Abstract: As a fractal model, percolation method can analyze salt rock failure process and damage evolution under triaxial compression by creating a percolation model to locate the salt rock failure grid by using acoustic emission (AE). Percolation model shows that the slope of relationship curves between cluster number and the largest cluster occupancy for each specimen percolation model is basically equal after 70% of peak stress. The extension of the largest cluster along the specimen axially agrees with the evolution of cracks; and the critical failure probability is determined by the end of the largest cluster in axial direction, which is very important for the study of the sharply increased permeability point. The starting point of damage is determined by the correlation length of percolation, the average distance between two points in the same clusters, and the damage factor is calculated by percolation model failure probability which is accordance with the damage factor calculated according to AE parameters such as AE count numbers or AE energy . The research results show that percolation is a promising method to describe rock failure process and damage status so as to provide a novel approach for studying the rock failure and its cracks propagation.

Key words: salt rock, percolation, damage, cracks, acoustic emission (AE), correlation length