Characterisation of turbulent duct flows: experiments and direct numerical simulations

Owolabi, BE
(2018) Characterisation of turbulent duct flows: experiments and direct numerical simulations. PhD thesis, University of Liverpool.

[img] Text

Download (6MB)


Turbulent duct flows are encountered in a wide range of engineering applications; a fundamental physical understanding of such flows is thus very important for making predictions about heat transfer, mixing and skin friction drag. Previous studies have focused mainly on the purely-pressure driven (Poiseuille) case at relatively large Reynolds numbers (Re), hence the flow characteristics at low Re are not well understood. Limited Poiseuille flow direct numerical simulation (DNS) data show that in non-circular ducts, there exists an interesting phenomenon at Re close to transition to turbulence. Specifically in ducts of square cross-section, the flow field is observed to switch between two states characterised by different secondary flow patterns, thus potentially having serious implications for heat and mass transport. This finding has never been verified experimentally. Another important area in the study of turbulent duct flows, which is yet to be fully understood, is the drag reduction (DR) obtained by seeding with long-chain flexible polymers having high molecular weights. These are known to modify the turbulence field in duct flows when added even at minute concentrations, causing a massive decrease in skin friction; yet it has never been possible to relate the degree of DR to a measurable fluid property. In this study, turbulent duct flows of Newtonian and non-Newtonian fluids over a wide range of Reynolds numbers are investigated both experimentally and numerically. The geometries considered include a square duct, rectangular channel and circular pipe. In purely pressure-driven flow in a square duct, the onset criteria for transition to turbulence is first examined. In so doing, the potential importance of Coriolis effects on this process for low-Ekman-number flows is highlighted. Experimental data on the mean flow properties and turbulence statistics at relatively low Reynolds numbers are then obtained. The alternation of the flow field between two states in the "marginal turbulence" regime in a square duct, originally predicted by the DNS of Uhlmann et al. (2007) is confirmed by bimodal probability density functions of streamwise velocity at certain distances from the wall as well as joint probability density functions of streamwise and wall-normal velocities which feature two peaks highlighting the two states. By applying Taylor's hypothesis of frozen turbulence to the data, it is shown that there is also a spatial switching along the length of the duct. Similarly, direct numerical simulations of zero-net-flux wall-driven (Couette) flow in a square duct reveal an alternation between two states, thus indicating that the phenomenon is not unique to Poiseuille flows. The secondary motions are observed to be closely related to the near-wall ejection and sweep events. Furthermore, the side walls are found to have a stabilising effect on the flow, the critical Reynolds number for transition being much higher than that in plane Couette flow. For an experimental investigation of square duct Couette-Poiseuille flows, a new test section with one moving wall has been designed and constructed. Preliminary Laser Doppler Velocimetry (LDV) measurements of the velocity profiles in the laminar regime show that the fully developed analytical solution can be accurately reproduced in the facility. Suggestions for future turbulent flow studies in the new test-section have been given. Finally, the polymer DR problem has been revisited, and for the first time, a correlation which allows for quantitative predictions of DR from the knowledge of a single measurable material property of a polymer solution, independent of the geometry, concentration, and other experimental variables is obtained.

Item Type: Thesis (PhD)
Divisions: Fac of Science & Engineering > School of Engineering
Depositing User: Symplectic Admin
Date Deposited: 21 Aug 2019 10:10
Last Modified: 09 Jan 2021 04:27
DOI: 10.17638/03031825
  • Dennis, David
  • Poole, Robert