Bridge displacement estimates from measured acceleration records

Girder displacement is an important component of bridge design and evaluation because it is directly related to bridge stiffness and flexibility. The complex interaction between vehicle and bridge dynamics will result in vibration and deflection greater than that under equivalent static loading. However, field measurement of girder displacement generally is non-trivial. Different sensors and instrumentation systems can be used to measure bridge displacement. Some of these provide direct measurements, whereas others indirectly provide displacement through measurements of the angle of rotation, velocity, or acceleration of the girder. Indirect methods, however, require additional signal processing analyses to remove the effect of small errors in recordings caused by sensor drift, unknown initial bridge conditions, and signal noise. These errors become sufficiently large through successive integrations and greatly distort the integrated velocity and displacement signals. Two correction methods are examined, namely, the velocity estimation method and the linear baseline correction method (BCM), for minimizing such errors in obtaining displacements indirectly from acceleration records. Independently measured girder acceleration and displacement records from a three-span continuous bridge under controlled live load testing are used to evaluate the accuracy of each method. It is found that the linear BCM results in a corrected displacement profile that more reasonably approximates the measured trace under various loading patterns. The choice of integration boundaries is also shown to affect the accuracy of both methods. An objective approach based on the energy content of the signal is proposed.