利用小型振動台模擬邊坡受震情形之研究

Abstract

921集集大地震，不但造成台灣人民傷亡、房屋倒塌、更引發許多邊坡產生滑動破壞，為更深入了解邊坡受震時破壞發生機制，林美聆與王國隆自2003年起於國家地震中心進行了多組大型邊坡受震模擬試驗，以了解邊坡受震中之動態行為模式。為能於大型試驗進行前預先模擬，因此王國隆學長於2004年開發小型振動台。在本研究中針對其步進馬達輸出功率加以提昇，以使更適用於探討邊坡滑移機制。 本研究期望藉由小型振動台，有效地模擬出邊坡滑動之行為。在試驗過程中以改變邊坡特定因子坡高與坡角準備六種不同試體進行試驗，發現在以坡角改變時，觀察得邊坡滑動發展主要由坡頂開始向下滑移推擠下方的坡面造成整體性的坡面滑動。而在改變不同坡高時，因在振動時由於坡高高時坡面較不穩定因此破壞型式以坡面快速滑落現象為主。 另外，本研究以影像分析方法運用PIV分析坡面位移場關係，了解整體坡面在產生滑動破壞時，坡面位移場分佈變化情況。利用雷射掃瞄方式，對於坡面在試驗前後進行掃瞄動作，根據掃瞄所得試驗前後之坡面平均高程曲線來進行坡面變化定量分析，由初步成果來看，在坡面有明顯變化時，利用掃瞄技術確實可以進行定量分析之用。最後以數值軟體FLAC模擬物理模型試驗結果，由數值模擬結果來對照與PIV分析坡面滑動時時間與位移之關係，結果相當接近。

The Chi-Chi earthquake struck Taiwan in 1999 which induced numerous landslides.Using pseudo static and sliding block analysis for seismic slope analysis usually involve simplitied assumptions. In order to understand the behavior of slope subjected to earthquake, Wang (2003) used large shaking table to perform large scale model slope test in NCREE to observe the dynamic behavior of slope. The small shaking table was developed in laboratory in 2004 to facilitate the laboratory slope model test with the advantage of easy operation and specimen preparation. The aim of this study is to take advantage of the small shaking table to simulate and study the seismic slope behavior. The control factors such as slope height and slope angle have been changed during tests. The development of sliding initiated from the slope crest and moved toward the toe causing heave when the slope angle. Varies as the slope height increases , the slope became more unstable with rapid development of failure. The images of model slope behavior were recorded during tests and image analysis where performed using PIV to tract the displacement vectors when landslide initiated. The infrared rays were used to scan the surface of specimen, which measured the differences in geometry of specimen before and after test. The deformation of slope was measured based on the average height of two curves. The FLAC program was used to simulate displacements of the slopes in laboratory tests and a good agreement between the the results of numerical models and the measurements analyzed by PIV was achieved.