基于数值模拟与多场耦合的滑坡稳定性分析方法与破坏机理研究进展

Research Progress on Landslide Stability Analysis Methods and Failure Mechanisms Based on Numerical Simulation and Multi-Field Coupling

摘要: 滑坡稳定性分析是工程地质领域的核心科学问题。本文以2011~2026年发表的滑坡稳定性相关研究为重点，并补充纳入与研究主题密切相关的早期工程案例、经典理论及方法文献，从数值模拟方法、破坏机理及滑带力学特性三个维度进行综述。综述表明：(1) 数值方法已从单一极限平衡法向多平台耦合方向发展，GeoStudio-FLAC3D互馈耦合工作流将稳定性系数较单一平台降低10%~33%，双折减系数法揭示了黏聚力先于内摩擦角衰减的渐进破坏规律；(2) 渗流–剪切耦合效应被识别为水库型与降雨型滑坡的共性主导机制，渗流应力使滑带土峰值抗剪强度降低18%~35%，且内摩擦角的渗流敏感性显著高于黏聚力；(3) 正应力依赖性摩擦弱化在约200 kPa处存在机制转换——低应力下由黏土颗粒重定向主导，高应力下由颗粒破碎主导——这一发现为不同深度滑坡的差异化复活风险评估提供了细观力学判据。本文识别了实验–数值参数化衔接、方法基准缺失及预测验证赤字等五个关键研究空白，并提出相应的研究议程。

Abstract: Landslide stability analysis is a core scientific issue in the field of engineering geology. This review focuses on landslide stability studies published between 2011 and 2026, while also incorporating relevant earlier engineering case studies, foundational theoretical works, and methodological references. The literature is synthesized across three dimensions: numerical simulation methods, failure mechanisms, and the mechanical properties of slip-zone materials. The review shows that: (1) Numerical approaches have progressed from standalone limit-equilibrium methods to multi-platform coupling: the GeoStudio-FLAC3D mutual-feedback workflow reduces stability coefficients by 10%~33% relative to single-platform analyses, and the double reduction coefficient method reveals that cohesion degrades more rapidly than internal friction angle during progressive failure. (2) Seepage-shear coupling is the common governing process for both reservoir- and rainfall-induced landslides: seepage stress reduces peak shear strength of slide-zone soils by 18%~35%, with internal friction angle exhibiting higher seepage sensitivity than cohesion. (3) Normal-stress-dependent frictional weakening undergoes a mechanistic transition at approximately 200 kPa, shifting from clay-particle reorientation at low stress to grain crushing at high stress, which provides a meso-scale criterion for depth-differentiated reactivation risk assessment. We identify five critical research gaps, spanning the experimental-numerical parameterization disconnect, the absence of methodological benchmarks, and the deficit of prospective prediction validation, and outline a corresponding research agenda.