Digital twin of the laboratory bridge demonstrator
Source: BAM
Using digital twins for decision making is a promising concept that has gained significant attention in various fields. It involves merging powerful simulation models with real-world sensor data, offering valuable insights and support for maintenance decisions and reliability investigations.
The success of utilizing digital twins for decision making hinges on two crucial factors: the quality of the data and the quality of the digital twin itself. High-quality data is essential to ensure accurate and reliable predictions and recommendations. On the other hand, the digital twin's quality depends on precise modeling assumptions and accurate parameters that characterize the virtual twin.
The paper addresses the challenges faced when implementing the digital twin concept for a demonstrator bridge in a controlled laboratory environment. The authors discuss various critical aspects of this implementation, shedding light on data management, iterative development of the simulation model, and the identification of model parameters. Bayesian inference is a powerful statistical technique that allows for the incorporation of prior knowledge and real measurement data to estimate the model's parameters. However, when dealing with many parameters, this process becomes computationally intensive and challenging. The paper proposes to use variational methods to tackle this complexity effectively.
The study also investigates different scenarios related to the digital twin's application, including iterative identification of structural model parameters and damage identification. The researchers demonstrate how the digital twin can be utilized to monitor the bridge's health and detect potential damage.
In addition to addressing the technical challenges, the paper emphasizes the importance of reproducibility in research. The authors aim to provide all the models and data used in their study in a transparent and accessible manner. This approach allows other researchers to replicate their experiments, validate their methodologies, and build upon the existing work, fostering collaboration and advancement in the field.
Bayesian model calibration and damage detection for a digital twin of a bridge demonstrator
Thomas Titscher, T. van Dijk, Daniel Kadoke, Annika Robens-Radermacher, Ralf Herrmann, Jörg F. Unger
published in Engineering reports, Article number e12669, Pages 1–27.
BAM Department Safety of Structures
BAM Division Buildings and Structures
BAM Division Modelling and Simulation
BAM Department Non-destructive Testing
BAM Division Sensors, Measurement and Testing Methods