Evaluating hydraulic jump oxygen aeration by experimental observations and data driven techniques

Abstract

ABSTRACT Hydraulic jump occurs at the transition from shallow flow to deep flow which is followed by vigorous turbulent flow where a bulk of air from ambient atmosphere sucked into the water flow and manifests a huge quantity of bubbles. In this work, data driven-models were employed for estimating hydraulic jump oxygen aeration efficiency (E20) at under sluice gate using experimental data. The performance potential of these models was evaluated in the form of comparison with observed data through two evaluation metrics, coefficient of correlation (CC) and root mean square error (RMSE). The results of triangular shaped ANFIS (ANFIS_trimf) having highest value of CC = 0.9878 and lowest value of RMSE = 0.0151 have been found outperforming the other applied models; however, ANN with value of CC = 0.9877 and value of RMSE = 0.0159 is equally giving comparable satisfactory results but classical models except multiple linear regression with CC = 0.9872 and RMSE = 0.0156 leads to incredible error. Sensitivity analysis shows that Reynolds number was significant parameter in estimation of E20. These data driven-based models described in the study may help the field engineer and researcher in estimating E20, nevertheless a wide range of dataset from the same and different areas may be utilized to develop a more generic model.