Scholarlink

<p>The current study investigates the ultimate bearing capacity (<i>UBC</i>) of an eccentrically inclined loaded strip footing of width <i>B</i>, positioned above an unlined horseshoe-shaped tunnel in a rock mass, following the generalized Hoek–Brown failure criteria. The reduction factor (<i>R</i><sub><i>f</i></sub>), which evaluates the impact of the tunnel’s presence on the decrease in the <i>UBC</i> of the strip footing, is determined using upper and lower-bound finite element limit analysis. The effects of several parameters, including load eccentricity (<i>e/B</i>), load inclination (<i>β</i>), normalized horizontal and vertical distances of the tunnel (<i>H/B</i> and <i>Z/B</i>), tunnel size, and rock mass strength parameters, on <i>R</i><sub><i>f</i></sub> are analyzed. The findings reveal that when <i>e/B</i> ≥ 0.30, and <i>β</i> ≤ 60°, the <i>R</i><sub><i>f</i></sub> &gt; 0.95, indicating that the presence of the tunnel has an insignificant impact on the <i>UBC</i> of the footing. In this scenario, the failure of the footing is primarily governed by load eccentricity and inclination rather than the tunnel’s presence. The critical <i>Z/B</i> at which <i>R</i><sub><i>f</i></sub> ≈ 1, varies with <i>e/B</i>. Specifically, for <i>e/B</i> = 0, <i>Z/B</i> = 2.5; for <i>e/B</i> = 0.1, <i>Z/B</i> = 2; and for <i>e/B</i> = 0.3, and 0.4, <i>Z/B</i> = 1.5, which remains constant for all considered values of <i>β</i>. The study also incorporates soft computing techniques to develop a numerical model for accurately forecasting <i>R</i><sub><i>f</i></sub>. An optimal artificial neural network architecture predicts <i>R</i><sub><i>f</i></sub> accurately with <i>R</i><sup>2</sup> = 0.99898. Additionally, regression techniques such as optimal ensemble tree and optimizable Gaussian process regression demonstrate robust predictive capabilities with high <i>R</i><sup>2</sup> (0.94 and 0.96) and low root mean square (0.04204 and 0.0017673) values.</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 Eccentrically Inclined Loading on Foundations Above Tunnels: A Combined Finite Element Analysis and Machine Learning Approach

  • 期刊论文

  • 发表日期：

    2025年1月6日

查看全文
  • Aayush Kumar,
  • Vinay Bhushan Chauhan,
  • Piyush Kumar

摘要

The current study investigates the ultimate bearing capacity (UBC) of an eccentrically inclined loaded strip footing of width B, positioned above an unlined horseshoe-shaped tunnel in a rock mass, following the generalized Hoek–Brown failure criteria. The reduction factor (Rf), which evaluates the impact of the tunnel’s presence on the decrease in the UBC of the strip footing, is determined using upper and lower-bound finite element limit analysis. The effects of several parameters, including load eccentricity (e/B), load inclination (β), normalized horizontal and vertical distances of the tunnel (H/B and Z/B), tunnel size, and rock mass strength parameters, on Rf are analyzed. The findings reveal that when e/B ≥ 0.30, and β ≤ 60°, the Rf > 0.95, indicating that the presence of the tunnel has an insignificant impact on the UBC of the footing. In this scenario, the failure of the footing is primarily governed by load eccentricity and inclination rather than the tunnel’s presence. The critical Z/B at which Rf ≈ 1, varies with e/B. Specifically, for e/B = 0, Z/B = 2.5; for e/B = 0.1, Z/B = 2; and for e/B = 0.3, and 0.4, Z/B = 1.5, which remains constant for all considered values of β. The study also incorporates soft computing techniques to develop a numerical model for accurately forecasting Rf. An optimal artificial neural network architecture predicts Rf accurately with R2 = 0.99898. Additionally, regression techniques such as optimal ensemble tree and optimizable Gaussian process regression demonstrate robust predictive capabilities with high R2 (0.94 and 0.96) and low root mean square (0.04204 and 0.0017673) values.