A Data-Driven Damage Assessment Tool for Truss-Type Railroad Bridges Using Train Induced Strain Time-History Response

ABSTRACT In this paper, a non-parametric damage detection method for truss-type railroad bridges is presented. The method uses operational strain time-history responses to detect damage in truss elements, and change in support behaviour. Dynamic strain time-history responses obtained under baseline and unknown-state bridge conditions are used to compute the magnitudes of differences in strain values between two successive time-steps. A new damage-sensitive feature (DSF) is proposed as the change in percentage of the square root of the sum of squared values of these magnitudes. After establishing a threshold DSF based on the baseline bridge, further structural change or damage in the bridge could be detected and located by observing the values of the DSFs. The validity of the method is investigated through finite element analysis of a steel-truss railway bridge. It is demonstrated that the proposed method yields promising results for identifying, locating, and relatively assessing the damage, and could be useful even when different operational conditions (i.e. different train speeds and loads) and measurement noise influence the strain data. Therefore, the proposed method has the potential to assist in developing effective maintenance strategies for railway bridges.