Response Surface-Based Model Updating to Detect Damage on Reduced-Scale Masonry Arch Bridge

Abstract

Response surface (RS) methods, a combination of mathematical and statistical techniques, have been widely used in design optimization, response prediction, and model validation in structural engineering systems. However, its usage in structural damage identification, especially for historic structures has not been quite common. For this purpose, this study attempts to investigate damage detection in a masonry arch bridge. Within the scope of this, a reduced-scale model of a one span historical masonry arch bridge was built in a laboratory environment. To determine the modal parameters of the reduced-scaled bridge model, operational modal analysis (OMA) was performed under ambient vibrations. Signals originated by sensitive accelerometers were collected to quantify the vibratory response of the reduced-scaled model bridge. The experimental natural frequencies, mode shapes, and damping ratios resulting from these measurements were figured out by using the Enhanced Frequency Domain Decomposition (EFDD) technique. The three-dimensional model of the reduced-scale bridge was created in the ANSYS finite element (FE) software program to expose the analytical dynamic characteristics of the bridge model. The results obtained in the experimental application were compared with those of the finite-element analysis of the bridge model. The calibration of the numeric model was utilized depending on the experimental modal analysis results of the reduced-scale bridge by using the RS method. Design of experiments was constructed by using central composite design, and the RS models were generated by performing the genetic aggregation approach. The optimum results between the experimental and numerical analyses were found by using the RS optimization. Then, regional damages created on the scaled model and the changes of dynamic properties of the damaged case were evaluated. The damage location was approximately identified by using the RS method in the calibrated finite-element model. The results demonstrated that the RS-based FE updating approach is an effective way for damage detection and localization in masonry type structures.