Intelligent prediction of settlement ratio for soft clay with stone columns using embankment improvement techniques

Construction of highway roads, railways and other engineering structures on soft clay soils normally encounters problems related to excessive settlement issues. The conventional methods are inadequate to analyze and to predict the settlement behavior. Artificial neural network systems are included to predict settlement under embankment load using soft soil properties together with various geometric parameters as inputs for each stone column arrangement and embankment conditions. A case study site investigated field data are taken from a highway project Lebuhraya Pantai Timur2 in Terengganu, Malaysia. Actual angle of internal friction (ϕ), spacing ratio (s/D), cylindrical ratio (L/D) and height of the embankment (H) were used as the input parameters, while the settlement ratio was the main output. The properties of materials on a stone column (ϕ) have high relative importance (40.15 %) compared with the other parameters. Two techniques namely non-cross-validation (β _NCV) and ten-fold cross-validation (β _FCV) were used to build the ANN model. The β _FCV model gives higher efficiency of 0.985 for training and 0.939 for testing, while β _NCV model gives 0.937 and 0.905. The β _FCV model provides results of greater accuracy as compared to the β _NCV models.