Thickness control of cold-rolled aluminum strips is a complex process influenced by several coupled parameters. The acceleration stage refers to the transient period after strip threading, during which the mill speed gradually increases from the initial low-speed rolling condition to the steady rolling speed. During this stage, process variables such as rolling force, strip tension, roll gap, and main speed fluctuate significantly, weakening the control capability of the thickness control system. In this paper, a new roll gap optimization strategy was proposed, which integrates the TCN-LSTM-AM (temporal convolutional network-long short-term memory-attention mechanism) thickness prediction model and the JITL-XGBoost (just-in-time learning-extreme gradient boosting) optimization model. It enhances the thickness-control capability of the system during acceleration. The TCN-LSTM-AM model extends the hierarchical information sensing capabilities. It effectively captures time dependencies and enhances attention to important hidden information with temporal correlations. A JITL-XGBoost model was proposed to regress the roll gap during the acceleration based on historical superior data. Combined with the predicted future strip exit thickness, the roll gap optimization strategy was proposed. Experiments show that the TCN-LSTM-AM model has a mean absolute error of 0.259 m , significantly improving the thickness prediction accuracy. The JITL-XGBoost model has a regression mean absolute error of 3.7 m , which accurately estimates the optimal roll gap value under the current process conditions. The effectiveness of the roll gap optimization strategy has been confirmed in practical applications.
Rolling force prediction is crucial for process automation and quality enhancement in thick plate rolling. Existing velocity field models, however, fail to accurately characterize progressive metal flow and cannot reliably predict rolling forces. This study develops a kinematically admissible tangential velocity field derived from metal flow behaviour within the deformation zone. The geometric midline (GM) yield criterion is then incorporated to analytically derive the rolling energy. A comprehensive rolling force prediction model is established by formulating the power components of internal deformation, friction, and shear. The model is validated against experimental data and Sims’ classical model. Parametric analysis reveals that a higher friction coefficient shifts the neutral point toward the entry side. While both the reduction rate and the thickness-to-radius rate significantly influence the rolling force, the latter exerts a more pronounced effect. These findings demonstrate that integrating the tangential velocity field with the GM yield criterion substantially improves prediction accuracy, providing valuable theoretical insights for optimizing the rolling process.
Both conventional aging and pulsed magnetic field aging treatments were conducted on an extruded Mg-8Gd-4Y-1Nd-0.5Zr alloy by a self-developed magnetic field aging treatment device. The as-extruded alloy exhibits an ultimate tensile strength of 293.4 MPa, a yield strength of 229.8 MPa, and an elongation of 4.1
In view of the increasing urgent necessity to curtail the environmental footprint of cement production, there has been a rapid development in hybrid cement-based materials, which incorporate alternative binders, supplementary cementitious materials (SCMs), or eco-efficient admixtures. This research work discusses an iterative artificial intelligence assisted data-driven design framework for cement-based hybrid materials targeting enhanced mechanical performance and environmental sustainability. The proposed framework integrates five novel artificial intelligence (AI) approaches to optimizing hybrid cement composition, durability, and lifecycle performance. A self-supervised multi-modal feature selector (SSMM-FS) identifies critical input parameters from chemical, structural, or environmental datasets, thereby simplifying the model by 50
Bilateral rolling process (BRP) is widely used to correct machining distortions in aluminum alloy aerospace structural components. However, it alters the original microplastic deformation and residual stress, affecting fatigue performance. In this study, the finite element method (FEM) is employed to create a simulation model for rolling correction–fatigue analysis, and the effects of rolling parameters and positions on fatigue life are examined. First, a continuous simulation model was developed and experimentally validated. Afterward, the impact of the rolling parameters and positions under periodic external loads was analyzed using a three-compartment frame as the research object. Finally, the principle of stress superposition was applied to reveal the coupled impact of rolling-induced residual stress and operational stress on the fatigue life of workpieces. The results show that increased rolling depth decreases fatigue life, and stress concentration caused by gaps in rolling positions significantly reduces fatigue life. These findings confirm that the rolling process parameters and position significantly affect the fatigue life of aeronautical structural components, providing a theoretical basis for optimizing the BRP to enhance structural reliability of aeronautical structural components under the studied uniaxial cyclic loading.
The growing requirement for lightweight yet high-performance materials has resulted in a major progress in aluminium (Al) alloys, especially in optimizing their mechanical properties. Conventional predictive models struggle to capture the intricate relationships among alloy composition, processing conditions, and mechanical properties. To address this challenge, this research proposes a new hybrid model combining a Bidirectional Gated Recurrent Unit (BiGRU) and a Light Gradient Boosting Machine (LightGBM), optimized using AdamW-based Bayesian Optimization (AdamW-BO). The model is trained and tested on experimental data sets, accurately predicting key mechanical properties such as yield strength (YS), tensile strength (TS), and elongation. The proposed BiGRU-LightGBM model with and without AdamW-BO achieves superior performance, with an R2 value of 0.979 and 0.96, respectively and less errors, demonstrating its efficacy in accurately forecasting mechanical characteristics. The findings suggest that the hybrid approach can serve as a robust predictive tool for material design and quality control in Al alloy manufacturing.
In response to the growing demands for both sustainability and precision in metal forming, this study investigates the potential of an adaptive die system for cold forging processes. The system allows of control of die preload during the main forming and ejection phases, thus offering two degrees of freedom to influence product properties. Through a combination of experimental and numerical investigations, the interdependence between the final part diameter, axial residual stresses, and ejection forces is systematically analyzed. It is shown that increasing the preload during forming reduces the final diameter. Conversely, preload applied during ejection has a direct influence on the resulting ejection force and surface stresses. This decoupling capability enables targeted tuning of individual product properties. To experimentally represent the variability of material batches, three different steel grades were selected, spanning a broad range of flow stresses. The resulting process maps reveal how fluctuations in material properties affect forming outcomes, and how the adaptive die system can be used to compensate these effects. The experimental trends were confirmed by finite element simulations, which support the physical interpretation of preload-related elastic and plastic interactions within the tooling system. The study shows that adjusting the preload intelligently enables dimensional corrections and residual stress or ejection force optimization. The primary focus is on understanding and modeling the process-property relationships. The results lay the foundation for potential control strategies, such as inline or part-to-part adaptation. These strategies can be integrated into future forming lines for increased robustness and flexibility.
To address lateral-edge wrinkling in 0.1 mm-thick 316L stainless steel metallic bipolar plates during roller embossing, this study investigates the wrinkling mechanism and optimizes key forming parameters. A finite element wrinkling model was established and combined with energy-based analysis and shell bending theory to analyze wrinkle evolution and instability behavior during roller embossing. Experimental validation was further conducted under different mold and constraint conditions. The results indicate that wrinkling is mainly induced by compressive stress instability on both sides of the transverse deformation zone, while shear stress contributes to the inclined wrinkle morphology. Multi-channel forming coupling intensifies residual compressive stress concentration and promotes nonlinear wrinkle amplification. Parametric analysis shows that increasing the blank-holder diameter reduces wrinkling, whereas increasing the mold fillet radius aggravates wrinkling. In contrast, the influence of mold side clearance exhibits a non-monotonic trend. Based on response surface methodology, multi-objective optimization of the process parameters was performed. Statistical analysis shows that the blank-holder diameter is the most significant factor affecting wrinkling under the investigated conditions. The optimized parameter combination was obtained as follows: mold fillet radius of 0.1 mm, mold side clearance of 0.27 mm, and blank-holder diameter of 49.90 mm. Compared with the parameter set showing the lowest wrinkling level in the experimental design, the optimized condition reduced the wrinkle amplitude and fluctuation index by 6.12
Compound strip vibration cast-rolling (CVCR) has demonstrated excellent effectiveness in regulating interfacial bonding strength through mechanical vibration. However, as a highly coupled thermal-mechanical-flow process, the influence of vibration on the flow and solidification behavior of the molten pool remains insufficiently understood. To address this gap, thermal-fluid coupled (TFC) numerical simulation of the process was combined with experimental characterization of the solidification microstructure of the cast-rolled strips. Results show that vibration induces periodic evolution of the flow field, generating a stirring effect that refines the solidification microstructure and promotes dendrite equiaxation. Quantitative simulation analysis further reveals that increasing frequency and amplitude produce comparable stirring effects, whereas amplitude exerts a stronger influence on the solid-liquid interface (SLI) position, which is more likely to compromise process stability. Accordingly, a process control strategy of restraining amplitude while increasing frequency is proposed. This study provides theoretical guidance for the process design of CVCR and contributes to advancing its practical application.
In alloys susceptible to Strain-Induced Martensitic Transformation (SIMT) such as metastable austenitic steels, the extensive shearing imposed during Severe Plastic Deformation (SPD) processes accelerates transformation kinetics and enhances material strength. However, compressive mean stress inherent to such processes acts detrimentally to the transformation. The present study investigated the potential of Low-Speed High-Die-Angle (LSHDA) wire drawing as an SPD process for manufacturing ultra-high strength 304 L steel wires by leveraging a controlled state of positive mean stress using atypically high die semi-angles (20 ^∘ to 40 ^∘ ). The tribo-plasticity analysis of this process was conducted using slab method combined with a physically-based, regime-switching Thermal Mixed Lubrication (TML) model capable of predicting the onset of local lubrication breakdown under extremely high die pressures. Furthermore, a novel constitutive-geometric redundancy factor was proposed to accurately capture the coupled effects of highly localized deformation geometry and material strain hardening on the extent of redundant shearing. The theoretical framework was augmented with a Particle Swarm Optimization (PSO) scheme for inverse calibration of constitutive parameters of the proposed friction model and the redundancy coefficient against experimental measurements. The framework demonstrated high fidelity in deriving governing parameters of the frictional model, accurately predicting the onset of surface scoring due to high die pressure. Moreover, the calibrated redundancy factor provided new insights into the extreme inhomogeneous deformation caused by unconventionally high die angles. Experiments confirmed that maintaining low drawing speeds and efficient lubrication entirely prevented central bursting and surface defects, enabling the successful manufacture of defect-free, high-strength products via LSHDA wire drawing.
To address the problems of high forming torque, severe tap wear, and tap fracture during cold extrusion of large-diameter internal threads in high-strength materials, this study proposes a novel internal thread cold extrusion method using specially shaped preformed holes. An M20 × 2.5 internal thread made of Ti-6Al-4V titanium alloy was selected as a representative case. Considering material plasticity, frictional behavior, and strain hardening, a mathematical model of extrusion torque was established to clarify the torque generation mechanism and its dominant influencing factors. Based on this model, several preformed-hole geometries were designed, and finite element simulations were performed to investigate the effects of preformed-hole geometry on extrusion torque, forming temperature, axial load-bearing capacity, and tap wear. The results show that specially shaped preformed holes can effectively modify material flow behavior and the tap–workpiece contact state, thereby reducing extrusion torque and tap wear. By comprehensively considering forming difficulty, thread mechanical performance, tap wear, and manufacturing feasibility, the keyway-shaped preformed hole was identified as the optimal design. Compared with the conventional circular preformed hole, the keyway-shaped preformed hole reduced the extrusion torque, forming temperature, and tap wear by 14.79
Many machining parameters are directly influenced by tool wear and tool geometry evolution. In conventional tool condition assessment, the wear criterion V_b is commonly employed. However, this criterion merely reflects the flank wear width on the main flank face. This study aims to propose a novel methodology based on the integration of microscopic imaging and CAD technology. This approach enables the evaluation of the total flank wear area over the entire flank face, including the tool nose region. Notably, the tool nose accounts for approximately 30
The formability of tufted composite preforms is strongly dependent on the in-plane shear, out-of-plane bending and interlayer friction behaviors. However, owing to a lack of systematic studies concerning the influence of various tufting parameters on these forming behaviors, effective strategies for optimizing tufting parameters for double-curved forming process remain unavailable. In this paper, orthogonal experiments with range analysis and ANOVA are employed to investigate the relationships between the multiple tufting parameters and the corresponding forming behaviors. The results indicate that in-plane angle of tufting yarn dominates in-plane shear stiffness of preforms (contribution ratio is 99.69
Hot isostatic pressing (HIP) is an effective post-processing technique for eliminating internal porosity in metal injection molded (MIM) components; however, the optimization of process parameters still largely relies on empirical approaches, which limits process reliability and efficiency. In this study, a mechanism-oriented numerical framework is developed to investigate stress-assisted pore closure behavior in FD-0205 steel during HIP. Realistic pore morphologies extracted from metallographic images are reconstructed using image-based processing to establish representative two-dimensional microstructural models. A fully coupled transient finite element model incorporating temperature-dependent elastoplastic deformation and high-temperature creep is then implemented to simulate the evolution of pore closure throughout the complete HIP thermal cycle. The results show that densification is governed by the coupled interaction between plastic collapse and creep deformation, which is strongly dependent on pore size, morphology, and local stress redistribution. Larger pores tend to undergo earlier stress-driven collapse due to higher stress concentration, whereas smaller pores exhibit more gradual closure dominated by time-dependent creep deformation. The evolution of equivalent stress, plastic strain, and creep strain exhibits a non-monotonic behavior during heating, holding, and cooling stages, reflecting transitions in the dominant deformation mechanisms. Experimental HIP results validate the simulation predictions, achieving a maximum relative density of 97.82
Based on the analysis of the interfacial bonding state of industrially hot-rolled 1060 Al/AZ31 Mg laminated composites, isothermal hot-compression experiments were designed to simulate the rolling bonding process of Al/Mg and to investigate the interfacial bonding criteria for dissimilar metals. A thermo–mechanical coupled finite element method was employed to analyze the upsetting hot deformation behavior of Al/Mg composites under different processing parameters. By integrating the experimental results with numerical simulations, a novel interfacial bonding criterion for Al/Mg composites based on the coupled response of stress and strain was proposed. The results indicate that increasing temperature and deformation promotes interfacial bonding and improves overall deformation uniformity. Complete interfacial bonding was achieved at 250 °C/60 C_ε , C_σ ) based on strain and stress fields were established and validated through hot-compression experiments, providing a new approach for studying interfacial bonding criteria in dissimilar layered metals.
This study develops a reliability-aware machine-learning framework for compressive-strength prediction and inverse mix design of fly ash–GGBFS geopolymer concrete using a structured database of 672 specimens. Eight raw mix and curing variables were used after removing constant dosage-related columns. Thirteen regression models were first screened using 10-fold cross-validation, and the top models were further tuned using Bayesian optimisation. To move beyond point prediction, split conformal prediction, K-fold residual conformal prediction, and jackknife-after-bootstrap intervals were evaluated at 95
Self-piercing riveting (SPR) is increasingly employed for joining lightweight hybrid structures, particularly when traditional fusion welding is unsuitable for materials such as fibre-reinforced composites. However, the rivet-and-die design in SPR remains challenging due to the complex deformation and failure mechanisms involved in composite-metal joining. This study develops a hybrid, data-driven SPR rivet-and-die design framework that integrates machine learning (ML) with numerical simulations. The framework is demonstrated using the SPR of Glass Fibre-Reinforced Polymer (GFRP) composite sheets and Drawing Quality (DQ) steel sheets, illustrating their potential as dissimilar materials for automotive applications. SPR experiments are performed to validate numerical results and establish a reliable numerical simulation approach. The validated simulation setup is then used to perform multiple simulations with variations in rivet length, diameter, hardness, die diameter, die depth, pip height, friction coefficient, and sheet thickness, generating a comprehensive dataset. Additionally, parameter variations are generated using the Conditional Tabular Generative Adversarial Network (CTGAN) to produce statistically balanced synthetic data for the simulation input. The framework leverages simulation data to train predictive ML models and is validated by using experimental data that determine optimal die dimensions for various GFRP/DQ steel thickness combinations. The developed ML framework achieved strong predictive capability, with FE–ML correlation coefficients of r = 0.90, 0.95, and 0.89 for ID, BRT, and CD, respectively. The optimized design predicted by the Random Forest–L-BFGS-B framework showed close agreement with FE validation results, with deviations of only 1.1
This study experimentally investigated the plastic behavior of AA5083-O sheets under uniaxial tension and compression, tension–compression–tension, and loading–unloading tests. The flow stress under uniaxial compression was higher than that under uniaxial tension, indicating tension–compression asymmetry. In the tension–compression–tension test, the reversal of the loading direction decreased the flow stress, indicating a Bauschinger effect. After tensile prestraining, the unloading stress–strain curves exhibited a slight nonlinearity, and their slopes decreased below the Young’s modulus. A crystal plasticity model incorporating the non-Schmid effect was developed based on our previous model to describe the Bauschinger effect in aluminum alloy sheets (Yoshida K. 2024 International Journal of Solids and Structures, 291:112697). In addition to the resolved shear stress, plastic slip was influenced by the deviatoric stress component normal to the slip plane. The proposed model successfully reproduced the tension–compression asymmetry of flow stress without introducing pressure sensitivity. The reduction in the unloading slope was modeled by introducing a strain-dependent Young’s modulus. The model reasonably reproduced the tension–compression–tension response and approximately represented the overall unloading response. Finally, the model was applied to the springback simulation of L-bending to examine the effects of the crystal plasticity model, particularly its ability to describe the tension–compression asymmetry and unloading behavior. The simulation results showed that improving the accuracy of the crystal plasticity model improved the springback prediction. For the sheet examined in this study, the reduction in the approximated unloading slope had the largest influence among the modeling factors considered.
Recent need for enhanced mechanical properties in cementitious composites poses a demand for high performance along with functional infrastructure, particularly hybrid concrete reinforced with carbon nanotubes (CNTs) and nano-SiO2 particles. Such demands require improvements to be made in mechanical properties under modern requirements. This work develops a comprehensive multiscale hybrid model that is data-driven by incorporating stochasticity to grasp complex behavior in hybrid concrete materials under a wide array of loading conditions. As such, the model fuses Stochastic Kriging-based Hybrid Multiscale Model (SK-HMM) that encompass random field models of CNT and nano-SiO2 dispersion coupled with finite element analysis, thereby making it Viable for the probabilistic predictions of the effective Young’s modulus and tensile as well as compressive strength accurately. Bayesian neural networks (BNNs) with Monte Carlo Dropout are exploited for uncertainty quantification so that the outputs are probabilistic with uncertainty bounds quantified. When it comes to finding solutions to problems and making judgments, it is an essential and essential component. As a further point of interest, the constitutive models of stress-strain behavior adhere to scientific laws, such as the conservation of momentum and energy in physics-informed neural networks (PINNs). Because of this, it is possible to generate fairly accurate predictions regarding stress and strain even when the behavior is not linear. At the end of the day, the nonlinear response surface for crack growth rate and compressive strength can be modeled using gaussian process regression (GPR) with heteroscedastic noise. This allows for input diversity. It is far simpler to make an educated prediction regarding the tensile strength, fracture toughness, and wear life of a material than it would be otherwise. The purpose of this study is to develop a robust framework for the enhancement of hybrid concrete composites by making careful use of randomization, physical laws, and uncertainty quantification respectively. It is possible that these composites will result in structures that are more durable and better building materials that are more resistant to damages.
The thermally induced free vibration behavior of curved laminated composite and sandwich beams is investigated using a rigorously coupled computational–data-driven framework. Curved geometries introduce membrane–bending coupling and anisotropic interactions that significantly complicate dynamic response prediction, particularly under thermal loading where material degradation alters stiffness characteristics. To address this, a higher-order zigzag theory (HOZT)-based finite element formulation is developed to accurately capture interlaminar kinematics and transverse shear effects without ad hoc corrections. The numerical model is validated against literature benchmarks and ABAQUS simulations, demonstrating high fidelity. A large-scale parametric dataset (10,000 samples) is generated using Sobol sequence sampling across geometric, material, layup, boundary, and thermal variables. A hybrid physics-informed artificial neural network–support vector machine (ANN–SVM) surrogate is then constructed, embedding variational bounds, sensitivity constraints, and uncertainty quantification within the learning process. This is not just curve-fitting, the model is explicitly regularized by mechanics. Results show a consistent reduction in natural frequencies with increasing temperature due to stiffness degradation, while curvature and ply orientation induce nontrivial modal coupling effects. The proposed hybrid model outperforms standalone approaches, achieving R2 ≈ 0.98 (training) and 0.96 (testing), with unbiased residuals and well-calibrated prediction intervals.
