Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 97
Edited by: Y. Tsompanakis, B.H.V. Topping
Paper 23

Settlement Prediction for Deep Foundation Piles using Artificial Neural Networks

A. Abadkon and M.E. Akiner

Civil Engineering Department, Bogazici University, Istanbul, Turkey

Full Bibliographic Reference for this paper
A. Abadkon, M.E. Akiner, "Settlement Prediction for Deep Foundation Piles using Artificial Neural Networks", in Y. Tsompanakis, B.H.V. Topping, (Editors), "Proceedings of the Second International Conference on Soft Computing Technology in Civil, Structural and Environmental Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 23, 2011. doi:10.4203/ccp.97.23
Keywords: artificial neural networks, deep foundation, non-linear modelling, piles, settlement.

Concrete piles are special types of deep foundations used to transmit loads to deeper strata capable of supporting the applied loads. Foundation settlement is the primary design concern along with the bearing capacity of the soil. Accurate prediction of pile settlement is necessary to ensure appropriate structural performance. By performing pile load tests we would be able to confirm the pile lengths and hence contract costs. The most common types of test loading procedures are the constant rate of penetration (CRP) test and the maintained load test (MLT) [1]. The location of the pile load tests should be at the most critical area of the site, such as where the bearing stratum is deepest or weakest. This study has used the data of pile load tests in a fifty storey structure with four basements and one ground level. The project site was located at the trade centre, Dubai. According to the soil investigation report the dredged material consists of varying or irregular densities ranging from dense to loose, grayish brown to light grey or dark grey, slightly silty to silty, fine to medium, often becoming coarse, sand with shells and cemented pieces of sand with occasional sandstone fragments. The data obtained from experiments were used to generate a model based on an ANN. The multilayered perceptron (MLP) architecture were used to train with the backpropagation algorithm. Two distinct approaches were used in this research in order to predict pile settlement as close as possible to their measured values. The first approach was to use empirical equations [2] which were given in the literature for estimation of pile settlement. The second approach was to use an ANN model. When the results are investigated, it is clear that the proposed ANN models give more reliable results for settlement when compared with the results from empirical models. The determination coefficient between measured settlement and empirical value was calculated as 0.692. On the other hand the proposed ANN model is more realistic in terms of performance. The results also showed that the load is an important factor on the settlement and that the effect of the diameter is limited In comparison with the load.

Y. Xu, L.M.M. Zhang, "Settlement Ratio of Pile Groups in Sandy Soils from Field Load Tests", Journal of geotechnical and Geoenvironmental Engineering, 133(8), 1048-1054, 2007. doi:10.1061/(ASCE)1090-0241(2007)133:8(1048)
A.S. Mohamed, H.R. Maier, M.B. Jaksa, "Predicting settlement of shallow foundations using neural networks", Journal of Geotechnical and Geoenvironmental Engineering, 128(9), 785, 2002.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description