Diffusers are essential for flow deceleration and pressure recovery, but the adverse pressure gradient often induces boundary layer separation that degrades performance. This study employs large eddy simulation (LES) to investigate how inflow velocity influences the internal flow structures of a square-section diffuser with emphasis on separation phenomena. Three inflow velocities—5, 10.5, and 30 m/s—are considered, and the flow characteristics are analyzed in terms of time-averaged, instantaneous, and fluctuating behavior. Results show that increasing inflow velocity significantly enhances static pressure recovery and exponentially reduces total pressure loss, improving diffuser efficiency. Two types of separation are identified: (1) a centerline bubble near the diverging section entrance and (2) large-scale three-dimensional corner separations. The centerline bubble triggers shear layer instability and separation-induced transition, leading to the formation of hairpin vortices. In contrast, corner regions experience earlier instability accompanied by vortex leg crossover due to vortex interactions from adjacent walls. With higher inflow velocity, vortices become smaller and more coherent, and corner separation zones shrink, reducing interference with centerline vortices. Corner separation also weakens in the mid-to-downstream regions. Turbulent kinetic energy and Reynolds stresses increase, but relative kinetic energy decreases, suggesting enhanced main flow stability despite increased asymmetry and localized instabilities. This study provides high-fidelity insights into the mechanisms of flow separation and vortex formation in square-section diffusers. These findings contribute to the development of turbulence models and the optimization of diffuser performance.
Building similarity graph...
Analyzing shared references across papers
Loading...
Ling Zhou
Beijing Institute of Technology
Yue Suo
Beijing Institute of Technology
Boo Cheong Khoo
National University of Singapore
Physics of Fluids
National University of Singapore
Building similarity graph...
Analyzing shared references across papers
Loading...
Zhou et al. (Fri,) studied this question.
synapsesocial.com/papers/68c1a77a54b1d3bfb60e0b4b — DOI: https://doi.org/10.1063/5.0276962