Scholarlink

<p>This study investigates the sorption behavior of cement-stabilized earth block (CSEB) produced using lateritic soils, focusing on the influence of various soil properties such as fine content, sand content, plasticity index, maximum dry density, and optimum moisture content. Sorption encompasses the combined processes of absorption, where liquids are drawn into pores, and adsorption, where molecules cling to surfaces. Understanding sorption is essential for ensuring the durability of masonry structures constructed with cement-stabilized earth blocks. Nine lateritic soil samples from northern Sri Lanka were utilized to create CSEB with varying cement contents of 8%, 12%, and 16% by weight. Based on the experimental program, 648 data points were collected and utilized to create training and testing datasets for predicting the sorption rate of CSEB using machine learning models. Additionally, 112 data points from published literature were gathered to validate these machine-learning models. Machine learning models, including Artificial Neural Networks (ANN), k-nearest neighbors (KNN), random forest regression (RF), and Extreme Gradient Boosting (XGB), were employed to predict sorption behavior based on soil properties. The ANN model exhibited the highest accuracy for the training and testing datasets, achieving a root mean squared error (RMSE) of 0.262&#xa0;mm and 0.572&#xa0;mm, respectively, indicating strong predictive capability. However, the XGB model outperformed the other models on the validation datasets, achieving an RMSE of 0.656&#xa0;mm. These findings underscore the critical role of soil characteristics in the sorption behavior of CSEB and demonstrate the effectiveness of the developed machine learning model in predicting sorption based on these properties, providing valuable insights for engineers and builders.</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

Sorption behavior of cement-stabilized earth blocks: machine learning insights

  • 期刊论文

  • 发表日期：

    2025年10月17日

查看全文
  • Navaratnarajah Sathiparan,
  • Pratheeba Jeyananthan

摘要

This study investigates the sorption behavior of cement-stabilized earth block (CSEB) produced using lateritic soils, focusing on the influence of various soil properties such as fine content, sand content, plasticity index, maximum dry density, and optimum moisture content. Sorption encompasses the combined processes of absorption, where liquids are drawn into pores, and adsorption, where molecules cling to surfaces. Understanding sorption is essential for ensuring the durability of masonry structures constructed with cement-stabilized earth blocks. Nine lateritic soil samples from northern Sri Lanka were utilized to create CSEB with varying cement contents of 8%, 12%, and 16% by weight. Based on the experimental program, 648 data points were collected and utilized to create training and testing datasets for predicting the sorption rate of CSEB using machine learning models. Additionally, 112 data points from published literature were gathered to validate these machine-learning models. Machine learning models, including Artificial Neural Networks (ANN), k-nearest neighbors (KNN), random forest regression (RF), and Extreme Gradient Boosting (XGB), were employed to predict sorption behavior based on soil properties. The ANN model exhibited the highest accuracy for the training and testing datasets, achieving a root mean squared error (RMSE) of 0.262 mm and 0.572 mm, respectively, indicating strong predictive capability. However, the XGB model outperformed the other models on the validation datasets, achieving an RMSE of 0.656 mm. These findings underscore the critical role of soil characteristics in the sorption behavior of CSEB and demonstrate the effectiveness of the developed machine learning model in predicting sorption based on these properties, providing valuable insights for engineers and builders.