• Open access free of charge
  • Free and professional English polishing
  • Free and high quality figure editing
  • Free widest possible global promotion for your research

[Featured Article] Defect inspection technologies for additive manufacturing

  • Share:
Release Date: 2021-03-16 Visited: 

REVIEW ● OPEN ACCESSRead More

1. Introduction

Additive manufacturing (AM) technology is considered one of the most promising manufacturing technologies in the aerospace and defense industries. However, AM components are known to have various internal defects, such as powder agglomeration, balling, porosity, internal cracks and thermal/internal stress, which significantly affect the quality, mechanical properties and safety of final parts. Therefore, defect inspection methods are important for reducing manufactured defects, improving the surface quality and mechanical properties of AM components. Methods of detecting metal additive manufacturing can be divided into traditional non-destructive defect detection technology and defect detection technology based on machine learning. For the first type of detection technology, the monitoring of materials in the additive manufacturing process focuses on the abnormal phenomena of the materials by detecting the characteristic quantities, and has a certain predictive effect on the occurrence of defects. With the development of defect detection technologies, non-destructive techniques have evolved from infrared imaging defect detection, penetrant testing and eddy current method to ultrasonic inspection and X-ray testing. Recently, defect detection by machine learning has emerged as a technology that uses advanced equipment and deep learning methods to conduct in-process imaging for defect identification. However, machine learning detection technology is still in its preliminary stage in the field of additive manufacturing. Research teams by Prof. Lingbao Kong from Fudan University and by Prof. Richard Leach present a review on defect inspection technologies for additive manufacturing in IJEM. In this article, the authors have described defect inspection technologies with their application in AM processes. The architecture of defects in AM processes is reviewed. Traditional defect detection technology and the surface defect detection methods based on deep learning are summarized, and future aspects are suggested.

Highlights

The purpose of this study is to summarize the different categories of defects and defects inspection technologies in AM processes by introducing various defect detection methodologies in the application:

● A detailed description of different defects that appear in general AM processes are presented, and different traditional defect inspection technologies are summarized.

● Different machine learning defect detection technologies are also discussed and several latest research of defect detection technologies are presented.

● The future prospects of defect detection technology are also discussed, including industrial characteristics of additive manufacturing from the analysis and improvement of data sets, image acquisition methods, and detection scheme design, etc.

2. Background

At the start of the development of additive manufacturing (AM) technology in the 1990s when AM was called "rapid prototyping technology", researchers attempted to prepare non-metallic parts based on various rapid prototyping manufacturing methods. After that, the preparation of metal parts was realized through subsequent processes. Compared with traditional metal parts manufacturing technologies such as forging machining, forging, and welding, AM technology has the advantages such as no need for tools and molds, high material utilization, short product manufacturing cycle, and the easy realization of complex structure manufacturing. Although the development of AM technology has been relatively successful at attaining sufficient mechanical properties, actual component adoption in the industry is still limited by the defects and geometric accuracy. Some defects that appear in general AM processes are represented as cracking, residual stresses, porosity and balling. Furthermore, as defects often occur in the built components due to discontinuities originating during the generation process, the development of novel defect detection technologies has recently advanced in the AM field. In this article, Prof. Kong et al. provided a detailed introduction to different traditional detection technology and machine learning detection technology. Meanwhile, some latest research is presented and the future prospects of defect inspection technology are also proposed in this paper.

3. Recent Advances

AM systems work on the principal of building a structure additively from a substrate, while each method has advantages and disadvantages. According to the ASTM standard, additive manufacturing processes can be divided into seven categories: Binder Jetting, Material Extrusion, VAT Photo-polymerization, Material Jetting, Sheet Lamination, Directed Energy Deposition (DED), and powder bed fusion (PBF), which are able to handle various materials including polymers, metals, ceramics and composites. Various material discontinuities occur during the AM processes, and different defects may appear in the process of additive manufacturing. Although some post-processing techniques can reduce or eliminate defects in AM processing parts, improving the quality of parts inspection parts is important to meet the challenging of high-end industrial requirements.

Introduction to AM defects categories

In AM processes, the temperature of the metal powder varies considerably and thermal stresses easily form within the component, causing significant uncertainty with regards to final quality. When the stresses trapped inside the component are suddenly released, cracks emerge on the surface, affecting the performance and life of the component. As shown in Figure 1(a), porosity is a common phenomenon in both PBF and DED processes, and the pores also originate from lack of fusion in addition to the trapped gas, which can directly affect the density and mechanical properties of finished components and determine their performance, as shown in Figure 1(b). At the same time, melt ball formation, a.k.a. balling, occurs when the molten material solidifies into spheres instead of solid layers, which is a severe impediment to interlayer connection, as shown in Figure 1(c).

Fig. 1 AM defects: (a) The cracking phenomenon in the selective laser melting manufactured part; (b) Representative defect types in batches SLM-1b; (c) The balling phenomenon under particle sizes 0.05 mm.

Traditional non-destructive defect detection technology

For different types of defects, the traditional non-destructive defect detection methods mainly include: infrared imaging defect detection, penetration defect detection, eddy current defect detection and ultrasonic defect detection. The principle of infrared thermal imaging technology is mainly to display the shape and contour of defects through the thermal radiation intensity of the additively manufactured workpiece. Penetration defect detection uses capillary phenomena to inspect the surface defects of materials as shown in Figure 2. Eddy current testing uses the principle of electromagnetic induction to non-destructively evaluate certain properties of conductive materials by measuring the changes in induced eddy currents in the tested workpiece to find defects, as shown in Figure 3. Ultrasonic testing uses ultrasonic waves to inspect internal defects of metal components, as shown in the Figure 4.

Fig. 2 Results of PT of AM structure for aerospace: (a) Sample of rocket gas injector; (b) POGO–Z bezel.

Fig. 3 Eddy current detection results.

Fig. 4 B-scans of defects with different depths: (a) Defect depth 0.5mm; (b) Defect depth 0.1mm.

Defect detection technology based on machine learning

In recent years, machine learning has developed rapidly, and great progress has been made in industrial quality inspection fields. Based on the powerful learning ability and feature extraction of deep learning, many researchers have applied this technology to detect defects using machine learning and improve the overall detection efficiency and quality, such as Convolutional Neural Network, Auto-encode Network, Deep Residual Neural Network, Recurrent Neural Network, etc. Detection technologies by machine learning are widely used since they have the least influence on the AM part, an increased detection efficiency and are also highly automated compared to traditional inspection technologies.

4. Perspectives

Additive Manufacturing (AM) technology is considered as one of the most promising manufacturing technologies. Although the development of AM technology has been successfully applied to attain sufficient mechanical properties, actual components in the industry are still limited by the defects and geometric accuracy. At the same time, most of the existing defect detection methods can only be used in traditional industrial processing and manufacturing fields, and cannot replace the detection data and detection methods collected in the additive manufacturing. Therefore, further more defect detection technologies should be explored. Defect detection methods by machine learning based on the characteristics of additive manufacturing processes, with the methodology and experience of traditional machining defect detection, generate the most suitable defect detection approaches for additive manufacturing.

5. About the Authors

Prof. Lingbao Kong, PhD, Research Professor, has been engaged in ultra-precision manufacturing and metrology for many years. His research interests include ultra-precision intelligent manufacturing, freeform machining and measurement, machine learning, multiple spectrum inspection and data fusion, etc. He has been involved in many national key R&D and NSFC projects including Science Challenge Program as PIs or Co-Is. He has published more than 150 research papers, and got many granted patents and national & international awards.

Ms. Yao Chen was born in Liaoning, China, in 1995. She received the Bachelor degree from Dalian Polytechnic University, Liaoning, China, in 2018. She is currently pursuing the Master degree from Fudan University, Shanghai, China. Her research interests mainly focus on image processing, defect detection, multi-sensor data fusion algorithms.

  • Share: