Scholarlink

<p>The black cotton soil, which is of very high plasticity, exhibits excessive swelling-shrinkage behaviour and has low bearing capacity, posing a great problem in geotechnical applications. This paper explores the stabilisation of black cotton soil using hybrid nanomaterials, i.e., Nano-Silica (NS) and Carbon Nanotubes (CNT), in combination with state-of-the-art machine-learning-based predictive modelling. Nano-Silica was varied from 0 to 2.0%, and CNT from 0 to 0.20%, resulting in a total nanomaterial content of 2.20%. An all-inclusive experimental program was conducted to assess the effects of nanomaterials on the limits of consistency, compaction properties, California Bearing Ratio (CBR), and Unconfined Compressive Strength (UCS). The findings have shown considerable improvement in soil behaviour, reduction in the plasticity index (34 to 17) and optimum moisture content (20.5 to 17.8) as well as an increase in the maximum dry density (1.58 to 1.67&#xa0;g/cc). Strength parameters showed substantial enhancement, with unsoaked CBR increasing from 3.2% to 10.2% (218%), soaked CBR from 2.1% to 7.1% (238%), and UCS from 145&#xa0;kPa to 410&#xa0;kPa (183%). Optimal performance was observed at 1.5% NS and 0.20% CNT; at higher concentrations, slight decreases were observed, which could be attributed to nanomaterial agglomeration. Artificial Neural Network (ANN) models were designed to capture the complex nonlinear behaviour, with critical geotechnical and material parameters as inputs. The models were also highly predictive (R<sup>2</sup> = 0.9951 in training and R<sup>2</sup> = 0.9671 in testing), and CBR and UCS predictions were consistent, with R<sup>2</sup> = 0.83–0.85, making them a more probable choice compared to other models. Residual analysis showed unbiased, stable model performance, whereas SHAP-based interpretation revealed CNT (%) and total nano content as the most influential parameters. The paper highlights the synergistic influence of Nano-Silica and CNT in improving soil strength and establishes a powerful AI-based predictive model. The results show that stabilising hybrid nanomaterials can enhance soil performance by more than 200%, providing a sustainable and efficient approach to treating expansive black cotton soil without relying primarily on laboratory tests.</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

Experimental and ANN-based prediction method for hybrid Nano-Silica and carbon nanotube stabilisation of black cotton soil

  • 期刊论文

  • 发表日期：

    2026年6月2日

查看全文
  • Anirudh Shirke,
  • Navnath Raut,
  • Manoj Wagh

摘要

The black cotton soil, which is of very high plasticity, exhibits excessive swelling-shrinkage behaviour and has low bearing capacity, posing a great problem in geotechnical applications. This paper explores the stabilisation of black cotton soil using hybrid nanomaterials, i.e., Nano-Silica (NS) and Carbon Nanotubes (CNT), in combination with state-of-the-art machine-learning-based predictive modelling. Nano-Silica was varied from 0 to 2.0%, and CNT from 0 to 0.20%, resulting in a total nanomaterial content of 2.20%. An all-inclusive experimental program was conducted to assess the effects of nanomaterials on the limits of consistency, compaction properties, California Bearing Ratio (CBR), and Unconfined Compressive Strength (UCS). The findings have shown considerable improvement in soil behaviour, reduction in the plasticity index (34 to 17) and optimum moisture content (20.5 to 17.8) as well as an increase in the maximum dry density (1.58 to 1.67 g/cc). Strength parameters showed substantial enhancement, with unsoaked CBR increasing from 3.2% to 10.2% (218%), soaked CBR from 2.1% to 7.1% (238%), and UCS from 145 kPa to 410 kPa (183%). Optimal performance was observed at 1.5% NS and 0.20% CNT; at higher concentrations, slight decreases were observed, which could be attributed to nanomaterial agglomeration. Artificial Neural Network (ANN) models were designed to capture the complex nonlinear behaviour, with critical geotechnical and material parameters as inputs. The models were also highly predictive (R2 = 0.9951 in training and R2 = 0.9671 in testing), and CBR and UCS predictions were consistent, with R2 = 0.83–0.85, making them a more probable choice compared to other models. Residual analysis showed unbiased, stable model performance, whereas SHAP-based interpretation revealed CNT (%) and total nano content as the most influential parameters. The paper highlights the synergistic influence of Nano-Silica and CNT in improving soil strength and establishes a powerful AI-based predictive model. The results show that stabilising hybrid nanomaterials can enhance soil performance by more than 200%, providing a sustainable and efficient approach to treating expansive black cotton soil without relying primarily on laboratory tests.