基于累积动应力水平的间歇加载下超孔压预测

Predicting excess pore water pressure under cyclic loading with regular intervals based on cumulative dynamic deviator stress level

设计一系列不同动应力幅值、间歇时长、振动时长的波形,对杭州地铁沿线淤泥质软黏土进行不排水单向循环加载试验,精确观测孔压累积过程。引入累积偏应力水平描述应力状态,分析对比各试验方案下归一化孔压随其变化的发展规律;分析动应力比、间歇时长、振动次数对超孔压累积的影响,发现动应力水平和振次的影响最为关键;从孔压增长机制的角度分析了间歇对于超孔压累积的影响;利用双曲函数拟合关系曲线,建立预测模型;通过分步拟合各振次关系曲线函数,发现振次不同时该模型的常数取值略有不同,分析常数随振动次数变化的规律,提出常数获取方法;最后利用该模型和既有模型预测新试样的超孔压发展,发现该模型表现良好。

This paper has performed a group of undrained unidirectional cyclic loading tests on undisturbed clay samples near planned metro line of Hangzhou.To simulate subway load,this paper has considered the characteristic that the trains always leave at intervals,so cyclic load is designed to consider regular intervals.Under cyclic loading with different amplitudes,dynamic loading durations and intervals durations,excess pore water pressure accumulations are observed.In this paper,the critical dynamic amplitude under cyclic loading with intervals is obtained and is found to be similar with that under cyclic loading without intervals.After that,an explicit model to calculate excess pore water pressure is built with an independent variable that describes cumulative dynamic deviator stress level.It is a new stress state parameter that can describe the effect of long-term dynamic loading in a cumulative way.The influence regulation of cumulative dynamic deviator stress level,vibration number and interval duration are presented and the dynamic deviator stress level and vibration number are much more important to excess pore water pressure accumulation than intervals.This explicit model is in the form of hyperbolic function with one major variable and a few constants.When predicting excess pore water pressure under different vibration numbers,these constants are also different but has relation with vibration number.The method to calculate the constants is also presented.In the end of this paper,testing results of several samples are presented to verify this model.Comparing the results predicted by the new model with an existing model with the measured data,the new model performs well.