A welded connection consists of three main material zones, the base material (BM), the heat-affect zone (HAZ), and the weld metal (WM). The strength of HAZ depends on the BM grade and manufacturing process, electrode grade, and welding parameters. Under certain conditions, HAZ has the lowest material strength, especially for high-strength steel. Therefore, a semi-empirical methodology is proposed to establish a constitutive model of HAZ necessary for predicting the fracture position of welded connections. This methodology is based on an engineering approach to consider HAZ as an isotropic and homogeneous material, with no consideration of different volumetric fractions of microstructures within a HAZ. The equivalent material properties of HAZ in butt-welded hollow section connections were investigated experimentally and numerically. Hardness tests and microstructure investigations were conducted to determine the boundaries of material variations and the width of HAZ. The stress–strain relationship of HAZ was established and calibrated based on tensile coupon tests and finite element analyses. Using the calibrated HAZ stress–strain relationship, the effect of transverse constraint imposed by the adjacent and stronger material (BM and WM) on HAZ was evaluated in the welded connections. Finally, the new methodology of a semi-empirical constitutive model based on the Swift model was used to propose equivalent characteristics of HAZ as a function of the mechanical properties of BM for a specific welding procedure considered in the project.

This paper investigates the ultimate negative bending capacity of composite plate girders, whose ultimate hogging flexural state is governed by the local buckling of the bottom flange of steel girders. The Continuous Strength Method, a deformation-based design approach, which employs a continuous relationship between cross-sectional slenderness and inelastic local buckling deformation capacity, is adopted to accurately predict the ultimate negative bending capacity of composite plate girders. A large number of composite cross-sections, which could represent the conventional cross-section of composite plate girder bridges with medium spans, are designed along with various combinations of steel girder strengths and rebar strengths. For each composite cross-section, its ultimate negative bending capacity based on CSM, ultimate elastic bending capacity, and rigid plastic bending capacity are calculated and compared mutually. Both linear and quadratic interpolation equations are proposed for simplifying the negative bending capacity prediction of the composite section. The analysis results indicate that the linear interpolation bending capacity predicting equation can produce relatively conservative results for most Class 2 and Class 3 cross-sections but the equation is quite concise for application. The quadratic interpolation bending capacity predicting equation can offer more accurate results, especially for cross-sections of which the bottom flange governs the ultimate elastic flexural state rather than the rebar in concrete slabs. The rebar in concrete slabs, the top flange and bottom flange of steel girders are supposed to enter into yield state simultaneously for improving the elastic negative bending efficiency of the composite section. Meanwhile, the proposed quadratic interpolation equation could also provide an accurate negative bending capacity prediction.