Wire-arc additive manufacturing (WAAM) has gained popularity for large metallic components because of its cost-effectiveness and high deposition rates. However, this method often leads to coarse microstructures and subpar tribological properties because of substantial heat input during deposition. This study employed laser remelting as post-processing to enhance the surface quality and performance of low-carbon steel produced via gas metal arc welding-based WAAM. It investigated pulsed laser reeling’s impact on microstructural transformations, phase characteristics, and corrosion resistance. Laser remelting refined the microstructure, reducing average grain size from 9.52 ± 6.32 μm to 1.46 ± 1.04 μm. The predominant α-ferrite phase remained, while crystallite size diminished by 8.76
This study solves the critical problem of small-sample weld morphology prediction for laser-plasma arc coaxial hybrid welding of TC4 titanium alloy thin sheets. A lightweight multilayer perceptron (MLP) is developed with laser power, plasma current, welding speed, and assembly gap as inputs to predict weld depth and width. Based on 46 experimental groups, the model integrates Glorot initialization, Adam optimizer, and L1-Dropout dual regularization to enhance training stability, mitigate gradient vanishing, and suppress overfitting. The novelty of this work lies in the first integrated application of this lightweight strategy to coaxial laser-plasma hybrid welding under limited data, combined with quantitative contribution analysis and interpretable visualization of parameter coupling mechanisms, which converts the black-box network into an interpretable engineering tool. The optimal network structure of 4-20-2 with a batch size of 5 and 419 iterations, the model achieves R2 of 0.980 for the training set and 0.903 for the validation set, demonstrating high prediction accuracy and strong generalization. Visualization results clarify that weld depth is dominated by laser power, while weld width is co-controlled by laser power and plasma current. Higher welding speed reduces both depth and width nonlinearly. Within 0–0.2 mm assembly gap, weld width increases slightly while depth decreases due to arc bridging and weakened molten pool stability. The model reduces the need for complex physical modeling and costly trial-and-error experiments, providing a reliable and intelligent framework for parameter optimization and quality prediction in hybrid welding. It provides theoretical support for intelligent aerospace manufacturing of high-performance titanium alloy structures.
A three-dimensional numerical model of the single-arc double-wire gas metal arc welding (SD-GMAW) arc is established based on fluid dynamics and electromagnetic theory. The arc shape, temperature field, velocity field, and pressure field of SD-GMAW under different welding currents are systematically analyzed and compared with those of single-wire GMAW. The results show that the SD-GMAW arc forms a single coupled arc, resembling a Mexican hat shape in the x-z plane and a conical shape in the y-z plane. The maximum arc temperature of SD-GMAW is significantly lower than that of single-wire GMAW at 350 A, but slightly higher than that of single-wire GMAW at 175 A. The maximum plasma velocity of SD-GMAW is lower than that of single-wire GMAW. The arc pressure of SD-GMAW is lower than that of single-wire GMAW. As the welding current increases, the arc shape expands, the overall arc temperature and plasma velocity rise, and the arc pressure on the base metal surface gradually increases. The simulation results are consistent with the experimental findings, and the established model can provide a theoretical basis for in-depth investigation of arc mechanisms and optimization of welding process parameters.
This study experimentally measured shot velocity using an ultra-high-speed camera to provide reliable input data for developing numerical simulations of shot peening for welded joints. Shot peening was performed using straight and curved nozzles with two different distances between the nozzle and the specimen. Shot velocity was measured from afterimages of the trajectories of shot media captured by the camera. Measured shot velocity agrees well with a value generally reported in previous studies, despite the shot velocity increases with increasing the distance in case of a straight nozzle. A numerical simulation model for single shot was developed using an explicit dynamic finite element method and its model accuracy was confirmed through comparison with previous studies. Then, parametric and multiple shot analyses were conducted, revealing that compressive residual stresses introduced by shot peening are primarily governed by the kinetic energy of the shot. After those analyses, numerical simulation incorporating the measured shot velocity was conducted, and the simulated residual stress was compared with measurement results obtained on base materials near weld toes. The shot peening analysis successfully captured the depths associated with the minimum compressive residual stress and the plastic region. Furthermore, based on this analysis model, this study investigated the influence of shot angle, friction, and material hardening rules on the residual stress distribution. The investigation results showed that the shot angle and friction might be key influencing factors for better estimating the amount of compressive residual stress. Thus, the shot peening angle to weld toes should be certainly controlled.
