Abstract
Evaluation of the maximum resistance and corresponding pile-head displacement of a landslide prevention pile in arbitrary landslide mass, based on a few significant factors such as the geometrical shape of the landslide mass, the construction position of the pile and the mechanical properties of the ground, is very effective in design. Such a solution may be directly linked on a cost cut and safety disposition of landslide prevention projects. In this study, a parametric study of a simple landslide model based on the 3-dimensional elasto-plastic finite element method (FEM) is carried out. In each analytical case, the maximum resistance and corresponding pile-head displacement of a landslide prevention pile is estimated and a consideration about the relationship between those obtained values and various mechanical conditions is made. Moreover, some important indications are examined from comparison with the solutions based on the present design code.
Finally, in order to save the time cost spent on the 3-dimensional FEM, the neural network assembled based on the FEM results is built, and the idea for which such a method is used in business is proposed. Compared with the amount of work accompanying the analysis processing by FEM, execution of the neural network is much lighter. This induces the decisive difference at the time of examining the application possibility to a design. In this study, the possibility of development of the computer-aided design system of landslide prevention pile on network computing is shown.
