Page 35 - IJAMD-2-2

International Journal of AI for
Materials and Design ML-driven optimization in additive manufacturing

manufacturing approach builds objects layer by layer requirements remain key challenges. 39,40 Binder jetting,
from digital models, maximizing material efficiency and which selectively deposits a liquid binding agent onto a
enabling intricate geometries. Its ability to offer exceptional powder bed, offers a low-temperature alternative to PBF,
design flexibility has made it a transformative technology making it particularly useful for ceramic and metal-based
in industries such as aerospace, healthcare, and automotive, applications. 41,42 This method enables high-speed, cost-
where lightweight, customized, and functionally optimized effective large-scale fabrication, but the resulting parts
structures are essential for innovation and performance. often require sintering or infiltration post-processing
The ASTM standard classifies AM into seven categories, to achieve full density and mechanical strength. DED,
with material extrusion, vat photopolymerization, powder which directly deposits molten or semi-molten material
bed fusion (PBF), binder jetting, and directed energy using a focused energy source such as a laser or plasma
deposition (DED) being the most widely adopted. arc, is commonly used for metal repair, aerospace
35
While AM technologies offer significant advantages component restoration, and large-scale manufacturing.
over conventional manufacturing, each method presents While DED provides greater material efficiency and repair
unique processing requirements and challenges depending capabilities, it typically results in lower resolution and
on the material type. surface quality compared to PBF, necessitating additional
post-processing to improve dimensional accuracy.
43
Polymer-based AM technologies, such as material While AM technologies continue to advance, optimizing
extrusion and vat photopolymerization, are widely used due process parameters, improving material performance,
to their cost-effectiveness and broad material availability. and addressing scalability challenges remain critical for
Material extrusion, exemplified by fused deposition industrial adoption. Recent advancements in ML have
modeling (FDM) and direct ink writing (DIW), is one of demonstrated significant potential in enhancing process
the most accessible and scalable AM techniques. FDM, efficiency, real-time monitoring, and defect prediction, as
which utilizes thermoplastic filaments such as acrylonitrile summarized in Figure 1.
butadiene styrene (ABS) and polylactic acid (PLA), is
commonly used for rapid prototyping and functional 2.2. ML approaches
components. However, FDM parts often suffer from ML techniques have become increasingly integral to
anisotropic mechanical properties and limited resolution, AM process optimization, defect detection, and quality
requiring post-processing or parameter optimization control. ML approaches in AM are broadly classified
to enhance quality. DIW, in contrast, is particularly
36
advantageous for soft and bioinspired materials, such
as hydrogels and elastomers, but demands precise
rheological control to maintain printing accuracy. Vat
37
photopolymerization, which includes stereolithography
(SLA) and digital light processing (DLP), enables high-
resolution fabrication with smooth surface finishes, making
it particularly beneficial for biomedical applications,
dental prosthetics, and microfluidic devices. However,
photopolymer-based materials often exhibit brittleness
and require post-curing, which may limit their mechanical
performance in load-bearing applications. 38
Metal and ceramic-based AM technologies, such as PBF,
binder jetting, and DED, are essential for high-performance
applications that require superior mechanical properties
and thermal resistance. PBF processes, including selective
laser sintering (SLS) for polymers and selective laser melting
or electron beam melting for metals, utilize high-energy
sources to selectively fuse powdered materials, enabling the
production of complex, high-strength components. These
techniques are widely used in aerospace, automotive, and
medical implants, where precision and material integrity Figure 1. Overview of machine learning for additive manufacturing.
are critical. However, strict control of powder properties, Machine learning applications, learning types, and key roles in process
high energy consumption, and extensive post-processing optimization, quality prediction, and real-time monitoring.

Volume 2 Issue 2 (2025) 29 doi: 10.36922/IJAMD025130010