You are on page 1of 22

39th AIAA Fluid Dynamics Conference AIAA 2009-4210

22 - 25 June 2009, San Antonio, Texas

Simulations of a Low-Boom, Axisymmetric, External


Compression Inlet with Bleed

T. Coyne1, J. L. Koncsek2, E. Loth3,


Department of Aerospace Engineering, Urbana, IL, 61801

D. Davis4
Aerojet, Sacramento, CA, 95813

T. Conners5, D. Howe6
Gulfstream Aerospace Corporation, Savannah, Georgia, 31402

The NPARC designed code, WIND-US, has been used in a study of low boom axisymmetric external
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

compression inlets with and without throat slot bleed. To validate the Reynolds-averaged Navier-Stokes
(RANS) approach to predict such flows, two simulations of an experimental slot bleed flow (with and without
an oblique shock) from AIAA-95-0032 have been performed. A computational analysis was then conducted
on WIND-US of three different axisymmetric inlets, designated C0, A4, and K1, designed to achieve low
boom at Mach 1.70. All computations were run at Mach 1.67 to be consistent with the operational limits of
the 8’x6’ NASA Glenn wind tunnel to be used in future experimental testing. A variety of different operating
conditions and geometry modifications were investigated. It was found that the A4 inlet has the best recovery
characteristics. However K1 has the possibility of attaining a more stable flow field without significant
detriments in terms of pressure recovery. The inlet performance on the C0 is not significantly sensitive to
Reynolds numbers. Small changes in Mach number on the C0 led to variations in the overall total pressure
recovery, consistent with that expected from quasi 1-D shock relations. The overall ‘boom’ characteristics of
the inlet did not significantly change when either the Reynolds or Mach numbers were altered. It was found
that moving the cowl forward with respect to the centerbody reduced the boom signature of the inlet. The A4
was used to show that reducing the cowling angle significantly decreased the sonic boom signature. Given the
presence of shock boundary layer interactions within the inlet, slot bleed was implemented on the centerbody
of the K1 just downstream of the terminal normal shock. At the optimized plenum pressure, 25.1% of the
freestream stagnation pressure, the bleed and solid wall cane curves behaved vary similarly. The slot leading
edge geometry had little effect upon integrated recovery, but the elliptic leading edge was found to bleed more
efficiently. Boom performance was very similar to that found with the solid wall geometry, however bleed
has the ability to allow for stable operation at lower inlet mass flow ratios and thus lower boom levels.

Nomenclature
AIP = Aerodynamic Interface Plane – the site of the engine fan face
Boom = A measure of the far-field sonic boom characteristics
ELE = Elliptic Leading Edge – one of two analyzed leading edges for the bleed geometry
MFR = Mass Flow Ratio – the AIP mass flow rate normalized by the inlet capture mass flow rate
NLE = NACA 0012 Leading Edge – one of two analyzed leading edges for the bleed geometry
PP*/Pt0 = Static bleed plenum pressure normalized by freestream total pressure
Q = Bleed mass flow normalized by bleed slot choked mass flow
MFR-Bleed = Bleed mass flow normalized by inlet capture mass flow
REC = Area integrated AIP total pressure normalized by freestream total pressure
SW = Solid Wall

1 Research Assistant, AIAA Student Member


2 Engineering Specialist
3 Professor, Associate Fellow AIAA
4
Engineering Specialist, Senior Member AIAA
5 Principal Engineer – Propulsion, Preliminary Design Department, M/S R-07, Associate Fellow AIAA
6 Staff Scientist – Aerodynamics, Preliminary Design Department, M/S R-07, Associate Fellow AIAA

1
American Institute of Aeronautics and Astronautics

Copyright © 2009 by T. Coyne, J. L. Koncsek, E. Loth, D. Davis, T. Conners, D. Howe. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
I. Introduction
or the past sixty years the objectives of external compression inlet technology for supersonic aircraft have been
F high total pressure recovery for propulsion efficiency, low distortion for engine operability, and flow stability
for flight safety. However, a more recent area of study is design of inlets with sonic boom reduction as a key
design parameter. A few key design features have been incorporated into a new inlet design to help minimize the
boom signature projected to the ground. One such feature is the shaping of the external cowl. Traditionally, there is
a non-zero cowl angle in external compression inlets, which is also present in the cowl external surfaces. The
oblique shock produced by the external cowl surface may result in excessive boom signature on the ground. In
terms of inlet performance, nominally aligning the cowl with the flow produced by the ramp or centerbody provides
the highest performance. However, if sonic boom reduction is the key design constraint, a cowl with a near-zero
external cowl angle can significantly reduce the strength of the oblique shock wave yielding better boom
performance, though potentially at the expense of inlet performance. A near-zero cowl angle requires a relaxed
isentropic compression surface[2] approach which defocuses the shock near the inner cowl lip and reduces the flow
turning at the inlet cowl lip. To maintain an adequate diffusion profile with a flatter cowl surface, the centerbody
must turn inward more rapidly downstream of the throat. The more abrupt centerbody turning thickens the hub side
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

boundary layer and increases susceptibility to boundary-layer separation. Therefore, flow control may be required
to maintain high pressure recovery and ensure stable inlet operation.
One common method for shock boundary layer interaction control is bleed. The objective of bleed is to
suppress the boundary layer separation and stabilize the normal shock position on the centerbody and to provide a
mechanism to stabilize inlet flow during transient disturbances. Upstream disturbances can arise from freestream
gusts or flight maneuvers, while engine airflow demand transients can generate downstream disturbances. This
paper analyzed three different inlets designs making use of both zero cowl angle and a relaxed isentropic
compression surface. Additionally, the impact of bleed was determined.

II. Methodology
A. Inlet and flowfield specifications:
The axisymmetric low-boom inlet considered is a 12-inch diameter 1:4.85 scale model of the full geometry.
The scale model is intended for supersonic wind tunnel testing in the 8’x6’ wind tunnel at NASA Glenn Research
Center in Cleveland, Ohio. Inflow properties are specified to be consistent with wind tunnel test conditions. The
inflow conditions were Mach number of 1.67, a stagnation pressure of 21.85 psi, and stagnation temperature of
390.8°R. The engine fan face pressure was approximated in the CFD analysis by a static pressure outflow boundary
condition, meant to simulate the massflow plug to be used during the testing. Two different types of inlets were
analyzed: a solid wall geometry (SW) without any bleed and a geometry with a slot bled plenum section. Within
these classifications a variety of different geometries with slightly different aero lines were examined, however
focusing on the overarching categories (solid wall and bleed), figures 1 and 3 present each geometry, respectively,
with the important dimensions. Both figures also show Mach contours.
B. Grid Generation:
All computational meshes used in this study were generated using Gridgen V15. Structured meshes were
used to fully resolve boundary layers and capture flow features. Grid stretching factors were specified to ensure
nodal distributions capable of fully resolving the viscous sub-layer by setting the first grid point away from the wall
to coincide with y+~1 for the flow. Grid nodes were clustered around the inlet terminal normal shock region to
maximize computational accuracy. While multi-zone grids were utilized for parallel computation the zonal
boundaries were chosen so as to avoid shock/zonal boundary interactions. The WIND-US utility GMAN was
utilized to specify all boundary conditions. The solid wall grid was separated into 8 different domains and contains
approximately 410,000 grid points. The bleed grid contains 9 different zones with 525,000 grid points. A picture of
the bleed grid is provided as figure 3. The solid wall grid is not shown, but it is very similar to that of the bleed grid
except that the plenum is removed and there is a viscous solid wall boundary in its place.
C. Numerical Methodology
Simulations were run with the code WIND-US [1], a code developed by the NPARC alliance. The WIND-
US package supports the solution of the Euler and Navier-Stokes equations and allows for modeling of both
turbulent and reacting flows. An extensive study of various spatial-discretization schemes, time-integration
schemes, and CFL numbers were performed to determine the optimal numerical approach for both computational
accuracy and speed. A more detailed explanation of this study can be found in Rybalko, et al. 2009. The results of
the studies demonstrated that the approximate factorization alternating direction implicit (AF ADI) time integration
scheme, the Roe_over (an implementation of the OVERFLOW version of the Roe scheme), and a CFL number of
2.5 each achieved optimum computational accuracy in the shortest run time. A separate study was performed to
2
American Institute of Aeronautics and Astronautics
determine the optimum TVD limiter and the results indicated that both the Koren and MINMOD 2.0 limiters were
the best and yielded very similar results. The MINMOD 2.0 limiter was found to be slightly more accurate with the
bleed studies, and for consistency was used for all computational runs. All computations were performed using two-
dimensional grids coupled with the WIND-US axisymmetric keyword, which creates an axisymmetric domain with
a specified angle of revolution[1].
Convergence for solid wall runs was determined by plotting ratios of mass flow ratio (MFR) and stagnation
pressure recovery (REC) (see equations 1 and 2, respectively, in section V) as a function of the number of iterations.
Solutions were considered converged once the MFR values changed by less than ±0.001 and REC by less than
±0.005 every 2,000 iterations. This method of convergence monitoring was chosen after it was observed that the
flow field continued to change even after the L2 norms of the Navier-Stokes solver stabilized. For the solid wall
runs, convergence was achieved in 15,000 iterations. For bleed studies the bleed mass flow (MFR (Bleed)) (see
equation 6 in section V) was found to be the best indicator of convergence given that MFR and REC values
stabilized much quicker than values of MFR (Bleed). Once the change in MFR (Bleed) was less than 0.0001 every
2,000 iterations the runs were considered converged. For bleed studies, convergence was not achieved until after
35,000 iterations.
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

III. Fundamental Bleed Study


Two validations of experimental slot bleed flows were performed with WIND-US to examine computational
accuracy. Both validations were based upon experimental results in AIAA-95-0032, Flowfield Measurements in a
Slot-Bled Oblique Shock-Wave and Turbulent Boundary-Layer Interaction [7].
A validation of a Mach 2.46 oblique shock slot bleed flow from AIAA-95-0032 was performed. The slot
was 1 cm wide and had an incoming boundary layer of 3.06 cm, measured at an inflow plane 45.60 cm ahead of the
slot bleed section. In the experiment, an oblique shock wave generator was set at an angle of 8˚ degrees and it
generated a shock which interacted with the boundary layer just ahead of the slot bleed section. In the experiment,
flow measurements were taken inside the bleed slot and presented in a series of contour plots [7]. The computational
grid to simulate the experiment utilized had 50 horizontal grid nodes and 200 vertical. Figure 4 presents a schematic
of the experimental setup and a Mach contour plot of the computation domain. Results in figure 5 compare the
Mach number contours provided in AIAA-95-0032 to those generated using WIND-US and to the computational
results provided in AIAA-95-0033, Flow Characteristics in Boundary Layer Bleed Slots With Plenum, by Hamed,
Yeuan and Jun. The results indicate that WIND-US and the computational data from AIAA-95-0033 are very simlar
to one another. Both do a fair job in approximating the experimental flow of AIAA-95-0032. The results indicate
that WIND-US is a suitable tool to simulate a slot bleed flow.
A no shock validation of AIAA-95-0032 was also performed. Two Mach numbers were analyzed: 1.27 and
1.58. For the Mach 1.27 flow, the boundary layer at an inflow plane 45.60 cm ahead of the bleed slot was 2.25 cm.
For the Mach 1.58 runs it was 2.10 cm. The computational grid and Mach contours are presented in figure 6. The
grid contained 100 horizontal grid nodes and 450 vertical. Figure 7 compares the experimental results to those
approximated using WIND-US. The horizontal axis of the plot in figure 7 is ratio of the bleed plenum pressure
normalized by the freestream total pressure. The vertical axis presents the mass flow ratio of the bleed, Q (defined
in equation 7), or the ratio of the mass flow rate through the slot bleed section normalized by the maximum allowed,
or ideal choked mass flow rate. WIND-US shows good agreement with the two AIAA-95-0032 experiments,
validating WIND-US’ ability to accurately simulate a slot bleed flow.
Given the results of both bleed validations, it was determined that WIND-US was a suitable tool to use to
simulate slot bleed on an axisymmetric, low boom inlet.

IV. Inlet Geometries Investigated and Associated Grids


A. Compression Surface:
The success of a low boom inlet greatly depends upon the design of the compression surface. Typically, in
an external compression inlet the flow is compressed in such a way that at the throat there is a constant Mach
number upstream of the terminal shock. The aim of this type of design is to maximize the inlet’s recovery
characteristics. However, a drawback to this design is that there is significant turning of the flow at the cowl lip
because of the dense concentration of characteristic lines. The larger turning of the local flow requires a cowl lip at
more of an inward angle to the incoming flow. An angled cowl causes stronger external shocks and poor boom
performance.
If boom is the main design constraint, a redesign of the compression surface is required. The proposed
inlet compression surface designed by T. Conners[2][3][4][5][6][9] and others at Gulfstream calls for relaxed
3
American Institute of Aeronautics and Astronautics
compression, or a de-focusing of the characteristics, particularly in the area of the cowl. By de-focusing the
characteristics, there is significantly less turning of the flow in the cowl area, allowing for the use of cowl which is
virtually parallel to the incoming flow. Figure 8 presents the conceptual design. With relaxed compression the
normal shock at the entrance to the inlet becomes stronger as one moves outward from the compression cone to the
cowl. Figure 9 shows a plot of how the Mach number changes as one moves from the centerbody surface to the
cowl lip. Additionally, while this design is much better from a boom standpoint, there is a bit of a sacrifice of
recovery and flow control is needed to help ensure shock stability and improve recovery as discussed under Inlet
Characteristics.
B. Diffuser Surfaces:
Three different inlets utilizing relaxed compression are discussed in the paper. The first inlet, C0, had a
slope discontinuity where the compression surface transitioned to baseline diffuser. Figure 10 (left contour plot)
presents Mach contours of the C0 inlet with the shock in the optimum recovery position. A major disadvantage of
the C0 inlet was that if the normal shock fell before or after the optimum point seen in figure 10, it would trigger
significant flow separation downstream. To fix this problem multiple designs which smoothed out the slope
discontinuity were analyzed. In addition to minimizing flow separation at off design operating points, two important
objectives were kept in mind through the redesign process; increasing the peak pressure recovery and increasing the
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

mass flow ratio where the peak recovery occurs. Increasing the mass flow ratio at the peak recovery point helps to
minimize the boom signature at on-design operation. The most promising redesigns by D. Howe [10] at Gulfstream
were a series of inlets designated A1 through A4. Figure 11 zooms in upon the area of the slope discontinuity and
presents some of the geometry differences for a small selection of the curves analyzed, including the curves
designated A1 through A4. Figure 12 presents the inlet cane curves for each of these designs in addition to many
others. A computational study proved the optimum design to be the A4.
Also important was the design of the diffuser. Two of the parameters strongly affecting diffuser
performance are diffusion rate, and local wall curvature. Both of these parameters influence the adverse pressure
gradient on the diffuser surfaces and thereby boundary-layer development. Two diffusers were analyzed. The
baseline diffuser, designed as a simple polynomial profile, was used on both C0 and A4 and aimed to gradually
decelerate the flow down to core flow Mach number of 0.65 at the engine fan face. The second diffuser, exclusively
used on the K1 inlet, employed an area distribution that produces a near-linear Mach number distribution. Figure 13
summarizes the differences between the two.
C. Bleed Geometries:
The bleed study was implemented upon the K1 inlet. A 0.5 inch slot bleed section was input just
downstream of the geometric throat on the K1 solid wall inlet. Two different leading edge geometries were
analyzed in the study. The first had the shape of a NACA0012 airfoil nominally at zero angle of attack and was
designated NLE. The second leading edge analyzed was a 4.93:1 ellipse at 7.5˚ angle of attack relative to the inlet
centerline, designated ELE. The 4.93 aspect ratio was selected to approximate the aspect ratio of the NACA0012
shape. Figure 14 depicts the two different leading edge geometries examined. The bleed plenum extended all the
way down to the grid centerline and was run with different pressures. Figure 3 illustrates the right wall inside the
plenum where the constant static pressure outflow boundary conditions were implemented. While the study is still
being completed, analyzing plenum pressures (PP*/Pt0) of 28% and 32% of the freestream stagnations pressure is
proposed.

V. Inlet Characteristics
Inlet performance for mass flow ratios is presented in plots called cane curves. The plots have ratios of
mass flow on the horizontal and total pressure recovery on the vertical axes. Mass flow ratio is defined as the mass
flow entering into the engine normalized with respect to the mass flow in the freestream passing through an area
equal to the capture are of the inlet. Equation 1 defines the mass flow ratio.

m engine face
MFR = [1]
m inlet capture

Given that there is always some degree of normal shock spillage and there can also be bleed flow mass removal, the
ratio is less than one.

4
American Institute of Aeronautics and Astronautics
The inlet recovery is an area integrated measure of the total pressure at the engine fan face normalized by
the freestream total pressure. Equation 2 defines the pressure recovery.

Po engine face − area integrated


REC = [2]
Po freestream

Upstream conditions are held constant, while the static pressure at the engine face (AIP) is varied to generate the
different points defining the cane curve. We can define the ‘peak’ recovery position as the operating condition in
which REC is at a maximum.
The equation for the Reynolds number for the inlet is given in equation 3.

ρ ud
Re = [3]
μ
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Here the density (ρ), velocity (u), and viscosity (μ) are freestream inflow values. The diameter (d) was 12 inches.
A qualitative measure of boom signature can be seen when observing the degree of normal shock spillage.
Spill can be considered excess flow that is not pulled into the inlet and instead spills out into the freestream and can
contribute to a louder sonic boom. Examining figures 1 or 2, it is the green colored contour that flows around the
cowl. An additional measure of inlet performance attempts to quantify sonic boom signature. Given that a sonic
boom is essentially a high pressure spike above a freestream value followed by a sharp dip to a lower pressure, the
boom was determined by taking the difference between the maximum and minimum static pressure spikes and
normalizing them by the freestream value. Equation 4 illustrates the equation used to generate the boom plots.

Pstatic Max − Pstatic Min


Boom = [4]
Pstatic freestream

Figures 1 and 2 depict where the boom was determined. For each cane curve point a corresponding boom
measurement was also taken.
As seen in figure 3, the bleed plenum region has a constant static pressure outflow boundary on the right
wall. The convention for classifying the bleed plenum pressure is given in equation 5,

Pstatic bleed plenum


PP * / PT 0 = [5]
P0 freestream

Additionally, the mass flow, MFR (Bleed), is defined as the flow being pulled out through bleed region normalized
by the inlet upstream capture mass flow

m bleed plenum
MFR ( Bleed ) = [6]
m inlet capture

Finally, similar to equation 6, one can define the choked mass flow rate as

m bleed plenum
Q= [7]
m bleed choked

 bleed
Where m choked is the ideal mass flow rate through the choked bleed slot.

VI. Results and Discussion


For each of the three different solid wall inlets, cane curves were developed to determine each inlet’s
performance at different operating conditions. Figure 15 presents a cane curve plots for the solid wall C0, A4, and
5
American Institute of Aeronautics and Astronautics
K1 geometries. The A4 inlet has the highest peak recovery. The K1 has a recovery peak at a slightly lower value,
however both peaks are at roughly the same MFR (around 0.975), indicating that both have similar performance in
terms of boom. An advantage of the K1 inlet over that of the A4 is that as the MFR increases to right of peak, the
drop-off is much more gradual than the A4. A typical supersonic inlet will be operated over a range of mass flow
ratios, therefore the more gradual drop-off may be significant for flow stability. The final curve in figure 15 is for
the C0 configuration. The disadvantage of the C0 configuration is generally lower recovery. From a boom
standpoint, the fact that the peak recovery occurs at a lower MFR means more normal shock spillage for equal (or
less) recovery.
The effect of changing the Reynolds number was evaluated on the C0 inlet. The inlet was analyzed at three
different Reynolds numbers, 5.4 x 106, 10.8 x 106, and 5.4 x 107. Qualitatively, the contour plots for the three
different Reynolds numbers were very similar, indicating very small quantitative differences. The three cane curves
are provided in figure 16 and all fall essentially right on top of one another, signifying that the inlet is not
particularly sensitive to Reynolds number effects. A point to note is that the three cane curves fall in order of
highest Reynolds number on top and lowest on the bottom. Given that higher Reynolds numbers have the tendency
to thin boundary layers, one expects that when the inlet is run at higher Reynolds numbers, there will be a slightly
higher pressure recovery due to lower viscous losses.
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Examining the boom curves we see two details of interest. The first is that all the different points fall on
the same line. The significance of this is that boom characteristics are mainly a function of mass flow ratio into the
inlet. Notice that as the mass flow ratio increases, boom decreases: less normal shock spillage equals quieter
performance due to a weaker cowl shock. Secondly, the three “peak” recovery points all fall next to one another.
Sonic boom performance is essentially independent of Reynolds number.
The C0 inlet was analyzed at freestream Mach numbers of 1.67, 1.70, and 1.73. While significant
differences between the contour plots cannot be easily seen, there are large differences that arise upon examination
of the cane curve plot (figure 17). In figure 17 we can see that the pressure recovery properties of the inlet get better
with decreasing Mach number. Analysis, depicted in figure 18, showed that the differences in pressure recovery we
see in the cane curves compare very well with the differences one would expect with increasing normal shock
strength. Normal shock relation show that changes in Mach number of on the order of 0.03 should yield total
pressure recovery differences around 1.0%, which is what was seen computationally. Analysis of the boom
signature shows similar results to that of the Reynolds number study.
Another study performed upon the C0 inlet was extending the cowl forward (to the left). The cowl was
extended by two increments: 1.25 mm and 2.50 mm. Figure 19 presents Mach contours of the baseline, 1.25 mm,
and 2.50 mm cowl extension cases for the peak pressure recovery condition. For the three cases, as one moves the
cowl farther forward, the amount of spill around the cowl decreases. Additionally, the normal shock becomes more
and more angled as the cowl is pushed forward. Quantitatively (figure 20), we can see these two phenomena. As
the cowl is pushed forward, the cane curve shifts to the right, meaning the inlet is operating at a position of higher
mass flow ratio. Examining the 2.50 mm cane curve, we see that as one progresses to higher and higher mass flow
ratios, the curve eventually drops off. With a low enough engine face pressure the cane curves of all three cowl
extensions will eventually drop off, however as one moves the cowl forward this drop off point moves closer and
closer to the peak pressure recovery position. An inlet run with a cowl extended too far is much more susceptible to
becoming unstable because of engine flow fluctuations.
Examining the boom curves, we see an interesting phenomenon. As one moves the cowl forward, the
boom signature in the peak recovery position decreases. This is expected given that there is less spill with cowl
extension. Another point to note is that all the points essentially fall on the same line, once again illustrating that the
noise signature is a function of mass flow ratio. Extending the cowl forward allows the inlet to operate at a higher
mass flow ratio at close to the same recovery. Operation at a higher mass flow ratio has the benefits of reducing the
sonic boom, spillage drag, and inlet size (i.e. structural weight) for a given fan operating point.
The A4 inlet design was used to computationally analyze the effect of the external cowl’s shaping upon the
sonic boom signature. Two A4 inlets with different external cowl shapes were compared. The first inlet utilized the
‘stove-pipe’ zero angle cowl while the second incorporated a 10˚ degree external angle. Figure 21 presents Mach
number contours for the two inlets. There is a significantly stronger shock wave when using the angled cowl. In
figure 22 results of each inlet’s performance at the AIP are plotted along with the boom. The first point to notice is
that changing the external cowl shaping had virtually no effect upon AIP performance, as would be expected with
supersonic flow. Additionally, for all points of inlet operation, the angled cowl has a stronger boom than the zero-
angle cowl. Finally, the angled cowl boom pressure results are virtually constant for the entire range of inlet
operation, indicating that the oblique shock generated by the cowl bevel dominates the shock resulting from normal
shock spillage. In contrast, the zero-angled cowl boom signature reduces as the inlet mass flow increases. This
6
American Institute of Aeronautics and Astronautics
points to a key advantage of using the zero-angle cowl: the sonic boom projected to the ground can be controlled by
maximizing the inlet MFR thereby minimizing the inlet’s normal shock spillage.
A bleed study on the K1 geometry was also completed. Figure 23 shows the Mach contours for the K1-
ELE inlet operated at a plenum pressure of 25.1% of the freestream stagnation for the points indicated on the
associated cane curve and bleed plenum plot. For this particular bleed plenum pressure, differences between the
bleed and solid wall geometry cane curves are small. An interesting flow feature that develops when bleed is input
into this type of low cowl angle inlet is a secondary shock which sits just in front of the plenum leading edge. Even
with the terminal normal shock, the significant turning of the centerbody causes the flow to become transonic before
shocking down once again at the leading edge of the plenum. Note that a significant range of well behaved, stable
inlet operation is available (between just right of point C and slightly to the left of point B) before the terminal
normal shock jumps forward (point A) into a potentially unstable position accompanied by a sudden decrease in
total pressure recovery and mass flow ratio. From a boom mitigation perspective, the optimal operating MFR for
the K1-ELE inlet presented in figure 23 is near 0.988 (point C). The analysis indicates that a bleed system equipped
with a variable exit controlled to maintain a constant plenum pressure is an effective way to provide stable inlet
operation over a range of AIP mass flow ratios. The operating point can be set to a low bleed rate to minimize bleed
drag and the system will compensate for transient AIP mass flow reductions by increasing the bleed mass flow rate.
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

The value of 25.1% was found to be the optimum plenum pressure for this inlet. Other plenum pressure values were
examined, two of which are presented in figure 24. From the figure we see that dropping the plenum pressure from
a value of 32% to 25.1% shifts the overall inlet cane curve to high MFR values, signifying better boom performance.
Also, the lower plenum pressure allows for a higher MFR-Bleed up until the point where the 25.1% curve plateaus
in the area of peak recovery. The value of 25.1% as the optimum plenum pressure was determined by holding the
AIP outflow pressure constant while varying the plenum pressure. By examining the solutions in terms of both
optimum recovery, low MFR, and adequate degrees of MFR-Bleed the optimum pressure was determined.
Figure 25 compares the cane curves of the NLE and ELE geometries for a PP*/Pt0 of 32%. In terms of
integrated AIP recovery, the curves show that there is no discernable difference between using either of the leading
edge geometries. However, the bleed plenum plot shows that the bleed plenum with the elliptic lip pitched up 7.5
degrees relative to the inlet centerline (ELE) has a wider range of stable operation than the airfoil shaped lip at zero
pitch. This indicates that the pitched up lip provides a more efficient means of removing bleed to compensate for the
fan flow reduction. Figure 26 show both stagnation pressure profiles and axisymmetric shape factors at the AIP for
the points along the points indicated in the corresponding cane curves. In terms of AIP performance, none of the
variables examined had a significant difference on the AIP condition. This is a consequence of the long diffuser, as
all differences in the flow field have dissipated by the time they reach the AIP. A shorter diffuser would minimize
dissipation, however factors inherent to implementation into a full scale airplane have constrained the diffuser
length. Given that the main objective of the bleed system is to ensure normal shock stability, the lack of significant
improvements in AIP performance was moderately expected. Finally, figure 27 compares a boom plot for both the
K1 solid wall and elliptical leading edge bleed geometry. The plot shows similar trends to those seen in other boom
plots, however, an important point to note is that the bleed system allows for more stable operation. Given this, it is
expected that an inlet with a bleed system will have stable operation at higher AIP mass flows, and lower boom
levels.

VII. Conclusions
The WIND-US code was used to analyze the performance of three subsonic diffuser designs on an
axisymmetric relaxed compression inlet. Of the three, the A4 shows the most promise in terms of pressure recovery
and boom characteristics. To verify WIND-US’ ability to simulate a slot bled flow, two validations of existing
experiments from AIAA-95-0032 were performed. The first was a validation of a slot bleed section with an oblique
shock. The second validation performed was upon slot bled flow without any shocks. WIND-US was able to
approximate the experimental results to a high degree of accuracy. The design process yielded three different inlet
geometries: C0, A4, and K1. The A4 inlet has the highest peak recovery, followed by the K1, and C0. Both the A4
and K1 geometries achieved their peak recovery at similar mass flow ratios of 0.975 while the C0 peaked at a mass
flow ratio of 0.965. An advantage of the K1 is that it offers more in terms of stability than the other two inlets. The
C0 inlet’s performance was analyzed for differences in Reynolds number, Mach number, and cowl extension.
Reynolds number changes did not cause significant differences in inlet performance. Varying the Mach number
changed the recovery characteristics but differences were in accordance with what was expected from 1-D shock
relations. There were no significant differences in boom due to changing Mach number. Extending the cowl had
significant effects upon both MFR and boom. Cowl extension shifted the inlet’s peak recovery to a higher MFR.
7
American Institute of Aeronautics and Astronautics
This in turn significantly reduced the boom characteristics. The A4 was used to analyze the differences between an
angled and zero-angle cowl. It was found that a zero angle cowl has significantly less boom than an angled one.
Additionally, a zero-angle cowl’s boom is governed by the inlet MFR. An angled cowl has approximately constant
boom for all MFR. Finally a full bleed study on the K1 inlet for two different plenum pressures (PP*/Pt0 = 25.1%
and 32%) along with two different leading edge geometries (ELE and NLE) was performed. It was found that for
the cane curve for the optimized plenum pressure of PP*/Pt0 = 25.1%, there was not a significant difference to K1
solid wall geometry cane curve. For the operating point of 0.988 MFR the plenum does not bleed significantly.
With lower MFR, the plenum bleeds more to pull the shock back towards the design point. The elliptical leading
edge was determined to be the optimum. Finally, little differences can be seen between stagnation pressure profiles
or axisymmetric shape factor, or boom between any of the geometries analyzed, however the stability gained from
adding a bleed system to the solid wall geometry allows for stable operation at higher MFR and lower boom.
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

8
American Institute of Aeronautics and Astronautics
Figures
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Figure 1: Dimensioned solid wall geometry

Figure 2: Dimensioned bleed geometry

9
American Institute of Aeronautics and Astronautics
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Figure 3: Bleed geometry grid

Figure 4: AIAA-95-0032 experimental setup and WIND-US generated Mach contour plot the experimental
flowfield

10
American Institute of Aeronautics and Astronautics
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Figure 5: Mach number comparison of experimental results from the Mach 2.46 oblique shock slot bled flow in
AIAA-95-0032 [7] to those simulated by WIND-US and the computation results from AIAA-95-0033 [8]

Figure 6: Computational grid, Mach contours, and general information for AIAA-95-0032 (no shock) bleed
validation

11
American Institute of Aeronautics and Astronautics
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Figure 7: Validation of AIAA-0032 no shock with WIND-US [7]

Figure 8: Method of characteristics design for inlet compression surface [2][3][4][5][6][9]

12
American Institute of Aeronautics and Astronautics
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Figure 9: Mach number profile just before terminal normal shock

Figure 10: Comparison of C0 and A4 flow fields – zoomed in on shock region

13
American Institute of Aeronautics and Astronautics
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Figure 11: Geometry differences for some of the different inlet designs analyzed [10]

Figure 12: Cane curve plots for all inlets analyzed in optimization [10]

14
American Institute of Aeronautics and Astronautics
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Figure 13: Comparison of C0, A4, and K1 inlet geometries

Figure 14: Bleed leading edge geometries

Figure 15: C0, A4, and K1 solid wall cane curves

15
American Institute of Aeronautics and Astronautics
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Figure 16: Cane curves of inlet for different Reynolds numbers (C0 geometry)

Figure 17: Cane curves of inlet for different Mach numbers (C0 geometry)

16
American Institute of Aeronautics and Astronautics
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Figure 18: Increasing Mach number shock strength calculation (C0 geometry)

Figure 19: Mach contours for different degrees of cowl extension (C0 geometry)

17
American Institute of Aeronautics and Astronautics
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Figure 20: Cane curves of inlet for different degrees of cowl extension (C0 geometry)

Figure 21: Mach contours of 2 different cowl angles for peak recovery position (A4 geometry)

18
American Institute of Aeronautics and Astronautics
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Figure 22: Inlet performance for different angled cowls (A4 geometry)

Figure 23: K1 25.1% Pp*/Pt0 Mach contour plots for different cane curve and bleed plenum plots

19
American Institute of Aeronautics and Astronautics
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Figure 24: K1 ELE comparison for different values of Pp*/Pt0 (32% and 25.1%)

Figure 25: K1 32%Pp*/Pt0 inlet cane curve and bleed plenum plot for ELE and NLE leading edge geometry

20
American Institute of Aeronautics and Astronautics
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

Figure 26: K1 geometry stagnation pressure profile and axisymmetric shape factor comparison at the AIP for points
through dotted dashed line

Figure 26: K1 geometry boom comparison for solid wall and 25% Pp*/Pt0 slot bleed inlets

21
American Institute of Aeronautics and Astronautics
Acknowledgments

The support of National Center for Supercomputing Applications (NCSA), NASA Glenn Research Center,
NASA Fundamental Aeronautics Supersonics Project, Gulfstream Aerospace Corporation, and Rolls-Royce is
gratefully acknowledged.

References

[1] Bush, R.H., G.D. Power, and C.E. Towne. "WIND: The Production Flow Solver of the NPARC Alliance."
AIAA 98-0935. January 12, 1998.

[2] Conners, T.R. (personal communication) "Baseline Aerodynamic Surface Definition (Presentation)."
Savannah, GA. April 16, 2008
Downloaded by Cranfield University on January 17, 2024 | http://arc.aiaa.org | DOI: 10.2514/6.2009-4210

[3] Conners, T.R. and Howe, D.C., “Supersonic Inlet Shaping for Dramatic Reductions in Drag and Sonic
Boom Strength,” AIAA-2006-0030, 2006.

[4] Conners, T. R., Merret, J. M., Howe, D. C., Tacina, K. M., and Hirt, S. M., “Wind Tunnel Testing of an
Axisymmetric Isentropic Relaxed External Compression Inlet at Mach 1.97 Design Speed,” AIAA-2007-
5066, 2007.

[5] Conners, T.R., Howe, D.C., and Henne, P.A., Gulfstream Aerospace Corporation, Savannah, GA, U.S.
Patent Application for “Isentropic Compression Inlet for Supersonic Aircraft,” Application No. 11/639,339;
Filed 15 Dec. 2006.

[6] Conners, T.R., Gulfstream Aerospace Corporation, Savannah, GA, U.S. Patent Application for “Low Shock
Strength Inlet,” Application No. 12/000,066; Filed 7 Dec. 2007.

[7] Davis, D. O., B. P. Willis, and W. R. Hingst. "Flowfield Measurements in a Slot-Bled Oblique Shock-
Wave and Turbulent Boundary-Layer Interaction." AIAA-95-0032. Reno, NV, January 9–12, 1995.

[8] Hamed, A., J.J. Yeuan, and Y.D. Jun. "Flow Characteristics in Boundary Layer Bleed Slots With Plenum."
AIAA-95-0033. Reno, NV, January 9–12, 1995.

[9] Hirt, Stefanie M., Kathleen M. Tacina, Timothy R. Conners, Jason M. Merret, and Donald C. Howe. "CFD
Results for an Axisymmetric Isentropic Relaxed." AIAA 2008-92. Reno, NV, January 7–10, 2008.

[10] Howe, Donald.(personal communication) "Full Scale Zero-Angle Inlet Centerbody Shape Variations,
Unabridged (Presentation)." Atlanta, GA, October 21, 2008.

[11] Rybalko, Michael, Eric Loth, Rodrick V. Chima, Stephanie M. Hirt, and James R. DeBonis. "Micro Ramps
for External Compression Inlets." 39th AIAA Fluid Dynamics Conference. San Antonio, TX, 2009.

22
American Institute of Aeronautics and Astronautics

You might also like