Numerical simulation of ozonation in hollow-fiber membranes for wastewater treatment

Abstract

In this study, we developed a comprehensive modeling framework for simulation of ozonation process using combination of artificial intelligence and computational fluid dynamics (CFD). The process is carried out in a hollow-fiber membrane contactor in which the concentration data of ozone was obtained by solution of mass transfer equations, and then the results were used for artificial intelligence modeling. We used three different machine learning models to predict concentration of ozone (C) in a system based on its coordinates, i.e., r and z. The models were optimized using the Bat Algorithm (BA) and were trained on a dataset consisting of over 10,000 data points. The three models developed were Support Vector Regression (SVR), Decision Tree Regressor, and Orthogonal Matching Pursuit (OMP). These models were evaluated using three common metrics — Mean Squared Error (MSE), R-squared (R2), and Mean Absolute Error (MAE). Our results indicated that the SVR model overperformed the other two models in terms of all evaluation metrics. Specifically, the SVR model achieved an MSE of 0.003, an R2 of 0.998, and an MAE of 0.046. The Decision Tree Regressor and OMP models achieved less favorable results with MSEs of 0.007 and 0.221, R2 scores of 0.996 and 0.878, and MAEs of 0.056 and 0.359, respectively.