Edward M. J. Naylor
Cecilia Ortiz Dueñas
Robert J. Miller
Howard P. Hodson
Whittle Laboratory,
University of Cambridge,
Cambridge,
Cambridgeshire CB3 0DY, UK
Optimization of Nonaxisymmetric
Endwalls in Compressor
S-Shaped Ducts
This paper presents a new design methodology for strutted S-shaped compressor ducts
that allows for more aggressive designs while maintaining current levels of duct loss. A
baseline duct geometry was selected, which had a radius change to length ratio that is
34% larger than current engine design limits. A large-scale low-speed model of the
baseline duct was experimentally tested. The flow in the corner between the hub and the
strut was found to separate due to the high local diffusion causing an increase in duct
loss. Area ruling was applied to the baseline duct and was predicted to reduce the size
and extent of the strut-hub corner separation, but the duct design was compromised. The
duct loss coefficient at midpitch was predicted to increase compared with that of the
baseline design. Nonaxisymmetric endwall profiling was then used on the duct wall,
locally to the strut, to remove the strut-hub corner separation and thus reduce net duct
loss, without compromising the duct design away from the strut. The endwall geometry
was produced by numerical optimization. It was shown that the net duct loss was insen-
sitive to casing profiling but highly sensitive to hub profiling. The optimal hub geometry
was experimentally tested and shown to completely remove endwall strut-hub corner
separation. The profiling was found to reduce the net duct loss by 16%. The paper shows
that the key benefit to endwall profiling is that it can be used to safely increase the size of
the design space in which aeroengine duct designers can operate.
DOI: 10.1115/1.3103927
1 Introduction
In aeroengines, an S-shaped annular duct is used to connect the
high pressure HP compressor spool to the upstream spool. The
duct is required because the spools are designed with different
mean radii. The inlet to the duct is at the larger radius. Because
the inlet and exit directions of the duct are approximately axial,
there are two bends in the duct. The curvature in the S-shaped
duct generates static pressure gradients normal to the streamlines.
In the first bend, this results in a rise in the casing pressure and a
fall in the hub pressure and, in the second bend, the opposite
occurs. This results in the flow close to the hub wall being sub-
jected to a small acceleration, followed by a large deceleration and
then finally a small acceleration Fig. 1. The flow close to the
casing wall is subjected to the opposite trends, with a small initial
deceleration followed by a large acceleration and then a small
deceleration.
In addition to the pressure gradients caused by the curvature,
the flow in the duct is subjected to streamwise pressure gradients
generated by changes in area from inlet to exit. Typically, com-
pressor ducts have a small overall area reduction, which helps to
reduce the severity of adverse pressure gradients by accelerating
the flow. In practice the magnitude of area change is set by the
particular engine architecture.
In many aeroengines large nonlifting struts intersect the
S-shaped duct. These serve two purposes. First, they are a struc-
tural element. Second they allow services, such as oil, cooling air,
and the radial drive shaft, to traverse the duct. These constraints
mean that the struts have a minimum cross-sectional area. Limi-
tations on strut length, set by both upstream potential forcing lim-
its and by the duct length, result in the strut having a relatively
high thickness-to-chord ratio, which is typically between 0.2 and
0.3 in modern engines.
The presence of the strut causes additional pressure gradients to
be imposed on the duct wall local to the strut. At the front of the
strut the flow stagnates. The flow is then accelerated to a velocity
greater than that which occurs without the strut being present.
Finally the flow is decelerated along the rear portion of the strut
Fig. 1. Close to the casing wall, the deceleration caused by the
rear part of the strut usually occurs in the region of the S-shaped
duct where curvature has resulted in a large acceleration. This
means that the deceleration is unlikely to cause boundary layer
separation on the casing. Close to the hub wall, the opposite oc-
curs and the deceleration at the rear of the strut usually occurs
over the region where duct curvature already results in a large
flow deceleration. The result is that if the combination of strut
diffusion and duct diffusion is too high, a corner separation will
occur in the flow between the strut and the hub wall.
The nondimensional design space of annular S-ducts unstrut-
ted depends on four main parameters. The four parameters that
are most commonly used are R / L, h
in
/ L, A
out
/ A
in
, and R
in
/ h
in
.
Increasing the first three parameters by either increasing the
change in radius, R, or reducing the length, L, or increasing the
exit area, has a similar effect on the duct’s performance. The large
deceleration on the hub wall increases in magnitude and the flow
begins to separate. This is commonly referred to as increasing
duct “loading.” If a strut is added to the duct an extra nondimen-
sional design parameter is added. This is the thickness-to-chord
ratio, t / c. As this parameter is raised the local diffusion along the
strut-hub corner rises moving it toward separation.
During the aeroengine design process, there is a requirement to
fix the annulus line early. This encourages the designers to “play
safe” so as to avoid any chance of strut-hub corner separation.
This often limits the design space in which the duct designer
operates. The consequence of such design decisions can lead to
nonoptimal designs of the neighboring compressor stages. It can
also result in an unnecessary rise in engine length and thus weight
Contributed by the International Gas Turbine Institute of ASME for publication in
the JOURNAL OF TURBOMACHINERY. Manuscript received September 1, 2008; final
manuscript received December 17, 2008; published online September 17, 2009. Re-
view conducted by David Wisler. Paper presented at the ASME Turbo Expo 2008:
Land, Sea and Air GT2008, Berlin, Germany, June 9–13, 2008.
Journal of Turbomachinery JANUARY 2010, Vol. 132 / 011011-1 Copyright © 2010 by ASME
Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 09/10/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use