A random discrete element method for modeling rock heterogeneity

Due to variations in mineral composition and formation history, rock properties are characterized by heterogeneity and spatial variability. In this paper, a random discrete element method was used to model rock heterogeneity. A parallel bond model was used to simulate the connection between particles and the bond strength was spatially varied to account for the rock heterogeneity. The spatial variabilities of uniaxial compression strength (UCS) and failure mechanism of two natural rocks were analyzed using stochastic analysis. The results have indicated that the used method properly reproduces both the UCS uncertainties and the failure mechanism. Compared to deterministic analysis, the stochastic analysis yields a smaller mean UCS and more complex failure patterns. With an increase in the coefficient of variation of the bond strength, the rock becomes more fragmented, causing a mean UCS to decrease with the greater discrepancy, along with failure mechanisms with multiple fracturing. The mean UCS is minimized and macrocracks are the most discrepant when the scale of fluctuation (SOF) is approximately half the sample width. The UCS discrepancy increases with SOF until it is 20 times greater than the sample width; beyond that the SOF effect becomes negligible.

Automated summary

In this paper, a random discrete element method was used to model rock heterogeneity, and stochastic analysis was used to investigate the spatial variability of compressive strength and failure mechanisms in two natural rocks. The results show that stochastic analysis provides more accurate predictions of uncertain rock properties and complex failure patterns.