Heat-affected zone in welded cold-formed rectangular hollow section joints

High-strength steel (HSS) has higher strength but lower ductility than mild steel. The cross-section of the structural members may be reduced using HSS instead of mild steel, provided the buckling of elements does not govern the failure. The reduced member size benefits the environment and economy by means of less energy consumption, less carbon dioxide emission, and less labour work during the fabrication and structure construction stages.

The current design rules in prEN 1993-1-8 for welded hollow section joints are developed based on extensive experimental and numerical studies on joints made of mild steel (S235 and S355). A material factor Cf is stipulated to reduce the design resistance of the joint given the lower ductility of HSS than mild steel. In addition, the design yield strength of the material should be lower than 0.8 times the ultimate strength (fu) to calculate the resistance of punching shear failure and tension brace failure. However, these two strength restrictions are proposed based on limited experimental and numerical investigations on welded HSS tubular joints. The mechanical background behind the two restrictions is vague. Applying both Cf and the 0.8fu restriction would eliminate the benefits of using HSS, reducing the competitiveness in the market. Besides, the heat-affected zone (HAZ) often has the lowest strength in a weld region. The strength difference between HAZ and the base material (BM) is more significant for HSS than mild steel, indicating that HAZ plays a more critical role in welded HSS joints. Hence, the HAZ constitutive model should be considered in the numerical study of welded HSS joints in order to predict the load-deformation relationship and failure mode correctly.

This dissertation proposes a systematic approach to include HAZ in the finite element (FE) analysis of welded joints considering ductile failure mode. First, the mechanical and geometrical properties of HAZ were obtained from tensile tests on the milled welded coupon specimen, the low-force Vickers hardness test, and the microstructure observation. The full-field deformation of the milled welded coupon specimen was measured using the digital image correlation (DIC) technique. Using the DIC result, a method is proposed to identify the boundaries of different regions in the milled welded coupon specimen. The identified boundary matches the hardness result well. Based on the identified boundaries, the width of HAZ and the weld metal (WM) are determined, which provides geometric information for the measuring range of the virtual extensometer in DIC and for creating the FE model with different partitions (HAZ, BM, and WM). Due to the transverse constraint imposed by BM and WM, HAZ was under a biaxial or triaxial stress state during the tensile coupon test. The measured stress of HAZ is higher than that under the uniaxial stress state at a given strain. Hence, a method is proposed to correct the measured stress-strain relationship of HAZ. The modified stress-strain relationship is successfully validated against the tensile coupon test regarding the load-deformation relationship and the strain distribution on the specimen surface.

In order to accurately predict the load-deformation relationship and failure mode of welded joints, the Gurson-Tvergaard-Needleman (GTN) damage model is employed to simulate the failure of HAZ and BM. A computational homogenization analysis using representative volume element models was carried out to calibrate the yield-surface-related parameters (q1, q2, and q3). The effect of the hydrostatic pressure, the accumulated initial hardening strain, and the void volume fraction (VVF) on the yield surface were evaluated. An equation is proposed to describe the relationship between VVF and q1 value with a constant q2. The fracture-related parameters (fc and ff) were calibrated against the tensile coupon test. In addition, as the procedures for modifying the constitutive model and calibrating the damage model are rather complicated, a semi-empirical material damage model for HAZ correlating to the mechanical properties of BM is proposed to facilitate the FE analysis of welded joints.

Monotonic tensile tests were conducted on 18 welded cold-formed rectangular hollow section (RHS) X-joints made of S355, S500, and S700 to investigate the validity of Cf and the 0.8fu restriction. The test result shows that a conservative resistance is predicted using the current design rules without applying Cf and the 0.8fu restriction. The calibrated GTN damage model for HAZ and BM was implemented in the fracture simulation of welded X-joints. The FE results agree well with the experimental results concerning the load-deformation relationship and the failure mode. Based on the validated X-joint FE model, the importance of including HAZ in the FE model was revealed by the FE analysis without the HAZ constitutive model. Finally, the semi-empirical material damage model for HAZ was employed to predict the tensile behaviour of all 18 welded X-joints.