Proceedings of the Institution of Mechanical

Engineers, Part G: Journal of Aerospace

Engineering
http://pig.sagepub.com/

Numerical simulation and parametric study of a supersonic intake

Mohammad R Soltani, Javad Sepahi Younsi, Mohammad Farahani and Afshin Masoud
Proceedings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineering 2013 227: 467
originally published online 17 February 2012
DOI: 10.1177/0954410012436620

The online version of this article can be found at:

http://pig.sagepub.com/content/227/3/467

Published by:

http://www.sagepublications.com

On behalf of:

Institution of Mechanical Engineers

Additional services and information for Proceedings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineering
can be found at:

Email Alerts: http://pig.sagepub.com/cgi/alerts

Subscriptions: http://pig.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://pig.sagepub.com/content/227/3/467.refs.html

>> Version of Record - Feb 19, 2013

OnlineFirst Version of Record - Feb 17, 2012

What is This?

Downloaded from pig.sagepub.com at RYERSON UNIV on February 20, 2013

Original Article

Proc IMechE Part G:

J Aerospace Engineering
Numerical simulation and parametric 227(3) 467–479
! IMechE 2012

study of a supersonic intake Reprints and permissions:

sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0954410012436620
uk.sagepub.com/jaero

Mohammad R Soltani1, Javad Sepahi Younsi1,

Mohammad Farahani1,2 and Afshin Masoud2

Abstract
A computational fluid dynamics code was developed to compute the flow inside and around a supersonic external
compression axisymmetric intake. The code solves the Reynolds-averaged Navier–Stokes equations using an explicit
finite volume method in a structured grid and uses the Baldwin–Lomax algebraic model to compute the turbulent
viscosity coefficient. Experiments were performed to validate the predicted results and good agreements are achieved.
In the next part of the research, a parametric study was undertaken using the designed base case at a constant Mach
number of 2 and at 0 angle of attack. The effects of various important parameters such as free stream Mach number,
spike deflection angle, and back pressure ratio on the total pressure recovery, mass flow ratio, flow distortion, and drag
coefficient of the intake were then numerically investigated. The results showed that when the spike deflection angle of
the intake was changed from 28 (designed base case) to 30 , the intake drag coefficient was reduced up to 9%. In
addition, the intake performance degraded for very low values of the back pressure ratio.

Keywords
Supersonic intake, back pressure ratio, total pressure recovery, mass flow ratio, flow distortion, drag coefficient

Date received: 11 October 2011; accepted: 4 January 2012

Introduction all information needed to run the newly developed code

The performance of a ramjet engine is strongly affected for an existing supersonic external compression axisym-
by its intake flow characteristics. The main role of the metric intake. Once the code was calibrated and the
intake in this type of engine is to decelerate the super- predicted data were reliable, a parametric study to
sonic flow with minimum losses and provide a uniform investigate the effects of other important intake design
flow with a minimum flow distortion (FD) along with a variables was performed.
maximum total pressure recovery (TPR). The normal Usually, finite volume methods with upwind flux
and/or oblique shocks, shock-boundary layer interac- schemes are used to compute high-speed flows. Roe’s
tion, and flow separation might be present in a super- flux difference splitting scheme and advection upstream
sonic intake. The aforementioned phenomena make the splitting method) family schemes have been efficiently
flow field very complex and the analytical methods are used in this regard.1,2 Jain and Mittal3 solved the
not able to completely model the intake flow yet. In unsteady Euler equations using a finite element
addition, experimental tests are too expensive and method to compute the flow field of a two-dimensional
could be performed only for special cases, perhaps mixed compression intake. Akbarzadeh and Kermani4
just for validating the computational fluid dynamics also numerically computed the inviscid flow field for
(CFD) results. The main purpose of this study is to
develop a CFD code from the beginning which will 1
Department of Aerospace Engineering, Sharif University of Technology,
be capable of computing the entire intake flow field. Iran
2
Moreover, to validate the newly developed code, the Engineering Research Institute, Sharif University of Technology, Iran
authors conducted extensive experimental tests includ-
Corresponding author:
ing pressure measurements and shadowgraph flow visu- Mohammad R Soltani, Department of Aerospace Engineering, Sharif
alization technique using a newly designed supersonic University of Technology, Azadi Street, Tehran, Iran.
intake at various conditions. The authors could not find Email: [email protected]

Downloaded from pig.sagepub.com at RYERSON UNIV on February 20, 2013

468 Proc IMechE Part G: J Aerospace Engineering 227(3)

different intakes using Roe’s and MacCormack flux The intake performance can be assessed by exam-
schemes. However, ignoring the viscosity in the super- ining different variables. Das and Prasad5–7 selected
sonic intake flow simulation can yield high inaccura- TPR, mass flow ratio (MFR), and FD as the per-
cies. For example, Das and Prasad5–7 numerically formance parameters and investigated the effects of
simulated a supersonic mixed compression intake flow the cowl deflection angle, bleed, and back pressure
using both inviscid normal flow and Reynolds-averaged on these parameters for a mixed compression intake.
Navier–Stokes (RANS) equations via an upwind flux Dawei and Rongwei11 also experimentally investi-
scheme along with the k–! turbulence model. They gated the effects of the free stream Mach number,
showed that the inviscid simulation indicates no start- angle of attack, back pressure, and bleed on the
ing problem, while the simulation with RANS equa- MFR and back pressure ratio (BPR) for a mixed
tions and experimental tests indicated the starting compression intake. Jain and Mittal3 selected TPR
problem for the intake. Kwak et al.8 solved RANS and FD as the performance parameters and studied
equations using Roe and the central difference schemes effects of the back pressure on these variables. Singh
with Coakley’s q–! and Menter’s k–! shear stress tur- Thangadurai et al.9 also selected back pressure and
bulence turbulence models to compute the supersonic free stream Mach number as the design parameters
intake flow field. In addition, Singh Thangadurai et al.9 and TPR, MFR, and FD as the performance vari-
solved RANS equations using a finite volume discreti- ables for a mixed compression intake. In addition,
zation and k–" turbulence model in a supersonic mixed Gokhale and Kumar10 investigated effects of the free
compression intake integrated with a combustion stream Mach number and axial position of the spike
chamber and a nozzle. They studied different phenom- only on one parameter, TPR.
ena that cause various operation modes of a supersonic As noticed, the drag coefficient which is an
intake; subcritical, critical, and supercritical. important parameter in the aerodynamic perfor-
Further, finite difference methods can be used to mance of both intake and the flying vehicle has
numerically simulate the supersonic intake flow. For not been considered seriously. In this research, the
example, Gokhale and Kumar10 solved RANS equa- authors selected the drag coefficient, TPR, MFR,
tions using the explicit finite difference scheme of and FD as the performance parameters. In addition,
MacCormack. They further used Baldwin–Lomax alge- most of the previous investigations were focused
braic Model for computing the turbulent viscosity coef- on the mixed compression intakes, whereas in this
ficient to simulate the flow in a ramjet and scramjet research, an external compression intake is
intake. considered. Furthermore, the free stream Mach
It is evident from the above review of the litera- number, the spike deflection angle, and the BPR
ture that most of the investigations are focused on are selected as the design variables for the paramet-
the mixed compression intakes using the two- ric study. The free stream Mach number and the
equation turbulence models. In this study, a numer- angle of attack for the base case are 2 and 0 ,
ical code was developed that uses Baldwin–Lomax respectively.
turbulence model to solve the RANS equations for This study is considered to investigate the perfor-
an axisymmetric supersonic external compression mance characteristics of an intake from which the
intake. This model is accurate for flows without seri- intake performance curve will be obtained. This
ous separation and it is algebraic, i.e. it does not use curve can be used to control the intake performance
an extra partial differential equation. Thus, it is quickly. Further, this study will help the designer to
numerically efficient and suitable for parametric investigate the effects of the principal design param-
study where the code must be run so many times. eters and redesign some parts of the intake if
The governing equations were discretized by the necessary.
finite volume method and then applied in a struc- One of the parameters that is considered in this
tured grid generated via an elliptic grid generator. research is the effect of BPR variations. This param-
Since the flow is steady, the time step has been cal- eter has significant influences on the combustion
culated with the local time stepping method to accel- chamber pressure disturbances which affects the
erate the convergence and is used to proceed the intake operation considerably. TPR is defined as
calculations until the desired convergence is the ratio of the total pressure at the end of the
achieved, i.e. steady-state solution was attained. intake to that of the free stream. TPR is one of
The convective fluxes are computed by the Roe’s the most important parameters in the intake perfor-
upwind scheme since this scheme is very accurate. mance because it directly influences the vehicle thrust
Viscous fluxes are calculated by a finite volume level. MFR is the ratio of the actual intake mass
method consistent with the overall discretization flow rate to the mass flow rate that an intake can
method. capture, according to Figure 1

Downloaded from pig.sagepub.com at RYERSON UNIV on February 20, 2013

Soltani et al. 469

Figure 1. Spike deflection angle and the flow areas.

Figure 3. Computational blocks and boundary conditions used

in the CFD code.

The calculated drag coefficient in this investigation

includes both the pressure and viscous effects. The spike
deflection angle, , is also shown in Figure 1. The
designed base case parameters of the present intake
are M1 ¼ 2.0,  ¼ 28 , and BPR ¼ 5.56 and the static
pressure and temperature are held constant for all
cases.

Numerical methodology and code

validation
Figure 2 shows a picture and the schematic figure of the
intake that was used in this research. The designed
Mach number of the intake is 2. This Mach number,
M1 ¼ 2.0, was chosen such that the intake will operate
Figure 2. The intake used in this study: (a) mounted in the test
in its subcritical stage due to flow stability problems.
section and (b) schematic of the model. The physical domain has been divided into two
blocks, as shown in Figure 3. In each block, a struc-
tured grid was generated using an elliptic grid genera-
tor. In this grid generator, a system of elliptic partial
differential equations is numerically solved to find the
m_ i Ai location of grid points. The entire intake grid is shown
MFR ¼ ¼ ð1Þ
m_ 1 A1 in Figure 4.
Neglecting the body forces and heating, RANS
equations in terms of the mean flow quantities in two-
dimensional (planar and axisymmetric) conservative
FD presents the flow uniformity at the combustion form are12
chamber entrance and must be as low as possible for a Z Z Z Z Z
@ ~ ~
complete combustion. This parameter is defined as WdA þ Fc ds þ  V~c dA
@t A A
s
Z Z Z ð3Þ
ðPt Þmax  ðPt Þmin
FD ¼ ð2Þ ¼ F~v ds þ  V~v dA
ðPt Þavg A
s

FD is calculated at the exit plane of the intake. where

Downloaded from pig.sagepub.com at RYERSON UNIV on February 20, 2013

470 Proc IMechE Part G: J Aerospace Engineering 227(3)

By the explicit finite volume discretization method,

equation (1) becomes
 4 
ti, j X
W nþ1 n
i, j ¼ W i, j  ðFc Þk sk  ti, j Vc i, j
Ai, j k¼1 i,j
X 4 
ti, j
þ ðFv Þk sk þ ti, j Vv i, j
Ai, j k¼1 i, j

ð7Þ

The flow is steady; however, the time derivative term

in equation (7) is used to update the flow variables until
the steady-state solution is achieved. To accelerate the
convergence, the time step, ti, j, is calculated using the
local time stepping method.
Figure 4. Structured grid in internal and external The vectors Fc and Fv in equation (1) are convective
computational blocks. and viscous fluxes, respectively. They must be evaluated
at the cell face in equation (7). In this research, the
2 3 2 3 convective flux is computed by the Roe’s scheme as
 Vn mentioned previously. Viscous fluxes are calculated by
6 u 7 6 uV þ n P 7 a finite volume method which is consistent with the
~ ¼6
W 6 7,
7 6
F~c ¼ 6
n x 7
7
4 v 5 4 vVn þ nr P 5 overall discretization method. The problem is steady;
thus, the time discretizations have an accuracy of first
E HVn order. In addition, applying the monotone upstream-
2 3
0 centered schemes for conservation laws approach for
6n  þ n  7 higher order of spatial accuracy does not considerably
6 x xx r xr 7
F~v ¼ 6 7 improve the accuracy of the numerical result in the
4 nx xr þ nr rr 5
selected grid size; therefore, the spatial accuracy is
nx x þ nr r
2 3 2 3 also of first order.
v 0 Figure 3 shows the boundary conditions that are
16 7
6 uv 7 16
6 xr 7
7 used in the present code. The stress terms in the
V~c ¼ 6 2 7, V~v ¼ 6 7 ð4Þ RANS equations are computed by the following viscos-
r 4 v 5 r4 rr   5
ity coefficient
vH uxr þ vrr þ k @T
@r

 ¼ L þ  T ð8Þ
and

~ n ¼ nx u þ nr v, nx ¼ r
Vn ¼ V:~ where L and T are the laminar and the turbulent
s viscosity coefficients, respectively. The laminar viscosity
x pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
nr ¼  , s ¼ x þ r2 2 coefficient is the molecular viscosity and in this research
s has been computed by the Sutherland law. The turbu-
@T lent viscosity coefficient, however, has been calculated
x ¼ uxx þ vxr þ k
@x by the Baldwin–Lomax turbulence model. This alge-
@T braic model is based on the Cebeci–Smith model with
r ¼ uxr þ vrr þ k ð5Þ
@r some modifications to avoid the need for locating the
If  ¼ 0, the above equations are for two- edge of the boundary layer. This simple and numeri-
dimensional planar and if  ¼ 1, they are for axisym- cally efficient model has been successfully and widely
metric flows. A is the area of the two-dimensional cell, used for the numerical computation of the intake flow
s the length of the cell face, and Vn the velocity com- field.10,13–16 For all the cases considered in this investi-
ponent normal to the cell face. The equation of state gation, there is no serious flow separation and accord-
which is used as an auxiliary equation is ing to the aforementioned references, especially
reference,15 the Baldwin–Lomax model can precisely
calculate the turbulent viscosity coefficient for flows
P ¼ RT ð6Þ without or with moderate separation.

Downloaded from pig.sagepub.com at RYERSON UNIV on February 20, 2013

Soltani et al. 471

Figure 5. Effect of grid size on the static pressure distribution over (a) cowl surface and (b) spike surface.

Figure 6. Comparison of the numerical and experimental data for M1 ¼ 1.8 and BPR ¼ 5.56: (a) static pressure ratio distribution on
the cowl, (b) Mach number distribution in the outer rake, (c) static pressure ratio distribution on the spike, and (d) total pressure
distribution in the inner rake.

Downloaded from pig.sagepub.com at RYERSON UNIV on February 20, 2013

472 Proc IMechE Part G: J Aerospace Engineering 227(3)

Figure 7. Comparison of the numerical and experimental data for M1 ¼ 2.0 and BPR ¼ 2.82: (a) static pressure ratio distribution on
the cowl, (b) Mach number distribution in the outer rake, (c) static pressure ratio distribution on the spike, and (d) total pressure
distribution in the inner rake.

Grid resolution study was conducted to insure that in each block shows that the aforementioned grid sizes
the numerical solution is independent of the grid size. are sufficient and acceptable.
As shown in Figure 3, the physical domain was divided The tests for the code validation were performed in a
into two blocks. The static pressure distributions over continuous supersonic wind tunnel. The shock wave
the cowl surface, block 1, and over the spike surface, patterns over the intake were recorded via a shadow-
block 2, are plotted for various grid sizes in Figure 5. graph system and a high-speed camera with the record-
From these results, a grid of 100  60 points in block 1 ing speed of 1000 frames/s was used to capture the
and a grid of 400  40 points in block 2 were selected shock phenomenon.
for the present studies (the left number is the number of A picture of the model installed in the wind tunnel
nodes in the x direction and the right one the number of test section was shown in Figure 2(a). The model has a
nodes in the r direction). There exists viscous sublayer fixed geometry with an L/d (length/diameter) of 4.8.
near the wall within y þ < 2–8; thus, it is recommended The designed Mach number of the model is 2 and its
that the first cell centroid should be placed at a distance nose apex semi angle 28 . The mass flow rate passing
y þ4 1 from the wall. However, a higher yþ could be through the intake and as a result the intake back pres-
acceptable as long as it is made sure that this value is sure can be controlled via a plug located at the end of
well inside the viscous sublayer.17 The selected grids are the intake, as shown in Figure 2(b).
sufficiently wall resolving with yþ & 4 to capture the The cowl and spike surface pressure distributions
turbulent phenomena near the walls. In addition, com- were measured via several sensitive pressure trans-
paring the computed and experimental boundary layers ducers located at different positions. In addition, two

Downloaded from pig.sagepub.com at RYERSON UNIV on February 20, 2013

Soltani et al. 473

boundary layer rakes, one (outer) at x/d ¼ 4 (x is mea-

sured from the tip of the spike) located on the cowl
surface and another (inner) located on the spike surface
at the end of the intake were used to verify the capa-
bility of the CFD code in capturing the boundary layer
and turbulent phenomena. To reduce the experimental
errors (such as the instrumentation calibration, pres-
sure transducers, data acquisition system, system
noise, and human errors), each test was performed sev-
eral times and the maximum experimental error was
about 1.5%.
For the code validation, various flow situations and
data positions were considered and a few results are
shown in Figures 6 and 7. In all numerical computa-
tions, the free stream static pressure and temperature
were considered to be constant and the validations were
performed only for different values of the free stream
Mach number and back pressure. The BPR that were
Figure 8. Comparison of the numerical and experimental used in these figures is defined as the ratio of the static
shock pattern for M1 ¼ 2.0 and BPR ¼ 5.56.18 pressure at the end of the intake (back pressure) to that
of the free stream value. The capability of the CFD

Figure 9. Effect of M1 on the performance parameters, BPR ¼ 5.56 and h ¼ 28 . BPR: back pressure ratio; TPR: total pressure
recovery; MFR: mass flow ratio; FD: flow distortion.

Downloaded from pig.sagepub.com at RYERSON UNIV on February 20, 2013

474 Proc IMechE Part G: J Aerospace Engineering 227(3)

code in capturing the shock pattern is also examined

using the shadowgraph pictures, as shown in Figure 8.
From Figure 8, note that the upper surface shock is
predicted from the code, while the lower surface one
is from the shadowgraph system. As seen from
Figures 6 to 8, very good agreements between the
numerical and experimental results are achieved. The
inaccuracy seen in the inner rake data, Figures 6(d) and
7(d), is mostly due to the presence of the model struts
that are not considered in the computational model.
Based on the aforementioned results, the authors were
encouraged to extend the existing code in such a way to
study the effects of various important design parame-
ters and their variations on the intake performance.

Results and discussion

The parametric study began with analyzing the effect of
Figure 10. Intake performance curve, BPR ¼ 5.56 and h ¼ 28. the free stream Mach number on the performance
BPR: back pressure ratio; TPR: total pressure recovery; MFR: parameters, and the results are shown in Figure 9.
mass flow ratio. From Figure 9(a), it is clearly seen that as the free
stream Mach number, M1, increases, TPR decreases

Figure 11. Effect of h on the performance parameters, BPR ¼ 5.56 and M1 ¼ 2.0. BPR: back pressure ratio; TPR: total pressure
recovery; MFR: mass flow ratio; FD: flow distortion.

Downloaded from pig.sagepub.com at RYERSON UNIV on February 20, 2013

Soltani et al. 475

Figure 12. Intake shock position and type and the static
pressure distribution in general.

Figure 14. Intake performance curve, BPR ¼ 5.56 and

M1 ¼ 2.0. BPR: back pressure ratio; TPR: total pressure recov-
ery; MFR: mass flow ratio.

reduction in the additive drag at high Mach numbers.

Further from Figure 9(c), note that for this intake when
operates at the designed Mach number, the drag coef-
ficient has its maximum value. The reason for this
behavior is that external compression intakes should
usually operate at their subcritical Mach numbers.19
Hence, they will not encounter flow stability problems
during the flight.
Figure 9(d) shows the effect of M1 on the intake
FD. It is seen that as M1 increases, the defined flow
non-uniformity, FD, increases too. This is due to
increase in the boundary layer thickness caused by the
interaction of the strong shock wave with the intake
boundary layer.
Figure 10 shows the corresponding intake perfor-
Figure 13. Spike static pressure distribution for different values mance curve with various M1 values for a BPR of
of h, M1 ¼ 2.0, and BPR ¼ 5.56. 5.56. Note that this intake has a reasonable perfor-
mance characteristics for the designed Mach number
which is due to strengthening of the shock waves of M1 ¼ 2.0.
formed both in front and inside the intake. Note that The effects of nose angle on the performance param-
the shock pattern formed ahead of this intake in super- eters are shown in Figure 11. As seen from Figure 11(a),
sonic flow is shown in Figure 8 for both shadowgraph increasing the value of  results in an increase of the
visualization and CFD predicted one. As expected and TPR until  ¼ 30 (maximum TPR) and by further
according to Figure 9(b), increasing M1 increases increase of ,  > 30 , the TPR decreases sharply. To
MFR. This variation is due to the reduction of the describe this behavior, the position and type of the
nose oblique shock angle that will reduce the flow spill- intake shock waves must be first analyzed, Figure 12.
age at higher free stream Mach numbers. According to this figure and Figure 13, when  increases
Figure 9(c) shows the variations of the drag coeffi- from 24 to 30 , the normal shock weakens, whereas
cient with Mach number. It is seen that the drag coef- the strength of the oblique shock remains approxi-
ficient does not significantly change when the intake is mately constant. Note that as the strength of the
operated near its designed Mach number, M1 ¼ 2.0. normal shock decreases, the TPR becomes higher.
The reduction of drag coefficient for M1 > 2.0 is due For  > 30 , a second normal shock is formed around
to the decrease of the spillage flow and consequently the x/d & 0.8 that results in a rapid reduction of the TPR.

Downloaded from pig.sagepub.com at RYERSON UNIV on February 20, 2013

476 Proc IMechE Part G: J Aerospace Engineering 227(3)

Figure 15. Effect of BPR on the performance parameters for h ¼ 28 and M1 ¼ 2.0. BPR: back pressure ratio; TPR: total pressure
recovery; MFR: mass flow ratio; FD: flow distortion.

The reason for the variation of MFR with  seen

from Figure 11(b) is also due to the variation of the
shock pattern of the intake at various  values. As the
normal shock weakens by increasing , the amount of
the flow spillage decreases which results in an increase
of the intake mass flow rate (increasing the Ai values
shown in Figure 1). The rapid reduction of MFR for
 > 30 seen from Figure 11(b) is also due to the second
normal shock that forms inside the intake and prevents
the air to enter the intake.
There is a type of drag for the supersonic intake that is
called additive drag and is schematically shown in Figure
1. The amount of this drag is related to the amount of
flow spillage over the intake. Figure 11(c) shows the var-
iations of the drag coefficient with . It was mentioned
before that as  varies, the strength and pattern of the
intake shocks vary too. As a result of these variations,
the amount of the flow spillage varies for different 
values; thus CD varies, as shown in Figure 11(c).
As shown in Figure 11(d), FD is not significantly
Figure 16. Spike static pressure distribution for different values affected by the variation of . The variation as seen
of BPR, M1 ¼ 2.0, and h ¼ 28 . BPR: back pressure ratio. from Figure 11(d) is less than 0.5% for all  values

Downloaded from pig.sagepub.com at RYERSON UNIV on February 20, 2013

Soltani et al. 477

Figure 17. Mach number contours and velocity vectors inside a portion of the intake, BPR ¼ 2.82.

Figure 16. As seen from this figure, for BPR values of

2.82 and 5.19, a second normal shock is formed inside
the intake. This normal shock causes an additional loss
in the total pressure and as a result, the TPR will decrease.
In fact, when the back pressure becomes low, the flow
reaccelerates inside the intake that causes an additional
normal shock and consequently an additional loss.
According to Figure 16, as BPR increases, the
normal shock outside the intake moves upstream. In
this process, the flow spillage increases and MFR
decreases, as clearly seen from Figure 15(b). This
figure also shows that the intake mass flow rate is
very sensitive to the BPR values of 5.0 to 6.0.
The spike normal shock moves upstream and its
strength increases for BPR > 5.19; however, the static
pressure on the spike increases, as seen from Figure 16.
This phenomenon increases the drag coefficient of the
intake for BPR > 5.19 which is clearly shown in Figure
Figure 18. Intake performance curve (BPR varying) for h ¼ 28 15(c). In addition, the drag coefficients for values of
and M1 ¼ 2.0. TPR: total pressure recovery; MFR: mass flow BPR ¼ 2.82 and 5.19 are approximately identical,
ratio. Figure 15(c).
Figure 15(d) shows that FD does not vary much for
examined here. Figure 14 shows the performance curve different values of the BPR except for the lowest BPR,
of the intake for various  values. According to this BPR ¼ 2.82. In fact for this value of BPR, BPR ¼ 2.82,
figure and Figure 11, it is concluded that the intake an additional normal shock is located at about x/
has a better performance for  ¼ 30 and its drag coef- d & 2.6 which will cause a large separation region, as
ficient for this value of  decreases up to about 9% seen from Figure 17. This large separation is obviously
compared to the designed base case,  ¼ 28 . followed by a loss in the total pressure and will create a
Therefore, based on these information, it is recom- non-uniform total pressure distribution at the intake
mended to redesign the spike geometry with  ¼ 30 exit plane, both of which will cause the large value of
and consider other design requirements. FD seen from Figure 15(d).
The effect of variations of BPR on various perfor- Figure 18 shows the intake performance curve for
mance parameters is shown in Figure 15 for a constant various values of BPR. According to this figure for
Mach number of M1 ¼ 2.0 and  ¼ 28 . From Figure the designed value of the BPR, BPR ¼ 5.56, the intake
15(a), it is seen that as BPR increases, TPR increases has its optimum performance. It also shows that at a
too. To explain this variation, consider the intake very low value of the BPR, BPR ¼ 2.82, this supersonic
shock pattern for different values of BPR presented in intake does not operate efficiently.

Downloaded from pig.sagepub.com at RYERSON UNIV on February 20, 2013

478 Proc IMechE Part G: J Aerospace Engineering 227(3)

5. Das S and Prasad JK. Unstart suppression and perfor-

Conclusions mance analysis of supersonic air-intake adopting bleed
A CFD code was developed that is enable to compute and cowl bending. IE (I) J–AS 2010; 91: 27–35.
the flow field around and inside a supersonic axisymmet- 6. Das S and Prasad JK. Starting characteristics of a rect-
angular supersonic air-intake with cowl deflection.
ric external compression intake and the result was vali-
Aeronaut J 2010; 114(1153): 177–189.
dated with the wind tunnel data of an intake designed to
7. Das S and Prasad JK. Cowl deflection angle in a super-
operate at a free stream Mach number of 2.0.
sonic air intake. Def Sci J 2009; 59(2): 99–105.
Experimental studies for this specific intake were also 8. Kwak E, Lee H and Lee S. Numerical simulation of flows
performed by the authors and a good agreement around axisymmetric inlet with bleed regions. J Mech Sci
between the numerical and experimental results were Technol 2010; 24(12): 2487–2495.
achieved. Once the validation was insured, a parametric 9. Singh Thangadurai GR, Subhash Chandran BS, Babu V
study for optimizing the intake was performed. For this and Sundararajan T. Numerical investigation of the
study, the effects of the free stream Mach number (M1), intake flow characteristics for a ramjet engine with and
spike deflection angle (), and BPR on the TPR, MFR, without heat addition in the combustion chamber. Def
FD, and the drag coefficient of the intake were investi- Sci J 2004; 54(1): 3–16.
gated. The results show that M1 does not affect the 10. Gokhale SS and Kumar VR. Numerical computations of
intake drag coefficient and FD significantly; however, supersonic inlet flow. Int J Numer Methods Fluids 2001;
36: 597–617.
it has a strong effect on the TPR and MFR. In addition,
11. Dawei W and Rongwei G. Experimental investigation of
for a specific range of  ( < 30 ), an improvement in the
a fixed-geometry two-dimensional mixed-compression
intake TPR and MFR as well as a reduction in the drag supersonic inlet with sweep-forward highlight and bleed
coefficient are seen all of which are due to the weakening slot in an inverted ‘‘X’’-type layout. Chin J Aeronaut
effects of the normal shock formed at the inlet of the 2007; 20: 304–312.
intake. However, for  > 30 , the reverse effects on the 12. Soltani MR, Farahani M and Sepahi Younsi J.
performance parameters were noticed which is believed Performance study of a supersonic inlet in the presence
to be caused by the additional normal shock formed of a heat source. Sci Iran B 2011; 18(3): 375–382.
inside the intake. The spike deflection angle for the 13. Smith CF, and Smith GE. Two stage supersonic inlet
designed intake is about  ¼ 28 ; however, this paramet- (TSSI): 10-inch model calculations. NASA technical
ric study reveals that the spike should be substituted with paper no. CR-2005-213287, 2005. Washington, DC: The
the one having a deflection angle of  ¼ 30 . This inves- National Aeronautics and Space Administration.
tigation also shows that for the low values of BPR, an 14. Lu PJ and Jain LT. Numerical investigation of inlet buzz
flow. J Propul Power 1998; 14(1): 90–100.
additional normal shock is formed inside the intake. The
15. Sakowski B, Darling D, Roach RL, et al. Evaluation and
interaction of this normal shock with the boundary layer
application of the Baldwin-Lomax turbulence model in
causes a large flow separation region inside the intake. two-dimensional, compressible boundary layers. NASA
As a result, a large loss as well as a non-uniform distri- technical paper no. TM-105810, 1992. Washington, DC:
bution of the total pressure is created which reduces the The National Aeronautics and Space Administration.
TPR and increases FD. Increasing the value of BPR also 16. Kumar A. Numerical simulation of scramjet inlet flow fields.
causes the outside normal shock to move upstream and NASA technical paper no. 2517, 1986. Washington, DC:
becomes strong which will increase the intake drag coef- The National Aeronautics and Space Administration.
ficient and reduce the intake mass flow rate. 17. Blazek J. Computational fluid dynamics: principles and
applications. Chap. 7. London: Elsevier Science, 2001.
Funding 18. Soltani MR, Sepahi Younsi J and Farahani M.
This project was partially supported by the Engineering Investigation of a new flux scheme for the numerical sim-
Research Institute. ulation of the supersonic intake flow. Proc IMechE, Part
G: J Aerospace Engineering Epub ahead of print 8
References December 2011. DOI: 10.1177/0954410011422634.
1. Liou MS. A sequel to AUSM, part II: AUSMþ-up for all 19. Seddon J and Goldsmith EL. Intake aerodynamics, an
speeds. J Comput Phys 2006; 214: 137–170. account of the mechanics of flow in and around the air
2. Kim KH, Kim C and Rho OH. Methods for the accurate intakes of turbine-engined and ramjet aircraft and missiles.
computations of hypersonic flows I. AUSMPWþ scheme. London: Collins, 1985.
J Comput Phys 2001; 174: 38–80.
3. Jain MK and Mittal S. Euler flow in a supersonic mixed- Appendix
compression inlet. Int J Numer Methods Fluids 2006; 50:
1405–1423. Notation
4. Akbarzadeh M and Kermani MJ. Numerical simulation of A grid cell area, flow area
inviscid airflows in ramjet inlets. Trans Can Soc Mech Eng D
2009; 33(2): 271–296. CD drag coefficient ¼ 1 2
2 1 V1 S


Downloaded from pig.sagepub.com at RYERSON UNIV on February 20, 2013

Soltani et al. 479

d model diameter Subscripts

E total internal energy
c convective flux
H total enthalpy
i, j axial and radial counters of the grid cells
k thermal conductivity coefficient
r radial direction
m_ mass flow rate
n normal direction of the cell face
M Mach number
v viscous flux
n magnitude of the normal vector to the
x axial direction
cell face
P static pressure
 circumferential direction, spike deflection
Pt total pressure
angle
r radial coordinate
s length element along the cell face
t time Superscripts
T static temperature
n time level in discretization
u axial component of the velocity
v radial component of the velocity
x axial coordinate

" change of variable

 air density
 stress

Downloaded from pig.sagepub.com at RYERSON UNIV on February 20, 2013