Scholarlink

<p>Tensile failure and desiccation cracking critically undermine the long-term performance of geotechnical infrastructures built on expansive bentonite clays, a hazard that is acute in regions like South Asia, where seasonal drying repeatedly drives these soils to exceed their tensile limit. This necessitates a coupled experimental and predictive framework for these two phenomena. The previous models reported so far depend on extensive input and larger datasets, which are impractical where equipment sources are scarce. This study developed 140 experimental records for both surface crack ratio (SCR) and direct tensile strength (DTS), obtained from desiccation and direct tension tests on compacted bentonite clay. Tensile measurements were performed on self-developed ‘8-shaped’ direct tensile apparatus, which was validated for reliable tensile failure of bentonite clay. The experimental results showed statistically significant results conforming to the established soil mechanics of compacted clay. Four predictive frameworks: multiple linear regression (MLR), multiple nonlinear regression (MNLR), artificial neural network (ANN), and random forest (RF) were employed for these two parameters separately. ANN demonstrated superior performance for both DTS and SCR prediction. A robustness analysis with uncertainty quantification was also performed to ensure the model’s stability for moderate-sized noisy dataset with fewer input parameters. Besides, the results of SHAP analysis identified water content and dry density to be predominant features. Finally, a novel regression model was established correlating SCR to DTS, ensuring statistical reliability. These findings offer a quantitative tool for estimating cracking and tensile failure thresholds in compacted bentonite clays used in various engineering designs and earth structures.</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

Desiccation Cracking and Direct Tensile Strength of Compacted Expansive Bentonite Clay: Experimental Investigation and Predictive Modeling

  • 期刊论文

  • 发表日期：

    2026年7月28日

查看全文
  • Md. Nazmul Islam Rafi,
  • Amena Begum,
  • Mohammad Shahidur Rahman,
  • Shriful Islam,
  • Md. Daniel Hossain Tonmoy

摘要

Tensile failure and desiccation cracking critically undermine the long-term performance of geotechnical infrastructures built on expansive bentonite clays, a hazard that is acute in regions like South Asia, where seasonal drying repeatedly drives these soils to exceed their tensile limit. This necessitates a coupled experimental and predictive framework for these two phenomena. The previous models reported so far depend on extensive input and larger datasets, which are impractical where equipment sources are scarce. This study developed 140 experimental records for both surface crack ratio (SCR) and direct tensile strength (DTS), obtained from desiccation and direct tension tests on compacted bentonite clay. Tensile measurements were performed on self-developed ‘8-shaped’ direct tensile apparatus, which was validated for reliable tensile failure of bentonite clay. The experimental results showed statistically significant results conforming to the established soil mechanics of compacted clay. Four predictive frameworks: multiple linear regression (MLR), multiple nonlinear regression (MNLR), artificial neural network (ANN), and random forest (RF) were employed for these two parameters separately. ANN demonstrated superior performance for both DTS and SCR prediction. A robustness analysis with uncertainty quantification was also performed to ensure the model’s stability for moderate-sized noisy dataset with fewer input parameters. Besides, the results of SHAP analysis identified water content and dry density to be predominant features. Finally, a novel regression model was established correlating SCR to DTS, ensuring statistical reliability. These findings offer a quantitative tool for estimating cracking and tensile failure thresholds in compacted bentonite clays used in various engineering designs and earth structures.