Filipe Amarante dos Santos

This paper analyses and compares the dynamic behavior of superelastic shape memory alloy (SMA) systems based on two different constitutive models. The first model, although being able to describe the response of the material to complex uniaxial loading histories, is temperature and rate independent. The second model couples the mechanical and kinetic laws of the material with a balance equation considering the thermal effects. After numerical validation and calibration, the behavior of these two models is tested in single degree of freedom dynamic systems, with SMAs acting as restoring elements. Different dynamic loads are considered, including artificially generated seismic actions, in a numerical model of a railway viaduct. Finally, it is shown that, in spite of its simplicity, the temperature- and rate-independent model produces a set of very satisfying results. This, together with its robustness and straightforward computational implementation, yields a very appealing numerical tool to simulate superelastic passive control applications.

Nowadays, modern analysis of Civil Engineering structures implies the use of increasingly sophisticated computer models, designed not only to predict the response of actual structures to different loadings, but also to simulate the effects of eventual modifications in their structural configuration. Nevertheless, it is often discovered that, when the numerical simulations are compared with experimental data, the degree of correlation is weak preventing the use of the FE models with confidence in further analyses. In such cases, FE updating techniques are available to correct the FE models, based on dynamic response records of the real structures. These updating processes usually consist of four phases: a preliminary FE modeling, an experimental modal identification, a manual sensitivity analysis and, finally, an updating of the FE model. The present paper presents all four phases of a successful updating process for the FE model of a footbridge structure. It is shown how the last phase of the process can be performed fully automatically, by coupling an optimization routine with a commercial FE analysis program.