Scholarlink

<p>The random limit equilibrium method (RLEM) and the random finite element/difference method (RFEM/RFDM) are the two most commonly used methods for slope reliability analysis, but they often provide different reliability results for the same slope. As a result, the reliability of a specific slope would be uncertain, making geotechnical community face a new dilemma, which is obviously detrimental to the promotion and application of reliability analysis methods in geotechnical engineering. Indeed, such dilemma is mainly attributed to the different deterministic slope stability analysis models (or model error) being used by the two methods. Although a number of studies have explored the differences between the two models in traditional slope stability analysis, few studies have considered the impact of the model errors on slope reliability analysis, making consistent reliability analysis based on the two types of methods remain an open question. To fill the research gap, by leveraging the computational efficiency of RLEM and the computational accuracy of RFDM, this paper first proposes an efficient and consistent reliability framework for spatially variable soil slopes by considering model errors. Initially, the linear and bias models are used to quantitatively characterize the model error between RLEM and RFDM with the aid of parametric studies. Subsequently, the model error is incorporated into the RLEM by taking the RFDM as reference true model to form two new reliability analysis methods that consider the model error (denoted as ML-RLEM and MM-RLEM). The effectiveness and accuracy of the proposed methods are then demonstrated by two classical soil slopes through lots of parametric analyses against various statistics. The results indicate that the computational accuracy of the proposed ML-RLEM and MM-RLEM methods has been improved by an average of 34.49% and 58.13%, respectively, compared to the traditional RLEM; and the computational efficiency of the two methods has been improved by an average of 5.5 times compared to traditional RFDM.</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

Efficient and consistent reliability analysis of spatially varied soil slopes by considering model errors

  • 期刊论文

  • 发表日期：

    2025年4月11日

查看全文
  • Si-Qi Zhang,
  • Lei-Lei Liu

摘要

The random limit equilibrium method (RLEM) and the random finite element/difference method (RFEM/RFDM) are the two most commonly used methods for slope reliability analysis, but they often provide different reliability results for the same slope. As a result, the reliability of a specific slope would be uncertain, making geotechnical community face a new dilemma, which is obviously detrimental to the promotion and application of reliability analysis methods in geotechnical engineering. Indeed, such dilemma is mainly attributed to the different deterministic slope stability analysis models (or model error) being used by the two methods. Although a number of studies have explored the differences between the two models in traditional slope stability analysis, few studies have considered the impact of the model errors on slope reliability analysis, making consistent reliability analysis based on the two types of methods remain an open question. To fill the research gap, by leveraging the computational efficiency of RLEM and the computational accuracy of RFDM, this paper first proposes an efficient and consistent reliability framework for spatially variable soil slopes by considering model errors. Initially, the linear and bias models are used to quantitatively characterize the model error between RLEM and RFDM with the aid of parametric studies. Subsequently, the model error is incorporated into the RLEM by taking the RFDM as reference true model to form two new reliability analysis methods that consider the model error (denoted as ML-RLEM and MM-RLEM). The effectiveness and accuracy of the proposed methods are then demonstrated by two classical soil slopes through lots of parametric analyses against various statistics. The results indicate that the computational accuracy of the proposed ML-RLEM and MM-RLEM methods has been improved by an average of 34.49% and 58.13%, respectively, compared to the traditional RLEM; and the computational efficiency of the two methods has been improved by an average of 5.5 times compared to traditional RFDM.