基于多因素椭圆堆叠的岩土细观建模方法

天然岩土体具有复杂的细观结构和组成特征，现有细观建模方法难以解决任意粒径、长短轴比、倾角以及平面分布特征的多因素岩土体细观建模问题。为此基于椭圆的4段圆弧等效，结合改进的分组波前堆叠法，实现了任意粒径、长短轴比、倾角以及平面分布的快速高体积率椭圆堆叠，并以此建立了与堆叠椭圆粒径、长短轴比、倾角以及平面分布一致的凸多边形结构。与现有的椭圆堆叠ALA法和圆盘堆叠ODR方法相比，提出的方法具有更高的椭圆堆叠体积率和堆叠效率，MATLAB环境下椭圆堆叠效率为500～600个/s。与现有仅能考虑颗粒粒径分布的岩土细观建模GBM法相比，提出的方法可以同时考虑颗粒的粒径、长轴、倾角，以及平面分布4个因素，快速生成任意粒径、长短轴比、倾角以及平面分布的颗粒堆积，并建立与椭圆分布一致的岩石细观结构。通过两个岩土体细观建模案例分析表明，所提方法可以提高岩土体模型的细观建模精度与生成效率。

Mesostructure modeling for rock and soil based on ellipse packing with multiple prescribed distribution

Natural rock and soil have complex mesostructure and composition characteristics, and it is difficult to model geotechnical mesostructure, which has specific particle size, aspect ratio, dip angle, and plane distribution characteristics. For this reason, a method is introduced to use four connected arcs to approximate the shape of an ellipse, combining with improved advancing front method. A highly dense ellipse packing allowing any size, aspect ratio, orientation angle, and spatial distribution is generated by this method. Based on the ellipse packing generated, a complex geotechnical mesostructure with imposed elliptical particle size, aspect ratio, dip, and plane distribution characteristics is generated. A comparison between the proposed method and two traditional methods (optimized dropping and rolling method (ODR) and advancing layer algorithm (ALA)) shows that the ellipse packing generated by the proposed method has a higher density and faster speed than that generated by ALA and ODR, with a packing efficiency of 500-600 s in the MATLAB environment. Compared with the size distribution generated by GBM in modeling mesostructure of rock and soil, the method proposed in this paper can consider four factors (i.e., particle, size, long-to-short axis ratio, dip angle, and plane distribution). The method can construct a convex-based structure equivalent to real rock and soil. Two application cases show that the method proposed in this paper can improve the modeling accuracy and efficiency from the current geotechnical model.