An investigation of the effects of traffic induced local dynamics on global damping estimates using operational modal analysis

Simulations of estimating the modal damping on a bridge from hour-long records of traffic loading were conducted by combining physics-based finite element modeling and signal processing. The finite element method was used to model a bridge consisting of a series of stringer beams resting atop a larger girder. The traffic loads were separated into trains and cars, with the trains modeled as partially distributed moving masses traveling along the girder and the cars modeled as point loads moving along the stringers. From the acceleration time histories, different operational modal analysis (OMA) techniques were used to find estimates for the modal coefficients of damping. The results demonstrated that a quasi-periodic component in the traffic loading introduces significant error to the damping estimates. This error could be observed in the distortion of the peaks for the power spectral densities (PSD) generated from the responses to the traffic simulations. The main OMA technique explored for the damping estimates was Enhanced Frequency Domain Decomposition (EFDD), but it could not compensate or correct for any alterations to the PSD. Other techniques such as the Stochastic Subspace Identification (SSID) method and curve-fitting frequency domain analysis were evaluated, but they produced comparable damping ratio estimates to EFDD and similarly resulted in large errors for the distorted modes. The influence of quasi-periodic loads was perceptible, which means that for certain cases, the nature of traffic loads may result in damping estimates that are considerably inaccurate no matter what OMA technique is chosen.