In response to the pandemic caused by the COVID-19 virus, many countries adopted control and mitigation measures (e.g., lockdown, social distancing, intermittent attendance, etc.) that significantly increase the amount of time people spend inside the built environment. Nevertheless, the need to resume activities in attendance and ensure the continuity of businesses that cannot be stopped has led to the adoption of measures to reduce the risk of contagion, mainly involving increased ventilation of indoor environments. In this context, the management of Indoor Air Quality became crucial to ensure occupants' health while optimizing thermal comfort and buildings' energy demands. In the scientific literature, one of the most explored solutions to this optimisation problem is the adoption of a digital twin (DT) triggered by the broader diffusion of sensors and technological systems. DTs have three main domains: the physical one (the asset), the virtual one (the digital counterpart), and the cyber-physical (sensors and devices to send data to the digital domain) that binds the two together. However, installing a proper number of sensors in existing buildings may be too expensive or impossible due to technical problems. This study introduces a methodology to take advantage of DTs in existing buildings where technical and economic reasons hinder the deployment of a fully developed cyber-physical domain, replacing hard sensors with soft/virtual ones deploying Artificial Intelligence (AI) techniques. AI can leverage the cyber-physical part of DTs by retrieving useful information on occupant behavior from even few available data, thus allowing for a fully functional DT even in existing buildings with a limited number of installed sensors. The methodology has been tested on a school building in Italy, where a DT has been deployed. In the case study, low-cost and easy-to-install sensors are used to monitor CO2 levels within various rooms. Many control and mitigating protocols adopted during the pandemic set thresholds and limits for CO2 concentrations, requiring increased ventilation once a certain CO2 threshold is reached, with consequent repercussions on indoor comfort and energy consumption. In this study, the opening of windows is automatically detected with AI techniques by examining CO2 temporal trends, thus automatically recognizing building users' behaviors. Consequently, systems can be properly managed to limit energy consumption and thermal discomfort (e.g., turning off the heating while the window is on). Therefore, the introduced methodology enables a DT approach, overtaking difficulties often encountered in installing sensors in existing assets, with obvious benefits in terms of health, energy consumption, and economic savings.

ARTIFICIAL INTELLIGENCE ENABLING DIGITAL TWINS IN EXISTING BUILDINGS

Rampini, Luca;Re Cecconi, Fulvio
2021

Abstract

In response to the pandemic caused by the COVID-19 virus, many countries adopted control and mitigation measures (e.g., lockdown, social distancing, intermittent attendance, etc.) that significantly increase the amount of time people spend inside the built environment. Nevertheless, the need to resume activities in attendance and ensure the continuity of businesses that cannot be stopped has led to the adoption of measures to reduce the risk of contagion, mainly involving increased ventilation of indoor environments. In this context, the management of Indoor Air Quality became crucial to ensure occupants' health while optimizing thermal comfort and buildings' energy demands. In the scientific literature, one of the most explored solutions to this optimisation problem is the adoption of a digital twin (DT) triggered by the broader diffusion of sensors and technological systems. DTs have three main domains: the physical one (the asset), the virtual one (the digital counterpart), and the cyber-physical (sensors and devices to send data to the digital domain) that binds the two together. However, installing a proper number of sensors in existing buildings may be too expensive or impossible due to technical problems. This study introduces a methodology to take advantage of DTs in existing buildings where technical and economic reasons hinder the deployment of a fully developed cyber-physical domain, replacing hard sensors with soft/virtual ones deploying Artificial Intelligence (AI) techniques. AI can leverage the cyber-physical part of DTs by retrieving useful information on occupant behavior from even few available data, thus allowing for a fully functional DT even in existing buildings with a limited number of installed sensors. The methodology has been tested on a school building in Italy, where a DT has been deployed. In the case study, low-cost and easy-to-install sensors are used to monitor CO2 levels within various rooms. Many control and mitigating protocols adopted during the pandemic set thresholds and limits for CO2 concentrations, requiring increased ventilation once a certain CO2 threshold is reached, with consequent repercussions on indoor comfort and energy consumption. In this study, the opening of windows is automatically detected with AI techniques by examining CO2 temporal trends, thus automatically recognizing building users' behaviors. Consequently, systems can be properly managed to limit energy consumption and thermal discomfort (e.g., turning off the heating while the window is on). Therefore, the introduced methodology enables a DT approach, overtaking difficulties often encountered in installing sensors in existing assets, with obvious benefits in terms of health, energy consumption, and economic savings.
21st International Multidisciplinary Scientific GeoConference SGEM 2021
9786197603361
Digital Twin; Artificial Intelligence; Neural networks; School building
File in questo prodotto:
File Dimensione Formato  
31719_ART1_BOOK_6_2.pdf

accesso aperto

: Publisher’s version
Dimensione 3.34 MB
Formato Adobe PDF
3.34 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11311/1210600
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact