Scholarlink

<p>This study investigates the influence of compressive strength on the mechanical behavior of coarse and fine aggregates sourced from the Seti-River Basin in Gandaki Province, Nepal. Concrete mixes of M20, M25, and M30 grades were prepared using these river-derived aggregates, and their 28-day compressive strengths were evaluated through standardized laboratory tests. To further understand the material response, finite element simulations were performed using Abaqus CAE, capturing stress distribution and failure mechanisms in each mix design. Additionally, ensemble machine learning models, Random Forest, Gradient Boosting, and XGBoost were employed to predict 28-day compressive strength based on aggregate properties and mix proportions. The integrated approach revealed strong correlations between aggregate characteristics and overall concrete performance, with ensemble models achieving high prediction accuracy (R² &gt;0.90). This hybrid framework offers a reliable pathway for optimizing concrete mix design using naturally available aggregates, supporting sustainable construction practices in the Himalayan region.</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

Multi-scale modeling and prediction of concrete strength using river-derived aggregates: an integrated experimental, numerical, and AI approach

  • 期刊论文

  • 发表日期：

    2025年8月25日

查看全文
  • Bishwash Poudel,
  • Yagyanath Rimal,
  • Niroj Lamichhane,
  • Sundar Adhikari,
  • Amrit Poudel,
  • Ishwor Thapa

摘要

This study investigates the influence of compressive strength on the mechanical behavior of coarse and fine aggregates sourced from the Seti-River Basin in Gandaki Province, Nepal. Concrete mixes of M20, M25, and M30 grades were prepared using these river-derived aggregates, and their 28-day compressive strengths were evaluated through standardized laboratory tests. To further understand the material response, finite element simulations were performed using Abaqus CAE, capturing stress distribution and failure mechanisms in each mix design. Additionally, ensemble machine learning models, Random Forest, Gradient Boosting, and XGBoost were employed to predict 28-day compressive strength based on aggregate properties and mix proportions. The integrated approach revealed strong correlations between aggregate characteristics and overall concrete performance, with ensemble models achieving high prediction accuracy (R² >0.90). This hybrid framework offers a reliable pathway for optimizing concrete mix design using naturally available aggregates, supporting sustainable construction practices in the Himalayan region.