Landslide model set up and simulation using discrete element method

• Main Authors: Chan-Huan Weng, 翁晨桓
• Other Authors: 張國楨
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/2da2c5

碩士 === 國立臺北科技大學 === 土木與防災研究所 === 99 === This research takes Tsaoling landslide caused by Chi Chi Earthquake as an example, adopts from DTM data and also integrates the PFC 3D program of Particle flow discrete element method to analyze the sling mechanism and the landslide behaviors. To determine the affected area of landslide and to constraint the slid block, we use pre-disaster DTM and post-event DTM to build two topographic surfaces (upper topographic surface and lower topographic surface), thus the slid block between upper topographic surface and lower topographic surface could be defined. First of all, in this study several different methods are developed and tried to build the slidblock and the particle clusters to confine the slidblock, including NO SHADOW command Method, Pluviation Method, Triangular Prism Method, Gravity Elimination Method、Particle Detection Method, etc. The above-mentioned methods are compared according to the availability and feasibility. The Particle Detection Method is composed of several procedures: Firstly, prepare one dense and cubic particle clusters and then transpose to the mass center of the slid block; Secondary, the particle cluster should be rescaled in order to cover the entire slid block; Third, detect the particlesif the position is situated within the slid blocks confined by the upper and lower surface of the DTM before and after landslide, otherwise, the particle is been eliminated from the cluster. After the procedure, all there maining particles are bonded according to the material properties and served as the model for further landslide simulation. Therefore an almost realistic geometry concerning the thickness and shapes of slid block could be established, and the sliding surface also could be built from DTM data to match the real terrain features of topography. This study considers earthquake ground excitation by the strong motion acceleration data gather from the CHY080 seismostation situated just nearby the sliding area. Furthermore the existence of pore water pressure within the rock mass and the effect of friction coefficient attenuation caused by rock mass shaking and sliding are also integrated in this study. Compare the topography results from the simulation data with the real DTM, it says that the gained accumulation patterns of the simulation result could be mostly similar to real situation (DTM after landslide) at the friction coefficient of 0.087. The average difference between the DTM and PFC results is -3.2m and the standard deviation is about 17.5. When landslides occur, the largest sliding distance is about 3662 m, the maximum drop height is about 632m and the highest sliding speed is 87.2 m/s. The timing of maximum sling velocity coincides with seismograph of CHY080. The results shows that use the numerical method of three-dimensional discrete element particle flow can simulate the complete landslide process dynamically. Comparing with the limit equilibrium method and finite element method which are the most common to be used, the discrete element particle flow provides not only the states of instability but also the dynamical process of sliding. The result from this study states the effect of this simulation and states the inference of the related earthquake-sliding phenomena as well.