Probabilistic ground modelling and slope stability assessment based on random field theory

Read full text Summarise with VIRAT
Authors: Sheng Zhang
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Spatial variability in soil properties and subsurface stratigraphy is a major source of uncertainty in geotechnical analyses, posing challenges to reliable geohazard risk assessment. As geotechnical engineering increasingly adopts uncertainty aware approaches, there is a growing need for probabilistic subsurface characterization that explicitly accounts for these uncertainties and supports reliability based design. Grounded in random field theory, this thesis advances probabilistic ground modeling and explores its application to slope stability assessment. A Bayesian framework is first developed to integrate geotechnical and geophysical investigation data by exploiting both intra-site and inter-site correlations. The framework effectively reduces uncertainty by jointly accounting for transformation, statistical, and spatial variabilities within a unified probabilistic formulation. A generic random field approach is then proposed to quantify geological uncertainty, wherein the spatial existence of each formation is represented by a latent random field. Stratigraphic dips are interpolated and incorporated into the model, enabling more realistic simulation of subsurface stratigraphy. The random field–based characterization is further applied to probabilistic slope stability analysis, where stochastic stratigraphic realizations generated from sparse borehole data are mapped to finite element models. Monte Carlo simulations are performed to estimate reliability indices and mobilized soil mass under different geological conditions. Finally, the random field framework is extended to landslide susceptibility updating, where spatially variable soil parameters are modeled using Bayesian networks and calibrated against observed landslide inventories. Overall, this thesis establishes an integrated probabilistic framework for ground modeling and slope stability assessment, promoting more reliable and resilient geotechnical design and risk mitigation.

Fields of Study & Subjects

Geotechnical Engineering and Analysis Probabilistic and Robust Engineering Design Geotechnical Engineering and Soil Mechanics

Publication details

Journal / source
Warwick Research Archive Portal (University of Warwick)
Publication date
17 September 2026
Language
EN
Document type
Thesis