Scholarlink

<p>This study evaluates the fire resistance of mortar incorporating zirconia (ZrO₂) and rice husk ash (RHA) under elevated temperatures (25–400&#xa0;°C) using a cement–sand ratio of 1:3. Experimental results showed that the optimum mix with 3% zirconia and 40% RHA improved residual strength up to 200&#xa0;°C, after which strength declined due to microstructural degradation. Zirconia enhanced crack-bridging and thermal stability, while RHA improved pozzolanic reactivity and densification. A dataset of 468 samples was used to train Random Forest, LightGBM, and XGBoost regressors, with XGBoost achieving the highest accuracy (R² = 0.991, MAE = 0.245&#xa0;MPa, MAPE = 1.25%). SHAP analysis confirmed temperature as the most critical factor reducing strength, while curing duration and zirconia content contributed positively. These results demonstrate that integrating waste-derived additives with AI-driven modeling provides a sustainable and accurate framework for designing fire-resilient mortars with reduced dependence on empirical testing.</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

Fire-resistant mortar with waste-derived additives: machine learning prediction of compressive strength at elevated temperatures

  • 期刊论文

  • 发表日期：

    2025年10月29日

查看全文
  • Ankit,
  • Aditya Kumar Tiwary,
  • Abhishek Sharma

摘要

This study evaluates the fire resistance of mortar incorporating zirconia (ZrO₂) and rice husk ash (RHA) under elevated temperatures (25–400 °C) using a cement–sand ratio of 1:3. Experimental results showed that the optimum mix with 3% zirconia and 40% RHA improved residual strength up to 200 °C, after which strength declined due to microstructural degradation. Zirconia enhanced crack-bridging and thermal stability, while RHA improved pozzolanic reactivity and densification. A dataset of 468 samples was used to train Random Forest, LightGBM, and XGBoost regressors, with XGBoost achieving the highest accuracy (R² = 0.991, MAE = 0.245 MPa, MAPE = 1.25%). SHAP analysis confirmed temperature as the most critical factor reducing strength, while curing duration and zirconia content contributed positively. These results demonstrate that integrating waste-derived additives with AI-driven modeling provides a sustainable and accurate framework for designing fire-resilient mortars with reduced dependence on empirical testing.