摘要 本研究採用Ohsaki所提出的相位角產生方式和三角級數法，透過改變產生亂數的種數，建立十個能包絡日本兩種不同功能的岩盤設計反應譜的人工地震。所產生的人工地震則用來進行盆地地形和側向非均勻土層的彈塑性受震反應分析，並與相對應的一維土層模式的結果做比較。之後再以一維和二維土層模式所得的地表地震動進行一棟十樓建築物歷時分析，以最大層間變位角為參數，探討結構物受震反應。 研究的結果發現，在盆地地形的分析例中，於強震作用下，使用一維模式將會低估地表地震動反應與最大層間變位；而受到弱震作用時，在邊坡場址使用一維土層分析的結果比較保守，但在盆地中央則會低估地表受震反應。在側向不均勻土層例中，二維土層模式的地表受震反應大於一維土層模式。產生人工地震的種數對於層間變位角有相當大的影響；此外，規範中建議的使用多個人工地震產生的最大層間變位角之平均值來代表結構性能的作法，會受到所選取的樣本的影響，有必要做進一步的檢討，我們的建議是應該要加上標準偏差的規定。 ABSTRACT In this study, the concept proposed by Ohsaki to generate artificial earthquake is adopted. By varying seed numbers, ten artificial earthquakes which envelop Japan two-level bedrock design spectrum are generated. These ten artificial earthquakes are then used as input for the response analysis of a two-dimensional basin, a soil stratum with lateral inhomogeneity and associated 1D model. Each of the computed surface ground motion are then served as input to a ten-story steel building and the maximum story drift ratio are computed. For a basin subjected to strong bed-rock motion, the 1D model underestimates the surface motions and the maximum story drift ratio of the building; however, when the bed-rock motion is weak, the 1D analysis gives higher responses near the edge of basin, while the response of this trend is reversed in the middle of the basin. The results of 2D analysis are larger than those of 1D model on soil stratum with lateral inhomogeneity. The seed number has significant effect on maximum story drift ratio. On the other hand, the results of currently adopted average process given in the code are affected by how samples are chosen and a clear guideline should be developed. We suggested that a specification on Standard Deviation should be imposed.
