TY - CONF A1 - Meng, Xiangmeng A1 - Bachmann, Marcel A1 - Kising, Pascal A1 - Yang, Fan A1 - Rethmeier, Michael T1 - Prediction of weld pool and keyhole geometries in high-power laser beam welding through a physics-informed generative artificial intelligence approach N2 - The weld pool and keyhole geometries are critical characteristics in evaluating the stability of the high-power laser beam welding (LBW) process and determining the resultant weld quality. However, obtaining these data through experimental or numerical methods remains challenging due to the difficulties in experimental measurements and the high computational demands of numerical modelling. This paper presents a physics-informed generative approach for predicting weld pool and keyhole geometries in the LBW process. With the help of a well experimentally validated numerical model considering the underlying physics in the LBW, the geometries of the weld pool and keyhole under various welding conditions are calculated, serving as the dataset of the generative model. A Conditional Variational Autoencoder (CVAE) model is employed to generate realistic 2D weld pool and keyhole geometries from the welding parameters. We utilize a β-VAE model with the Evidence Lower Bound (ELBO) loss function and include Kullback-Leibler divergence annealing to better optimize model performance and stability during training. The generated results show a good agreement with the ground truth from the numerical simulation. The proposed approach exhibits the potential of physics-informed generative models for a rapid and accurate prediction of the weld pool geometries across a diverse range of process parameters, offering a computationally efficient alternative to full numerical simulations for process optimization and control in laser beam welding processes. T2 - International Congress of Applications of Lasers & Electro-Optics 2025 CY - Orlando, USA DA - 12.10.2025 KW - Laser beam welding KW - Generative artificial intelligence KW - Machine learning KW - numerical simulation PY - 2025 SP - 1 EP - 10 AN - OPUS4-65075 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Bachmann, Marcel A1 - Putra, Stephen Nugraha A1 - Yang, Fan A1 - Meng, Xiangmeng A1 - Pusbatzkies, Pablo A1 - Rethmeier, Michael T1 - Elucidation of the laser beam energy attenuation by the vapor plume formation during high-power laser beam welding N2 - In high-power laser beam welding, a common phenomenon is the formation of a keyhole caused by the rapid evaporation of the material. Under atmospheric pressure, this evaporation generates a vapor plume that interacts with the laser beam, leading to energy attenuation and scattering of the laser radiation along its path. These interactions affect the stability of the process and the overall weld quality. This study investigates the influence of the vapor plume on the weld pool and keyhole dynamics during high-power laser beam welding of AlMg3 aluminum alloy through experimental and numerical approaches. The primary goal is to identify key vapor plume characteristics, particularly its length fluctuations, and to improve the accuracy of the numerical models. To achieve this, an algorithm was developed for the automated measurement of the vapor plume length using high-speed imaging and advanced data processing techniques. The measured plume length is then used to estimate additional vapor heating and laser energy attenuation using the Beer–Lambert law. A refined numerical CFD model, incorporating 3D transient heat transfer, fluid flow, and ray tracing, was developed to evaluate the vapor plume’s impact. Results show that already the time-averaged plume length effectively captures its transient influence and aligns well with experimental weld seam geometries. Additionally, energy scattering and absorption caused by the vapor plume led to a wider weld pool at the top surface. The study also shows an increased percentage of keyhole collapses due to the reduced laser power absorption at the keyhole bottom, further highlighting the importance of accurately modeling vapor plume effects. T2 - International Congress of Applications of Lasers & Electro-Optics 2025 CY - Orlando, USA DA - 12.10.2025 KW - Laser beam welding KW - Vapor plume formation KW - Weld pool KW - Keyhole dynamics KW - Numerical modeling PY - 2026 DO - https://doi.org/10.2351/7.0001863 SN - 1938-1387 IS - 38 SP - 012001-1 EP - 012001-9 PB - Laser Institute of America AN - OPUS4-64949 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Bevilacqua, Tommaso A1 - Gumenyuk, Andrey A1 - Habibi, Niloufar A1 - Hartwig, Philipp A1 - Klawonn, Axel A1 - Lanser, Martin A1 - Rethmeier, Michael A1 - Scheunemann, Lisa A1 - Schröeder, Jöerg T1 - Large-scale thermo-mechanical simulation of laser beam welding using high-performance computing: A qualitative reproduction of experimental results N2 - Laser beam welding (LBW) is a non-contact joining technique that has gained significant importance in modern industrial manufacturing. One potential problem, however, is the formation of solidification cracks, which particularly affects alloys with a pronounced melting range. The aim of the present work is the development of computational methods and software tools to numerically simulate LBW. In order to obtain a sufficiently accurate solution, a large number of finite elements has to be used. Therefore, a highly parallel scalable solver framework, based on the software library PETSc, was used to solve this computationally challenging problem on a high-performance computing architecture. Finally, the experimental results and the numerical simulations are compared. They are found to be in good qualitative agreement, which confirms the validity of the numerical simulations and allows for a better interpretation of the experimentally observed strain distribution. KW - Laser beam welding KW - Termo-mechanical processes KW - Solidification cracking KW - High-performance computing KW - Domain decomposition methods PY - 2025 DO - https://doi.org/10.1016/j.rineng.2025.108827 SN - 2590-1230 SP - 1 EP - 33 PB - Elsevier B.V. AN - OPUS4-65290 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Yang, Chunliang A1 - Yang, Fan A1 - Meng, Xiangmeng A1 - Putra, Stephen Nugraha A1 - Bachmann, Marcel A1 - Rethmeier, Michael T1 - Phase-field simulation of the dendrite fragmentation by electromagnetic stirring in AA5754 aluminum alloy laser beam welding N2 - A phase-field model including magnetic field induced dendrite fragmentation was established and applied to the cases with different initial crystal nuclear positions for AA5754 aluminum alloy electromagnetic laser beam welding. Compare the calculated results that include dendrite fragmentation caused by the thermal electromagnetic Lorentz force with the results that consider only the thermal electromagnetic Lorentz force, without fragmentation, at the characteristic time instants. Both in the early and late stages, the small fragmentation at the dendrite tip promotes the number of higher-order branches and their growth, especially in the direction perpendicular to the solidification. The later stage fragmentation has the possibility of breaking one grain into several, which verifies the possibility of grain refinement caused by dendrite fragmentation. The fracture surface caused by fragmentation also makes more solid-liquid interfaces and their growth. In addition, the cases with different initial nuclear positions were compared. The grain growth in the low-temperature zone can be inhibited by the equiaxed grains' fragmentation at the high-temperature area (179.8 μm² and 14.7 % start at the center, 115.4 μm² and 9.4 % start at the high-temperature corner, 134.3 μm² and 10.9 % start at the low-temperature corner), which is another kind of grain refinement by the dendrite fragmentation. This kind of inhibition effect on grain growth in the low-temperature region will be enhanced with the increasing time interval between the two crystal nuclei’ appearance (179.8 μm² and 14.7 % when virtual grains appear at t = 4.3803 s and t = 4.3803 s, 134.3 μm² and 10.9 % at t = 4.0977 s and t = 3.9564 s, and 115.4 μm² and 9.4 % at t = 3.8151 s and t = 3.5325 s). KW - Laser beam welding KW - Electromagnetic KW - Aluminum alloys KW - Phase field method KW - Equiaxed grain KW - Dendrite fragmentation PY - 2026 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-656610 DO - https://doi.org/10.1016/j.cwe.2025.100014 SN - 3117-4159 VL - 35 IS - 1 SP - 1 EP - 12 PB - Elsevier B.V. AN - OPUS4-65661 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Meng, Xiangmeng A1 - Bachmann, Marcel A1 - Kising, Pascal A1 - Yang, Fan A1 - Rethmeier, Michael T1 - Prediction of weld pool and keyhole geometries in high-power laser beam welding through a physics-informed generative artificial intelligence approach N2 - The weld pool and keyhole geometries are critical characteristics in evaluating the stability of the high-power laser beam welding (LBW) process and determining the resultant weld quality. However, obtaining these data through experimental or numerical methods remains challenging due to the difficulties in experimental measurements and the high computational demands of numerical modelling. This paper presents a physics-informed generative approach for predicting weld pool and keyhole geometries in the LBW process. With the help of a well experimentally validated numerical model considering the underlying physics in the LBW, the geometries of the weld pool and keyhole under various welding conditions are calculated, serving as the dataset of the generative model. A conditional variational autoencoder model is employed to generate realistic 2D weld pool and keyhole geometries from the welding parameters. We utilize a β-variational autoencoder model with the evidence lower bound loss function and include Kullback–Leibler divergence annealing to better optimize model performance and stability during training. The generated results show a good agreement with the ground truth from the numerical simulation. The proposed approach exhibits the potential of physics-informed generative models for a rapid and accurate prediction of the weld pool geometries across a diverse range of process parameters, offering a computationally efficient alternative to full numerical simulations for process optimization and control in laser beam welding processes. KW - Laser beam welding KW - Generative artificial intelligence KW - Machine learning KW - Numerical simulation KW - Weld pool KW - Keyhole dynamics PY - 2026 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-656621 DO - https://doi.org/10.2351/7.0001862 SN - 1042-346X VL - 38 IS - 1 SP - 1 EP - 8 PB - Laser Institute of America AN - OPUS4-65662 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Meng, Xiangmeng A1 - Bachmann, Marcel A1 - Yang, Fan A1 - Rethmeier, Michael T1 - Porosity prediction in laser beam welding with a multimodal physics-informed machine learning framework N2 - 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. KW - Laser beam welding KW - Porosity defect KW - Physics-informed machine learning KW - Multimodal model KW - Simulation PY - 2026 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-657589 DO - https://doi.org/10.1016/j.aei.2026.104611 SN - 1474-0346 VL - 74 SP - 1 EP - 12 PB - Elsevier Ltd. AN - OPUS4-65758 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER -