Numerical simulation of laboratory strength tests using a stochastic approach

Heterogeneity and discontinuity significantly affect rockmass strength. Past studies have largely used arbitrary scaling approaches to produce rockmass strength from laboratory rock tests. This paper proposes a stochastic approach to produce rockmass strength from laboratory rock tests. Based on the laboratory data, MATLAB with an extreme value stochastic model generated a database for each physico-mechanical property of the rock. Then finite-difference software FLAC2D simulated laboratory-sized rock specimens. Random material properties in MATLAB were generated using the grids developed in the numerical model. The random properties were then applied to the final FLAC2D model. Model runs simulated the approach performed in the laboratory. The results from the model indicate that a stochastic approach produced compressive strengths that were lower than those from the deterministic approach. Failure modes for each specimen were different from those observed in the laboratory. In addition, random density also influenced the failure mode, highlighting the importance of stochastic analysis in rocks.