Application of flat DMT and ANN to Korean soft clay deposits for reliable estimation of undrained shear strength

Abstract

The flat dilatometer test (DMT) is a geotechnical tool used to estimate in-situ properties of various types of ground materials. Undrained shear strength is known to be one of the most reliable and useful parameters that can be obtained by the flat DMT. However, a successful application of the existing relationships that have been established for other local deposits depends on regional geotechnical characteristics. In addition, although flat DMT data are interpreted using 3 intermediate indices-the material index (I-D), the horizontal stress index (K-D), and the dilatometer modulus (E-D)-undrained shear strength is estimated using only the horizontal stress index (K-D). In this paper, the applicability of the flat DMT to Korean soft clay deposits is investigated. An artificial neural network (ANN) model is developed to predict undrained shear strength, based on p(0), p(1), p(2) and sigma(v)', without using the K-D. The ANN model adopts the back-propagation algorithm and is trained by using DMT data obtained from Korean soft clays. To investigate the feasibility of the ANN model, the prediction results were independently evaluated by data that had not been used to train the ANN model. They were also compared with data obtained using conventional relationships.