PSI - Issue 64

Chongjie Kang et al. / Procedia Structural Integrity 64 (2024) 1232–1239 Chongjie Kang, Maria Walker, Steffen Marx/ Structural Integrity Procedia 00 (2019) 000 – 000

1235

4

Another way to classify bridge digital twins is to use the source of information provided. This source of information is called the digital entity. This classification is suitable for structuring the papers that were systematically researched in this study because, in contrast to the classification according to Fig. 2, it allows a detailed review analysis with regard to the advantages and disadvantages of the use cases in bridge construction. This analysis will be done in the next section. 3.2 Digital entity for bridge digital twins The work further focuses on the digital entity of DT, which functions as a source of information and is the basis for the implementation of the bridge digital twin. According to Honghong et al. (2023), five basic types of digital entities are commonly used to build bridge digital twins: information models, data-driven surrogate models, analysis models, 3D surface models, and federated models, which combine several approaches to the above model types. Tab. 2 lists the aforementioned model types, their functionalities, and commonly used technologies for creating digital twins of bridges. For each modelling approach, two research examples are presented.

Tab. 2. The model types, their functions, and used technologies in Bridge DT Model type Used Technologies

Examples

BIM model

Ontologies, Semantic Web, knowledge graphs

Herbers et al. (2024); Lazoglu et al. (2023)

Data-driven surrogate model

Monitoring data from IoT sensors, Data Analytics, machine learning, Cloud Computing

Braml et al. (2022); Hagen and Andersen (2024)

Analysis model

FEM, numerical analysis methods

Jeon et al. (2024); Smarsly et al. (2022)

3D surface model

Laser scanning, photogrammetry

Taraben et al. (2022); Morgenthal and Helmrich (2023)

Federated model

Combination of FEM, BIM, 3D point cloud

Chacón et al. (2023); Shim et al. (2019)

The function of an information model is to store and manage information about the entire lifecycle of the bridge. Using technologies such as BIM and ontologies, data is integrated into the bridge digital twin and linked to its components. With a data-driven bridge digital twin, rapid analysis of the bridge condition based on available data, identification of hidden relationships, prediction, and early warning systems become possible. In this case, the assessment of the bridge condition is purely data-driven, without consideration of physical relationships. The evaluation of the data can be done either with conventional methods of data analysis or with machine learning; refer to Hagen and Andersen (2024) and Braml et al. (2022). Physics-based analysis models, such as finite element modeling (FEM) or numerical models, are used to analyze the structural condition and provide the results of mechanical calculations. The use of digital twins, which combine physics-based models with measurement data, often involves validating or updating an analysis model using monitoring data. High-resolution 3D surface models from laser scanning and photogrammetry represent the current surface condition of a structure. In Taraben et al. (2022) and Morgenthal and Helmrich (2023), the possibilities of supporting bridge inspections with unmanned aerial systems (UAS) are investigated. By linking discrete images taken at different times, a condition history is created and automatically evaluated in bridge digital twins. Combined approaches use several of the aforementioned model types and link them to a federated model. In Shim et al. (2019), BIM models, 3D surface models, and FE models are federated. In Chacón et al. (2023), an environment is created in which BIM, sensor data, and structural analysis are combined. Tab. 3 summarizes the advantages and limitations of the presented model types that are used for the implementation of bridge digital twins.