Journal of Advanced Joining Processes (Mar 2020)
Finite element modeling of friction stir welding (FSW) on a complex curved plate
Abstract
For decades, Friction Stir Welding (FSW) has been used to decrease the weight of structures. During this time the application of the curved surfaces has been extensively increasing in numerous applications like FSW, 5 axis milling computer numerical controlled (CNC) machines, and elsewhere. However, for finite element modeling of the abovementioned operations, difficulty arises in defining the tool's perpendicular movement on the curved surface because the basic principle of the finite element software is based on a single point movement for the tool. To explain the difficulty, the tool should follow a pattern and has to have a position perpendicular to the surface at each point, however because of the modeling difficulties the literature simulated only the tool's single point movement. Consequently, the software should be modified for accurately simulating a representation of perpendicular movement. In this paper Altair HyperworksⓇ and ABAQUSⓇ software are employed to simulate the process on a curved plate. Mathematical formulations solve the governing equations of the perpendicular tool movement (movement in X direction, Y direction and the angular movement). Then, VDISP user-defined subroutine and the optimized input parameters (from the previous part) are incorporated in the software for analyzing the FSW thermal behavior on a curved surface. It should be noted that, the letter “V” shows the analysis type that is explicit and “DISP” indicates that this subroutine can be appropriate when complicated displacement boundary conditions should be applied to the model. The results showed that there is a significant increase for heat generation, resulting in the expansion of the shear zone. That leads to a peak temperature of almost 300 °C after 3 s. During the dwelling step (t = 3 s to t = 5 s), the generated heat was stable and the shoulder moved the material up, because the material velocity at the upper surface is higher. In addition, a non-symmetrical temperature distribution is found at the cross section. By the end of the welding (t = 12.8 s), the temperature pattern was asymmetrical, while at the step time of 19.6 s an asymmetrical temperature was observed. Consequently, the outcomes of this paper indicates that the model agrees well with the literature.