Behaviour of steel joints under dynamic actions

Abstract

The extensive use of steel framed structures as a solution for fast, efficient and well-controlled construction process has stimulated the research on the behaviour of steel connections under static and seismic loading regimes. Nonetheless, there is a limited amount of knowledge around the response of steel joints subjected to transient dynamic loads. Exceptional loading situations due to explosions or impact can lead to the failure of structural elements directly exposed or in the proximity of the loading source. When a joint between two or more structural elements has failed, potential progressive collapse of the structure may occur as a result of the propagation of initial failures. Therefore, to make the built environment safe, it is important to evaluate accurately the resistance capacities and ductility of steel joints. Current design standards provide a very limited guidance for the design of joints subjected to transient loads, mainly due to the lack of investigations conducted on this topic. The present thesis addresses these aspects by exploring several methods of assessment of the joints nonlinear response under quasi-static and impact loading. Experimental, analytical and numerical approaches are employed to predict the observed real behaviour of steel connections. The analytical and numerical procedures are validated against tests results available from the experimental campaign conducted within the research project RobustImpact. More than 20 specimens of double-sided beam-to-column joint configurations were tested under quasi-static and dynamic loading regimes. The analytical procedures used for the characterisation of the behaviour of steel joints under monotonic loads tend to estimate accurately the response experimentally observed. A specific limitation of the analytical model implemented in Eurocde 3 has been highlighted in this study. For static conditions, the equivalent T-stub model is likely to underestimate the ultimate resistance capacity when it is used for components with thin plates. The FE modelling techniques allow for the accurate simulation of the response of steel joints under both static and impact loading scenarios. Within the framework of this thesis several FE models that incorporate the material rate sensitivity were developed and validated against experimental results. A good agreement between the simulations and the physical tests results was obtained. The strain rate effects expressed in terms of Dynamic Increase Factors were quantified relying on two assessment methods - analytical and numerical. The analytical method used for the estimation of maximum impact forces leads to generally accurate predictions. Consequently, this method is used for the estimation of strain rate effects. The values found for the DIFs were in partial-to-good agreement with the ones estimated from the FE results, allowing for the identification of basic active components mostly susceptible to be affected by the deformation rate.