Page 74 - MSAM-1-1

Materials Science in Additive Manufacturing A ML model for AM PSP of Ti64

1. Introduction Kalidindi (2015) summarized the data-based methods
that related to the accelerated development of advanced
Metal additive manufacturing (AM) technologies provide hierarchical materials . However, this research reveals
[4]
a flexible, efficient, and rapid means to fabricate complex that the correlation functions and physical descriptors
and customized products through a layer-by-layer require high computational cost; hence, a low-dimensional
approach. Two main categories of metal AM technologies method needed to be explored in the PSP linkage
predominantly used in industrial applications are powder framework. Popova et al. (2017) investigated the AM
bed fusion (PBF) and directed energy deposition (DED). process parameters and microstructure PSP correlations
These processes constitute over 90% of production-grade [5]
metal AM systems installed worldwide. PBF can be divided based on a reduced-ordered ML method . Their
into two categories depending on the energy source and research developed process-structure (P-S) linkage with
processing conditions. Electron beam PBF (EB-PBF) a low-dimensional representative of AM heterogeneous
microstructure. However, on the other side, metal AM
requires a vacuum environment, and laser PBF (L-PBF)
operates under an inert atmosphere. In addition, Frazier parts’ mechanical and machining behavior had not been
(2014) highlighted the major differences in the cooling rates investigated. Greitemeier et al. (2015) noticed that the
across metal AM processes ranging from 10 K/s in DED chemical composition, microstructure, and mechanical
3
to 10 –10 K/s in L-PBF and EB-PBF. Interactions between properties in AM Ti-6Al-4V are highly dependent on the
4
6
[6]
as-AM, post-processing conditions, and final properties AM processes . Furthermore, the uncertainty in metal
have always been of interest, since different cooling AM parts’ material properties and mechanical properties
rates lead to varying material structures and mechanical need to be understood. Hence, Markl and Körner (2016)
[1]
properties . Trelewicz et al. (2016) found that inherent declared that it is critical to establish a novel modeling
differences in metal AM processing conditions directly framework based on data science and numerical methods
impact the cooling rate and thermal gradient during the that can bridge to a critical knowledge gap between
build, resulting in highly heterogeneous and AM process- process, material microstructure, and material behavior to
specific dominant textured microstructure (e.g., dominant overcome the current limitations due to uncertainties in
[7]
columnar microstructure and microsegregation) . PSP linkages . The present study aims to develop a novel
[2]
PSP ML model for metal Ti-6Al-4V AM alloys that can
In metal AM production cycles, post-processing steps accurately predict material behavior in post-processing,
such as heat treatment and machining are often necessary that is, machining behavior.
to achieve desired material properties, tolerances, and
surface finish. However, it should be noted that the unique Li et al. (2020) clarified that titanium alloys are widely
heterogeneous microstructure and resulting mechanical used in multiple mechanical, aerospace, and biomedical
properties of metal AM also significantly affect their applications due to their high strength-to-density ratio and
[8]
machinability. Hence, it is critical to investigate this excellent corrosion resistance . However, titanium alloys
complex interaction between grain morphology (size, are often difficult to cast and process through subtractive
density, orientation, residual stress, and phase fraction) machining (e.g., strain hardening). Liu and Shin (2019)
of metal AM parts and resulting material behavior, that showed that metal AM technologies provide an alternative
is, specific cutting power during machining, which is of near-net-shape fabrication capability that allows for
interest in this study. The goal of this study is to establish a titanium product manufacturing to become relatively more
[9]
validated PSP linkage that could be extended to other AM cost effective for high-performance applications . Hence,
material properties, such as corrosion behavior, wear, and to better understand the titanium alloys performance in the
mechanical strength. AM field, a PSP linkage is required to reveal the structure
and machining behavior correlation of AM titanium alloys.
A low-dimensional process-structure-property (PSP)
linkage that captures the effects of processing conditions on To build the PSP linkages, descriptors are critical
critical material structure that ultimately affects material components to the necessary data science-based ML
behavior has always been of interest to the research and approaches. These descriptors quantitatively represent
manufacturing communities. Bostanabad et al. (2016) recorded information on material microstructure using
provided a stochastic microstructure characterization statistical methods and appear as correlation functions,
reconstruction method using supervised machine learning physical descriptors, and spectral density functions for
(ML) . The microstructure reconstruction method a given grain morphology. Corson (1974) described the
[3]
in their research indicates that the correlation feature 2-point correlation function representing heterogeneous
[10]
extraction methods are accurate and efficient to represent material microstructure . However, the limitation of
microstructure characterization statistically. Moreover, 2-point correlations in the representation of heterogeneity

Volume 1 Issue 1 (2022) 2 https://doi.org/10.18063/msam.v1i1.6