You are on page 1of 20

ISABE-2019-24429 1

Powered, Aerodynamic
Simulations of an Airbreathing,
Hypersonic Vehicle
Alexander D. T. Ward1 and Michael K. Smart2
alexander.ward1@uq.net.au
1PhD Candidate, 2Professor

Centre for Hypersonics


The University of Queensland
Brisbane
Australia

Airbreathing, hypersonic vehicles powered by scramjets offer significant and unique


benefits over traditional rockets but also produce complex aerodynamic interactions.
At hypersonic Mach numbers, their engines must be large and tightly integrated with
the airframe thus coupling the aerodynamic and propulsive forces and moments. This
paper presents 3D simulation results for a proposed hypersonic accelerator at a
trimmed, lift equals weight flight condition at Mach 8. The vehicle geometry is built
parametrically and automatically for CFD simulations using the inviscid solver,
Cart3D, with viscous forces calculated using a boundary layer code.

A key parameter for thrust production is the exit area of the scramjet nozzle. In this
work, the nozzle expansion area was varied by increasing the frontal area around the
engine modules. The increased expansion was observed to create additional thrust,
however the detrimental effects of reduced angle of attack on engine performance
and trim drag resulted in reduced overall net thrust.

For meaningful results, airframe-integrated vehicles can and should be analysed at a


realistic flight condition such as trimmed and lift equals weight. Furthermore,
aerodynamic and propulsive contributions must be considered concurrently. Future
work includes applying this methodology to investigations of the boat tail geometry.

Keywords: hypersonic; airbreathing; accelerator; airframe-integration;


scramjet; nozzle; thrust margin; trimmed; Cart3D

ISABE 2019
2 ISABE 2019

NOMENCLATURE
α Angle of attack, °
Elevon deflection angle, °, positive down
Estimated discretisation error
Trajectory angle, °
Mesh refinement functional
, Nozzle exit area, m
, Combustor exit area, m
Drag coefficient
Lift coefficient
Moment coefficient about centre of gravity
Mach number
Dynamic pressure, kPa
Thrust, N
Velocity, m/s
Weight, N

1.0 INTRODUCTION
If reliable and reusable vehicles for the launch of small satellites are developed, an
increase in launches is expected, potentially with a commensurate decrease in cost [1, 2].
A technology shift to hypersonic accelerators equipped with airbreathing propulsion can
fundamentally change access-to-space. The primary benefit is in their improved
operability, which mimics a modern-day airliner compared to traditional, expendable
rockets [3]. This stems from fundamental design differences, relatively high fuel
efficiency (specific impulse), improved reusability and responsiveness. In particular, the
supersonic combustion ramjet, or scramjet, is an attractive engine choice for the
hypersonic portion of a staged launch system [4].
The benefits of reusability are beginning to be explored with the propulsive boost-back
and vertical landing of the SpaceX Falcon 9 [5]. In 2019, SpaceX have reused the same
orbital booster three times and have a goal of flying that same booster twice more before
the end of the year [6]. Although revolutionary, further advancement in rocket
performance is difficult because modern rockets operate close to theoretical limits [7].
Further improvement of airbreathing engines is expected given their technological
infancy.
There have been numerous, notable efforts to develop a reusable, airbreathing launcher.
For example the National Aero-Space Plane (NASP, USA, 1986-1995, [1, 2]), HOrizontal
Take Off and Landing (HOTOL, England, 1982-1989, [9]), Sänger II (Germany, [3, 4])
and SKYLON (England, [12]). For various technical, economic and political reasons
however, an airbreathing launch vehicle has not yet been realised [13].
1.1 Airframe Integration
One such impediment to their development is the unique need for the careful, synergistic
integration of the airbreathing engine into the airframe. As flight Mach number increases,
the relative size of the inlet and nozzle must increase to extract a given fraction of
available thrust from the propulsive stream [14]. At hypersonic speeds, the vehicle must
effectively become a flying engine. The forebody forms part of the inlet, capturing and
compressing the freestream and the high pressure exhaust is further expanded over the
aftbody.
WARD & SMART 24429 3

This complicates the analysis and design through tightly coupling the aerodynamic and
propulsive components and hence disciplines. The high pressure exhaust gases in
particular mean small changes to nozzle design can have a large effect on overall system
performance (acceleration capability). This is because the thrust margin of a hypersonic
accelerator tends to be the small difference between the two relatively large forces, thrust
and drag [11, 12]. The exhaust also contributes strongly to lift and pitching moment.
Much work has been done on scramjet nozzle integration to various classes of vehicles.
For example, Edwards (1988) highlighted the importance of cowl geometry on vehicle
performance for a planar scramjet engine module and concluded 3D simulations were
required because of the complex interactions between separate exhaust flows and the
vehicle geometry [17]. Nozzle integration on accelerating waveriders was studied by
O’Neill and Lewis [18] while Weidner et al. and Henry et al. explored the same but for
winged accelerators with planar scramjet flowpaths [19, 20].
This paper presents a methodology and preliminary results for powered, aerodynamic
simulations of a scramjet-powered accelerator for the purposes of evaluating nozzle
design and integration problems. This vehicle was proposed by Smart and Tetlow in [4]
and subsequently the subject of vehicle and trajectory optimisation studies [20, 21]. It
therefore provides an established candidate geometry. Furthermore, as a conical vehicle,
it provides unique geometry which has not been studied in this way or for these purposes
before.
1.2 Objectives
Our objectives were 1) to present a methodology for the simulation of realistic hypersonic
vehicles suitable to the preliminary design stage and 2) to demonstrate this methodology
in evaluating a design change to the external shroud of the vehicle’s scramjet modules.
This evaluation is the first step in determining a desirable shape for the nozzle exit.
1.3 Approach
The general approach is as follows.
1. Generate the geometry parametrically in individual, logical components using
the Python library of the CAD package, Rhinocerous (Section 2.0).
2. Intersect each of these individually watertight components to obtain a single,
watertight surface mesh of the entire vehicle.
3. Simulate the inviscid, 3D flowfield around the powered vehicle using Cart3D
and correct for viscous drag with a boundary layer code (Section 3.1).
4. Simulate the internal scramjet isolator and combustor using a quasi-1D cycle
analysis code to obtain combustor exit conditions (Section 3.3).
5. Perform simulations at a given Mach number and a trimmed, lift equals weight
condition. The trim forces are approximated externally (Section 3.2 and 3.4).

2.0 VEHICLE MODEL


The vehicle studied, SPARTAN, was originally modelled on the Winged Cone Vehicle
of Shaughnessy et al. [23]. It has a conical forebody, cylindrical fuselage and truncated
boat tail. The wings and tail are swept at 70° and 50° respectively and have a diamond
cross section with a root thickness to chord ratio of 4%. The four scramjet engines are
underslung and tightly integrated with shape transition inlets designed specifically for a
conical vehicle [24]. The thrust-producing, expansion surfaces, are comprised of a
straight, internal nozzle and the boat tail. This is similar in form to a truncated, plug nozzle
with clustered, ring, modules [25].
For this work, a new geometry tool has been developed. The vehicle geometry is defined
entirely within a Python script using the functions and tools of the CAD package,
Rhinocerous 6.0, made available through the Python library, RhinoPython. The python
script is executed within Rhino, similarly to [26]. This gives a precise definition of the
vehicle geometry.
4 ISABE 2019

Scripting the generation of the geometry with Python has several benefits. It allows
precise variation of a specific parameter with the ability to include logical rules for how
to generate dependent geometry. This allows a robust parametric model. Practically, the
Python script becomes a record of that specific vehicle and other information such as key
parameters (e.g. exit area) can be calculated and written to a logfile.
The CFD solver used (Cart3D, introduced in Section 3.1), requires a single watertight
mesh. The robust mesh intersection tools included with Cart3D [27] were used to create
the final geometry from the individual components. This retains the definitions of the
component surfaces and allows a straightforward breakdown of forces. Figure 1 shows
the baseline vehicle examined in this work coloured by the 24 individual vehicle
components.

Figure 1 Baseline vehicle geometry coloured by components. The vehicle is made up of 24


individual components.

2.1 Vehicle Geometry Investigated


The effect of varying one parameter on the net thrust of the vehicle was investigated. This
parameter was the radius of the shroud , in between neighbouring engine modules
(Figure 2). Although geometrically simple, this dimension results in a complex change to
the scramjet shroud. Furthermore, from experience, we know the engine shroud
contributes significantly to the vehicle forces.
The bodyside geometry is kept constant with an expansion angle of 15° along the engine
centreline in the nozzle and boat tail. As shown in Figure 2, can vary between
two extremes defined by the radius of the cylindrical fuselage (1.05 m) and the radius of
the engine module shroud at the engine centreline (1.77 m). This meant the exit area of
the internal nozzle ( , varied from 0.519 to 0.819 (58% increase).

The centreline shroud geometry was designed to cant inwards at 2°, the vehicle’s nominal
angle of attack. This was chosen to minimise the frontal area. However, the rear of each
engine module does come out of aerodynamic shadow when they are integrated
circumferentially around the axisymmetric vehicle and it is at some angle of attack. This
is shown in the bottom row of images in Figure 3.
WARD & SMART 24429 5

Figure 2 The radius of the shroud ( ) in between the engine modules was varied in the current study.

(a) , 0.519 m (b) , 0.647 m (c) , 0.861 m


Figure 3 The minimum, intermediate and maximum (left to right) nozzle expansion vehicle geometries studied.
The bottom row of images show the vehicle at its nominal angle of attack, 2°.

3.0 AERODYNAMIC MODEL


The aerodynamics of the vehicle were calculated using Cart3D and a boundary layer code
for skin friction drag. Whenever required, the freestream air properties were all calculated
using the 1976 U.S. Standard Atmosphere [28].
3.1 External Aerodynamics
The high-fidelity, CFD package Cart3D was used to simulate the 3D, external flowfield
[29, 27]. Cart3D solves the Euler equations for inviscid flow using a perfect gas model
on a Cartesian mesh with embedded boundaries. The flux splitting technique of Van Leer
[30] and minmod limiter function [31] were used to obtain steady-state solutions. The
spatial discretisation was second-order accurate.
The volume mesh was an unstructured, regular mesh of hexahedrons except for a layer of
body-intersecting cut cells. This mesh was built and refined automatically using adjoint
error estimates [32]. The adjoint solver was first-order accurate.
The inviscid flow solution on the body was used as boundary layer edge conditions for a
viscous correction code [33]. Laminar and turbulent skin friction coefficients were
modelled using the Smart-Meador reference temperature [34] and van Driest II [35] flat
plate correlations respectively. The local running length was estimated as the Euclidean
distance to the local stagnation feature. No coupling or viscous-interaction effects were
considered.
This aerodynamic analysis method was validated against wind-tunnel data for different
configurations at supersonic and hypersonic Mach numbers and high Reynolds numbers.
6 ISABE 2019

These results and further details may be found in [33]. For the slender bodies investigated,
the viscous drag was 15‐25% of the total drag. By computing the viscous drag and adding
it to the Cart3D value, the total drag predictions were within 6% of the experimental
values on average. Lift and pitching moment are generally predicted well by Cart3D and
within 5% of experimental values [36, 37].
The propulsive boundary conditions developed in [38] were used to model the flow
conditions at the combustor exit. Previous work has shown that 3D, inviscid computations
are adequate to predict core flow features. However, depending on wake structure, the
inviscid Euler equations cannot accurately predict the flowfield around the base with
strong interactions between the inviscid and viscous flow regimes [41, 42, 43].
3.2 Deflected Elevon Aerodynamics
The vehicle elevons were used to trim the pitching moment created by the airframe and
engines. They were placed such that the undeflected trailing edge was coincident with the
wing trailing edge. The elevons were originally sized in [22] based on the experimental
work of [42] to have an area and chord of 15% of the wing planform area and chord
respectively. Following later vehicle and trajectory analyses using the methodology
developed in [22], the elevon area was reduced slightly to 10% (1.66 m ). The elevon
chord was kept at 15% (1.38 m) giving a span of 1.20 m.
The aerodynamic forces on the deflected elevons are approximated using classical shock-
expansion theory and assuming 1.4 [43]. This approximation was made to avoid
modifying the vehicle geometry during a simulation. The swept wing and deflected
elevon are modelled in 2D and in isolation. That is, the flow seen by the wing leading
edge is the freestream and interaction effects with the forebody and fuselage and edge
effects of the finite elevon span are neglected. A schematic of this model is shown in
Figure 4a.
For compression faces, the weak oblique shock angle caused by flow turning by , is
found by solving the oblique shock ‐ ‐ equation (Eq. 1) using the secant root-finding
method.
2 sin 1
tan …( 1 )
tan cos 2 2

The downstream Mach number , and pressure , are calculated from the oblique
shock relations given in Eq. 2 and 3.

1 1 sin 1 /2 …( 2 )
sin sin 1 /2

2
1 sin 1 …( 3 )
1

For expansion through a turning angle of , the Mach number downstream , of the
expansion fan is calculated from the Prandtl-Meyer function either side of the fan:

…( 4 )

Where is defined in Eq. 5 and solved using the secant method:

1 1 …( 5 )
arctan 1 arctan 1
1 1
WARD & SMART 24429 7

The downstream pressure , is then is then calculated:

1 1 /2 …( 6 )
1 1 /2

The resultant aerodynamic loads were assumed to act at the centroid of the deflected
elevon. Care was taken to remove the forces of the undeflected elevon from the Cart3D
solution. No viscous forces were included at this stage but viscous drag was calculated
on the entire wing (i.e. including undeflected elevon) and included in the vehicle forces.
The predicted aerodynamics of the deflected elevons for 15° 15° at 8 and
2° are shown in Figure 4.

(a) Schematic of the modelled wing (b) Elevon lift

(d) Elevon moment about vehicle centre


(c) Elevon drag
of gravity.
Figure 4 Aerodynamics of deflected elevons on the vehicle at its nominal 2°.
8 ISABE 2019

3.3 Propulsion
The scramjet isolator, injector block and combustor were modelled using quasi-1D, cycle
analysis [44]. This code solves the differential equations for conservation of mass,
momentum and energy. The properties of the air flowing into the duct and therefore the
inflow to the 1D code were calculated a priori with 3D CFD simulations. The different
capture and pre-compression of the inboard and outboard inlets was accounted for by
running separate cases. An average skin friction coefficient of 0.002 was assumed and
gaseous hydrogen was injected at an equivalence ratio of 1.00.
The cycle code provided the flow properties of the real exhaust gases at the combustor
exit ( , , , . This flow is a hot, combusting mixture of air, fuel and combustion
products ( 1.4). The method of Pindzola [45] was followed to simulate this flow with
cold air in Cart3D (perfect gas, 1.4).
This requires that the initial expansion angle, jet to freestream pressure ratio ( / ) and
a similarity parameter (Eq. 7) be preserved. The similarity parameter (derived from
conserving the thrust coefficient) scales the exit Mach number [45]:


…( 7 )

This scaling was used in powered Cart3D simulations of Skylon and computed
aerodynamic loads compared well with independent results obtained with engineering
methods [13].
The combustor exit boundary condition was deliberately placed normal to the engine
centreline and at combustor exit, away from the nozzle exit [46]. This assumes there is
no further combustion in the nozzle and the flow exiting the divergent combustor may be
modelled as one dimensional – both limitations on the current methodology.

Figure 5 A cut through the vehicle at the engine centreline showing the throat and combustor
the vehicle geometry (black) and exit boundary conditions (BC) (red). Note the flowpath
(grey) is shown only for reference and was not part of the simulated geometry. The vehicle is
scaled by 0.75 longitudinally in this figure.

3.4 Steering Methodology


Figure 6 outlines the overall methodology for the aerodynamic simulations. The lift
coefficient steering was handled internally by Cart3D. After a set number of initial
iterations (200), if the lift coefficient was outside the given tolerance (0.0002, 0.5% of
the target lift coefficient), the angle of attack was incremented (0.025°). The direction of
this first step was made assuming / 0. The solver proceeds and updates the
angle of attack as required every 50 iterations using a current estimate of / .
While Cart3D was computing the adjoint error estimate and refining the volume mesh,
the power boundary condition update and pitch trimming scripts were called. The
combustor exit boundary conditions were updated for the current angle of attack by
linearly interpolating in the propulsion database. The pitch trimming script read the
vehicle’s current pitching moment and added the pitching moment of internal flow. It
then calculated the elevon deflection required to give an equal and opposite moment. The
target lift coefficient was updated to the original target minus the lift generated by the
deflected elevons.
WARD & SMART 24429 9

Figure 6 The general process for the trimmed, powered simulations at lift equals weight.

4.0 SIMULATION VERIFICATION


Our goal was to obtain aerodynamic forces and moments, surface properties and plume
characteristics that are independent of the surface and computational meshes. This was
achieved through adjoint-based mesh adaptation driven by an integrated force objective
function (functional). This process automatically builds a volume mesh that is specifically
tailored to the geometry and specific flight condition to minimise the error in the
functional.
For the aerodynamic simulations, the functional was a weighted, linear summation of the
integrated forces, lift and drag, 0.1 0.8 . The drag converged slower hence it
was given a higher weighting. Figure 7a presents a representative cell distribution along
the vehicle symmetry plane using this functional.
For visualisation cases, the functional included a contribution (around 40%) from
pressure signals measured along four lines, 0.33 body-lengths downstream of the trailing
edge. These were added to resolve the plume structure. This necessarily resulted in more
cells and was therefore not used for the majority of simulations. Figure 7b shows an
example mesh on two planes, a horizontal plane below the vehicle and the vehicle
symmetry plane.
10 ISABE 2019

(a) Driven by aerodynamic forces (30.0 million cells).

(b) Driven by pressure signals on downstream line sensors (48.9 million cells) for plume
visualisation. The aerodynamic forces were within 8% of those calculated on the mesh in (a).
Figure 7 Two representative examples of the adaptively refined meshes 8.0, 2.7° . The
cell edges are coloured by cell volume.

A representative example of adaptation and simulation convergence is presented for the


baseline geometry at 8.0, 2.7° (Figure 8). For all simulations, the L residual
of density converged only four orders relative to their maximum values, stalling at around
10 . Therefore, we relied on convergence of the lift, drag and moment (i.e. the
engineering quantities of interest) to demonstrate simulation convergence. This is shown
concisely in Figure 8a which illustrates the convergence (with mesh refinement) of the
functional. The error bars shown come from the adjoint solve and estimate the remaining
discretisation error in the functional. The dashed line is the Richardson extrapolation
value. The functional converges steadily and error estimates reduce with each subsequent
mesh adaptation. This implies the simulation is refining cells to reduce error in the forces
and moments thus obtaining a mesh independent solution.
Figure 8b and c show steady convergence of the lift, drag (net) and moment coefficients
normalised by their final values. The change in moment and drag between the final two
meshes is 5% and less than 2.5% respectively. Note that the lift coefficient is held nicely
at the target value throughout the simulation.
WARD & SMART 24429 11

(a) Convergence with mesh resolution of the functional 0.1 0.8 . The final value is
within 0.9% of the Richard extrapolation (dashed line).

(b) Aerodynamic coefficients with mesh (c) Aerodynamic coefficients with solver
resolution. iterations on the final three meshes.
Figure 8 Representative convergence history at Mach 8, 2.7° for the baseline geometry. The
final mesh was 30.0 million cells. The aerodynamic coefficients have been normalised by their final
values, 0.0413, 0.0310, 0.0034.

5.0 RESULTS AND DISCUSSION


Two sets of results are presented; 1) specified angle of attack and 2) specified lift
coefficient and pitch trimmed, both at Mach 8. The reference flight condition was taken
from a trajectory simulation of the baseline vehicle using the in-house code STALKER
[22]. At Mach 8 the vehicle is flying at 2.03° and 30324 m ( 50 kPa) and
has burnt 51% of its fuel. This gives a current weight of 8589.1 kg and therefore
lift coefficient of 0.0413. The trajectory angle , is 0.3° and therefore assumed to
be zero. The fuel remaining in the two tanks is modelled giving a centre of gravity at
13.99, 0, 0.147 m. For all results presented, the reference area and length were
40.837 m and 8.762 m respectively.

5.1 Mach 8, . °
The net thrust (normalised by reference area and dynamic pressure) is plotted for the five
vehicle geometries in Figure 9. The error bars on the – plot come from the adjoint
error estimate in the functional (Figure 8a). Because is also a function of lift, this gives
a conservative estimate of the remaining discretisation error in the net thrust.
12 ISABE 2019

Noting the vertical scale, it is interesting that even with such a seemingly large change to
the vehicle geometry, the actual performance doesn’t change. The increased drag of the
shroud appears to be largely balanced by the increased thrust of the internal nozzle.

Figure 9 Thrust margin for the geometries studied ( 8.0, 2.03°

Unlike the thrust margin, the lift and pitching moment coefficient increase substantially
as the shroud gets larger (Figure 10). The larger shroud has a large compression surface
which generates a large amount of lift. Furthermore, the shroud creates a shock that
pressurises the fluid under the wings (see Figure 13a versus Figure 13c).

The increase in pitching moment may be attributed to an increase in the shroud’s nose up
pitching moment and a decrease in the restoring moment from the boat tail.

Figure 10 Lift coefficient and moment coefficient about the centre of gravity (
8.0, 2.03°
WARD & SMART 24429 13

5.2 Mach 8, Trimmed,


Owing to the complexity of interactions between thrust, lift and pitching moment, it is
difficult to draw conclusions from the results of Section 5.1. In evaluating the
performance, one must consider these changes concurrently. Therefore, we propose
simulating the vehicles at a trimmed, lift equals weight flight condition.
The same results are presented but now showing angle of attack and elevon deflection
rather than lift and pitching moment coefficients. To achieve the required lift, Cart3D
predicts a higher angle of attack at 2.8° compared to the engineering methods (strip
theory, shock expansion) of STALKER. The increase in angle of attack actually means
the net thrust of the small nozzles has increased. This implies the engines produce more
thrust quicker than the drag of the vehicle increases as angle of attack increases.

Figure 11 Thrust margin ( 8.0, 0.0413, 0

All vehicles produced a positive, nose-up pitching moment requiring a positive, elevon
down deflection. The deflected elevons increase the L/D of the wings. This partially
offsets the drag penalty of the elevons because the vehicle now flies at a lower angle of
attack. As the pitching moment increases, the required elevon deflection and lift increase.
This decreases the required angle of attack and consequently decreases the airframe drag
but also engine thrust. The elevon drag further decreases performance; ,
0.00002 for , 0.519 m and , 0.0023 for , 0.861 m .
14 ISABE 2019

Figure 12 Angle of attack and elevon deflection ( 8.0, 0.0413, 0 .

5.3 Visualisation Mach 8, Trimmed,


Flowfield visualisations for three of the vehicles are presented in Figure 13 and Figure
14. These show pressure contours on the vehicle surface and on planes at 0 and
0 respectively. The extremely complex wave structures formed between the vehicle and
individual exhaust streams expanding over the boat tail are immediately evident. At this
flight condition, it appears that the exhaust gases impinge on the tails. This requires
further investigation and was not an objective of the current study.
Recompression regions are visible on the boat tail for all cases. These are particularly
visible (for the scale chosen) in Figure 13b and Figure 13c. It is expected that these arise
from the convergence of the four expanding flows over the external nozzle surfaces. This
behaviour is shown schematically in Figure 17 of [39].
A comparison of Mach number contours through the engine symmetry is presented for
the minimum and maximum area nozzles in Figure 15. These results are for the inboard
engines and the topology was observed to be qualitatively similar to an outboard engine.
The contours in these Figures must be questioned however because of the presence of a
seemingly viscous base flow. We propose this is because of numerical dissipation
imitating fluid viscosity arising from our selection of the highly dissipative but robust
flux calculator, van Leer.
Although erroneous, it is suggested that the effects on the vehicle forces are small. Firstly,
the areas are small. Additionally, the base region took the form of a closed wake structure
with wall pressures around 50% of ambient. This is a conservative result compared to an
open wake structure which has higher base pressures similar to ambient [39, 47]. This
flow topology may be predicted by a viscous analysis with the ability to predict
separation. Finally, this flow is at the back of the vehicle and therefore does not influence
the upstream vehicle solution.
WARD & SMART 24429 15

(a) Minimum expansion nozzle, , 0.519 m

(b) Baseline vehicle, , 0.647 m

(c) Maximum expansion nozzle, , 0.861 m

Figure 13 Pressure fields on plane at 0 and vehicle underside for 8.0, 0.0413
and 0.0 (pressure shown in Pa).
16 ISABE 2019

(a) Minimum expansion nozzle, , 0.519 m

(b) Baseline vehicle, , 0.647 m

(c) Maximum expansion nozzle, , 0.861 m

Figure 14 Pressure fields on symmetry plane at 8.0, 0.0413 and 0.0


(pressure shown in Pa).
WARD & SMART 24429 17

,
a) Minimum expansion nozzle, , 0.519 m , 9.4

,
b) Maximum expansion nozzle, , 0.861 m , 4.3
Figure 15 Mach number contours through the inboard engine centreline 8.0,
0.0413 and 0.0.

6.0 CONCLUSIONS
A scramjet powered, hypersonic accelerator has the potential to decrease the cost of
access-to-space, primarily through operability and reusability benefits. For adequate
acceleration performance, these vehicles require relatively large engines which must be
integrated within the airframe. Inviscid simulations were performed using adjoint-driven
mesh refinement to obtain verified results.
By varying the size of the external scramjet shroud, a trade-off between thrust-producing,
expansion area and extra frontal area was investigated. At constant angle of attack, the
difference in net thrust between designs was small but commensurate changes in lift and
pitching moment coefficient complicate comparisons. However, the difference became
more pronounced when simulations were performed at the more realistic condition of
pitch-trimmed at a specified lift coefficient. The benefits of increased expansion were
primarily offset by the detrimental effects of trim drag and reduced angle of attack on
engine performance.

These results emphasise the importance of accounting for the coupling between
aerodynamic and propulsive components in the analysis of airframe-integrated vehicles.
Future work includes investigating this geometry change at different Mach numbers on
the accelerator’s trajectory (different freestream to combustor exit pressure ratios).
Finally, the applicability of the inviscid solver in the base region and the effects of this
region on aerodynamic forces and moments must be considered.
18 ISABE 2019

ACKNOWLEDGMENTS
A. Ward gratefully acknowledges the financial support of an Australian Government
Research Training Program (RTP) Scholarship. The authors are greatly appreciative of
the contributions of C. Pierens and R. Palmer for the mass modelling of SPARTAN; D.
Curran for the CFD simulations of the CREST forebody and inlet and cycle analyses of
the CREST scramjet flowpath; and finally, R. Jones for the CREST inlet geometry post-
processing tools.

REFERENCES
[1] A. Butrica, “Reusable Launch Vehicles or Expendable Launch Vehicles? A
Perennial Debate,” in Critical issues in the history of spaceflight, no. August, S.
J. Dick and R. D. Launius, Eds. Government Printing Office, 2006, pp. 301–42.
[2] FAA, “The Annual Compendium of Commercial Space Transportation: 2018,”
2018.
[3] A. Dissel, A. Kothari, V. Raghavan, and M. Lewis, “Comparison of HTHL and
VTHL Airbreathing and Rocket Systems for Access to Space,” 40th
AIAA/ASME/SAE/ASEE Jt. Propuls. Conf. Exhib., no. July, 2004.
[4] M. K. Smart and M. R. Tetlow, “Orbital Delivery of Small Payloads Using
Hypersonic Airbreathing Propulsion,” J. Spacecr. Rockets, vol. 46, no. 1, pp.
117–125, 2009.
[5] M. Vozoff and J. Couluris, “SpaceX Products - Advancing the Use of Space,” in
AIAA SPACE 2008 Conference & Exposition, 2008, no. September 2008, pp. 1–
9.
[6] M. Wall, “SpaceX Plans to Fly a Single Rocket 5 Times by the End of the Year,”
SPACE.com, 2019. [Online]. Available: https://www.space.com/spacex-fly-
falcon-9-five-times-2019.html. [Accessed: 02-Jul-2019].
[7] U. B. Mehta, “Strategy for developing air-breathing aerospace planes,” J. Aircr.,
vol. 33, no. 2, pp. 377–385, Mar. 1996.
[8] B. Augenstein et al., “The National Aerospace Plane (NASP): Development
Issues for the Follow-on Vehicle,” 1993.
[9] R. Hannigan and D. Webb, “Spaceflight in the aero-space plane era,” in 3rd
International Aerospace Planes Conference, 1991, pp. 102–103.
[10] D. E. Koelle, “SÄNGER II, A Hypersonic Flight and Space Transportation
System.” ICAS-88-1.5.1, 1988.
[11] D. E. Koelle and H. Kuczera, “Sänger II, An Advanced Launcher System for
Europe,” Acta Astronaut., vol. 19, no. 1, pp. 63–72, 1989.
[12] R. Longstaff and A. Bond, “The SKYLON Project,” vol. 2, no. April, pp. 1–10,
2012.
[13] U. Mehta, M. Aftosmis, J. Bowles, and S. Pandya, “Skylon Aerospace Plane and
Its Aerodynamics and Plumes,” J. Spacecr. Rockets, vol. 53, no. 2, pp. 340–353,
2016.
[14] K. G. Bowcutt, “Physics Drivers of Hypersonic Vehicle Design,” 22nd AIAA Int.
Sp. Planes Hypersonics Syst. Technol. Conf., no. September, pp. 1–22, 2018.
[15] P. J. Johnston, J. M. Cubbage, and J. P. Weidner, “Studies of Engine-Airframe
Integration on Hypersonic Aircraft,” J. Aircr., vol. 8, no. 7, pp. 495–501, 1971.
[16] J. Haney and M. Bradley, “Waverider nozzle integration issues,” in 33rd
Aerospace Sciences Meeting and Exhibit, 1995.
WARD & SMART 24429 19

[17] T. A. Edwards, “The Effect of Exhaust Plume/Afterbody Interaction on Installed


Scramjet Performance,” Moffett Field, 1988.
[18] M. K. L. O’Neill and M. J. Lewis, “Design Tradeoffs on Scramjet Engine
Integrated Hypersonic Waverider Vehicles,” J. Aircr., vol. 30, no. 6, pp. 943–
952, 1993.
[19] J. P. Weidner, W. J. Small, and J. A. Penland, “Scramjet Integration on
Hypersonic Research Airplane Concepts,” J. Aircr., vol. 14, no. 5, pp. 460–466,
1977.
[20] J. R. Henry and G. Y. Anderson, “Design Considerations for the Airframe-
Integrated Scramjet,” Washington, D.C., 1973.
[21] S. O. Forbes-Spyratos, M. P. Kearney, M. K. Smart, and I. H. Jahn, “Trajectory
Design of a Rocket–Scramjet–Rocket Multistage Launch System,” J. Spacecr.
Rockets, vol. 56, no. 1, pp. 53–67, 2018.
[22] D. Preller and M. K. Smart, “Reusable Launch of Small Satellites Using
Scramjets,” J. Spacecr. Rockets, vol. 54, no. 6, pp. 1317–1329, 2017.
[23] J. D. Shaughnessy, S. Z. Pickney, J. D. McMinn, C. I. Cruz, and M.-L. Kelley,
“Hypersonic Vehicle Simulation Model: Winged-Cone Configuration. NASA
TM-102610,” 1990.
[24] J. Kunze, M. K. Smart, and R. J. Gollan, “A Design Method for Shape Transition
Nozzles for Hypersonic Vehicles,” no. September, 2018.
[25] G. Hagemann, H. Immich, T. Van Nguyen, and G. E. Dumnov, “Advanced
Rocket Nozzles,” J. Propuls. Power, vol. 14, no. 5, pp. 620–634, 2008.
[26] A. Sobester, “Self-Designing Parametric Geometries,” in 56th
AIAA/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials
Conference, 2015, no. January, pp. 1–8.
[27] M. J. Aftosmis, M. J. Berger, and J. E. Melton, “Robust and Efficient Cartesian
Mesh Generation for Component-Based Geometry,” AIAA J., vol. 36, no. 6, pp.
952–960, 1998.
[28] NASA, “U.S. Standard Atmosphere,” 1976.
[29] M. J. Aftosmis, “Solution Adaptive Cartesian Grid Methods for Aerodynamic
Flows with Complex Geometries,” von Karman Institute’s 28th Comput. Fluid
Dyn. Lect. Ser., 1997.
[30] B. van Leer, “Flux-Vector Splitting for the Euler Equation,” in Upwind and High-
Resolution Schemes, Hussaini M.Y., B. van Leer, and J. Van Rosendale, Eds.
Springer Berlin Heidelberg, 1997.
[31] M. Berger, M. Aftosmis, and S. Muman, “Analysis of Slope Limiters on Irregular
Grids,” pp. 1–22, 2013.
[32] M. Nemec, M. Aftosmis, and M. Wintzer, “Adjoint-Based Adaptive Mesh
Refinement for Complex Geometries,” in 46th AIAA Aerospace Sciences Meeting
and Exhibit, 2008, pp. 1–23.
[33] A. D. T. Ward, M. K. Smart, and R. J. Gollan, “Development of a rapid inviscid-
boundary layer aerodynamics tool,” in 22nd AIAA International Space Planes
and Hypersonics Systems and Technologies Conference, 2018.
[34] W. E. Meador and M. K. Smart, “Reference Enthalpy Method Developed from
Solutions of the Boundary-Layer Equations,” AIAA J., vol. 43, no. 1, pp. 135–
139, 2005.
[35] E. R. van Driest, “The Problem of Aerodynamic Heating,” Aeronaut. Eng. Rev.,
20 ISABE 2019

vol. 15, no. 10, pp. 26–41, 1956.


[36] D. G. Sagerman, M. P. Rumpfkeil, B. M. Hellman, and N. Dasque, “Comparisons
of Measured and Modeled Aero-thermal Distributions for Complex Hypersonic
Configurations,” in 55th AIAA Aerospace Sciences Meeting, 2017.
[37] D. Almosnino, “Assessment of an Inviscid Euler-Adjoint Solver for Prediction of
Aerodynamic Characteristics of the NASA HL-20 Lifting Body,” in 34th AIAA
Applied Aerodynamics Conference, 2016.
[38] S. Pandya, S. Murman, and M. Aftosmis, “Validation of Inlet and Exhaust
Boundary Conditions for a Cartesian Method,” 22nd Appl. Aerodyn. Conf. Exhib.,
no. January, pp. 1–16, 2004.
[39] G. Hagemann, H. Immich, and M. Terhardt, “Flow Phenomena in Advanced
Rocket Nozzles - The Plug Nozzle,” in 34th AIAA/ASME/SAE/ASEE Joint
Propulsion Conference and Exhibit, 1998.
[40] G. Hagemann, “Flowfield Simulations in Advanced Rocket Nozzles,” in 3rd
European Symposium on Aerothermodynamics for Space Vehicles, 1999, pp. 89–
96.
[41] T. Rommel, G. Hagemann, C.-A. Schley, G. Kr-uacute, lle, and D. Manski, “Plug
Nozzle Flowfield Analysis,” J. Propuls. Power, vol. 13, no. 5, pp. 629–634, 2008.
[42] R. Ramesh, “Aerodynamics of Delta Wings with Flaps at Hypersonic Speeds,”
Aeronaut. J., vol. 106, no. 1060, pp. 293–312, 2002.
[43] J. D. Anderson, Modern Compressible Flow: With Historical Perspective, 3rd
Editio. McGraw-Hill Education, 2003.
[44] M. K. Smart, “Scramjets,” Adv. Propuls. Technol. High-Speed Aircr., vol. RTO-
EN-AVT, no. 9, pp. 9-1-9–38, 2008.
[45] M. Pindzola, “Jet simulation in ground test facilities,” AGARDograph, vol. 79,
no. 1, 1963.
[46] M. Gusman, J. Housman, and C. Kiris, “Best Practices for CFD Simulations of
Launch Vehicle Ascent with Plumes - OVERFLOW Perspective,” 49th AIAA
Aerosp. Sci. Meet. Incl. New Horizons Forum Aerosp. Expo., pp. 1–15, 2011.
[47] M. Onofri et al., “Plug nozzles: summary of flow features and engine
performance,” in 40th AIAA Aerospace Sciences Meeting & Exhibit, 2002.

You might also like