Use of genetic optimization in parameter identification of reinforced concrete bridge girders

  • PDF / 3,105,411 Bytes
  • 13 Pages / 595.276 x 790.866 pts Page_size
  • 85 Downloads / 184 Views

DOWNLOAD

REPORT


PRACTICE-ORIENTED PAPER

Use of genetic optimization in parameter identification of reinforced concrete bridge girders Fatima El Hajj Chehade1,2,3 · Rafic Younes3 · Hussein Mroueh1 · Fadi Hage Chehade2 Received: 11 March 2020 / Accepted: 10 July 2020 © Springer Nature Switzerland AG 2020

Abstract The real response of a given structure, such as deflection, generally differs from prediction obtained from theoretical models. Hence, the need of periodic or long-term structural monitoring has become mandatory. The present paper describes an approach to localize the most suitable period to perform deflection monitoring in case of reinforced concrete bridges. The searched period allows making the best updates of theoretical models to better represent the structural behaviour with time. So, the genetic optimization is employed to update an existing law of flexural rigidity in order to minimize the difference between the measured and predicted deflections; measurements will be considered from different time periods. The proposed methodology is applied to a representative set of 21 RC T-beam bridges with variable parameters. Simulated measurements of static deflections based on weigh-in-motion data collected in some European countries are used in the application. Keywords  Genetic optimization · Deflection · RC bridges

Introduction Model updating is the popular name for using measured data to identify or to update the parameters of a theoretical or finite element model. There are two main types of model updating methods: residual minimization and Bayesian model updating. In the residual minimization method, the values of the searched parameters are chosen to minimize the difference between predicted and measured values while in Bayesian model updating the Bayesian conditional probability is used to update the prior knowledge of model parameters [1]. Model updating methods have been successfully applied in many civil engineering-related studies. For example, Yuen et al. [2] used an extension of a Bayesian system identification approach to identify the stiffness parameters for a 15-story building based on available dynamic data. Hajela * Fatima El Hajj Chehade [email protected] 1



Laboratory of Civil Engineering and geo‑Environment (LGCgE), Lille 1 University, Villeneuve‑d’Ascq, France

2



Modelling Center Doctoral School of Science and Technology, Hadath, Lebanon

3

Faculty of Engineering, Lebanese University, Hadath, Lebanon



and Soeiro [3] applied the output error method for local damage identification in truss members based on simulated measurements of static deflection. Sanayei and Scampoli [4] used static parameter identification algorithm to predict the plate bending stiffnesses for a reinforced concrete pier deck from simulated static deflection data. A probabilistic global search algorithm PGSL was used in [5] to identify the boundary conditions that lead to match, as closely as possible, deflection data from a static load test on a bridge. Catbas et al. [6] were based on static strain data to