Evaluation of Aeration Efficiency over Broad-Crested Stepped Spillways Using an Intelligent Computational Approach

Stepped spillways can enhance dissolved air flow, which increases the rate of energy dissipation and reduces the size of the energy dissipater at the downstream of the spillway. This paper was concerned with the application of a feedforward two-layer artificial neural network (ANN) to evaluate the aeration efficiency over broad-crested stepped spillways corresponding to different types of the flow regime. After the training of different networks, the best network, which includes two layers and one neuron at output layer, was selected. According to obtained results, the aeration efficiency at 20° C can be evaluated by having the discharge per unit width over the spillway, the chute slope angle, the step height, and the total number of all the steps. The obtained network was validated using laboratory data, and the proper conformity was clearly observed statistically. Furthermore, ANN results were compared with the Adaptive Network Based Fuzzy Inference System (ANFIS) results. Finally, the sensitivity analysis of the aeration efficiency against the change of each parameter of geometry of the spillways was done. Results indicated the behavior of the aeration efficiency against investigated parameters well.