Scholarlink

<p>Concrete is the most extensively used construction material worldwide due to its strength and durability. Concrete requires proper compaction to perform effectively. The compaction process requires skilled workers, extra time, and vibrators to minimize voids in the concrete and achieve the necessary strength and durability. Using self-compacting concrete (SCC) instead of normal concrete results in higher compressive strength (CS) and durability. Self-compacting concrete significantly benefits the construction process, including cost and time reduction, minimized labor, and improved overall performance. However, evaluating the compressive strength is crucial to ensure the SSC’s durability. Based on a sensitivity analysis chart in this study, some factors significantly influence the CS, such as age, coarse aggregate, and fly ash. This study investigates the effect of those factors on the CS of SCC by applying several modeling techniques for 123 different mixtures. The predictive modeling techniques used to predict the CS of SCC include Linear Regression, Non-Linear Regression, Multi-Linear Regression, Logarithmic, Pure Quadratic, Artificial Neural Network, and the M5P-tree. The independent variables in the obtained dataset are Cement with its value ranging between 141.5–530 kg/m<sup>3</sup>, Limestone 0–200 kg/m<sup>3</sup>, Fly Ash 0–275 kg/m<sup>3</sup>, Fine Aggregate 464.4–1014 kg/m<sup>3</sup>, Coarse Aggregate 480–957 kg/m<sup>3</sup>, Water 49.53–252 kg/m<sup>3</sup>, Super Plasticize 0.30–4.70 %, Fiber 0–80 kg/m<sup>3</sup>, and Age 1–56 days. The value of CS, which is a dependent variable in this study, is between 19.8 and 75.2 MPa. Among the models that have been evaluated, the Artificial Neural Network (ANN) model demonstrated the highest accuracy in predicting compressive strength, with superior results across all evaluation criteria. Also, the Multi-Linear Regression model (MLR) showed a high performance. The evaluation of residual error confirmed that the ANN model provided the smallest error compared to the other models. This study's findings highlight the models that accurately predicted the CS of SCC, along with the factors that had the most tremendous impact on the CS of SCC. </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

Sensitivity-based prediction of self-compacting concrete strength using hybrid modeling techniques

  • 期刊论文

  • 发表日期：

    2025年6月7日

查看全文
  • Sham Hara Mohammed,
  • Lalan Barzan Hussein,
  • Ahmed Salih Mohammed

摘要

Concrete is the most extensively used construction material worldwide due to its strength and durability. Concrete requires proper compaction to perform effectively. The compaction process requires skilled workers, extra time, and vibrators to minimize voids in the concrete and achieve the necessary strength and durability. Using self-compacting concrete (SCC) instead of normal concrete results in higher compressive strength (CS) and durability. Self-compacting concrete significantly benefits the construction process, including cost and time reduction, minimized labor, and improved overall performance. However, evaluating the compressive strength is crucial to ensure the SSC’s durability. Based on a sensitivity analysis chart in this study, some factors significantly influence the CS, such as age, coarse aggregate, and fly ash. This study investigates the effect of those factors on the CS of SCC by applying several modeling techniques for 123 different mixtures. The predictive modeling techniques used to predict the CS of SCC include Linear Regression, Non-Linear Regression, Multi-Linear Regression, Logarithmic, Pure Quadratic, Artificial Neural Network, and the M5P-tree. The independent variables in the obtained dataset are Cement with its value ranging between 141.5–530 kg/m3, Limestone 0–200 kg/m3, Fly Ash 0–275 kg/m3, Fine Aggregate 464.4–1014 kg/m3, Coarse Aggregate 480–957 kg/m3, Water 49.53–252 kg/m3, Super Plasticize 0.30–4.70 %, Fiber 0–80 kg/m3, and Age 1–56 days. The value of CS, which is a dependent variable in this study, is between 19.8 and 75.2 MPa. Among the models that have been evaluated, the Artificial Neural Network (ANN) model demonstrated the highest accuracy in predicting compressive strength, with superior results across all evaluation criteria. Also, the Multi-Linear Regression model (MLR) showed a high performance. The evaluation of residual error confirmed that the ANN model provided the smallest error compared to the other models. This study's findings highlight the models that accurately predicted the CS of SCC, along with the factors that had the most tremendous impact on the CS of SCC.