Determination of undrained shear strength parameters of undistrubed samples using the data obtained by resedimentation method

In this study, an estimation model was developed to determine the shear strength parameters of disturbed soil samples using resedimented samples. In this context, plastic and index property determinations and UU triaxial compressive strength tests were carried out on undisturbed alluvial soil samples obtained by core drillings from Sahilkent region of Bafra district (Samsun). These samples, which were disturbed after the initial experiments, were resedimented in the second stage of the study by reflecting their original characteristics in the field. Consolidometers with a diameter of 9 cm, designed and produced within the scope of this study, were used in the resedimentation process. The UU triaxial compressive strength tests were repeated on the resedimented samples. As a result of multivariate regression analysis, empirical equations with coefficients of correlation varying between 0.795 and 0.850 were developed for the estimation of undrained shear strength, cohesion, and internal friction angle, whereas equations with coefficients of correlation ranging from 0.875 to 0.974 were proposed using artificial neural networks. Although the coefficients of correlation of the equations developed with artificial neural networks are higher than the equations obtained by multivariate regression analysis, it is recommended that the equations obtained by multivariate regression analyzes should be preferred because the number of independent variables used in artificial neural networks is too large and the process of obtaining empirical equations is difficult.