A Study of The Internal Aerodynamics of The Concorde Inlet: John W. Slater
A Study of The Internal Aerodynamics of The Concorde Inlet: John W. Slater
2020-3770
August 24-28, 2020, VIRTUAL EVENT
AIAA Propulsion and Energy 2020 Forum
This paper summarizes a study of the aerodynamic design and analysis of the inlet for the
Concorde aircraft. The supersonic inlet design and analysis (SUPIN) tool was used to model
the Concorde inlet based on publicly-available information for the geometric and
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
aerodynamic properties of the inlet. Computational fluid dynamics (CFD) methods were then
applied to compute the flowfield about and through the inlet at the design cruise condition.
Inlet performance metrics of total pressure recovery and distortion were obtaine d from the
CFD simulations and compared to available wind-tunnel data for the Concorde inlet. The
design and analysis methods were able to approximate the Concorde inlet and achieve similar
aerodynamic performance. This study provides useful information for the development of
inlets for future commercial supersonic aircraft.
Nomenclature
A = Area
D = Diameter
h = Height
L = Length
M = Mach number
p = Pressure
w = Width
W = Flow rate
x, y, z = Cartesian coordinates
( )0 = Freestream property
( )L = Local inflow conditions
( )1 = Property at the cowl lip / inlet entrance station
( )2 = Property at the engine-face station
( )cap = Property associated with the reference capture area
I. Introduction
The Aérospatiale (Sud Aviation) / Bristol Aircraft Corporation (BAC) Concorde aircraft was a commercial
supersonic airliner developed during the 1960’s with a first flight in 1969 [1]. A total of 20 aircraft were produced,
of which 14 were in commercial service with British Airways and Air France between 1976 and 2003. The Concorde
could accommodate 92 to 128 passengers at a cruise speed of Mach 2.0. Its primary route was between Europe and
the east coast of North America. Its end-of-service was prompted by limited economic success and stricter noise and
emission standards for commercial aircraft. Yet the Concorde was a remarkable technical achievement and stands out
as an example of a fully-operational commercial supersonic airliner. Further, the development and operation of the
Concorde provides a wealth of information for the development of future commercial supersonic aircraft.
In this study we are interested in the propulsion system, and specifically, in the aerodynamic flowfield and
performance of the inlet. The objectives of the study were to recreate the geometric design of the inlet and perform
computational aerodynamic analysis to understand the flowfield of the Concorde inlet and to verify that current design
and analysis methods can model the flowfield and aerodynamic performance of the Concorde inlet.
1 Research Aerospace Engineer, Propulsion Division / Inlets and Nozzles Branch, AIAA Associate Fellow.
This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
Section II provides a description of the Concorde inlet and includes a summary of the Concorde aircraft and how
the propulsion system with the inlet was integrated with the aircraft. This provides an understanding of the flow
conditions ahead of the inlet. The inlet provided airflow to the Olympus turbojet engine, and so a description of the
engine is provided to understand the geometry and flow conditions at the engine face of the inlet. The geometric
features and flow conditions of the inlet are then discussed to understand the inlet design. Information on the geometry
and flow conditions for the Concorde aircraft were obtained from various publicly-available references and
discussions with other researchers. However, in some cases, approximate geometric and flow were used.
Section III discusses the formulation of the computational model of the inlet for aerodynamic analysis. The
Supersonic Inlet Design and Analysis (SUPIN) Tool [2] was used to perform the modeling and generate the surfaces
of the inlet from the leading edge to the engine face. SUPIN also provided estimates of the aerodynamic performance
of the inlet at the design condition. SUPIN also generated three-dimensional, multi-block, structured grids of the flow
domain about and through the inlet for the computational fluid dynamic (CFD) analyses of the flowfield.
Section IV discusses the CFD methods for the analysis of the inlet flowfield as performed using the Wind-US flow
solver [3]. Included is a discussion of Wind-US, the flow domain, boundary conditions, grid generation, initial
flowfield, solution convergence, and inlet aerodynamic performance metrics.
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
Section V discusses the results of the computational analyses. Included are images of the flowfield and the
aerodynamic performance metrics, which include the flow rates, total pressure recovery, and total pressure distortion.
2
inlet from the angle-of-attack variations of the aircraft, which is known as wing shielding. Thirdly, the horizontal
orientation of the inlet ramps resulted in a downward deflection of the external inlet flow which contributed to the lift
of the aircraft.
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
Figure 1. Concorde aircraft (left, public domain image, Wikipedia) and propulsion system
beneath portside wing (right, permission to use from www.ConcordeSST.com).
In placing the inlet under the wing, the inlet had to be placed far enough downstream of the leading edge to avoid
interference of the inlet flow with the formation of the shocks and vortices from the leading edge of the wing. The
flow along the under-surface of the wing had a slight outward direction, and so the inlets were canted slightly inward
so that their inlet axes were in line with the local flow. The inner inlet was canted inward of about 2 degrees while
the outer inlet was canted inward 4 degrees. The canted orientation of the inlets may be noticeable in Fig. 1.
A viscous boundary layer also formed along the surfaces of the wing. The inlet was thus offset away from the
undersurface of the wing to avoid having the inlet ingest the entire boundary layer, which would decrease the total
pressure available for the inlet and engine. The offset or diverter height was at most 70% of the bounda ry-layer
thickness [4]. Thus, the outer 30% of the boundary layer was ingested; however, the throat bleed subsequently
removed this flow from entering the subsonic diffuser. The start and initial angle of the diverter structure were
designed to avoid disturbing the flow entering the inlet [4]. The diverted flow was likely a mix of subsonic and
supersonic flow, and so, easily prone to choking and the feeding forward of disturbances within the subsonic portion
of the boundary layer.
B. Olympus 593 Engines and Nozzles
The propulsion system included four Rolls Royce / Snecma Olympus 593 dual-spool turbojet engines. Each
engine was provided airflow by an inlet positioned ahead and in line with the engine. Afterburners provided up to
18% additional thrust at take-off and transonic acceleration [1]. Above Mach 1.7, the afterburners could be turned off
and thrust maintained for supersonic cruise. Each engine had an integral nozzle that extended to the trailing edge of
the wing. The nozzle used variable-geometry to adapt to the flight conditions. At take-off and subsonic flight, the
nozzle could be made to be converging with partial ejection of secondary flow. For supersonic flight, the nozzle could
be made converging-diverging for acceleration of the nozzle flow to higher supersonic speeds. Upon landing, the
nozzle could be configured for thrust reversal. Figure 2 shows some details of the engine core, afterburner, and nozzle.
Figure 3 shows the variable geometry of the nozzle.
The modeling of the inlet required the dimensions of the engine face that formed the internal outflow boundary of
the inlet as it delivered airflow to the engine. Here we model the engine face of the turbojet engine as a planar annular
cross-section with the outer diameter equal to the radial extent of the initial guide vanes (D2) and the inner diameter
equal to the diameter of the spinner covering the hub of the engine (Dspinner). The picture at the top of Fig. 2 shows
the dimensions of the engine face needed for the inlet modeling. At the start of the study, a value of D2 = 1.206
meters (3.957 feet) was used based on Ref. [1]. This value was reported as the maximum diameter of the Olympus
engine and images suggested that this dimension occurred at the front of the engine. The diameter of the hub or
spinner (Dspinner) was estimated by measurements from images of the engine and the above value for the engine-face
3
diameter (D2). It was estimated that the ratio of the spinner diameter to the engine face diameter (hub-to-tip ratio) was
Dspinner / D2 = 0.238, which resulted in Dspinner = 0.942 feet. The resulting flow area of the annular engine face was A2
= 11.599 ft2. The profile of the spinner was approximated a circular arc. These values were used for the Inlet A
simulations reported in this paper.
Later in the study, discussions were made with researchers at Boom Supersonic who presented a study of the
Concorde inlet in an invited talk at the 2020 AIAA SciTech Forum. They indicated they used an engine face diameter
of D2 = 4.0 feet. In another attempt at refining the dimensions of the engine face, the author contacted Prof. Gregory
Elliott of the University of Illinois at Urbana-Champaign. The Talbot Laboratory of the university has an Olympus
engine on loan from Rolls-Royce. Scott Dalby of the university was kind enough to measure the diameters of the
engine face and spinner with resulting values of D2 = 3.80 feet and Dspinner = 1.19 feet, which resulted in a hub-to-tip
ratio of Dspinner / D2 = 0.314. The resulting engine face was A2 = 10.235 ft2. These values were used for the Inlet B
simulations reported in this paper.
Along with the dimensions of the engine face, the computational modeling required information on the flowrate
of the engine at the cruise design condition to complete the specification of the downstream boundary condition for
the inlet. For a turbojet engine, this flowrate information takes the form of a corrected flow rate, which can be
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
equivalently expressed as an engine-face Mach number. The flow rate at the engine face for the Olympus 593 was
indicated to be 186 kg/s (410 lbm/s) [1]. This engine flow rate was assumed to be the full-throttle corrected flow rate
(WC2). Since no firm data was available with regards to the installed flow rate or engine lapse of the flow rate with
flight Mach number, the inlet design efforts here assumed a corrected flow rate of WC2 = 410 lbm/s at the cruise design
conditions. For an engine face area of A2 = 11.60 ft2, this corrected flow rate corresponds to an engine face Mach
number of M2 = 0.472. Several references on the Concorde indicated an engine face Mach number of M2 = 0.5, which
was likely a rounding of the actual Mach number. These conditions were used for the initial modeling and simulations
of inlet A. The data from the University of Illinois indicated a smaller engine-face area of A2 = 10.235 ft2. A corrected
flow rate of WC2 = 410 lbm/s resulted in an engine-face Mach number of M2 = 0.565, which is significantly higher
than reported values. A value of M2 = 0.5 along with the area based on the University of Illinois dimensions, a
corrected flow rate of WC2 = 377 lbm/s was obtained, which was used for the modeling and simulations of inlet B.
C. Concorde Inlet
The purpose of the inlets was to capture the amount of airflow required by the engines and condition the airflow
to match the flow conditions and flow quality required at the engine face. The right image of Fig. 1 shows the two
portside inlets. The Concorde inlets were considered two-dimensional since the external supersonic compression
occurred using nominally two-dimensional shock and Mach waves. Further, much of the forward part of the inlet
could be illustrated as profiles on the vertical plane of symmetry.
In Fig. 4, the schematic of the inlet is shown beneath the profile of the wing. The airflow is from left to right, as
indicated by the arrow on the left. The boundary layer diverter is shown as an offset of the inlet from the undersurface
4
of the wing. The leading edge of the inlet is clearly downstream of the leading edge of the wing. The general features
of the inlet are identified in Figs. 3 and 4 for the compression system at the cruise conditions. The forward part of the
inlet performed the external supersonic compression and consisted of three stages. The first stage was a fixed initial
ramp that formed an external oblique shock wave that passed forward of the cowl lip. The second stage was a curved
segment that formed a series of Mach waves that focused just ahead of the cowl lip. The third stage was a ramp
connected to the second stage. Both the second and third stages rotated about a hinge located near the upstream end
of the second stage. The level of external supersonic compression could be adjusted by the rotation of the second and
third stages. Figure 3 also shows this variable-geometry aspect of the inlet.
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
Figure 3. Variable geometry of the inlet and nozzle at take-off and cruise.
Sidewalls were used to contain the compression along the external supersonic diffuser. The outer sidewalls of the
inlet pair had “cut-backs” that positioned the leading edges of the sidewalls above a line from the leading edge to the
cowl lip. These cut-backs reduced internal boundary-layer separation and improved the performance of the inlets at
sideslip conditions. Between two adjacent inlets, a splitter plate was used to reduce spillage from one inlet interfering
with the performance of the adjacent inlet. The splitter extended slightly forward and downward ahead of the inlet.
The right image of Fig. 1 shows the sidewalls and splitter plate. The planform of the splitter plate can be seen in Figs.
3 and 4.
The cowl lip was the forward-most leading edge of the cowl and indicated the start of the internal ducting of the
inlet, as shown in Fig. 4. The cowl lip featured a corner cut-out near the splitter plate that also alleviated shock wave
/ boundary layer interactions at angle-of-sideslip. This cut-out can be seen in the right image of Fig. 1.
The cross-section through the external supersonic diffuser was rectangular in keeping with the two-dimensional
nature of the inlet. The capture cross-section of the inlet was bounded at the top by the leading edge of the inlet, at
the bottom by the cowl lip edge, and on the sides by the sidewall and splitter plate leading edges. The capture cross -
section for the Concorde was indicated to be square [5].
The external supersonic compression was terminated with a oblique shock generated by the cowl lip, as illustrated
in Fig. 4. The Concorde inlet involved mostly external supersonic compression with some internal supersonic
compression. This approach allowed for a stable shock system while taking advantage of some of the greater
efficiencies that resulted in lower cowl wave drag. The behavior of the inlet shock system will be discussed in greater
detail later in the discussions of the modeling of the external supersonic diffuser and the flowfield computations. This
approach of partial internal supersonic compression was made possible due to the use of a wide ramp bleed slot and a
rotating rear ramp. The rear ramp rotated about a hinge located at the downstream end of the ramp. Figures 3 and 4
show this rotating rear ramp. The gap between the front and rear ramps formed the bleed slot. Inlet flow was extracted
through the ramp bleed slot to remove low-momentum boundary layer flow and to control in part the amount of flow
delivered to the engine. The bled flow was mixed with the nozzle flow as part of the inlet control system. Downstream
of the rear ramp, the cross-section of the inlet transitioned from a rectangle at the end of the rear ramp to a circle at
the engine face. This transition section formed the subsonic diffuser. The images of Fig. 5 show the ramps, bleed
slot, and subsonic diffuser.
The inlet was required to operate at speeds from take-off to cruise and then back down to landing. The rotating
front and rear ramps allowed variation of the cross-sectional area through the inlet and allowed the inlet flow to adjust
to the conditions and airspeed of the airflow entering the inlet. Figure 3 illustrates the positions of the front and rear
ramps at the take-off and cruise speeds of the aircraft. At take-off and subsonic conditions, the ramps were collapsed
5
to open the cross-sectional area at the throat to allow the passage of the flow without acceleration to supersonic speeds
or choking of the flow. At cruise conditions, the ramps were rotated downward to provide the external and internal
supersonic and subsonic compression. A dump door on the bottom surface of the inlet, as illustrated in Fig. 4, allowed
excess captured airflow to be dumped overboard when not needed for the engine. The use of the dump door minimized
disturbances of the flow at the cowl lip as the engine flow rate was reduced. The left image of Fig. 5 shows a front
view of the starboard inlets with the dump doors opened. Inset within the dump door was an auxiliary door that could
be used to allow additional flow into the inlet at take-off conditions. At this condition, the dump doors were closed.
The auxiliary door is shown in the upper image of Fig. 3 and is labeled “spill door” instead of “dump door”. In Fig.
5, the auxiliary door can be seen with the marking “DANGER”. The auxiliary door was spring-loaded and deflected
inward by the lower pressures within the inlet caused by the drawing and acceleration of flow into the inlet at the low -
subsonic flight speeds. The additional passage for airflow avoided choking of the flow within the throat of the inlet,
which could result in shock wave / boundary-layer interactions that decrease total pressure within the inlet flow.
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
Figure 4. Concorde inlet geometric features (From Ref. 4, Rettie and Lewis, 1968; reprinted
by permission of the American Institute of Aeronautics and Astronautics, Inc.).
Figure 5. Photographs of the starboard inlets looking straight into the inlets (left,
www.airandspace.si.edu) and a view within the inlet (www.heritageconcorde.com).
6
The remaining subsections of this section discuss the modeling of the separate components of the inlet, which included
the external supersonic diffuser, throat section, and the subsonic diffuser.
A. Supersonic Inlet Design and Analysis Tool (SUPIN)
The Supersonic Inlet Design and Analysis (SUPIN) Tool [2] was used to create a computational model of the
Concorde inlet based on the geometry and flow condition information discussed in Section II. SUPIN is a FORTRAN
95 program that reads in a text-based input data file that contains values for the various input factors for the inlet
model. These factors are geometric items such as surface angles, lengths, and areas, and flow conditions such as
Mach numbers or pressure ratios. SUPIN uses analytic, empirical, and computational methods to model the inlet and
estimate the flow rates, total pressure recovery, and drag for an isolated inlet. SUPIN generates the surfaces of the
inlet and creates a Plot3D file [6] of the surface grid of the inlet. SUPIN can also automatically generate a multi-
block, structured grid for a flow domain about the inlet for flow analysis using CFD methods.
Images of the inlet surfaces created for the Concorde inlet as modeled by SUPIN are shown in Fig. 6. The inlet
was divided into components that include the external supersonic diffuser, throat section, and subsonic diffuser. Also
shown are the bleed slot and plenum. The key stations within the inlet flowpath are shown in Fig. 6. Station L
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
represents the local flow approaching the inlet. The external supersonic diffuser started at the leading edge and ended
at station 1, which was the start of the internal ducting of the inlet and contained the cowl lip. The throat section
started at station 1 and extended to station SD, which was the start of the subsonic diffuser. The subsonic diffuser
started at station SD and extended to station 2, which was the engine face. The engine face was annular and contained
a spinner. The numeric stations were consistent with the SAE standard for propulsion system numbering [7]. The
cowl exterior provided the outer surfaces of the bottom and sides of the inlet. The next three sub-sections discuss the
modeling of the external supersonic diffuser, throat section, and subsonic diffuser, respectively.
7
The modeling of the external supersonic diffuser required specification of the flow conditions ahead of the inlet at
station L. As mentioned in Section II, the cruise conditions were M0 = 2.0 with an altitude of h0 = 60000 feet.
Reference [5] indicates that the flow beneath the wing and approaching the inlet was ML = 1.915, which was used for
the modeling. The thermodynamic properties at station L were obtained using the Standard Atmosphere for an altitude
of h0 = 60000 feet. The resulting total pressure and temperature used for the modeling were ptL = 1027.2 psf and TtL
= 675.0 oR, respectively. A reference capture flow rate (Wcap) could be established using the reference capture area
(Acap) and the flow conditions at station L (ML, ptL, and TtL).
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
8
normal shocks provides an understanding of the trade-offs for the terminal shock. For a MEX = 1.38 two-dimensional
flow with 3.5-degree deflection, the weak oblique shock solution yielded downstream conditions of M1 = 1.253, shock
angle of 51.4 degrees, total pressure ratio of 0.999, and static pressure ratio of 1.188. The strong oblique shock solution
yielded downstream conditions of M1 = 0.760, shock angle of 84.5 degrees, total pressure ratio of 0.964, and static
pressure ratio of 2.035. A normal shock at MEX = 1.38 yielded downstream conditions of M1 = 0.748, shock angle of
90.0 degrees, total pressure ratio of 0.963, and static pressure ratio of 2.055. Thus, the strong oblique shock allows
some internal supersonic compression with similar performance of a normal shock.
The strong oblique cowl lip shock leaned downstream and into the throat section. The wide bleed slot acted to
turn the flow of the front ramp toward the axis of the engine. This accelerated the flow slightly, which resulted in the
cowl lip shock curving downstream as it approached the bleed slot. This helped to alleviate shock-wave / boundary-
layer interactions due to the cowl lip shock. The bleed slot operated with near stagnant flow within the bleed plenum,
which resulted in mostly uniform static pressure, and so, the gap across the bleed slot formed a constant-pressure
boundary. At the critical operating condition of the inlet, the core inlet flow near the bleed slot remained at supersonic
speeds and created a normal, terminal shock located near the end of the bleed slot. Most of the slot bleed flow entered
the plenum downstream of the normal terminal shock in the form of a supersonic jet injected into the plenum. The
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
differences between the oblique cowl shock at the bottom of the inlet duct and the normal terminal shock at the top of
the inlet duct resulted in a gradient of total pressure. However, downstream of the normal terminal shock there was a
turning of the inlet flow due to the bleed flow that set up radial pressure gradients that resulted in a mostly uniform
total pressure profile across the entrance to the subsonic diffuser at station SD. At supercritical inlet operation, the
normal terminal shock increased to span a greater portion of the throat section. At subcritical inlet operation, the
normal terminal shock disappeared, and the cowl lip shock angle increased and detached from the cowl lip. At more
severe subcritical conditions, the cowl lip shock could interact unfavorably with the ramp boundary layer and the
leading-edge oblique shock to result in buzz, which involves rapid upstream and downstream motion of the shock.
The bleed slot also helped control the shock system within the throat by acting to balance some differences between
the flowrate at station 1 and the engine flowrate. If the engine flowrate decreased slightly, the static pressures in the
subsonic diffuser would increase and move the normal terminal shock forward. This increased the bleed flowrate by
an amount similar to the reduction in the engine flowrate. This reduced the movement of the cowl lip shock. The
flow bled through the bleed slot was ducted downstream along the engine to be mixed with the nozzle flow.
The cross-section for the throat section was modeled within SUPIN as being rectangular and constant-width equal
to wcap through the length of the throat section. Thus, the centerbody (lower) and cowl interior (upper) surfaces of the
throat section were defined by planar profiles as shown by the schematic of Fig. 8. The throat section model consisted
of several design factors that defined the planar profiles, as shown in Fig. 8 and listed in Table 1. The profile for the
centerbody of the throat section started at point (x,y)cb1, which was also the end of the external supersonic diffuser.
The slope at this point was θcb1 and matched the slope at the end of the centerbody. Point (x,y)cb1 was also located on
a line that intersected the cowl lip and was perpendicular to the end of the external supersonic diffuser. Point (x,y)cb1
and the cowl lip point (x,y)clip also defined the cowl lip plane.
The profile of the centerbody through the throat section consisted of three planar segments. The first segment
was a line of length Lcb1sh that started at point (x,y)cb1 and had a slope of θcb1. The second segment was the shoulder
and was formed using a non-uniform rational B-spline (NURBS) that started at point (x,y)cbsha with the slope of θcb1
and ended at point (x,y)cbshb with the slope of θcbSD. The third segment was a line of length LshSD that started at point
(x,y)cbshb and ended at point (x,y)cbSD, which was the end of the centerbody in the throat section and the start of the
subsonic diffuser.
The cowl interior started at the cowl lip interior point (x,y)clin, which was the downstream end of the interior elliptic
profile of the cowl lip. The planar profile of the cowl interior for the throat section was defined by establishing points
(x,y)cwTH with slope θcwTH and (x,y)cwSD with slope θcwSD and then fitting NURBS curves between the points of the cowl
interior. The point (x,y)cwTH was located along a vertical line that also passed through point (x,y)cbTH which was the
point on the centerbody that had zero slope. The height between points (x,y)cbTH and (x,y)cwTH was established by a
specified area ratio ATH/A1. Point (x,y)cwSD was established along a line passing through and normal to point (x,y)cbSD.
The height between points (x,y)cbSD and (x,y)cwSD was established by a specified area ratio ASD/A1.
The values of the design factors listed in Table 1 were established through estimations obtained from various
planar images of the Concorde inlet like that of Fig. 4. The process involved viewing the image within Microsoft
Paint® and locating points along the geometry in terms of pixel location. Using known or best estimates of a
dimension, such as the engine face diameter, a conversion was be made between pixel distances and dimensions.
Coordinates (x,y) were established by selecting the nose of the external supersonic diffuser as y = 0 and the cowl lip
point as x = 0. From the coordinates, angles between points were calculated using the point-slope formula for a line.
9
D. Subsonic Diffuser
The subsonic diffuser served to transition the shape of the cross-section of the internal ducting of the inlet from a
rectangle at the start of the subsonic diffuser to a circle at the engine face. The increase in cross-sectional area diffused
the subsonic flow from an inflow value of MSD to the desired value of M2. The axial length of the subsonic diffuser
was obtained from images and schematics of the inlet and estimated to be Lsubd/D2 = 2. This established the axial
coordinate of the engine face (x2). A two-dimensional inlet allows a vertical offset of (– ySD)/Lsubd of the engine face.
Using the images of the inlet, the vertical coordinate of the engine face was estimated to be y2 = 1.755 feet.
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
Figure 8. Geometry model for the throat section of the two-dimensional inlets.
10
The spillage flow rate was estimated from the above description of the external supersonic diffuser dimensions to
be Wspillage/Wcap = 0.021. The bleed rate used for the inlet sizing was specified as Wbleed/Wcap = 0.06 [1]. The engine-
face flow rate (W2) is related to the corrected flow rate (WC2) as
𝑇
√ 𝑡2⁄𝑇𝑡 (2)
𝑟𝑒𝑓
𝑊𝐶2 = 𝑊2 𝑝
𝑡2
⁄𝑝𝑡𝑟𝑒𝑓
The reference values were Ttref = 518.69 oR and ptref = 2116.2 psf. The flow was assumed to be adiabatic through the
inlet, so Tt2 = Tt0. Thus, for a value of the total pressure at the engine face (pt2), Eq. 2 was solved for W2 and Eq. 1
was solved for Wcap. For a value of Wcap, the capture cross-sectional area Acap was computed. The sizing operation
was performed by SUPIN as a fixed-point iteration on pt2. An initial guess for pt2 was provided by MIL-STD-5008C
[9] and for the intermediate inlet size, the total pressure losses through the inlet were estimated. The total pressure
losses were computed from analytic shock formulas and models for total pressure loss through subsonic ducts. This
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
process was repeated until convergence of the pt2, and so, Acap.
The summary of the inlet sizing is presented in Table 2. Inlet A was the designation given to the initial inlet
designed with an engine-face diameter of D2 = 3.957 ft. Inlet B was the designation given to the inlet with the updated
and smaller engine face diameter measured by the University of Illinois with D2 = 3.802 ft. Both inlets were modeled
to operate at different corrected flow rates or engine-face Mach numbers, as discussed above. The dimensions are
different between the two inlets because the modeling was scaled by the engine-face diameter.
11
of the inlets. The internal and external surfaces of the inlet formed a portion of the boundary of the flow domain
where non-slip, adiabatic viscous wall boundary conditions were imposed. A bleed boundary condition was imposed
on the back wall of the bleed plenum to extract the bleed flow entering the plenum through the bleed slot. The flow
domain only contained the starboard half of the inlet since the inlet had geometric symmetry about the vertical plane
through the center of the inlet and flow symmetry was assumed. Symmetry or reflective boundary conditions were
imposed at the symmetry boundary.
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
Figure 9. Flow domain and boundary conditions for the CFD simulations.
The flow domain for the exterior of the inlet had a width equal to the capture width, wcap, as shown in the right-
most image of Fig. 9. This was done to reduce the size of the flow domain, and so, the computational cost of the flow
simulations. The exterior flow domain boundaries on the starboard side were planar and parallel to the symmetry
boundary. This boundary was also on the same plane as the sidewall of the external supersonic diffuser. This choice
of the flow domain boundary assumed that for supersonic flow about the inlet at zero angle-of-sideslip, there was very
little flow being spilled off to the side of the inlet. This was the case at critical and supercritical conditions; however,
at subcritical conditions, this assumption could result in inaccurate modeling of the subsonic spillage that would
normally occur over the cowl lip and sidewalls. Some of the CFD flow simulations used an inviscid wall boundary
condition for this side boundary, which prevented any spillage past the sidewalls. Other simulations used a freestream
boundary condition for this side boundary. The freestream boundary condition allowed outflow and for subsonic
outflow, the freestream pressure was imposed on the side boundary. A more accurate flow domain could be obtained
by extending the flow domain off to the side of the inlet; however, this was not done for this paper.
The inflow and farfield boundaries of the flow domain had freestream boundary conditions imposed in which the
Mach number, pressure, temperature, and angle-of-attack were specified. The inflow and farfield boundaries were
positioned just upstream of the leading-edge oblique shock so that the uniform freestream conditions could be imposed
on those boundaries. At the downstream end of the cowl exterior, the domain had an external outflow boundary where
an extrapolation boundary condition was applied for supersonic outflow.
Downstream of the engine face, a converging-diverging, outflow nozzle section was added to the flow domain to
set the flow rate within the inlet. The nozzle section moved the internal outflow boundary condition downstream of
the engine face, which reduced possible interference from the boundary condition on the flow at the engine face. The
outflow nozzle is shown in Fig. 9. The converging-diverging portion is preceded by a portion of constant-area. The
length of the outflow nozzle section was twice the diameter of the engine face, which was found to be sufficient for
this inlet. A longer length may be required for an inlet in which significant boundary-layer separation extends into
the engine face. The cross-sectional area of the nozzle throat was set by specifying the ratio of the diameter of the
nozzle throat to the diameter of the engine-face (Dnoz/D2) and was set to form choked flow at the throat. Upstream of
the nozzle throat and into the subsonic diffuser, the flow was subsonic and created the necessary back-pressure to
12
support the terminal shock at the inlet throat section. Reducing the outflow nozzle throat area increased the back-
pressure, and so, reduced the inlet flow rate. Downstream of the outflow nozzle throat, the flow was supersonic, and
so, an extrapolation boundary condition could be applied at the internal outflow boundary.
C. Computational Grid
The computational grid for the flow domain was generated by dividing the flow domain into multiple blocks and
generating structured grids for each block. SUPIN generated the blocks and grid points using an automated process.
SUPIN also created the boundary condition file for Wind-US. The inputs to the process include some factors to
determine the extents of the flow domain and the resolution of the grid points. The grid resolution factors include the
grid resolution of the first grid point away from the wall (swall), the grid resolution within the throat section in the
streamwise direction (sx), and the grid resolution at the symmetry boundary (ssym). SUPIN then imposed these grid
resolution values along the edges of the inlet geometry and flow domain to compute the required number of grid points
along those edges. A grid block topology was assumed for the inlet to form the edges into faces and those faces into
blocks. SUPIN generated grids along the edges, on the surfaces, and within the interior volume of each block. The
interior block boundaries abutted with other block boundaries. For most blocks, the grid lines were contiguous across
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
block boundaries, but some non-contiguous boundaries were used to facilitate the structured topology. Figure 10
shows an example of the multi-block topology in the forward portion of the inlet. The blocks are wrapped about the
cowl lip and entrance to the bleed slot to cluster grid points about those features.
The simulations presented in this paper used a wall grid spacing of swall = 2.0x10-5 feet, which resulted in values
of y+ 1.0 for the first grid point off the wall. The axial spacing in the throat were sx 0.06 feet. The typical grid
generated by SUPIN resulted in 53 blocks with a total of approximately 28 million grid points.
13
E. Inlet Performance Metrics
The flow solutions from the CFD simulations were used to obtain the aerodynamic performance metrics of the
inlet. The four inlet performance metrics used to characterize the inlet included the inlet capture flow ratio (W1/Wcap),
the inlet total pressure recovery (pt2/pt0), the inlet circumferential (IDC) and radial (IDR) distortion indices. The inlet
flow ratio was defined as the inlet flow rate (W1) divided by the reference capture flow rate (Wcap). The inlet flow rate
(W1) was obtained by summing the engine-face and bleed flow rates (W1 = W2 + Wbleed). The engine-face flow rate
(W2) was obtained from the CFD simulation by integrating the rate of flow passing through the cross-stream grid
planes of the outflow nozzle. The total pressure at the engine face (pt2) was computed as the mass-average of the total
pressures at the grid plane at the engine face.
The third and fourth metrics of inlet performance were indices of the inlet circumferential (IDC) and radial (IDR)
total pressure distortion at the engine face as defined by General Electric [11]. The distortion indices were defined on
a standard 40-probe rake array of the Society of Automotive Engineers (SAE) Aerospace Recommended Practices
(ARP) 1420 document [12]. The rake array consisted of eight radial rakes each containi ng five total pressure probes.
In the circumferential direction the eight probes were located at a constant radius and formed a ring about the
circumference of the engine face. Each ring was placed radially at the centroid of equal-area sectors. The flowfield
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
from the CFD simulation was interpolated onto the locations of the probes to obtain the total pressure at the probe
location. The IDR index is identical to the radial distortion index defined in the SAE ARP 1420, Ref. [12]. The
method for computation of the IDC index for each ring can be found in Ref. [11]. The IDC indices reported in this
paper were computed as the average of the IDC indices computed on the two outer rings because the total pressure
distortion was predominately a tip distortion.
V. Results
The results of the CFD simulations of the Concorde inlet at the design cruise conditions are presented in this
section with comparison to available experimental data and descriptions of the flowfield features reported in the
references. The station L conditions for the inflow to the inlet for the simulations were a freestream Mach number of
ML = 1.915 and total pressure and total temperature of ptL = 1027.18 psf and TtL = 675.99 oR, respectively, which
corresponded to conditions for the standard atmosphere at an altitude of h0 = 60000 feet. The CFD simulations were
performed for a sequence of outflow nozzle areas to sweep over a range of inlet flow ratios (W1/Wcap).
The results will be reported for three sets of CFD simulations. The first set is for Inlet A in which the side boundary
containing the sidewall was specified with an inviscid wall boundary condition to impose no spillage past the sidewall.
The second set is also for Inlet A, but the side boundary was specified with a freestream boundary condition to allow
flow to be spilled through that boundary. The third set of CFD simulations was for Inlet B in which the freestream
boundary condition was applied on the side boundary.
Figure 11 shows plots of the performance data for the three sets of CFD simulations. The upper left plot s hows
the characteristic curve relating the total pressure recovery (pt2/ptL) to the inlet flow ratio (W1/Wcap). The data from
Ref. [4] by Rettie and Lewis was from wind tunnel tests of scale models of the Concorde inlet and used the term
“Capture Mass Flow Ratio”, which here is assumed to be the inlet flow ratio (W1/Wcap). Thus, the wind tunnel data
suggests a spillage of 4.5% for the supercritical leg of the characteristic curve. The inlet modeling assumed a spillage
of 2.1% as indicated by Ref. [5]. The supercritical simulations for Inlet A resulted in a spillage of 3.2% when the
reflection boundary condition was used for the side boundary and 3.6% when the freestream boundary condition was
used. This suggested that the freestream boundary condition on the side boundary allowed some spillage at the
supercritical condition. The supercritical simulations for Inlet B using the freestream boundary condition at the side
boundary also had a spillage of 3.6%.
The critical condition for the CFD simulations was established as the simulation for which the inlet flow ratio was
near the supercritical ratio while maximizing the total pressure recovery. The simulations of Inlet A indicated a critical
total pressure recovery of about pt2/ptL = 0.956. The simulation of Inlet B indicated a critical total pressure recovery
of pt2/ptL = 0.954. These values compare well to the wind tunnel critical value of pt2/ptL = 0.949. However, these
critical operating points occurred at engine-face Mach numbers below that of the design corrected flow rate for the
inlets as listed in Table 2. Inlet A had a design engine-face Mach number of M2 = 0.472, but the critical simulation of
Inlet A with the reflection side boundary condition had an engine-face Mach number of M2 = 0.439 and with the
freestream side boundary condition of M2 = 0.424. Inlet B had a design engine-face Mach number of M2 = 0.500, but
for the critical simulation of Inlet B with the freestream side boundary conditio n had an engine-face Mach number of
M2 = 0.467. These results suggest that the modeled inlets have too large of capture area (Acap). The correct capture
area could easily be computed if the capture width (wcap) was known; however, that value remains unknown to the
author at this time.
14
The subcritical simulations indicate large changes in the inlet performance with small changes in the engine face
Mach number. As the inlet becomes subcritical, the terminal shoc k is pushed forward of the bleed slot and eventually
becomes a normal terminal shock that interacts with the boundary layer on the front ramp. This results in a small
region of boundary layer separation on the ramp. The Mach number contours of Fig. 12 shows this interaction. With
a slight decrease in the engine-face Mach number, the terminal shock is pushed further upstream as shown in the
bottom-right image of Fig. 12. The wind-tunnel data of Fig. 11 indicated a gradual reduction in total pressure recovery
as the inlet becomes more subcritical. However, the CFD simulations indicated a more drastic subcritical behavior
with lower total pressure recoveries. The use of the freestream boundary condition for the side boundary did allow
greater sideways spillage that allowed for a more realistic subcritical behavior. Clearly, better modeling of the sidewall
and flow domain about the side of the inlets is needed to properly model the subcritical portion of the characteristic
curve for this inlet.
The top-right plots of Fig. 11 show the variation of the slot bleed (Wbleed/Wcap) with the inlet flow ratio (W1/Wcap).
In the supercritical simulations, the bleed increases with decreased engine-face Mach number. However, in the
subcritical simulations, the bleed rate becomes nearly constant.
The bottom-left plots of Fig. 11 show the variation of the engine-face Mach number (M2) with the inlet flow ratio
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
(W1/Wcap). As mentioned above, for the subcritical simulations, the engine-face Mach number only changes slightly
for large changes in the inlet flow ratio.
The lower-right plots of Fig. 11 show the variation of the distortion indices for the simulations. While the values
show some variation, the important check is that the values are below the limits established for the engine. The IDC
and IDR limits for a military-grade turbofan engine are 0.2 and 0.1, respectively [13]. Thus, the distortion indices
indicated by the CFD simulations are below these limits.
Images of the Mach number contours on the symmetry plane and cross-planes of the inlet are shown in Fig. 12.
Figure 13 shows the Mach number contours on the symmetry plane about the external supersonic diffuser and in the
throat section for a critical operating point. The oblique shock from the leading edge of the inlet can be clearly seen.
The shock passes ahead and below the cowl lip to indicate the presence of supersonic spillage. The isentropic section
of the external supersonic diffuser creates several visible Mach waves that are focused at the cowl lip. Mostly uniform
flow is shown about the forward ramp leading to the terminal shock. The terminal shock appears to be a strong oblique
shock that is slightly titled back into the throat section and interacting with the bleed slot. The red-dashed line indicates
15
the theoretical strong oblique shock angle for the critical condition. Figure 13 clearly shows the supersonic jet directed
into the bleed plenum that is responsible for most of the bleed flow. For the simulations, the bleed flow leaves the
bleed plenum at the back of the plenum, as indicated by the images of arrows indicating the flow at the boundary.
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
Figure 12. Mach number contours for the CFD simulations of the Concorde inlet.
Figure 13. Mach number contours on the symmetry plane about the external
supersonic diffuser and throat section for the inlet at the critical operating condition.
VI. Conclusion
The geometric and aerodynamic modeling of the Concorde inlet was accomplished using publicly-available
information and the SUPIN tool. CFD simulations were performed to compute the flowfield through the inlet and
obtain inlet performance metrics for the cruise design condition of M0 = 2.0. The simulations were able to obtain the
reported peak recovery of the inlet of approximately pt2/ptL = 0.96. However, the simulated critical operating point
was at lower engine-face Mach numbers or engine corrected flow rates than the design conditions. While the published
references on the Concorde inlet provided much information on the geometry and flow conditions, there remains
several uncertainties that prevent a definitive modeling of the inlet. Key among these is the capture width (wcap) and
16
length of the inlet. The author has contacted a museum with a Concorde on display with the intent on having such
measurements taken from the inlet. Another point of uncertainty is the actual design corrected flow rate or engine -
face Mach number for the engine. The two inlets studied in this paper used two different values; however, similar
inlet performance was obtained. The modeling of this paper approximated the side boundary of the inlet. A more
correct modeling is needed to model the geometry of the sidewall and splitter plate and flow domain about the side of
the inlet. Such modeling will be a focus of future work.
Acknowledgments
This work was funded by the NASA Commercial Supersonic Technology (CST) Project. Some of the CFD
simulations were performed using the High-End Computing Capability of the NASA Advanced Supercomputing
(NAS) Facility. The author would like to thank Melita Wiles, a former high-school volunteer intern, for extracting
some of the geometry and performance data for the Concorde. The author would also like to thank Gordon Roxburgh,
Joe Wilding, Greg Elliot, and Scott Dalbey for their assistance with verifying some of the geometry of the Concorde
inlet.
Downloaded by CORNELL UNIVERSITY on August 19, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-3770
References
[1] Candel, S., “Concorde and the Future of Supersonic Transport,” AIAA Journal of Propulsion and Power, Vol. 20,
No. 1, January-February 2004, pp. 59-68.
[2] Slater, J. W., “SUPIN: A Computational Tool for Supersonic Inlet Design,” AIAA Paper 2016-0532, January
2016.
[3] Yoder, D. A., ‘Wind-US User’s Guide: Version 3.0,” NASA TM 2016-219078, March 2016.
[4] Rettie, I. H. and Lewis, W. G. E., “Design and Development of an Air Intake for a Supersonic Transport Aircraft,”
AIAA Journal of Aircraft, Vol. 5, No. 6, November-December, 1968, pp. 513-521. doi: 10.2514/3.43977.
[5] Cain, T., “Design and Evaluation of High Speed Intakes,” NATO EN-AVT-195-03, September 2011.
[6] Walatka, P. P., Buning, P. G., Pierce, L., and Elson, P. A., “PLOT3D User’s Manual,” NASA-TM-101067, March
1990.
[7] Society of Automotive Engineering (SAE), “Aircraft Propulsion System Performance Station Designation and
Nomenclature,” Aerospace Standard (AS) 755, December 1997.
[8] Leyneart, J., Surber, L. E. and Goldsmith, E. L., “Transport Aircraft Intake Design,” Practical Intake Aerodynamic
Design, Seddon, J. and Goldsmith, E. L., editors, AIAA Education Series, New York, 1993. Ch. 5.
[9] MIL-STD-5008C, “Engines, Aircraft, Turbo-Jet, and Turbofan, Model Specifications For,” December 1965.
[10] Menter, F. R., “Two-Equation Eddy-Viscosity Turbulence Models for Engineering Applications,” AIAA Journal,
Vol. 32, No. 8, 1994, pp. 1598–1605. doi: 10.2514/3.12149.
[11] Moore, M. T. “Distortion Data Analysis”, Report AFATPL-TR-72-111, February 1973.
[12] Society of Automotive Engineers (SAE), “Gas Turbine Engine Inlet Flow Distortion Guidelines,” SAE ARP
1420, Rev. B, February 2002.
[13] Steenken, W. G., Williams, J. G., Yuhas, A. J., and Walsh, K. R., “An Inlet Distortion Assessment during Aircraft
Departures at High Angle of Attack for an F/A-18A Aircraft,” NASA TM-104328, March 1997.
17