Robustness of Steel Connections Loaded at High Rates

Robustness of structures to prevent progressive collapse has been considered by different building codes since the Ronan Point accident in the UK in 1968 and research on this type of collapse was brought to the fore after the building collapses at the World Trade Center on 9/11 2001. In the case of losing a column, to maintain the integrity of the building large end rotation of the beams is likely to be required in order to transfer the vertical loads from the floor above by redistribution of loads through the remaining structure. For this reason connections are important in enabling the structure to bridge over the lost column. Connections also need to be capable of resisting high strain rates arising from the dynamic redistribution of moments and tension loads which means they require high ductility to dissipate energy and undergo deformation without failure.
This thesis presents results from both experimental tests conducted on connections under quasi-static and rapidly applied loading (performed as part of an EPSRC funded research project [1] Prior to the PhD research), and a series of ANSYS LS-DYNA finite element analyses which have been developed to model the connection response. Web-cleat and End-plate connection models, with their complex geometry, has been produced and validated against experimental data. The effect of rate of applied loading on connection response has been investigated both experimentally and by numerical modelling. It is important to understand the effect of loading rate on the response of connections in dynamic frame response to column loss. The results indicate that web angle cleat connections are relatively insensitive to the rate of loading within the range considered here. This suggests that static characterisation of web angle cleat connections may yield suitable data for use in dynamic analysis of frame response in column loss scenarios. End-plate connections behaved differently and this type of connection was sensitive to loading rates and this showed that static characterisation of this connection is not enough for dynamic analysis. This study highlight a gap in knowledge of welded material properties, and provided the results of parametric studies (different loading types, rates and connection geometry) on these two types of connections.