The failure mechanism and influence area of granular soil slopes under dry and rainfall conditions

本研究藉由乾砂邊坡滑動破壞和降雨入滲引致邊坡破壞的模型試驗結果，配合能處理土壤顆粒互制行為與大變位的離散元素法(顆粒流數值分析)軟體進行模擬，以探討顆粒性土壤邊坡的破壞機制與破壞後的影響範圍，並依研究成果提出災害防治之建議。 乾砂邊坡滑動破壞試驗考慮細砂之相對密度、坡角與坡高，探討乾砂快速滑動行為。試驗時，藉由支撐砂土之擋板迅速拉起，使邊坡產生破壞，破壞面由土體滑動自然產生，同時利用質點影像測速技術，進行土砂運動過程中的影像分析，建立土砂顆粒運動過程之位移軌跡及質點速度。並將乾砂滑動行為簡化，以影像分析得之土壤顆粒前端速度代表，再依據牛頓運動定律與能量守恆方程式，推導得簡化之滑動距離及堆積長度預測公式，並與試驗值及案例比較，獲得不錯之結果。分析顯示，影響顆粒流前端速度之主要因子為邊坡坡角與土壤摩擦角，故邊坡破壞後之最終坡面可予以簡化，並以土壤內摩擦角φ表示之。 另一方面，藉由降雨入滲試驗探討高低降雨強度(287與78 mm/hr)入滲坡面引致邊坡破壞之機制。坡體土壤由不同細粒料含量(0~12%)組成，坡面採用常見之凹型及凸型坡，底層具一傾斜不透水層，以模擬破壞比例頗高之順向坡地層及上覆崩積土壤之情形。試驗時，觀察滑動土體滑移及堆積過程，並記錄土壤內含水量與孔隙水壓力之變化且加以分析探討，以瞭解邊坡破壞機制的特性。 試驗結果顯示，破壞機制以細粒料含量10%為界，分別為沖蝕及坡趾管湧的破壞現象。當細粒料低於10%之土壤，不同細料含量影響其初始破壞時間、破壞面長度，其原因為細料含量愈高，土壤內摩擦角較低，且單位重較高所致。岩盤面上之水壓分布，因滲流受傾斜岩層影響，以坡趾處之水力坡降最高，且水壓力甚至有高過坡面之現象。凹型坡較凸型坡破壞時間為短，而破壞長度較長，其與水力坡降及土層之厚度相關。值得注意的是，凸型坡在高/低降雨強度下的破壞長度比值與堆積長度比值，恰與高/低降雨強度的比值相近。若坡面土量足夠時，則高降雨強度將導致較多的土方下滑。 數值分析雙軸壓縮試驗得數值率定之微觀參數勁度比為(kn/ks=120)，但並不適用於乾砂滑動試驗，因真實運動型態有顆粒的滑動、轉動與碰撞等，與雙軸壓縮試驗之微小變形行為截然不同。由數值分析結果知，微觀參數以摩擦係數最為敏感：當kn/ks=1~15、摩擦係數為tanφ、臨界阻尼比依據材料回彈係數估算時，堆積長度之誤差在10%以內。將這些微觀參數用於現地案例的模擬，可合理求得堆積長度，並可估算衝擊力、滑動速度、與運動時間等。 有滲流之土坡主要受滲透係數(k)影響，而k值與顆粒的有效粒徑d10平方成正比。故應用數值分析程式模擬降雨入滲的破壞機制時，如模擬流速與滲透係數，須考量模型相似性之尺度因素，其中顆粒粒徑為重要影響因子。由數值分析結果知，土壤不飽和強度對降雨入滲邊坡之破壞機制影響很大，而基質吸力可以平行鍵結之微觀參數模擬之。 關鍵詞：邊坡、顆粒性材料、模型試驗、破壞機制、質點影像分析、PFC數值分析The purpose of this study was to understand the failure mechanism of granular soil slopes and the influence area resulting from the slope failures. Two kinds of model tests were carried out, i.e., sliding of dry granular soil slope and the slope failure caused by rainfall infiltration. Combined with the test results, numerical analysis was also employed and the software used was PFC (Particle Flow Code), which employs discrete element method to simulate the interaction of soil particles. For the sliding of dry granular soil slope, tests were conducted under various combinations of slope angle, slope height, and the relative density of soil. The displacement of the soil was measured by a particle image velocimetry (PIV). These series of images clearly displayed the movement and development of granular flows. The final profile of the soil and the shape of deposition area were then measured. Based on the measured displacement of flow, the front velocity of the flow was obtained. Among various parameters, slope angle and the internal friction angle of soil are found to be the most significant factors influencing the front velocity. Furthermore, a simplified equation based on Newton’s law of motion and energy conservation is developed to predict the run-out of flow. Comparison is then made between the measured and predicted run-out distances, and they are generally in good agreement. Finally, the profile of slope after failure can be simplified and represented by the friction angle of the soil. For slope failure caused by rainfall infiltration, the model test was conducted on slopes with convex and concave profiles, and the slopes were non-homogeneous and had a dip stratum underneath. In addition, different fines contents (0-12%) and rainfall intensities (78 and 287 mm/hr) were considered as variables. During the experiment, variations in pore water pressure and volumetric water content in the soil were measured. The characteristics of the failure mechanism and the responses of pore pressure and water content in the model slopes are compared and discussed as functions of variables. For samples with fines content of 12%, failure was more likely initiated by surface erosion. In contrast, for samples with fines content less than 10%, slope failures were initiated near the toe. The failure length and the time when piping was observed also varied with fines content and rainfall intensity. The reason could be due to the samples with more fines content having lower friction angle and higher unit weight. The profiles of total head above the impervious stratum were generally nonlinear and the pressure head near the toe of the slope could be even higher than the equivalent head of overburden pressure. Moreover, the initial failure in concave slopes occurred sooner than that in convex slopes; it was attributed to the thin soil layer near the toe of the slope, so that a high hydraulic gradient was induced to cause piping to be initiated. It is also note worthy that these ratios for failure length and run-out distance under high and low rainfall intensities approximate the ratio of rainfall intensity. This result also implies that the rainfall would have caused as much soil to fail as it could if there was enough soil. In the sensitivity analysis on micro-parameters, the coefficient of friction of the soil was found to be the most critical. Furthermore, with the proposed values of the micro-parameters from this study, the error in predicting run-out distance is less than 10%. The simulation of a real slope failure was also in good agreement. Hence, this numerical model is suitable to be applied in real cases to estimate the velocities of a moving mass, its impact force on structures, and the travel distance and time of the soil mass. The stability of a slope under seepage is mainly influenced by the permeability of the soil, which in turn is proportional to the square of the effective size of the soil particle. Therefore, geometrical similarly must be considered in the numerical analysis on the seepage velocity and the permeability because the size of soil particle is a dominant factor. The result from the numerical analysis shows that the strength of unsaturated soils has great influence on the failure mechanism of the slope, in which the matric suction can be simulated by the micro-parameter of parallel bonds. Keyword: slope, granular soil, model test, failure mechanism, PIV analysis, PFC numerical analysis