1 |
Implementation of thermomechanical laser welding simulation : Predicting displacements of fusing A AISI304 T-JOINTRolseth, Anton, Gustafsson, Anton January 2021 (has links)
Laser welding is an advanced joining technique with the capability to form deep, narrow, and precise welds. Numerical models are used to simulate the process in attempts of predicting distortions and stresses in the material. This is done to reduce physical testing, optimize processes and enable integrated product- and process development. The Virtual Manufacturing Process research group at University of Skövde wishes to increase their knowledge on modeling options of thermomechanical simulations to grant local industries these benefits. A numerical model for the laser welding process was developed in ABAQUS. This was done by examining the macrograph structure of a simple weld and applied to a stainless-steel T-joint welding application. The macrograph data was used to calibrate a mathematical heat source model. User subroutine DFLUX was used to enable movement of the heat source and element activation was used to simulate the fusion of the two parts. A T-joint welding experiment was carried out to measure deflection and the result was compared to numerical simulations. Different combinations of heat source models, coupling type and element activation was compared in relation to predicting the deflection. Computational time and modeling complexity for the techniques was also considered.The results showed that a 3D Gaussian heat source model will imitate the keyhole weld achieved superior to the compared 2D model. The 3D model provides greater flexibility since it enables combinations of any geometrical bodies. It was shown that element activation has a significant contribution on part stiffness and thus resulting distortions. To implement element activation a fully coupled analysis is required. The deflection of the fully coupled 3D simulation with element activation showed a 9% deviance in deflection compared with experiments.
|
Page generated in 0.0199 seconds