Scholarlink

<p>The accelerated expansion of the construction industry has heightened the demand for concrete and the production of construction-demolition waste, prompting the utilisation of recycled fine aggregate (RFA) to preserve natural sand resources. This work experimentally assesses the compressive strength of GGBS-blended concrete using 0-100% RFA (by mass of fine aggregate) across three curing environments: normal water, 3% HCl, and 5% H₂SO₄, over curing durations of up to 90 days. The compressive strength improved with curing age across all mixtures; an optimal replacement of 60% RFA resulted in the highest strength, exhibiting an increase of up to 6.44% at 90 days (normal water) relative to the control. Conversely, higher replacement levels (80–100%) resulted in a decrease in strength due to increased porosity and diminished interfacial bonding. To generalise these findings and minimise further testing, machine-learning models utilising Regression Trees (RT), Support Vector Machines (SVM), and Gene Expression Programming (GEP) were created to predict compressive strength based on mix proportions and curing parameters; the GEP model exhibited superior performance (R² = 0.990, RMSE = 1.207, MAE = 0.974, MAPE = 2.426%) and offers an equation for design-oriented applications within the examined parameter space. The model is calibrated on the current experimental database; thus, its use should be confined to tested ranges, and transferability was assessed using grouped k-fold cross-validation. The research indicates that up to 60% RFA can be used into GGBS-blended concrete, and that interpretable machine learning models can proficiently assist in the mix design of RFA concretes.</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

AI-driven strength prediction and design of recycled fine-aggregate cementitious composites

  • 期刊论文

  • 发表日期：

    2026年3月24日

查看全文
  • Rishabh Tyagi,
  • Pradeep Kumar Goyal

摘要

The accelerated expansion of the construction industry has heightened the demand for concrete and the production of construction-demolition waste, prompting the utilisation of recycled fine aggregate (RFA) to preserve natural sand resources. This work experimentally assesses the compressive strength of GGBS-blended concrete using 0-100% RFA (by mass of fine aggregate) across three curing environments: normal water, 3% HCl, and 5% H₂SO₄, over curing durations of up to 90 days. The compressive strength improved with curing age across all mixtures; an optimal replacement of 60% RFA resulted in the highest strength, exhibiting an increase of up to 6.44% at 90 days (normal water) relative to the control. Conversely, higher replacement levels (80–100%) resulted in a decrease in strength due to increased porosity and diminished interfacial bonding. To generalise these findings and minimise further testing, machine-learning models utilising Regression Trees (RT), Support Vector Machines (SVM), and Gene Expression Programming (GEP) were created to predict compressive strength based on mix proportions and curing parameters; the GEP model exhibited superior performance (R² = 0.990, RMSE = 1.207, MAE = 0.974, MAPE = 2.426%) and offers an equation for design-oriented applications within the examined parameter space. The model is calibrated on the current experimental database; thus, its use should be confined to tested ranges, and transferability was assessed using grouped k-fold cross-validation. The research indicates that up to 60% RFA can be used into GGBS-blended concrete, and that interpretable machine learning models can proficiently assist in the mix design of RFA concretes.