Search

The use of industrial robots for machining large parts has attracted more and more attention. Previous studies have shown that ball-end milling is greatly affected by the process parameters. Besides, robotic machining is also affected by various posture-dependent robot performances. However, these two critical aspects are usually treated separately in many works for studying robotic machining. In this paper, a redundant robotic system consisting of a six-axis industrial robot, a two-axis positioner and a linear track was developed for machining. The combined effects of milling process parameters and robot postures on the machining results were experimentally investigated. Grey relational analysis-based multi-objective optimization was conducted for lower cutting force and better surface quality. A group of process parameters and robot posture was obtained as the optimal combination that generally yields the best milling performance. The results allow to preliminarily determine the main factors that affect the quality of robotic machining.

Cracks and keyhole pores are detrimental defects in alloys produced by laser directed energy deposition (LDED). Laser-material interaction sound may hold information about underlying complex physical events such as crack propagation and pores formation. However, due to the noisy environment and intricate signal content, acoustic-based monitoring in LDED has received little attention. This paper proposes a novel acoustic-based in-situ defect detection strategy in LDED. The key contribution of this study is to develop an in-situ acoustic signal denoising, feature extraction, and sound classification pipeline that incorporates convolutional neural networks (CNN) for online defect prediction. Microscope images are used to identify locations of the cracks and keyhole pores within a part. The defect locations are spatiotemporally registered with acoustic signal. Various acoustic features corresponding to defect-free regions, cracks, and keyhole pores are extracted and analysed in time-domain, frequency-domain, and time-frequency representations. The CNN model is trained to predict defect occurrences using the Mel-Frequency Cepstral Coefficients (MFCCs) of the lasermaterial interaction sound. The CNN model is compared to various classic machine learning models trained on the denoised acoustic dataset and raw acoustic dataset. The validation results shows that the CNN model trained on the denoised dataset outperforms others with the highest overall accuracy (89%), keyhole pore prediction accuracy (93%), and AUC-ROC score (98%). Furthermore, the trained CNN model can be deployed into an in-house developed software platform for online quality monitoring. The proposed strategy is the first study to use acoustic signals with deep learning for insitu defect detection in LDED process., 36 Pages, 16 Figures, accepted at journal Additive Manufacturing

Surface monitoring is an essential part of quality assurance for additive manufacturing (AM). Surface defects need to be identified early in the AM process to avoid further deterioration of the par...

Support structures are critical to the successful printing of the overhang structures in selective laser melting. The heat transfer performance of support structures has significant influence on the temperature distribution and cooling rate within the overhang structures which in turn determine the microstructure and residual stress. In the present study, functionally graded support structures have been proposed and their thermal performance has been numerically investigated, with the consideration of different materials, cooling times and gradedness values. It has been found that functionally graded support structures can maintain a higher temperature level than the conventional uniform support structure at the bottom of overhang, which is equivalent to an extra pre-heating effect. The temperature fluctuation and cooling rate at the bottom of overhang can also be reduced by adopting the functionally graded support structures.

The short article reviews the state-of-the-art of multi-axis toolpath planning, especially for polymer extrusion. The existing approaches for geometry-based and performance-based toolpath planning are emphasized and highlighted. The future perspectives of continuous fiber-reinforced composites are addressed to overcome the research gaps as well as improve the performance of products induced by polymer extrusion.

Optical laser head is a key component used to shape the laser beam and to deliver higher power laser irradiation onto workpieces for material processing. A focused laser beam size and optical intensity need to be controlled to avoid decreasing beam quality and loss of intensity in laser material processing. This paper reports the multiphysics modeling of an in-house developed laser head for laser-aided additive manufacturing (LAAM) applications. The design of computer experiments (DoCE) combined with the response surface model was used as an efficient design approach to optimize the optical performance of a high power LAAM head. A coupled structural-thermal-optical-performance (STOP) model was developed to evaluate the influence of thermal effects on the optical performance. A number of experiments with different laser powers, laser beam focal plane positions, and environmental settings were designed and simulated using the STOP model for sensitivity analysis. The response models of the optical performance were constructed using DoCE and regression analysis. Based on the response models, optimal design settings were predicted and validated with the simulations. The results show that the proposed design approach is effective in obtaining optimal solutions for optical performance of the laser head in LAAM. Published version

The classification and recognition of underwater target signal are based on the samples. In the practical field, the amount of available training samples is limited. In this paper, based on analyzing the radiated noise signals of two types of underwater targets, the Bayesian dynamic modification model for underwater acoustic signals is established, which follows the idea of virtual sample generation. According to idea of virtual sample generation, a Bootstrap-Bayesian dynamic modification model for the small-sample-size problem is proposed, based on Bootstrap method and Bayesian parameter estimation. Through the processing of experimental data and comparative analysis of the experimental results, the effectiveness of the feature extraction method and the dynamically modified classification model proposed in this paper are verified. The applicability of the classification model for two types of underwater acoustic signals has also been confirmed.

Vision-based surface monitoring is critical for surface defects and geometric distortion detection during additive manufacturing (AM) processes. It is the pre-requisite of quality assurance and process control to maintain and improve the quality of AM produced parts. However, current surface monitoring solutions are not efficient enough to provide in-situ monitoring feedback during the relatively long AM fabrication process, which has become a significant barrier to AM automation. This paper presents a new surface monitoring method integrated in a robot-based laser-aided additive manufacturing cell, with in-situ point cloud processing capability. During surface scanning, a multiprocessing technique is used to run both sensor data capturing and point cloud processing programs. Point cloud data captured within predefined time intervals are filtered and segmented automatically to extract the parts surface to be monitored. Experimental results are presented to verify the effectiveness of the proposed surface monitoring method.

A kinematical redundant robotic system is developed for laser aided additive manufacturing (AM). The eight-axis robotic system composed of a six-axis robot synchronizing with a two-axis positioner. For the developed robotic system, the post-processing of an AM path is a challenging problem due to the functional and intrinsic redundancies. A profuse kinematic model of the robotic system is required during the development of dedicated offline programming (OLP) software. In this study, a heuristic kinematic analysis is conducted for the robotic system to address the redundant problem considering the special requirements and constraints of the AM process. With the developed model, a postprocessor is programmed into an internally-developed OLP software package. An example is presented to show the effectiveness of the kinematic model and application of the OLP software.

Closed-loop control is desirable in direct energy deposition (DED) to stabilize the process and improve the fabrication quality. Most existing DED controllers require system identifications by experiments to obtain plant models or layer-dependent adaptive control rules, and such processes are cumbersome and time-consuming. This paper proposes a novel data-driven adaptive control strategy to adjust laser voltage with the melt pool size feedback. A multitasking controller architecture is developed to incorporate an autotuning unit that optimizes controller parameters based on the DED process data automatically. Experimental validations show improvements in the geometric accuracy and melt pool consistency of controlled samples. The main advantage of the proposed controller is that it can adapt to DED processes with different part shapes, materials, tool paths, and process parameters without tweaking. System identification is not required even when process conditions are changed, which reduces the controller implementation time and cost for end-users. Agency for Science, Technology and Research (A*STAR) Published version This research was funded by A*ccelerate, grant number ACCL/19-GAP077-R20A.

Additive manufacturing (AM) technologies provide more freedom to functional part design in various industries. One of the unique capabilities of AM is that multi-material parts can be produced with material compositional and geometric complexity. Multi-material parts have the advantage of achieving multiple performance requirements. In the research, we propose a framework for designing multi-material parts using AM processes. The proposed framework is composed of four interacting modules, including design requirement identification, primary material selection, AM process selection, material composition, and part geometry determination. Rules and guidelines for AM are integrated into the proposed framework with AM processes’ capabilities and constraints compiled in databases. We also introduce databases to assist in decision-making and ensure manufacturability of the designed multi-material part in various product design phases. The proposed framework is applied to a case study involving a conceptual design of a multi-material battery pack cooling plate. NRF (Natl Research Foundation, S’pore) ASTAR (Agency for Sci., Tech. and Research, S’pore) Accepted version

Laser aided additive manufacturing (LAAM), a blown powder additive manufacturing process, can be widely adopted for surface modification, repair and 3D printing. A robust numerical model was developed to simulate convective fluid flow and balancing of surface tension forces at the air-fluid interface to predict melt-pool free surface curvature and solidified clad dimensions. The free surface physical interface was calculated using the Arbitrary Lagrangian Eulerian (ALE) moving mesh approach. Powder deposition efficiency was considered by activating mesh normal velocity at melted regions based on localized powder mass flux intensity from the discrete coaxial powder nozzles. The heat flux equation used for representing the laser heat source considered attenuation effect from the interaction between the powder jets and laser as well as heat sink effects of un-melted powder particles entering the melt-pool. The predicted thermal gradient directions agree well with grain growth orientations obtained from electron backscatter diffraction (ESBD) analysis in three different cross-sectional orientations. Experimental validation of clad width, height and melt-pool depth shows a maximum error of 10% for a wide range of processing parameters which consider the effects of varying laser power, laser scanning speed and powder feeding rate. Keywords: Laser aided additive manufacturing, Numerical simulation, Temperature gradient, Fluid dynamics, Microstructure

In this paper, we report that a selective laser melted (SLM) AlSi10Mg alloy has unique hierarchical microstructures. There are micro-sized Al grains and transgranular ultrafine cellular structures around primary Al, and the cell boundary is made up of alternate eutectic Si and eutectic Al phases. The SLM AlSi10Mg alloy has both high tensile strength and high strain-hardening capability, which are superior to Al-Si alloys fabricated by conventional powder metallurgy and cast methods. Post-tensile high-resolution microscopy analyses provide further evidence supporting Orowan looping as a strengthening mechanism of the SLM Al-Si alloy.

The CoCrNi medium entropy alloy with different amounts of oxides was fabricated by laser aided additive manufacturing (LAAM). The cryogenic tensile properties and microstructure evolution during tensile deformation were investigated. For the Sample B with higher oxide content (2.03 vol%), the yield strength (YS), ultimate tensile strength (UTS) and elongation (El) were all inferior to those of the Sample A with much fewer oxides (0.47 vol%). The lower YS of Sample B was mainly attributed to the lower initial dislocation density. The oxides contributed slightly to the increase in YS, while reducing the El significantly. The El of Sample B was comparable at 298 K and 143 K, owing to the compensation effect from twin boundaries (TBs). Higher YS and UTS were obtained at 143 K for both samples. With decreasing temperature from 298 K to 143 K, the YS and UTS of Sample A increased almost linearly, whereas the El decreased. Though a large amount of TBs were formed during the tensile deformation, they were unevenly distributed among the grains near the fractured location. Under higher stress at cryogenic temperature, the interaction of grain boundaries with massive TBs caused micro-voids to initiate more readily along the grain boundaries, resulting in premature failure.

In the robotic machining process, the external force due to cutter-workpiece interactions and gravity force due to link weights can cause considerable deviations of the desired trajectory. Besides, gravity compensators can be found in many heavy-duty industrial robots to balance the gravity force. However, the effects of gravity force and balancing force are rarely considered in the deflection analysis. This paper proposes an effective method for the stiffness modeling of heavy-duty industrial robots by taking account of the joint/link compliances, link weights and gravity compensator. Firstly, the static equilibrium equations of each substructure are formulated to derive the expressions of reaction force exerted on each joint. Secondly, a linear map of linear/angular deflections between the joints/links and the end-effector is developed. Finally, the stiffness model of the whole robot is formulated by integrating the joint/link compliances. The robot deflections at several configurations are analyzed, and the deflection distributions throughout the workspace are demonstrated. With this model, it is possible to analytically evaluate the contributions of external force, link weights and gravity compensator to the total deflections. The correctness of the proposed model is also experimentally verified by comparing the calculated and the measured deflections under the same conditions.

Purpose This paper aims to present a hybrid machine learning algorithm for additive manufacturing (AM) design feature recommendation during the conceptual design phase. Design/methodology/approach In the proposed hybrid machine learning algorithm, hierarchical clustering is performed on coded AM design features and target components, resulting in a dendrogram. Existing industrial application examples are used to train a supervised classifier that determines the final sub-cluster within the dendrogram containing the recommended AM design features. Findings Through a case study of designing additive manufactured R/C car components, the proposed hybrid machine learning method was proven useful in providing feasible conceptual design solutions for inexperienced designers by recommending appropriate AM design features. Originality/value The proposed method helps inexperienced designers who are newly exposed to AM capabilities explore and utilize AM design knowledge computationally.

In this study, the CoCrNi medium entropy alloy (MEA) was successfully fabricated by laser aided additive manufacturing (LAAM), a directed energy deposition (DED) process. Results indicate that the ultimate tensile strength (UTS), yield strength (YS), and elongation after fracture (δF) are 873.5 MPa, 620.5 MPa, and 44.8 % respectively. Without any post heat treatment, the as-built state CoCrNi MEA exhibits an excellent combination of strength and ductility. The high YS is mainly attributed to the combination effects from lattice friction stress, boundary strengthening, and dislocation strengthening. Deformation twinning is activated during room temperature tensile deformation, inducing the steady work hardening behavior. Hence, fracture is postponed to higher deformation strain due to the synergistic effect of dislocation gliding and deformation twinning. LAAM is expected to be a versatile and efficient method to fabricate CoCrNi MEA and some other high-performance alloys.

Powder bed fusion (PBF) represents a class of additive manufacturing processes with the unique advantage of being able to fabricate functional products with complex three-dimensional geometries. PBF has been broadly applied in highly value-added industries, including the biomedical device and aerospace industries. However, it is challenging to construct a comprehensive knowledgebase to guide material selection and process optimization decisions to satisfy the product standards of various industries based on a poor understanding of process-structure-property/performance relationships for each type of thermoplastic. In this paper, an intelligent optimization system is proposed to establish quantitative relationships between process parameters and multiple optimization objectives, including mechanical properties, productivity, energy efficiency, and degree of material degradation. Polyurethane is considered as a representative thermoplastic because it is sensitive to thermal-induced degradation and has a relatively narrow process window. Material and powder properties as functions of temperature are investigated using systematic material screening. Numerical models are created to analyze the interactions between laser beams and polymeric powders by considering the effects of chamber thermal conditions, laser parameters, temperature-dependent properties, and phase transitions of polymers, as well as laser beam characteristics. The theoretically predicted features of melting pools are validated experimentally and then utilized to develop quantitative relationships between process parameters and multiple optimization objectives. The established relationships can guide process parameter optimization and material selection decisions for polymer PBF.

IN100 Ni-based superalloy fabricated by micro-laser aided additive manufacturing (micro-LAAM) was investigated in this study. After solution treatment and aging (STA) of the micro-LAAMed IN100 alloy, hierarchical γ′ phases were recognized and characterized, which contributed significantly to the high ultimate tensile strength (~1050 MPa) and acceptable ductility (5%) at 25 °C and 600 °C. The tensile fractured surfaces of the as-built and STAed IN100 were characterized by dimples and dimples/cleavages, respectively. The high cycle vibration fatigue (HCVF) behavior was preliminarily studied through simulating the service conditions of some cantilever structures in automobile and aerospace industries. Compared with the cast IN100, the micro-LAAMed IN100 superalloys (as-built or STAed) both exhibited inferior HCVF lives. The HCVF behavior was discussed and correlated with the microstructure characteristics, such as the preferred growth direction of the grains in micro-LAAMed IN100 and the massive interfaces existing in the final obtained material. In the present study, the hierarchical γ′ phases were beneficial to the static tensile property, whereas detrimental to the HCVF behavior of the final obtained IN100 to some extent.

The microstructural and mechanical properties of AlSi10Mg alloys fabricated by additive manufacturing (AM) and powder metallurgy (PM) routes were investigated and compared. The microstructures were examined by scanning electron microscopy assisted with electron-dispersive spectroscopy. The crystalline features were studied by x-ray diffraction and electron backscatter diffraction. Room-temperature tensile tests and Vickers hardness measurements were performed to characterize the mechanical properties. It was found that the AM alloy had coarser Al grains but much finer Si precipitates compared with the PM alloy. Consequently, the AM alloy showed more than 100% increment in strength and hardness compared with the PM alloy due to the presence of ultrafine forms of Si, while exhibiting moderate ductility.

Challenges exist in developing new product family design methodologies that can utilize the additive manufacturing (AM) advantages to improve the product family competitiveness in a cost efficient way. In this research, the author attempts to compensate the performance compromise in conventional product families by proposing the novel concept of an additive manufactured variable platform (VP); and the major objective of this thesis is to develop a design framework that helps designers implement additive manufactured VP modules within product families. During the development of the design framework, the following methodologies have been proposed: 1) a representative mathematical model to formulated the VP-based product family configuration; 2) a hybrid machine learning approach to select appropriate AM design features for addtive manufactured VP modules; 3) a fuzzy time-driven activity-based costing method to predict the AM production cost variation as a function of VP design adjustments, 4) a novel evaluation metric for VP-based product family designs, and 5) a multiobjective design optimization technique to identify the modular configuration and engineering design parameters of a product family. A case study of an R/C racing car family design is used throughout this thesis to demonstrate the usefulness of the proposed methodology. The implementation of additive manufactured VP modules shows a significant improvement in a product family’s performance at the price of a slight decrease in design commonality. As the major contribution of this research, the proposed design framework provides designers and enterprises with a guideline in developing high-performance and cost-efficient product families by utilizing AM-enabled design flexibility. The two previously separate research domains, i.e. AM and product family design, are integrated using the proposed framework and the constituent methodologies. Doctor of Philosophy (MAE)

Selective laser melting (SLM) is a powder-bed fusion additive manufacturing process that fabricates metallic parts in a layer-by-layer manner. IN718 is a nickel-based austenitic (γ) superalloy with broad applications in aerospace, chemical, and nuclear industries. In this study, fully-dense IN718-matrix composite reinforced by TiC nanoparticles were fabricated in-situ by SLM. As-built and heattreated IN718/TiC nanocomposite was compared against pure IN718 in terms of microstructures and tensile properties. Columnar and cellular microstructure morphologies in as-built pure IN718 and IN718/TiC nanocomposite were demolished by heat treatment. Microstructure refining and grain boundary dislocation pinning effect of TiC nanoparticles improved the tensile strength of as-built IN718/TiC nanocomposite. In both as-built IN718/TiC and pure IN718, γ and Laves phase are the major constituents while the other precipitates were suppressed by rapid solidification in SLM. Heat treatment dissolved Laves precipitates and grew γ'/γ'', δ, and MC/ M23C6 carbides that strengthened both pure IN718 and IN718/TiC nanocomposite. In heat-treated IN718/TiC, the TiC nanoparticles and M23C6 carbides degenerated from TiC counteracted the deficiency of δ precipitates. Therefore, heat-treated IN718/TiC nanocomposite had similar tensile strength but lower ductility than heat-treated pure IN718. NRF (Natl Research Foundation, S’pore) ASTAR (Agency for Sci., Tech. and Research, S’pore) Accepted version

Additive manufacturing (AM) has evolved from prototyping to functional part fabrication for a wide range of applications. Challenges exist in developing new product design methodologies to utilize AM-enabled design freedoms while limiting costs at the same time. When major design changes are made to a part, undesired high cost increments may be incurred due to significant adjustments of AM process settings. In this research, we introduce the concept of an additive manufactured variable product platform and its associated process setting platform. Design and process setting adjustments based on a reference part are constrained within a bounded feasible space (FS) in order to limit cost increments. In this paper, we develop a cost-driven design methodology for product families implemented with additive manufactured variable platforms. A fuzzy time-driven activity-based costing (FTDABC) approach is introduced to estimate AM production costs based on process settings. Time equations in the FTDABC are computed in a trained adaptive neuro-fuzzy inference system (ANFIS). The process setting adjustment's FS boundary is identified by solving a multi-objective optimization problem. Variable platform design parameter limitations are computed in a Mamdani-type expert system, and then used as constraints in the design optimization to maximize customer perceived utility. Case studies on designing an R/C racing car family illustrate the proposed methodology and demonstrate that the optimized additive manufactured variable platforms can improve product performances at lower costs than conventional consistent platform-based design. ASTAR (Agency for Sci., Tech. and Research, S’pore) MOE (Min. of Education, S’pore)

Additive manufacturing (AM) is projected to have a profound impact on mass customisation. In order to benefit from this new technology, we need to incorporate AM into design processes. This paper addresses that need by introducing an AM process model for product family design. The proposed model reflects the ability of AM to produce customised and complex parts without tooling efforts. By utilising AM, we eliminate all constraints which arise in conventional product family designs from finding a compromise between commonality and performance. The proposed model starts by identifying design requirements and constraints. Subsequently, we use topology optimisation to determine an optimal design for each product. Next, finite element analysis and cost analysis are performed. We combine the analysis results in one three-dimensional plot, which displays the merits of the individual component realisations. Thus, a fair and competitive component evaluation is possible and the most suitable product family design can be selected. The final designs are fabricated using fused desposition modelling. A case study is conducted to illustrate how the proposed model facilitates the benefits of AM. The results show that the proposed model has the potential to provide affordable customisation. MOE (Min. of Education, S’pore) Accepted version

The purpose of this paper is to investigate the effects of TiC nanoparticle content on microstructures and tensile properties of the IN718/TiC nanocomposites fabricated by selective laser melting (SLM). 0.5wt%, 1.0wt%, and 2.0wt% of TiC nanoparticles are added to the IN718 powders. The bulk-form IN718/TiC nanocomposites with different TiC contents are fabricated in-situ by SLM using the same process settings. The evolution of microstructures and tensile properties as the effect of changing the TiC content is studied using the optical microscopy, scanning electron microscopy, X-ray diffraction analysis, and tensile testing. The increase of TiC content refines the microstructure, promotes the formation of the cellular morphology, and reduces the size and continuity of Laves precipitates. Increasing the TiC content improves the yield strength and ultimate tensile strength but decreases the ductility. The grain refinement, dislocation bowing, dislocation punching, and the reduction in Laves precipitate contribute to the strengthening effect in the IN718/TiC nanocomposites. NRF (Natl Research Foundation, S’pore) Published version

Additive manufacturing (AM) has evolved from prototyping to functional part fabrication for a wide range of applications. AM process settings have significant impact to both part quality and production cost, which makes the process setting adjustment a key consideration during product development and manufacturing. This research aims to investigate the relationship among process setting adjustments, costs, and component design parameters. Platform-based product family design and process family planning are used in this research as the strategy to provide product diversity while controlling cost. In this paper, the concept of a variable product platform and its corresponding AM process setting variants are proposed to describe the characteristics of additive manufactured platform modules. AM production cost drivers are identified. A Fuzzy Time-Driven Activity-Based Costing (FTDABC) approach is proposed to estimate the cost increment due to process setting adjustments. Time equations in the FTDABC are computed in a trained Adaptive Neuro-Fuzzy Inference System (ANFIS). The process setting adjustment’s feasible space boundary searching is formulated as an optimization problem, with minimizing the cost increment and maximizing the design parameters’ variability as objective functions. The upper and lower limits of variable platform module’s design parameters are mapped from process setting adjustments in a Mamdani-type expert system. The proposed methodology is illustrated in the analysis of a honeycomb-shaped bumper, which is taken as a variable platform module for a family of R/C racing cars. The result provides boundaries for design parameters, which confines the AM-enabled design space for product platform modules.Copyright © 2015 by ASME

Additive manufacturing (AM) techniques are ideal for producing customized products due to their high design flexibility. Despite the previous studies on specific additive manufactured customized products such as biomedical implants and prostheses, the simultaneous optimization of components, materials, AM processes, and dimensions remains a challenge. Multidisciplinary design optimization (MDO) is a research area of solving complex design problems involving multiple disciplines which usually interact with each other. The objective of this research is to formulate and solve an MDO problem in the development of additive manufactured products customized for various customers in different market segments. Three disciplines, i.e. the customer preference modeling, AM production costing, and structural mechanics are incorporated in the MDO problem. The optimal selections of components, materials, AM processes, and dimensional parameters are searched with the objectives to maximize the functionality utility, match individual customers' personal performance requirements, and minimize the total cost. A multi-objective genetic algorithm with the proposed chromosome encoding pattern is applied to solve the MDO problem. A case study of designing customized trans-tibial prostheses with additive manufactured components is presented to illustrate the proposed MDO method. Clusters of multi-dimensional Pareto-optimal design solutions are obtained from the MDO, showing trade-offs among the objectives. Appropriate design decision can be chosen from the clusters based on the manufacturer's market strategy. Highlights An optimization problem for additive manufactured customized products is solved. Three disciplines are incorporated in multidisciplinary design optimization (MDO). The selection of component, material, additive manufacturing process and dimensions are optimized. A multiobjective genetic algorithm is applied to solve the MDO problem. Pareto-optimal solutions with different utilities and costs are obtained.