A building that becomes a scientific instrument.

Published

1. Journal article Open access 2026

   Occupancy-Aware Digital Twin for Sustainable Buildings

   Ivan Smirnov, Fulvio Re Cecconi

   Buildings, vol. 16, no. 8, art. 1629 · MDPI, 2026 · doi:10.3390/buildings16081629

   Abstract

   This paper proposes a human-centric digital twin (DT) framework balancing energy efficiency with occupant well-being in existing buildings, addressing the lack of actionable insights in data-driven facility management and comfort issues common in fully automated systems. A “Human-in-the-loop” approach using dual-KPIs integrates real-time IoT data and visualization to evaluate sustainable energy use via Indoor Environmental Quality (IEQ). A novel occupancy-inference method tracks efficiency in legacy buildings without granular metering, implemented through a case study of 26 office rooms. Results indicate that the framework successfully identifies significant energy wastage and comfort anomalies without compromising well-being. Integrating real-time analytics with human oversight enables more resilient management than fully automated alternatives, particularly for detecting non-operational heating waste. The occupancy inference method was validated against ground truth, achieving 81% accuracy, with limitations regarding decay lag discussed. This research offers a cost-effective diagnostic tool for legacy buildings lacking sub-metering, lowering DT adoption barriers, and shifting maintenance from reactive to data-driven strategies. The framework leverages human expertise and infers occupancy-normalized energy metrics from standard IEQ sensors, proposing a human-centric DT framework to bridge the gap between raw sensor data and actionable facility management insights.

   Cite (BibTeX)

   @article{smirnov2026occupancy,
     title     = {Occupancy-Aware Digital Twin for Sustainable Buildings},
     author    = {Smirnov, Ivan and Re Cecconi, Fulvio},
     journal   = {Buildings},
     volume    = {16},
     number    = {8},
     pages     = {1629},
     year      = {2026},
     publisher = {MDPI},
     doi       = {10.3390/buildings16081629}
   }

   Copy

2. Conference paper Open access 2025

   Transforming Live IoT Data into Actionable Information for Asset Management via Digital Twins

   Ivan Smirnov, Sebastiano Maltese, Nicolò Molinari, Fulvio Re Cecconi

   IOP Conference Series: Earth and Environmental Science, vol. 1554, 012051 · SBE25 Zurich — Sustainable Built Environment Conference, ETH Zurich · IOP Publishing, 2025 ·doi:10.1088/1755-1315/1554/1/012051

   Abstract

   The rapid expansion of the Internet of Things (IoT) has generated an overwhelming amount of dynamic data, challenging asset management in the built environment. The sheer volume and complexity of these raw data often prevent managers from extracting actionable insights necessary for sustainable practices. This paper aims to transform live IoT data into meaningful information to enhance sustainable asset management. Focusing on the segment ‘data-to-information’ of the pyramid of data-information-knowledge-wisdom, it explores how integrating IoT data with digital twin technology can facilitate environmentally conscious decision making.

   Authors propose a cohesive framework that leverages data analytics and machine learning algorithms within a digital twin environment. The methodology involves deploying a data processing pipeline that aggregates IoT data streams, enhancing relevance for asset managers focused on sustainability objectives. Visualization tools present this information in an accessible format, aiding informed decision-making.

   Through two case studies, it is demonstrated how this approach leads to improved operational efficiency and strategic asset utilization, all contributing to minimized environmental impact. The findings highlight the critical role of effective data interpretation in driving decisions that align with organizational values and green initiatives. By transforming raw IoT data into actionable insights within a digital twin framework, organizations can make informed decisions to promote sustainable asset management.

   The paper includes the design and implementation of the data processing pipeline, integration strategies with digital twin technology, application of machine learning algorithms for data analysis, and development of visualization tools. Issues beyond the data-to-information conversion, technical specifics of IoT hardware deployment and cybersecurity aspects of IoT data transmission are not addressed.

   Cite (BibTeX)

   @article{smirnov2025transforming,
     title     = {Transforming Live {IoT} Data into Actionable Information for Asset Management via Digital Twins},
     author    = {Smirnov, Ivan and Maltese, Sebastiano and Molinari, Nicol{\`o} and Re Cecconi, Fulvio},
     journal   = {IOP Conference Series: Earth and Environmental Science},
     volume    = {1554},
     number    = {1},
     pages     = {012051},
     year      = {2025},
     publisher = {IOP Publishing},
     doi       = {10.1088/1755-1315/1554/1/012051}
   }

   Copy

3. Book chapter 2025

   Neural Network Tree Identification from Street View Images to Estimate CO₂ Sequestration

   Sam Wilcock, Ivan Smirnov, Ornella Iuorio

   In Envisioning the Futures — Designing and Building for People and the Environment, Lecture Notes in Civil Engineering, pp. 836–852 · Springer, 2025 · doi:10.1007/978-3-032-06978-8_45

   Cite (BibTeX)

   @inproceedings{wilcock2025neural,
     title     = {Neural Network Tree Identification from Street View Images to Estimate {CO2} Sequestration},
     author    = {Wilcock, Sam and Smirnov, Ivan and Iuorio, Ornella},
     booktitle = {Envisioning the Futures -- Designing and Building for People and the Environment},
     series    = {Lecture Notes in Civil Engineering},
     pages     = {836--852},
     year      = {2025},
     publisher = {Springer Nature Switzerland},
     address   = {Cham},
     doi       = {10.1007/978-3-032-06978-8_45}
   }

   Copy

4. Journal article Open access 2025

   Enabling Sufficiency Through Smart Locks: Transforming Office Occupancy and Building Management for Energy Savings

   Andrea Giovanni Mainini, Francesco Pittau, Elena Casolari, Matthieu Simon Majour, Matteo Cavaglià, Riccardo Riva, Giulia Amendola, Alberto Speroni, Juan Diego Blanco Cadena, Tiziana Poli

   Buildings, vol. 15, no. 5, art. 669 · MDPI, 2025 · doi:10.3390/buildings15050669

   Abstract

   In the aftermath of the global pandemic, the widespread embrace of flexible working models has led to suboptimal occupancy levels in office buildings. Despite this shift, traditional space management practices persist, contributing to increased energy consumption per person. This study investigates how integrating smart lock systems can enhance space utilization within flexible working environments, ultimately reducing energy use. A case study of an office building in Milan, Italy, is used to evaluate the proposed approach. The methodology includes a comprehensive assessment of building design and functionality, coupled with impact analyses using Building Energy Modeling and Life Cycle Assessment. The results indicate that innovative occupancy management strategies can achieve energy savings of from 9% up to 14% compared to baseline operational energy use, leading to a reduction in CO₂ emissions of 7.5 to 17.6 kgCO₂eq/m² depending on occupancy scenarios. The life cycle assessment reveals that, while smart locks introduce an initial embodied carbon footprint of approximately 2 tons of CO₂, that is recovered through the savings obtained after a few months of installation. The findings demonstrate that this methodology is effective in buildings that allow both functional and temporal flexibility, enabling partial shutdowns and the redirection of certain services when not in use, ultimately improving energy efficiency through lean interventions.

   Cite (BibTeX)

   @article{mainini2025enabling,
     title     = {Enabling Sufficiency Through Smart Locks: Transforming Office Occupancy and Building Management for Energy Savings},
     author    = {Mainini, Andrea Giovanni and Pittau, Francesco and Casolari, Elena and Majour, Matthieu Simon and Cavagli{\`a}, Matteo and Riva, Riccardo and Amendola, Giulia and Speroni, Alberto and Blanco Cadena, Juan Diego and Poli, Tiziana},
     journal   = {Buildings},
     volume    = {15},
     number    = {5},
     pages     = {669},
     year      = {2025},
     publisher = {MDPI},
     doi       = {10.3390/buildings15050669}
   }

   Copy

Forthcoming

• Accepted Conference paper 2026

  IDS-Constrained Text-to-Graph Querying for IFC Models

  Ivan Smirnov, Tanya Bloch, Fulvio Re Cecconi, Sebastiano Maltese, Lavinia Chiara Tagliabue, Silvia Meschini, Shabtai Isaac

  EC3 2026 — European Conference on Computing in Construction, Corfu, Greece

  Politecnico di Milano · Technion – Israel Institute of Technology · SUPSI · University of Turin · Ben-Gurion University of the Negev

  Abstract

  Accessing Building Information Model (BIM) data encoded in Industry Foundation Classes (IFC) remains difficult for non-expert stakeholders due to heterogeneous property sets and nested relationships. This paper presents a constrained text-to-graph querying framework that transforms IFC data into a labeled property graph, queried via Cypher, while project-specific Information Delivery Specification (IDS) requirements define admissible entity labels, relationships, and property keys. A regular-expression grammar restricts Large Language Model (LLM) generation to executable, schema-compliant queries. Evaluation shows grammar constraints eliminate schema hallucinations, achieving 100% Semantic Compliance and improved execution accuracy, with competitive performance from smaller open-weights models.

• Under review Journal article 2026

  Why BIM?

  Ivan Smirnov, Sebastiano Maltese, Nicolò Molinari, Fulvio Re Cecconi

  Abstract

  Digital Twins (DTs) are increasingly adopted in the built environment to support monitoring, prediction, optimisation, and lifecycle decision-making. However, the role of Building Information Modelling (BIM) within operational DTs remains insufficiently defined, particularly because quasi-static BIM information and near-real-time operational data evolve at fundamentally different temporal scales. This study critically examines when BIM contributes tangible value to DT applications, how BIM-derived information can be combined with Internet of Things (IoT) data for Facility Management (FM), and which DT benefits require specific levels of BIM information. Evidence from the literature, documented applications, BIM update practices, and illustrative implementation scenarios is synthesised to characterise the temporal mismatch between BIM and operational data, identify IoT-BIM data-fusion patterns, and map DT benefits to their minimum BIM requirements.

  The results show that BIM dependency is strongly use-case specific. Geometry-dependent applications, including retrofit planning, compliance assessment, and spatially constrained interventions, require accurate and maintained BIM information, whereas monitoring, occupancy analytics, and predictive maintenance primarily depend on asset identifiers, spatial relationships, system topology, and other semantic information. Accordingly, continuous synchronisation of a high-fidelity BIM model is neither necessary nor economically justified for many operational DT applications.

  The principal contribution is an integrated decision framework that jointly relates use-case-specific BIM dependency, the minimum geometrical, alphanumerical, and documentary information required under ISO 7817-1:2024, the refresh rate at which that information must remain current, and the lifecycle cost consequences of maintaining it. The framework is operationalised through a selective integration strategy in which BIM-derived information is maintained in a lightweight semantic layer, while detailed geometry is retained and updated only where justified by the intended use case. The findings reframe BIM from a universal DT prerequisite to a purpose-driven information resource whose level of integration should be determined by operational requirements, information refresh rates, and lifecycle maintenance costs.