The Karhunen-Loeve (KL) expansion and probabilistic collocation method (PCM)
are combined and applied to an uncertainty analysis of rock failure behavior by integrating a self-
developed numerical method (i.e., the elastic-plastic cellular automaton (EPCA)). The results
from the method developed are compared using the Monte Carlo Simulation (MCS) method. It
is concluded that the method developed requires fewer collocations than MCS method to obtain
very high accuracy and greatly reduces the computational cost. Based on the method, the elasto-
plastic and elasto-brittle-plastic analyses of rocks under mechanical loadings are conducted to
study the uncertainty in heterogeneous rock failure behaviour.