AIAA AVIATION Forum 10.2514/6.

2016-3559
13-17 June 2016, Washington, D.C.
34th AIAA Applied Aerodynamics Conference

Influence of near-leading edge curvature on the

performance of aero-engine intake lips at
high-incidence

A. Coschignano∗ and H. Babinsky†

Department of Engineering, University of Cambridge, Cambridge, CB2 1PZ, UK

C. Sheaf‡ and E. Platt‡

Installation Aerodynamics, Rolls Royce Plc., Derby, DE24 8BJ, UK
Downloaded by Cranfield University on February 5, 2025 | http://arc.aiaa.org | DOI: 10.2514/6.2016-3559

This paper describes the investigation into the flow over the lip of subsonic engine intakes
at incidence, focusing on the shock wave-boundary layer interaction occurring over the
inner lip. A baseline geometry is considered along with two variations, characterised by a
sharper and a blunter intake highlight (i.e.: nacelle leading edge) respectively. Results to
date reveal a relatively benign interaction for the baseline model, with small or no shock-
induced separation reported under on-design conditions, which correspond to typical take-
off or climb circumstances. The alternative geometries reveal a considerable influence of
near-highlight curvature on the flow development. In particular, a blunter nose leads to the
formation of a larger supersonic region, terminated by a consequently stronger shock, which
shows a greater degree of shock-induced separation and increased total pressure losses and
unsteadiness. The sharp nose, on the other hand, resulted in the compression occurring via
three separate shock-waves, all of which weak. Overall, none of the three intake geometries
showed inherently unsteady behaviour. However, this is expected to occur as the engine
flow demand increases. Further testing is in progress to assess off-design performance and
to produce a complete operational envelope for intakes at incidence.

Nomenclature
κ Surface curvature
ρ Density
ṁ Mass flow
mx Modified super ellipse x exponent
n Modified super ellipse y exponent
a Modified super ellipse major axis
b Modified super ellipse minor axis
U flow velocity
M Mach number
P Static Pressure
Subscript
0 Property upstream of the shock - Stagnation value
1 Property at the tunnel entry
Abbreviations
AR Intake aspect ratio
LDV Laser Doppler Velocimetry

∗ PhD Student, Department of Engineering, University of Cambridge, AIAA Student Member.

† Professor of Aerodynamics, Department of Engineering, University of Cambridge, AIAA Associate Fellow.
‡ Installation Aerodynamics, Rolls Royce Plc, Derby.

1 of 13

American
Copyright © 2016 by Andrea Coschignano and Holger Babinsky. Published Institute
by the American of Aeronautics
Institute of Aeronautics and Astronautics
and Astronautics, Inc., with permission.
 I. Shock-Boundary Layer Interactions in Subsonic Engine Intakes
When travelling at an incidence, such as during take-off and climb, the substantial mass flow demand by a
turbofan engine is sufficient to accelerate the flow over the intake lip to supersonic conditions. This faster-
than sound flow pocket is terminated by a near-normal shock wave. The adverse pressure gradient imposed by
this disturbance on the boundary layer can cause the latter to separate, introducing large scale unsteadiness
and an increase in viscous losses. These losses have a direct negative repercussion on the overall engine
efficiency as the total pressure reaching the fan face is reduced. Moreover, if the separated boundary layer
does not reattach before the engine face, the unsteadiness, characteristic of separated flows, may increase the
stress on the fan, which can ultimately reduce component lives. A schematic representation of the typical
flow structure developing over an intake cross section is shown in Figure 1.
Although a significant amount of research has gone into reducing the detrimental effects associated with
shock-induced separation in transonic flight, the majority of these efforts have been limited to aerofoil design.
In fact, the formation of shock waves on the inside lip of engine intakes has often been overlooked and thus
there is insufficient understanding and a lack of data for CFD validation. Recently, engine manufacturers are
moving towards larger engines and opting for slimmer nacelles to enhance overall efficiency.1 The reason for
the latter is to be found in the reduced wetted area and, consequently, reduced drag. However, slimmer intake
Downloaded by Cranfield University on February 5, 2025 | http://arc.aiaa.org | DOI: 10.2514/6.2016-3559

lips naturally imply higher curvature near the leading edge, which may result in greater flow acceleration
and ultimately leading to stronger shocks. The consequences of novel design choices on the nature of the
interaction between the shock and the boundary layer is yet to be investigated, as the few parametric studies
performed on engine intakes to date were limited to purely incompressible flows.
The research currently undertaken at Cambridge University Engineering Department (CUED) not only
assesses the viability of a slimmer nacelle, but also investigate the response of shock-boundary layer inter-
actions (SBLIs) to geometry changes. The scope of this paper is to present the effect of near-leading edge
curvature on the flow-field over the inlet lip and its effect on the SBLI. Ideally, this will result in useful de-
sign criteria for next-generation engine intakes, which will not only satisfy the novel thickness requirements,
and thus reduce parasitic drag by design, but also minimise the stagnation pressure losses and unsteadiness
related to compressibility effects.

Fan Face
Centre-Line
Streamtube

Fan face
Shocklets
Supersonic Flow Shockwave Flow Reattachement

SWBLI

Separated Shock-Induced
Flow Separation
Diffuser
Stagnation
Point

Figure 1. Schematic representation of shock-induced boundary layer separation over an intake lip cross-section
during high-incidence flight.

II. Experimental set-up and methodology

A. Experimental facility
All experiments are performed at the high-speed aerodynamics laboratory at the University of Cambridge.
In particular, a blow-down wind tunnel, powered by two 50 kW compressors, is used. The flow is fed from
the compressors into the settling chamber, where it is passed through a number of flow straighteners and
turbulence grids before going through a 18:1 contraction. In the current configuration, no nozzle is used

2 of 13

American Institute of Aeronautics and Astronautics

 and the entry velocity is varied by changing the stagnation pressure and the effective area of the second
throat where the flow is chocked. By altering the cross-sectional area of the second throat by means of
an aerofoil (see Figure 2) the overall mass flow rate is adjusted, controlling the entry Mach number with
an accuracy of ±0.001. The working section was designed by Makuni2 exclusively for the investigation of
shock wave boundary layer interactions in engine intakes. The working section is depicted in Figure 2 and
features curved upper and lower walls, forming a stream-tube divided by a 1/7th scale intake lip model. The
stream-tube design is based on computational results performed for a real intake.2 The upper boundary
is a streamline of the computed flow far enough from the supersonic region to avoid choking in the upper
channel. The mass flow rate into the lower channel is controlled by choking the flow with an adjustable plug
as indicated in Figure 2. For a given stagnation pressure, by manually reducing the lower channel area, more
mass flow is forced into the upper channel mimicking a greater mass flow demand by the engine. As the ratio
between the upper and lower channel mass flows increases, the stagnation streamline is lowered, resulting
in stronger acceleration and thus a greater supersonic region. This allows the assessment of performance at
off-design conditions.
Downloaded by Cranfield University on February 5, 2025 | http://arc.aiaa.org | DOI: 10.2514/6.2016-3559

Streamtube 2nd throat airfoil

Flow from Settling 1
Chamber Upper
Channel
2 αa
3
Lower
Channel

Model
Dividing Streamline

Low. Ch. Choking

114mm 3

2 70mm
25mm 69mm
95mm

319mm

Figure 2. Representation of the blow-down wind tunnel working section. Stream-tube designed based on
computed flow streamlines. The position of the three points where pressure measurements are taken are
highlighted in red. Tunnel entry Mach number is measured at the location of Port No.1 and so was free
stream turbulence

The experimental operating range is portrayed in Figure 3. The rig is capable of achieving entry conditions
of M1 = 0→0.45, for α up to 29 degrees. The characteristic length used for calculating the Reynolds number
is the maximum intake lip thickness. In the range M1 = 0.25→0.45, the experiment is capable of matching
full scale Reynolds numbers for altitudes greater than 5000ft and 25000ft for a small and large engine
respectively. The highest supply pressure possible is 2.4 bar.2 A pair of windows on each side of the working
section allows high speed Schlieren photography and estimation of static wall pressure over the intake cross-
section by exploiting the properties of pressure sensitive paint (PSP).
The settings listed in Table 1 result in a flow field closely matching the target flow provided by both
experiments and Rolls Royce computational efforts based on a real intake travelling at a free-stream Mach
number of 0.3 and representative angle of attack, typical for take-off conditions. The wind-tunnel free stream
turbulence was measured using a hot wire at the tunnel entry. For M1 = 0.435, the turbulence level was
found to be < 1%.
The boundary layer displacement thickness upstream of the shock could not be measured directly by

3 of 13

American Institute of Aeronautics and Astronautics

 Absolute p0 (bar)
1.0 1.7 2.1 2.4 3.4
0.6

0.55
On design
0.5 operating point
Design limit
0.45

0.4

Mentry
0.35
Isobars
0.3
moverall
= 10 kg/s
0.25

0.2 Contours of constant

mass flow rate
0.15

0.1
0 5 10 15 20 25
5
Remax thickness (x10 )
Downloaded by Cranfield University on February 5, 2025 | http://arc.aiaa.org | DOI: 10.2514/6.2016-3559

Figure 3. Wind tunnel operational envelope in terms of entry Mach number (or equivalently mass flow rate).
Adapted from Makuni.2

Table 1. Tunnel entry conditions for on-design investigation

ṁ (kg/s) M1 AoA (deg.) P0 (kPa) T0 (K) I%

8.68 0.435 23 211.6 290±4 0.88

LDV due to poor data rate in the immediate proximity of the wall. From qualitative data observation, this
is estimated to be <0.2mm. This is particularly important when determining the effective aspect ratio of
experimental facilities and assessing the impact of corner flows on centre span separation. According to
the findings by Bruce et al.,7 the resulting δ ∗ /w ratio, where w is the tunnel width (114mm for the CUED
facility), would suggest the absence of strong 3D effects and the corner flows are, thus, not expected to
affect the onset of shock-induced separation in the centre-span. Further support to this hypothesis will be
considered when discussing experimental results.
A number of different techniques is used to investigate the flow. In particular, a Schlieren technique is
used to visualize the features typical of supersonic flows such as shock waves. A light source is placed at
the focus of a mirror, which produces collimated light that is passed through the tunnel working section,
where changes in density occurs as a consequence of compressibility effects in the transonic regime. The
light is then collected by a second mirror that focuses all the rays. However, the density gradients within the
working section have produced a deflection of some of the rays. As a result, these will not be concentrated
at the focal point. A knife edge is placed here to intercept some of the rays, while other will go through
without being obstructed. This ultimately results in dark patches where the beams were blocked by the
obstruction and brighter patches where the rays deviated by the density gradients were not intercepted by
the knife edge. In such a system, the resulting light intensity is proportional to the first derivative of the
light retardation, allowing the visualization of shock waves, as well as expansion and compression waves.3
Pressure measurements are taken during the run to both characterise the flow and to assess the experi-
mental repeatability. These measurements are obtained by using pressure taps drilled along the surface of
the intake model and connected via tubing to a differential pressure transducer. Furthermore, a number of
these pressure readings are used to calibrate pressure sensitive paint. The photons reflected when the latter
is excited by UV light are directly proportional to the pressure acting on it. An optical device is used to
measure such luminescence, resulting in a number of pressure readings equivalent to the camera resolution
available. The relationship between luminescence intensity and pressure is determined by the Stern-Volmer

4 of 13

American Institute of Aeronautics and Astronautics

 relation:4
Iref P
= A(T ) + B(T ) (1)
I(P,T ) Pref
Iref is the luminescence taken at a reference pressure and temperature. The obvious choice is to measure
the intensity at normal atmospheric conditions with the tunnel off. A second image is taken with the flow
on. A no flow - no light background image is subtracted to both images to reduce noise from external light
sources. Finally, accordingly to the Stern-Volmer equation, the flow-off reference image is divided by the
flow-on image to obtain the pressure ratio. By using the pressure taps located on the model, the values of
the constants A(T ) and B(T ) can be determined and the absolute pressure values can be extracted from
raw light intensity. To minimise the calibration error, estimated to be <2% for a typical calibration curve in
the current investigation, 5-6 points are required.4 Given the rig configuration and the position of the access
windows, the images are captured at an angle and are, thus, characterised by severe prospective distortion.
For this reason, along with the fact that, due to the curved nature of the intake, extracting pressure readings
along the surface requires the photo to be flattened, all the PSP images have been processed via the same
un-warping algorithm and will be presented as a two dimensional contour plots.
Furthermore, streak-lines are visualised by means of a dense mixture applied on the model prior to the
Downloaded by Cranfield University on February 5, 2025 | http://arc.aiaa.org | DOI: 10.2514/6.2016-3559

run. This results in the skin-friction lines being deposited on the surface, offering an accurate depiction of
the flow structures. Shock waves will cause an accumulation of oil near the foot. Similarly, the stagnation
point will be characterised by greater oil residuals. Most importantly, this technique becomes extremely
valuable when determining the extent of shock induced separation. In fact, the flow recirculation, typical
of separated boundary layers, is captured by the oil mixture. The mixture is a combination of Titanium
Dioxide (TiO2 ) powder, used as a pigment, and Kerosene (or paraffin), ideal for its low kinematic viscosity.
Generally, a small amount of oleic acid is added to improve the mixing and reduce the coagulation of TiO2
powder.
Finally, direct velocity measurements are taken by using Laser Doppler Velocimetry (LDV). A two com-
ponent LDV system by TSI is used. Two pairs of coherent laser beams, with a wavelength of 542nm and
532nm respectively, are focused inside the working section to form the interference pattern of the ellipsoidal
working volume, measuring 130µm in diameter. Kerosene particles, with a diameter of approximately 0.5µm,
are used to seed the flow and allow velocity measurements to be recorded via a proprietary software. The
laser emitting head and receiving optics are mounted on a traverse capable of moving in one direction with
a user defined velocity. Typical measurement accuracy is ±0.1% of Umax (∼580m/s).
Stagnation temperature is recorded by using 4 T-type thermocouples placed in the settling chamber.
A linear drop of stagnation temperature from ∼294K to ∼286K is recorded during an average 30 second
long run. This variation in stagnation temperature is taken into account when converting absolute velocity
measurements from LDV to local Mach number to minimise the error involved, which would otherwise peak
1.7%.

B. Intake Lip Design

The intake lips considered here have been designed by using a modified super ellipse profile. Mathematically,
a super ellipse is defined as:
 m  
x−a y n
+ =1 (2)
a b
where a and b are the major and minor axis of the ellipse, respectively controlling the position and the size
of the ellipse highest point. In intake terms, such point indicates where the intake is at its thickest, i.e.: the
throat. The ratio ab is defined as the aspect ratio AR of the ellipse. The powers m and n, on the other hand,
set the locus of the point of maximum curvature with smaller values associated to sharper noses. Moreover,
to reduce the number of parameters involved in the geometry definition, a modified super ellipse (MSE)
approach was preferred for the current parametric study as used by Lin5 and Schrader,6 amongst many, for
their flat plate boundary layers investigations. A modified super ellipse is defined similarly to Equation (2).
However, to further reduce the number of variables, m is not chosen arbitrarily, but varies between 2 and 3
along the major axis of the ellipse according to the equation:5
 x 2
m(x) = 2 + (3)
a

5 of 13

American Institute of Aeronautics and Astronautics

 Combining Equation (2), (3) and keeping the aspect ratio constant, a number of modified elliptical
profiles can be obtained, defining a family of curves for different values of the exponent n only. This type of
ellipse results in a continuous reduction in curvature, which falls to zero at the point of maximum thickness.
This characteristic has made the super ellipse a good candidate for leading-edge design of flat plates or other
similar purposes where a continuous curvature distribution is required.5 However, if an subsonic engine intake
is considered, it is noted that, at the point of maximum thickness (i.e.: the intake throat), the curvature does
not drop to zero but the surface simply changes concavity. After this point, beyond which a super ellipse is
not defined, the intake lip is smoothly connected, by means of a bridge spline, to the rest of the inner intake,
which uniformly decelerates the flow in the diffuser section. This is done by forcing the first two derivatives
to match at the joint between the front and the rear part of the intake model. Figure 4 shows how the
parameters defined in Equation (2) relate to the design of a real intake lip. The baseline geometry is defined
by a modified super ellipse of aspect ratio AR < 3 and a value of n = 1.98. Figure 4 also shows a plot of the
theoretical curvature distribution along the surface of the three lips considered in the current investigation.
Note that the term leading edge and highlight, referring to the intake models, are used interchangeably in
the following paper.
Center-Line
Downloaded by Cranfield University on February 5, 2025 | http://arc.aiaa.org | DOI: 10.2514/6.2016-3559

RHighlight
RThroat
Lip ds b
Profile
Fan face Decreasing n
b a
a

Diffuser
0.08
n=1.70
n=1.98
0.06 n=2.20
curvature

0.04
Decreasing n
0.02

00 0.2 0.4 0.6 0.8 1

s/s
a
Higlight Throat

Figure 4. Variation in curvature distribution along the nacelle lip upper surface for decreasing highlight
bluntness.

III. Results
A. Baseline Flow
As aforementioned, the target flow was provided by validated computations performed on 3D intakes. The
stagnation pressure achieved in the wind tunnel, equivalent to 2.1bar, results in a Reynolds number of ∼ 106 ,
representative of a medium size engine at take off. The area ratio between the second throat and the tunnel
entry was chosen to give a Mach number at the tunnel entry of 0.435. The latter is set with an accuracy of
±0.001 and was determined based on computations. Furthermore, the stagnation point, and, consequently,
the shock wave position were finely tuned to match the target by reducing the area of the underside passage,
effectively shifting the balance between upper and lower mass flows.

6 of 13

American Institute of Aeronautics and Astronautics

 The resulting flow field is shown in Figure 5. The flow is rapidly turned around the intake highlight and
accelerates beyond the local speed of sound via expansions fans created near the front part of the intake. As
these intersect the constant pressure boundary at the free-stream interface, they are reflected as compression
waves, which coalesce into the physical plane to form a shock-wave. The interaction between such a shock-
wave and the boundary layer, from Schlieren analysis only, appears relatively benign. No clear lambda shock
pattern, characteristic of interactions resulting in flow separation, is present. The absence of a clear lambda
shock wave is confirmed by LDV investigation, which shows a peak local Mach number of 1.47 in the inviscid
flow region upstream of the shock. However, adverse pressure gradients characterising high incidence flight
start decelerating the flow upstream of the shock, resulting in an effective Mach number immediately before
the shock foot of 1.4 (Figure 6). This is somewhat beyond the generally accepted separation limit of Mach
1.3. Recirculation regions may therefore be present but confined to a small region of space, which makes
their identification more challenging. Surface oil flow visualization suggests the presence of a shock-induced
separation line, defined by a number of critical points, namely saddles and nodes, and three-dimensional
bubbles recirculating around their foci. This is partially shown in Figure 7, which combines wall static pres-
sure from PSP with relevant streak-lines extracted directly from oil-flow sediments. An actual photograph
of the oil flow deposit on the surface is offered in the following section for comparison with the two intake
Downloaded by Cranfield University on February 5, 2025 | http://arc.aiaa.org | DOI: 10.2514/6.2016-3559

variations (Figure 12b)

Expansion fan Shock-wave

Figure 5. Schlieren photograph depicting the flow development over an engine intake lip travelling at an
incidence of 23 degrees during take-off conditions. No flow image subtracted to eliminate impurities and
increase contrast.

Mach
15 1.5

1.4

1.3

1.2
z (mm)

1.1

0.9

0.8
0
0.7
44 50 62
x(mm) * 56

* From leading edge 0.1 0.12 0.14

x/c

Figure 6. Local Mach number contour obtained by 12 normal LDV measurements. Noise near the wall due
to difficulties in obtaining good seeding as a result of the highly curved surface.

Furthermore, the severity of the pressure jump imposed by the shock-wave, would suggest that the
boundary layer might have transitioned to a turbulent nature somewhere upstream of the shock foot in
order to withstand the pressure gradient.

7 of 13

American Institute of Aeronautics and Astronautics

 P/P0
0.5 0.65

0.4
0.6
0.3
0.55
0.2

Span-wise direction 0.1 0.5

0
0.45
-0.1
0.4
-0.2

-0.3 0.35

-0.4
0.3
-0.5
20 40 60 80 100 120 140
dS*
Downloaded by Cranfield University on February 5, 2025 | http://arc.aiaa.org | DOI: 10.2514/6.2016-3559

Figure 7. Wall static pressure from processed PSP images. Black streamlines extracted from oil flow. Three
dimensional bubbles appear to be localised in small region of space. Recirculation in the outer region is due
to the interaction between the side-wall boundary layer and the shock.

Overall, it can be concluded that the flow for a typical on-design case, representative of take-ff conditions,
is relatively benign with only small separated flow areas caused by the presence of a Mach 1.4 shock wave.
This is expected to change as the mass flow demand by the engine is increased and it will be investigated
in the upcoming future. The effect of varying highlight geometry will now be investigated and compared to
the baseline flow.

B. Impact of highlight (leading-edge) sharpness

The effect of varying the highlight geometry on the strength and position of the shock-wave and, most
importantly, on the nature of its interaction with the incoming boundary layer will be now discussed. The
stagnation pressure, relative angle of incidence and entry Mach number were kept constant, as well as the
ratio between upper and lower channel.
Figure 8 shows a side by side comparison between the three different geometries characterised by an
increasing highlight bluntness, defined by a greater MSE exponent. A qualitative analysis can be started
based on Schlieren photographs alone. Whereas the flow that develops over the bluntest intake geometry
considered appears to differ only slightly from the baseline, the shock structures characterising the sharper
nose are significantly different. Here, three separate shock-waves are observed. The first sits very close to
the highlight and is a consequence of the abrupt acceleration of the flow by the strong pressure gradients
characterising sharp turns. The presence of a small lambda structure suggests that the interaction is sep-
arated. The flow reattaches and undergoes further acceleration before reflected expansion fans cause the
second shock. The latter would appear to be of turbulent nature. However,quantitative measurements of
the shock strength are required to reinforce this hypothesis. No lambda is observed. As a result of the
supposedly weak shock, the flow is just below the sonic limit and becomes faster than sound again as a
consequence of a third acceleration stage, which ultimately result in a third, even weaker shock wave further
downstream.
The blunter nose, on the other hand, appears similar to the baseline flow. The shock position along the
stream wise axis is not significantly different, which is to be expected given the similar nature of surface
curvature in this region. However, a closer inspection of the interaction region shows a lambda structure
appearing, suggesting a degree of boundary layer separation. The viscous layer downstream of the interaction
appears to be thicker as a consequence of this. This qualitative evidence would suggest a higher Mach number
at the shock foot. This hypothesis is confirmed by quantitative measurements from laser velocimetry and
isentropic values from pressure readings, both of which are discussed below.
Velocity measurements were taken by means of both stream-wise readings across the inviscid region of

8 of 13

American Institute of Aeronautics and Astronautics

 n: 1.7 Baseline n: 1.98 n: 2.2

Bluntness +

Figure 8. Qualitative flow comparison between the three different geometries from Schlieren photography.
Downloaded by Cranfield University on February 5, 2025 | http://arc.aiaa.org | DOI: 10.2514/6.2016-3559

the shock-wave, and by a series of normal traverses to obtain the velocity field in the shock-boundary layer
interaction region. The former are depicted in Figure 9. Velocity contour plots for the region of interest of
each shape are, on the other hand, depicted in Figure 10. The increased noise as the measurement volume
approaches the surface is due to the difficulties encountered with seeding the flow near the wall. Improved
seeding and acquisition techniques are currently being attempted with the purpose of reducing noise.
These measurements show good agreement with the velocity estimate from static pressure measurements
and provide means of characterising the shock wave strength. Due to the high curvature near the sharp
nose leading edge, no LDV measurement across the first interaction could be obtained. The only available
measurement is a pressure tap. The location of the latter is immediately upstream of the shock-wave and
recorded a value corresponding to an isentropic Mach number of 1.3. This is not expected to be able to
separate a turbulent boundary layer. Thus, the interaction might be laminar. Further support to this
theory comes from the extent of the interaction length. The latter is defined as the distance between the
point where the incoming boundary layer starts to feel the upcoming pressure rise and the location of an
ideal inviscid shock. A high value for the ratio between interaction length and the local boundary layer
thickness, such as the one inferred from qualitative data, is generally associated with a laminar interaction.
The central shock-wave, on the other hand, is a relatively weak Mach 1.3 shock, as measured by both stream
wise traverses and velocity contours in the SBLI region. The resulting downstream velocity is just below
Mach 1 and accelerates to approximately Mach 1.15 before a combination of adverse pressure gradients and
a very weak final compression through a shock decelerate it to subsonic condition in the diffuser portion of
the intake.
As seen from qualitative measurements, the two less aggressive blunter nose geometries, on the other
hand, present only one strong shock-wave. For the baseline case (n=1.98), the peak Mach number in the
inviscid stream is just short of 1.5, before undergoing a short diffusion due to adverse pressure gradients
prior to coalescence of the compression waves into ∼ Mach 1.4 shock-wave. The velocity distribution in
Figure 10 suggests a smeared shock foot. It is possible that there is a small lambda but this is not resolved
by the measurements. In contrast, for n=2.2, corresponding to a blunt nose, a distinct lambda, albeit small
in size, is reported. Similarly to the baseline case, the flow undergoes a small degree of deceleration before
going through the shock. However, for the rounder nose both the peak Mach number upstream, and the one
at the shock foot are higher, sitting at ∼1.57 and ∼1.48 respectively, consequence of the greater extent of
favourable pressure gradient around the blunter nose.
The diffuser geometry is identical for all the variations considered and the flow velocity sufficiently
downstream of the shock approaches the same velocity, just above Mach 0.9, as confirmed by both LDV
(Figure 9) and static pressure measurement (Figure 11). Immediately downstream of the shock, however, the
baseline and the blunter variation nose geometries show a greater degree of local flow acceleration. Shocklets
are formed as a consequence of substantial acceleration over the thicker viscous layer. This is ultimately
reflected in static wall pressure. Figure 11 provides a good example by depicting isentropic Mach number
estimated directly from pressure sensitive paint analysis. The Mach line, highlighted in black, is substantially

9 of 13

American Institute of Aeronautics and Astronautics

 1.6
n=2.20
1.5 n=1.98
n=1.70

1.4

Mach Number
1.3

1.2

1.1

0.9
0 0.03 0.06 0.09 0.12 0.15 0.18
x/c

Figure 9. Stream-wise LDV measurements across the inviscid core of the shock-wave for the three different
geometries. Both the second and third compression captured for the sharper lip.
Downloaded by Cranfield University on February 5, 2025 | http://arc.aiaa.org | DOI: 10.2514/6.2016-3559

Mach
15 17 1.5
11
1.4
1.3
10 1.2

z (mm)
z(mm)

z(mm)

5.5 1.1
1
5 0.9
0.8
0 0.7
0 0
0 3 6 9 0 9 18 0 7.5 15
x(mm) x(mm) x (mm)
(a) Sharp, n=1.7 (b) Baseline, n=1.98 (c) Blunt, n=2.2

Figure 10. Mach number contour plots of the main shock wave boundary interaction for the three different
models. Mach number obtained directly from LDV measurement after temperature correction. a) depicts the
central shock wave occurring over the surface of the sharp nose.

downstream of the effective pressure jump due to the shock, reflecting the presence of a small supersonic
pocket characterised by secondary shocklets. It should also be noted that both isentropic Mach from PSP
and direct LDV measurements are in good agreement.
Wall static pressure, obtained from PSP analysis (Figure 11), provides a tool to assess flow two dimen-
sionality and further strengthen the observation that side-wall effects are limited and do not greatly affect
the main flow along the centreline, as anticipated in Section IIA. Pressure sensitive paint images show nearly
non-existent corner flows for the sharp geometry, which is consistent with a weak centre-span interaction.
Similarly, recirculation regions near the wall are not found to compromise the baseline flow as these are
confined to a small region of space. These tunnel effects become more evident as the centre span SBLI grows
stronger, as it can be inferred from Figure 11c. Nonetheless, the corner flows are found to be dominant
in less than 30% of the total span-wise distance and are not thought to have major impact on the extent
of centre-span separation. Oil flow visualization offers a clearer depiction of the vortical structures char-
acterising corner flows for both the baseline and the blunt nose (Figure 12b and 12c) and strengthens the
assumption of negligible impact of wind-tunnel effect on centre-span separation onset. On another hand, oil
flow also confirms the presence of shock-induced separation lines for two of the three geometries considered.
As depicted in Figure 12b, the baseline flow streak lines present a distinct separation line defined by sad-
dle and nodal points in correspondence with the location of the shock-wave. Whereas this is a fairly two
dimensional attribute, the reattachment appears to be dominated by three-dimensional features. A number
of owl-face separations are present, with four foci clearly distinguishable. The reattachment lines at those

10 of 13

American Institute of Aeronautics and Astronautics

 span-wise locations where owl-face separations are not present are not clearly captured by the oil mixture.
The blunter nose shows similar flow features. Overall, separation appears more pronounced for the blunter
geometry, which is consistent with the other measurements. Finally, for the sharpest intake geometry, shock
induced separation is present in the form of extremely small bubbles in the immediate proximity of the
highlight. A subtle saddle point would appear to be present in the centre-span, as a result of the second
shock-wave. The reattachment line is not distinguishable but believed to be immediately downstream of the
separation point.

IV. Conclusion
The flow development around an aero-engine intake travelling at an incidence has been investigated and the
results for a baseline case and two geometry variations are presented here. The baseline flow, for which target
was provided by 3D CFD on nacelles, showed a shock-wave boundary layer interaction with a strength of
M=1.4, which is around the onset of shock induced separation. Oil flow visualisation suggests that a small
separation exists in the form of small, localised, 3D pockets. Two leading edge geometries have been tested
for equivalent flow conditions. The experimental measurements reveal a substantial influence of near-leading
Downloaded by Cranfield University on February 5, 2025 | http://arc.aiaa.org | DOI: 10.2514/6.2016-3559

edge curvature on the supersonic flow development. Compared to the baseline case, a sharper nose results
in strong local acceleration, which leads to a first terminal shock in the immediate proximity of the leading-
edge. The flow undergoes two further re-acceleration stages, both terminated by shock waves of decreasing
intensity. The flow shows an exceptional two-dimensionality and separation bubbles are absent for the two
SBLI occurring over the upper part of the lip. Conversely, the favourable pressure gradients characterising
blunter noses favour the development of a larger supersonic region, terminated by a stronger shock, which
appears to be associated with a degree of flow separation. At this stage stagnation pressure recovery has
not yet been measured. However, there is sufficient evidence to support the hypothesis that a sharper nose
would reduce the wave drag losses as compression occurs via a number of shock, thus closer to an isenstropic
phenomena.
Nonetheless, for the flow conditions considered in the current investigation, chosen to investigate the
interaction between the shock and the boundary layer during typical high incidence manoeuvring, appear
to be relatively benign for all of the shape considered. Ultimately, none of them showed inherently un-
steady characteristics associated with largely separated flows. Future research will try to assess the onset of
unsteadiness before attempting its control.

Acknowledgements
The authors wish to acknowledge David Martin, Sam Flint, Anthony Luckett and John Hazlewood for
operating the CUED blow-down wind tunnel and Kevin Bullman for the manufacturing of the models used
in this investigation. Moreover, they would like to thank Rolls Royce Plc and the Engineering and Physical
Sciences Research Council (EPSRC) for funding the current research.

References
1 DAGGET, D. L., BROWN, S. T., and KAWAI, R. T., “Ultra-Efficient Engine Dimaeter Study,” NASA Technical Report

No. 212309 , 2003.

2 MAKUNI, T. E., Experimental Investigations of the Interaction between Normal Shock Waves and Transitional Boundary-

Layers in Subsonic Intakes at High Incidence, First Year Report, University of Cambridge, august 2013.
3 FERRI, A., Elements of Aerodynamics of Supersonic Flows, Macmillan Co., 1949.
4 GREGORY, J. W., ASAI, K., KAMEDA, M., LIU, T., and SULLIVAN, J. P., “A review of pressure-sensitive paint for

high-speed and unsteady aerodynamics,” Proc. IMechE Vol. 222 Part G: J. Aerospace Engineering, 2007.
5 LIN, N., REED, H. L., and SARIC, W. S., “Effect of leading-edge geometry on boundary-layer receptivity to freestream

sound,” Instability, Transition, and Turbulence, 1992, pp. 421–440.

6 SCHRADER, L.-U., BRANDT, L., MAVRIPIIS, C., and HENNINGSON, D. S., “Receptivity to free-stream vorticity of

flow past a flat plate with elliptic leading edge,” Journal of Fluid Mechanics, Vol. 653, 2010, pp. 245–271.
7 BRUCE, P. J. K., BURTON, D. M. F., TITCHENER, N., and BABINSKY, H., “Corner effect and separation in transonic

channel flows,” Journal of Fluid Mechanics, Vol. 679, 2011, pp. 247–262.

11 of 13

American Institute of Aeronautics and Astronautics

 Mach
0.5
1.6

1
0.4

1.5
0.3

1
0.2 1.4

Span-wise direction
1

0.1 1.3
1

0
1.2
-0.1
1.1

1
-0.2
1
-0.3
1

1
-0.4 0.9
1

-0.5 0.8
20 40 60 80 100 120 140
dS
Downloaded by Cranfield University on February 5, 2025 | http://arc.aiaa.org | DOI: 10.2514/6.2016-3559

(a) Sharp highlight, n=1.7.

Mach
0.5
1.6
1

0.4
1.5
0.3
1

0.2 1.4
Span-wise direction

0.1 1.3
1

0
1.2
-0.1
1

1.1
-0.2
1 1
-0.3
1

1
1
1

0.9
-0.4 1 1
1

-0.5 0.8
20 40 60 80 100 120 140
dS

(b) Baseline, n=1.98.

Mach
0.5
1
1.6
0.4 1

1.5
0.3
1

0.2 1.4
Span-wise direction

0.1 1.3
1

0 1
1.2
1

-0.1 1

1
1.1
1

-0.2 1

1
-0.3 1

0.9
-0.4 1

-0.5 0.8
20 40 60 80 100 120 140
dS

(c) Blunt highlight, n=2.2.

Figure 11. Comparison between isentropic Mach number calculated from wall static pressure along the model
surface measured from PSP. Mach line highlighted in black. Top right corner and b) affected by poor lighting.

12 of 13

American Institute of Aeronautics and Astronautics

 Small separated pockets

Saddle point

Final compression stage

Downloaded by Cranfield University on February 5, 2025 | http://arc.aiaa.org | DOI: 10.2514/6.2016-3559

(a) Sharp highlight, n=1.7.

Shock induced
separation line Corner separation
Saddle point
focus

Owl face
separation

(b) Baseline, n=1.98.

Saddle
Corner flows

Separation line

(c) Blunt highlight, n=2.2.

Figure 12. Oil flow visualization image. Pressure tap located at identical stream wise positions.

13 of 13

American Institute of Aeronautics and Astronautics