Finite element model updating of a simply supported skewed PSC I-girder bridge using Hybrid Genetic Algorithm

Abstract

Hybrid Genetic Algorithm (HGA) which combines the genetic algorithm as a global optimization and the simplex method as a local optimization is proposed for a finite element model updating of a real prestressed concrete bridge structure. In order to minimize the updating error between the measurement and the finite element model updating result, objective functions which are combinations of fitness functions based on the natural frequency, the mode shape and the static displacement are introduced. And an interface tool is also developed in order to utilize various element library and numerical analysis tools which are provided by commercial finite element and numerical analysis programs. A simply supported skewed PSC girder bridge which has 30 m span length is selected for the verification of the proposed FE model updating algorithm. Static vehicle loading test and forced vibration test by traveling vehicle as well as ambient vibration test were carried out to obtain the reference measurement data for numerical updating. A grillage model is used for the finite element analysis. Effect of the spring element to simulate the realistic support condition which is not perfectly free or restrained in real situation as well as that of the objective function on the updating accuracy are studied. From the result of parametric study, it is investigated that the use of spring element for support condition is effective to minimize the updating error for natural frequency and mode shape. Furthermore, including the static displacement fitness function together with those of dynamic properties may improve the global behavior of updated finite element model. It is concluded that the hybrid genetic algorithm proposed in this study is a very effective finite element model updating method to find an accurate result in updating real bridge structure based on measured data.