Scholarlink

<p>Ground tire rubber (GTR) as a modifier notably improves asphalt binder performance while simultaneously addressing environmental concerns associated with tire disposal. As electric vehicle (EV) adoption rises, scrap tire generation is expected to increase, highlighting the growing need for sustainable tire disposal solutions. To address this issue, this study modifies two asphalt binders, unmodified PG 64–22 and SBS-modified PG 76–22, with ground tire rubber (GTR) in combination with trans-polyoctenamer (TOR) as a dispersing agent. To evaluate the effects of these modifications, 72 asphalt binder samples were subjected to laboratory aging and advanced rheological testing. A two-way ANOVA test was also carried out to evaluate the statistical significance of modifier effects on binder rheological performance. The laboratory findings showed that modification with GTR and TOR enhanced the rheological performance of the binders and improved resistance to oxidative aging under both unaged and aged conditions. For PG 64–22, increasing modifier content upgraded the binder to PG 76–22, while higher dosages in SBS-modified PG 76–22 further elevated its grade to PG 82–22. The ANOVA results also showed that both modifiers had statistically significant effects on the viscosity and unaged rheological properties (<i>p</i> &lt; 0.001), while only GTR remained significant after aging. In parallel, machine learning models including Artificial Neural Network (ANN), Random Forest (RF), and Multiple Linear Regression (MLR) were used to predict complex shear modulus (G*) and phase angle (δ) of modified binders under unaged and RTFO-aged conditions. Among them, ANN demonstrated superior predictive performance, achieving R<sup>2</sup> values of 0.978 for G* and 0.908 for δ on the testing dataset, compared with RF and MLR models. The results indicate the robustness of ANN in capturing the nonlinear relationships between input variables and rheological properties. This integrated approach provides a practical framework for optimizing GTR–TOR-modified asphalt binders and identifying reliable machine learning models for predicting their rheological behavior.</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 evaluation and predictive modeling of asphalt binders modified with ground tire rubber

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月24日

查看全文
  • Azharul Islam,
  • Abujar Gifari,
  • Muhammad Mutahir,
  • Junan Shen,
  • Xiaoming Yang

摘要

Ground tire rubber (GTR) as a modifier notably improves asphalt binder performance while simultaneously addressing environmental concerns associated with tire disposal. As electric vehicle (EV) adoption rises, scrap tire generation is expected to increase, highlighting the growing need for sustainable tire disposal solutions. To address this issue, this study modifies two asphalt binders, unmodified PG 64–22 and SBS-modified PG 76–22, with ground tire rubber (GTR) in combination with trans-polyoctenamer (TOR) as a dispersing agent. To evaluate the effects of these modifications, 72 asphalt binder samples were subjected to laboratory aging and advanced rheological testing. A two-way ANOVA test was also carried out to evaluate the statistical significance of modifier effects on binder rheological performance. The laboratory findings showed that modification with GTR and TOR enhanced the rheological performance of the binders and improved resistance to oxidative aging under both unaged and aged conditions. For PG 64–22, increasing modifier content upgraded the binder to PG 76–22, while higher dosages in SBS-modified PG 76–22 further elevated its grade to PG 82–22. The ANOVA results also showed that both modifiers had statistically significant effects on the viscosity and unaged rheological properties (p < 0.001), while only GTR remained significant after aging. In parallel, machine learning models including Artificial Neural Network (ANN), Random Forest (RF), and Multiple Linear Regression (MLR) were used to predict complex shear modulus (G*) and phase angle (δ) of modified binders under unaged and RTFO-aged conditions. Among them, ANN demonstrated superior predictive performance, achieving R2 values of 0.978 for G* and 0.908 for δ on the testing dataset, compared with RF and MLR models. The results indicate the robustness of ANN in capturing the nonlinear relationships between input variables and rheological properties. This integrated approach provides a practical framework for optimizing GTR–TOR-modified asphalt binders and identifying reliable machine learning models for predicting their rheological behavior.