Modeling and Prediction of Time-Dependent Deformations of the I-35W St. Anthony Falls Bridge

To properly identify anomalous data in a structural monitoring system, the expected behavior of the structure under changing environmental conditions must be accurately determined. The St. Anthony Falls Bridge, a post-tensioned concrete box girder bridge, was examined as a case study for testing a data normalization scheme for temperature-and time-dependent deformations of an in situ structure. A methodology based on linear regression was developed to extract time-dependent behavior from the measured data. An Arrhenius-adjusted time was proposed to correct for the slowing and accelerating creep and shrinkage rates during the winter and summer, respectively. The extracted time-dependent deformations were compared to estimates using the finite element method. Bayesian regression was used to update the finite element model predictions using the measured data, substantially narrowing the expected bounds and facilitating anomaly detection.