Layer-wise spatial modeling of porosity in additive manufacturing

Liu, Jia; Liu, Chenang; Bai, Yun; Rao, Prahalada; Williams, Christopher B.; Kong, Zhenyu

doi:10.1080/24725854.2018.1478169

Cited by 35 publications

(8 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There were more smaller particles in the recycled powder compared to its virgin counterpart probably due to the blowing of smaller particles away from the melt pool during processing, as for this a lower level of kinetic energy would be required [27]. Apart from the particle size distribution analysis, extensive testing was carried out using a variety of image processing methods for quantitative pore analysis of the powders [28,29,30]. Most literature reports on pore evaluation in manufactured parts (and not powders) using the XCT technique [18].…”

Section: X-ray Tomography Analysismentioning

confidence: 99%

“…Fig. 2 represents the image processing steps that were applied such as threshold adjustment (gray scale threshold), colored classifying, edge cutting, disconnecting the powders and some other manual image processing [18,28,29]. In the virgin powder specimen, pores of 5 m size contributes to 16.28% of the entire pore population.…”

Section: X-ray Tomography Analysismentioning

confidence: 99%

See 1 more Smart Citation

A new method for assessing the recyclability of powders within Powder Bed Fusion process

Gorji

Saxena

Corfield

et al. 2020

Materials Characterization

View full text Add to dashboard Cite

Recycling metallic powders used in the additive manufacturing (AM) process is essential for reducing the process cost, manufacturing time, energy consumption, and metallic waste. In this paper, the focus is on pore formation in recycled powder particles of stainless steel 316L during the selective laser melting process. We have introduced the concept of optimizing the powder bed's printing area in order to see the extent of the affected powders during the 3D-printing process. X-ray Computed Tomography (XCT) is used to characterize the pores inside the particles. The results from image processing of the tomography (rendered in 3D format) indicate a broader pore size distribution and a higher pore density in recycled powders compared to their virgin counterparts. To elucidate on this, the Electron Dispersion spectroscopy (EDX) analysis and Synchrotron-based Hard X-ray Photoelectron Spectroscopy (HAXPES) were performed to reveal the chemical composition distribution across the pore area and bulk of the recycled powder particles. Higher concentrations of Fe, Cr, and Ni were recorded on the interior wall of the pore in recycled particles and higher Mn, S and Si concentrations were recorded in the outer layer around the pore area and on the surface of the recycled particle. The pore formation in recycled powder is attributed to out-diffusion of Mn, S and Si to the outer surface as a result of the incident laser heat during the AM process due to higher electron affinity of such metallic elements to oxygenation. HAXPES analysis shows a higher MnO concentration around the pore area which impedes the in-diffusion of other elements into the bulk and thereby helps to creates a void. The inside wall of the pore area (dendrites), has a higher concentration of Fe and Cr oxide. We believe the higher pore density in recycled powders is due, at least in part to composition redistribution, promoted by laser heat during the AM process. Nanoindentation analyses on both virgin and recycled powder particles shows a lower hardness and higher effective modulus in the recycled powder particles attributed to the higher porosity in recycled powders.

show abstract

Section: X-ray Tomography Analysismentioning

confidence: 99%

Section: X-ray Tomography Analysismentioning

confidence: 99%

A new method for assessing the recyclability of powders within Powder Bed Fusion process

Gorji

Saxena

Corfield

et al. 2020

Materials Characterization

View full text Add to dashboard Cite

show abstract

“…Third, the artificial hollow area, such as the strut-based lattice design in the SLM, can also introduce disturbances in model estimation. It is because the characteristics of the pixel on hollow and voids areas are similar to each other on the layer-wise image [20].…”

Section: Introductionmentioning

confidence: 91%

Pyramid Ensemble Convolutional Neural Network for Virtual Computed Tomography Image Prediction in a Selective Laser Melting Process

Wang

Chen

Henkel³

et al. 2021

Journal of Manufacturing Science and Engineering

View full text Add to dashboard Cite

Additive manufacturing (AM) is a type of advanced manufacturing process that enables fast prototyping to realize personalized products in complex shapes. However, quality defects existed in AM products can directly lead to significant failures in practice. Thus, various inspection techniques have been investigated to evaluate the quality of AM products, where X-ray computed tomography serves as one of the most accurate techniques to detect defects. Taking a selective laser melting process (SLM) as an example, voids can be detected by investigating CT images after the fabrication of products. However, limited by the sensor size and scanning speed issue, CT is difficult to be used for online (i.e., layer-wise) voids detection, monitoring, and process control to mitigate the defects. As an alternative, optical cameras can provide layer-wise images to support online voids detection. The intricate texture of the layer-wise image restricts the accuracy of void detection in AM products. Therefore, we propose a new method called pyramid ensemble convolutional neural network to efficiently detect voids and predict the texture of CT images by using layer-wise optical images. The proposed PECNN can efficiently extract informative features based on the ensemble of the multiscale feature-maps from optical images. Unlike deterministic ensemble strategies, this ensemble strategy is optimized by training a neural network in a data-driven manner to learn the fine-grained information from the extracted feature-maps. The merits of the proposed method are illustrated by both simulations and a real case study in SLM.

show abstract

“…They also developed a regression model to predict the amplitude of the desired signal (dimensionless units) using nozzle temperature and magnetic field strength. Other existing methodologies are focused on tracking general process changes via statistical process control [6], in situ monitoring of melt pool images to understand microstructure formation [7], tracking porosity via thermal distribution [8], or modeling layer-wise spatial distribution [9]. However, process instability is difficult to eliminate, making it impossible to avoid porosity altogether.…”

Section: Introduction 1background Of Laser Metal Deposition and Chall...mentioning

confidence: 99%

A Physics-Informed Convolutional Neural Network with Custom Loss Functions for Porosity Prediction in Laser Metal Deposition

Gawade

Guo

2022

Sensors

View full text Add to dashboard Cite

Physics-informed machine learning is emerging through vast methodologies and in various applications. This paper discovers physics-based custom loss functions as an implementable solution to additive manufacturing (AM). Specifically, laser metal deposition (LMD) is an AM process where a laser beam melts deposited powder, and the dissolved particles fuse to produce metal components. Porosity, or small cavities that form in this printed structure, is generally considered one of the most destructive defects in metal AM. Traditionally, computer tomography scans measure porosity. While this is useful for understanding the nature of pore formation and its characteristics, purely physics-driven models lack real-time prediction ability. Meanwhile, a purely deep learning approach to porosity prediction leaves valuable physics knowledge behind. In this paper, a hybrid model that uses both empirical and simulated LMD data is created to show how various physics-informed loss functions impact the accuracy, precision, and recall of a baseline deep learning model for porosity prediction. In particular, some versions of the physics-informed model can improve the precision of the baseline deep learning-only model (albeit at the expense of overall accuracy).

show abstract

Layer-wise spatial modeling of porosity in additive manufacturing

Cited by 35 publications

References 44 publications

A new method for assessing the recyclability of powders within Powder Bed Fusion process

A new method for assessing the recyclability of powders within Powder Bed Fusion process

Pyramid Ensemble Convolutional Neural Network for Virtual Computed Tomography Image Prediction in a Selective Laser Melting Process

A Physics-Informed Convolutional Neural Network with Custom Loss Functions for Porosity Prediction in Laser Metal Deposition

Contact Info

Product

Resources

About