大規模崩塌有著歷時性短、毀滅性高的特性,可能造成重要的保全設施的損壞與人員的傷亡。預測邊坡崩塌體運移的方法可略分為三類,分別為經驗模型、解析方法與數值方法。經驗模型係經由迴歸同類型崩塌體資料而來,需要收集與過濾大量崩塌體資料所得之迴歸模型方能可靠; 數值方法則需要掌握必要之輸入參數,且往往需要花較長時間建立模型與進行運移分析。本研究旨在建構簡化解析方法以進行運移分析,將崩塌體簡化為單質點,經由半解析方法建立模型,期能快速概估邊坡崩塌後之運移行為。 本研究共建置了三種質點運移模型,分別是單參數模型、雙參數模型與三參數質點運移模型。採用2012年暴雨導致南部六龜某邊坡滑動破壞為案例,先以PFC2D分析該案例的邊坡滑動運移,紀錄邊坡運移不同位置球體的位移歷程。再以PFC2D分析結果為標的,以各模型進行非線性擬合PFC2D模擬結果的位移歷程。透過非線性擬合的方式,可得對應PFC2D模擬結果的質點運移模型參數。最後並比較對應相同的滑動運移歷程下,簡化質點運移模型參數與PFC2D參數的概略關係。本研究建置之簡化運移模型,可用於快速概估邊坡崩塌後可能的運移影響距離與位移速度歷程,以供崩塌潛在災害之初步評估。
Large-scale landslides have the characteristics of short duration and devastation; these type of natural hazard may result in the damage of important facilities and cause human casualties. The prediction of the run-out distance of a landslide is one of key issues for disaster prevention. The methods for predicting the run-out of landslides can be divided into three categories: namely empirical models, analytical methods and numerical methods. An empirical model is usually constructed on the basis of field data of the landslide cases, its reliability replies on the correctness of the landslide data sets. The numerical methods for run-out analysis usually require a set of input parameters not necessarily easy to determine; also, the time for model construction and computation may be intensive. This study aims to develop a few simplified analytical models for landslide run-out analysis. In these simplified models, a landslide body is simplified into a lumped mass (mass point); the run-out simulation is carried out using a semi-analytical method. Three mass-point run-out models are developed in this thesis: (1) single-parameter model, (2) two-parameter model and (3) three-parameter of t model. Using a landslide case study caused by the 2012 heavy rainstorm in the southern Taiwan as an illustration, this study conducted a PFC2D simulation for the landslide case and recorded the displacement history of the run-out of the sliding body. Using the results of the PFC2D simulation as the calibrated target, this study then developed an optimization procedure to calibrate the model parameters of the simplified run-out models in order to match the simulated run-out results from the PFC2D simulation. Subsequently, this study explored the relations between the calibrated parameters of the simplified run-out models and the PFC input parameters for a same run-out history. The proposed simplified models may serve as a quick and approximate approach for the determination of a landslide run out and for the preliminary assessment of potential landslide disaster.