Scholarlink

<p>This study investigates the energy absorption performance of thin-walled hollow polymeric tubes with various cross-sectional geometries fabricated by Fused Deposition Modelling (FDM). Tubes made from natural PLA, ABS, and PETG were designed with triangular, square, pentagonal, hexagonal, and circular profiles, all standardized to a constant wall thickness and an equivalent circumscribed diameter. Quasi-static compression tests were conducted to evaluate crashworthiness indicators including peak crushing force, mean crushing force, specific energy absorption (SEA), crushing force efficiency (CFE), stiffness and stiffness per unit area. Experimental results revealed that performance was strongly influenced by both material type and geometric configuration. Hexagonal and pentagonal sections generally demonstrated superior crashworthiness, while triangular sections consistently underperformed due to unstable failure modes. PLA exhibited the highest stiffness and PCF, but also exhibited brittle failure, whereas PETG showed a favorable balance of ductility and energy absorption. To complement experimental findings, a machine learning model based on Categorical Boosting regression was developed and interpreted with Shapley Additive Explanations (SHAP) to quantify the influence of material and geometry on energy absorption metrics. The model revealed that material selection predominantly governs stiffness (89.3% influence), while geometry had a greater effect on CFE (63.0%). SEA and absorbed energy were influenced by both factors nearly equally. The SHAP heatmap showed that triangular geometries negatively affected performance, whereas hexagonal and pentagonal shapes provided consistently positive contributions. These findings offer valuable design guidance for developing lightweight, crashworthy polymeric structures, and demonstrate the combined power of experimental testing and interpretable machine learning in advanced structural optimization.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Investigating material and geometric effects on the crashworthiness of 3D-Printed polymeric tubes through machine learning quantification

  • 期刊论文

  • 发表日期：

    2025年12月10日

查看全文
  • Emir Avcioglu,
  • Melih Eltis

摘要

This study investigates the energy absorption performance of thin-walled hollow polymeric tubes with various cross-sectional geometries fabricated by Fused Deposition Modelling (FDM). Tubes made from natural PLA, ABS, and PETG were designed with triangular, square, pentagonal, hexagonal, and circular profiles, all standardized to a constant wall thickness and an equivalent circumscribed diameter. Quasi-static compression tests were conducted to evaluate crashworthiness indicators including peak crushing force, mean crushing force, specific energy absorption (SEA), crushing force efficiency (CFE), stiffness and stiffness per unit area. Experimental results revealed that performance was strongly influenced by both material type and geometric configuration. Hexagonal and pentagonal sections generally demonstrated superior crashworthiness, while triangular sections consistently underperformed due to unstable failure modes. PLA exhibited the highest stiffness and PCF, but also exhibited brittle failure, whereas PETG showed a favorable balance of ductility and energy absorption. To complement experimental findings, a machine learning model based on Categorical Boosting regression was developed and interpreted with Shapley Additive Explanations (SHAP) to quantify the influence of material and geometry on energy absorption metrics. The model revealed that material selection predominantly governs stiffness (89.3% influence), while geometry had a greater effect on CFE (63.0%). SEA and absorbed energy were influenced by both factors nearly equally. The SHAP heatmap showed that triangular geometries negatively affected performance, whereas hexagonal and pentagonal shapes provided consistently positive contributions. These findings offer valuable design guidance for developing lightweight, crashworthy polymeric structures, and demonstrate the combined power of experimental testing and interpretable machine learning in advanced structural optimization.