Scholarlink

<p>This study evaluates the effectiveness of Cement Deep Mixing (CDM) columns installed outside the excavation zone in reducing diaphragm wall displacement. The analysis considers a deep excavation in very soft clay, characterized by a highly compressible layer approximately 32&#xa0;m thick. A back-analysis based on the Finite Element Method (FEM) was first conducted to calibrate soil constitutive parameters using field monitoring data, thereby establishing a reliable numerical modeling framework. Subsequently, Artificial Neural Network (ANN) and eXtreme Gradient Boosting (XGBoost) models were used to evaluate the influence of key design parameters of the external CDM system on diaphragm wall displacement. The results show that the CDM block depth (L) is the dominant factor, followed by the horizontal distance to the diaphragm wall (a), the block width (B), and the elastic modulus (E). Based on the ANN model, an explicit predictive equation for estimating the maximum lateral displacement of diaphragm walls was developed, with a high coefficient of determination (R<sup>2</sup> = 99.5%). The proposed framework provides a quantitative basis for optimizing externally installed CDM systems and improving deformation control in deep excavations on soft ground.</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

Influence of Cement Deep Mixing Columns Used to Reinforce Roadbeds Adjacent to Deep Excavations on Diaphragm Wall Displacement: A Comparative Study Using FEM, ANN, and XGBoost

  • 期刊论文

  • 发表日期：

    2026年5月13日

查看全文
  • Trong Nghia Le,
  • Huynh Huy Nguyen,
  • Minh Trung Nguyen,
  • Kien Trung Nguyen

摘要

This study evaluates the effectiveness of Cement Deep Mixing (CDM) columns installed outside the excavation zone in reducing diaphragm wall displacement. The analysis considers a deep excavation in very soft clay, characterized by a highly compressible layer approximately 32 m thick. A back-analysis based on the Finite Element Method (FEM) was first conducted to calibrate soil constitutive parameters using field monitoring data, thereby establishing a reliable numerical modeling framework. Subsequently, Artificial Neural Network (ANN) and eXtreme Gradient Boosting (XGBoost) models were used to evaluate the influence of key design parameters of the external CDM system on diaphragm wall displacement. The results show that the CDM block depth (L) is the dominant factor, followed by the horizontal distance to the diaphragm wall (a), the block width (B), and the elastic modulus (E). Based on the ANN model, an explicit predictive equation for estimating the maximum lateral displacement of diaphragm walls was developed, with a high coefficient of determination (R2 = 99.5%). The proposed framework provides a quantitative basis for optimizing externally installed CDM systems and improving deformation control in deep excavations on soft ground.