At its core, the University of Toronto Center for Intelligent Buildings Digital Twinning (IBDT) aspires to bring about a transformative shift in how artificial intelligence is used in the built environment.

Overall, the IBDT intends to guide the processes of design, construction, and operational planning of buildings (in U of T and beyond) by means of understanding the data generated by the building (i.e., the what); analyzing the impact of operational parameters on performance (i.e., the why); using machine learning to predict future conditions (i.e., what's next); simulating the impacts of changes in policy and operational scenarios on performance (i.e., the how); and doing all of this in the virtual world before taking action in the real world.

In addition to conducting research work, the IBDT is creating a data portal for UofT facility data, which will enable facility users and other researchers at UofT to access and use data about each building, their performance, and operational characteristics.

Intelligent Buildings and Digital Twins

Before exploring how the IBDT intends to achieve its goal of virtualizing built assets, it is important to establish a common definition for two fundamental concepts: intelligent buildings and digital twins.

In simple terms, a “connected” building is one which is equipped with sensors and IoT systems for data collection, while a “smart” building uses analytics of such collected data to control building operations. On the other hand, an intelligent building employs data analytics to profile its users' activities and become aware of its own conditions. An intelligent building aims to interactively adapt to a building users' needs and provides them with suggested actions to promote both wellbeing and optimized productivity, all while enhancing the overall sustainability of the building.

The definitions of digital twin vary between researchers and practitioners. At IBDT, we adopt a definition that describes a digital twin as a digital replica of the actual conditions and possible states of the three key facets of a facility: physical, (decision) processes, and people. A digital twin can be understood as a tool which combines analytics, process modeling, and building-user interaction to imagine alternative decision-making protocols, examine scenarios for optimizing energy and user wellbeing, and, most importantly, a tool to engage building users in such a way that empowers them to lead the innovation of new operational schemes (i.e., by means of providing insights obtained from data analytics and artificial intelligence).

Why Do We Need Digital Twinning?

Advancing the built asset industry in Canada has extensive potential. To begin, the stakes are high. According to NRCan ^[1] , Canada has over 400,000 institutional buildings, accounting for 18% of emissions. Investments in rehabilitation by the Federal Government only are estimated at $3 billion.

The efficiency of current asset management tools is lagging. As a case in point, in 2012-2018 and – despite substantial investments in asset management tools for water systems, the watermain break rates in North America increased by about 27% ^[2]. The social costs of these events have been found to reach as high as 400% of construction costs on certain projects ^[3].

Finally, the rewards for increased efficiency in decision making in the built asset industry are significant. Proactive maintenance is 12-18% cheaper than reactive maintenance^[4]. The financial return on investment on timely maintenance decisions is estimated to be 10 times that of delayed reactive maintenance ^[4].

The built asset sector pathway to enhance productivity, advance operational efficiencies and to decarbonization is complicated due to systemic problems.

• Complex decision making: Owners (especially of residential and small commercial units), most contractors and many consultants lack knowledge about how to decarbonize their facilities and how to make decisions and how to manage projects effectively.
  • The sociotechnical nature of decision making: The decision making is based not only on technical or even financial considerations, but on a complex social, regulatory, and economic set of criteria.
  • Context-dependency: stakeholders consider and evaluate a multitude of options, which are highly dependent on the context of the setting, the nature of the facility and the profile of owners.


• Outdated supply chain:
  • Fragmented industry structures: to design and rehabilitate a facility, an owner has to navigate an industry that is dominated by short-term, project-based thinking.
  • Risk-transfer contracting: for long, the industry has been dominated by a bidding system that is based on contentious transfer of risks between parties, which promotes mistrust; and punishes innovation.
  • ad hoc processes: Even if a technology solution is clear for a specific context, a set of ill-defined processes must be managed to realize the decarbonization solution. These range from outdated permitting processes, complicated contracting, and a construction process laden with schedule and budget overruns.


• Digitization backlog: obviously, digitization can be a great help in addressing the decarbonization challenges. However, this is faced with several key challenges:
  • Level of usage: the construction industry has some of the lowest levels of utilization of information technology resulting in inefficient performance. In fact, it is the only economic sector that has lost productivity over the last 40 years, while every other sector gained.
  • Orientation: the majority of IT systems in the construction industry target engineers and architects and focuses on increasingly deeper technical analyses. Very little work has been directed at other stakeholders, particularly owners and facility users.
  • Technical vs business tools: Limited work has been done on the business side of the industry (managing supply chains for example). There is also a significant lack of awareness of the socio-economic nature of the problem (the need to profile users and address their expectations)
  • The dominance of top-down thinking. The domain, particularly BIM, has been dominated by a belief in the role of experts in setting data models and in developing rules. This also has been dominated by a thinking oriented towards using structured data only (and even structuring unstructured data as the only solution to handle text and other unstructured data). This goes against the prevailing belief that unstructured data is a key source for knowledge capturing; and in contrast to the ad hoc and subjective nature of construction data.
  • Outdated data sharing and governance systems: there is a common resistance to sharing data in the domain—possibly because of the perception that sharing data can reduce competitiveness. Similar concerns exist in other domains and rational solutions were introduced to highlight that data sharing (and even coopetition) is better for business and for the whole sector in the long run. The lack of utility of IT systems to non-technical teams, and the neglect of the business and socio-economic aspects of the sector did not help too.

The Opportunity with the University of Toronto’s Digital Twin

The following features make our dataset unique:

• Data triangulation: longitudinal structured data (e.g., IoT data), unstructured data (e.g., complaints, and maintenance logs), and building contextual data (e.g., weather and community profile).
• Data reliability: The engagement of facility operators will provide researchers with a "ground truth” view of data, which can help overcome data reliability and completeness issues.
• Higher-value data: providing access to the data for other researchers working on other topics such as energy and air quality will generate new sets of insightful data (e.g., generate new simulations of possible scenarios).
• Occupants & their data: For a long time, informatics research in the domain has focused on professionals. The IBDT will have access to the (typically illusive) occupant data; and to occupants themselves (i.e., innovative stakeholders with unique knowledge profiles who can create ideas and identify new applications).