Abstract: The characters of FLAC3D are studied and compared with finite element method. FLAC3D method uses node motion equation and traces the medium motion process from being loaded to reaching equilibrium state. While the finite element method solves from mechanical equilibrium equation directly. The above is the basic difference between the two methods. Because FLAC3D does not use the real medium damping property and node mass, the real process experiencing by the medium is not given and the influence of medium motion process on result cannot be reflected rightly. So the physical meaning of stress and deformation given by FLAC3D is not very clear. The medium vibration during solving process is a noise rather than a more real simulation to medium mechanical behavior and could cause calculation error for elastoplastic medium. The study shows that FLAC3D has a very high precision in determining the stress and deformation of elastic medium at equilibrium state and the vibration in solving process almost does not affect the final result. The advantages of FLAC3D method are that its mathematical operation is simple, and the solving process is convergent, and it is convenient in treating cases of medium large deformation, elastoplastic property, non-associated flow rule, excavation and supporting, etc. compared with finite element method. The defects of FLAC3D are that its calculation time is long, and the efficiency is low, and in some cases the number of time steps is high amazingly which may result in accumulating error. Studies in four aspects are also made through examples. (1) Comparisons between Drucker-Prager yield criterion and Mohr-Coulomb yield criterion. (2) Influence of dilation angle value on calculation result. (3) Influence of large deformation mode and small deformation mode on calculation result. (4) Influence of calculation precision setting on result.

Key words: rock and soil medium, Lagrangian numerical analysis method, finite element method, yield criterion, dilation angle