Analysis and Estimation of Seepage through Homogeneous Earth Dams Using Neural Network and Empirical Equation

Abstract

In this investigation, the quantity of seepage through a homogeneous earth dam with a vertical drain on a permeable foundation was studied using the SEEP/W program. Moreover, the effects of the geometrical and geotechnical parameters (upstream slope, dam height, top width, free board, and permeability ratio) on leakage flow were analyzed with SPSS 20. To begin with, leakage flow values through an earth dam were estimated using SEEP/W by taking three values of each parameter, and then statistical analysis was performed with the help of dimensional analysis. In addition, an empirical equation and a neural network model dubbed q_ANN were developed to calculate leakage flow through an earth dam with a vertical drain. The validation of these models (empirical equation and q_ANN) is made by their applications to real case studies such as El-Haimeur and Boubrik dams located in Ghardaia City. In conclusion, the values of these models can prove that the performance of the models is high and accurate to predict leakage flow through a homogeneous earth dam with a vertical drain on a permeable base.