Applying statistical models optimize the process of multi-pass narrow-gap laser welding with filler wire

Laser welding of thick plates has been used in automobile and shipping industry owing to its focusing heat input, little welding deformation, and high productivity. Using narrow-gap technique can decrease the filling volumes of wire and increase the welding efficiency. However, its process is more complicated since it introduces filler wire to narrow-gap weld configurations. Usually the welding path, planning mainly depends on man's experiences. Human factors have big influences on the quality of the joint. The aim of this paper is to study the interactions between welding parameters and the geometry of single bead using statistical methods. Then, the founded models were employed to plan the welding path, the number of filling passes, and layers in the multi-pass narrow nap joining. These results will provide supports for automatic multi-pass laser welding. The welding system includes a 10-kW Nd:YAG laser system, a KUKA robot, a Fronius wire feeder, and a 20-mm-thick high-strength ship steel with narrow gap. Laser power, welding speed, and wire feed rate are input variables. The responses include the transversal area of the added metals, bead width, bead height, and the ratio of bead width to height. Verification experiments indicate that all these models can forecast the responses within the factor domain. Transversal section micrographs of the joints show that this method can get welding joint with less defects.