Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

18th AIAA/3AF International Space Planes and Hypersonic Systems and Technologies Conference AIAA 2012-5872

24 - 28 September 2012, Tours, France

Preliminary Performance Analysis of the LAPCAT-MR2


by means of Nose-to-Tail Computations

T. Langener 1 and J. Steelant 2


1
ESA-ESTEC, Keplerlaan 1, 2200 AZ Noordwijk, Netherlands
P. Roncioni 3, P. Natale 4 and M. Marini 5
CIRA, Centro Italiano Ricerche Aerospaziali, Capua, Italy
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

Hypersonic airbreathing vehicle designs are challenging due to the complexity of


airframe integration and engine cycle design able to operate over a wide Mach number
range. Airframe integration needs to deal with the mutual effect of the aerodynamics on
the engine performance. On the other hand, the engine performance affects the vehicle
design and operation. This paper first presents the followed design methodology
including a description of the different approaches used for the layout of a M=8
hypersonic cruiser vehicle. The simulation of combustion process with the Dual Mode
Ramjet engine is still based upon on 1D model whereas a dedicated 3D combustor CFD-
analysis assesses the reliability of the 1D combustion hypotheses. An assessment of the
thrust minus drag balance is carried out on a conceptual vehicle design for different
Mach numbers using three-dimensional nose-to-tail CFD simulations. Here, the loosely
coupled one-dimensional code for the estimation of the Dual Mode Ramjet performance
is applied as well. It could be shown that for the trajectory points investigated both
lateral and axial acceleration is possible.

Nomenclature

ATR = Air Turbo Rocket


CFD = Computational Fluid Dynamics
DMR = Dual-Mode Ramjet
EC = European Commission
EINO = Emission Index of NO
LAPCAT = Long-Term Advanced Propulsion Concept and Technologies
RBCC = Rocket Based Combined Cycles
TBCC = Turbo Based Combined Cycles

I. Introduction

L APCAT II (J. Steelant 2009) is a follow-up of the previous EC-project LAPCAT I (J. Steelant 2008). The
objective is to develop different vehicle concepts enabling the potential reduction of antipodal flight times to
about 4 hours along with the critical technologies and know-how to realize this ambitious goal. Among the several
studied vehicles in LAPCAT I, only two concepts, one each for a Mach 5 and 8 cruise flight, were retained for
further evaluation in LAPCAT II .
As the Mach 5 concept had a higher maturity in its conceptual design, it was retained for the next phase of
LAPCAT. The same level of maturity could not be achieved for the Mach 8 vehicle as the iterative design loop had
to shift from a RBCC towards a TBCC cycle to guarantee a long range capability. Moreover, as the Mach 8 vehicle

1
Dr.-Ing., Aerothermodynamics Engineer, Aerothermodynamics and Propulsion Analysis Section TEC-MPA.
2
PhD, Senior Research Engineer, Aerothermodynamics and Propulsion Analysis Section TEC-MPA, AIAA Member.
3
PhD, Research Engineer, Propulsion Division, Combustion Unit.
4
Aerospace Engineer, Propulsion Division, Combustion Unit.
5
PhD, Senior Research Engineer, Propulsion Division, Combustion Unit Head.

Copyright © 2012 by T. Langener, J. Steelant, P. Roncioni, P. Natale and M. Marini. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
is intrinsically a highly integrated design and its performance much dependent on its architecture, it was decided to
let different LAPCAT II teams propose different concepts.
Preliminary parametric studies within the project have shown that a Mach 8 vehicle concept based on liquid
hydrogen fuel with a long-haul range is potentially achievable. However, confidence and credibility when proposing
a fully integrated vehicle that complies with the mission goals is largely dependent on validated advanced design
tools to provide quick aerodynamic, propulsion, structural and trajectory data. In particular estimates on
aerodynamic and propulsion performances are very crucial at this phase of the study and small variations can lead to
large errors in trajectory assessment. While the vehicle designs were gradually converging to a final consolidated
geometry, also the quick engineering tools related to aerodynamics and propulsion performance estimation for
hypersonic vehicles were gradually replaced towards full three-dimensional CFD simulations.

Integrated design of airframe and engine throughout the whole trajectory is a must to guarantee the performance of
a preliminary hypersonic airbreathing vehicle design. This means that both off- and on-design conditions for the
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

aerodynamics and the different parts of a combined cycle need to be well understood, modelled and validated. A
turbo-based engine is used to assure better performance and fuel consumption during acceleration. Above Mach 4, a
dual mode ramjet takes over the last part of the acceleration phase before switching to scramjet mode for the Mach
8 cruise. Important points to be addressed to realize these goals are:
a. Proper development and validation of engine-airframe integration tools and methodology.
Engineering tools for a quick assessment of vehicle performance are required allowing for swift conceptual
vehicle changes during the first design iterations. Various levels of detail in aerodynamics and propulsion
performance simulation can be used at this stage. However, the complex interaction among the different
components and disciplines (Fig. 1) should be optimally addressed for a fast and efficient vehicle design and
assessment. A key element in this approach is the proper definition of interfaces and control volumes of the
different elements, in particular for the external and internal flow paths. Proper validation and cross-
comparison are needed on existing vehicles supported by properly chosen experiments within the project.
b. High-Speed Airbreathing Cycle Analysis
Closely linked to engine-airframe integration tools is the correct prediction of engine performance throughout
its operational domain. Several new and particular cycles need to be implemented and investigated such as
several variants on the air-turbo rocket/ramjet engine. Though a dual mode ramjet cycle was investigated in
LAPCAT I, amongst other further level of detail on the chemistry, combustion efficiency is needed, in
particular for operation in off-design conditions. These cycle models form a basic module into the above
described engine-airframe integration tool.

Fig. 1: Hypersonic Vehicle Design System (McClinton 2007)

2
American Institute of Aeronautics and Astronautics
c. Dedicated experiments to evaluate the design in various operation points.
Validation of the integration tools will be based on experiments dedicated on both individual components such
as intakes and combustors, but also on integrated models for the internal flowpath with and without
combustion as well as a complete model with external and internal flow aerodynamics and interaction. This
means that the wind tunnel models will be equipped with both detailed diagnostics as well as integrated force
and/or moment balances. Also scaling issues will be addressed experimentally in order to extrapolate the
validity of the tools to the design of larger vehicles, not able to be tested in available ground based facilities.
Though intended within LAPCAT II, this experimental part has yet to start and hence will not be addressed
here.

This paper will rather focus on the performance analysis based on one-dimensional propulsion modelling and CFD
analysis to perform an assessment on the thrust minus drag balance on an airbreathing vehicle proposed by ESA-
ESTEC. Also, the combustion chamber with the actual injection concept has been simulated by CIRA. The overall
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

concept is based on a dorsal mounted engine in combination with a volumetrically efficient waverider.
Methodologies involved in the design of LAPCAT-MR1 were described previously (Murray, Steelant and Mack
2008). Within LAPCAT II, different evolutions of this concept were proposed but they all have an inward turning,
highly 3D intake in combination with a highly integrated nozzle. This is in contrast with a simple 2D intake and a
SERN-based nozzle of the MR1 design. This fully integrated vehicle is known as the LAPCAT-MR2.4 aiming for a
passenger capacity of 300 and range of about 18,000 km. The fuel used is liquid hydrogen with a combined
propulsion unit comprising of an Air Turbo Rocket (ATR) and a Dual Mode Ramjet (DMR).

II. LAPCAT MR2.4 Vehicle Description

The LAPCAT MR2.4 conceptual design is based upon a waverider with a dorsal mounted engine. The intake is
constructed using streamtracing methods from an axisymmetric inward turning compression surface. The air capture
cross section is elliptical with a ratio of semi-major to semi-minor axes of 3. It feeds a dual mode ramjet/scramjet
combustion chamber and is foreseen to operate between M=4.5 and M=8. Below M=4.5 an accelerator engine is
required. This is incorporated into the intake by sliding doors and including a turbine based engine flow-path in the
vehicle body. The analysis of this flowpath is not part of the present study. The nozzle is constructed in two parts:
the first isentropic 2D nozzle has an inviscid area ratio of three, thus bringing the elliptical combustor cross section
to a circular cross-section and does not expand in lateral direction. The second nozzle itself was streamtraced from
an axisymmetric isentropic expansion and truncated to a suitable length. Both nozzles were designed for cruise
conditions. The final vehicle is shown below in Fig. 2. The nozzle geometries are depicted in Fig. 3.

Fig. 2: Final TBCC Mach 8 MR2 Concept

3
American Institute of Aeronautics and Astronautics
III. Nose-to-Tail Assessment via Computational Fluid Dynamics

A. Numerical Modeling

For the nose-to-tail simulation of the full-scale vehicle several CFD codes were used. At ESTEC the hybrid
structured/unstructured DLR-Navier-Stokes solver TAU, which is validated for a wide range of steady and unsteady
subsonic, transonic and hypersonic flow cases (http://tau.dlr.de/), is used. The TAU code is a second order finite-
volume flow solver for the Euler and Navier-Stokes equations in the integral form using eddy-viscosity, Reynolds-
stress or detached- and large eddy simulation for turbulence modeling. For the present investigations, the Spalart-
Allmaras one-equation eddy viscosity model is used. The AUSMDV flux splitting scheme is applied together with
MUSCL gradient reconstruction to achieve second order spatial accuracy. The combustion models in the DLR Tau
code is based on detailed chemical kinetics. The flow is considered to be a reacting mixture of thermally perfect
gases. A transport equation is solved for the partial density of each individual species. The chemical source term in
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

this set of transport equations is computed from the law of mass action by summation over all participating
reactions. The forward reaction rate is computed using the modified Arrhenius law and the backward rate is obtained
from the equilibrium constant which is directly derived from partition functions. A modified Jachimowski reaction
mechanism for hydrogen-air mixtures (Gerlinger, Möbus and Brüggemann 2001) was applied for this investigation.
This mechanism includes hydrogen peroxide (H2O2) and the perhydroxyl radical (HO2) and is valid over a wide
range of pressures, densities and equivalence ratios. The applicability of this numerical approach for accurate
simulation of supersonic combustion phenomena was successfully validated on the basis of representative reference
configurations such as the HyShot II Scramjet (Hannemann, et al. 2010), (Karl 2011). The initial mesh for the nose-
to-tail simulations had a size of approximately nine million nodes. After several gradient based refinement steps to
better resolve for example shock structures, this number grew to approximately 14 million mesh points.
At CIRA the numerical code used to carry out the aerothermodynamic analysis of the LAPCAT-MR2 vehicle is the
parallel code SPARK (formerly C3NSDB) that solves, on a multi-block structured grid, the Reynolds-Averaged-
Navier-Stokes equations in a density-based finite volume approach with a cell centred, Flux Difference Splitting
second order ENO-like upwind scheme for the convective terms (Battista, Cutrone and Ranuzzi 2008). The code has
been run on the a DELL cluster of 128 cores (16 nodes with two Intel XEON 5560 2.8GHz each, with an aggregated
memory of 48 GB per node, 1.4 TFLOPS aggregated). Particular features of the code are, in addition to what already
explained above, the multilevel approach allowing a grid sequencing and finally the multi-block/multi-inlet enabling
the handling of particular initial/boundary conditions. The physical modelling available inside the numerical code
includes non-viscous (Eulerian) fluid flow and viscous laminar/transitional/turbulent flow with the work fluid being
air modelled as an ideal gas or in both thermo-chemical non equilibrium and equilibrium, or a general multi-
component reacting mixtures (Arrhenius formulation). In particular, for the present simulations Jachimowski air-
hydrogen chemical schemes have been used including the Zel’dovich mechanisms for NOx formation (Jachimowski
1998). Several turbulence models are available inside the SPARK code for eddy viscosity calculation: the one-
equation Spalart-Allmaras model both in incompressible (Spalart and Allmaras 1994) and compressible (Forsythe,
Hoffmann and Squires 2002) version and the two-equation k-ε turbulence model. For all the models the laminar-to-
turbulence transition is imposed across surface lines (i.e. a transition front). Different versions of the two-equation k-
ε model are available: standard, RNG along with compressibility effects correction for high speed turbulent flows
simulations (Grasso, et al. 2001).
In addition to the CIRA code, the commercial code Fluent®13 (Ansys Inc. 2011) has been also used in order to
perform simulation on unstructured grids that are necessary for the full combustor configuration. The same
Jachimowski chemical schemes written in Chemkin format, and used within the SPARK code, have been imported
into Fluent®13. This allowed us to have a direct code-to-code comparison for the simulation conducted on simpler
geometries where structured grids are feasible.

B. Boundary Conditions

It was decided to perform the nose-to-tail simulations at two different trajectory points. The free stream conditions at
a Mach number of 8 and 6 are given in Tab 1. The parameters were set as a farfield boundary condition in the CFD
solver. All the external walls and the intake wall were assumed to be in radiative thermal equilibrium with isotropic
spectral emissivity of ε = 0.8. The combustion chamber wall and nozzle walls were set to constant temperature at
1000 K (see Fig. 3).

4
American Institute of Aeronautics and Astronautics
Table 1: Investigated free stream conditions

M [-] T [K] p [Pa] ρ [kg/m3] u [m/s] q [Pa] ReL=1·10-6[-]


8 222.7 845.1 0.01322 2,392.8 37,845.5 2.18
6 222.7 2,152.7 0.03369 1,794.6 54,250.8 4.16

The injector system of the LAPCAT MR2 flight concept is a V-shaped array of full-strut fuel injectors as depicted in
Fig. 4 through Fig. 12. Every strut is equipped with a multitude of injection holes. In order to model the actual
injection, mixing and combustion process, these features need to be resolved by the numerical grid. This requires a
very high numbers of mesh points. Together with the thin boundary layers which have to be resolved by prismatic
layers on the outer surfaces of the vehicle and the required spatial resolution of the full scale concept, this would
lead to a mesh size which is currently unfeasible to be simulated with the available hardware.
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

Therefore, all nose-to-tail simulations in this study were performed with a loosely coupled approach between CFD
simulations using the TAU code and a one-dimensional supersonic combustion tool. This tool has been developed to
compare and predict experimental campaigns in the field of Scramjet combustion on the lab scale and has been
verified against own experiment and data available from literature (Scheuermann, Chun and von Wolfersdorf 2008).
The tool is able to simulate the wall heat losses and skin friction momentum losses as well as staged injections
schemes. A separate CFD study on the fully 3D combustion chamber is given in Section IV, isolated from the rest of
the vehicle concept.
The inlet conditions for the 1D tool are extracted from an averaging plane (see Fig. 3) and the results at a
downstream location with respect to the averaging plance of x = 5.64 m (nozzle inlet plane, see Fig. 3) are extracted
from the 1D tool. Here, static pressure, temperature, velocity and the chemical specie’s mass fractions are given.
The combustion chamber outlet plane is used in the CFD simulations for having the flow exit the numerical domain.

Wall boundary conditions Combustor averaging and outlet plane;


nozzle inlet plane.
Fig. 3: Boundary conditions of the external and internal flowpath

Fig. 4: Top View of injector array, (Vellaramkalayil, et al. 2012)

5
American Institute of Aeronautics and Astronautics
First, the nose-to-tail simulations were carried out in fuel-off conditions. This means that the conditions at the outlet
plane were simply transferred to the nozzle inlet plane. Once a converged solution was reached, the flow properties
at the averaging plane were extracted following the stream-thrust averaging technique (Baurle and Gaffney 2008),
which ensures the conservation of mass, momentum, and total enthalpy during the averaging of the CFD data. These
values are given in Table 2 a) for the M=8 freestream condition and were used as a boundary condition for the one-
dimensional modelling of the scramjet combustor. In Fig. 5 the Mach number and static pressure distribution within
the engine are given for both investigated trajectory points for an angle of attack of zero. The x-coordinate is relative
to the averaging plane presented in Fig. 3. The fuel is entering the domain at x=0.274 m (leading edge first strut in
the V-shape injector array) and it is being injected at the actual location of the struts in the full-scale concept. For the
M=8 condition, finite rate chemistry was used leading to an ignition delay of approximately 1 m. At the end of the
combustion chamber the combustion processes are complete. The values to be used at the inlet plane of the nozzle in
the CFD studies are extracted at x=5.64 m and are given in Table 2. For the trajectory point at M=6 equilibrium
chemistry is used because the average temperature at the averaging plane is below the auto-ignition temperature of
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

hydrogen. In reality, hot pockets are present in the highly 3D dominated flowfield, which are well above the auto-
ignition temperature ensuring the mixture will ignite. Classically the combustion heat process is then gradually
released via a mixing function. In the combustion tool, this is done in a gradual exposure of the fuel with the air, by
adding a certain portion of the injected fuel per spatial step size into the reaction mechanism. This can be seen by the
pressure rise starting immediately at the injector location as opposed to the results for the M = 8 case and the gradual
increase of pressure related to the prescribed fuel mixing. For the investigated cases, the set mixing efficiency will
practically equal the combustion efficiency for both the finite rate chemistry and equilibrium chemistry formulation
due to the long combustion chamber. This also shows the limitation of the 1D-modelling approach: in reality,
complex shock structures and recirculation zones (hot pockets) are available in the combustion chamber, which will
aid to ignite the injected fuel.
For the case at M=8 an equivalence ratio of 0.65 was chosen whereas for the lower trajectory point at M=6 it was set
to 0.7. A larger value for the latter was chosen to anticipate on the needed acceleration up to cruise speed. The
combustion efficiency, which controlled by the mixing efficiency function in the 1D-tool, was limited to 85 %, i.e.
15% fuel remains unburnt independent of the mixture efficiency This value seemed feasible after three-dimensional
simulation of the combustion chamber (Vellaramkalayil, et al. 2012). For the M=6 trajectory point the combination
of the equivalence ratio and high combustion efficiency typically lead to a thermal choking. While the equivalence
ratio of 0.7 was kept constant, the combustion efficiency, again controlled by the mixing efficiency function, was
reduced until the Mach number within the domain stayed above unity. The unburnt hydrogen was included into the
mass-fractions at the nozzle inlet plane, so that the remaining fuel was able to burn in the last part of the combustion
chamber and within the first part of the nozzle at a finite rate in the CFD simulatons.

Table 2: Stream-thrust averaged conditions

M=8 Combustor Inlet (fuel off / fuel on) M=8 Combustor Outlet
AoA = 0 AoA = -2 AoA = 2 AoA = 0 AoA = -2 AoA = 2
m [kg/s] 1,193.3 1,327.2 973.2 m [kg/s] 1,215.9 1,352.3 991.6
p [bar] 0.271 0.297 0.214 p [bar] 1.201 1.316 0.978
T [K] 815.2 805.3 787.5 T [K] 2,510.2 2,509.1 2,522.9
u [m/s] 2,111.1 2,132.3 2,124.3 u [m/s] 1,719.8 1745.0 1,732.7
γ [-] 1.353 1.354 1.356 γ [-] 1 262 1.262 1.262
R [J/kg/K] 288.3 288.3 288.3 R [J/kg/K] 329.5 329.4 329.9

a) Combustor inlet (averaging plane) b) nozzle inlet

6
American Institute of Aeronautics and Astronautics
4
3
2.5
3.5
M=8, AoA = 0 M=6, AoA = 0
2.5
2
3

p [bar]
p [bar]

M [-]

M [-]
p [bar]
M [-] 1.5 M [-] 2
2.5

1 1.5
2

0.5
1.5 1
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

4 6 8 0 2 4 6 8
x [m] x [m]
Fig. 5: Pressure and Mach number distribution with the one-dimensionally modelled scramjet combustor.
Left: M=8, finite rate chemistry, right: M=6, equilibrium chemistry.

C. Results

In this section information of the flow field around the LAPCAT MR2.4 vehicle concept is given followed by a
summary of the aerodynamic performance. Data is available from simulations performed by ESTEC as well as from
CIRA. The Mach number contour lines and contour for the trajectory point at M = 8 are given in Fig. 6 and Fig. 7
respectively. First, one can see the leading edge shock is aligning well with the wings of the vehicle outlining the
successful waverider design for cruise conditions. Additionally, the shock originating from the first compression
ramp of the intake is captured just under the intake’s lip, which leads to zero intake mass-flow rate spillage (see Fig.
10 b). These two features are the most important influencing factors for the high aerodynamic performance of this
concept. A contour plot of static pressure is given in Fig. 7. First, the high static pressure region at the shock
impingement location of the intake cowl caused by a recirculation zone can be seen. This shock is reflected and
impinges again on the bottom wall of the combustion chamber. In the vehicle concept, the injectors would be
located at this location. The presence of the compression shock will aid in the ignition of the fuel-air mixture.

Fig. 6: Contour lines of Mach number for M=8 freestream conditions and zero angle of attack.

Fig. 7: M=8 flowfield from NtT computations: Mach contours (left) and pressure field (right).

7
American Institute of Aeronautics and Astronautics
Fig. 8: M=6 flowfield from NtT computations: Mach contours (left) and pressure field (right).
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

The results for the trajectory point at M = 6 are given in Fig. 8. First, one notices how the shock on the intake is not
any longer captured under its cowl due to the higher wave angle at the lower free stream Mach number. This also
has an influence on the size and strength of the recirculation zone under the cowl. More importantly, one can see a
low Mach number zone above the intake cowl, which depicts the off-design spillage mass-flow rate originating from
the gap between the intake cowl leading edge and the first intake shock.
The definition of the drag coefficient CD, the lift coefficient CL, and the pitching moment coefficient CM are given in
Eqn. 1 through Eqn. 3. The dynamic pressure q is computed from the free stream and is given in Table 1. The
reference area Aref has been in all cases set to one m2 whilst the reference length for CM is Lref = 1 m. The origin in
the reference frame is located at the leading edge on the symmetry plane of the vehicle, which is also the reference
point for the calculation of the pitching moment. The coefficients include both pressure and viscous components. A
negative drag coefficient indicates thrust in the given co-ordinate system. The drag D, the lift L, and the pitching
moment M were extracted from the CFD simulations. The actual planform area is 2365m2.

D
CD = (1)
Aref q
L
CL = (2)
Aref q
M
CM = (3)
Aref Lref q

Fig. 9: Lift coefficient, drag coefficient and moment coefficient for both M=8 freestream conditions (left) and
M=6 freestream conditions (right)

8
American Institute of Aeronautics and Astronautics
All three coefficients have been plotted in Fig. 9 for both cases at M=6 and M=8. The dotted line gives the
distribution of the pitching moment coefficient, the solid line the drag coefficient, and the dashed line the lift
coefficient, all as a function of the angle of attack. First, one can notice the effect of the angle of attack on the
waveriders lift. Changing the angle by plus or minus two degrees basically doubles or halves the vehicle’s lift,
which is almost a linear function. When delivering higher lift, also higher vehicle drag is being generated. This
becomes too high to be overcome by the propulsion system at the given equivalence ratio (positive CD). For negative
angle of attack the vehicle concept is predicted to be able to provide enough thrust to accelerate. At cruise conditions
the T=D balance is achieved for a slightly negative angle of attack (-2<AoA<0). In red the results of CIRA’s
simulation have been plotted in the same figure. The results compare very well to ESTEC’s computations.
The values for total lift and thrust or drag are given in Table 3. One can see that for the trajectory point at M = 6 the
vehicle with a foreseen take-off mass of 400 t is well able to provide normal and axial acceleration. At the beginning
of the cruise (trajectory point M=8) at zero angle of attack the lift will balance approximately the weight of the
vehicle. Any remaining drag component can be easily compensated, for example, by a slightly higher equivalence
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

ratio.

Table 3: Summary of result from NtT computations. (‘ext’ = external; ‘int’ = internal flowpath, propulsion
unit; ‘p’ = pressure; ‘visc’ = viscous)

M [-] AoA [deg] regime ER [-] L [kN] D [kN] Dext [kN] Dint [kN] Dp [kN] Dvisc [kN]
8 0 turbulent 0 3,453 643 467 176 337 306
8 -2 turbulent 0.65 1,493 -88 308 -396 -468 379
8 0 turbulent 0.65 3,245 51 467 -416 -318 369
ESTEC

8 2 turbulent 0.65 5,267 492 790 -298 96 394


6 0 turbulent 0 5,533 1009 769 240 610 399
6 -2 turbulent 0.70 2,095 -489 512 -1001 -1,008 540
6 0 turbulent 0.70 5,330 -237 767 -1004 -733 496
6 2 turbulent 0.70 8,747 945 1274 -329 443 525
8 0 laminar 0.6 3767 -265 334 -598 -327 62
CIRA

8 0 turb k-ε 0.6 3355 -167 365 -535 -347 180


8 0 turb SA 0.6 3405 116 577 -461 -365 481
8 0 turb SA0 0.6 3372 44 511 -467 -364 408

M=8 AoA = -2 AoA = 0 AoA = 2


m [kg/s] 1,327.2 1,193.3 973.2
ηin ESTEC [-] 0.88 1.00 1.11
ηin [-] CIRA - 0.996 -

M=6 AoA = -2 AoA = 0 AoA = 2


m [kg/s] 2,457.5 2,125.9 1,804.3
ηin [-] 0.85 0.93 1.08

a) Aerodynamic Efficiency b) Intake Efficiency

Fig. 10: Vehicle performance at M=8 and M=6

From the above given Table 3one is also able to compute the net-thrust of the propulsion unit and its fuel specific
impulse by looking only at installed thrust, which in our case is also the internal drag, here -Dint=T = 416 kN. At an

9
American Institute of Aeronautics and Astronautics
equivalence ratio of 0.65 and the given free-stream condition the engine consumes m f = 22.59kg / s of hydrogen as
a fuel. This means a specific impulse at the cruise trajectory point of I sp = T /( gm f ) = 1877 s. For the M=6
trajectory point at an ER=0.7, we obtain I sp = 2361s.
To further evaluate the performance of the vehicle, the aerodynamic efficiency and the intake efficiency for both
simulated trajectory points are shown in Fig. 10. The aerodynamic efficiency is defined as the ratio of lift over drag.
To calculate this quantity, as opposed to Table 3, only the external surfaces and not the surfaces of the internal flow
path (intake, combustion chamber, nozzle) have been used. For both trajectory points, L/D reaches its maximum at
the design point of zero angle of attack as expected. It decreases significantly when pitching down the vehicle as
leeward side of wings and fuselage are tilted into the freestream creating a downward lift. If needed, this can be
easily alleviated by adapting the relative position of engine with respect to wings and fuselage. The same is valid for
the curve at M=6, which is slightly higher than for M=8. This is caused by the higher unit Reynolds number at this
trajectory point (see Table 1) leading to relatively lower viscous drag on both wings and fuselage.
The intake efficiency ηin has been computed for all investigated trajectory points and angle of attack. It is defined as
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

the ratio of the actual swallowed mass-flow rate versus the mass-flow rate, which can theoretically be captured. Last
is based upon the frontal, freestream elliptical surface of the intake and the free stream conditions. At an angle of
attack this mass-flow rate needs to be corrected, because the intake height and therefore the frontal surface changes.
This follows the geometrical relationship of a rotation around the leading edge by the angle of attack:

 ref = m
m  ref , AoA=0
1
(H in cos( AoA) − Lin sin ( AoA)) (4)
H in
Here, Hin is the height of the intake at zero angle of attack and Lin the length of the intake from the leading edge up
to the beginning of the combustion chamber. For negative angle of attack, the intake is turned into the freestream
meaning higher capture area for the mass-flow rate. For positive angle of attack, the opposite happens.
First, one can see from Fig. 10 that at zero angle of attack the intake works perfectly fine for the on-design condition
(M=8) with an intake efficiency of unity, for the simulations by ESTEC as well as CIRA. At M=6 seven percent of
the captured mass-flow rate is spilled. This value is comparatively low, because this high-speed intake has been
optimized for off-design conditions by enlarging the cowl, which was pulled further upstream. This ensures a high
mass-capture, even though the shock is not hitting the lip anymore. In Fig. 7 for the on-design condition one can see
how the first intake shock impinges well under the top surface, and it is not exactly shock-on-lip. The intake
performance at negative angle of attack is showing an efficiency lower than at zero angle of attack. This is because
the turning angle, and therefore the wave angle of the main intake shock, is higher. Also, on the flat surface at the
bottom of the air intake, which is now inclined with respect to the flow, another shock is created. Together, they
cause some mass-flow rate spillage, leading to the decrease in intake efficiency. In contrast, the intake is able to
capture more mass-flow rate as compared to the reference mass-flow rate (which is basically only based on a single
geometrical assumption) at positive angle of attack. Here, the enlarged cowl area allows to capture more freestream
streamlines.

Table 4: Summary of results from turbulence models sensitivity analyses performed by CIRA. (‘ext’ =
external; ‘int’ = internal flowpath, propulsion unit; ‘intake’ = engine’s intake; ‘cc’ = combustion chamber;
‘nozzle’ = engine’s nozzle)

regime L [kN] D [kN] Dext [kN] Dint [kN] Dintake [kN] Dcc [kN] Dnozzle [kN] L/D [-]
LAM 3412.0 -263.7 334.3 -598.0 176.7 14.1 - 10.21
TURB k-e 3352.0 -169.3 365.3 -534.5 202.2 33.4 -770.2 9.18
TURB-SA 3405.3 116.0 576.7 -460.7 233.3 43.8 -737.8 5.91
TURB-SA0 3372.3 44.0 511.2 -467.2 224.3 42.3 -733.9 6.60
LAM -327.2 300.9 -628.1 170.1 8.7 -806.9
Pressure
Drag

TURB k-e -347.2 301.6 -648.7 177.0 7.1 -832.8


TURB-SA -364.7 315.5 -680.2 185.1 4.7 -870.1
TURB-SA0 -364.2 310.2 -674.5 182.9 5.4 -862.8
LAM 63.6 33.4 30.1 6.6 5.5 18.1
Viscous
Drag

TURB k-e 177.9 63.7 114.2 25.3 26.3 62.6


TURB-SA 480.7 261.1 219.6 48.2 39.1 132.2
TURB-SA0 408.2 201.0 207.2 41.4 36.9 128.9

10
American Institute of Aeronautics and Astronautics
In order to have a code-code comparison the nose-to-tail simulations have also been performed by CIRA. In
particular, the M=8 cruise conditions and several flow modeling hypotheses (laminar and turbulence models) have
been considered. The results are reported in Table 3 and have been highlighted in bold font. A good comparison can
be found for what concerns lift (4.7% difference). A larger difference occurs for both the pressure and viscous drag
which both amounts to some 19% in the external drag for the same turbulence model (TURB SA in Table 3) .
A similar difference is noted for the net thrust which is the difference between engine thrust and external drag. In
fact this is the aero-propulsive balance and is a critical point because at cruise conditions (constant velocity) must be
equal to zero. The CIRA turbulent cases have been conducted using three different turbulence models. From the
results reported in Table 3, one can see that the best agreement for the M=8 is obtained with the compressible
Spalart-Allmaras turbulence model (Forsythe, Hoffmann and Squires 2002). All ESTEC nose-to-tail simulations
have been performed by using the original version of the Spalart-Allmaras turbulence model (see Section III).
The main results of the this turbulence sensitivity analysis can be found in Table 4. It reports in particular the
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

splitting of the total drag into pressure/wave part and friction part for the laminar and the three turbulent cases.
From the analysis we can deduce that the pressure/wave drag is quite the same for the laminar and the three
turbulent cases, the external turbulent friction drag and the aerodynamic efficiency vary from laminar to turbulent
cases and also from the k-ε turbulence model to Spalart-Allmaras one. It is worthy to note as this strong dependence
of the friction drag on the turbulence model strongly affects the aero-propulsive balance.

IV. Full 3D Combustor Analysis

The limitation of the above Nose-to-Tail computations is driven by hypotheses of the 1D combustion tool. The
mixing or combustion efficiencies need to be prescribed relying on best practice. To assure this can be achieved, a
3D injector layout was designed and optimized for best combustion efficiency. A full 3D CFD analysis was carried
out and is discussed in present paper. Assumptions for the 1D analysis will be compared with the full combustor
analysis, this with the goal to assess the reliability of the 1D combustion hypothesis.
The combustor in standalone configuration has been considered. Fig. 4 shows the combustion chamber from the top
with the whole struts array arrangement and the orientation towards flight direction. Fig. 11 shows a more detailed
geometry used to generate mesh-grid. As it can be noticed, inlet section has been advanced by one meter for solving
meshing issues and inlet conditions have been extracted from nose-to-tail simulations at M=8, AoA=0, averaging the
flow properties at this section plane.

Fig. 11: extracted CAD of combustor with surfaces description

Full geometry (with port hole as injector) leads to a very expensive mesh-grid. Moreover, it is not feasible to create
a multi-block structured mesh. For these reasons, a simpler geometry has been derived for preliminary analyses (see

11
American Institute of Aeronautics and Astronautics
Fig. 12). Essentially a slice substitutes the array of port holes for each side of the struts. Slices areas have been
calculated keeping propellant mass flow characteristics as nominal. Starting from this simplified geometry, a 2D
domain has been also considered extracting a slice from an intermediate z-wise plane (see Fig. 12). The consequent
2D mesh-grid, characterized by a lower number of cells, was adopted to perform simulations to compare physical
models and to define a convergence strategy. While the 3D “simplified” simulations (from 3D simplified geometry)
have been conducted to obtain order of magnitude for quantities in outlet section and to preliminary investigations
on EINO.
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

Fig. 12: modified geometry by sliced struts (with detail on struts) (on left) and extracted 2D section (on right)

A 3D full geometry simulation (with port holes) was also carried out (see Fig. 13). While 2D and 3D simplified
geometries have been modelled as structured mesh-grids, it was possible to model the 3D full geometry only as an
unstructured mesh. For these last simulations, the well-known commercial code Fluent®13 described in section III
has been used.

Fig. 13: Full geometry mesh with detail on strut leading and trailing edge and port holes

12
American Institute of Aeronautics and Astronautics
Table 5: Mesh characteristics comparison

solver topology cells blocks interior faces


2D Fluent unstructured (from structured) 63,344 96 125,484
2D SPARK structured 71,408 96 -
3D-Simple SPARK structured 2,469,952 290 -
3D-Full Fluent unstructured 10,897,878 - 25,033,427

Table 6: Comparison between quantities averaged on outlet section for different runs
Solver M [-] p [bar] ρ [kg/m3] T [K] ptot [bar] ξH2 [-] ξO2 [-] ξH2O [-] ξNO [-] EINO [-] ηcomb [-]
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

2D Fluent® 2.07 1.1832 0.142 2,586.6 10.23 6.193E-04 0.084 0.141 0.00624 377 0.964
2D SPARK 1.93 1.1125 0.142 2,522.1 8.12 5.421E-04 0.123 0.100 0.00750 451 0.968
3D-Simple SPARK 1.88 1.1573 0.144 2,561.3 7.78 7.055E-04 0.118 0.103 0.00899 546 0.959
3D-Full Fluent® 1.99 1.0679 0.126 2,625.0 8.24 6.245E-04 0.084 0.139 0.00790 477 0.964
ESTEC 1D 1.68 1.1898 0.145 2,510.2 5.52 2.908E-03 0.095 0.133 n/a n/a 0.850

In Table 5 the differences between the different meshes adopted can be appreciated. As it can be seen, a 3D full
detailed geometry (with port holes) requires a huge number of cells (almost one order of magnitude greater than
simpler one – by slices). In Table 6 it is possible to compare results calculated as area-weighted averages at the
outlet section of combustor. It is possible to note the high value of the combustion efficiency at outlet section for all
simulations that seems justify the design choices of the struts shape and distributions along the combustion chamber.
Furthermore, the values obtained with the one-dimensional tool are relatively close to the ones obtained from the 2D
Fluent simulation, but do not reach the same combustion efficiency because this parameter was an input for the
simulations. This also explains the comparatively high mass-fraction of hydrogen. Furthermore, for the simulation
performed by ESTEC no Nitrogen was included in the reaction scheme, both within the 1D-tool as well as for the
CFD simulations. Another important predicted result is the value of EINO. Even though there is a significant spread
of results, all the simulations give the same order of magnitude. A description of combustor predicted flow features
is given in the following.

Fig. 14: Static pressure plotted on 2D mesh-grid by SPARK code

Fig. 15: Mass fraction of H2O (main reaction product) on 2D mesh-grid by SPARK code

13
American Institute of Aeronautics and Astronautics
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

Fig. 16: Pressure iso-surfaces in volume and temperature contours on outlet surface

Due to struts supersonic interactions, in Fig. 14 a pressure waves coalescence can be observed in 2D simplified
simulation carried out by SPARK code. This pattern creates an over-pressure zone near symmetry plane (y=0) that
reflects towards chamber wall creating another overpressure zone. In Fig. 15 a stratification of H2O fraction can be
also noticed. This behaviour has been found also in 3D simplified geometry (Fig. 16), since slices, adopted instead
of port holes, are not enough efficient to mix reactants. In Fig. 16 a detail on block located downstream of the struts
is highlighted. A full three-dimensional reflection pattern can be observed. Moreover, an outlet stratification of
temperature distribution can be also seen.
From Fig. 17 to Fig. 20 results of full detailed 3D mesh-grid (full geometry with port holes) are reported. In Fig. 17
a pressure pattern that impinges on symmetry plane and reflects on chamber wall (such as in Fig. 14) can be
observed. This distribution can be noticed also in temperature distribution as a consequence of this
aerothermodynamic pattern (Fig. 18), and there is a production of NO (Fig. 19) that persists also after temperature
lowering (downstream the reflected waves), reaching its higher concentration near outlet section. In Fig. 20, where
H2O mass fraction distribution is reported, it can be seen how port holes are the better choice for mixing reactants
and energizing products (if compared to slices solution, see Fig. 15), i.e. the stratification has disappeared.

Fig. 17: Static pressure plotted on middle section of 3D full detailed mesh-grid

14
American Institute of Aeronautics and Astronautics
Fig. 18: Static temperature plotted on middle section of 3D full detailed mesh-grid
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

Fig. 19: NO mass fraction plotted on middle section of 3D full detailed mesh-grid

Fig. 20: H2O mass fraction plotted on middle section of 3D full detailed mesh-grid

V. Conclusions

The focus of this study was the aerodynamic assessment of a hypersonic cruiser concept and the performance
assessment of its Scramjet combustion chamber using CFD. First, nose-to-tail simulations were carried out using a
loosely coupled approach for the propulsion unit. A one-dimensional supersonic combustion tool was used to
simulation the processes within the engine. The results obtained here were fed back into the CFD simulations, which
were carried out for on-design and off-design trajectory points and angle of attack.
It could be shown, that the current concept will be able to accelerate at a Mach number of 6 and be able to sustain its
cruise, by choosing a slightly higher equivalence ratio for the engine as simulated here. Also, if the beginning of
cruise weight allows, pitching the vehicle slightly down will allow sustained hypersonic flight. Both studies carried
out by ESTEC and CIRA underline this.
Furthermore, a detailed simulation of the combustion chamber processes was carried out by CIRA. Here, the
complete injector array with its multitude of injection holes was resolved by the numerical mesh. It could be shown
that the injection concept exhibits almost no ignition delay. Also, due to the length of the combustion chamber the
chemical reactions and the mixing processes are nearly completed at the outlet of the domain, leading to a very high
combustion efficiency. Also, the comparison to the one-dimensional approach used for the loose propulsion
coupling is very reasonable.

15
American Institute of Aeronautics and Astronautics
VI. Acknowledgments

The development part of the work was performed within the ‘Long-Term Advanced Propulsion Concepts and
Technologies II’ project investigating high-speed transport. LAPCAT II, coordinated by ESA-ESTEC, is supported
by the EU within the 7th Framework Programme Theme 7 TRANSPORT, Contract no.: ACP7-GA-2008-21 1485.
Further info on LAPCAT II can be found on http://www.esa.int/techresources/lapcat_II.

VII. References

Ansys Inc. Ansys Fluent User’s and Theory Guides, release 13.0. Canonsburg, PA, 2011.
Battista, F. , L. Cutrone, and G. Ranuzzi. “C3NS Solver for Supersonic Combustion Simulations. CIRA LAPCAT
Downloaded by PENNSYLVANIA STATE UNIVERSITY on December 5, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-5872

Report.” CIRA-CF-08-0484, 2008.


Baurle, R.A., and R.L. Gaffney. “Extraction of One-Dimensional Flow Properties from.” Journal of Propulsion and
Power, 2008: 704-714.
Forsythe, J.R. , K.A. Hoffmann, and K.D. Squires. “Detached-Eddy Simulation with Compressibility Corrections
Applied to a Supersonic Axisymmetric Base Flow.” AIAA 2002-0586, 2002.
Gerlinger, P., H. Möbus, and D. Brüggemann. “An Implicit Multigrid Method for Turbulent Combustion.” Journal
of Computational Physics, Vol. 167, 2001: 247-276.
Grasso, F., and D. Falconi. “Shock-Wave/Turbulent Boundary-Layer Interactions in Nonequilibrium Flows.” AIAA
Journal, Vol. 39, No. 11, 2131-2140, 2001.
Hannemann, K., S. Karl., J. Martinez Scharmm, and J. Steelant. “Methodology of a Combined Ground Based
Testing and Numerical Modelling Analysis of Supersonic Combustion Flow Paths.” Shock Waves, Vol. 20,
No. 5 (Springer), 2010: 353-366.
http://tau.dlr.de/. Homepage of the DLR TAU Code, DLR Institute of Aerodynamics and Flow Technology. 2012.
Jachimowski, C.J. “An Analytical Study of Hydrogen-Air Reaction Mechanism with Application to Scramjet
Combustion.” NASA-TP-2791, 1998.
Karl, S. Numerical Investigation of a Generic Scramjet Configuration. Dissertation, TU Dresden, http://nbn-
resolving.de/urn:nbn:de:bsz:14-qucosa-68695, 2011.
Murray, N., J. Steelant, and A. Mack. “Conceptual Design of a Mach 8 Hypersonic Cruiser with Dorsal Engine.”
Sixth European Symposium on Aerothermodynamics for Space Vehicles. Versailles, France, 2008.
Scheuermann, T., J. Chun, and J. von Wolfersdorf. “One-Dimensional Modelling of a Scramjet Combustor Reacting
Flow.” 15th AIAA International Space Planes and Hypersonic Systems and Technologies Conference.
Dayton, OH: AIAA, 2008.
Spalart, P.R., and S.R. Allmaras. “A One Equation Turbulence Model for Aerodynamic Flows.” La Recherche
Aerospatiale, Vol. 1, 1994: pp. 5-21.
Steelant, J. “Achievements obtained for sustained hypersonic flight within the LAPCAT project.” 15th AIAA
International Space Planes and Hypersonic Systems and Technologies Conference. Dayton, OH: AIAA
2008-2578, 2008.
Steelant, Johan. “Sustained Hypesonic Flight in Europe: Technology Drivers for LAPCATII.” 16th
AIAA/DLR/DGLR International Space Planes and Hypersonic System Technologies Conference. Bremen,
Germany: AIAA 2009-7240, 2009.
Vellaramkalayil, J., T. Langener, J. Steelant, and J. von Wolfersdorf. “Injector Layout Optimization for the
LAPCAT MR2 Mach 8 Cruiser.” AAAF-ESA-CNES Space Propulsion 2012. Bordeaux, France, 2012.

16
American Institute of Aeronautics and Astronautics

You might also like