中文摘要：

土-水特征曲线是非饱和土力学研究中的一条重要曲线,其拟合参数对于分析非饱和土的各种性质具有重要意义。基于渗析法及滤纸法试验,测定了合肥市某工地非饱和膨胀土的土-水特征。以van Genuchten模型为基础,采用科学计算语言Matlab中的Lsqcurvefit函数对试验结果进行了曲线拟合,分析了土-水特征曲线拟合参数的多解性及其影响因素。拟合结果表明:渗析法试验得到的吸力范围偏低,拟合参数多解性较为明显;综合2种试验结果,可以得到较大吸力范围内的土-水特征曲线,且参数多解性较小;参与土-水特征曲线拟合的试验数据点的吸力范围应足够大,试验数据点应能够代表整条曲线的形状,试验点不必太过于密集,否则土-水特征曲线拟合参数的多解性较为明显;另一方面,只要试验数据点能代表整条曲线的形状,试验数据点的个数不必太多。这一研究结果对提高土-水特征曲线拟合参数的精度、提高试验效率有重要指导意义。

英文摘要：

Soil-water characteristic curve (SWCC) is a very important curve in the study of unsaturated soil mechanics;and the curve fitting parameters of SWCC are significant for analyzing the properties of unsaturated soils. To obtain the curve fitting parameters of SWCC accurately and efficiently, the experimental design and the curve fitting method of SWCC are studied. Based on the osmotic method and the filter paper method, a SWCC of an unsaturated expansive soil in a construction site in Hefei is measured. The van Genuchten model is used to represent the relationship between gravity water content and matric suction of the tested soil. The Lsqcurvefit function of the scientific language Matlab is adopted for carrying out the curve fitting of SWCC because the Lsqcurvefit function can set the lower limits and upper limits of fitting parameters. The problem of multi solutions of curve fitting parameters of SWCC and its influence factors are studied. It is demonstrated that when only the experimental data of the osmotic method is used, the range of suction is relatively small;and the multi solution phenomenon of SWCC’s fitting parameters is obvious. However, if the experimental data of both the osmotic method and the filter paper method are used, the range of suction is relatively large. Then, the multi solution phenomenon of SWCC’s fitting parameters can be diminished. And if the range of suction is large enough to represent the shape of the whole SWCC, the number of experimental data points need not be too much. Based on these conclusions, the experiment for measuring the soil-water characteristic curve can be designed more efficiently;and then the curve fitting parameters can be obtained more accurately.