Effects of Stochastic Variability of Soil Properties on Liquefaction Resistance

9523092尽管在过去30年中进行了广泛的研究，但土壤液化现象仍然是造成大地震灾难性破坏的主要原因之一。本项目的目的是为了更好地了解土壤液化的机理，旨在提出更符合实际的液化风险评估方法和适当的抗震设计对策，并强调土壤性质所表现出的空间变异性的重要性。天然土壤的性质多种多样。野外试验结果的稀缺性以及土壤性质表现出的无序程度，导致使用统计方法来描述这些性质的分布。在此背景下，将各种地质力学性质模拟为多变量、多维(MV-ND)随机场的分量。随机场的特征是基于统计上有意义的“现场”测量结果集来评估的。采用DYNAFLOW程序中建立的多屈服本构模型，对水平成层砂土以及可液化地基上的结构在地震荷载作用下的行为进行了二维和三维有限元模拟。将用随机输入本构参数得到的预测超孔压累积、液化指数和液化变形的数值模拟结果与确定性输入模拟结果进行了比较，并对空间相关距离、土性概率分布、各种土性之间的相关性、有限元离散网格大小以及二维简化和真实三维行为的影响进行了检验和量化。通过与过去地震记录的“现场”行为的比较，验证了本构参数校准方法和分析程序的性能。将建议更新基于可液化土壤的结构的现行分析和设计程序，以考虑土壤特性的空间变异性程度。这项研究将为液化风险评估和表征土壤性质的空间变异性提供良好的基础。***

9523092 Prevost Though extensively studied during the last 30 years, the phenomenon of soil liquefaction is still one of the main causes of catastrophic damage during major earthquakes. The aim of this project is to obtain a better understanding of soil liquefaction mechanisms, aimed at more realistic liquefaction risk assessment methods and appropriate seismic design countermeasures, and emphasizing the importance of spatial variability exhibited by soil properties. Natural soils are variable in their properties. The scarcity of field test results, and the degree of disorder exhibited by soil properties, leads to the use of statistical methods to describe the distribution of those properties. In this context, the various geomechanical soil properties are modeled as the components of a multi-variate, multidimensional (mV-nD) stochastic field. The characteristics of the stochastic field are evaluated based on statistically significant sets of "in- situ" measurement results. Two- and three-dimensional finite element simulations of the behavior of horizontally layered sandy soil, as well as of structures founded on liquefiable soil and subjected to seismic loads, are performed using the multi-yield soil constitutive model implemented in the computer code DYNAFLOW. The numerical simulation results for predicting excess pore- pressure buildup, liquefaction index, and liquefaction induced deformations -- obtained using stochastic input constitutive parameters -- are compared to deterministic input simulation results; and the influences on these results of spatial correlation distances, probability distributions of soil properties, cross-correlations between various soil properties, finite element discretization mesh size, as well as the effects of the 2-D simplification vs. true 3-D behavior are examined and quantified. Verification of the method for constitutive parameter calibration and for the performance of the analysis procedure is made by comparison with "in-situ" behavior recorded during past earthquakes. Recommendations will be made for updating current analysis and design procedures for structures founded on liquefiable soils, in order to account for the degree of spatial variability of soil properties. This research will provide a sound basis for liquefaction risk assessment and for characterization of spatial variability of soil properties. ***