Through the collection of data via sensors in smart buildings, value can be created by assessing the collected data, and thus derive the potential energy-saving measures, reduce water consumption, and more. Metadata exists about the sensors installed, but the Linked Metadata which would allow to easily understand the dependencies of the data collected, e.g., on the thermal level, is missing. This results in a lack of knowledge about the relationship of the data recorded, the physical location of the sensors, and what is measured in detail. Consequently, a lot of implicit expert knowledge is associated with the data.

The task was to evaluate how a prototype can be developed, utilizing a mixed research approach, combining quantitative and qualitative research. During the qualitative research, papers, videos, and web seminars were used to grasp the concept of the ontologies. Quantitative research focused on numerical data, such as statistics on how many concepts the ontologies can cover. As a result, existing smart building ontologies were compared and evaluated. Furthermore, an appropriate ontology was chosen to create a model to evaluate various use cases in the context of the NEST building’s data points.

While many different smart building ontologies can be utilized to depict the relationships and entities of a building, a comparison of available ontologies has been conducted. Additionally, the underlying technology to implement a digital twin was presented. The results showed that as of today, the industry leading ontologies provide similar descriptive classes and properties to define components of buildings. These ontologies are currently under development to become more interoperable and extendable with another, merging the best aspects of the currently existing descriptions. Brick Schema was the fitting ontology chosen to consequently create a model to run the use cases. The approach on how the generation of such a model can be created with various tools was provided. With the use of the generated model, the defined use cases were evaluated, and the existing gaps were shown. The remaining partially fulfilled use cases would need further steps to provide the user with the specific answers. Furthermore, an outlook was provided on how the model may be extended, and how an interactive data analysis measurement application at Empa could look like in the future.