Thermal error is one of the key factors affecting the motion accuracy of the linear motor driven feed axis (LMFDA), so it is necessary to model and predict it. The complexity of the underlying physics makes mechanistic models hard to build and solve, and their predictions often deviate from real-world measurements. Empirical models require high-quality data in training, and their training process lacks physical interpretability. To address these challenges, we propose a physics-informed neural network (PINN)-based thermal error modeling method that integrates governing partial differential equations into a deep learning framework. We conduct a mechanistic analysis of LMFDA’s operation, derive the governing temperature-field PDEs under coupled electromagnetic-thermal-flow conditions, and build a PINN model tailored to LMFDA. During the neural network training process, prior physical information such as PDE, initial condition (IC), and boundary condition (BC) are embedded to regularize learning, thereby achieving good physical interpretability. We validate the proposed method on a fully direct-drive three-axis machine tool. Experimental results demonstrate that by combining mechanistic analysis with data-driven learning, the proposed method is well-suited for thermal error problems characterized by limited training data but well-defined physical mechanisms.
This study introduces a multi-point adaptive magnetorheological polishing technique employing an enhanced Halbach array for the ultra-precision finishing of Ti-6Al-4 V turbine blades, addressing the challenges associated with machining complex freeform surfaces. An improved Halbach magnetic configuration was devised to concentrate the magnetic field within the polishing zone, achieving a peak strength of 1.78 T. A mechanism-based material removal model, integrating Preston theory with magnetic-force-induced abrasive interactions, was developed to predict material removal distribution on intricate curved surfaces. A shape-adaptive optimization strategy was proposed to determine the optimal interaction angle, ensuring uniform magnetic pressure and stable polishing conditions. The experimental results indicate significant enhancements in surface quality, reducing the surface roughness from approximately 90–100 nm to 5–9 nm with high uniformity. The magnetic field simulation closely aligned with the experimental measurements. This method offers an efficient and scalable solution for finishing complex freeform components, particularly lightweight titanium alloys, in aerospace applications.
The modern industrial automation needs intelligent coordination of Robotic Distribution Systems (RDS) to reduce downtime and maximize the use of resources. This paper suggests an AI-based predictive scheduling model to combine a hybrid model of Genetic Algorithm-Particle Swarm Optimization (GA-PSO) and ensemble-based downtime predictions. The framework is a hybrid which incorporates predictive intelligence and adaptive scheduling to reach dynamic optimization in robotic task allocation. The down time events are predicted with an impressive accuracy of 98.30
Ball burnishing is an efficient and innovative process for improving surface integrity through controlled plastic deformation. This study proposes the simultaneous modeling and multi-objective optimization of surface quality, surface hardness, and productivity during the ball burnishing of high-strength AA2024-T3 aluminium alloy. The effects of burnishing force, workpiece rotational speed, and feed rate were investigated. Quadratic regression models for surface roughness improvement rate and surface hardness were developed using Response Surface Methodology (RSM), while production time was independently determined using an analytical model. Statistical analysis showed that feed rate and burnishing force predominantly govern roughness improvement, whereas surface hardness is mainly influenced by the applied force and rotational speed. Significant interactions between process parameters were also identified. The developed RSM models were experimentally validated and, together with the analytical production-time model, integrated into a multi-objective optimization framework to identify optimal trade-off solutions. The results showed that surface quality improvement generally evolves in opposition to surface hardness and productivity. Three representative solutions were selected: (i) a productivity- and hardening-oriented solution, achieving a 60.99
Titanium alloys are widely deployed in aerospace, biomedical and marine applications owing to their high strength-to-weight ratio, corrosion resistance and biocompatibility. Recent progress in additive manufacturing (AM) has enabled the production of near-net-shape Ti-based components; however, fusion-based AM processes impose steep thermal gradients and rapid cooling rates that generate coarse columnar prior-β grains, strong build-direction textures and pronounced mechanical anisotropy, restricting industrial uptake. Ceramic-particle inoculation has emerged as a compelling route to overcome these limitations by providing heterogeneous nucleation sites during solidification. This review consolidates and critically appraises the peer-reviewed literature on ceramic inoculation of Ti alloys processed by directed energy deposition (DED-Arc, DED-LB) and powder bed fusion (PBF-LB, PBF-EB). The inoculant families examined include borides (TiB, LaB₆, boron nitride nanotubes), nitrides (TiN, ZrN), carbides (B₄C, SiC) and oxides (Y₂O₃, TiO₂), with selected non-ceramic systems (B, Si, TiAlNb) added for comparison. When the inoculant loading is optimised, prior-β grain refinement to the sub-millimetre scale is consistently reported, whereas over-loading commonly increases porosity, anisotropy or embrittlement. In contrast to earlier broader nano-inoculant surveys spanning multiple base metals, this review focuses specifically on ceramic inoculants for Ti-alloy AM, offering a mechanistic and process-oriented framework to guide future alloy and process design.
The thermal field inside industrial Czochralski (CZ) silicon crystal growth systems strongly influences the crystal quality, oxygen incorporation, interface stability, intrinsic point-defect behavior, and furnace energy efficiency. A fully coupled three-dimensional thermal-fluid model was developed to investigate the influence of a furnace’s thermal-field design on industrial CZ silicon crystal growth under realistic operating conditions. The numerical framework simultaneously considers conductive, convective, and radiative heat transfer, as well as melt convection, turbulence transport, argon gas flow, and interface heat transfer. A comprehensive simulation database consisting of 2588 operating conditions was generated by systematically varying seven major thermal-field design parameters related to the heater geometry, heater position, insulation configuration, thermal shielding, and heater power redistribution. Four characteristic thermal outputs were extracted for thermal-field evaluation and machine-learning (ML) prediction: the average crucible temperature, total heater power consumption, average V/G ratio at the crystal–melt interface, and the radial nonuniformity of V/G , where V is the crystal pulling rate, and G is the axial thermal gradient near the solid–liquid interface. Multiple ML algorithms were evaluated for predicting the complex nonlinear thermal behavior of the CZ system: linear regression (LR), artificial neural network (ANN), support vector machine (SVM), random forest (RF), gradient boosting (GB), and stochastic gradient descent (SGD) models. The GB model consistently achieved the highest prediction accuracy, particularly for interface-sensitive thermal parameters. Furthermore, explainable artificial intelligence analysis based on SHapley Additive exPlanations (SHAP) was used to identify the dominant thermal-control mechanisms that govern the furnace’s thermal behavior and interface stability. The results show that the dominant parameters were the main-heater length, main-heater vertical position, thermal gap between the melt free surface and thermal shield, and heater-power ratio. These parameters controlled the global furnace thermal behavior and the localized interface thermal uniformity. Smaller thermal gaps increased the asymmetry of radiative cooling near the crystal shoulder region and the radial V/G nonuniformity, whereas larger thermal gaps led to smoother thermal distributions at the interface. The total power consumption of the furnace varied significantly from 52.6 to 76.9 kW for different thermal-field configurations, which indicates that there is considerable potential for industrial energy optimization through redesign of the thermal field. The proposed COMSOL–ML–SHAP framework provides both highly accurate thermal-field prediction capability and physically interpretable insight into the dominant thermal mechanisms that govern the crystal quality, energy consumption, and defect-related growth behavior in such growth systems.
Manual assembly workstations still devote a relevant share of their cycle time to non-value-added picking-and-placing activities, which may also expose operators to awkward postures when bulky parts are stored in rear locations. This paper studies whether collaborative robots (cobots) can support large-part picking in Human–Robot Collaboration (HRC) assembly systems while jointly improving productivity, ergonomics, and economic feasibility. A structured offline decision-support framework is proposed. The method combines storage-location- and part-attribute-dependent estimation of Picking-and-Placing Time (PPT), a Mixed-Integer Linear Programming (MILP) task allocation model for two operators and one cobot, a time-weighted Rapid Entire Body Assessment (REBA) indicator, and an iso-payback economic analysis. The framework is applied to a 26-task assembly case study. Compared with the human-only baseline, the makespan-oriented solution reduces the cycle time from 553 s to 493 s (about 11
Aluminum alloys are extensively used in new energy vehicles and aerospace components owing to their high specific strength and low density. However, adhesive wear remains the primary failure mode of polycrystalline diamond (PCD) tools during the dry turning of these alloys. This study systematically investigates the influence of tool rake angle on adhesive wear evolution using PCD tools with rake angles of 3°, 6°, 9°, 12°, and 15°. Wear morphologies in the critical region near the cutting edge on both rake and flank faces were characterized using scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDS). Cutting forces and temperatures were measured to support the analysis of wear mechanisms. The results show that the three-component cutting forces decreased with increasing rake angle, with the principal cutting force reduced by 22.94
Additive manufacturing has recently emerged as a viable approach for producing prototype injection moulds for low-volume manufacturing. PolyJet-printed photopolymer inserts offer rapid lead times and cost efficiency, but their limited thermal resistance, low thermal conductivity, and tendency to creep are fundamental constraints on mould lifetime and process stability. In this study, the suitability of the MED615RGD biocompatible photopolymer for injection mould tooling was investigated and its performance was compared to the widely used VeroWhite photopolymer resin. Comprehensive thermal and mechanical characterization was conducted, including dynamic mechanical analysis (DMA), differential scanning calorimetry (DSC), heat deflection temperature (HDT) and Shore D hardness testing, creep analysis, and 3D optical scanning. These tests help to evaluate the fundamental thermal and mechanical properties of the mould, which govern its durability. After the material tests, mould inserts were printed from VeroWhite and MED615RGD. The inserts were subsequently built into a custom-designed, experimental injection mould, equipped with strain gauges and an embedded thermocouple, which enables real-time monitoring of deformation and thermal loads during the injection moulding tests. The experimental results showed that MED615RGD exhibits a higher glass transition temperature (Tg = 78.3 °C) and improved stiffness compared to VeroWhite (Tg = 70.5 °C), while maintaining similar hardness. During injection moulding, both materials suffered from progressive heat accumulation due to low conductivity, resulting in dimensional deviations and eventual edge failure. Tool degradation began in the 40–50 cycle range, consistent with previously reported PolyJet-based tooling lifetimes. The findings confirm that MED615RGD is suitable for rapid tooling applications where short manufacturing cycles, dimensional stability, and design iteration speed are the primary objectives. In this study, performance limitations, degradation mechanisms and practical process window recommendations are discussed for future industrial use of photopolymer rapid tooling.
Wire-feed laser additive manufacturing (WLAM) offers significant potential for producing bimetallic structures (BMs) by combining the unique properties of dissimilar alloys. This study investigates the microstructural evolution, mechanical properties, and environmental performance of a BM composed of Inconel 625 (IN625) and Stainless Steel 316 (SS316) fabricated using a WLAM system. Monolithic IN625, monolithic SS316, and a 50/50 IN625-SS316 BM were characterized using optical microscopy, scanning electron microscopy (SEM), X-ray diffraction (XRD), and mechanical testing. Microstructural analysis revealed a smooth transition zone with good metallurgical bonding, exhibiting epitaxial grain growth in the form of columnar dendrites and equiaxed grains with effective elemental mixing and no sharp discontinuities. XRD confirmed the stability of face-centered cubic (FCC) phases throughout the bimetallic structure, with no detectable peaks corresponding to brittle secondary phases. Mechanical testing showed that the BM exhibited intermediate properties, with a UTS of 506 MPa and an average hardness of 145 HV, corresponding to 27.5
Powder feedstock is the primary input to metal additive manufacturing processes such as laser powder bed fusion. Its characteristics are crucial for producing high-quality components. Realization of the waste reduction and cost benefits that this technology offers requires reclaiming, processing, and reusing powder feedstock. While these benefits can be significant, there are challenges related to the industrial implementation of powder reuse, including the variability in powder quality that can propagate through to the produced material. This variability can be both negative and positive, but ultimately leads to uncertainty for both users and regulators. This means material and process qualifications are difficult to achieve when powder feedstock is being repeatedly reused. This study reports the effects of 175 Ti-6Al-4V ELI powder reuse cycles, characterized at 23, 104, and 175 reuses, in a certified medical device production facility, whereby powder was topped up with virgin powder from the same production lot after each build cycle. To the best of our knowledge, this is the longest reuse campaign reported in the literature. During this campaign, flowability improved with Hall flow rate decreasing by 5.4
This study investigates the influence of cyclic low-temperature exposure on the crashworthiness response of embossed thin-walled structures made of AA6063-T6 aluminum alloy. The experimental program included static tensile tests on standardized dog-bone specimens and dynamic axial crushing tests on square, thin-walled columns equipped with spherical, concave crush initiators. The specimens were subjected to cyclic thermal exposure in the temperature range of 0–30 °C, corresponding to long-term operating conditions of roadside energy absorbers. Optical microscopy observations and X-ray diffraction (XRD) analysis revealed local microstructural changes in the embossed regions, including partial alignment and redistribution of Mg2Si precipitates, as well as changes in diffraction peak intensities associated with residual stress relaxation and recovery. Static tensile tests demonstrated a reduction in the yield strength and ultimate tensile strength after cyclic thermal exposure, whereas the largest decrease was observed for specimens subjected to the longest exposure period (T3 and T4). Dynamic crushing tests revealed a gradual reduction in the peak crushing force (Pmax) of 10
Manufacturing organizations are increasingly expected to improve product quality and operational efficiency while reducing production losses. In this context, zero-defect manufacturing (ZDM), supported by Industry 4.0 and Quality 4.0, promotes proactive and data-driven quality improvement. However, existing quality-management approaches remain largely focused on defect detection, occurrence frequency, or prediction accuracy, offering limited support for translating quality data into explainable, actionable improvement priorities. This study proposes the quality failure impact framework (QFIF) as an impact-oriented and explainable decision-support approach for defect prioritization. QFIF integrates the quality failure impact index (QFII) for historical frequency–impact assessment, SHapley Additive exPlanations (SHAP) for model-based defect-contribution analysis, and one-dimensional sensitivity analysis (1DSA) for response-sensitivity evaluation. These outputs are subsequently integrated through an integrated priority score (IPS) to establish the final cross-defect improvement ranking. The framework was validated through a wiring harness manufacturing case study. The results identified incorrect cut length, feeder slip, and insulation damage as the highest-priority defects. Following targeted improvement actions, these defects decreased by approximately 48.88
In this study, we propose a shape optimization method for multi-stage preform design in hot forging. The method is intended to provide a computationally efficient alternative to conventional nonlinear large-deformation simulations, which are often costly when repeated during design optimization. Our approach models each forging step by a linear elastic deformation and combines this approximation with a gradient-based shape optimization framework. A key ingredient is a distance-function-based formulation for prescribing boundary displacements on the workpiece surface, which enables the deformation analysis to be carried out efficiently using the standard finite element method. The optimization is performed by a level set method driven by the shape derivative of a cost functional that evaluates the mechanical performance of the deformation process. In the present study, we focus on axisymmetric workpieces and formulate the corresponding optimization problem for intermediate preform shapes in a multi-stage forging sequence. Numerical examples are presented to demonstrate the behavior of the proposed method and to confirm that the obtained preform designs can improve the deformation process while keeping the computational cost low. These results suggest that the proposed framework is a promising tool for practical preform design in hot forging.
Deploying AI-based monitoring systems in laser powder bed fusion (PBF-LB/M) requires upstream decisions concerning process knowledge, data quality, and operational constraints. Many studies reviewed here report these decisions as task-specific choices, which makes it difficult to relate model performance to deployment requirements. This paper presents a standards-integrated systems engineering framework for AI monitoring in PBF-LB/M that combines ISO/IEC 25059, ISO/IEC 5259–4, ASTM E3353, and ISO/IEC 5338. The framework formalizes monitoring-target definition, label-taxonomy derivation, deployment-oriented data partitioning, and latency specification before model development and links these decisions to explicit verification evidence. We evaluate one instantiation through paired compliant and non-compliant conditions using the same data and model architectures. The comparisons show that upstream design choices affect deployment-relevant capability even when model capacity is held constant. Binary labeling yields a higher aggregate classification score but cannot distinguish the opposite corrective actions associated with keyhole-prone and lack-of-fusion-prone conditions. A mixed scan-pattern split obscures shifted-condition performance, while monitoring-scale analysis changes which architectures satisfy the latency constraint. Of the six quality-characteristic requirements specified for the case study, five satisfy the measured acceptance criteria. Robustness remains unmet because classification performance degrades under the tested compound shift, although a feature-space detector identifies the tested shifted condition with low computational overhead. Evidence for Intervenability is limited to the model-independent computational decision path because controller communication and physical actuation are not measured. The annotation protocol is evaluated through sensitivity analysis, process-scale consistency checks, and an independent expert assessment, but direct volumetric defect ground truth is unavailable. These results show how the framework converts upstream design decisions into traceable verification targets and exposes deployment limitations that aggregate model metrics can obscure. The framework is intended as a reusable design structure, while empirical validation in this study is limited to the present PBF-LB/M case.
We introduce a novel hybrid Pulsed-Electric Vibratory Machining process that integrates high-frequency ultrasonic vibrations with an externally applied electric field to enhance the machinability of aerospace-grade alloys, demonstrated here on Ti-6Al-4 V. Electrically-assisted (EA) manufacturing leverages the electroplastic effect, wherein the application of electric current facilitates plastic deformation by reducing flow stress and enhancing material ductility, thereby improving machinability. The novelty is the integration of electrically-assisted electroplasticity with ultrasonic vibratory turning into a single process. We investigate the influence of high and low current densities and varying current frequencies, including continuous current profiles, on the turning process, in conjunction with mechanical ultrasonic vibration. The significance is evidenced by substantial reductions in cutting force, improved surface integrity, and favourable thermal profiles. This work highlights, for the first time, the potential of EA-ultrasonic hybrid machining as an effective strategy for improving machining operations for difficult-to-machine materials.
Manufacturing companies require large numbers of high-performance cutting tools, which are often made from cemented carbides due to their high hardness and toughness. Cobalt (Co) is typically used as a binder because of its favorable mechanical properties and the high solubility of tungsten carbide (WC) in Co. However, Co is considered a critical raw material, as it is both economically important and subject to high supply risks due to political instabilities in its mining regions. In addition, Co mining and handling pose environmental and health risks to workers. These concerns motivate the search for alternative binder materials in cemented carbides. Nickel (Ni), iron (Fe), and a nickel–chromium-cobalt mixed binder (NiCrCo) are investigated in this work as potential binders containing less or no Co. The grinding behavior of cemented carbides has so far mainly been studied for Co-bonded systems. For cemented carbides with alternative binders, the grinding behavior in dependency of the material characteristics and grinding parameters remains unclear. This work characterizes these cemented carbides using hardness and critical fracture toughness measurements at various temperatures, and mesoscale structure analysis. Furthermore, analogy surface grinding experiments on the mentioned materials are carried out. Combining the material characterization and grinding analysis, conclusions on the material removal behavior are derived. The study shows that alternative binders lead to comparable, but not equal material properties and grinding behavior. These results contribute to a knowledge-based grinding of cemented carbides with alternative binders, supporting more economical, ecological, and human-friendly production of cemented carbide cutting tools.
Next-generation carbon fiber-reinforced thermoplastics (CFRTPs), such as carbon fiber-reinforced polyetheretherketone (CF/PEEK), offer a promising and sustainable alternative to conventional thermoset carbon fiber-reinforced plastics (CFRPs). However, their surface milling performance and the influence of cooling strategies on machining behavior remain insufficiently understood. In this study, the machinability of CF/PEEK during surface milling was systematically investigated, focusing on the evolution of cutting forces, cutting temperature, and surface quality at different fiber orientations, feed rates, and machining inclination angles. The effects of flood cooling, supercritical CO₂ (scCO₂) cooling, and cold air cooling on the surface milling performance of CF/PEEK were comparatively analyzed. The results showed that cutting force, cutting temperature, and surface roughness increased with increasing feed rate. The highest cutting force and temperature were observed at a fiber orientation of 135°, whereas the poorest surface quality occurred at 0°. As the machining inclination angle increased, surface roughness initially decreased and subsequently increased, reaching a minimum at 20°. Among the three cooling strategies, scCO₂ cooling provided the greatest cooling capacity, whereas flood and cold-air cooling were more effective in improving surface quality. Overall, considering machining performance, economic efficiency, and green manufacturing requirements, cold air cooling emerged as the preferred strategy for the surface milling of CF/PEEK. These findings provide further insight into the surface milling behavior of CF/PEEK and offer practical guidance for developing high-performance and sustainable manufacturing processes.
The manufacturing industry is facing increasing pressure to improve efficiency and sustainability in response to energy constraints and environmental regulations. Within this context mold cleaning represents a critical and energy-intensive step in resin-based component manufacturing. Although conventional manual sandblasting is widely adopted it is affected by poor repeatability limited process optimization and significant material and energy waste. The objective of this work is to develop and validate a predictive numerical tool capable of supporting the optimization and digitalization of the sandblasting process enabling higher efficiency and improved sustainability. A validated multiphase Computational Fluid Dynamics (CFD) model based on the Discrete Phase Model (DPM) was developed to simulate particle-laden jets interacting with mold surfaces. The methodology involved an accurate characterization of the nozzle flow field to determine particle velocity and distribution at the nozzle exit. The CFD model was rigorously validated through comparison with experimental sandblasting tests evaluating residual material on the mold surface under different nozzle–mold distances and jet inclination angles. Furthermore, a simplified plate-based CFD model was employed to derive optimal nozzle trajectories and pass strategies reducing computational cost while preserving predictive capability. The numerical framework was designed to be integrated into a digital manufacturing environment for a real-time camera-based monitoring. The results demonstrate a strong agreement between numerical predictions and experimental measurements confirming the reliability of the CFD model in describing material removal mechanisms. The proposed approach enables the rational optimization of process parameters and nozzle paths leading to a significant improvement in cleaning efficiency and process repeatability. By reducing unnecessary sand usage process time and rework the methodology contributes to lower energy consumption and material waste. This work introduces an innovative digital tool for sandblasting process optimization supporting the transition toward smarter and more sustainable mold maintenance operations.
