal and HVAC system. Models that accurately describe the operating behaviour of the real system are fundamental for training of RL models. These learn the system responses and define new improved operating strategies.
Digital twins of the thermal architecture and the cabin were created within this project. To improve the prediction accuracy of the cabin model and thus fully exploit the potential of the AI controls, a new method for modelling the cabin is in development. It will be able to use 3D computational fluid dynamics (CFD) data for fast-running models (FRM) to calculate the cabin climate with increased accuracy.
It has been proven that intelligent control has enormous potential. Considering online big data, user characteristics, current and future information on road conditions and current and future weather information can be combined to derive smart control strategies and thus improve the energy-efficiency of heat management systems for electric vehicles (EVs). Furthermore, such self-adaptive software solutions can significantly reduce manual calibration effort on the control systems by human experts, and ultimately, optimize operational robustness in real-world conditions.
The integration of all systems is strengthened in SmartCorners. Additional individual information (on the route, weather and user status, etc.) is made available to the control algorithm in order to derive a holistic and predictive heat management system with improved adaptation to the ambient conditions.