Prediction of lateral effective stresses in sand using artificial neural network


Creative Commons License

Uncuoglu E., Laman M., Saglamer A., KARA H. B.

SOILS AND FOUNDATIONS, cilt.48, sa.2, ss.141-153, 2008 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 48 Sayı: 2
  • Basım Tarihi: 2008
  • Doi Numarası: 10.3208/sandf.48.141
  • Dergi Adı: SOILS AND FOUNDATIONS
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Sayfa Sayıları: ss.141-153
  • İstanbul Teknik Üniversitesi Adresli: Evet

Özet

Predicting the lateral effective stress and the coefficient of lateral earth pressure at rest values is an important task in geotechnical engineering since it is used in the design and analysis of earth retaining structures, slope stability, piles and pier foundations. It needs sophisticated test procedures. The laboratory and in situ tests are also expensive and time consuming. In this study, an artificial neural network model is developed to predict the sigma(h)', lateral effective stress in cohesionless soils. Back propagation neural networks are used for function approximation and model has been trained by Levenberg-Marqurdt (LM) learning algorithm. The data used in the running of network models have been obtained from extensive series of oedometer tests on Kilyos, Ayvalik and Yalikoy sands. Tests were carried out on loose, medium dense and dense state of compactness in normal loading, unloading and reloading conditions. The test results demonstrate that there is a linear relationship between vertical and lateral stresses for normally loaded cohesionless soils under K-0 conditions. K-0 values obtained for the loose state of cornpactness are higher than for the dense state of compactness. The results of the artificial neural network model indicate that the model serves as simple and reliable tool to predict sigma(h)' and also K-0 in cohesionless soils. The variation of K-0 values with internal friction angles is obtained and a simple expression is derived from this relationship.