5% - eTheses Repository - University of Birmingham

Section 3.4 considers suspension from the stoss slope downstream
of the crest. However, the interaction between shear layer
vortices and the stoss slope downstream of the trough was
particularly difficult to study here because of bedload motion
during start-up. Thus, particles pre-spread over the stoss slope
were washed away during the start-up flows in our flume, even
when the flume was full before the flow was started and
regardless of the rate of increase in flow velocity. A partial
solution was secured by attaching a grid of galvanised meshing
to the centreline of the bedform. The height of the grid above
the perspex bedform was about 4mm.
Even with the grid present the starting flow washed away most of
the particles at higher flow velocities, however enough remained
to allow reasonable estimation of the suspension fluxes. No
estimates were made of the concentration distribution due to lack
of particle numbers so we give a brief description here of the
seven runs in which the flow velocity U and particle fall speed
V were varied through U=0.20-0.36m/s at V=0.034m/s and U=0.20-
0.43m/s at V=0.60m/s.
3 EXPERIMENTAL RESULTS
3.1 Visualisation of the Coanda-flapping events
The series of plates in figure 2 show the Coanda-flapping
phenomenon with great clarity. This sequence was taken over the
bedforms described above at a flow velocity of about 0.6m/s
4-10