Abstract: Based on different probability distribution functions， a stochastic planes simulation is applied in engineering rockmass. Three different probability distribution density functions: dual average density distribution， dual normal distribution and Fisher distribution are used to generate the stochastic structural planes. Meanwhile， a back analysis method is applied to acquire the distribution parameters based on spatial attitude. By inputting the data of deterministic structral planes and stochastic structural plane of a rock slope project in Liaoning Province， the slope modeling can be built and the key blocks can be identified by GeoSMA-3D(geotechnical structure and model analysis-3D). Then a discreteness index is put forward to evaluate the effectiveness and difference of different stochastic structural plane simulations. The results show that key blocks’ volume will increase when the discreteness index becomes smaller within a certain range.

Key words: stochastic structural plane, dual normal distribution, Fisher distribution, stereographic projection, discreteness