Laser beam welding offers high productivity and relatively low heat input and is one key enabler for efficient manufacturing of sandwich constructions. However, the process is sensitive to how the laser beam is positioned with regards to the joint, and even a small deviation of the laser beam from the correct joint position (beam offset) can cause severe defects in the produced part. With tee joints, the joint is not visible from top side, therefore traditional seam tracking methods are not applicable since they rely on visual information of the joint. Hence, there is a need for a monitoring system that can give early detection of beam offsets and stop the process to avoid defects and reduce scrap. In this paper, a monitoring system using a spectrometer is suggested and the aim is to find correlations between the spectral emissions from the process and beam offsets. The spectrometer produces high dimensional data and it is not obvious how this is related to the beam offsets. A machine learning approach is therefore suggested to find these correlations. A multi-layer perceptron neural network (MLPNN), support vector machine (SVM), learning vector quantization (LVQ), logistic regression (LR), decision tree (DT) and random forest (RF) were evaluated as classifiers. Feature selection by using random forest and non-dominated sorting genetic algorithm II (NSGAII) was applied before feeding the data to the classifiers and the obtained results of the classifiers are compared subsequently. After testing different offsets, an accuracy of 94% was achieved for real-time detection of the laser beam deviations greater than 0.9 mm from the joint center-line.

Fluid dynamics models for laser material processing with metal fusion in conduction mode generally assume a constant absorptivity. This parameter is known to govern the process. However, it used to be pre-set by extrapolating absorptance measurements made at different conditions or adjusted to reproduce experimental bead shapes. In this study a new approach is developed. It consists in predicting the absorptance as a function of local surface conditions, including the surface temperature. The proposed absorptance model is applied to the metal alloy Ti-6Al-4V. It is found that the absorptance of this alloy changes with surface temperature over a wide range of beam incidence angles. Thermo-fluid simulations with tracking of the free-surface deformation are performed for conduction mode beam welding test cases with a Yb fibre laser and different travel speeds. It is found that the absorptivity coefficient commonly used for this process clearly underestimates the absorptance and the melt pool geometry predicted for the process conditions of this study. The computational results are also compared against experimental results and good quantitative agreement of the melt pool depth, width, length, free surface contour geometry, and the curvature of the end depression left after re-solidification at the laser switch-off location is obtained. The results show that the absorptance field predicted depends on the melt pool development stage, on the spatial location within the beam spot, and on the process conditions.

Metal additive manufacturing (AM) facilitates the production of geometrically complex components, yet its broader industrial use remains limited by the risk of defect formation and uncertainties in their detection, originating from the highly dynamic and high-temperature process environment. To make additive manufacturing more reliable and establish high-quality parts, it is important to understand how these defects form and how their characteristics appear during the process. This review explains the main causes of common defects, such as cracking, porosity, lack of fusion, and inclusions in metal AM processes, including Powder Bed Fusion and Directed Energy Deposition. It also connects main defect formation mechanisms to the optical, thermal, acoustic, and spectroscopic signals that can be measured during the process. Moreover, it is described how commonly used in situ monitoring systems work and how their signals correspond to melt pool dynamics, vapor plume, particle movement, and the solidification process for each kind of defect. An overview is provided of how data from these systems are analyzed, including the extraction of features from images, the evaluation of temperature fields, and the use of time and frequency domain techniques for various signals. By linking the physics of defect formation to measurable process signals, the interpretation of sensor data is enabled, and potential strategies for monitoring specific problems are outlined. Finally, recent developments are examined, including the integration of multiple sensors, advanced feature-representation approaches, and real-time data interpretation coupled with adaptive control. Together, these directions represent promising advances towards more intelligent and reliable monitoring systems for the future of metal AM.