High-Resolution Numerical and Experimental Study of Dense Jets

Abstract

The rapid global expansion of seawater desalination has emerged as a critical solution to freshwater scarcity, yet it generates large volumes of saline and thermal effluents that pose challenges for coastal and marine environments. The near field hydrodynamics and mixing of desalination discharges are governed by the delicate interplay between jet momentum, buoyancy, and ambient conditions. An accurate prediction of these processes is vital for designing outfalls that ensure regulatory compliance and minimize ecological impacts. Despite significant progress in experimental and numerical modeling, notable knowledge gaps remain in (i) the specification of realistic boundary conditions for high-fidelity simulations, (ii) the representation of jet behavior in complex stratified and crossflow environments, and (iii) the integration of experimental and numerical data within predictive, computationally efficient modeling frameworks. This thesis addresses these challenges through a coordinated program of advanced experiments, high-resolution simulations, and multi-fidelity statistical modeling. The first part of the thesis investigates the role of inlet boundary conditions in Large Eddy Simulation (LES) of negatively buoyant jets relevant to desalination outfalls. Multiple inflow turbulence treatments, including mapped, synthetic, and stochastic methods, were systematically compared against high-resolution Particle Image Velocimetry (PIV) measurements. The study demonstrated that mapped inflow and synthetic turbulence approaches are most effective in reproducing near-field turbulence generation, velocity statistics, and dilution performance, while oversimplified uniform inflows led to systematic underprediction of mixing. This highlights the decisive role of physically consistent boundary specification for LES reliability. The second part examines dense jets in crossflow and stably stratified environments using LES. A suite of simulations covering a wide parameter space of densimetric Froude numbers, velocity ratios, and stratification strengths was conducted. Results revealed that ambient stratification strongly suppresses vertical penetration and enhances asymmetric entrainment, while crossflow drives trajectory deflection and modifies coherent turbulent structures. These findings extend the current understanding of jet–ambient interactions and provide physically grounded benchmarks for the design of desalination outfalls in variable environments. The third component of the thesis develops a Bayesian hierarchical multi-fidelity framework that combines low-fidelity Reynolds Averaged Navier-Stokes (RANS), high-fidelity LES, and experimental Particle Image Velocimetry (PIV) and Laser Induced Fluorescence (LIF) data within a Gaussian Process Regression (GPR) model. This approach exploits the complementary strengths of different fidelities, achieving robust predictions of terminal rise height, return point dilution, and centerline concentration while substantially reducing computational cost. The framework demonstrates the feasibility of integrating simulation and measurement data into unified, uncertainty-aware predictive models for environmental engineering applications. Finally, Direct Numerical Simulation (DNS) of a positively buoyant thermal jet is presented as a benchmark. The DNS resolves turbulence and scalar transport at all relevant scales, producing high-fidelity datasets of mean flow, turbulence statistics, and entrainment. These results elucidate fundamental turbulence–buoyancy interactions and serve both as validation for LES subgrid-scale models and as calibration input for the multi-fidelity framework. The thesis makes four main contributions: (1) it establishes best practices for LES inlet boundary conditions in desalination jet simulations; (2) it quantifies the interplay between crossflow, stratification, and dense jet dynamics; (3) it introduces and validates a Bayesian multi-fidelity modeling framework for efficient prediction of discharge behavior; and (4) it generates high-resolution DNS datasets of thermal jets that advance physical understanding and provide a benchmark for future model development. Collectively, these contributions enhance predictive capabilities for desalination discharges, offering methodological and physical insights that support the design of environmentally sustainable outfalls.