Developing a surrogate RC Model for Optimal HVAC Control Applications in Modelica-Based Building Digital Twin

¹ Department of Architecture, Faculty of Engineering, The University of Tokyo, Tokyo 113-8656, Japan
² Institute of Industrial Science, The University of Tokyo, Tokyo 153-8505, Japan

^* Corresponding author: This email address is being protected from spambots. You need JavaScript enabled to view it.

Abstract

In the development of building Digital Twin frameworks for HVAC systems, EnergyPlus simulations via the Spawn interface have been used for building load calculation. However, although such systems can technically be exported as Functional Mock-up Units (FMU), their runtime dependence on the EnergyPlus engine limits portability and flexibility in diverse deployment environments and makes control optimization difficult to implement efficiently. To address this issue, we develop a surrogate modeling approach based on a resistance-capacitance (RC) thermal model. The RC model was first fitted using data from a building simulation model based on architectural drawings design parameters, and then RC model was calibrated using nighttime natural cooling data. This calibration ensured that the adjusted RC model primarily reflects the building's inherent thermal properties, with little influence from HVAC operation. In this study, the target system is a two-story office building located in the cold region of Sapporo, Japan. It features energy efficient building systems, including a groundwater-source heat pump and a thermal storage tank. To accurately represent HVAC dynamics, we developed a detailed HVAC system model using Modelica. Furthermore, the RC model was calibrated using data from a Modelica-Spawn co-simulation model. This co-simulation model had been validated against measured data. This calibration process enabled the surrogate RC model to approximate the load calculation results of Spawn when coupled with the Modelica HVAC system. In addition, after replacing Spawn with the RC model to form the Digital Twin framework, we evaluated its integration into an optimal HVAC control setting, with particular attention to computational efficiency, response fidelity, and compatibility with control algorithms. Overall, the developed surrogate approach provided a physically grounded, computationally efficient, and FMU-compatible solution that successfully constructed a building digital twin and used it to examine the application of optimal HVAC control algorithms.