Experimental Investigations and Multi-Objective Optimization of Selective Inhibition Sintering Process Using the Dragonfly Algorithm

Introduction

Additive manufacturing (AM) is a new class of layered manufacturing technique used for fabricating intricate customized components in short span, manifest automation and significantly reduced manufacturing cost (Rajamani & Balasubramanian, 2019a). Therefore, AM can be effectively used in fabricating net shaped components in various fields such as aerospace, automotive, bio-medical and defence industries, etc. (Rajamani & Balasubramanian, 2019b). In the past few decades, several AM techniques such as stereolithography, fused deposition modeling (FDM), selective laser sintering (SLS), laser engineered net shaping (LENS), etc., have been developed to fabricate different parts in various applications.

Each AM technique has its unique features such as processing capabilities, method of fabrication and diversity of raw materials. Among all these techniques, powder-based AM processes are recently used in versatile applications of medium-to-high volume series production (Esakki et al., 2017). Direct metal laser sintering (DMLS), three-dimensional printing (3DP) and Selective laser sintering (SLS) are the most-widely used powder-based AM techniques in fabricating plastic, metal and ceramic parts. The utilization of high-cost heating mechanisms such as laser and electron lead the machine and processing cost. Eliminating laser and electron heating elements in AM processes has incredible impression on reducing the machine cost and speed of the process increased.

In view of this, high speed sintering (HSS) and selective inhibition sintering (SIS) processes are evolved to significantly reduce the processing cost by eliminating the costlier heating elements. HSS involves cost-effective infra-red heater to sinter the powder particles. The heat is transferred to the powder surface through a radiation absorbing material (RAM), which absorbs the as-received heat radiation from infra-red heater and transfer it for achieving effective sintering (Majewski et al., 2008). However, the incidence of RAM and more consumption of polymer powder material are foremost challenges to accustom HSS.

To overcome this issue, SIS system was built at University of Southern California, USA (Khoshnevis et al., 2003). In the SIS process, powder particles are sintered with desired part peripheries which is defined through precise delivery of inhibitors. The SIS has the key advantages of eliminating expensive tooling and support structure, processing various indigenously available raw materials such as polymer, metal and ceramics which makes the process cost-effective.

Among the several advantages, SIS has few drawbacks such as compatibility issues with NC tool path generation, compaction of powder particles, effective usage of raw materials and improving the quality characteristics of sintered parts such as strength, surface quality, dimensional stability, etc. (Rajamani, Balasubramanian, Arunkumar et al, 2018). The quality and performance features of the SIS parts can be improved by appropriate selection of process parameters. Several SIS process parameters such as thickness of powder layer, supplied heat energy, part bed temperature, feedrate of heater and inhibitor printer and particle size are having prominent influence on aforementioned functional qualities of sintered parts (Asiabanpour et al., 2007). Hence, it is necessary to investigate the influence of process parameters and identifying optimal values of parameters to improve the quality of sintered parts is very much essential.

Presently various modeling and optimization techniques such as statistical (desirability, GRA and TOPSIS) and metaheuristic approaches (GA, SA, PSO, ABC, etc.,) have been used to enhance the performance of AM process and quality of fabricated parts (Boschetto et al., 2011; Calignano et al., 2012; Gholaminezhad et al., 2016; Mahapatra & Sood, 2011; Raju et al., 2018; Strano et al., 2011). However, statistical optimization techniques solutions are often discrete combinations of profound ranges of process parameters and they may fall on local optima (Javed et al., 2018). Therefore, researchers are keenly looking for appropriate optimization technique to obtain global optimal solutions. Although, several researchers have attempted the hybridization of optimization techniques to arrive at optimal process parameters (Panda et al., 2014; Peng et al., 2014; Sood et al., 2009; Vijayaraghavan et al., 2014).