Effects of connections on structural behaviour in fire

Abstract

The behaviour of connections in fire has become of particular interest to the structural engineering community following the possible link of connection failure to the collapse of the World Trade Centre building 7 and the failures and huge distortion of some connections after the Cardington full scale tests. In order to widen the understanding of the complex behaviour of connections in fire this thesis discusses a number of specific issues relating to connections in fire and their influence on structural response. The first part of this work presents a finite element model for predicting connection temperature profiles. A parametric study is then carried out to investigate which factors have the greatest influence on temperature prediction. This method is compared to the currently available methods for connection temperature prediction presented in the Eurocodes: using a percentage of the beam mid-span lower flange temperature to estimate the temperature across the connection and a lumped capacitance method to calculate average joint temperature based on the mass of material and surface area. In each case, along with the predicted temperatures, the influence on connection material strength is also presented. The three methods have varying levels of accuracy. The finite element model provides detailed and accurate results due to the thorough consideration given to the input parameters. The percentages method gives reasonable estimates in the heating phase but is less accurate in cooling and the lumped capacitance method is only suitable for crude estimations. The remainder of the thesis is concerned with how a number of phenomena affect the overall structural behaviour of buildings: the inclusion of detailed connection models within larger, less complex, finite element models; the effects of connection rotational capacity and the composite beam-slab shear connection. A finite element model for isolated joints is presented in detail for a number of heating regimes and connection types. The influence of the bolt shear and tensile properties is considered in detail and the need for further testing on bolts at high temperatures is discussed. The model has the capacity to predict a number of failure modes and also shows a good comparison between experimental and theoretical deflected shapes. This connection model is then inserted into a large model. It is shown that whilst the inclusion of the shell connection has a small influence on the residual deflections of a structure after cooling when compare to a model where connections are simple and fixed, the difference between heating and not heating the connection does effect structural deflections. Following on from the previous full scale model, simple connections are then exclusively included where the connection rotational capacity is varied. Results show that there is not a large effect on the structural deflections or beam axial and shear forces when rotational behaviour is changed. However column bending moments are hugely increased during heating both in the fire compartment and away from it and fixed connections result in larger bending moment that pinned ones. Finally, the shear interaction between the slab and beams is investigated. The detailed development of both an ambient temperature and then an elevated temperature model of a beam-slab system including explicit shear studs are presented. A study is then carried out looking at the effects on deflections and beam forces when the strength and ductility of the studs are altered. It is found that more ductile studs with a high shear capacity are beneficial for reducing forces in beams and limiting their deflections. Finally the shear studs are included in the larger model used in previous chapters where results are similar to those seen in the beam-slab model, but are less pronounced.