The term “Digital Twin” pops up every now and then in articles that discuss the use of information models in maintenance.

I find the term extremely accurate because it makes us think about the environment in which buildings are analysed and managed. The term itself is not new and has been in circulation for years—for example, in the manufacturing industry. In building maintenance, if we want to manage the massive amount of data generated by sensors in an understandable form, we need a digital twin of the building.

After collecting information on the building and creating new information through analysis, we need a way of visualising the information for the user. At its simplest, visualisation could mean painting the space according to the measured temperature. So, for example, people immediately understand that the redder the colour, the hotter the space.

A digital twin is created through APIs and platform thinking

In order to create a digital twin, we need at least the following building blocks:
1. A graphic information model of the building, with standardised or known data content.
2. A platform that connects the graphic view with cloud-based data content.
3. Documented (or open) interfaces that different systems used for communication.

Figure 1: An outline of a development environment created using documented APIs.

One important thing to understand is the difference between static and dynamic information. An example of a service area that generates static information is ventilation. It doesn’t change unless some concrete alterations are made in the building.

Temperature, on the other hand, is an example of dynamic information. It is constantly measured and the measurement results indicate whether everything in the space is satisfactory. If everything is not as it should be, the static information from the ventilation service area can provide additional details, such as exhaust air temperature.

This means a connection is built within a service area that covers one or more spaces, allowing the system to automatically find answers to the question: “Why is it hot in the space?”

The system must be able to analyse the problem and provide the user with possible causes for it. Information is also collected outside of the building automation system—for example, from IoT sensors that use APIs. The digital twin then uses this information to generate an overview of the building.

Figure 2: An example of a Granlund Manager Metrix prototype where the building automation system enters temperature data to a cloud-based information model.

Vision: “A virtual property where the user and the building communicate with each other.”

Granlund has released its Innovation strategy for 2017–2021.

The vision in our innovation strategy is to create: “A virtual property where the user and the building communicate with each other.”

When we connect the virtual property with the available static and dynamic building information, we can use the outcome to generate new information. Further analyses will make reporting on the property easier.

The next analysis step is to predict the behaviour of the property. At this stage, we can automatically simulate, for example, energy consumption and indoor conditions for the following day, based on the weather forecast, expected number of people and timetables. This information allows us to control the system proactively without manual adjustments.

This process takes us one step further towards a learning property. Over the years, the property will develop its own form of AI, learning from its own mistakes—for example, if the predicted situation did not match the actual conditions of the property.

A learning property needs feedback from users. With the help of such feedback the property can optimise conditions for those users. Artificial intelligence cannot tell us what the user wants—we need user feedback for that. Furthermore, by monitoring and analysing the user of the property anonymously, we can gain an overview of how users affect the functioning of the property. The building services control system can learn a lot from knowing the utility rates of the property at different times—not to mention how useful this information is for company management in determining how suitable the property is for the company’s business.

However, we cannot look past the physical features of the property; the property must function as it was intended to function. In Granlund’s strategy for 2020, our mission is “Well-being in the built environment”.

Our innovation strategy also follows this mission: “Everything we do is guided by use”.

Figure 3: In spring 2017, Granlund released its innovation strategy, which emphasises users and well-being.

 

Writer Tero Järvinen works in Granlund as Technology Director.

Share This