A Comprehensive Study of Buoyant Confined Jets by Integrating Laboratory Experiments, CFD, and Machine Learning

Abstract

This thesis presents a comprehensive study of the behavior and dispersion characteristics of vertically confined buoyant jets using an integrated approach combining laboratory experiments, computational fluid dynamics (CFD) and advanced machine learning techniques. The overall objective of this study is to accurately determine and predict the dynamic properties of buoyant jets under different confinement conditions in order to enhance the environmental management strategies for wastewater discharge systems. In the experimental part, a laser-induced fluorescence (LIF) technique was used to capture the detailed concentration field of the buoyant jet. This method enables accurate visualization and quantification of the jet behavior under different flow scenarios characterized by varying degrees of lateral confinement, Froude number and Reynolds number. These experimental results provide a reliable benchmark for the validation of subsequent numerical simulations. Numerical analysis using the open-source computational fluid dynamics (CFD) software OpenFOAM evaluates three commonly used Reynolds-averaged Navier-Stokes (RANS) turbulence models: the standard k-ε model, the RNG k-ε model, and the SST k-ω model. Among them, the standard k-ε model shows excellent predictive ability and is in high agreement with experimental observations. The model consistently has high coefficient of determination (R²) and low normalized root mean square error (NRMSE), which proves its reliability in simulating complex buoyant jet phenomena. To further improve the prediction accuracy and computational efficiency, three state-of-the-art machine learning models are employed in this study: Adaptive Neuro-Fuzzy Inference System (ANFIS), Extreme Learning Machine (ELM), and Multivariate Adaptive Regression Splines (MARS). The ELM model is efficient and maintains high accuracy, making it ideal for scenarios that require fast computation. MARS provides an interpretable framework that greatly facilitates the understanding of jet dynamics through explicit mathematical formulas derived from data-driven basis functions. Overall, this integrated approach improves the accuracy of predictions of buoyant jet behavior and is expected to significantly reduce the environmental impacts of wastewater management practices by enabling more informed decision making regarding the design and operation of wastewater discharge systems, ultimately supporting sustainable environmental management.