Optimization and management of microgrids in the built environment based on intelligent digital twins

The interplay between the built environment and energy use has profound implications for global energy consumption, emissions, and the transition towards sustainable systems. Recent analyses reveal that buildings are responsible for consuming approximately 40% of global energy, contributing to over 70% of electricity consumption and nearly a third of all carbon emissions. As the energy paradigm shifts, so does the dynamic between infrastructure and energy sources. Distributed Energy Resources (DERs), including microgrids, nanogrids, and behind-the-meter energy storage, offer both challenges and opportunities. Additionally, innovations such as grid-interactive efficient buildings and Electric Vehicle (EV) integration pathways (like EV-to-grid and EV-to-home) are broadening the horizons of demand response, heralding an age marked by heightened demand flexibility. The past decade's advancements on the Internet of Things (IoT) have opened an era of interconnected sensors and devices. This transition from limited to expansive data repositories facilitates advanced anomaly detection, predictive maintenance, and smart features based on Machine Learning (ML) and other Artificial Intelligence (AI)-powered methods. In this scenario, the emergence of the Digital Twin paradigm (DT) is redefining how we model the contexts in which these innovations operate, even though this concept remains under-explored in the energy and smart grid sectors. This research was conducted with the cooperation of ENEA (Italian National Agency for New Technologies, Energy and Sustainable Economic Development) and delves into the challenges and potential of integrating AI with Digital Twins to support and enhance energy management and optimization in built environments. Using a testbed at the Faculty of Architecture of Sapienza University of Rome known as the SmartLAB—a building provided with an IoT network, a photovoltaic system, and an EV charging station—the microgrid is monitored and regulated via an AI-driven energy management system (EMS). Through the deployment of IoT devices, vast data streams are generated enabling insights for an Intelligent Digital Twin (IDT) framework built on open-source methods, integrating Building Information Modeling (BIM), IoT, and AI within a Proof of Concept (PoC) Data Ecosystem. The research's core aims are to craft an autonomous system capable of discerning energy loads using Naïve Bayes Classifier with Association Rule Mining ensuring efficient energy consumption management and optimized distribution among distinct loads/users. Moreover, it seeks to enhance the well-being of SmartLAB occupants by overseeing CO2 levels, lowering volatile organic compound (VOC) concentrations through IoT-guided, data-informed procedures. This approach demonstrates adaptability across multiple contexts, from grid interactions to smart ecosystems, improving real-time control and providing data-driven energy optimization strategies. Moreover, the proposed prototype embraces scalable and cross-disciplinary strategies, paving the way for further integrations, such as space management systems and predictive maintenance. The findings highlight the advantages of embracing DT technologies within the built sector, while ongoing advancements are also discussed towards DT-based Smart Cities and Energy Communities for a more sustainable built environment.