Data-driven investigation of pore formation mechanisms in laser welding of Al-Cu

Laser welding of dissimilar materials poses a crucial challenge in understanding the pore formation mechanisms due to its complex physical phenomena. This paper explores the pore formation mechanisms in the laser welding of Al-Cu lap joints, employing a multidisciplinary approach that integrates data science, three-dimensional image analysis, and computational fluid dynamics. Micro X-ray computed tomography imaging revealed three-dimensional morphological and positional data of pores in laser-welded samples, which were subsequently subjected to data clustering to distinguish three distinct pore types, i.e., fusion zone pores, small fusion boundary pores, and large fusion boundary pores. Guided by these data-driven insights, the study delves deeper into the understanding of pore formation mechanisms through computational fluid dynamics simulations, which illustrates the movement of gas bubbles and their intricate interactions with the molten pool, providing insights into the governing mechanisms of different pore types. The influence of process parameters on the pore formation mechanisms was further explored, revealing the effects of process parameters on pore generation in the laser welding of Al-Cu.