Thesis title: Strategia di ottimizzazione del monitoraggio e controllo ambientale degli edifici basata sul Digital Twin
In the current scenario of the Architecture, Engineering, and Construction sector (AEC), the transition towards intelligent and sustainable building management paradigms has become a strategic imperative. In this context, the introduction of Digital Twins (DT) represents a cutting-edge methodology for optimizing the operational performance of buildings, offering innovative tools for monitoring and control. This study is situated within this framework, exploring the application of such principles in a complex case study, with the ambition to establish a replicable model that can be extended to future developments, such as the Rome Technopole project.
This study analyzes the development and validation of a Digital Twin prototype, named QTwin, applied to the advanced management and monitoring of specific environments within a historic and institutional building located in Rome. This study aims to pursue multiple objectives, including improving energy efficiency and indoor air quality, while promoting predictive maintenance and optimizing space utilization. The research addresses intrinsic complexities related to technological integration in a constrained building context, characterized by obstacles to wireless connectivity and the need to adhere to rigorous cybersecurity and interoperability standards, an aspect for which academic literature is still lacking.
The adopted methodology involved a systematic selection and evaluation of platforms for Digital Twin development, Internet of Things devices, and their communication protocols. A system architecture was implemented based on the framework developed within the SmartLab research project, an experimental demonstrative environment located within the Faculty of Architecture of Sapienza University of Rome. This architecture integrates a vast network of IoT sensors for the real-time collection of essential data, such as energy consumption, indoor air quality, thermal comfort, and space occupancy levels.
This data was transmitted and managed through standard protocols and processed by a backend infrastructure that includes storage systems and visualization via interactive dashboards. For data interpretation, prediction formulation, and anomaly detection, advanced Machine Learning algorithms were used, with processing also occurring locally (edge computing) to ensure timely responses. The system also includes a "Human-in-the-loop" mechanism for actions requiring human intervention, and it was rigorously validated through the comparison of real data and predictive simulations, as well as stress tests.
The obtained results highlight the effectiveness of the integration between IoT technologies, automation systems, and Machine Learning algorithms within the platform. This synergy allows for a dynamic and real-time representation of data, enabled by the visualization of graphical and informational models developed according to the Building Information Modeling methodology and the use of interactive dashboards.
The Machine Learning models developed for energy consumption prediction showed high accuracy, successfully identifying both general trends and seasonal fluctuations. The analysis of carbon dioxide (CO₂) concentration allowed for estimating occupancy and identifying periods of suboptimal indoor air quality, in line with current regulations. The monitoring of temperature (°C) and relative humidity (HR) provided essential data for informed decision-making regarding thermal comfort and efficiency, while data analysis techniques allowed for identifying consumption patterns, promoting more conscious energy management strategies.
These outcomes attest to the validity and replicability of the framework in complex and heterogeneous building contexts, highlighting the scalability and adaptability of the Digital Twin approach for sustainable and intelligent building management. The QTwin project offers a deep understanding and crucial decision-making tools for timely data-driven management, foreshadowing the possibility of extending this methodology to other urban or infrastructural areas. Future research directions include the refinement of Machine Learning models, the optimization of human-machine interaction, further investigation into cybersecurity, and standardization for greater interoperability.