Bridging BIM and operational sensor data through knowledge graphs: an interoperable digital twin for building energy maintenance

The operational phase of buildings requires the integration of highly heterogeneous data streams, including Building Information
Modelling (BIM) models, Internet of Things sensors, computerised maintenance management systems, and technical documentation, yet
current BIM deliverables remain poorly suited to this task. This paper presents the JUMENGI project, which investigates how such
heterogeneous data can be structured and articulated within an interoperable digital twin dedicated to building energy maintenance. An
ontology-based information architecture, instantiated as a knowledge graph, is designed following a Competency Questions methodology and
implemented in a graph-oriented database. The approach is tested on two contrasting real-world case studies, a residential project with
available BIM models and sensor data, and a tertiary building initially lacking an exploitable BIM model, and operationalised through a digital
platform combining three-dimensional visualisation, sensor monitoring, and alert management. The results demonstrate the technical
feasibility of unifying multiple operational data sources within a single semantic model, while revealing persistent limitations of existing BIM
deliverables for operation and maintenance. The contribution lies in a validated, industrially grounded information architecture that bridges
design-oriented BIM practices and operational energy maintenance needs, with direct relevance to interoperability and digital twin research
within the built environment.