Prediction of Energy Dissipation and Length of Jump over Modified Stepped Spillway using Empirical and Adaptive Neuro-fuzzy Inference System Technique

Abstract

This study aims to enhance energy dissipation efficiency by introducing an alternative design to conventional stepped spillways. To achieve this, modified stepped spillway (MSS) with reflector were used and analyzed experimentally for energy dissipation and length of jump. Empirical models were developed to characterize the hydraulic jump with good R² values of 0.86, 0.96 and 0.86 for sequent depth ratio, relative energy loss and relative length of jump respectively. Results indicate jump characteristics are greatly influenced by curved steps, flow regime and turbulence. Higher Froude number increases roller length, jump length and energy loss. Adaptive Neuro-Fuzzy Inference System (ANFIS) model is also utilized for modified stepped spillway to predict flow characteristics for training and validation of experimental results. It is found that ANFIS predicts the non-linear transition effectively by achieving RMSE, R² and MSE training values as 0.06, 0.98 and 0.43 respectively, which are a favorable performance metrics. 3D surface plot provides thorough understanding of parametric dependencies facilitating the identification of key input factors that influence hydraulic modeling. All three models are found to be capable of optimizing the flow and efficient in controlling the length of roller and length of jump. It is concluded that stepped spillways with reflectors and higher number of steps have more energy dissipation rates at various discharge. Also, ANFIS can be trained with experimental or field data, allowing it to adapt to specific site conditions for design configurations.