Thesis (M.Sc.) - Cairo University - Faculty of Engineering - Department of Civil Engineering

In-situ tests generally investigate a much greater volume of soil more quickly than possible for sampling and laboratory tests, and therefore they have the potential to realize both cost savings and increased statistical reliability for foundation design. The principle objective of this study is to demonstrate the feasibility of using artificial neural networks (ANNs) to predict different correlations from CPT and SPT results considering the uncertainties and non-linearity of the problem. ANN is used to predict CPT results from SPT results and to estimate the angle of internal friction and soil modulus of elasticity, which are important soil parameters, from CPT and SPT results considering the additional factors to improve the correlations. A large amount of field and experimental data including SPT/ CPT results, grain size distribution of different samples and a calculated data of overburden pressure was obtained. This data was used for the training and the validation of the neural network. A comparison has been made between the obtained results from artificial neural networks (ANNs) approach, and some common traditional correlations of predicting CPT results (qc) from SPT results (N), angle of internal friction and soil modulus of elasticity from SPT results and angle of internal friction and soil modulus of elasticity from CPT results with respect to the actual results of the collected data

Issued also as CD