Research On Back-analysis Method Of Soil Parameters With Great Differences

Back-analysis of soil parameters is an important research subject for geotechnical engineering.It is based on what we measure to calculate soil parameters which close to the site condition and predict mechanical behavior of geotechnical structure.also.it can direct informatization construction.As we know. Back-analysis on soil parameters of geotechnical engineering always has some characteristics such as multiple parameters which are great different and every parameter is resticted(for example. Poisson's ratio limits (0-0.5),Young's modulus and permeability is greater than zero). As a result, it makes current Optimization theory very difficult to calculate multiple soil parameters at one time.This paper considers the soil parameters characteristic fully, putting the multiple calculation parameters problem which are great different and every parameter is resticted into unconstrained problem by transformation function.Then, this article couples the Classical optimization method(simplex method and powell method)and finite element method to build back-analysis about multiple soil parameters with great differences.Finally,it explores the feasibility of this paper method by researching the results of three examples. This paper gets following main achievements.1.It finds out 3 transformation functions(exponential function, trigonometric func-tion and hyperbolic function)。These functions can make soil parameters with different dimension. Inequality constraints and different order of magnitude become unconstrain-ed soil parameters.2. According to elastic consolidation problems and saturated soil nonlinear elastic consolidation problem,this article finds the best [D] matrix and transformation function of back-analysis about soil parameters by Comparing the combination results of 6 mod-els [D] matrix and 5 transformation function.(1). For one layer example,every model and every transformation function can obt-ain excellent results,also,the analytical results hardly rely on the initial value. (2). For multi- layer example, every model and every transformation function can obtain excellent results except G-E model,and the calculated results are quite approach with each other.At the same time,when this paper use E-μmodel and G-μmodel with the first,second,fourth,and the fifth transformation function, the initial value has less influe-nce on the back analysis results.(3). According to the number of iteration about the analytical results and the analy-tical result whether or not rely on the initial value.this paper considers E-μmodel and G-μmodel are excellent model.and the second transformation function is the best functi-on.Finally, according to this thesis method, a project example's data is used to carry on the back-analysis. The parameters that calculated by the program tally with the rule which consolidation process goes on. The forecasted displacements calculated by the parameters back-analyzed are also tally with the measured displacements. It is proved that sand drained ground consolidation back-analysis method which this thesis proposed is quite successful and effective.