Scholarlink

<p>This study presents an integrated experimental–analytical–AI approach to evaluate the long-term hydrostatic performance of PE-RT/Al/PE-RT multi-layer pipes. Hydrostatic pressure tests were conducted on four pipe sizes (16, 20, 25, and 32&#xa0;mm) at temperatures of 20 °C, 60 °C, 95 °C, and 110 °C, with failure times ranging from 1&#xa0;h to over 10,000&#xa0;h. The stress distribution within each layer was determined using stress equilibrium in pipe layers, revealing that the aluminum layer carries 81% to 84% of the internal pressure, while the inner and outer PE-RT layers contribute 12% to 16% and &lt; 2%, respectively. A novel <i>K</i>-coefficient, ranging from 0.76 to 0.89, was introduced to relate aluminum hoop stress to its yield strength, simplifying the structural design process. To complement the analytical framework, an artificial neural network model was developed using 6 input features (diameter, layer thicknesses, temperature, failure time), a single hidden layer with 25 neurons, and trained on about 100 experimental samples. The ANN achieved a correlation coefficient <i>R</i><sup>2</sup> &gt; 0.99, RMSE = 0.93 bar, enabling accurate prediction of rupture pressure, stress profile, and <i>K</i>-coefficient for unseen configurations. This study provides a robust and time-efficient predictive tool to estimate long-term pressure capacity without requiring extensive physical testing. The proposed model supports more efficient and cost-effective design of PE-RT/Al/PE-RT pipes Additionally, a comparative analysis between PE-RT/Al/PE-RT and PE-X/Al/PE-X pipes was conducted to evaluate differences in material performance and long-term durability.</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

Experimental and ANN-based prediction of long-term failure in metal-polymer composite pipes

  • 期刊论文

  • 发表日期：

    2025年6月9日

查看全文
  • Asieh Atarodi-Kashani,
  • Maziar Khademi,
  • Shahram Delfani,
  • Nima Mafi,
  • Roxana Ghelich

摘要

This study presents an integrated experimental–analytical–AI approach to evaluate the long-term hydrostatic performance of PE-RT/Al/PE-RT multi-layer pipes. Hydrostatic pressure tests were conducted on four pipe sizes (16, 20, 25, and 32 mm) at temperatures of 20 °C, 60 °C, 95 °C, and 110 °C, with failure times ranging from 1 h to over 10,000 h. The stress distribution within each layer was determined using stress equilibrium in pipe layers, revealing that the aluminum layer carries 81% to 84% of the internal pressure, while the inner and outer PE-RT layers contribute 12% to 16% and < 2%, respectively. A novel K-coefficient, ranging from 0.76 to 0.89, was introduced to relate aluminum hoop stress to its yield strength, simplifying the structural design process. To complement the analytical framework, an artificial neural network model was developed using 6 input features (diameter, layer thicknesses, temperature, failure time), a single hidden layer with 25 neurons, and trained on about 100 experimental samples. The ANN achieved a correlation coefficient R2 > 0.99, RMSE = 0.93 bar, enabling accurate prediction of rupture pressure, stress profile, and K-coefficient for unseen configurations. This study provides a robust and time-efficient predictive tool to estimate long-term pressure capacity without requiring extensive physical testing. The proposed model supports more efficient and cost-effective design of PE-RT/Al/PE-RT pipes Additionally, a comparative analysis between PE-RT/Al/PE-RT and PE-X/Al/PE-X pipes was conducted to evaluate differences in material performance and long-term durability.