Distributed bending stiffness estimation of bridges using adaptive inverse unit load method

Abstract

Bending stiffness has important applications in bridge damage detection, model updating, and load bearing capacity assessment. However, accurately predicting bending stiffness is a challenging task. In this paper, a two-phase methodology named the adaptive inverse unit load method (AIULM) is developed for distributed reconstruction of bending stiffness from onboard deflection data. In the first phase, the bending stiffness is roughly estimated by uniform discretization. A global stiffness smoothing method is applied to eliminate nonphysical discontinuities of stiffness at interelement boundaries. Then, the smoothed results are applied in the second phase to define a node optimization function to adaptively adjust the element node distribution. An accurate bending stiffness distribution is provided by performing AIULM analysis again based on optimized discretization. Since only real bridge deflection influence lines (DILs) are utilized in the formulation, the AIULM provides a C0-continuous stiffness distribution without any information on internal force. The adaptive optimization of the node distribution allows the practical application of AIULM to different bridges, especially old bridges. Numerical and experimental validation cases, which demonstrate the excellent predictive capability and practical usefulness of the AIULM, have been performed.