A numerical analysis of a corrugated channel flow by Large Eddy Simulation

This thesis presents a detailed numerical analysis of ?ow through a corrugated channel. The geometry has a rectangular cross section. The bottom wall is corrugated with periodic cavities which is a two-dimensional representation of a commercial stainless steel ?exible pipe, and the top and two side walls are ?at plates. The Reynolds number based on bulk ?ow velocity and the hydraulic diameter of the channel is 5300. The principal objective is to understand the ?ow ?eld in this corrugated channel, which bene?ts the future design of engineering equipment with corrugated wall. Several Sub-grid Scale (SGS) models are ?rst validated on plane channel ?ows with di?erent grid densities. Then a Large Eddy Simulation (LES) is performed on the corrugated channel with 128 corrugations and the synthetic turbulence inlet. Flow features of di?erent zones in the channel are analysed in detail, focusing on the time-averaged results, development of the boundary layer, and mechanism of transition. The principal message emerging from this analysis is that the ?ow in the corrugated channel is more complicated than suggested previously by the experimental study and the two side walls have profound e?ect on the ?ow behaviour. The discrepancies between LES predictions and experimental data are also discussed. After a detailed examination, a number of problems and open questions are raised concerning the experimental results and setups. Meanwhile, the LES results are validated from various aspects. In addition, a study of this corrugated channel with 16 corrugations and a periodic boundary condition is also conducted, with the focus on hydrodynamic interaction and vortex evolution. It shows that the case for 16 corrugations with the periodic boundary condition can reproduce the ?ow characteristics of the fully turbulent region predicted in the case of 128 corrugations. The features of vortex evolution shown in the experiment are reproduced and understood by current LES.