Issue 78

C. Popa et alii, Fracture and Structural Integrity, 78 (2026) 118-128; DOI: 10.3221/IGF-ESIS.78.07

Optimal shape determination of shear specimens using Machine Learning

Cosmin Popa, Liviu Marsavina University Politehnica Timisoara, Department of Mechanics and Strength of Materials, Romania

cosmin.popa@upt.ro, https://orcid.org/0000-0001-8046-4523 liviu.marsavina@upt.ro, http://orcid.org/0000-0002-5924-0821 Cosmin Marsavina, Calin Popa University Politehnica Timisoara, Department of CTI, Romania cosmin.marsavina@upt.ro, http://orcid.org/ 0009-0004-0648-8155 calin.popa@cs.upt.ro, http://orcid.org/ 0000-0003-4445-8091

Citation: Popa, C., Marsavina, L., Marsavina, C., Popa, C., Optimal shape determination of shear specimens using Machine Learning, Fracture and Structural Integrity, 78 (2026) 118-128.

A BSTRACT . Shear strength represents an important mechanical property. For the additively manufactured polymers nowadays, there is not a valid standard to determine the shear strength. Finite element analysis of 145 shear specimen geometries was performed. Then an AI-based optimal shape prediction methodology was developed using supervised machine learning, where a Multiple Linear Regression model was trained on selected design parameters to accurately predict shape performance and efficiently explore new design configurations. The ratio between maximum shear stress and maximum normal stress in the shear plane was considered the parameter to be optimized. K EYWORDS . Shear test, Fused Deposition Modeling, Machine learning, FEM simulation

Received: 10.06.2026 Accepted: 12.07.2026 Published: 11.08.2026 Issue: 10.2026

Copyright: © 2026 This is an open access article under the terms of the CC-BY 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

I NTRODUCTION used Deposition Modeling (FDM), is one of the most widely used additive manufacturing technologies for polymeric materials due to its low production cost, geometric flexibility, and accessibility [1]. However, the mechanical behavior of FDM-printed components is strongly influenced by process parameters such as raster orientation, layer thickness, infill density, printing temperature, and the presence of external contours or perimeter shells [2–5]. The layer-by-layer F

118