节点文献
邓肯-张EB模型参数求解的二次优化法
Two Step Optimization Method for Parameters of Duncan-Chang EB Model
【摘要】 邓肯-张非线性弹性模型是土石坝工程中最常用的本构模型。水利行业《土工试验规程》中根据应力水平75%和90%两点法进行计算时,得到的结果往往并不合理,有时n值还可能出现负数。一般的适线法仅仅对单个试样结果进行优化,而并不是针对整组试验结果,因此无法得到最优结果。提出了一种二步优化的参数计算方法,首先对每级围压下单个试样的试验成果采用适线法优化,得到每级围压下的参数a、b。在此基础上,计算得到参数K、n、Rf的初值。然后以邓肯-张理论为基础,根据获得的参数初值针对整组试验成果进行二次优化,以理论计算与试验的应力应变曲线差的平方和最小为目标函数,从而得到EB模型的主要参数。该方法简单实用,能够快速和准确地获得邓肯-张模型参数,并结合糯扎渡大坝堆石料三轴试验数据,对方法进行了验证。
【Abstract】 Duncan-Chang nonlinear elastic constitutive model is the most used one in embankment dam engineering. The Specification of Soil Test in hydraulic industry proposes a computational method based on the values of two points from the stress-axial strain curve of the triaxial testing results. The stress levels of these two points are 75% and 90% respectively.However the proposed method cannot obtain reasonable results all the times, and sometimes even the parameter n maybe negative. Curve fitting methods make some progress, but still could not gain the optimal value for the parameters because these methods only based on single sample result. A two step optimization method for acquiring the optimal values of Duncan-Chang model is presented herein. First, the traditional curve fitting method is adopted to obtain the values of parameters a and b under each confining pressure. Then the parameters K, n and Rfare obtained. Using these parameters as initial values, a second optimization procedure is carried out to fit all the results of triaxial test to gain the parameters of Duncan-Chang model, in which, the minimum square sum of the differences of stress and strain curves of theoretical calculation and experiment is taken as the objective function. The method is simple and practical, and can quickly and accurately obtain the parameters of DuncanZhang model. The method is validated based on the triaxial test data of Nuozhadu Dam.
【Key words】 rockfill; Duncan-Chang model; optimization method; embankment dam;
- 【文献出处】 水力发电 ,Water Power , 编辑部邮箱 ,2017年08期
- 【分类号】TV41
- 【被引频次】4
- 【下载频次】330