Deformation and Failure Charateristics of Tunnel Excavation Based on GPU Accelerated Random Field

The random distribution of parameter of the surrounding rock have obvious effect on the stability and this paper investigates the influence of the random spatial distribution of rock mass parameters on the stability of surrounding rock. A random field generation method based on the covariance matrix decomposition and GPU acceleration method is proposed. A function that transforms the random field to IMASS constitutive model parameter were successfully established. Afterwards, the effect of the random field under different conditions and the influence on the subsequent simulated excavation were compared. The results show that the random field generated by the Exp correlation function is relatively stable and scattered, and its maximum displacement value varies within a small range; the random field generated by the SExp correlation function is more compact, with a larger fluctuation range of the maximum displacement value, and more dangerous displacements occur occasionally. It can be found from the X, Y correlation length that the distribution of the maximum displacement points is similar. During excavation, the middle, bottom and vault on both sides of the chamber are dangerous areas, and the damage locations are mainly concentrated in the cavern side wall, the junction of the vault and the middle of the side wall.