Towards Building Energy Digital Twins with a Low-Cost Portable Mobile Mapping System

Abstract

Building Energy Digital Twins require structured, simulation-ready representations integrating building geometry, topology, semantics, thermal properties and energy-related systems. For existing buildings, however, these representations are still produced largely through manual processes, particularly for the identification, 3D localization and parameterization of energy-relevant devices. This work presents a mapping-to-BIM/BEM workflow for generating simulation-ready Building Energy Models (BEMs) directly from low-cost mobile mapping data, providing a pathway towards scalable Building Energy Digital Twins of existing buildings. Structural geometry and spatial organization are reconstructed from 3D point clouds, while energy-related devices are detected in registered multi-view imagery and consolidated into unique 3D instances through cross-view semantic and geometric consistency. Technical information visible on device nameplates is extracted using Optical Character Recognition (OCR) and associated with the corresponding physical instances. Spaces, devices and their attributes are subsequently integrated into a typed semantic graph encoding spatial, topological and functional relationships. This intermediate representation is then translated into EnergyPlus epJSON and a BIM environment for downstream simulation and interoperability. The workflow demonstrates how heterogeneous observations can be transformed into structured, simulation-ready building information through computer vision methods and computational geometry, establishing a traceable and robust connection between 3D reality-capture data and Building Energy Simulation to support scalable Building Energy Digital Twins for existing buildings.

Main findings

...

Additional materials

...