Professional Documents
Culture Documents
Numerical Simulation Capsule Vehicle PDF
Numerical Simulation Capsule Vehicle PDF
DOI 10.1007/s00193-005-0003-0
O R I G I N A L A RT I C L E
R. C. Mehta
Received: 24 June 2005 / Accepted: 31 August 2005 / Published online: 1 February 2006
C Springer-Verlag 2006
Abstract The flow fields over ARD (ESA’s atmospheric a result the development of high aerodynamic drag which
reentry demonstrator), OREX (orbital reentry experiments) is needed for aerobraking. Highly blunt configurations are
and spherically blunted cone-flare reentry configurations generally used to decelerate spacecraft for safe returning on
are numerically obtained by solving time-dependent, ax- the Earth after performing the experiments. The bow shock
isymmetric, compressible Navier–Stokes equations for wave is detached from the blunt forebody and is having a
freestream Mach numbers range of 1.2–6.0. The fluid dy- mixed subsonic–supersonic region between them. The sur-
namics are discretized in spatial coordinates employing a face pressure distribution, the location of the sonic line and
finite volume approach which reduces the governing equa- the shock stand-off distance on the spherical cap have been
tions to semi discretized ordinary differential equations. analytically calculated at very high speeds with an adiabatic
Temporal integration is performed using the multistage index near to unity which gives a singular point at 60◦ from
Runge–Kutta time-stepping scheme. A local time step is the stagnation point [1, 2]. The analytical approach for the
used to achieve steady-state solution. The numerical simu- high-speed flow-past the blunt-body is considerably difficult
lation is carried out on a structured grid. The flow-field fea- and complex [3]. The flow-field over the reentry module be-
tures around the reentry capsule, such as bow shock wave, comes further complicated due to the presence of corner at
sonic line, expansion fan and recirculating flow in the base the shoulder and the base shell of the capsule.
region are obtained. A good agreement is found between Aeroassist flight experiment (AFE) configurations have
the calculated value of aerodynamic drag coefficient of the been analyzed using two different Navier–Stokes flow
spherically blunted cone/fare reentry configuration with the solvers by Venkatapathy et al. [4]. Aerodynamic heating
experimental data. The effects of geometrical parameters, and pressure along with the forebody and wake-flow struc-
such as radius of the spherical cap, half cone angle, with ture during atmospheric entry of the Mars Pathfinder space-
sharp shoulder edge and with smooth shoulder edge on the craft have been computed by Haas and Venkatapathy [5] us-
flow-field have been numerically investigated for various ing the commerically available general atmospheric simula-
reentry configuration which will be useful for optimization tion program (GASP 2.2). An aerodynamic analysis of the
of the reentry capsule. commercial experiment transport (COMET) reentry capsule
has been carried out by Wood et al. [6] solving the lami-
Keywords Supersonic flows · Computational fluid nar thin layer Navier–Stokes LAURA code for low super-
dynamics · Reentry capsules sonic to hypersonic speeds. The LAURA code is an upwind,
PACS 47.11.Df, 47.40.Ki point implicit, second-order-accurate fluid dynamics solver
based on an extension of the Roe flux-difference-splitting
scheme. Yamomoto et al. [7] carried out flow-field com-
1 Introduction putation over the OREX (orbital reentry experiments) us-
ing computational fluid dynamics approach coupled with the
A high-speed flow-past a blunt-body generates a bow shock thermal response of the heat shield material using finite el-
wave which causes a rather high surface pressure and as ement method in conjunction with the aerodynamic flight
data. Tam [8] has computed flow field at hypersonic Mach
Communicated by K. Takayama numbers over Viking, Biconic and AFE reentry capsules us-
ing IEC3D (inviscid equilibrium computation in three di-
R. C. Mehta (B)
Aerodynamics Division, Vikram Sarabhai Space Centre,
mension). Menne [9] has computed flow field over Viking
Trivandrum 695022, India (Apollo type) and Biconic cases for freestream Mach num-
E-mail: atulm@md4.vsnl.net.in bers 2.0 and 3.0 by solving Euler equations. The inviscid
32 R. C. Mehta
analysis is unable to predict the flow separation in the back forebody region, the fluid decelerates through the bow shock
region of the reentry module. The aerodynamic characteri- wave depending upon the cruise speed and altitude. At the
zation of the CARINA reentry module in the low supersonic shoulder of the capsule, the flow turns and expands rapidly,
Mach regimes has been performed employing numerical and and the boundary layer detaches, forming a free shear layer
experimental approach [10]. The flow-field simulation past that separates the inner recirculating flow region behind the
the ARD (ESA’s atmospheric reentry demonstrator) module base from the outer flow field. The latter is recompressed
has been computed numerically by Walpot [11] at hyper- and turned back to freestream direction, first by the so-called
sonic speeds. Developments relating to base pressure and lip shock, and further downstream by recompression shock.
base heating prediction for wide range of cone and wedge At the end of the recirculating flow past the neck, the shear
have been reviewed for supersonic and hypersonic Mach layer develops in the wake trail. A complex inviscid wave
numbers by Lamb and Oberkampf [12]. Recently, flow- structure often includes a lip shock wave (associates with
field solutions past Beagle-2 spacecraft have been obtained the corner expansion) and wake trail (adjacent to the shear-
by Liever et al. [13] using commerically available CFD- layer confluence). The corner expansion process is a modi-
FASTRAN code for low supersonic to hypersonic speeds. fied Prandtl–Mayer pattern distorted by the presence of the
A numerical simulation code has been used for super or- approaching boundary layer. Figure 1 shows schematic fea-
bital reentry flow and has been applied to the flow-field pre- tures of the flow field over OREX and DART.
diction around the MUSES-C reentry capsule [14]. Super- The sonic line is located on the OREX-type configura-
sonic and hypersonic flow over a slender cone [15] has been tion on the shoulder whereas in the case of DART’s reentry
numerically obtained by solving Navier–Stokes equations module, the location of the sonic line is at the junction of the
using an explicit multi-stage Roe’s scheme. The flow field spherical blunt cone. Thus, it is seen that the flow field over
around blunt reentry capsules [16] were numerically stud- the reentry capsule needs a high drag with good static stabil-
ied in order to understand the mechanism of dynamic in- ity margin which leads to the selection of an axisymmetric
stability of the capsule at freestream Mach number of 1.3. shape of large angle sphere-cone combination.
The blunt and short reentry capsule tends to be dynami- The forebody geometry of the ARD configuration is hav-
cally unstable at low supersonic Mach number attributed ing a spherical cap; the OREX module is having combina-
primarily to the delay in base pressure [17]. Numerical tion of spherical cap with cone. The other capsule consists
studies have been carried out by Ottens [18] using a lam- of spherically blunted cone/fare. These capsules cover most
inar Navier–Stokes flow solver for two different types of of the existing reentry modules. The main aim of the present
delft aerospace reentry test (DART) demonstrators reentry paper is to numerically analyze three types of reentry mod-
modules. ules such as ARD, OREX and DART configurations in order
The above literature survey reveals that the forebody to understand flow-field behavior and its influence on pro-
shape of reentry capsules can be classified either using as files of the surface pressure and skin friction coefficients and
a spherical cap as in the case of Apollo and ARD, or com- also aerodynamic drag for freestream Mach numbers range
bination of the spherical nose with cone as in the case of of 1.2–6.0. The numerical solution of solve-axisymmetric
OREX and Beagle-2, or a spherical blunt-cone/flare config- laminar-compressible time-dependent Navier–Stokes equa-
uration as proposed by DART. The flow-field features over tions is carried out employing a multi-stage Runge–Kutta
the reentry capsule can be delineated through the experimen- time stepping scheme. The numerical scheme is second-
tal and theoretical investigations at high speeds. The signif- order accurate in space and time. A local time step is used
icant flow features are described by the following: In the to obtain a steady-state solution. The computation is carried
Numerical simulation of supersonic flow past reentry capsules 33
To suppress the tendency for odd- and even-point decoupling The spatial discretization described above reduces the gov-
and to prevent the appearance of oscillations in regions con- erning flow equations to semidiscrete ordinary differential
taining severe pressure gradients near shock wave and stag- equations. The integration is performed using an efficient
nation points, the finite volume scheme must be augmented mutistage three-stage Runge–Kutta time-stepping scheme
by the addition of artifical dissipation terms. Therefore, [20]. The time-step advance of each variable W from time
Eq. (4) is replaced by (t) to time (t + t) can be written as
∂Wi, j W(0) = W(t)
Ai, j + Res(Wi, j ) − D(Wi, j ) = 0 (5)
∂t t
where the residual Res(W ) is given by W(1) = W(t) − 0.6 Res(0) − D(0)
A
Res(Wi, j ) = Q(Wi, j ) − V (Wi, j ) + Ai, j Hi, j (6) t
W(1) = W(t) − 0.6 Res(1) − D(0) (12)
where Di, j denotes the dissipative terms which are gener- A
ated by dissipative fluxes. The approach of Jameson et al. t
[20] is adopted to construct the dissipative terms consisting W(1) = W(t) − 0.6 Res(2) − D(0)
A
of a blend of second and fourth differences of the vector
conserved variables Wi, j with coefficients that depend on W(t + t) = W(3)
the local pressure gradient. The dissipation term consists of
the following operators in each direction. In order to minimize the computation time and increase
the stability margin for the dissipative terms, the expensive
D = (Dx + Dr )Wi, j (7) evaluation of the artificial dissipation terms is carried out
The dissipative fluxes in each direction are only at the first intermediate stage (0) and then frozen for the
subsequent stages. A conservative choice of the Courant–
Dx Ui, j = di+ 1 , j − di− 1 , j Friedrichs–Lewy (CFL) number (1.4) is taken to achieve a
2 2
Dr Ui, j = di, j+ 1 − di, j− 1 . stable numerical solution. Local time steps are used to ac-
2 2
clerate convergence to a steady-state solution by setting the
The dissipative flux di+ 1 , j is defined as time step at each point to the maximum value allowed by
2
the local CFL condition. The present numerical algorithm is
( A)i+ 1 , j
(2) (4)
validated with many test cases [21, 22].
di+ 1 , j = 2
d 1 −d 1 (8)
2 ( t)i+ 1 , j i+ 2 , j i+ 2,j
2 2.3 Initial and boundary conditions
with
(2) The freestream conditions for each trajectory point are enu-
d =
(2) (Wi+1, j − Wi, j ) merated in Table 1, which are used as the initial conditions.
i+ 12 , j
(4) (9)
d 1 =
(4) (Wi+2, j − 3Wi+1, j + 3Wi, j − Wi−1, j ). The subscript ∞ represents freestream value in Table 1.
i+ 2 , j Four types of boundary conditions are required for the
The terms di− 1 , j , di, j+ 1 and di, j− 1 are calculated in an computation of flow field, i.e. wall, inflow, outflow and sym-
2 2 2
metric conditions. They are prescribed as follows:
analogus manner.
(2) and
(4) are adaptive coefficients and At the solid wall, no-slip condition is enforced by setting
defined as
domain. At the centre line of the computational domain, the The spherically blunted-cone/flare configuration is illus-
following symmetric conditions are prescribed. trated in Fig. 2c. The conical forebody has R N = 0.51 m,
D = 2.03 m, L = 1.67 m and α N = 20◦ . The flare has a
ρν = 0 half-angle cone of 25◦ and is terminated with a right circu-
∂u ∂T ∂p (14) lar cylinder and is geometrically similar to the REV of the
= = =0 DART demonstrator [18].
∂r ∂r ∂r
At the center line of the reentry capsule, the respective
cell faces are having zero surface area; therefore, it is simple
to implement the symmetric conditions. 3.2 Computational grid
where ri,w and ri,0 are wall and outer surface points, respec- mospheric reentry demonstrator), OREX (orbital reentry ex-
tively, β is the streching factor. n x and nr are total number of periments) with a smooth and a sharp shoulder edge and
grid points in x and r directions, respectively. These streched a spherically blunted cone-flare reentry modules and for
grids are generated in an orderly manner. The typical com- freestream Mach numbers range of 1.2–6.0.
putational space of the reentry module is defined by a num-
ber of grid points in the cylindrical coordinate system. Using
these surface points as the reference nodes, the normal co-
4.1 Flow characteristics
ordinate is then described by exponentially streched struc-
tured field points, extending up to an outer computational
Figures 4–7 show the enlarged view of the computed veloc-
boundary. The streched grid points in the direction is ob-
ity vector field over the above-mentioned vehicles at various
tained using exponentially stretched relation. These grids are
freestream Mach numbers M∞ . It can be visualized from the
generated in an orderly manner. Grid independence tests are
vector plots that all the significant flowfield features such as
carried out taking into consideration the effect of the com-
bow shock wave, rapid expansion fans on the corner, flow re-
putational domain, the streching factor to control the grid
circulation region with converging free shear layer and for-
intensity near the wall, and the number of grid points in the
mation of the vortex flow in the aft region of the capsule.
axial and normal directions. The outer boundary of the com-
The wake flow field, immediately behind the capsule base,
putational domain is varied from 5 to 12 times the maximum
exhibits complex flow characteristics. The formation of the
diameter of the capsule. The grid streching factor in the ra-
bow shock wave on the forebody of the capsule is observed,
dial direction is varied from 1.5 to 5. The present numerical
which depends on R N and α N and the value of M∞ . The
analysis is carried out on 132 × 52 grid points. The grid
bow shock wave moves close to the forebody with the in-
streching factor is selected as 5, and the outer boundary of
creasing M∞ and the stand off distance between bow shock
the computational domain is kept about 4–7 times the maxi-
wave and the forebody decreases with the increasing M∞ .
mum diameter of the reentry module. In the downstream di-
rection, the computational boundary is about 6–10 times the
maximum diameter of the capsule. The minimum grid spac-
ing at the wall is about 2 × 10−5 − 8 × 10−5 m, sufficient
to resolve the boundary layer and complex flow field which
gives resulting Reynolds number after bow shock wave for
this minimum grid spacing as 33 − 41. The coarse grid helps
in reducing the computer time. A close-up view of the com-
putational grid over different capsules can be seen in Fig. 3.
This grid arrangement is found to give a difference mea-
sured and computed values of ±1.5% in the drag coefficient.
The convergence criterion is based on the difference in den-
sity values, ρ, at any grid point between two successive iter-
ations, that is, |ρ l+1 − ρ l | ≤ 10−5 , where l is the iterative
index.
Fig. 5 Enlarged view of velocity field over OREX (with smooth shoul-
der edge)
In Fig. 7, the bow shock wave does not follow the body con-
tour in the case of the spherically blunted cone-flare config-
uration, which is attributed to small values of R N and α N as
compared to OREX. A gradual flow turning can be visual-
ized in the case of OREX with smooth shoulder edge as seen
in Fig. 5 whereas a sharp flow turning is found in the sharp
shoulder edge of the OREX as noticed in Fig. 6. The ap-
proaching supersonic boundary layer separates at the corner
and the free shear layer is formed in the wake region. The
wake flow feature also depicts vortex attached to the corner
with a large recirculating flow behind the vehicle adjacent to
the axis of symmetry which depends on α B and M∞ . The
separation point moves downstream from the shoulder to-
wards the base with the increase in M∞ . Similar flow-field
features were observed in the analysis of the bulbous pay-
load shroud of the heat shield of the launch vehicle [24]. Fig. 8 Mach contours over ARD
38 R. C. Mehta
Fig. 11 Mach contours over spherically blunted cone-flare module Fig. 12 Variation of pressure coefficient along the surface (ARD)
Numerical simulation of supersonic flow past reentry capsules 39
Fig. 13 Variation of pressure coefficient along the surface (OREX with Fig. 15 Variation of pressure coefficient along the surface (spherically
smooth shoulder edge) blunted cone-flare module)
Fig. 17 Variation of skin friction coefficient along the surface (OREX Fig. 18 Variation of skin friction coefficient along the surface (OREX
with smooth shoulder edge) with sharp shoulder edge)
5 Conclusion 7. Yamamoto, Y., Yoshioka, M.: CFD and FEM coupling analysis
of OREX aerothermodynamic flight data. AIAA Paper 95-2087
(1995)
The flow field over various reentry axisymmetric configu- 8. Tam, L.T.: LU-SGS Implicit Scheme for entry vehicle flow com-
rations are studied numerically by solving time-dependent putation and comparison with aerodynamic flight data. AIAA pa-
compressible Navier–Stokes equations. The governing fluid- per 95-2671 CP (1992)
flow equations are discretized in spatial coordinates employ- 9. Menne, S.: Computation of non-winged vehicle aerodynamics in
the low supersonic rane. In: Proceedings of the Second European
ing a finite volume approach which reduces the equations Symposium on Aerothermodynamics for Space Vehicles, ESTEC,
to semi-discretized ordinary differential equations. Tempo- Noordwiik, The Netherlands, pp. 73–78, 21–25 November 1994.
ral integration is performed using the two-stage Runge– 10. Solazzo, M.A., Sansone, A., Gasbarri, P.: Aerodynamic charac-
Kutta time-stepping scheme. A local time step is used to terization of the Carina reentry module in the low supersonic
regimes. In: Proceedings of the Second European Symposium on
achieved steady-state solution. Flow field around the cap- Aerothermodynamics for Space Vehicles, ESTEC, Noordwiik, pp.
sules have been calculated in the freestream Mach number 41–47, 21–25 November 1994
range of 1.2–6.0 for different configurations. The essential 11. Walpot, L.: Numerical Analysis of the ARD capsule in S4 wind
flow-field features around the capsules are captured for vari- tunnel. In: Proceedings of the Fourth European Symposium on
Aerothermodynamics for Space Applications, Capua, Italy, pp.
ous reentry capsules. The effects of geometrical parameters, 523–527, 15–18 October ESA, 2001
such as radius of the spherical cap, half cone angle, with 12. Lamb, J.P., Oberkampf, W.L.: Review and development of base
the sharp shoulder edge and with the smooth shoulder edge pressure and base heating correlations in supersonic flow. J.
on the flow field have been numerically investigated for var- Spacecraft Rockets 32(1), 8–23 (1995)
ious reentry configuration. The ARD, the OREX capsules 13. Liever, P.A., Habchi, S.D., Burnell, S.I., Lingard, J.S.: Computa-
tional prediction of the Beagle 2 aerodynamic database. J. Space-
have the sonic line over the forebody shoulder whereas the craft Rockets 40(5), 632–638 (2003)
spherically blunted cone-flare module is having sonic line 14. Osu, H., Abe, T., Ohnishi, Y., Sasoh, A., Takayama, K.: Numerical
over the spherical cap region. The flow field behind the bow investigation of high-enthaly flow generated by expansion tube.
shock wave is either subsonic or mixed subsonic–supersonic AIAA J. 40(12), 2423–2430 (2002)
15. Tai, C-S., Kao, A-F.: Navier–Stokes solver for hypersonic flow
region depending upon the geometrical parameters of the over a slander cone. J. Spacecraft Rockets 31(1), 215–222 (1994)
reentry configuration. Thus, the reentry configurations can 16. Teramoto, S., Hiraki, K., Fujii, K.: Numerical analysis of dynamic
be distinguished by the location of the sonic line over the stability of a reentry capsule at transonic speeds. AIAA J 39(4),
fore-body at supersonic Mach number which influences wall 646–653 (2001)
17. Teramoto, S., Fujii, K.: Mechanism of dynamic instability of a
pressure and skin friction variations and also aerodynamic reentry capsule at transonic speeds. AIAA J 40(12), 2467–2475
drag coefficient. The shoulder edge affects the corner expan- (2002)
sion wave. The pressure coefficient, the skin friction coeffi- 18. Otten, H.B.A.: Preliminary computational investigation on aero-
cient variation along the surface and the integrated value of dynamic phenomena on DELFT aerospace reentry test vehicle. In:
pressure coefficient will be useful quantities for optimization Proceedings of the Fourth Symposium on Aerothermodynamics
for Space Applications, Capua, Italy, pp. 207–213, ESA, 15–18
of the reentry capsule. October 2001
19. Peyret, R., Viviand, H.: Computational Methods for Fluid Flow.
Springer-Verlag, New York (1993)
References 20. Jameson, A., Schmidt, W., Turkel, E.: Numerical solution of Eu-
ler equations by finite volume methods using Runge-Kutta time
1. Chester, W.: Supersonic flow past a bluff body with a detached stepping schemes. AIAA Paper 81-1259 (1981)
shock. J. Fluid Mech. 1, 353–365 (1956) 21. Mehta, R.C.: Comparitive study of surface pressure fluctuations
2. Freeman, N.C.: On the theory of hypersonic flow past plane and over bulbous heat shield at mach 0.95. Comput. & Fluids 30(6),
axially symmetric bluff bodies. J. Fluid Mech. 1, 353–365 (1956) 689–709 (2001)
3. Lighthill, M.J.: Dynamics of a dissociating gas, Part 1: Equilib- 22. Mehta, R.C.: Numerical heat transfer study over spiked blunt
rium flow. J. Fluid Mech. 2, 1–32 (1957) bodies at Mach 6.8. J. Spacecraft Rockets 37(5), 700–703
4. Venkatapathy, E., Palmer, G., Prabhu, D.K.: AFE base compu- (2000)
tations including base heating predictions. AIAA Paper 91-1372 23. Shang, J.S.: Numerical simulation of wing-fuselage interference
(June 1991) aerodynamic. AIAA Paper 81-0084 (1981)
5. Haas, B.L., Venkatapathy, E.: Mars Pathfinder computations in- 24. Mehta, R.C.: Flowfield over bulbous heat shield in transonic and
cluding base heating predictions. AIAA Paper 95-2086 (1995) low supersonic speeds. J. Spacecraft Rockets 35(1), 102–105
6. Wood, W.A., Gnoffo, D.F.G., Rault, D.F.G.: Commercial Experi- (1998) AIAA Paper 97-2256 (1997)
ment Transport reentry capsule. J. Spacecraft Rockets 33(5), 643– 25. Bertin, J.J.: Hypersonic Aerothermodynamics. p. 297. AIAA Ed-
646 (1996) ucation Series, AIAA Press, VA, USA pp. 297 (1994)