Stability Analysis of an Undrained Slope with a Single Weak Layer

邊坡穩定為台灣重要的工程問題，並常使用極限平衡法 (LEM) 或有限元素法 (FEM) 等數值軟體分析其穩定性。然而，此過程往往十分耗時，故為了減少分析所花費的時間，常搭配使用穩定數分析，藉由幾何形狀判別和查圖的方式，能快速取得邊坡於均質情況下之安全係數與破壞行為，實為相當便利的分析方法。 但對於真實的邊坡，在破壞時常有沿弱層滑動破壞之情形，卻鮮有完整的分析成果，故本研究考量邊坡在含有弱層的情況下，將建立含弱層邊坡分析法，並搭配穩定數，以使分析法便於使用。 本研究分析法之訓練案例主要利用有限元素分析軟體ABAQUS 2D進行穩定分析，並為了解數值模擬的正確性，會先對均質不排水黏土邊坡(不含弱層)進行驗證。經模擬分析後，利用強度折減法 (SRM) 與無因次位移量求得安全係數，並與前人分析結果進行比對，以驗證邊坡模型和運算過程是否正確。 模型經驗證後，即進行邊坡含單一弱層的模擬，建立不同弱層深度、強度比及傾角之訓練案例，並分析其穩定性。經大量訓練案例模擬分析後，計算各案例之穩定數在均質邊坡與含弱層邊坡情況下之修正因子，並使用支持回歸向量機 (SVR) 與支持分類向量機 (SVC) 分析此修正因子與邊坡破壞行為，由此建立出完整的含弱層邊坡分析法 預期此分析法在未來使用上，能簡易利用均質邊坡穩定數圖，搭配本研究含弱層邊坡分析法，得到更貼近真實邊坡的分析結果，並預測出安全係數與破壞行為。Slope stability analysis is an important engineering problem in Taiwan. We usually use Limit Equilibrium Method (LEM) or Finite Element Method (FEM) to analyze the safety of slope. However, this analysis process could cost lots of time in finding the weakest section of the slope if we want to get more accurate results. Thus, improving efficiency of the analysis is one of the most important issue of the slope stability. For real slope, it is common to see a situation that the slope slides along a weak layer which shear strength is weaker than the soil around it and causes failure, but it still does not have complete study to explain the effect of this situation. Therefore, in this study, we would consider the slope with a weak layer and analyze the influence of the slope stability. In addition, we built a number of the training sample of the slope and use ABAQUS 2D, FEM numerical software, to analyze these sample. After analyzing all training sample, we would use these results to build an analysis method of the slope with a weak layer by correction factor of Ns, support vector regression (SVR), and support vector classification (SVC). In the future, we hope that the analysis method can not only estimate slope stability more easily and more quickly, but also understand the type and the effect range of the slope failure, and get more accuracy assessment with slope stability