Data-driven digital twining of ventilation systems for performance optimization: A university building case study

This study introduces the creation and application of a data-driven digital twin for building ventilation systems, focusing on a university building as a case study. It employs a grey-box energy modelling framework to accurately forecast, simulate, and monitor the ventilation system's efficiency under diverse conditions. The study collects a substantial dataset to reflect various usage patterns and environmental influences, which serves to test and validate the component models of the ventilation system. These models are integrated into a digital twin platform, providing a comprehensive overview of the system's performance and critical indicators in real time. The digital twin facilitates informed decision-making for facility managers regarding energy consumption, inefficiency identification, and the recommendation of custom retrofitting actions specific to the building's characteristics and use. The findings confirm that digital twins are effective as a tool to continuously commission and detect anomalies in buildings. The study offers a ventilation modelling and monitoring method capable of recognizing rule-based control behaviours and changes in systems that occur in cycles, like system shifts from winter to summer, and can estimate total air mass flow rate with a correlation exceeding 80%