Tensile Testing and Mechanical Property Prediction of 3D Printed PLA Using FEA and Machine Learning

Abstract

This paper presents an integrated methodology combining physical tensile testing, finite element analysis (FEA), and machine learning (ML) to evaluate and predict the mechanical behaviour of 3D printed polylactic acid (PLA). ASTM D638 Type I specimens were fabricated using a Raise3D E2 printer and tested with an INSTRON 5582 Universal Testing Machine. Numerical simulations were performed using ANSYS Mechanical, and tensile strength was predicted using an Elastic Net Regression model trained on publicly available data. The proposed hybrid approach demonstrates effective parameter optimization and provides valuable insights into the mechanical performance of PLA components produced via additive manufacturing.

Introduction

The mechanical properties of materials, especially tensile strength, are crucial for the design and integrity of engineered components. Tensile strength is commonly measured via uniaxial tensile testing, producing stress–strain curves from which properties like Young’s modulus, yield strength, and elongation at break are derived.

Additive manufacturing (AM), particularly fused deposition modelling (FDM), is increasingly used in industries like aerospace and biomedical due to its ability to produce complex, customized parts with minimal waste. Polylactic Acid (PLA), a biodegradable thermoplastic, is a popular FDM material because of its good printability, mechanical strength, and environmental benefits.

In this study, tensile specimens were fabricated following ASTM D638 Type I standards using a Raise3D E2 printer with PLA filament. Mechanical testing was performed on an INSTRON 5582 Universal Testing Machine to obtain accurate tensile data. Since experimental testing can be time-consuming and costly, a machine learning model was trained on public data to predict tensile strength based on printing parameters and validated experimentally. Additionally, finite element analysis (FEA) in ANSYS simulated stress distribution under load to correlate with failure regions.

Specimen Design and Modeling:

• 2D CAD drawings were created in AutoCAD, adhering to ASTM dimensions.

• 3D models were developed in SolidWorks and exported as IGS files for FEA and STL files for 3D printing, ensuring consistency between physical and simulated specimens.

3D Printing Parameters:

• Specimens were sliced using IdeaMaker with fixed parameters (e.g., 0.2 mm layer height, 100% infill, 205°C nozzle temperature).

• Printing was done on Raise3D E2 with a single extruder, within a temperature-controlled chamber to ensure dimensional stability.

Tensile Testing:

• Conducted according to ASTM D638 Type I using INSTRON 5582.

• Key parameters included a 57 mm gauge length, 5 kN load capacity, and 0.1126 mm/s strain rate.

• Results yielded Young’s modulus, yield strength, ultimate tensile strength, and strain at break.

This integrated experimental, numerical, and data-driven approach enables efficient characterization and optimization of the mechanical performance of 3D-printed PLA components.

Conclusion

This study presented an integrated methodology for characterizing the tensile behaviour of 3D-printed PLA specimens by combining experimental testing, finite element analysis (FEA), and machine learning (ML). The synergy between physical validation, numerical simulation, and data-driven modelling offers a robust framework for mechanical analysis and performance prediction of additive-manufactured components. The key outcomes of this work are summarized as follows: 1) Experimental Characterization: Three ASTM D638 Type I PLA specimens were printed using a Raise3D E2 and tested on an INSTRON 5582 UTM, yielding an average tensile strength of 32.35 MPa and modulus of 1308 MPa, confirming PLA’s brittle behaviour with elongation under 15%. 2) Fracture Analysis: Failure consistently occurred in the fillet region, identified as the zone of highest stress concentration. This was observed visually and further validated through simulation. 3) FEA Validation: ANSYS-based simulations accurately replicated the stress distribution observed during physical testing, with peak Von Mises stress (~34 MPa) localized at the fillet. This alignment between simulation and experimental outcomes demonstrates the predictive accuracy of FEA for stress localization in 3D-printed components. 4) Machine Learning Prediction: An Elastic Net Regression model trained on a publicly available Kaggle dataset achieved a prediction error of just 5.07%, effectively estimating tensile strength from print parameters. This demonstrates the potential of ML in optimizing print settings without exhaustive physical testing. In conclusion, the combined use of physical experimentation, simulation, and ML offers a time-efficient and cost-effective approach for evaluating and improving the mechanical properties of 3D-printed PLA. The proposed methodology can be extended to other polymeric or composite materials, making it a scalable blueprint for hybrid experimental–computational–data-driven material characterization in additive manufacturing.

References

[1] Popescu, A. Zapciu, C. Amza, F. Baciu, and R. Marinescu, “FDM process parameters influence over the mechanical properties of polymer specimens: A review,” Polymer Testing, vol. 69, pp. 157–166, 2018. [Online]. Available: https://doi.org/10.1016/j.polymertesting.2018.05.020 [2] ASTM International, Standard Test Method for Tensile Properties of Plastics, ASTM D638-14, West Conshohocken, PA, 2014. [Online]. Available: https://www.astm.org/d0638-14.html [3] S. Farah, D. G. Anderson, and R. Langer, “Physical and mechanical properties of PLA, and their functions in biomedical applications,” Advanced Drug Delivery Reviews, vol. 107, pp. 367–392, 2016. [Online]. Available: https://doi.org/10.1016/j.addr.2016.06.012 [4] S. Banerjee, “3D Printing Parameters Dataset,” Kaggle, 2020. [Online]. Available: https://www.kaggle.com/datasets/sounakbanerjee/3d-printing-parameters [5] J. M. Chacón, M. A. Caminero, E. García-Plaza, and P. J. Núñez, “Additive manufacturing of PLA structures using fused deposition modelling: Effect of process parameters on mechanical properties and their optimal selection,” Materials & Design, vol. 124, pp. 143–157, 2017. [Online]. Available: https://doi.org/10.1016/j.matdes.2017.03.065 [6] K. L. Lawrence, ANSYS Workbench 14: Release Guide. SDC Publications, 2013. [7] X. Zhang, M. Li, Y. Zhou, and Y. Wu, “Mechanical property prediction of 3D-printed PLA using machine learning,” Journal of Materials Research and Technology, vol. 15, pp. 2489–2500, 2021. [Online]. Available: https://doi.org/10.1016/j.jmrt.2021.09.092 [8] B. Shahrjerdi and M. R. Barzegari, “FEA analysis of mechanical behavior of 3D printed PLA under tensile load,” in Proc. ASME Int. Mech. Eng. Congr. Expo., 2018, paper 51934. [Online]. Available: https://doi.org/10.1115/IMECE2018-88132 [9] J. Jordan, 3D Printing: The Essential Guide to 3D Printers. CreateSpace Independent Publishing Platform, 2019. [10] C. Sebastian, Machine Learning for Beginners: A Comprehensive Guide to Understanding Machine Learning Algorithms. Independently Published, 2021.

Copyright

Copyright © 2025 Angad Riat, Kaushal Singhania, Sujan Banerjee. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.