Developing a numerical model for sediment transport in channel network by considering multigrade bed load sediment

Abstract

ABSTRACT Predicting transport rates is difficult for the morphological features of marine, river, and coastal ecosystems. To design a model bed load sediment transport rate along with the surface layer, a mesh-based numerical method is used, and for the sediment transport the governing equations are employed and depth-averaged is used to create open channel flow and in three-dimensional horizontal space the evaluation is performed. By utilizing the governing equations, the open channel flow discretization and the numerical hydrodynamic model of sediment transport rate are calculated, and in the flow channel by using the solution method the settling velocity is determined. The complex issues involved nonlinear equations for the numerical modelling with solution methodology. The chicken swarm optimization algorithm is revealed by multi-objective metaheuristics optimization, and a better outcome is acquired by optimization. Finally, if polygonal mesh modelling is used in the diversion flow channel and curved bend channel, several experimental models were employed in the model experiments. A numerical measured and improved model is used to estimate the results in comparison, and the optimally developed model forecast rates with accuracy. In some open-channel flows such as reservoirs and rivers, the struggling flow and bed load sediment transport were dealt with in this model. Based on the results obtained, the proposed technique demonstrates significant improvements in both accuracy and prediction performance compared to the existing techniques. The proposed technique achieves an accuracy of 96.87%, outperforming the BCNN technique with an accuracy of 89.77% and the H-ANN-FA technique with an accuracy of 84.65%. Moreover, the proposed technique exhibits a lower RMSE value of 0.12, indicating a substantial reduction in the average deviation between predicted and observed values. Based on the NSE and R2 results, the proposed technique demonstrates remarkable improvements in prediction accuracy compared to the other techniques. The proposed technique achieves an NSE value of 0.998 and R2 value of 0.99, compared to the existing technique with an improvement of 46.47% and 11.24%.