Abstract: In order to achieve 3D visualization simulation of rock fragmentation prediction，the key techniques were proposed.According to the distribution regularity of rock mass joint geometrical parameters，including joint inclination，dip angle and joint spacing，Boost Random library was used to realize the Monte Carlo random simulation of joint surface，so as to establish joint sketch file for original fragmentation prediction，and then the correction coefficient of caved fragmentation was introduced to create the joint sketch file for caved fragmentation prediction.3D joint surface network was calculated according to the data of joint sketch file.Then the joint network was used for creating the 3D fragmentation model by cutting the cuboids model that covering the simulation region.Volume coefficient and colinearity of each block were calculated by 3D fragmentation model and used for block shape classification.The algorithms above were implemented with VTK library and HOOPS 3D.Fragmentation simulation for block caving method was performed at Pulang copper mine to study the fragmentation distribution law at initial mining district.The quartiles of the block volume are 0.164 m3，0.403 m3 and 0.853 m3，respectively，and when the fragmentation equivalent size is in the range of 0.5~3.0 m，the blocks occupy a larger volume percentage.Furthermore，the size of draw hole was proposed as 5 m×5 m according to fragmentation equivalent size distribution curve.The result show that the method can simulate the distribution of joints in rock mass according to the survey results of on-site joints，and then prediction and statistical analysis the fragmentation are made by 3D fragmentation model.It provides an important basis for the study of block caving mining method.

Key words: block caving method, fragmentation prediction, Monte Carlo simulation, joint network, fragmen-tation model, VTK, block shape classification