Application of artificial neural network for calculating anisotropic friction angle of sands and effect on slope stability
来源期刊：中南大学学报(英文版)2015年第5期
论文作者：Hamed Farshbaf Aghajani Hossein Salehzadeh Habib Shahnazari
文章页码：1878 - 1891
Key words：anisotropy; artificial neural network; sand; principal stress rotation; slope stability
Abstract: The anisotropy effect is one of the most prominent phenomena in soil mechanics. Although many experimental programs have investigated anisotropy in sand, a computational procedure for determining anisotropy is lacking. Thus, this work aims to develop a procedure for connecting the sand friction angle and the loading orientation. All principal stress rotation tests in the literatures were processed via an artificial neural network. Then, with sensitivity analysis, the effect of intrinsic soil properties, consolidation history, and test sample characteristics on enhancing anisotropy was examined. The results imply that decreasing the grain size of the soil increases the effect of anisotropy on soil shear strength. In addition, increasing the angularity of grains increases the anisotropy effect in the sample. The stability of a sandy slope was also examined by considering the anisotropy in shear strength parameters. If the anisotropy effect is neglected, slope safety is overestimated by 5%-25%. This deviation is more apparent in flatter slopes than in steeper ones. However, the critical slip surface in the most slopes is the same in isotropic and anisotropic conditions.
Hamed Farshbaf Aghajani, Hossein Salehzadeh, Habib Shahnazari
(School of Civil Engineering, Iran University of Science and Technology, Narmak, Tehran, P.O.B.16756-163, Iran)
Abstract:The anisotropy effect is one of the most prominent phenomena in soil mechanics. Although many experimental programs have investigated anisotropy in sand, a computational procedure for determining anisotropy is lacking. Thus, this work aims to develop a procedure for connecting the sand friction angle and the loading orientation. All principal stress rotation tests in the literatures were processed via an artificial neural network. Then, with sensitivity analysis, the effect of intrinsic soil properties, consolidation history, and test sample characteristics on enhancing anisotropy was examined. The results imply that decreasing the grain size of the soil increases the effect of anisotropy on soil shear strength. In addition, increasing the angularity of grains increases the anisotropy effect in the sample. The stability of a sandy slope was also examined by considering the anisotropy in shear strength parameters. If the anisotropy effect is neglected, slope safety is overestimated by 5%-25%. This deviation is more apparent in flatter slopes than in steeper ones. However, the critical slip surface in the most slopes is the same in isotropic and anisotropic conditions.
Key words:anisotropy; artificial neural network; sand; principal stress rotation; slope stability
