International Journal of Heat and Mass Transfer 70 (2014) 8190

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

journal homepage: www.elsevier.com/locate/ijhmt

Shock wave boundary layer interactions in hypersonic ows

Bibin John, Vinayak N. Kulkarni , Ganesh Natarajan
Department of Mechanical Engineering, Indian Institute of Technology Guwahati, Guwahati 781 039, India

a r t i c l e i n f o a b s t r a c t

Article history: Shock-wave boundary layer interaction (SWBLI) and associated changes in wall properties for ramp
Received 11 July 2013 induced ow breakdown have been considered in the present studies. A two dimensional nite volume
Received in revised form 23 October 2013 based CFD solver has been developed and implemented successfully to study the SWBLI. Pressure mea-
Accepted 27 October 2013
surements are invariantly considered in the literature for qualitative prediction of various SWBLI param-
Available online 20 November 2013
eters. Hence efforts are made herewith to understand the laminar boundary layer separation in the
presence of ramp induced shock wave through surface heat transfer rates, wall skin friction coefcient
Keywords:
and wall pressure distributions. Effect of variation of freestream and wall properties along with geometric
Shock boundary layer interaction
Hypersonics
changes is considered in present studies. It has been observed from present limited investigations that
Separation ratio of wall temperature to freestream stagnation temperature is the governing parameter for SWBLI
Laminar hypersonic ow instead of the individual temperatures. Increase in Mach number is found to suppress the upstream inu-
ence which results in decrease in extent of separation. Efforts are also made to study the effect of leading
edge bluntness on the SWBLI. These studies are found useful to conrm the earlier reported experimental
observations regarding turbulent re-attachment.
2013 Elsevier Ltd. All rights reserved.

1. Introduction sure gradient near the region of ow deceleration or oblique shock.

Research in the eld of high-speed aerodynamics is centred on
shocks and their interactions that alter the oweld. These altera-
tions affect the forces and heat loads on bodies under consideration
[13]. A topic of great importance and interest in this regard is
shock wave boundary layer interaction (SWBLI), which essentially
deals with inviscid-viscous interaction [4]. The presence of SWBLI
may lead to boundary layer separation [5,6], enhanced heating
loads [7] or even turbulent re-attachment [8]. Internal as well as
external ow aerodynamics gets affected by this interaction. Vari-
ous components or subsystems like wing body junctions, engine
inlet, nozzle, control surfaces etc., experience such shock wave
boundary layer interactions. Therefore conscious efforts must be
undertaken in the design of several subsystems of a space vehicle
or a missile to account for the effect of these interactions. Numer-
ous experimental and computational ndings along with the allied
development of various ow control techniques are reported in the
open literature as an outcome of the research investigations on this
topic [9,10].
A typical schematic diagram representing the ramp induced
SWBLI is shown in Fig. 1. The abrupt deection of ow, in the pres-
ence of ramp, leads to generation of shock wave emanating from
the compression corner, which interacts with the boundary layer
over the wall. Hence, the boundary layer experiences adverse pres-
Corresponding author. Tel.: +91 361 258 2655; fax: +91 361 2690762.
E-mail address: vinayak@iitg.ernet.in (V.N. Kulkarni).
Separation of the ow in the presence of adverse pressure gradient
depends on many parameters. Separation of laminar boundary
layer depends on several factors including Mach number, Reynolds
number, ramp angle, wall temperature and boundary layer stabil-
ity. Amongst which the incipient separation angle of a particular
case can be identied from the relation given by Needham and
Stollery [11].
q
L
M1 hi:s 80 X 1
where X L is the viscous interaction parameter at the at plate-ramp
junction, and is given by,
p p
 L M3 C = ReL where C lw T 1
X 1 2
l1 T w
If the deection angle is higher than the incipient separation angle,
then boundary layer separation takes place. The separation of the
ow occurs at station S (shown in Fig. 1), well ahead of the com-
pression corner, depending on the viscous interaction parameter.
In such a case, separation shock forms ahead of the separation re-
gion due to the coalescence of compression waves induced by the
separation process [12]. The nearly constant pressure region down-
stream of the separation point is invariantly considered as the indi-
cator to identify the separation bubble during pressure
measurement experiments. This recirculation zone extends up to
the reattachment point R (Fig. 1), where the ow reattaches on
the ramp surface. Distance between the point of separation and

0017-9310/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2013.10.072
82 B. John et al. / International Journal of Heat and Mass Transfer 70 (2014) 8190

Nomenclature

L
X viscous interaction parameter q heatux
q density r leading edge radius
h ramp angle Re1 Reynolds number/meter (q1 U1/l1)
his incipient separation angle ReLref Reynolds number (Re1 Lref)
l coefcient of dynamic viscosity St Stanton number
s shear stress T temperature
C ChapmanRubesin constant Tr reference temperature
Cf skin friction coefcient U conservative variable vector
Cp pressure coefcient u velocity in x-direction
E total energy v velocity in y-direction
EI x-component of convective ux vector s stress
FI y-component of convective ux vector
Ev x-component of viscous ux vector Subscripts
Fv y-component of viscous ux vector 1 free stream conditions
H total enthalpy 0 total or stagnation values
Lref reference length (forward at plate length in present r reference values
studies) w wall properties
M Mach number CR critical
P pressure
Pr Prandtl number

the reattachment point is called as the length of separation bubble.
A ramp angle smaller than the incipient separation angle, can lead
to signicant alteration in the basic at plate laminar boundary
layer prole ahead of the ramp-foot, without separation. Such up-
stream ow alteration due to the presence of the ramp can be
termed as upstream inuence. The distance from the ramp-foot to
the most upstream location, which experiences the inuence, is
generally dened as extent of upstream inuence.
In the past, SWBLI has been investigated by several researchers
owing to its growing signicance. There are many reported inves-
tigations in this eld for both high speed laminar and turbulent
ows. The earliest literatures that give a physical understanding
of SWBLI in supersonic and hypersonic ow regime are the inves-
tigations of Chapman et al. [5] and Needham and Stollery [11].
Through a series of experimental studies in both laminar and tur-
bulent ow regimes, Chapman et al. [5] rst developed the free
interaction concept in the presence of various ramp angles. Hol-
den [13,14] performed theoretical and experimental studies, to
understand the inuence of Mach number, Reynolds number,
wedge angle and leading edge bluntness on SWBLI phenomena.
Coet et al. [8], during their studies, focused on the effect of leading
edge bluntness and associated entropy layer on the ramp induced
shock wave boundary layer interaction. Rizzeta and Mach [15] con-
ducted numerical investigation of ramp induced SWBLI in the lam-
inar hypersonic (M1 = 14.1) oweld by employing four different
numerical algorithms. Marini [16] carried out experimental
investigations to study the effect of geometrical parameters on
the shock induced separation. Davis and Bradford [17] performed
studies in high enthalpy non-equilibrium ow and computed the
separation length both numerically and experimentally. SWBLI
has also been computationally studied by Layland [18]. Recently
Marini [19] has extensively studied and reviewed ramp induced
SWBLI in the laminar hypersonic ow regime. In this review,
numerical studies which reveal the inuence of different ow
parameters were presented along with discussion of former exper-
imental studies. Numerical investigations of Reinartz et al. [20] is
the latest reporting about the effect of wall temperature on ramp
induced SWBLI. In this study, authors employed two different Rey-
nolds Averaged NavierStokes solvers to analyse SWBLI in laminar
as well as transitional ow regimes.
In spite of extensive numerical and experimental studies per-
formed to understand SWBLI, there exist only a limited number
of ndings [14,16] that have noticed adverse effect of decrease in
freestream Mach number and stagnation enthalpy on this vis-
cousinviscid interaction. Furthermore, the difculties involved
in measurement of wall shear stress in high-speed facilities neces-
sitate the use of pressure and heat transfer measurements to
understand the phenomenon of shock-induced separation. Results
on separation length from pressure or heat transfer measurements
must be interpreted qualitatively and quantifying these predic-
tions is possible only through high-resolution computations. In line
with this, effect of small leading edge bluntness radii on SWBLI has

Fig. 1. Schematic diagram representing the two dimensional high speed ow over a compression corner with SWBLI.
 B. John et al. / International Journal of Heat and Mass Transfer 70 (2014) 8190 83

not been addressed in the open literature. In light of these observa-
tions, the present studies focus on the following objectives.
1. To understand the effect of Mach number and total temperature
on upstream inuence and boundary layer separation as well as
on heat transfer.
2. To compare the qualitative predictions of separation length
from pressure and heat transfer rate measurements with quan-
titative data of skin friction computations and to comment on
the validity of conclusions made from experiments in high-
speed ows.
3. To numerically predict and validate the correlation for incipient
ow separation as well as experimental observations on SWBLI
in blunted leading-edge ramps.
4. To study the effect of very small leading edge bluntness on the
size of separation bubble
Fluid is assumed to be an ideal gas for present studies. Coef-
cient of dynamic viscosity (l) is calculated by making use of Suth-
erlands law [22], while Prandtl number (Pr) is taken as 0.71. Finite
volume formulation is considered to develop the present solver.
Time integration is achieved using an Euler explicit scheme with
a Courant number in the range of 0.20.4, to ensure numerical sta-
bility. The inviscid uxes are calculated using upwind AUSM
scheme [23], while the viscous uxes at the interface are obtained
directly using the face gradients calculated using the strategy pro-
posed by Blazek [24]. Linear reconstruction suggested by Barth and
Jespersen [25] is incorporated to achieve the higher order accuracy
using Venkatakrishnans limiter function [26,27]. Boundary condi-
tions like inlet, outlet, symmetry and viscous isothermal or viscous
adiabatic wall have been implemented in the solver. Generic nat-
ure of the solver allows the use of both structured and unstruc-
tured meshes for computations through face based algorithm [24].

Hence the present investigations focus on the effect of various

parameters such as ramp angle, freestream Mach number, free- 3. Numerical investigations
stream stagnation enthalpy, wall temperature and leading edge ra-
dius on the separation bubble dynamics in hypersonic ows. Numerical investigations of SWBLI have been carried out to
Therefore the major thrust of present SWBLI studies is to under- study the effect of ramp angle, wall temperature, freestream total
stand the boundary layer separation due to the adverse pressure enthalpy, Mach number and leading edge bluntness. Grid indepen-
gradient in the presence of ramp induced shock wave which re- dence studies are carried out to avoid existence of grid based error
quires an analysis of quantities such as wall heat transfer, wall skin in the solution in these cases. Hence a typical mesh independence
friction coefcient and wall pressure. The details of the viscous test is discussed herewith. A at plate of length 0.05 m attached
ow solver and the overall solution methodology are presented with 15 ramp along with freestream conditions of M1 = 6.0 and
in Section 2. Section 3 is devoted to the discussions on the numer- Re1 = 8  105 m1 are considered herewith to demonstrate the
ical simulations for SWBLI and its dependence on various parame- grid independence studies. Four different meshes, viz. 180  60,
ters. The conclusions from the present study and scope of future 180  90, 240  120 and 360  180, with sufcient clustering near
research are outlined in Section 4. the wall to resolve the boundary layers, are employed for the com-
putations. Grid points are also clustered at the leading edge of at
2. Flow solver and solution methodology plate and ramp-at plate junction. Such sample grid used for the
present study, marked with boundary conditions, is shown in
A cell-centred nite volume ow solver, based on unstructured Fig. 2. Details of the grids employed for the present study are given
data structure, has been developed in-house to solve the compress- in Table 1. The convergence histories of all four grid levels are also
ible NavierStokes equations. The non-dimensional form of these given in Fig. 3. It is clear from this gure that, the cost involved in
equations for the Unstructured Solver for Hypersonic Aerothermo- the computation enhances with increase in number of cells. Fig. 4a
dynamic Simulations (USHAS) [21] are given as, shows the variation of skin friction coefcient (Cf) along the wall,
obtained using four grid levels. Percentage error in Cf is also eval-
@U @EI @F I @Em @F m
3 uated using the reference Cf values of 300  150 grid in order to
@t @x @y @x @y clearly represent the error level in skin friction prediction for dif-
where, ferent grid sizes. Thus obtained Cf residues are given in Fig. 4b. It
3
2 2 3 2 3 is clear from this gure that the error reduces with grid renement,
q qu qm which in turn supports the grid convergence. The skin friction coef-
6 qu 7 6 qu2 p 7 6 qum 7 cient value at a particular location x = 0.035 m from the leading
6 7 6 7 6 7
U6 7 EI 6 7 FI 6 2 7
4 qm 5 4 qum 5 4 qm p 5 edge and separation bubble size (Lb) obtained with different grid
levels are also compared in Fig. 5. It is evident from Figs. 4 and 5
qE quH qmH
that the 180  90 mesh gives almost grid-independent solution
2 3 2 3 with lesser computational cost. Such mesh independence studies
0 0
6s 7 6 sxy 7 are performed for all the test cases and mesh independent results
6 xxp 7 6 7 are considered for understanding of SWBLI.
Em 6 7 Fm 6 7 4
4 sxy 5 4 syyp 5
usxxp msxy  qx usxy msyyp  qy

The terms in these vectors are,

   
l 4 @u 2 @ m l 4 @ m 2 @u
sxxp  ; syyp  ;
ReLref 3 @x 3 @y ReLref 3 @y 3 @x
 
l @u @ m
sxy ;
ReLref @y @x

l @T l @T
qx ; qy
ReLref Prc  1M21 @x ReLref Prc  1M 21 @y
q1 U 1 Lref
where, ReLref Re1 Lref l1 ; Pr 0:71 Fig. 2. Meshed computational domain marked with boundary conditions.
84 B. John et al. / International Journal of Heat and Mass Transfer 70 (2014) 8190

Table 1
Details of grids used for parametric studies of Marinis [16] test case where all dimensions are in m.

Overall Maximum boundary No. of grid points in the Minimum element size Minimum element size at the Minimum element size at the at
   
grid size layer thickness boundary layer near the wall (Dymin) leading edge Dxminle plate-ramp junction Dxminjunc

180  60 0.003 15 9  105 7  104 6  104

180  90 0.003 20 5  105 5  104 4  104
240  120 0.003 25 3  105 3  104 2  104
360  180 0.003 35 1  105 9  105 8  105

Fig. 3. Convergence histories of different grid levels.

Fig. 4b. Grid convergence test for wall Cf distribution.

Fig. 5. Comparison of Cf value at x = 0.035 m and separation bubble size (Lb)

obtained with different grid levels.

the experiments of Marini [11] on two test models. These congu-

Fig. 4a. Grid independence study for a ramp based boundary layer separation test
case.
3.1. Effect of ramp angle
Studies for the effect of ramp angle on shock induced boundary
layer separation are carried out to verify the incipient separation
condition [11] and also to enumerate the observations of Marini
[16] from the pressure measurement experiments. The freestream
conditions considered for these simulations are Mach 6, Reynolds
number 8  105 m1, wall temperature 300 K and freestream stag-
nation temperature 1080 K. These test conditions are same as in
rations had same initial plate length (L1) of 0.05 m but different an-
gles viz. h = 10 and h = 15 respectively (Fig. 1). In addition, a
model with ramp angle of 12.5 has been considered to verify
the effect of numerics on the incipient separation angle (his), which
for the chosen test conditions is 13.5. The surface pressure distri-
bution from the present computations is compared with the exper-
imental results of Marini [11] in Fig. 6. Encouraging agreement
between the experimental and computational results can be no-
ticed in this gure. Pressure distribution for the ramp angles 10
and 12.5 exhibit same trend unlike for the ramp angle 15 where
marginal constant pressure region, upstream of the ramp indicate
the presence of a separation bubble. The variation of heatux along
the plate and ramp for all the three cases is as shown in Fig. 7. This
 B. John et al. / International Journal of Heat and Mass Transfer 70 (2014) 8190
Fig. 6. Effect of ramp angle on pressure distribution.
Fig. 7. Effect of ramp angle on heatux distribution.
distribution is seen to follow the typical V-shaped curve in the
vicinity of the ramp for the two lower angled test models whereas
the heat ux distribution of ramp angle 15 follows a diffused V or
U shaped curve at the same location. Increase in upstream inu-
ence, noted from early rise in pressure and heatux values ahead
of the ramp, with increase in ramp angle can be depicted from both
Figs. 6 and 7. However it is not possible to obtain further informa-
tion with regard to separation from these data which otherwise is
evident from the skin friction distribution on the wall (Fig. 8) ob-
tained from computations. The points at which the curve crosses
the zero line give the locations of separation and reattachment
points. Details of numerical results obtained for ramp angle varia-
tions are given in Table 2. From the present simulations, it is seen
that the 10 ramp does not lead to separation while separation
bubble of length 17.75 mm is observed for the ramp angle of 15.
Interestingly, a separation zone of length 9.2 mm is observed for
the 12.5 ramp, in contrast to Needham and Stollerys correlation
which predicts no separation for ramp angles below 13.5. The
authors are of the opinion that, since the numerical solutions are
grid-independent and high-resolution, the correlation must be al-
luded to only for cases that are well-separated. The latter ows
85
Fig. 8. Effect of ramp angle on skin friction distribution.
are those for which there exists a discernable plateau in the pres-
sure measurements. In other words, for ramp angles within the
range of incipient separation predicted by the correlation, numer-
ical studies are necessary to reveal the physics of the actual ow.
3.2. Effect of wall temperature
Numerical simulations are also carried out to understand the ef-
fect of wall temperature on shock-boundary layer interaction phe-
nomenon. A 15 ramp has been considered for this study as it has
the maximum separation length among congurations considered
herein. The freestream conditions are chosen as M1 = 6.0, Re1 = 8
 105 m1 and T0 = 1080 K along with two different wall tempera-
tures viz. 300 K and 500 K. In addition, the effect of an adiabatic
wall as opposed to an isothermal wall is also investigated. Surface
variation of pressure coefcient, heatux and skin friction coef-
cient are shown in Figs. 911 respectively. Numerical results for
wall temperature variations are also summarized in Table 2. From
the present studies, it is clear that the upstream inuence increases
with increase in wall temperature. Hence the adiabatic wall
boundary condition has the largest upstream inuence. Wall heat
ux and skin friction coefcients are seen to be decreased at all
the locations for increased wall temperature case. The reason for
decrease in these wall properties lies in the fact that the increased
wall temperature increases the viscosity which in turn increases
the hydrodynamic and thermal boundary layer thicknesses. Hence,
decrease in near wall gradients leads to decrease in heatux and
skin friction coefcient. This is the reason for least skin friction
at all the spatial locations for adiabatic wall boundary condition
for which wall heat ux is zero. However, thicker boundary layer
is more susceptible for separation, therefore early separation at
x = 0.03825 m can be noticed for the increased wall temperature
(500 K) case. Most upstream separation can be seen for adiabatic
wall condition at x = 0.03225 m. Point of reattachment exhibits a
trend of downstream shift with increase in wall temperature.
Hence the separation bubble length is seen to be increased from
17.75 mm of 300 K case to 27 mm for 500 K wall temperature
while the same for the adiabatic wall case is 46.25 mm. That is,
the longest separation bubble among the three cases is for the
adiabatic wall which is consistent with the results on upstream
inuence as well.
86 B. John et al. / International Journal of Heat and Mass Transfer 70 (2014) 8190

Table 2
Summary of parametric studies of SWBLI.

Parameter M1 Re1 (m1) T0 (K) Tw (K) h() r (mm) Lb (mm) Inference

5
Ramp angle 6 8  10 1080 300 10 0.0 0.0 h " Lb "
6 8  105 1080 300 12.5 0.0 9.2
6 8  105 1080 300 15 0.0 17.75
Wall temperature 6 8  105 1080 300 15 0.0 17.75 T w " Lb "
6 8  105 1080 500 15 0.0 27
6 8  105 1080 Adiabatic 15 0.0 46.25
Freestream enthalpy 6 8  105 1800 300 15 0.0 13.2 Tw
" Lb "
T0
6 8  105 1080 300 15 0.0 17.75
6 8  105 1800 500 15 0.0 18.2
6 8  105 1080 500 15 0.0 27
Mach number 5 8  105 1080 300 15 0.0 23 M1 " Lb ;
6 8  105 1080 300 15 0.0 17.75
7 8  105 1080 300 15 0.0 13.25
8 8  105 1080 300 15 0.0 8.6
Leading edge radius-study 1 10 8.36  106 1100 300 15 0.0 97.2 r < r CR
10 8.36  106 1100 300 15 2.5 55.5 r " Lb "
10 8.36  106 1100 300 15 5.0 38.3
Leading edge radius-study 2 6 8  105 1080 300 15 0.0 17.75 r > r CR
6 8  105 1080 300 15 0.1 28.2 r " Lb #
6 8  105 1080 300 15 0.5 31.3
6 8  105 1080 300 15 1.0 18.99
6 8  105 1080 300 15 1.5 12.9
6 8  105 1080 300 15 2.0 11.08

Fig. 9. Effect of wall temperature on pressure distribution.

Fig. 10. Effect of wall temperature on surface heatux distribution.

3.3. Effect of variation of freestream total enthalpy

Investigations have been performed with ows having two dif-
ferent total temperatures to explore the effect of freestream total
enthalpy (or total temperature). The conditions corresponding to
total temperature 1080 K are referred here as low enthalpy test
conditions while conditions corresponding to total temperature
1800 K are referred here as high enthalpy test conditions. For these
simulations, freestream Mach number and unit Reynolds number
are 6 and 8  105 m1 respectively. Computations are carried out
with two different wall temperatures as in Section 3.2. The varia-
tion in freestream total temperature affects the properties at the
edge of the boundary layer, while wall temperature controls the
temperature distribution within the boundary layer. Therefore
the focus herein is to study the combined effect of these quantities
on SWBLI. The surface variation of pressure coefcient, skin friction
coefcient and non-dimensional heatux are shown in Figs. 1214
respectively. Here, non-dimensional heat ux at a location has
been obtained using reference maximum surface heat ux. For a gi-
ven wall temperature, increase in freestream stagnation enthalpy
is seen to decrease upstream inuence and consequently the ex-
tent of separation. For instance, the separation length for 300 K
wall temperature reduces from 17.75 mm at low enthalpy condi-
tions to 13.2 mm for high enthalpy conditions. The main reason
for this observation is that, the increased stagnation enthalpy in-
creases the kinetic energy of the uid and provides a greater resis-
tance to the adverse pressure gradient. A similar observation can
also be made for wall temperature of 500 K and is evident in
Fig. 13. An interesting observation in this study is that the non-
dimensional heat ux distribution, surface pressure distribution
as well as the skin friction distribution is nearly identical for the
conditions of low stagnation enthalpy with low wall temperature
as well as high stagnation enthalpy with high wall temperature.
It follows that in either case, the ratio of totalto-wall temperature
 B. John et al. / International Journal of Heat and Mass Transfer 70 (2014) 8190 87

Fig. 11. Effect of wall temperature on skin friction distribution. Fig. 13. Effect of freestream enthalpy on skin friction distribution.

Fig. 12. Effect of freestream enthalpy on pressure distribution.

Fig. 14. Effect of freestream enthalpy on non-dimensional heatux distribution.

is the same and therefore this non-dimensional ratio can be con-

sidered as an important indicator of SWBLI rather than the stagna-
tion and/or the wall temperature themselves. Hence, from the
present limited investigations, it is hypothesized that the parame-
ter T0/Tw is the best indicator to make a meaningful comparison of
SWBLI. Summary of these results is also provided in Table 2.

3.4. Effect of variation of freestream Mach number

Four different Mach numbers viz. 5, 6, 7 and 8 have been used in

the present study to explore the Mach number effect on SWBLI. In
this case, a xed value of Reynolds number Re1 = 8  105 m1 has
been used along with constant freestream temperature of 131.7 K
and wall temperature of 300 K. The incipient separation condition
analysis shows that, the critical angle is a direct function of free-
stream Mach number and also it decreases with increase in Mach
number. However the ramp angle under consideration (15) is suf-
ciently above the incipient separation angle for the lowest Mach
number considered for present studies. This is easily identied
from the plateau in the pressure plot (Fig. 15) as well as the skin
friction distribution (Fig. 16). Effect of freestream Mach number Fig. 15. Effect of freestream Mach number on pressure distribution.
88 B. John et al. / International Journal of Heat and Mass Transfer 70 (2014) 8190
Fig. 16. Effect of freestream Mach number on skin friction distribution.
variation on the interaction is given in Table 2. The points of sepa-
ration and reattachment from skin friction distribution show that
the longest separation bubble occurs at M = 5 while at M = 8 the
ow cannot be considered as well-separated. This observation
is along the expected lines because a higher freestream Mach num-
ber in the present case mean higher kinetic energy of the ow for
same internal energy which in turn leads to a smaller separation
bubble.
3.5. Effect leading edge bluntness
Hypersonic vehicles have blunt nosed conguration to reduce
the surface heating rate. From a practical viewpoint, it is impera-
tive to understand how leading edge bluntness may affect the
SWBLI. Moreover, the entropy layer formation associated with de-
tached bow shock in the presence of blunt leading edge has signif-
icant effect on shock boundary layer interaction. Hence exploration
of the same is the objective of present study. Experimental studies
were performed by Coet et al. [8] to investigate this effect. They
carried out pressure and heat ux measurements for Mach number
10 and Re1 = 8.36  106 m1 ow over 15 ramp attached with at
plate having three different leading edge curvatures viz. sharp,
r = 2.5 mm and r = 5 mm. In the present study, computations are
carried out to reveal the effect of leading edge bluntness for the
same congurations that were experimentally investigated by Coet
Fig. 17. Mach contours of ramp induced SWBLI (a) blunted leading edge (r = 5 mm) (b) sharp leading edge.
Fig. 18. Effect of leading edge bluntness on pressure distribution.
et.al. The computational domain for the cases of leading edge ra-
dius zero, 2.5 mm and 5 mm are meshed with quadrilaterals and
have sizes 510  90, 480  90 and 470  90, respectively. The
numerical investigations showed that the leading edge shock is a
detached bow shock in case of blunted leading edge, where as it
is an attached oblique shock in case of sharp leading edge
(Fig. 17). It should also be noticed here that, for the same free-
stream Mach number, the ramp-approach Mach number reduces
in the presence of bow shock. Studies in Section 3.5 have shown
that SWBLI would be prominent at lower Mach numbers. However,
the presence of the bow shock, for a blunted leading-edge ramp,
creates an entropy gradient normal to the ow behind the shock.
This entropy gradient in turn results in a strong vortical activity
just outside the boundary layer and provides a stabilizing effect
which counters the destabilizing effect of reduced Mach number.
This explains the reduction in the intensity of SWBLI with in-
creased leading edge bluntness (Table 2). The surface pressure dis-
tributions predicted from computations are shown in Fig. 18 and
agree well with experimental results. Interestingly, it can be no-
ticed that the peak pressure downstream of the reattachment point
in the case of the blunt leading edge ramp is nearly half of its value
as compared to a sharp leading edge ramp and may be attributed
to the fact that the separation shock is much stronger in the latter
case. Numerical predictions of the Stanton number (Fig. 19) show
excellent agreement with the results of Coet et al. [8] for all cases
 B. John et al. / International Journal of Heat and Mass Transfer 70 (2014) 8190 89

Fig. 19. Effect of leading edge bluntness on heatux distribution. Fig. 20. Effect of leading bluntness on distribution of skin friction coefcient.

except in the post-reattachment regions. The computations consis- similar observations have been reported by Neuenhahn and Olivier
tently under predict the heat transfer rate downstream of the reat- [28] from the experimental studies. In all, present results depict
tachment point and this discrepancy between the numerical and that the thin submerged entropy layer, at lower nose radii, should
experimental results is more prominent for the sharp leading-edge be accounted for separation bubble widening while the thick en-
ramp as compared to the blunted leading-edge ramps. This points tropy layer in the presence of moderate nose radii should be cred-
to the fact that the assumption of laminar ow in the computations ited for separation bubble shrinking.
is not valid in the post-reattachment regions and is consistent with
the experimental observations of Coet et al. [8] who have hinted at
4. Conclusion
the possibility of a turbulent reattachment. Furthermore, the heat
transfer rates from experimental measurements are higher than
The effect of various ow and geometric parameters on the
those from numerical predictions, which also add credence to the
ramp-induced SWBLI in laminar hypersonic ows has been inves-
theory of possible turbulent reattachment in the case of sharp
tigated using a high-resolution viscous ow solver. The studies
leading-edge ramp. The reattachment in the case of blunted lead-
show that while pressure and heat transfer measurements provide
ing-edge ramps could also be turbulent or even transitional and
only a qualitative picture of ow separation and reattachment,
a closer investigation is necessary to conclusively establish the nat-
quantitative estimates of separation bubble length and upstream
ure of reattachment.
inuence can be obtained by analysing the wall shear stress distri-
Present solver is also considered to reveal the effect of very
bution. The correlation for predicting incipient separation is criti-
small leading radius on the SWBLI. These simulations are carried
cally analysed and shown to work well only for well-separated
out using 15 ramp angle conguration considered by Marini
ows. The separation bubble length is found to increase with in-
[16] for Mach 6 ow conditions. This sharp leading edge geometry
crease in the ramp angle and wall temperature but decreases with
has been modied to incorporate bluntness at the leading edge.
increasing Mach number and freestream stagnation enthalpy. The
Five different leading edge radii viz. 0.1 mm, 0.5 mm, 1 mm,
present investigations provide limited evidence that the total-
1.5 mm and 2 mm have been considered. The results obtained with
to-wall temperature ratio would be a more relevant quantity for
various blunt nose radii are compared in terms of skin friction coef-
separation bubble studies. The effect of leading edge bluntness
cient distribution and are shown in Fig. 20. It is evident from this
on separation dynamics shows that the ow is laminar only
gure that the small leading edge bluntness as well alters the ow-
up-to the point of separation and supports the experimental obser-
eld signicantly. Upstream shift of separation point and down-
vations of possible turbulent reattachment. Investigations on the
stream shift of reattachment point has been noticed for smaller
effect of leading edge radius on separation bubble length show that
leading edge radius of 0.1 mm. Such enhancement in separation
there is a strong correlation between these parameters which
bubble size is seen to be more prominent for 0.5 mm leading edge
warrants a critical study in the future. Future works will also ex-
case. However, further increase in leading edge radius to 1 mm is
plore the dynamics of different methodologies for suppressing
found to push the reattachment point in the downstream direction
separation as well as evaluate the accuracy and efcacy of existing
with almost unaltered separation location. Decrease in separation
correlations in predicting separation bubble dynamics.
bubble size by downstream shift of separation point and upstream
shift of the reattachment point has been observed with increase in
leading edge radius to 1.5 mm. Such reduction in separation bub- References
ble size is more prominent for 2 mm leading edge radius. Hence
[1] R. Kimmel, D. Adamczak, D. Gaitonde, A. Rougeux, J. Hayes, HIFiRE-1 boundary
these results point that; initial increase in leading edge radius
layer transition experiment design, in: Proceedings of the 45th AIAA Aerospace
tends to enhance the separation zone for very small radii, whereas Sciences Meeting and Exhibit, Reno, NV, USA, 2007.
increase in leading edge radius, at moderate radii, reduces the sep- [2] T.J. Horvath, S.A. Berry, B.R. Hollis, Boundary layer transition on slender cones
aration bubble size (Table 2). Hence, it conrms the existence of in conventional and low disturbance Mach 6 wind tunnels, in: 32nd AIAA Fluid
Dynamics Conference and Exhibit, St. Louis, Missouri, 2002.
critical nose radius for given freestream conditions beyond which [3] S.P. Schneider, Hypersonic laminar-turbulent transition on circular cones and
decrease in separation bubble size can be assured. Recently, very scramjet forebodies, Prog. Aerosp. Sci. 40 (12) (2004) 150.
90 B. John et al. / International Journal of Heat and Mass Transfer 70 (2014) 8190
[4] J.D. Anderson Jr, Hypersonic and High Temperature Gas Dynamics, McGraw-
Hill, New York, 1989.
[5] D.R. Chapman, D.M. Kuehn, H.K. Larson, Investigation of separated ows in
supersonic and subsonic streams with emphasis on the effect of transition,
NACA TN 3869, NACA Rep. 1356, 1957.
[6] J.R. Sterrett, J.C. Mery, Experimental separation studies for two-dimensional
wedges and curved surfaces at M = 4.8 to 6.2, NASA TN D-1014, 1962.
[7] M.S. Holden, J.R. Moselle, Theoretical and Experimental Studies of the Shock
Wave-Boundary Layer Interaction on Compression Surfaces in Hypersonic
Flow, ARL 70-0002, Aerospace Research Laboratories, Wright-Patterson AFB,
OH, 1970.
[8] M.C. Coet, B. Chanetz, J.M. Delery, Shock-Wave Boundary Layer Interaction
with Entropy Layer Effect in Hypersonic Flow, ONERA, Colloque sur les
Ecoulements Hypersoniques, Garchy, France, 1992.
[9] M.R. Myers, A.B. Jorge, M.J. Mutton, D.G. Walker, Using ultrasound and the
extended kalman lter for characterizing aerothermodynamic environments,
AIAA J. 51 (10) (2013) 24102419.
[10] M.R. Myers, A.B. Jorge, M.J. Mutton, D.G. Walker, High heat ux point source
sensitivity and localization analysis for an ultrasonic sensor array, Int. J. Heat
Mass Transfer 55 (910) (2012) 24722485.
[11] D.A. Needham, J.L. Stollery, Boundary-layer separation in hypersonic ow, in:
AIAA paper 66-455, 1996.
[12] F. Grasso, M. Marini, Analysis of hypersonic shock wave laminar boundary-
layer interaction phenomena, Comput. Fluids 25 (6) (1996) 561581.
[13] M.S. Holden, Boundary-layer displacement and leading-edge bluntness effects
on attached and separated laminar boundary layers in a compression corner.
Part I: Theoretical study, AIAA J. 8 (l) (1970) 21792188.
[14] M.S. Holden, Boundary-layer displacement and leading-edge bluntness effects
on attached and separated laminar boundary layers in a compression corner.
Part II: Experimental study, AIAA J. 9 (l) (1971) 8493.
[15] D. Rizzetta, K. Mach, Comparative numerical study of hypersonic compression
ramp ows, in: AIAA Paper 89-1877, 1989.
[16] M. Marini, Effects of ow and geometry parameters on shock-wave boundary-
layer interaction phenomena, in: AIAA Paper 98-1570, 1998.
[17] J.P. Davis, S. Bradford, Separation length in high-enthalpy shock/boundary-
layer interaction, Phys. Fluids 12 (10) (2000) 26612687.
[18] P. Layland, Shock-wave/boundary layer interaction at hypersonic speed by an
implicit NavierStokes solver, Int. J. Comput. Fluid Dyn. 6 (1) (1996) 7187.
[19] M. Marini, Analysis of hypersonic compression ramp laminar ows under
sharp leading edge conditions, Aerosp. Sci. Technol. 5 (2001) 257271.
[20] B.U. Reinartz, J. Ballmann, B. Russell, Numerical investigation of wall
temperature and entropy layer effects on double wedge shock/boundary
layer interactions, in: Proceedings of the 14th AIAA/AHI Space Planes and
Hypersonic Systems and Technologies Conference, 2006.
[21] B. John, G. Sarath, V. Kulkarni, G. Natarajan, Performance comparison of ux
schemes for numerical simulation of high-speed inviscid ows, Prog. Comput.
Fluid Dyn. Int. J. (2013) (in press).
[22] W. Sutherland, The viscosity of gases and molecular force, Philos. Mag. S. 5 (36)
(1893) 507531.
[23] M.S. Liou, C.J. Steffen Jr, A new ux splitting scheme, NASA TM-104404 1991, J.
Comput. Phys. 107 (1) (1993) 2339.
[24] J. Blazek, Computational Fluid Dynamics: Principles and Applications, rst ed.,
Elsevier, 2001.
[25] T.J. Barth, D.C. Jespersen, The design and application of upwind schemes on
unstructured meshes, in: 27th AIAA Aerospace science Meeting, Reno, NV,
AIAA Paper 89-0366, 1989.
[26] V. Venkatakrishnan, On the accuracy of limiters and convergence to steady
state solutions, in: AIAA Paper 93-0880, 1993.
[27] V. Venkatakrishnan, Convergence to steady-state solutions of the euler
equations on unstructured grids with limiters, J. Comput. Phys. 118 (1)
(1995) 120130.
[28] T. Neuenhahn, H. Olivier, Inuence of the wall temperature and the entropy
layer effects on double wedge shock boundary layer interactions, in: 14th
AIAA/AHI Space Planes and Hypersonic Systems and Technologies Conference,
AIAA Paper 2006-8136, 2006.