5% - eTheses Repository - University of Birmingham

for h/D which varied between 0.21 and 0.27. The flow development
length to depth ratio (LDEV/D) ranges from 24 to 19 which is low,
but not unprecedented since Lyn (1993) reports as yet unpublished
work carried out at Delft where a value as low as 17 was used.
This parameter was not as important as the others because even
if the flow had arrived at the first bedform fully developed it
would have undergone an acceleration up the first stoss slope.
This situation would be repeated over each bedform so a 'fully
developed flow' was never a realistic proposition. Indeed,
Soulsby (1989) reports accelerating velocities along sandwave
stoss slopes which form low level jets near the crest,
comfortingly similar to those in figure 12. We add here that the
interesting 'Coanda-flapping' effect studied in chapter 3 is an
unsteady flow feature present in separated flows regardless of
the extent of flow development length (we reported identical
findings in report FAST9108 .RJO\NHT -appendix 1- at LDEV/D about
60) as are the discrete shear layer vortices present in the flow
after every crest.
The ratio of the flow width to depth varies from 2.2 to 2.8, less
than the desired value of 4, so we proceed with a caution that
all of our results should be derived from the central region of
the flow. Qualitative evidence from the lee slope capture
studies in chapter 3 show that particles have a very slow drift
rate towards the corner of the rig which is insignificant
compared to the convection velocity (less than 1%) .
2-11