Simulation of the welding process of steel tube joints made of S355 and S690

Welding is an important joining method for the work with metals. During the welding process, the material is heated and deforms plastically which leads a change in microstructure in the heat affected zone, to complex residual stress distributions, and distortions. The residual stresses can have a major impact, among other phenomena, on the fatigue resistance of a welded structure. The aim of this work was to develop a model with the finite element software Abaqus for the simulation of a welding process for butt welded tubes and the calculation of the resulting welding residual stress field. The residual stress field depends on several input parameters, like the geometry, the heat input, the welding speed, the material data, and the number of weld passes. The chosen dimensions of the tubes are realistic for tubular steel bridge constructions. Since, no experimental welding procedures were conducted, the welding parameters like the heat input and the welding speed were estimated in order to obtain a realistic temperature distribution in the weld and the heat affected zone. The assignment asked to place a main focus on the influence of the material data and the number of weld passes. Within the framework of this work, tensile tests at three different temperatures were conducted in order to find the deformational behavior of the two different steel grades S355 and S690. The results show generally good agreement with the data given by the European Standard for the structural fire design of steel structures [11], that is noted Eurocode 3 (EC3) in the course of this work. According to the test results, the yield strength at room temperature are underestimated by the Eurocode 3. Based on the experiments and the data given in the Eurocode 3, the influence of the mechanical properties were studied in several Abaqus models. For the description of the temperature dependent material behavior, a bilinear stress-strain curve was implemented. The resulting stress distributions show a similar trend, but the offset values between the stress curves of two models are dependent on the position in circumferential direction and the distance in axial direction. During the literature study, different welding simulation programs were compared. With specific welding simulation programs, mostly two-dimensional models can be realised. A major advantage of these programs is, that phase transformations can easily be considered and therefore a more realistic material model can be implemented provided that the temperature dependant behavior of the different microstructural phases are known. The influence of the number of weld passes was studied by comparing a single- and a multi-pass model with the material data from the Eurocode 3. The total heat input and the welding speed was the same for all the models. The tendency of the resulting stress distributions varied greatly and therefore implies that a multi-pass model cannot be approximated by a single-pass model.