Chinedum E. Okwudire

The number of publications on digital twins went from 3.000 in 2017 to over 70.000 in 2024. This paper analyses the use and status of digital twins for machine tools as well as the necessary modelling and identification for enabling optimization, process planning, process control and predictive maintenance on the machine tool level and the fleet level. Recent research on digital twins for machine tools leveraging AI or greybox models is also presented. This paper gives an overview of the application and the technology behind digital twins used for machine tool design, commissioning, and operations.

Repeated machining passes (i.e., continuous toolpaths) are common in CNC manufacturing, including multi-level machining of prismatic parts and iso-contour passes in contour machining. They present an opportunity to exploit pass-to-pass learning to improve productivity without sacrificing quality through feedrate optimization. Traditional iterative learning methods provide a means to exploit pass-to-pass learning for quality improvements, but they are not well-suited to feedrate optimization because the reference trajectories change as the feedrate increases. In the authors’ prior work, learning-based feedrate optimization was demonstrated for repeated machining along identical toolpaths. This paper extends that concept to the more challenging case of similar but non-identical cutting paths, as encountered in contour machining. A pass-to-pass learning strategy is proposed in which corresponding sections of non-identical iso-contour passes are identified using a contour-matching method based on geometric similarity. Bayesian linear regression models are then used to learn and predict contour error and spindle torque across passes, with uncertainty explicitly quantified through credible intervals. These predictions are embedded in a window-based feedrate optimization framework solved via sequential linear programming, enabling feedrate maximization subject to kinematic, contour-error, and spindle-torque constraints. The proposed approach is experimentally validated on a 3-axis desktop CNC milling machine through multiple 2.5D contour machining case studies. Results show that the method can rapidly approach near-optimal feedrates after only a few passes, culminating in up to 16.4% increase in productivity compared to an equivalent learning-based feedrate optimization approach for identical toolpaths.

Laser powder bed fusion (LPBF) enables fabrication of complex metal components but remains limited by residual stress accumulation and part deformation. Most existing scan sequence generation strategies for LPBF rely on heuristic rules or empirical optimizations that are suboptimal, difficult to generalize across geometries, and insensitive to the underlying physics of the problem. The SmartScan framework was developed to overcome these limitations through model-based and optimization-driven scan sequence generation. SmartScan 1.0 employed a thermal model to optimize temperature uniformity, leading to significant reductions in residual stress and distortion compared to state-of-the-art heuristic approaches. However, its formulation ignored the mechanical aspects of residual stress and deformation. To address this deficiency, a preliminary study introduced SmartScan 2.0 (Pre) which utilized a decoupled linear thermomechanical formulation for scan sequence optimization for 2D geometries. Building on this foundation, this paper proposes SmartScan 2.0 based on a sequentially coupled linear thermoelastic model that simultaneously solves temperature and displacement fields to minimize thermally induced elastic deformation in 3D geometries. The computational efficiency of SmartScan 2.0 is enhanced through nondimensional scaling. Experimental validation on 3D LPBF specimens shows that SmartScan 2.0 achieves up to 69.0% reduction in residual stress and 17.4% reduction in deformation relative to SmartScan 1.0, and up to 60.6% reduction in residual stress and 12.8% reduction in deformation compared with SmartScan 2.0 (Pre). This work establishes the superiority of scan sequence optimization using coupled linear thermomechanical models over the existing thermal-only or decoupled thermomechanical approaches, without significantly sacrificing computational efficiency.

Modeling the track cross-sectional profile (CSP) in deposition processes is critical for assessing and controlling deposition quality. This article focuses on modeling time-invariant deposition (TID) processes, where deposited material does not move after impact with the evolving surface, and deposition efficiency remains constant. For a TID process, the CSP can be computed from the mass flux distribution using the Abel integral transform. The model is validated for cold spray (CS) and aerosol jet printing. Using the TID assumption enables the modeling of CS and AJP tracks from dot deposition data with height and width errors down to 11 and 22% for CS, and 7 and 21% for AJP. The error of this model when considering short and curved tracks is discussed, as well as the effects of nozzle standoff distance and tilt. Fast methods for arbitrary CSP computations and a fast CS method considering varying deposition efficiency are also discussed.

Additive Manufacturing (AM) has gained wide attention in the past two decades and emerged as a significant method in the manufacturing sector. Advancements in AM have enhanced productivity, reduced lead times, and improved part quality while maintaining cost-effectiveness. Despite advancements in materials, technologies, and parameter optimization, the widespread AM adoption is limited by a lack of skilled workforce. This research presents a hybrid learning approach to address this gap through curricula and hands-on training. The proposed framework includes hybrid educational and training approaches in polymer-based FFF, SLA, SLS, and Metal FFF technologies toward the development of a workforce skilled in AM

Parts produced by laser powder bed fusion (LPBF) are susceptible to residual stress, deformation, and other defects that are strongly associated with non-uniform temperature distribution during the printing process. The authors, in their prior work, have proposed SmartScan, an intelligent scan sequence optimization approach that uses purely thermal models to achieve uniform temperature distribution, as an indirect means for achieving reduced residual stress and deformation in parts produced via LPBF. SmartScan was shown to outperform state-of-the-art heuristic scan sequences in reducing residual stress, distortion, and geometric errors in 3D printed parts. However, the thermal-model-only approach to SmartScan fails to account for the complex thermomechanical interactions that map non-uniform temperature distribution to residual stress and distortion in LPBF. This paper presents an initial (2D) investigation into a new version of SmartScan, called SmartScan 2.0, that integrates both thermal and mechanical models to better optimize scan sequences. To achieve this, local temperature gradients are extracted from a thermal model and combined with local compliance information extracted from a mechanical model to create a composite objective function that is optimized using an efficient control-theoretic approach. The effectiveness of SmartScan 2.0 in comparison with the thermal-only version of SmartScan (i.e., SmartScan 1.0) is evaluated in simulations and experiments involving laser marking of thin stainless steel 316L plates clamped in different configurations, as a proxy for 2D layers of parts built using LPBF. While SmartScan 2.0 demonstrated worse thermal uniformity than SmartScan 1.0, it consistently achieved lower maximum deformation of the marked plates (by up to 23.7%) compared to SmartScan 1.0. This supports the well-known fact that better thermal uniformity does not necessarily imply lower maximum deformation in LPBF, and motivates further research into SmartScan 2.0. We also show that, despite its use of thermomechanical models, SmartScan 2.0 is computationally efficient thus facilitating its usefulness in practice.

Metal binder jetting (MBJ) is a promising additive manufacturing technology, offering advantages including higher printing speed and lower residual stress in printed parts compared to other metal additive manufacturing technologies. However, to attain the full commercial potential of MBJ, there is strong interest from industry to substitute high cost spherical powder with lower cost non-spherical powder, like water atomized powder (WAP). The problem is that the non-sphericity of WAP leads to low powder bed packing fraction (i.e., the quotient of the powder bulk density and the particle density), which in turn increases the porosity and shape distortion of the printed parts. The packing fraction and final quality of MBJ parts printed using WAP could be improved by implementing real-time closed-loop control on the packing fraction. However, there are no suitable approaches to measure packing fraction in situ during MBJ to facilitate closed-loop control. This paper addresses this shortcoming by hypothesizing that recoating force in MBJ is a suitable signal for accurately measuring packing fraction in situ. To test this hypothesis experimentally, an apparatus consisting of a moving powder bed compacted by a recoater attached to a torque sensor was designed and prototyped. In the experiment, the packing fraction of the powder bed was varied to 15 different levels and the recoating force was measured for each case using the apparatus. The mean recoating force was seen to have an almost linear relationship with the packing fraction (with R 2 values of 0.91 and 0.93 for two cases investigated) thus supporting our hypothesis. However, anomalous data exist and the powder preparation process involves human-induced error, indicating the need for further work to improve the precision of the measurements.

In recent years, Additive Manufacturing (AM) has rapidly emerged as a significant manufacturing method. Recent advances in AM, especially in materials and processing techniques, have reduced lead times, enhanced productivity, improved print resolution, and increased cost-effectiveness. Despite these advancements, knowledge gaps remain a barrier to AM’s adoption in industry and academia, emphasizing the need for workforce training to equip users with the skills to address its various challenges. This research proposes a framework through hybrid educational programs using a novel training-based approach to address knowledge and skill gaps in AM. The proposed approach trains users in Design for Additive Manufacturing (DfAM), focusing on Computer-Aided Design (CAD) and simulation to optimize technology and process parameter selection. Additionally, the framework facilitates comprehensive AM education and hands-on training in Fused Filament Fabrication (FFF), Stereolithography (SLA), Selective Laser Sintering (SLS), and Laser Powder Bed Fusion (LPBF), promoting adoption through skill-based training in technology selection, material handling, and decision-making in additive manufacturing processes. Future work suggests integrating virtual and augmented reality modules to enhance user experience with interactive learning.

Laser powder bed fusion (LPBF) is an additive manufacturing technique that has gained popularity thanks to its ability to produce geometrically complex, fully dense metal parts. However, these parts are prone to internal defects and geometric inaccuracies, stemming in part from variations in the melt pool. This paper proposes a novel vector-level feedforward control framework for regulating melt pool area in LPBF. By decoupling part-scale thermal behavior from small-scale melt pool physics, the controller provides a scale-agnostic prediction of melt pool area and efficient optimization over it. This is done by operating on two coupled lightweight models: a finite-difference thermal model that efficiently captures vector-level temperature fields and a reduced-order, analytical melt pool model. Each model is calibrated separately with minimal single-track and 2D experiments, and the framework is validated on a complex 3D geometry in both Inconel 718 and 316L stainless steel. Results showed that feedforward vector-level laser power scheduling reduced geometric inaccuracy in key dimensions by 62%, overall porosity by 16.5%, and photodiode variation by 6.8% on average. Overall, this modular, data-efficient approach demonstrates that proactively compensating for known thermal effects can significantly improve part quality while remaining computationally efficient and readily extensible to other materials and machines.

Abstract In automated de-powdering of binder jetting additive manufactured parts, robots must quickly and accurately calculate how to grasp fragile, brittle green parts without shattering them. Previous work has used non-linear models to find a grasp configuration (specifying the robot gripper position, orientation, and distance between gripper fingers) that minimizes part stress and the likelihood of shattering under the gripper force. However, this process is time-intensive and impractical for additive manufacturing applications, where robots must work quickly. This paper proposes a novel method for accurately and quickly calculating safe grasps for fragile green parts. Our method involves modeling parts using a finite element linear elastic model and calculating part deformation caused by a surface traction force from the gripper’s fingers in each grasp configuration. The grasp configuration resulting in the lowest part deformation is chosen as the least likely to shatter the part. For an elastic and brittle material, part stress is approximately proportional to strain before fracture, meaning the grasp with the lowest strain for the elastic model corresponds to the lowest stress for the brittle part with the same geometry. We integrated our algorithm into a prototype binder jetting de-powdering robot and used the algorithm to choose grasps for binder jetting manufactured green parts. With optimizations such as parallelization, our algorithm takes 18 minutes to calculate grasps per part. During testing, our algorithm’s chosen grasp configuration was used to pick up and hold test green parts without shattering them at 80 N, as measured by the gripper’s contact force sensors, despite the parts shattering under forces as low as 2N in some grasps. Handling fragile green parts is one of the main barriers hindering the automation of the de-powdering process in binder jetting. Our grasping algorithm’s success in enabling the safe handling of green parts will drive progress toward automated de-powdering.

Fleets of networked manufacturing machines of the same type, that are collocated or geographically distributed, are growing in popularity. An excellent example is the rise of 3D printing farms, which consist of multiple networked 3D printers operating in parallel, enabling faster production and efficient mass customization. However, optimizing process parameters across a fleet of manufacturing machines, even of the same type, remains a challenge due to machine-to-machine variability. Traditional trial-and-error approaches are inefficient, requiring extensive testing to determine optimal process parameters for an entire fleet. In this work, we introduce a machine learning-based collaborative recommender system that optimizes process parameters for each machine in a fleet by modeling the problem as a sequential matrix completion task. Our approach leverages spectral clustering and alternating least squares to iteratively refine parameter predictions, enabling real-time collaboration among the machines in a fleet while minimizing the number of experimental trials. We validate our method using a mini 3D printing farm consisting of ten 3D printers for which we optimize acceleration and speed settings to maximize print quality and productivity. Our approach achieves significantly faster convergence to optimal process parameters compared to non-collaborative matrix completion.

Bayesian optimization (BO) is a sequential decision-making tool widely used for optimizing expensive black-box functions. Recently, Large Language Models (LLMs) have shown remarkable adaptability in low-data regimes, making them promising tools for black-box optimization by leveraging contextual knowledge to propose high-quality query points. However, relying solely on LLMs as optimization agents introduces risks due to their lack of explicit surrogate modeling and calibrated uncertainty, as well as their inherently opaque internal mechanisms. This structural opacity makes it difficult to characterize or control the exploration-exploitation trade-off, ultimately undermining theoretical tractability and reliability. To address this, we propose LLINBO: LLM-in-the-Loop BO, a hybrid framework for BO that combines LLMs with statistical surrogate experts (e.g., Gaussian Processes (GP)). The core philosophy is to leverage contextual reasoning strengths of LLMs for early exploration, while relying on principled statistical models to guide efficient exploitation. Specifically, we introduce three mechanisms that enable this collaboration and establish their theoretical guarantees. We end the paper with a real-life proof-of-concept in the context of 3D printing. The code to reproduce the results can be found at https://github.com/UMDataScienceLab/LLM-in-the-Loop-BO.

Abstract Laser powder bed fusion (LPBF) is an additive manufacturing technique that is gaining popularity for producing metallic parts in various industries. However, parts produced by LPBF are prone to residual stress, deformation, cracks, and other quality defects due to uneven temperature distribution during the LPBF process. To address this issue, in prior work, the authors have proposed SmartScan, a method for determining laser scan sequence in LPBF using an intelligent (i.e., model-based and optimization-driven) approach, rather than using heuristics, and applied it to simple 2D geometries. This paper presents a generalized SmartScan methodology that is applicable to arbitrary 3D geometries. This is achieved by (1) expanding the thermal model and optimization approach used in SmartScan to multiple layers, (2) enabling SmartScan to process shapes with arbitrary contours and infill patterns within each layer, (3) providing the optimization in SmartScan with a balance of exploration and exploitation to make it less myopic, and (4) improving SmartScan’s computational efficiency via model order reduction using singular value decomposition. Sample 3D test artifacts are simulated and printed using SmartScan in comparison with common heuristic scan sequences. Reductions of up to 92% in temperature inhomogeneity, 86% in residual stress, 24% in maximum deformation, and 50% in geometric inaccuracy were observed using SmartScan, without significantly sacrificing print speed. An approach for using SmartScan for printing complex 3D parts in practice, by integrating it as a plug-in to a commercial slicing software, was also demonstrated experimentally, along with its benefits in significantly improving printed part quality.

Chinedum OkwudireChinedum OkwudireThe 19th ASME International Manufacturing Science and Engineering Conference (MSEC 2024), sponsored by the Manufacturing Engineering Division (MED) of ASME, was held from June 17, 2024 to June 21, 2024 in Knoxville, TN. New to the MSEC 2024 technical program was the introduction of brief papers in addition to full-length papers. Brief papers are styled after the technical briefs of the Journal of Manufacturing Science and Engineering (JMSE). Brief papers undergo full peer review and are published in the conference proceedings in the same manner as full papers. However, they are shorter in length than full papers and, therefore, can be used to report preliminary research results for early feedback from the manufacturing community.MSEC 2024 received 238 submissions—102 (42%) brief papers and 137 (57%) full papers. After rigorous peer review, 204 technical papers—91 (45%) brief papers and 113 (55%) full papers—were accepted for publication. The technical papers had global representation, with authors from 22 countries across five continents, including the US, China, Germany, Australia, Brazil, and South Africa.Among the accepted technical papers, MSEC symposium organizers nominated 16 full papers to be fast-tracked to the JMSE. All of the papers, together with their reviews, were sent to the JMSE Editor-in-Chief for a new round of journal paper review. A total of eight top MSEC full papers received positive journal reviews and were compiled and published in this JMSE Special Section on MSEC 2024. The papers selected for this special section cover a wide range of topics. They come from technical tracks of the ASME MED, including advanced materials manufacturing, biomanufacturing, life cycle engineering, manufacturing processes, nano/micro/meso manufacturing, and quality and reliability.As a leading international conference held annually on manufacturing process technology, MSEC acts as a global bridge between industry, government laboratories, and academic institutions. This Special Section showcases recent manufacturing research advancements presented at MSEC 2024. It also provides a platform for researchers and practitioners to widely disseminate their research findings and innovative practices that may inspire future scientific and technological breakthroughs.Albert ShihAlbert ShihWe would like to thank all of the symposium organizers of MSEC 2024 for their dedicated management of the symposia and for guarding the quality of the papers to be fast-tracked, which has contributed a great deal to the success of this Special Section. We would also like to thank all of the reviewers of the paper submissions for their detailed suggestions to improve the papers' quality. Special thanks are to the ASME MED Executive and Technical Committees and the ASME staff, especially Lori Lee and Elizabeth Bruce. Their outstanding contributions in managing the submitted technical papers ensured the high-quality publication of this Special Section for MSEC 2024.JMSE continues to seek close partnerships with MED and MSEC to serve our manufacturing community. Authors of accepted JMSE papers published between Mar. 2023 and Feb. 2024 were offered the opportunity to present their papers at MSEC 2024. A record number of 20 authors selected this option and presented their journal-quality research work in MSEC 2024. This has expanded the opportunity for colleagues in our manufacturing community to submit their top research papers to JMSE and then disseminate them in a presentation to our manufacturing community at MSEC.This Special Section marks another important step for JMSE to connect with MED and MSEC. As the brief papers get more established at MSEC, we envision future brief papers being fast-tracked to JMSE technical briefs. Future Technical Program Chairs and JMSE Editors will continue to make this a seamless process. JMSE seeks top research papers from our colleagues and strives to serve our community as a platform for the timely publication of high-impact research work.

Laser powder bed fusion (LPBF) is an additive manufacturing technique that is gaining popularity for producing metallic parts in various industries. However, parts produced by LPBF are prone to residual stress, deformation, cracks and other quality defects due to uneven temperature distribution during the LPBF process. To address this issue, in prior work, the authors have proposed SmartScan, a method for determining laser scan sequence in LPBF using an intelligent (i.e., model-based and optimization-driven) approach, rather than using heuristics, and applied it to simple 2D geometries. This paper presents a generalized SmartScan methodology that is applicable to arbitrary 3D geometries. This is achieved by: (1) expanding the thermal model and optimization approach used in SmartScan to multiple layers; (2) enabling SmartScan to process shapes with arbitrary contours and infill patterns within each layer; (3) providing the optimization in SmartScan with a balance of exploration and exploitation to make it less myopic; and (4) improving SmartScan's computational efficiency via model order reduction using singular value decomposition. Sample 3D test artifacts are simulated and printed using SmartScan in comparison with common heuristic scan sequences. Reductions of up to 92% in temperature inhomogeneity, 86% in residual stress, 24% in maximum deformation and 50% in geometric inaccuracy were observed using SmartScan, without significantly sacrificing print speed. An approach for using SmartScan for printing complex 3D parts in practice, by integrating it as a plug-in to a commercial slicing software, was also demonstrated experimentally, along with its benefits in significantly improving printed part quality.

The structural flexibility of industrial robot arms makes them vibrate when they are commanded to move at fast operation speeds. Among the control strategies, feedforward control stands out as an interesting approach to suppress vibration since it does not create stability issues and works for repeating and non-repeating tasks. Currently, the state-of-the-art feedforward controller dedicated to suppressing residual vibration in robot arms is time-varying input shaping (TVIP). However, TVIP falls short in trajectory tracking tasks since the method adds delays in the commands creating errors in tracking and thereby contouring trajectories. Therefore, this paper proposes the use of an alternate feedforward method, known as the filtered B-splines (FBS) approach, to suppress vibration in six DOF robots while maintaining tracking accuracy. Since time-varying FBS (TVFBS) requires full frequency response functions (FRFs), compared to only natural frequencies and damping ratios for TVIP, we propose a framework for estimating the FRFs of serial kinematic chain 6-degree-of-freedom robots. Residual vibration reduction experiments and trajectory tracking experiments, in which the dynamics of a UR5e collaborative robot change considerably, were carried out to validate the model prediction framework. TVFBS reduced the end-effector vibration by 87% while improving tracking performance in both the y (22%) and z (29%) directions. On the other hand, TVIP worsened the tracking performance (-683.43% for the y and -662.37% for the z direction) despite the excellent vibration reduction (98%). Hence, TVFBS demonstrated significantly better tracking performance than TVIP while retaining comparable vibration reduction.

Abstract Over the last few decades, globalization has weakened the US manufacturing sector. The COVID-19 pandemic revealed import dependencies and supply chain shocks that have raised public and private awareness of the need to rebuild domestic production. A range of new technologies, collectively called Industry 4.0, create opportunities to revolutionize domestic and local manufacturing. Success depends on further refinement of those technologies, broad implementation throughout private companies, and concerted efforts to rebuild the industrial commons, the national ecosystem of producers, suppliers, service providers, educators, and workforce necessary to regain a competitive, innovative manufacturing sector. A recent workshop sponsored by the Engineering Research Visioning Alliance (ERVA) identified a range of challenges and opportunities to build a resilient, flexible, scalable, and high-quality manufacturing sector. This paper provides a strategic roadmap for regaining US manufacturing leadership by briefly summarizing discussions at the ERVA-sponsored workshop held in 2023 and providing additional analysis of key technical and economic issues that must be addressed to achieve dynamic, high-value manufacturing in the USA. The focus of this presentation is on discrete manufacturing of production of structural components, a large subset of total manufacturing that produces high-value inputs and finished products for domestic consumption and export.

Abstract Fused filament fabrication (FFF) has gained widespread recognition across diverse industries owing to its rapid prototyping and cost-effectiveness advantages. As a result, it is the most prevalent modality for desktop 3D printing. However, FFF can be susceptible to a variety of printing defects and jeopardize the printing quality. Monitoring when defects occur during 3D printing and promptly stopping faulty printing remains a significant challenge. To address this challenge, engineers have developed techniques for detecting and characterizing defects during the FFF printing process. They can be categorized into contact and non-contact detection methodologies. Non-contact methods usually rely on computer vision or laser scanning. However, computer vision needs the assistance of machine learning and demands a substantial amount of training data for accurate detection. Moreover, computer vision is susceptible to ambient light conditions. The laser scanning method detects the printing defects by comparing the point cloud obtained from scanning the printed object with the CAD model. However, this approach heavily depends on the precision of the laser scanner, and achieving high accuracy often entails a significant financial investment for a good laser scanner. To improve accuracy and cost-effectiveness, a low-cost contact-based detection system called MTouch was developed in prior work. However, using contact sensors carries a risk of damaging fragile prints and leading to printing failures. In response, this paper introduces a non-contact, cost-effective, and robust detection method, MScan, to detect defects during the printing process. In the MScan setup, a laser-camera sensor is designed with a laser stripe emitter and a camera module based on laser triangulation to assess the absence of the printed object during the printing process. Additionally, MScan employs an effective and straightforward image processing and data acquisition algorithm to ensure its robustness and computational efficiency. The effectiveness of MScan is demonstrated experimentally by deploying it on an Ender 3 desktop FFF 3D printer. A fault detection accuracy of over 95% is achieved. Furthermore, MScan’s robustness to lighting variations is experimentally demonstrated.

Abstract Metal paste extrusion is an emerging metal additive manufacturing process at the intersection of fused filament fabrication and binder jet printing categories. It produces parts using extruded beads of a metal-and-binder paste and achieves final densification in a sintering furnace cycle. In the metal paste extrusion process, reducing build time is crucial for improving process economics. Similar to fused filament fabrication, the toolhead positioning in metal paste extrusion can induce vibrations in the printer frame and components, especially at high acceleration and speed needed to reduce build time. These vibrations can lead to defects in the deposited part. In this study, a commercially available vibration compensation algorithm, initially developed for fused filament fabrication, is applied to metal paste extrusion for the first time to reduce build time. The algorithm uses the natural frequencies and damping ratios of the machine’s motion system to predict and compensate for vibrations. Three test conditions are employed to produce sample parts in stainless steel 316L for geometric, material, and mechanical property evaluation: nominal printing settings without vibration compensation (print speed 40 mm/s), faster printing settings without vibration compensation (print speed 70 mm/s), and the same fast settings with vibration compensation. The test part consists of three sections: a solid (∼100% infill) section, a low-infill section (∼40%) where the infill and meshing are visible, and another low-infill section with two top layers. The resulting samples are analyzed for geometric error, density, and hardness effects. Geometrical errors are observed using macro photography, particularly in the outer perimeters at the corners. Voids (gaps between beads) within the top layer are observed using 3D scanning. Density following ASTM B311 standards and subsequently porosity are measured. The final test involves Rockwell B hardness testing at three specified points on the samples. Without vibration compensation, the outer perimeter exhibits ringing. Faster printing with vibration compensation demonstrates savings in build time without negatively affecting the quantified geometric, material, or mechanical properties.

Abstract The depowdering process in Binder Jetting Additive Manufacturing is indispensable for the integrity of 3D-printed objects, necessitating the removal of excess powder from their surfaces. However, this step poses a formidable challenge to the productivity of powder-based AM, as traditional depowdering methods reliant on manual labor prove time-consuming, labor-intensive, monotonous, and hazardous for human operators. Existing automated depowdering solutions exacerbate this challenge, with some systems demanding preliminary manual depowdering, nullifying automation benefits, while others lack the needed adaptability for diverse 3D-printed objects. In response, this manuscript proposes a robotic depowdering system designed to handle fragile 3D-printed components with precision beyond manual or less sophisticated automated methods. The proposed robotic depowdering system comprises two robots: Robot 1, responsible for generating trajectories to eliminate powder from the powder bed environment using fixed STL files, and Robot 2, which fine-tunes force and pressure while handling delicate, printed 3D objects, safeguarding their intricate structures. Preliminary experimental results confirm that the proposed depowdering system effectively handles fragile 3D objects from the powder bed environment and removes residual powder to a substantial degree. This accomplishment is of paramount significance as it directly contributes to the quality of the final product. By successfully addressing both of these critical aspects, the system demonstrates its capacity to ensure the structural integrity and overall quality of the end product, marking it as a valuable tool in the realm of additive manufacturing.

Abstract Frog-leg robots are widely used for wafer-handling in semiconductor manufacturing. A typical frog-leg robot uses a magnetic coupler to achieve contactless transmission of motion between its driving motors, which operate at atmospheric pressure, and its end effector (blade) which operates within a vacuum chamber. However, the magnetic coupler is a low-stiffness transmission element that induces residual vibration during fast motions of the robot. Excessive residual vibration can cause collisions between the fragile wafer carried by the robot and cassette, hence damaging the wafer. While this problem could be solved by slowing down the robot, it comes at the cost of reduced productivity, which is undesirable. Therefore, this paper reports a preliminary investigation into input shaping (a popular vibration compensation technique) as a tool to reduce residual vibration of a frog-leg robot during high-speed motions. Two types of motions of the robot are considered: rotation and extension. A standard input shaper is shown to be very effective for mitigating residual vibration caused by rotational motion but is much less effective for extensional motion. The rationale is that the resonance frequencies of the robot are constant during rotation but they vary significantly during extension, hence reducing the effectiveness of standard input shaping. This necessitates the use of more advanced input shapers that can handle varying resonance frequencies to mitigate residual vibration during extensional motion in future work.

The future of intelligent manufacturing machines involves autonomous selection of process parameters to maximize productivity while maintaining quality within specified constraints. To effectively optimize process parameters, these machines need to adapt to existing uncertainties in the physical system. This paper proposes a novel framework and methodology for feedrate optimization that is based on a physics-informed data-driven digital twin with quantified uncertainty. The servo dynamics are modeled using a digital twin, which incorporates the known uncertainty in the physics-based models and predicts the distribution of contour error using a data-driven model that learns the unknown uncertainty on-the-fly by sensor measurements. Using the quantified uncertainty, the proposed feedrate optimization maximizes productivity while maintaining quality under desired servo error constraints and stringency (i.e., the tolerance for constraint violation under uncertainty) using a model predictive control framework. Experimental results obtained using a 3-axis desktop CNC machine tool and a desktop 3D printer demonstrate significant cycle time reductions of up to 38% and 17% respectively, while staying close to the error tolerances compared to the existing methods.

The future of intelligent manufacturing machines involves autonomous selection of process parameters to maximize productivity while maintaining quality within specified constraints. To effectively optimize process parameters, these machines need to adapt to existing uncertainties in the physical system. This paper proposes a novel framework and methodology for feedrate optimization that is based on a physics-informed data-driven digital twin with quantified uncertainty. The servo dynamics are modeled using a digital twin, which incorporates the known uncertainty in the physics-based models and predicts the distribution of contour error using a data-driven model that learns the unknown uncertainty on-the-fly by sensor measurements. Using the quantified uncertainty, the proposed feedrate optimization maximizes productivity while maintaining quality under desired servo error constraints and stringency (i.e., the tolerance for constraint violation under uncertainty) using a model predictive control framework. Experimental results obtained using a 3-axis desktop CNC machine tool and a desktop 3D printer demonstrate significant cycle time reductions of up to 38% and 17% respectively, while staying close to the error tolerances compared to the existing methods.

The future of intelligent manufacturing machines involves autonomous selection of process parameters to maximize productivity while maintaining quality within specified constraints. To effectively optimize process parameters, these machines need to adapt to existing uncertainties in the physical system. This paper proposes a novel framework and methodology for feedrate optimization that is based on a physics-informed data-driven digital twin with quantified uncertainty. The servo dynamics are modeled using a digital twin, which incorporates the known uncertainty in the physics-based models and predicts the distribution of contour error using a data-driven model that learns the unknown uncertainty on-the-fly by sensor measurements. Using the quantified uncertainty, the proposed feedrate optimization maximizes productivity while maintaining quality under desired servo error constraints and stringency (i.e., the tolerance for constraint violation under uncertainty) using a model predictive control framework. Experimental results obtained using a 3-axis desktop CNC machine tool and a desktop 3D printer demonstrate significant cycle time reductions of up to 38% and 17% respectively, while staying close to the error tolerances compared to the existing methods.

Laser powder bed fusion (LPBF) is an additive manufacturing technique that is gaining popularity for producing metallic parts in various industries. However, parts produced by LPBF are prone to residual stress, deformation, cracks and other quality defects due to uneven temperature distribution during the LPBF process. To address this issue, in prior work, the authors have proposed SmartScan, a method for determining laser scan sequence in LPBF using a model-based and optimization-driven approach, rather than using heuristics, and applied it to simple 2D geometries. This paper presents a generalized SmartScan methodology that is applicable to arbitrary 3D geometries. This is achieved by: (1) expanding the thermal model and optimization approach used in SmartScan to multiple layers; (2) enabling SmartScan to process shapes with arbitrary contours and infill patterns within each layer and (3) providing SmartScan with global foresight to make it less myopic in its optimization. Sample 3D parts are printed using the proposed generalized SmartScan and compared to those printed using standard heuristic scan sequences. Reductions of up to 93% in temperature inhomogeneity, 87% in residual stress, and 26% in maximum deformation were observed, without significantly sacrificing print speed. However, SmartScan was found to cause minor (<5%) to significant (up to 20%) increases in surface roughness compared to the heuristic approaches, depending on the scan pattern used.

The future of intelligent manufacturing machines involves autonomous selection of process parameters to maximize productivity while maintaining quality within specified constraints. To effectively optimize process parameters, these machines need to adapt to existing uncertainties in the physical system. This paper proposes a novel framework and methodology for feedrate optimization that is based on a physics-informed data-driven digital twin with quantified uncertainty. The servo dynamics are modeled using a digital twin, which incorporates the known uncertainty in the physics-based models and predicts the distribution of contour error using a data-driven model that learns the unknown uncertainty on-the-fly by sensor measurements. Using the quantified uncertainty, the proposed feedrate optimization maximizes productivity while maintaining quality under desired servo error constraints and stringency (i.e., the tolerance for constraint violation under uncertainty) using a model predictive control framework. Experimental results obtained using a 3-axis desktop CNC machine tool and a desktop 3D printer demonstrate significant cycle time reductions of up to 38% and 17 respectively, while staying close to the error tolerances compared to the existing methods.