Framework for the fracture analysis of structural steels and connections

The paper presents an overview of a recent research project to establish a framework for the prediction of fracture in steel construction, notably in welded and bolted connections of steel frames. The paper first describes a method for obtaining the full-range true stress vs true strain curve, including the post-necking material response, for which a new model is proposed. The model allows to obtain the accurate calculation of the true stress and strain states at large plastic strains up to the point of necking, as required for the precise prediction of fracture. The recently proposed Lode Modified Void Growth fracture initiation Model (LMVGM) is then described, and recommendations are given for selecting specialised coupon tests to determine the free parameters of the model. The coupon selection is guided by achieving widely spaced points on the fracture surface, defined in terms of the stress triaxiality and the Lode angle parameter. Subsequently, tests on miniature tensile coupons are presented with coupons cut from the base metal, weld and heat-affected zone (HAZ) to enable fracture to be predicted in all the three materials. Finite element predictions of fracture initiation and propagation are then compared with tests on welded and bolted connections showing excellent agreement, including fracture propagation traversing across disparate materials at welds, and thus demonstrating the capacity of the presented post-necking stress-strain model and the LMVGM model to accurately predict the initiation and propagation of fracture in steel structures. In summary, the purpose of the paper is to describe a rational and accurate framework for obtaining the full-range true stress-strain curve and fracture initiation properties for steel materials, HAZ and welds, suitable for engineering computations in both research and practice.