Stability analysis of buried pipelines under combined uplift and lateral forces using FELA and ANN

Abstract

In this paper, the effective intelligent system based on artificial neural networks (ANNs) and the finite element limit analysis (FELA) is developed to predict the undrained uplift capacity of buried pipelines in clays. The numerical results of buried pipelines under inclined force are evaluated using lower and upper bound methods from the FELA approach. The adhesion factor at the interface between the pipeline and surrounding soil is set to be varied from smooth to rough conditions. In FELA, the influences of four input parameters on the uplift capacity of buried pipelines are investigated. The parameters considered are the buried depth ratio of the pipe, the inclination angle of applied load, the soil strength ratio, and the adhesion factor. Using the FELA results, a machine learning technique based on artificial neural networks (ANNs) is adopted in the study. Based on the optimal ANN model, the proposed equation is verified and sensitivity analysis is performed. The obtained results showed that the proposed equation gives an excellent agreement between numerical and predicted results. The proposed hybrid soft computing approach benefits from the effective ability of the ANNs as well as the rigorous bound solutions from FELA which can be conveniently used to accurately predict the uplift capacity of pipelines under inclined loading. The results of sensitivity analysis and proposed empirical are produced for practical uses in offshore geotechnical engineering and ocean engineering.