PSI - Issue 47

V. Giannella et al. / Procedia Structural Integrity 47 (2023) 892–900 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

897

6

3. Results and Discussion Standing the usage of the “birth -and- death” method, the generation of the weld bead was explicitly modelled during the simulation. This can be seen in Figure 5, in which the model change during the advancing of the welding torch was highlighted. It is worth noting that such variation allowed to, simulate the process accurately. Concerning the temperature fields, these were measured for those nodes located in correspondence of the thermocouples adopted during the experimental tests (Figure 1). In Figures 6-8, a comparison on temperature trends predicted at these nodes with the temperatures recorded experimentally at the same locations was provided. Such comparison of the experimental and numerical temperature levels was required for validating the numerical model, in turn allowing the welding engineers to speculate on further numerical results such as residual stresses and distortions.

Fig. 5. Temperature distributions [°C] during welding at various simulation frames.

From Figures 6-8, a very good agreement was observed among the numerical and the two experimental datasets (one for each plate), both in terms of maximum values at temperature peaks and during the cooling phases. In Figures 6-8 it can be noticed that, before that welding started, temperature was mostly equal to room temperature (~ 25 °C). Then, when half of the first pass was completed ( t ~ 40 s), a first peak of temperature was reached immediately for the thermocouple closest to the weld bead (Figure 5) and after nearly 15 seconds for the farthest one. At the end of the first pass, the temperature started to decrease for the whole joint and reached an overall temperature of nearly 130 °C, close to the weld bead, after the interpass time, just before starting the second pass. Similarly, when half of the second pass was completed, ( t = 250 s), the second peak of temperature was recorded by the closest thermocouple and, after nearly 30 s, by the farthest one. As the torch furtherly proceeded with the welding, temperature at all thermocouples decreased monotonically and at different rates. Finally, after the last step of overall cooling, room temperature was mostly recovered for the whole joint (not shown in Figures 6-8). It was worth noting that a very good agreement was provided between numerical and experimental temperatures for all the measuring positions. Experimental and numerical curves appeared to be overlapped with small differences only at the peak values and for the measuring point closest to the weld bead (Figure 6). This was attributed to some