ISSN 2319-5347, ISI Impact Factor: 0.763

VOLUME 06 NO. 04 OCTOBER 2017:

Title:       Predicting the Settlement in Raft and Piled-Raft Foundations using neural models
Authors: Vikas Kumar, Arvind Kumar
Abstract: Piled-raft in recent years has been accepted as an economical and efficient form of foundation which can sustain high loads in structures such as high rise buildings. In this paper, load-settlement behaviour of the raft and piled-raft foundation has been investigated. To study the behaviour, a model with raft and piled-raft foundation system was developed and subjected to quantum of loads. The various tests were conducted by varying number of piles, diameter of piles, Length by width ratio of piles and relative densities of soil. In order to transform the results of this whole series of experiments on computing platform artificial neural networks (ANN) on Matlab was used to predict the settlements and was compared with experimental results. To develop ANN model, 687 data points were used. The process of developing neural model includes selection of various internal parameters to obtain better predictive ANN model. It was observed that ANN was able to give results much closer to experimental results and also at some instances it has behaved like an experimental set up. The sensitivity analysis done in this study gives the effect of each parameter in percentages on settlement. The number of piles in the model used contributes 26.82% for settlement.

Keywords: Artificial Neural Networks, Piled-Raft, Bearing Capacity, Settlement, Density, Load
Pages:       134-143

Download Full Paper