Artificial intelligence-based estimation of flushing half-cone geometry

An accurate estimation of half-cone geometry (i.e., volume and length) created by pressure flushing operation in dam reservoirs is required for sediment management in the reservoir storage. In this study, two artificial intelligence techniques namely, Artificial Neural Network (ANN) and Adaptive Neuro-fuzzy Inference System (ANFIS) were utilized to estimate the volume and length of flushing half-cone based on influential variables, i.e., mean flow velocity through bottom outlet (u), water depth in reservoir (H"w), mean grain diameter of deposited sediments (d"5"0), thickness of deposited sediment (H"s) and bottom outlet diameter (D). Experimental data in both dimensional and non-dimensional forms were used to train and test ANN and ANFIS models. The results of the intelligence-based models were also compared with those of existing studies. The outcomes indicated that both ANN and ANFIS models predict the volume and length of flushing half-cone more accurately than existing studies. Also, it was found that the ANN model provides a better estimation of the geometry of flushing half-cone compared to the ANFIS model. Finally, sensitivity analysis was conducted to determine the most and the least influential variables affecting the flushing half-cone geometry. It was found that the sediment characteristics (H"s and d"5"0) and fluids properties (H"w and u) have respectively the most and the least effect on flushing half-cone volume and length.