Prediction model for soil permeability based on fractal characteristics of particles

In order to predict the permeability characteristics of soils, a fractal recognition algorithm of soil particles and a seepage channel reconstruction algorithm are proposed based on the microstructure, and a Monte Carlo prediction model for permeability coefficient of soils is established by combining the geometric reconstruction model with the finite element method. Firstly, according to the microstructural characteristics of soils, the ellipticity, roughness, gradation, porosity and long-axis angle of soil particles are identified by the fractal characteristic identification method. Then, based on these characteristic parameters, the fractal channel reconstruction method is used to reconstruct the microstructural model. Based on the generated microstructure model, the finite element method and the Monte Carlo method are combined to calculate the permeability coefficient with statistical significance. Compared with the experimental results, the rationality of the prediction model is verified (the error is less than 5%). Through the multi-factor analysis, the influences of ellipticity, roughness, gradation, porosity and long-axis angle on permeability coefficient of soils are studied. The order of magnitude relationship is gradation > porosity > long-axis dip angle > ellipticity > roughness. The pearson correlation coefficients are -0.3512, 0.3065, -0.101, -0.042 and -0.010, respectively. Through the analysis of the seepage channel, it is found that the "width" and "tortuosity" of the seepage channel are mainly affected by the gradation and porosity. The ovality, roughness and long axis dip angle mainly affect the "angle" and "length" of the seepage channel.