Date of Award


Degree Type


Degree Name

Master of Science in Civil Engineering (MSCE)


Civil Engineering

First Advisor

Philippe L. Bourdeau

Committee Chair

Philippe L. Bourdeau

Committee Member 1

Antonio Bobet

Committee Member 2

Daniele Perissin


Many communities around the world have been established in areas of ongoing, as well as ceased, underground mining activity. Ground movements induced by ore extraction methods and the collapse of abandoned cavities have long been recognized as a hazard to surface structures. A number of approaches have been proposed for the prediction of subsidence in underground mining regions, and their integration to Geographic Information Systems (GIS) can produce a powerful risk management tool. Nevertheless, this application is often limited by either a lack of generality or excessive computational cost of the methods available.

In this work, the stochastic subsidence model proposed by Litwiniszyn (1964) was investigated. Conceptually, the model assumes the ground mass as a discontinuous medium, in which particle displacement towards a collapsing cavity is treated as a Markovian process. The accumulation of the discrete movements amounts to the Komolgorov diffusion equation which is then employed to compute surface displacements.

In order to gain better understanding of the mechanism at granular scale and test the stochastic diffusion model in controlled conditions, subsidence in a granular medium was simulated via the Discrete Element Method (DEM). Using a frictional-elastic constitutive law for inter-particle contact, large three-dimensional assemblies of gravel-size grains were generated with a range of microstructural and bulk properties, these were then subjected to trapdoor experiments. In each simulation, particle displacements, ground surface deflections, as well as stresses and changes in the granular matrix structure were monitored and provided detailed information about the phenomenon.

The behavior of the granular matrix undergoing subsidence was shown to be highly dependent on both its microstructural and bulk properties. A thorough evaluation of the impact of porosity, particle size dispersion, inter-particle friction and contact stiffness on behavior of the material is presented. The parameters for the stochastic model were calculated based on the displacements obtained from DEM. The stochastic diffusion model and the DEM experiments were found in very good agreement for medium-dense simulated deposits, while in more contrasted types of either loose or dense materials, discrepancies were observed.