Performance of composite steel/concrete members under earthquake loading. Part II: Parametric studies and design considerations

SUMMARY In this paper, an advanced analytical model proposed and verified in a companion paper (Part I: Analytical model) is used to conduct parametric investigations of ductile partially-encased composite beam-columns. Particular attention is given to a new configuration of composite member developed and tested at Imperial College. A detailed account of the analytical results is given, and the most significant observations are highlighted and discussed. Quantitative assessments of plastic moment capacity, rotation and displacement ductility and plastic hinge length are undertaken. The implications of the main findings on the seismic design process are also presented. Finally, a procedure for ductility-based design according to modern codes of practice is outlined. Although experimental work is essential in the understanding of the non-linear behaviour of structures, as well as in the verification of analytical models, laboratory testing is time-consuming and capital-intensive. Consequently, accurate and efficient analytical tools are invaluable in enriching and directing experimental investigations. Appropriate analytical models can be employed in the design of testing schemes, as well as in achieving better understanding of complex structural behaviour. Furthermore, reliable analytical methods can provide an abundance of information through parametric studies on realistic engineering problems, at a fraction of the cost of experimental investigations. In Part I of the current study (Analytical model),' an advanced non-linear dynamic analysis tool, developed by the writers to represent the behaviour of composite members and frames, was described. The proposed model was verified and compared with results of experiments conducted by the writers and their co-workers on conventional and novel configurations of partially-encased composite members. It was shown that the developed computer code provides accurate predictions of the non-linear behaviour of composite members under various combinations of loading. It is, therefore, applicable to refined non-linear dynamic analysis required by modern codes of practice for design verification of structures of special importance or those classified as irregular. For regular structures, however, simplified approaches are used, whereby idealized elastic response spectra are scaled appropriately using a structural behaviour factor to account for inelastic response. This requires not only an evaluation of the available local ductility in critical members but also a relationship between the local member ductilities and the overall structural behaviour factors. In Eurocode 8,2 the role of the behaviour factor is assumed by the parameter 'q' which is used to modify the elastic design spectrum. It accounts implicitly for inelastic behaviour and other force-reduction effects. This factor can be derived as the average ratio between the seismic forces inducing an ultimate limit state in the structure taking into account its actual non-linear behaviour, and the design seismic loads obtained by conventional linear analysis. The evaluation of the behaviour factor depends on the structural configuration,