In-situ monitoring of selective laser melting process

• Main Author: Lu, Qingyang
• Other Authors: Tuan Tran
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Mechanical engineering
• Online Access: https://hdl.handle.net/10356/143157
• Institution: Nanyang Technological University

Additive manufacturing (AM) has progressively transited from being regarded as a tool for rapid prototyping to the efficient production of individually customised, and highly complex functional parts. AM of metals, in particular, is an emerging field of interest with exponential growth in metallic parts manufactured for end-use applications via laser melting processes. Powder bed fusion (PBF) and directed energy deposition are the main processes for metal printing. One of the most promising PBF processes is selective laser melting (SLM), which is capable of fabricating metal components near full density with little post-processing. While SLM allow design freedoms unequalled by conventional manufacturing methods, production techniques require strict control of processing parameters and conditions to build a metallic part of high quality. The printing process is vulnerable to defects generation and is inevitable. In-situ measurements can be utilised to optimise part-to-part reproducibility. The implementation of in-situ monitoring systems (IMSs) in AM printers render both time and cost savings as the process can be stopped if a failure is detected. A myriad of in-situ monitoring studies has been carried out focusing on identification of defects and irregularities during the printing process. Despite the advancement of IMSs in defects detection, the impact of these defects on the printed part’s structural integrity remains unknown. The effects of these defects identified during printing on the density and mechanical properties of the printed part have yet to be adequately investigated by researchers and industrial experts. This thesis focuses on the development of a self-developed in-situ monitoring system (IMS) and the establishment of relationships between features (i.e., process signatures) captured in images during fabrication and the mechanical properties of the printed parts. The establishment of a correlation between features captured in images during printing and the actual defects present in the printed parts has also been explored. The self-developed IMS proposed in this research has several advantages over existing commercial IMSs. It is portable as it is an independent system and does not require any modification of the printer’s existing optical components. This makes the IMS flexible for installation on any SLM printers. The self-developed IMS captures high-resolution optical and thermal images, allowing for easy identification and evaluation of features. Coupled with the fast and robust algorithm, real-time image-based measurements of features can be correlated with mechanical properties of the printed part. Simultaneous monitoring of the powder bed is possible through the use of optical and infrared sensors. It enables different types of defects to be detected during the printing process. The design of the proposed IMS allowed monitoring of the powder bed via optical and infrared cameras. Ten sets of 316L stainless steel specimens were fabricated by the SLM process. Each set of specimens consisted of a cylinder and a sub-size tensile coupon. The specimens were printed with varying energy densities. Features captured in optical images of the powder bed were examined. They are represented by pixels that are either brighter or darker than that of the metal powders. A relationship has been established between features captured in optical images during fabrication and the mechanical properties of the printed parts. Cylinders with percentage area of features greater than 2% exhibit lower densities, ultimate and yield strengths, and vice versa. These cylinders have densities lower than 7.82 gcm-3, ultimate strength lesser than 565 MPa and yield strength lesser than 462 MPa. The implementation of the IMS based on optical imaging enables end users to distinguish between specimens with relatively higher density and better mechanical properties and specimens with relatively lower density and weaker mechanical properties. A correlation has been established between features captured in optical images during printing and the actual defects present in the printed part. Micro computed tomography (CT) was used to validate the correlation of features in two-dimensional optical images to actual three-dimensional defects in the printed samples through reconstruction of the CT images, and more importantly, to justify the viability of using optical imaging as a solution for evaluating defects in printed parts. A positive correlation exists between the percentage area of features in each resultant optical image, , and the percentage area of defects in each resultant CT image, , for the ten cylinders. Their Pearson correlation coefficients range between 0.30 and 0.78. Similarities in trends between and were established across the layers of the cylinders, where a higher translates to higher . This enables end users to directly evaluate the area of defects present in parts based on real-time analysis of optical images, eliminating the need for long hours of defects detection in printed parts by CT. A relationship between regions with high normalised temperature gradients captured in thermal images and the densities of selectively laser melted parts has been demonstrated. Cylinders with more pixels of high normalised temperature gradients across their cross sections exhibit lower densities. This finding potentially aids in the fabrication of parts with improved density via thermal imaging. Based on the established relationships, the proposed IMS has a strong potential in aiding the production of parts with better mechanical properties. Printing parameters can be adjusted in-situ to improve the quality of the printed part if the number of features or thermal signatures present in optical or thermal images correlate to weak mechanical properties, rendering both time and cost savings. The results also illustrated the capability of the system in differentiating the extent of defects present in specimens fabricated with varying energy densities. The findings presented in this thesis also lay the foundation for the development of closed-loop IMSs associated with the SLM process.