Effects of measurement statistics on the detection of damage in the Alamosa Canyon Bridge

This paper presents a comparison of the statistics on the measured model parameters of a bridge structure to the expected changes in those parameters caused by damage. It is then determined if the changes resulting from damage are statistically significant. This paper considers the most commonly used modal parameters for indication of damage: modal frequency, mode shape, and mode shape curvature. The approach is divided into two steps. First, the relative uncertainties (arising from random error sources) of the measured modal frequencies, mode shapes, and mode shape curvatures are determined by Monte Carlo analysis of the measured data. Based on these uncertainties, 95% statistical confidence bounds are computed for these parameters. The second step is the determination of the measured change in these parameters resulting from structural damage. Changes which are outside the 95% bounds are considered to be statistically significant. It is proposed that this statistical significance can be used to selectively filter which modes are used for damage identification. The primary conclusion of the paper is that the selection of the appropriate parameters to use in the damage identification algorithm must take into account not only the sensitivity of the damage indicator to the structural deterioration, but also the uncertainty inherent in the measurement of the parameters used to compute the indicator