| Summary: | The flow in multistage axial flow compressors is
particularly complex in nature because of the proximity of
moving bladerows, the growth of end-wall boundary layers and
the presence of tip and seal leakages and secondary flow.
The problems associated with these phenomena are at their
most acute in the latter, subsonic stages of the core
compressor, where Reynolds numbers are modest and the
blading has low aspect ratio. Indeed, much of the
inefficiency of axial stages is believed to be associated
with the interaction between blading and end-wall flows.
The fact that the end-wall flow phenomena result in
conditions local to the blade which are quite different
from those over the major part of the annulus was
appreciated by many of the earliest workers in the axial
turbomachinery field. However, experiments on blading
designs aimed specifically at attacking the end-loss have
been sparse.
This thesis includes results from tests of conventional
and end-bent blading in a four-stage, low-speed, axial
compressor, built specifically for the task, at a scale
where high spatial measurement resolution could be readily
achieved within the flowpath. Two basic design styles are
considered: a zero a0 stage with DCA aerofoils and a
low-reaction controlled-diffusion design with cantilevered
stators.
The data gives insight into the flow phenomena present in
'buried' stages and has resulted in a much clearer
understanding of the behaviour of end-bent blading. A
3D Navier-Stokes solver was calibrated on the two
low-reaction stators and was found to give good agreement
with most aspects of the experimental results. An improved
design procedure is suggested based on the incorporation
of end-bends into the throughflow and iterative use of the
3D Navier-Stokes solver.
|
|---|
