Page 8 - ESAM-1-4

Engineering Science in
Additive Manufacturing Experimental statistics in AM

with traditional methods, particularly in the fabrication validation experiments to ensure the models represent
of high-value parts with metal alloys using laser-based the physical reality with appropriate fidelity. Therefore, a
powder bed fusion of metals (PBF-LB/M). Some of these rigorous statistical analysis on data collected from a well-
characteristics include strength, reliability, geometric designed experiment is the best way to find out what
tolerances, cosmetic aspects, and cost. Experiments will parameters affect the AM process, set these parameters
3
play a fundamental role in the achievement of these goals. to obtain the best output sought, and determine if the
Thus, it is important that the right tools and procedures are resulting process can meet the required specifications.
used to properly design these experiments, analyze their Thus, given the pervasiveness of AM as well as the
results, and report critical findings. uncertainty about the quality of the fabricated objects,
To begin, we provide a brief overview of AM generally experiments and statistics should be used in tandem
and PBF-LB/M specifically. AM is defined as a process of to assess potential sources of variability and to ensure
joining materials to make objects from three-dimensional consistency of the final product. This issue is particularly
(3D) model data, usually in a constructive manner, i.e., salient within PBF-LB/M, which is considered one of the
layer upon layer, as opposed to subtractive manufacturing most promising processes for fabricating critical structural
methodologies. The goal of AM is to quickly and efficiently components in high-end applications such as aerospace
4
20
fabricate complicated objects that may be inconvenient to and medicine. In this paper, we assess the current state-
produce with traditional machining techniques. While of-the-art practices of experimental statistics, focusing
5
the specifics of every AM technique differ, the complexity on the design of experiments (DOE 21-23 ) and response
of potential builds enforces a sophisticated approach to surface methodologies (RSM 24,25 ). To do this, we evaluate
manufacturing 3D objects. For instance, one form of PBF the frequency at which statistical techniques and different
requires raking a metallic powder onto a build plate, using experimental design types are used within AM generally
a laser beam as an energy source to melt the powder into a and PBF-LB/M specifically, and evaluate how these
desired shape, lowering the build plate, and then repeating practices have been changing over time. We also compare
the process until a 3D structure is formed, referred to as the evolution of the use of these methods within PBF-
6
PBF-LB/M. However, the field also employs dozens of LB/M and a highly regulated medical field (orthopedics),
3
other AM techniques as well, utilizing a wide diversity and summarize the use of these methods in the best of
of materials such as polymers, metals, and ceramics ; the sampled papers. We then compare current practices
7
the American Society for Testing and Materials (ASTM) to ideal practices to provide recommendations on how
has classified the myriad AM technologies into seven the field can improve experimentally, offering a roadmap
overarching process categories. AM is used within fields that engineers can follow to maximize the efficacy of their
8
such as aerospace engineering, construction, robotics, experiments.
10
11
9
and medicine. The end application of the fabricated
12
objects can also be very different, from rough plastic 2. State-of-the-art practices for statistics
prototyping to operations that require higher performance and DOE
and reliability. 13,14 Although AM has been around for the last 30+ years, it has
Regardless of the end goal of the process and the type mostly remained in the areas of prototyping and, in a few
of AM used, the complexity of manufacturing allows for cases, in small production of parts. However, it is widely
error at many different processing steps. In addition, believed that the field should aim to develop systems and
15
each manufacturing process is highly sensitive to the processes that result in the manufacturing of parts that
parameter settings of the machine, causing uncertainty are ready for use in industrial applications. A primary
16
in the quality of the final products. In the case of PBF- challenge in AM, particularly for metal components, is
LB/M, these factors include the type of feedstock used, ensuring that printed parts achieve reliability comparable
the type of commercial AM system, laser power, scan to those manufactured through conventional methods.
speed, and more than a hundred additional factors. 17,18 It is Demonstrating this equivalence necessitates rigorous
difficult to know a priori which factors will have an effect testing, thorough experimentation, and robust methods
and what that effect will be, especially within PBF-LB/M for assessing quality and consistency. Experimental
where the relevant factors are very often not known and statistics offers valuable tools to address these needs by
are not controlled. Simulating the process with tools such quantifying variability, optimizing process parameters,
as finite element modeling can help in the absence of real and validating material properties and mechanical
data, but accurate models can be challenging to design, performance. Moreover, statistical methods can accelerate
19
computationally expensive, part-specific, and require the development and certification stages by efficiently

Volume 1 Issue 4 (2025) 2 doi: 10.36922/ESAM025340021