Open Access

Laser beam welding (LBW) in keyhole mode enables high-productivity joining for modern manufacturing processes, yet its industrial deployment is hindered by porosity defects that degrade weld quality and process reliability. This work presents a physics-informed optimization framework designed to systematically mitigate porosity in aluminum LBW by integrating multi-physics modeling, experimental data, and machine-learning-based predictive analytics. The framework couples a series of predictive physics-informed machine learning models (penetration predictor, porosity predictor, and physics estimator) with evolutionary and Bayesian optimization strategies to identify optimal process parameters across a wide operating space of laser power, welding speed, beam diameter, and focal position. High-fidelity thermal–fluid simulations and comprehensive experiments were used to train and validate the predictive models. The framework consistently converged toward parameter sets that achieve target penetration depths while suppressing porosity, revealing the inherent trade-off between weld penetration and defect formation. Beyond accurate prediction and optimization, the approach provides clear interpretability by quantifying key physical factors, such as keyhole stability, weld pool morphology, and local solidification rates, that govern porosity formation. The results demonstrate the potential of physics-informed machine learning as a scalable tool for quality-driven process control and intelligent optimization in advanced manufacturing processes.

Solidification cracking in laser beam welding (LBW) is governed by local thermal conditions at the weld-pool boundary, yet the sub-micron microstructural response across the full mushy zone remains poorly understood. This study employs an automated phase-field (PF) simulation workflow, integrated within the Kadi4Mat research data management platform, to conduct a systematic, FAIR-compliant parametric study of dendritic solidification in the quaternary EN 1.4301 (Fe–Cr–Ni–C) alloy. Thermal conditions—thermal gradient (𝐺), solidification velocity (𝑉𝑠), and grain misorientation angle (𝜃𝑅)—are extracted from a thermocouple-validated ANSYS Fluent weld-pool model and used as inputs to 2D and 3D PACE3D phase-field simulations spanning 𝐺 from 100 to 900 K/mm, 𝑉𝑠 from 5 to 40 mm/s, and 𝜃𝑅 from 0◦ to 45◦. Key findings demonstrate that 𝜃𝑅 is a critical parameter—alongside 𝐺 and 𝑉𝑠—governing the cellular-to-dendritic morphological transition, secondary den drite arm formation, and the topology of inter-dendritic liquid (continuous films versus isolated pockets), each carrying distinct solidification cracking risk pathways. The Kadi4Mat workflow reduces manual pre-processing effort substantially while ensuring data reproducibility and reuse. Quantitative validation against electron probe micro-analysis confirms primary dendrite arm spacing predictions within 12% at the upper weld surface and within 4% at the mid-section.

An easy-to-use methodology for a prior estimation of the overlapping track profile during wire arc directed energy deposition is in ever-demand to assess the dimensional consistency of the fabricated part. A novel analytical framework is proposed here to compute the cross-section of a deposited track, considering the spread of molten filler wire volume on a substrate until solidification. The final track cross-section is obtained as a function of the surface tension force, viscous force, and the contact angle between the liquid filler wire droplet and the substrate. The profile of multiple overlapping tracks is estimated further, considering the remelting of the adjacent tracks. The analytically computed build profiles are compared with the experimentally measured results for different process conditions. The model predicts the width and height of the single-track deposits with average errors of approximately 5% and 14%, respectively, while the multi-track widths are estimated with an error ranging between 2% to 9%. The proposed computational framework serves as a reliable mechanistic model for an effective design of the wire arc directed energy deposition process with improved part quality.

Laser beam welding (LBW) of metallic components is a knowledge‑intensive manufacturing process whose quality depends on the complex multi‑physics. However, its engineering application is often hindered by the occurrence of porosity defects. Achieving a thorough understanding and reliable prediction of porosity defects remains difficult because it demands robust representation and reasoning over nonlinear and hard‑to‑observe physical information. In this study, we propose an integrated multimodal physics-informed machine learning (PIML) framework with the help of multi-physical modelling and experimental data to predict the porosity defects in laser beam welding of aluminum alloys. The whole framework contains a multimodal PIML model for predicting the porosity ratio and an ML-based estimator for relevant physical information. By utilizing the scalar welding parameters and high-dimensional physical information (probability of keyhole collapses, cumulative existing time of collapses, and molten pool geometry) as inputs, the multimodal PIML model shows great superiority in predicting the porosity ratio, with a reduction of the mean square error by 45%, compared with the ML model trained only with welding parameters. The ML-based estimator constructed with an encoder‐decoder architecture can accurately reproduce the critical physical information within a timeframe of seconds. By integrating these two ML models, the proposed framework advances engineering informatics by offering a scalable, physics-knowledge‑centric solution for fast and accurate porosity prediction in LBW manufacturing.

This paper showcases how a holistic approach to digitalisation enables data-driven welding applications, exemplarily for a gas metal arc welding (GMAW) laboratory. The workflow integrates advanced process monitoring, synchronised multi-sensor data acquisition and tools for data analytics. A welding domain-specific data exchange format weldx is presented that unifies and aggregates the data sets acquired during process monitoring with final component quality metrics, supporting reuse, traceability, and reproducibility. Two case studies illustrate the approach. First, GMAW parameters are adaptively adjusted according to local seam geometry to compensate joint-preparation deviations from nominal values typical for large-scale steel fabrication. Second, the seamless data aggregation along the welding production chain enables an automatic life-cycle assessment (LCA), quantifying the environmental impacts of additive manufacturing with DED Arc/M and attributing the dominant contributors to the carbon footprint. Collectively, the results indicate that a fully integrated experimental set-up together with standardised data structures and scalable analytics can couple monitoring, control, and sustainability, thereby realising the potential of digitalisation for high-quality and environmentally informed welding production.