The evaluation of deteriorated members with corrosion, and consequent potential degradation of strength,is found in existing plate girder bridges. This difficult engineer’s task is based in a low level of informationconcerning the shear behaviour of corroded web panels. This insecurity in the evaluation of deteriorationmembers of plate girders bridges must be minimized by using models based on sound theoreticalprinciples, validated by extensive and reliable data.In this paper are presented the results of degradation of strength, for critical and ultimate shear load, ontapered plate girders with local corrosion (web or flanges), using a nonlinear finite element analysis. In thisway, it is also presented the development of a data base with several scenarios of degradation for thecreation of predictive models for critical and ultimate shear stress in tapered webs, taking into accountdifferent scenarios of thickness reduction. In order to analyze the data generated in the finite elementmodel, Data Mining techniques (e.g., Neural Networks) have been used.
展开▼
