PREDICTION OF NONLINEAR STRUCTURAL EARTHQUAKE RESPONSE BASED ON HYBRID METHOD CONSIDERING LOCAL SEISMIC ACTIVITIES AND UNDERGROUND SOIL CHARACTERISTICS

为研究1995年兵库县地震期间神户市重灾区地震动的放大特征，采用直接2.5维边界元方法计算了三维Haskell型断层引起的二维沉积盆地中的强地面运动。详细研究了不同类型的沉积盆地断面对盆地内强地表运动的影响，讨论了拉普拉斯方程中裂隙问题的三维快速多重边界积分方程法。该方法利用Hrycak和Rokhlin提出的新的多极子展开式，结合配置求解裂纹问题的离散化超奇异边界积分方程组。用广义最小残差法(GMRES)结合快速多极子方法(FMM)求解得到的数值方程。结果表明，当未知数大于1300时，该程序比常规程序更快。本文提出了一种计算桩顶水平激励时变非线性动土阻力的方法。在该程序中，只假定桩上某些点的弯矩是通过观测预先给出的。特别考虑了一种数据处理方法，该方法结合了对给定数据的平滑过程和引入正则化与惩罚矩阵，以提高从包含测量误差的观测数据估计的精度。通过数值算例验证了该方法的有效性。

A direct 2.5 dimensional boundary element method is applied to evaluate strong ground motions in two-dimensional sedimentary basin due to a three-dimensional Haskell-type fault to estimate the amplification characteristics of the ground motions in the heavily damaged belt zone in Kobe City during the 1995 Hyogo-ken earthquake. The effects of different types of cross-section of the sedimentary basin on the strong gound motions in the basin are examined in detail.We discuss a three-dimensional fast multiple boundary integral equation method for crack problems in Laplace's equation. The proposed implementation uses a new multipole expansion proposed by Hrycak and Rokhlin in conjunction with collocation in the solution of a discretised hypersingular boundary integral equation for crack problems. The resulting numerical equation is solved with GMRES(generalised minimum residual method) in connection with FMM(fast multipole method). It is found that the obtained code is faster than a conventional one when the number of unknowns is greater than about 1300.We present a method to evaluate time-varying nonlinear dynamic soil resistance to a pile when subjected to horizontal excitations on the top of the pile. In the procedure only bending moments at some points on the pile are assumed to be given in advance by observation. Special consideration to a data processing method, which combines a smoothing procedure for given data and introduction of regularization with a penalty matrix, is made for improving the accuracy of the estimation from the observed data including measurement errors. The validity of the proposed procedure is studied and confirmed through numerical examples.