Finite element model updating for the dynamic response of continuous deck railway viaducts leading to efficient digital twins

Bridges and viaducts are key infrastructure elements within high speed railway lines, for which dynamic actions due to traffic are one of the main actions for design of new structures or evaluation of existing ones. This paper describes techniques based on experimental measurements and numerical finite element models leading to efficient digital twins, to be used for predictive maintenance, within the framework of an innovative prototype for a continuous predictive maintenance system installed in the La Marota 381 m long viaduct, in the high-speed line between Córdoba and Málaga in Spain. This system includes both model-based methods as well as data-based methods for maintenance decisions. Here we focus on the work for model-based acceleration performance, comparing in real time measurements from accelerometer sensors and predictions from a digital twin. The digital twin is an autonomous structural modal analysis software derived from a finite element model previously calibrated with Operational Modal Analysis experimental data. Results obtained from 3 representative high-speed trains are shown and compared in terms of root-mean-square (RMS) and moving window RMS of accelerations between the measured and numerical predictions. These allow to establish criteria and thresholds for predictive maintenance alerts.