Predicting Hydraulic Properties of Circular Buoyant Jets in the Static Ambient Flow Using ANN and ANFIS

This article presents an artificial intelligence approach for predicting hydraulic properties of circular buoyant jet entering in the static ambient flow by coupling artificial neural networks and ANFIS. In many cases due to high concentrations of pollutants, or critical toxicity, or even very high temperature wastewaters from nuclear reactors, dilution of wastewater to reach the limit concentration should be done in the shortest time. One way to make quick dilution is using submerged jets in rivers or seas that can relatively quickly dilute large amounts of pollution and due to its mixture of high turbulence conditions, the destructive effects will rapidly reduce. Jet behaviors are very important in the environmental field and real ambient flow. Characteristics of jet diffusion, mixing length, distribution of concentration, jet core velocity, and trajectory of jet are variables that should be considered in the buoyant jets. The flow of jet is heavily dependent on the velocity and on the geometry of the diffuser and the physical properties of ambient flow. In this study, drawdown trajectory of jet has been investigated. To achieve goals of this research program, four input variables have been chosen, including relative length of jet trajectory, X/dp, in which X is length of positive buoyancy and dp is port diameter; jet convergence angle θc; geometry number of jet, Di/dp; and Densimetric Froude Number, Frd. In addition, relative height of jet trajectory, Z/dp was selected as output variable. In this article, a series of experiments (from the physical model in the hydraulic laboratory of Shahid Chamran University, Ahwaz, Iran) were used as the learning data for artificial neural networks and ANFIS. The best-trained network of ANN was compared with results of ANFIS. It should be noted that different topologies have been undertaken by using ANN and ANFIS for upper and lower boundaries of jet flux. Thus, two values for Z were predicted. Findings show that ANFIS can predict upper and lower boundaries values of jet trajectory better than ANN. By considering the results, the values of RMSE for the best-selected topology are equal to 0.0051, 0.0076, and 0.0076 in training, validating, and testing mood respectively for upper boundary, moreover 0.0047, 0.009, 0.0058 in training, validating, and testing mood respectively for lower boundary.