Scholarlink

<p>This study investigates the ultimate bearing capacity (UBC) of shallow foundations resting above single and dual subsurface cavities in cohesive–frictional soils. Adaptive Finite Element Limit Analysis (AFELA) is employed for the numerical investigation. The soil behaviour is characterized using the Mohr–Coulomb criterion, and the influence of shear strength parameters (c′, ϕ′), the unit-weight factor (γB/c′), and cavity geometry expressed through normalized offset (M/B), depth (N/B), and spacing between dual cavity (S/B) is assessed through the stability number (N<sub>s</sub>). The proximity of the cavity is shown to markedly reduce N<sub>s</sub>, with the most critical condition occurring at M/B = 0. The adverse influence diminishes progressively and becomes negligible beyond M/B = 3–4.5 and N/B = 3.5–4.5, depending on soil friction, indicating reduced vulnerability to void-induced settlements and sinkhole-type geohazards at greater embedment depths. The parameter γB/c′ significantly governs the stability response. Low γB/c′ values (γB/c′ = 0.5–1.0) produce stable and convergent solutions, whereas high γB/c′ values (γB/c′ = 2) leads to non-convergence in low-ϕ′ soils due to self-weight-controlled failure. For dual cavities, significant interaction occurs for S/B &lt; 2, whereas stability is restored when the spacing increases to S/B ≥ 3. Higher friction soils (ϕ′ = 15°–20°) consistently exhibit enhanced N<sub>s</sub> and reduced sensitivity to cavity location, whereas low-ϕ′ geomaterials remain susceptible to instability under natural hazard scenarios such as rainfall softening, subsidence, or seismic disturbances. Machine-learning evaluation based on the FELA dataset confirms the superior predictive performance of nonlinear models, particularly ANN and GPR, achieving R<sup>2</sup> &gt; 0.99.</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

Assessment and Mitigation of Cavity Induced Sinkhole Effects on Shallow Foundations Using FELA and Machine Learning

  • 期刊论文

  • 发表日期：

    2026年6月22日

查看全文
  • Piyush Kumar,
  • Anupam Kumar Singh,
  • Aayush Kumar

摘要

This study investigates the ultimate bearing capacity (UBC) of shallow foundations resting above single and dual subsurface cavities in cohesive–frictional soils. Adaptive Finite Element Limit Analysis (AFELA) is employed for the numerical investigation. The soil behaviour is characterized using the Mohr–Coulomb criterion, and the influence of shear strength parameters (c′, ϕ′), the unit-weight factor (γB/c′), and cavity geometry expressed through normalized offset (M/B), depth (N/B), and spacing between dual cavity (S/B) is assessed through the stability number (Ns). The proximity of the cavity is shown to markedly reduce Ns, with the most critical condition occurring at M/B = 0. The adverse influence diminishes progressively and becomes negligible beyond M/B = 3–4.5 and N/B = 3.5–4.5, depending on soil friction, indicating reduced vulnerability to void-induced settlements and sinkhole-type geohazards at greater embedment depths. The parameter γB/c′ significantly governs the stability response. Low γB/c′ values (γB/c′ = 0.5–1.0) produce stable and convergent solutions, whereas high γB/c′ values (γB/c′ = 2) leads to non-convergence in low-ϕ′ soils due to self-weight-controlled failure. For dual cavities, significant interaction occurs for S/B < 2, whereas stability is restored when the spacing increases to S/B ≥ 3. Higher friction soils (ϕ′ = 15°–20°) consistently exhibit enhanced Ns and reduced sensitivity to cavity location, whereas low-ϕ′ geomaterials remain susceptible to instability under natural hazard scenarios such as rainfall softening, subsidence, or seismic disturbances. Machine-learning evaluation based on the FELA dataset confirms the superior predictive performance of nonlinear models, particularly ANN and GPR, achieving R2 > 0.99.