基于标贯试验的砂土液化概率判别法

合理地对建设场地进行液化判别是降低液化灾害的基础。中国规范通过比较临界标贯击数与实测标贯击数的相对大小来判别砂土液化,但由于建立经验判别准则的过程中存在大量的不确定性因素,故这种确定性方法不是衡量砂土液化的准确指标。针对这个问题,基于中国标贯试验的液化案例库,利用极大似然法对4种广义线性模型进行参数标定,建立适用于中国的液化概率判别模型。结果表明:当液化概率较小时,4种广义线性模型差异显著;当液化概率在30%~70%之间时,4种模型的预测结果相近。模型比较表明,Log-log模型对案例库的拟合效果最好,给出了基于Log-log模型的液化概率计算公式及指定液化概率下的标贯击数临界值计算公式。回判分析表明,提出的液化概率判别模型的总体回判成功率高于现行建筑抗震设计规范。

Probabilistic evaluation of sand liquefaction potential based on standard penetration test

Reasonable liquefaction assessment of construction sites is the basis of mitigating liquefaction hazards. The current Chinese seismic design code evaluates the liquefaction potential by comparing the critical value of standard penetration test blow count with the measured value. However, due to the significant uncertainties associated with the process of establishing empirical liquefaction criteria, the deterministic method is not an accurate measurement for evaluating the liquefaction potential. To solve this problem, the maximum likehood method is used to calibrate the parameters of four generalized linear models based on the liquefaction case base of Chinese standard penetration test, and four probabilistic evaluation models applied to China are established. The results show that the four generalized linear models differ significantly when the liquefaction probability is small, and the prediction results of the four models are similar when the liquefaction probability is between 30% and 70%. Comparison of the models shows that the Log-log model fits the database best. The formulas for calculating the liquefaction probability and the critical value of standard penetration test blow count under specified liquefaction probability based on the Log-log model are provided. The results of verification analysis show that the overall judgment success ratio of the proposed probabilistic evaluation model of liquefaction is higher than the current Chinese seismic design code.