Continuing Education

Topology Optimization of Top Lateral Bracing for Steel Tub Girder Systems Using Genetic Algorithm

The use of steel trapezoidal box girder systems has gained increasing popularity in bridge applications as they feature high torsional stiffness and are aesthetically appealing. However, during the concrete deck pour, the top flanges are in compression and the entire girder is susceptible to the failure mode of global lateral torsional buckling (LTB). Several incidents have occurred ranging from the excessive deformation to the complete bridge collapse. Usually a lateral truss system is installed at the top flange level to form a ""pseudo-closed"" section and to help resist LTB before concrete hardens. However, the installation of top lateral bracing along the entire girder span, albeit common in current practices, might not be efficient given the differential girder shear deformation distribution along the length. This paper presents a general approach for the topology optimization of the top lateral bracing configuration for the steel tub girder system. The optimization is formulated based on a modified genetic algorithm (GA) in conjunction of the 3D finite-element analysis implemented in Python-ANSYS APDL AAS coupling programming environment. The truss member number and connectivity are encoded in real-valued chromosomes and the objective function of the optimization is to minimize the total weight of the top lateral bracing system subjected to buckling constraints using the penalty function. Case studies are carried out and the optimized bracing configurations are compared with those from the previously-published studies. The results show that the proposed approach allows successful optimization of partial top lateral bracing system with improved efficiency and buckling resistance. The approach can also be used for the optimization of the lateral bracing of other long and slender girder systems.

  • Date: 4/2/2019 - 4/5/2019
  • PDH Credits: 0


Liwei Han; CHI Consulting Engineers; Summit, NJ; Yang Wang; The University of Texas at Austin; Austin, TX

View content