Abstract

This study presents new empirical equations to estimate the initial shear modulus of clean sands under low strains by using soft computing methods A series of resonant column tests were conducted on clean sand specimens The effect of various factors such as effective stress saturation degree void ratio and shear strain levels were considered by using fuzzy expert systems neural networks and regression analysis A new empirical equation was developed to determine the initial shear modulus of clean sand samples and compared with the existing empirical relationships in the literature Success of the new equation was increased by considering boundary conditions at the shear strain ranges given by the ASTM standards for the resonant column test It was found that the new formulation has a high level accuracy for determining the initial shear modulus of Toyoura sand samples And also it can be used for clean sandy soils having similar properties It was suggested that new approaches can be developed by using soft computing techniques to identify the dynamic shear modulus of sandy soils