Issue 47

P. Ferro et al., Frattura ed Integrità Strutturale, 47 (2019) 221-230; DOI: 10.3221/IGF-ESIS.47.17 223 Metallographic analysis was carried out by means of optical microscope (OM) and Environmental Scanning Electron Microscope (ESEM). Vickers microhardness profiles were performed across the FZ to support the microstructure interpretation. N UMERICAL M ODEL he welding simulation was carried out by the numerical code Sysweld®. In order to reduce the long computational time required for a multi-pass welding thermo-mechanical transient analysis, a 2D model was adopted under generalized plane strain condition . A half cross-section was modeled according to the scheme shown in Fig. 1. The mesh, shown in Fig. 2 and obtained as a result of the sensitivity analysis, consists of 5436 four nodes elements. In order to take into account the effect of filler metal the geometrical variation of the weld toe induced by TIG-dressing, different element groups are created. By referring to Fig. 2, the black elements belong to the parent metal and are always active during simulation; the red ones define the filler metal and are activated during welding, while the blue elements groups are activated only during TIG-dressing and are used to simulate the smoothing effect induced by the remelting of the material. Figure 2 : Mesh of the model with elements groups The weld toe was modeled with a V-notch angle (2  ) equal to 135° [16]. The toe-radius (  ) is equal to 0 mm in the as- welded condition and 6.7 mm after TIG-dressing treatments. This last value was obtained by averaging the measurements carried out on real samples. It is noted that a high-density mesh is used near the weld toe in order to capture the residual stress singularity induced by welding (the minimum element size was 5x10 -5 mm). The different welding and TIG-dressing operations were sequentially performed according to Fig. 3. Thermo-metallurgical and mechanical properties of both filler and parent metals were taken from Sysweld data-base. Filler and parent metals alloys were chosen according to the chemical compositions reported in Tab. 1. With the aim to include the metallurgical effects, in the present analysis the following microstructural constituents were considered: martensite, bainite, ferrite-pearlite, tempered martensite, tempered bainite and austenite. They were modelled by means of the Leblond-Devaux [17] and Koistinen-Marburger [18] equations according to their diffusional or non- diffusional feature. The simplifying assumptions were made that tempered bainite has the same properties as ferrite and that tempered martensite is similar in properties to bainite. The heat source was modelled using a double ellipsoid power density distribution function given by Goldak et al. [19] (Eqn. 1) that has been used previously in published literature for arc welding simulation [14]. T