Computational fluid dynamics and artificial neural networks for modelling lined irrigation canals with low-density polyethylene and cement concrete liners

Abstract

This study numerically investigated the lining effect on the discharges and seepage losses of five reaches which belong to the El-Sont Canal, Asyut, Egypt, using FLOW-3D and Slide2 models, respectively. Two lining materials were considered, cement concrete (CC) and CC with low-density polyethylene (LDPE) film. A cost analysis was performed to explore the feasibility of the proposed lining materials. Moreover, a parametric study was conducted by the Slide2 model to investigate the effect of canal geometry and liner properties on seepage losses. An artificial neural network (ANN) model was developed based on the Slide2 model scenarios to estimate the seepage losses from lined irrigation canals. The results showed that reach the discharge calculated from the FLOW-3D model increased by 92%–97% and 149%–156%, while the calculated seepage losses from the Slide2 model decreased by 81%–87% and approximately 97% under CC and CC with LDPE film liners, respectively. Cost analysis revealed that the overall cost of CC with LDPE film was higher by 14% than CC. Relying on the importance of saving irrigation water and conveying water to the last reaches, CC with LDPE film is recommended for lining irrigation canals. A parametric study showed that the seepage losses were reduced by more than 96% when the ratio between liner and soil hydraulic conductivities was less than 0.01. A thick liner could maximally decrease the seepage losses by 68%, regardless of the canal geometry. As the developed ANN model showed a close agreement with the Slide2 results with coefficient of determination (R²) and mean squared error values of 0.99 and 0.05, respectively, the ANN model is recommended as a robust and rapid tool for estimating seepage losses from lined irrigation canals.