An overview of discrete element modeling for slope stability

Slope failures are among the most damaging geohazards worldwide, causing not only loss of life and infrastructure but also severe and often irreversible environmental consequences: destruction of vegetation cover and arable soil, blockage and contamination of watercourses, increased downstream sedimentation, and long-term disturbance of mountain and coastal ecosystems. Reliable stability assessment is therefore both an engineering and an environmental priority. The Discrete Element Method (DEM) has emerged over the past two decades as a powerful complement to continuum approaches for the analysis of slope stability, offering direct access to grain-scale mechanisms — bond breakage, fabric evolution, force chain reorganisation, and progressive failure — that remain difficult to resolve with the Finite Element Method (FEM). This paper provides a critical overview of DEM applications to geotechnical slope stability, covering: (i) the theoretical foundations of particle dynamics and contact constitutive laws, with a systematic comparison of models applicable to granular sands, cemented rocks, and cohesive clays; (ii) micro-to-macro calibration strategies, from Hertz–Mindlin elastic contacts to the Parallel Bond Model (PBM) and critical-state frameworks such as the Modified Cam-Clay (MCC); (iii) the Shear Strength Reduction (SRM) method as reformulated at the micro scale, including an exact Bolton-based micro-to-macro mapping showing that proportional reduction of contact friction does not, in general, produce the same proportional reduction of macroscopic strength; importantly, the sign of the micro–macro safety-factor difference is shown to be state-dependent rather than universal. For cohesive bonded assemblies, a damage-dependent framework is introduced in which cohesion and friction are mobilised at different stages of bond degradation, and an explicitly multivariable safety-factor mapping is formulated without assigning a universal sign; (iv) GPU-accelerated and multi-scale (DEM–FEM) implementations enabling slope-scale simulations; and (v) emerging DEM–machine learning surrogates for probabilistic stability assessment. Drawing on the literature surveyed here, the review identifies seven open challenges, including the absence of a standardised MCC-based DEM calibration protocol for cohesive soils, the scarcity of DEM back-analyses against documented natural landslides in developing countries, and the need for hydro-mechanically coupled frameworks under transient pore-pressure loading. A macro-scale proof-of-concept back-analysis of a landslide in northern Morocco illustrates both the potential and the present limitations of the framework. The review provides a roadmap for researchers aiming to combine the grain-scale resolution of DEM with the engineering reliability demands of slope stability practice, identifying the conditions under which DEM genuinely resolves micro-mechanical phenomena rather than acting as a macro-scale continuum surrogate.