Data-driven tools applied on measured structural responses enable extraction of valuable information on the behavior of inservice structures and identification of the actual system in its “natural habitat”.

An output-only testing campaign on two railway bridges is in the focus of this study. The specifics of the dynamic behavior of the structures are studied under three different scenarios. More precisely, the measured accelerations at relevant points of the structures, for unloaded and loaded conditions, are herein employed with the following objectives:

I) Structural dynamics identification via estimation of the first natural bending frequency of the structures during ambient unloaded conditions using the AutoRegressive Moving Average (ARMA) method and the time-domain Subspace System Identification (SSI) method

II) Assessment of traffic safety aspects based on the maximal amplitudes of vibrations during loaded operating conditions of the structure, for two separate train induced vibrations (train velocity of 40 km/h and 60 km/h)

III) Evaluation of the dynamic effects through a dynamic amplification factor estimated from the recorded responses during dynamically loaded structure (train velocity of 40 km/h and 60 km/h).

By utilization of the response-based data analysis methods the first natural frequencies of the steel structures are successfully identified and compared with FEM numerical results. The acceleration data is further assessed by preprocessing filtering techniques and extracted features are assessed according to EN 1990:2002 or UIC (International Union of Railways) code 776-2 (2009).