Model updating for structural health monitoring using static and dynamic measurements

Abstract

Structural health monitoring is tracking static or dynamic characteristics of a structure to identify and localize stiffness reductions for damage detection. Different damage indicators are used and any indicator presents advantages and drawbacks. Hence the idea comes up to combine them in a model-updating procedure using a finite element model. In a first step, a model is fit to match the healthy reference state of the examined structure. Therefore it relies on minimizing a special objective function adding and weighting the differences between measured and calculated static and dynamic structural characteristics. For doing structural health monitoring the measurements are repeated in distinct time intervals and the finite element model is updated again, using the same objective function and minimization procedure. Damage can be identified and localized by highlighting reductions in the stiffness matrix of the model compared to the initial model. The efficiency of the method is illustrated by in-situ tests, where a single beam is examined that was part of a real prestressed concrete bridge. For static tests, 8 displacement transducers were disposed along the length of 40m, while the beam was mass-loaded and the deflection line is analyzed. Modal analysis was performed with swept sine excitation with constant force amplitude to identify eigenfrequencies and mode shapes. Stepwise artificial damage was provoked by cutting multiple prestressed tendons inside the concrete beam. A finite element model with a mapped mesh was created, allowing a variation of Young’s modulus in grouped sections. On real bridges temperature is neither homogenous nor constant over time, which often has a considerable influence on measured static and dynamic characteristics as the stiffness of asphalt and/or bearings can be affected. The proposed methods show their efficiency when temperature effects were excluded or compensated after measurement, which is a topic on its own and not discussed here.