关键词: 勘探数据量, 岩土参数, 统计特征值, 变异性, 模型不确定性

Abstract: Accurate determination of statistical characteristic values (e.g. mean and standard deviation) of geotechnical design parameters is an important prerequisite for geotechnical reliability analysis and design. In this paper, under the condition of satisfying the accuracy of statistical design of geotechnical design parameters, a new method is proposed to determine the minimum test data quantity of geotechnical engineering. The relative error and relative variability index are defined to measure the accuracy of the statistical characteristics of geotechnical design parameters. The influence of the static cone penetration test data on the calculation accuracy of the statistical characteristic value of effective internal friction angle of sand soil is systematically analyzed. Moreover, the minimum cone penetration test data quantity is determined based on the relative error and relative variability index. The results show that the relative error of mean value of internal friction angle of sand estimated indirectly from static cone penetration test is small. The relative variability index of sand internal friction angle decreases with the increasing in the data quantity of static cone penetration test. The ratio of uncertainty caused by insufficient cognition to total variability decreases with the increasing data quantity of static cone penetration test. When the allowable relative variability is less than 0.2, the variability of sand internal friction angle is in the range of 5% and 20% COV, which satisfies the predetermined requirement of minimum static cone penetration test data quantity of 10-100. If the relative variability is allowed to be less than 0.3, the minimum static cone penetration test data quantity is 5-43. In addition, when estimating the geotechnical parameters, the uncertainty of empirical regression model has a significant influence on the minimum data quantity. The minimum data quantity of static cone penetration test increases with the increasing uncertainty of empirical regression model. Therefore, the test data should be collected as widely as possible and the calculation model should be selected with higher accuracy to obtain more accurate statistical characteristic values of geotechnical design parameters.

Key words: test data quantity, geotechnical parameters, statistical characteristic values, variability, model uncertainty