利用直接剪力試驗求取土壤的摩擦角是相當常見的方法，但傳統的直剪試驗不易觀察直剪盒內部在受剪過程的微觀破壞機制，且前人研究(Terzaghi、Peck，1948、Saada、Townsend，1981)提出直剪試驗在剪動面應力分布不均勻。因此，本研究利用離散元素法進行數值模擬，從中監測剪動過程直剪盒整體的應力分布與孔隙率，並分析剪動面上平均應力之變異程度，也針對剪動面特定區域觀察應力路徑與主應力平面角度旋轉情形。此外，以單一尺寸粒徑與顆粒堆加入顆粒群(clump)兩種方法，改善微觀參數提高到偏高之數值摩擦角也無法提高的現象。 由以上研究可知：(1)顆粒粒徑增大與加入顆粒群(clump)皆會提高摩擦角。(2)直剪模擬體積的剪脹多來自剪動帶。(3)剪應力尖峰值出現於剪動帶正中央。(4)以應力路徑來看，剪動帶正中央的應力分布比靠近邊界的應力狀態單純許多。(5)主應力平面旋轉角度從初始剪動到破壞時會逆時針旋轉約45度。(6)剪應力之變異係數遠大於水平與垂直應力之變異係數。 Direct shear test has been among common laboratory tests to obtain the engineering properties of soils, especially for dry granular soils. However, the microscopic behavior during direct shear test of this type of material is not always easy to monitor What’s more, the non-uniformity of the stress and strain can be developed along the failure plane have been brought up by several previous studies by Terzaghi and Peck (1948), Saada and Townsend (1981). Therefore, a discrete element method was employed to study the microscopic behavior of the dry granular material and understand the stress and porosity distribution during direct shear test. Variations of average stress, stress path and major principal plane along the pre-determined failure plane were also observed. Furthermore, even though the microscopic parameters gets unreasonable larger, the friction angle of the particle assembly only increases to a certain bounding value. In this study, the issue was solved by using particles with larger sizes and clump-models. The following summaries are addressed：(1) It is to say that the larger friction angle can also be by using the lager particles and the clump. (2) The dilation of the overall particle assembly comes from the dilation of the particles in the shear zone. (3) The maximum shear stress occurs in the middle of the pre-determined failure plane. (4)From the result of the stress path, the stress state in the middle of the shear plane is less complicated than that close to the edge of the shear box. (5) The directions of major principal plane are various between horizontal and 45 degrees in the counterclockwise direction until the test specimen has reached failure. (6)The coefficient of variation of the horizontal and vertical stress is much smaller than the shear stress.
