比較兩種循環荷載組成律模式於可液化地盤之分析-以地下管線上浮為例

Abstract

因應台灣城市化和人口快速增長民衆生活水平重要性日益突出，地下維生管線與交通隧道等作爲民衆生活機能以及交通運輸等公用設施，其安全性和可靠性對於城市和非城市地區的正常運行至關重要。然而，台灣處於歐亞板塊與菲律賓板塊交界處的位置，特殊的地理環境導致地震發生的機率極爲頻繁，土壤液化等地震災害事件隨之而來。因此，土壤液化對於地下維生管線的影響相關研究十分重要。 本研究使用有限元素法軟體PLAXIS 2D 平台内建之PM4Sand與UBC3D-PLM兩種循環荷載組成律模式，進行地下管線於可液化土層之動態數值模擬分析。首先，使用Python 直譯器模擬不排水試驗的單元分析 (Cyclic direct simple shear, CDSS)，以擬合真實試驗之抗液化曲線為目標，來進行組成律模式參數率定與敏感度分析，其率定之結果作爲後續數值模型土層之使用參數。本研究之分析分別爲無埋設地下管線與有埋設地下管線兩種數值模型。無埋設地下管線數值模型的目的是驗證所選擇的兩種組成律模式能否準確模擬土壤液化行為，並通過與試驗結果比對加速度和超額孔隙水壓的時間歷時進行驗證。根據無管線模型試驗的比對結果，PM4Sand和UBC3D-PLM都能有效地捕捉到液化趨勢。因此，本研究進一步分析埋設地下管線數值模型的案例，並比較兩種組成律模式對於管線上浮量的預測能力。 比較兩種組成律模式所擬合之試驗抗液化曲線，PM4Sand模式表現優異，能夠較準確地模擬出超額孔隙水壓比與單軸向應變3%時的情況。而UBC3D-PLM模式的表現較差，僅可模擬出近於超額孔隙水壓比，但無法達到單軸向應變3%。經過一系列試驗加速度與水壓歷時結果比對，PM4Sand模式相對於UBC3D-PLM模式更接近試驗結果的趨勢，可模擬液化發生時水壓激發與加速度縮小的變化。從管線上浮量之模擬結果可發現，PM4Sand和UBC3D-PLM兩種模式均有低估上浮量的現象。 關鍵字：土壤液化、循環荷載組成律模式、參數率定、動態數值模擬、管線上浮。

In response to the increasing importance of urbanization and rapid population growth in Taiwan, the significance of improving people’s living standards has become more prominent. Underground infrastructure, such as utility lines and transportation tunnels, serves as vital components of public facilities and transportation, playing a crucial role in ensuring the safety and reliability of operations in both urban and rural areas. However, Taiwan’s unique geographical location at the junction of the Eurasian Plate and the Philippine Sea Plate leads to a high probability of frequent seismic activities, which are often accompanied by earthquake-related disasters like soil liquefaction. Therefore, it is essential to study soil liquefaction’s impact on underground infrastructure. This study investigates two cyclic constitutive models for dynamic numerical analyses in PLAXIS 2D: PM4Sand and UBC3D-PLM. Firstly, the parameter calibration process of the consolidated undrained Cyclic Direct Simple Shear Test (CDSS) is simplified using a Python compiler. Based on liquefaction strength curves, parameter sensitivity analyses and model parameter calibrations are conducted for the two specified constitutive models. The calibrated results serve as the basis for subsequent analyses. Secondly, two numerical models are simulated, one without a buried underground pipeline and the other with a buried underground pipeline. The numerical results are then compared with the results of shaking table tests. Through these numerical models, the comparison focuses on the acceleration, excitation of excess pore pressure, and prediction of pipeline uplift with the shaking table test results. The conclusions can be drawn in the following. During the parameter fitting for the consolidated undrained CDSS test, the UBC3D-PLM model fails to achieve a cyclic axial strain of 3% and shows a fixed oscillation phenomenon, rendering it hard to determine the number of cycles for the liquefaction strength curve. The PM4Sand model offers good simulations in the element tests. This constitutive model in PLAXIS 2D also can generate a similar trend in acceleration oscillation and pore pressure excitation compared to the shaking table test results without a pipeline. The PM4sand model outperforms the UBC3D-PLM model for simulating the liquefaction strength curve and the model without considering a pipeline. However, compared to shaking table results considering a pipeline, both constitutive models underestimate pipeline uplift displacement.

Keywords: Soil liquefaction, cyclic constitutive models, parameter calibration, dynamic numerical simulation, pipeline uplift.