Sub-project C06 - Modular experimental structure completed
The basis for the model development will be a comprehensive BIM model, from which FE models will be extracted automatically. For this purpose, an interface between IFC and Ansys will be used and extended accordingly. First interface extensions for 1D structures have already been created and tested.
For the creation of model predictions, the load on the structure must be determined. This is to be done with artificial neural networks. For this purpose, a numerical example bridge was created and loaded by different vehicle crossings. Then, the accelerations at different locations of the structure were used to identify the space and time dependent parameters of the load. Convolutional neural networks (CNNs) are used for the load identification process. First results show a very good convergence and possibility to reconstruct the load.
These reconstructed loads will be used as input for the model predictions in order to apply the data assimilation algorithms. Uncertainties in the measurement as well as inaccuracies in the loading will be considered.
All methods developed and yet to be developed will first be tested on a substitute structure in the laboratory. For this purpose, a modular bridge made of wood and steel was designed and built as part of the project. At the end of August, the individual parts of the bridge will be bolted together, and the bridge geometry will be measured by Prof. Neitzel and his team (project A03) as part of a cooperation.