Abstract The brittle–ductile transition depth is a key parameter for understanding material-removal mechanisms and achieving damage-free surfaces in the machining of brittle single crystals. However, existing predictive models often neglect crystallographic anisotropy and the scale-dependent effects associated with the tool edge radius in micro-cutting. To address these limitations, a multi-parameter anisotropic model is developed to predict the critical depth of cut, incorporating directional variations in hardness, elastic modulus, shear modulus, fracture toughness, and specific surface energy. Edge-radius-related corrections are introduced into the effective rake and shear angles to account for the local stress fields present in micro-cutting. The model is validated through multi-directional scratch tests on CaF2, single-crystal Si, and single-crystal Ge. On the Ge (1 0 0) plane, the predicted maximum and minimum critical depths differ from experimental results by only 2 nm, representing a 60% improvement compared with models that ignore edge-radius effects. Additional experiments show that larger tool nose radii produce higher critical depths of cut, and this trend is accurately reproduced by the model.
Abstract Operational condition monitoring of manufacturing equipment for predictive maintenance is crucial to enable early detection of abnormal behavior and prevention of costly downtime. In industry settings, normal operation data are abundant, whereas abnormal or failure data are rarely available, limiting the development of reliable data-driven diagnostic models. Generative modeling provides a promising approach to address this imbalance by learning the latent distribution of normal operation and defining the boundary of potential anomalies. This study proposes a generative latent-space framework for sound-based condition monitoring of a stamping machine using an internal sound sensor (ISS). The framework incorporates MTConnect for machine context as conditional information to guide the generative process. Sound sensing offers a low-cost, nonintrusive, and high-resolution modality, while MTConnect context provides synchronized operational states. The latent representation of sound features is organized around part-level representative signals, forming structured regions that capture characteristic acoustic patterns of normal operation. This structure guides the generative process toward realistic variations of normal sound while limiting deviations toward implausible samples. The proposed framework integrates two complementary generative architectures, a conditional variational autoencoder (CVAE) and a conditional diffusion model (CDM), to construct a hybrid augmented normal dataset. Experimental results show that retraining the encoder with the synthesized data enhances separation of part-specific sound representations and improves discrimination between normal and abnormal operation states. These findings demonstrate that the proposed framework effectively leverages normal sound data to define anomaly boundaries and strengthen the robustness of sound-based predictive maintenance systems for stamping operations.
Abstract NASA's in-situ resource utilization (ISRU) initiative aims to eliminate Earth-dependency by developing sustainable, on-site manufacturing solutions for deep-space endurance and colonization missions to the Moon and Mars, and even beyond. Regolith–resin composites, which combine abundant extraterrestrial soil with UV-curable polymers, represent a promising class of materials for lightweight, energy-efficient, and on-site fabrication. This research investigates the feasibility and performance of regolith–resin composites with solid loadings up to 85 wt%, using Lunar Highlands Simulant (LHS-1) Regolith and Mars Global Simulant (MGS-1) Regolith as fillers. Two UV-assisted approaches, pressure compaction and blade coating, were employed to fabricate composite specimens in a layer-by-layer style. The fabricated samples were then evaluated for producing robust, thermally insulating, and radiation-shielding components suitable for extraterrestrial environments. Mechanical testing demonstrated compressive strengths up to 8.65 MPa and tensile strengths up to 1.63 MPa, with pressure compaction generally yielding higher strength due to improved compaction. Thermal conductivity measurements ranged from 0.05 to 0.107 W m−1 K−1, indicating strong heat insulation properties. Radiation-shielding simulations using NASA's OLTARIS tool showed that the composites achieved dose-equivalent performance. This research demonstrates the potential of regolith–resin composites as a practical solution for in-space manufacturing, combining high regolith utilization with favorable multifunctional performance for deep-space missions and long-term missions.
Abstract Precision grinding is a key technology for thinning silicon carbide substrates. However, the surface accuracy often deteriorates due to total thickness variation (TTV), which critically affects subsequent device fabrication and packaging. Existing research primarily focuses on correcting preset geometric alignment, such as the tilt between vacuum chucks and grinding wheels, while insufficiently considering the influence of force-induced structural deformation during grinding. In this article, a mechanics-based model is developed to predict TTV in silicon carbide substrate grinding based on an equivalent wheel-deflection model. The model quantitatively incorporates grinding process parameters and the material and geometric properties of the grinding wheel and substrate. By equivalently representing the force-induced compliance of the grinding system as wheel deflection, the model captures the resulting change in effective inclination angle and its influence on non-uniform material removal. Experimental validation shows that the model achieves high accuracy, with an average deviation of less than 4%. This work provides a mechanics-based basis for surface-form prediction and parameter-sensitive TTV compensation in precision grinding.
Abstract This article presents a new framework for modeling machine tool dynamics to enable their integration into machine tool digital twins. The framework is formulated as a dynamic Bayesian network, in which model parameters, sensor and process observations, and decision variables are represented explicitly and updated sequentially using Bayesian inference. The framework incorporates a library of candidate models and enables model selection, uncertainty quantification, and physical system state estimation through probabilistic inference. These properties enable scalable digital twin creation at both the machine and fleet levels. The proposed framework is designed to be broadly adoptable across a wide range of machine tool dynamics problems. In this article, we demonstrate its implementation through a case study in feed-drive dynamics, where the generated model interfaces with a 3D printer’s controller to adaptively produce optimal trajectories that compensate for structural vibrations during printing. Bidirectional digital–physical communication loop is established and experimentally validated, confirming the framework’s ability to predict and adapt to system variations. Results indicate improved vibration compensation and enhanced surface quality under varying dynamic conditions in this case study, while computational challenges for enabling general implementation are discussed.
