An approach to estimate unsaturated shear strength using artificial neural network and hyperbolic formulation

Abstract

Since most soils exist above the ground water table, negative pore water pressures develop in unsaturated soils. This negative pore water pressure, known as matric suction, causes increased shear strength. Therefore, it is required that the effect of the increase in shear strength should be included in geotechnical analyses. However, experimental studies on unsaturated soils are generally costly, time-consuming, and difficult to conduct. Therefore, it is better to have an empirical method that is able to predict the unsaturated shear strength with respect to the matric suction in a more convenient way. For that purpose, we formulated a nonlinear unsaturated shear strength relationship with the matric suction in a hyperbolic form. In the formulation, conventional saturated soil parameters (c', phi') and an ultimate increment of apparent cohesion (C-max) are required. A method is also developed wherein C-max can be predicted using an ANN (artificial neural network) in reference to data obtained from tests conducted in this study and published in references. (C) 2003 Elsevier Science Ltd. All rights reserved.