Page 115 - MSAM-4-3

Materials Science in Additive Manufacturing Sustainable manufacturing composite material optimization

speed was held constant at 30% to regulate solidification of This continuous monitoring enabled the identification of
the extruded material, thereby preventing overheating and energy-intensive stages and supports the development of
supporting interlayer bonding. The test specimens were more energy-efficient testing and printing processes. To
printed in the XY-plane to ensure mechanical stability normalize energy usage across specimens of varying infill
and were subjected to compression, flexural, and tensile densities and print durations, energy consumption was
testing. Real-time power consumption during printing was reported per unit mass of printed material (Wh/g).
also monitored using a power meter, providing real-time Figure 1 illustrates the complete fabrication and
data on the energy demands of TPU 95A processing. This characterization of TPU 95A specimens by FDM.
energy analysis is essential for evaluating the sustainability Experimental data, including mechanical characteristics,
of FDM printing with TPU, particularly in comparison to energy consumption, and process efficiency, were
conventional manufacturing techniques. analyzed accordingly. By integrating AI-assisted

Following the printing process, the mechanical process optimization, real-time energy monitoring,
properties of printed TPU 95A specimens were evaluated and mechanistic testing, this study aimed to establish a
in accordance with standardized test procedures. The sustainable and high-performance system for the AM of
samples were conditioned and tested based on ASTM TPU-based components.
D638 Type V for tensile strength, ASTM D790 for flexural
properties, ASTM D695 for compression, and ASTM 2.2. ANN implementation
E99 for wear resistance using a pin-on-disc tribometer. In this study, an ANN was developed to predict key
Tensile, flexural, and compressive tests were conducted mechanical properties, namely, tensile strength, flexural
using a Tinius Olsen Universal Testing Machine (UTM) strength, compression strength, and wear resistance,
(Norway). The experiments were conducted at a strain based on FDM process parameters. The ANN model also
rate of 5 mm/min to ensure consistent stress application served as a decision-support system within the broader
and precise mechanical response. In addition, a 10 kN unit AI-MCDM optimization framework. 18,50-53
load was used to determine the tensile strength, Young’s
modulus, flexural strength, and compressive strength 2.2.1. Model architecture and setup
of the TPU 95A specimen. Wear resistance was assessed The ANN architecture consisted of the following
using a pin-on-disc tribometer, evaluating the adhesive components:
and abrasive resistance of the material when subjected to (i) Input layer: Four neurons representing the main
regulated friction conditions. Each mechanical test was FDM parameters – layer thickness, infill density, shell
replicated three times to ensure statistical validity and thickness, and print speed.
reduce experimental variability. Energy consumption (ii) Hidden layers: Two hidden layers were employed: the
during the testing process was monitored continuously, first consisted of 16 neurons, and the second had eight
providing insights into the power consumption of the neurons; both layers used the Rectified Linear Unit
tribometer and UTM during mechanical characterization. (ReLU) activation function to introduce non-linearity.

Figure 1. Experimental workflow for sustainable fuse deposition modeling (FDM)-based printing of TPU 95A components. India

Volume 4 Issue 3 (2025) 7 doi: 10.36922/MSAM025200033