Abstract Textured work rolls are utilized in skin-pass rolling to achieve the specific surface finish of the strip, affecting the product properties, such as friction coefficient and paintability. Therefore, to achieve effective property control, it is essential to control surface finish while maintaining geometric accuracy. However, both geometric accuracy and surface finish are influenced by the rolling force, necessitating decoupling for independent control. In this work, asymmetric strip tension is considered as an additional actuator, alongside the roll gap actuator, to investigate the potential to control the surface roughness without influencing the thickness reduction, thereby obtaining the desired product properties. For this purpose, a numerical study is conducted using a finite element multi-scale model with different thickness reductions and strip tensions. Results indicate that, compared to the mean roughness of 2.34 μm at the symmetric case with backward and forward tensions of 0.2 kN, the roughness increases by 7.7% with increasing the forward tension to 1.2 kN and decreases by 2.1% with increasing the backward tension to 0.6 kN, at 7% thickness reduction. This variation in roughness is attributed to the effect of strip tension on the relative movement between the work roll and the strip material. Therefore, applying asymmetric strip tension emerges as an effective strategy to ensure both surface finish and geometric accuracy simultaneously in skin-pass rolling.

Abstract This study benchmarks multiple data-driven methodologies for predicting relative density (RD) of 316 L stainless steel fabricated via Powder Bed Fusion–Laser Beam (PBF-LB), as part of the ESAFORM Benchmark 2025 AMDmodel initiative. Two datasets (DS-01 and DS-02), each with 256 specimens from a 4-factor, 4-level design of experiments, were produced on different PBF-LB systems equipped with equivalent in-situ infrared (IR) melt-pool pyrometry. Failed builds (RD = 60%) were retained to allow models to learn from both nominal and catastrophic processing conditions, a scenario rarely addressed in PBF-LB machine learning (ML). Statistical analysis of variance (ANOVA) confirmed that conventional process parameters alone are weak predictors (R² ≈ 0.49). In contrast, sensor-driven supervised ML models using melt-pool thermal descriptors performed substantially better. Recursive feature elimination highlighted the interquartile range and mode of thermal signatures as dominant predictors; an XGBoost model using only these achieved R² = 0.93 on DS-01. Hybrid models combining parameters and IR descriptors performed slightly worse (R² = 0.92), indicating mild redundancy. Cross-system transferability was limited: ML models trained on DS-01 underperformed on DS-02 due to IR input-domain divergence despite RD distributions between both domain sources showing high inter-laboratory consistency. To address this, a physics-informed ML framework (PIML) using symbolic regression (QLattice) embedded dimensionless physical priors. Resulting compact expressions dominated by normalized laser power and volumetric energy density achieved R² = 0.83–0.93 under cross-system validation. Overall, sensor-driven ML models are effective for machine-specific monitoring and layer-wise closed-loop control, whereas PIML provide system-agnostic process parameter-window estimation for design-stage optimization.

Abstract Accurate prediction of the Forming Limit Curve (FLC) is essential for the design of sheet metal stamping processes; however, its experimental determination is costly and limited by data availability. This work investigates the use of Machine Learning techniques to predict the FLC of Dual Phase (DP) steels based on mechanical properties obtained from uniaxial tensile tests. To overcome the scarcity of experimental data, a synthetic database was developed based on statistical consistency and physical constraints, using Kernel Density Estimation, PCA projections, and controlled probabilistic interpolation, followed by the application of physicometallurgical plausibility criteria. The models use physics-based descriptors as input variables, which reflect known metallurgical mechanisms associated with plastic instability, without explicitly incorporating differential equations into the training process. The results show that all models were able to reproduce the characteristic geometry of the FLC, with errors on the order of 10⁻³–10⁻². Among the investigated techniques, Random Forest exhibited the best performance (MAE = 0.0052; MSE = 0.00011; R² = 0.943), followed by XGBoost, while the Neural Network showed greater variability and a tendency toward overfitting. The results demonstrate that the combination of physics-based descriptors, statistically validated synthetic expansion, and ensemble machine learning methods constitutes a robust and efficient strategy for modeling FLCs of DP steels.

Abstract Continuous fiber-reinforced polymers (CFRPs) offer high strength-to-weight ratios, which makes them an attractive choice for applications in transportation, biomedical devices, and sports equipment. Additive manufacturing presents new opportunities for producing CFRPs with improved geometric freedom and digital fabrication flexibility. However, achieving adequate fiber impregnation and strong interfacial bonding remains a major challenge. This paper presents a novel rotating impregnation die, patented by some of the authors, designed to produce fiber-reinforced polymer filaments at a die speed of 15 rad/s. These filaments, characterized by a final diameter of 0.65 mm and a fiber volume fraction of 5.4%, are compatible with fused deposition modelling for 3D printing. The die is engineered to improve polymer–fiber interaction during filament fabrication. Specifically, its rotating geometry induces a swirling flow pattern in the molten polymer, which enhances fiber wetting and promotes partial fiber interlacing. The performance of the system was evaluated through both numerical simulations and experimental tests. In the computational fluid dynamics analysis, an inlet velocity of 5 mm/s was imposed, showing that the rotational motion generates a tangential velocity component that improves fiber-polymer interaction and locally reduces viscosity at the fiber surface, leveraging the shear-thinning behaviour of the polymer. This results in improved impregnation efficiency without affecting the internal pressure of the die. Two filament configurations were produced for comparison: one using the rotating impregnation die PLAGF-B (PolyLactic Acid – Glass Fiber Braided) and one using a static die PLAGF-UB (PolyLactic Acid – Glass Fiber UnBraided). The produced filaments consisted of three glass-fiber bundles impregnated with PLA resin and were subjected to standard tensile testing, after being pulled at a controlled speed of 6 mm/s. The PLAGF-B samples exhibited higher tensile strength (~ 70 MPa vs. ~60 MPa) and elongation at break (~ 0.023 mm/mm vs. ~0.018 mm/mm), attributed to enhanced twisting and compaction induced by the die’s rotation.