Finite element model updating and damage identification using semi-rigidly connected frame element and optimization procedure: An experimental validation

Abstract

Owing to ever-increasing complexity of engineering structures, developing a methodology for the early detection of defects has become crucial to ensure their long-term safety and reliability with the least amount of expense. There are always discrepancies between experimental and numerical modal data because of unknown structural parameters and uncertainties. The finite element (FE) model updating techniques attempt to minimize the differences by adjusting the unknown parameters of the FE model. Therefore, the FE model updating methods are essential for developing a baseline model and accurate damage identifications in subsequent steps. This paper employs the semi-rigidly connected frame element (S-RCFE) instead of the standard Euler-Bernoulli beam element for assembling the FE model of the experimental beam and establishing a high-fidelity numerical model. The S-RCFE with extra design parameters, including the end fixity factor of all connections, enables us to achieve a reasonable agreement between experimental and numerical models through the optimization-based procedure. In FE model updating step, two objective functions based on modified total modal assurance criterion (MTMAC) and changes in natural frequencies are used to minimize by three optimization algorithms, viz, grey wolf optimizer (GWO), gradient-based optimization (GBO), and an improved version of GWO (IGWO). The influence of the S-RCFE and standard Euler-Bernoulli beam on the model updating accuracy is also examined, and the efficiency of S-RCFE is evaluated. The statistical results reveal that GWO-MTMAC and IGWO-MTMAC can be successfully implemented for FE model updating with almost the same performance. However, IGWO provides the most reliable results with a relatively extensive computation time for damage identification in all scenarios. In some damage scenarios, the GWO and GBO perform comparably with very similar running time. Data used in this article can be found at https://github.com/Samir-Khatir/Data-modal-analysis-of-healthy-and-cracked-beam.git