Degree Name

Doctor of Philosophy


School of Civil, Mining, and Environmental Engineering


The prediction of the structure of turbulent flow in a 3-D channel is very difficult, thus hydraulic engineers often divide the flow region into sub-regions to simplify its calculation, and then a complex 3-D problem can be treated using a 1-D technique. This treatment has been found effective in estimating some real world hydraulic parameters, such as the boundary shear stresses. In practice, engineers generally separate the flow regions into three sub-regions associated with a channel bed and its side-walls. It is widely believed that the theory of flow partitioning is an effective mathematical tool to simplify the hydraulic calculation without any physical meaning, and the practice shows that such treatment can significantly enhance the accuracy for estimation of bedload transport, the bedform resistance, and the pollutant transport. Therefore, it is necessary to experimentally ascertain the existence of the division lines in a channel flow.

This research attempts to address these research gaps with respect to turbulent structures and flow partitioning in flows and explain why the flow region is dividable. An intensive laboratory investigation was carried out. The major part of this study involved the development and use of a sophisticated instrumentation system based on a new 2-D Laser Doppler Anemometer (LDA) system by Dantec. Special attention has been paid to the time-averaged velocity, from which the location of division lines can be observed. The channel bed was specially designed to observe the variation of division lines corresponding to the bed’s curvature, and channel bed-form has been fabricated as the flat bed, convex bed and concave bed.

The literature review shows that although the flow partitioning argument is intensive, no similar experiments have been conducted to verify the existence of division line in a channel flow, and none of the previous researchers have examined the relationship between the division lines and mean velocity profiles. The main novel contribution of this thesis is to examine the existence of division lines by analysing the mean velocity distribution in a flume with a flat or curved bed. Two new methods have been developed to detect whether division lines actually exist from experiments, one uses the condition of zero total shear stress and the other uses the log-law. The feasibility of these two methods developed for division lines have been discussed from author’s experimental data, as well as available data in literature. Throughout this research, it can be confirmed that the division lines indeed exist in a 3-D flow, and they can be determined from the mean velocity distribution in a flume for a flat or curved bed. Moreover, the experimental research verified those previously proposed mathematical methods, which can yield the most accurate division line locations.

The experiments were classified into three channel shapes. Detailed measurements of instantaneous velocities were carried out for the three different channel shapes at various depths of water. Using the LDA technique, a comprehensive set of high quality data of the 2-D turbulence structure have offered valuable information for further understanding of open channel flow with a convex or concave boundary.

Determination of local boundary shear stress from a very thin boundary layer is difficult as it requires special skills and instruments. Hence, a new method called Momentum Balance Method (MBM) is developed to estimate the local boundary shear stress using the main flow data. A theoretical relationship between the boundary shear stress and parameters of main flow region has been established. The results obtained from MBM agree reasonably well with other methods, indicating that the new method is workable in closed duct flows.