NASA Technical Reports Server (NTRS) 20040046889: Shock Waves Mitigation at Blunt Bodies Using Needles and Shells Against a Supersonic Flow

5.7 MB·English

by  NASA Technical Reports Server (NTRS)

. . . *..
. . .
. .
. .
. . . .
..
. .
. .
. . .
. . .
.
..'...:
. . . : ',NASA: .. .. . . ...
..... . ' .. .. .. . ... ... ... . . . .
'. . . , . . ... .. .
.....
. . ..
. . . . .
... . ... . ... .. ..
1
Shock Waves Mitigation at Blunt Bodies Using Needles and
Shells against a Supersonic Flow.
M. Gilinsky*
Hampton University, USA
I.M. Blankson**
NASA Glenn Research Center, USA
V.I. Sakharov***
and
A.I. Shvets****
Institute of Mechanicshloscow State University, Russia
ABSTRACT
The paper contains some experimental and of Mechanics of Moscow State University,
numerical simulation test results on cylindrical Russia.
blunt body drag reduction using thin spikes or
shell mounted in front of a body against a I. INTRODUCTION
supersonic flow. Experimental tests were A thin spike mounted in front of a blunt body
conducted using the Aeromechanics and Gas can be used to decrease the drag and the heat
Dynamics Laboratory facilities at the Institute transfer in a high oncoming flow velocity. The
of Mechanics of Moscow State University spiked blunt body at an attack angle decreases
(IMMSU). Numerical simulations utilizing not only the drag-although not so effectively
NASA and IM/MSU codes were conducted at as the case of zero attack angle-but also
the Hampton University Fluid Mechanics and increases the lift. Therefore, a nose spike may
Acoustics Laboratory. The main purpose of be used for take-off and other flight
this research is to examine the efficiency of conditions. By varying the length and diameter
application of multiple spikes for drag of the protruding probe (spike or needle), flow
reduction and flow stability at the front of a characteristics can be controlled. The thin
blunt body in different flight conditions, i.e. straight probe is a handy and simple control
Mach number, angle of attack, etc. The device, compared to variable geometry skirts.
principal conclusions of these test results are: However, its applicability is somewhat
multiple spikeheedle application leads to limited, unless a proper arrangement is made
decrease of drag reduction benefits by to eliminate the unfavorable effect of flow
comparison with the case of one central oscillation on the probe, which may cause an
mounted needle at the front of a blunt bodyi aerodynamic disturbance. The summary of the
but increase lift benefits. The research was early experimental and theoretical papers
conducted under the NASA Faculty Research where flow past a body with a spike was
Award for Hampton University, Hampton, studied is contained in the well-known book
VA, grant # NAG-3-2422, grant # NAG-3- by P.K. Chang (19 70) and in the Russian book
2495, and under the Civil Research [l]. Six regimes of supersonic flow past a
Development Foundation (CRDF) grant, # cylinder with conical nose and a spike
RE1-2068, for joint research between the protruding from it are distinguished for
NASA Glenn Research Center and Institute specified Mach and Reynolds numbers of the
oncoming flow which includes the regime of
* Research Professor, Senior Member AIAA, USA stationary flow with drag reduction and the
**
Senior Scientist, AIAA Associate Fellow, USA
regimes of unsteady pulsating and fluctuating
*** Leading Senior Scientist, Russia
**** Professor, IM/MSU Laboratory Chief, Russia flows.
2
Since 1950, the application of spike-nosed and numerical simulations were conducted at
blunt bodies has drawn the attention of the Hampton University FM&AL during the
scientists. In the USA and the former USSR, a visit of two IM/MSU scientists to the USA for
number of researches were conducted in this joint research.
area. In the first group of experimental tests,
the main aerodynamic characteristics of 2. EXPERIMENTAL SETUP
different blunt body shapes with needles were The main purpose of this research is to
obtained. In the following stage, numerical estimate the influence of the application of
simulations of this problem were conducted. several needles mounted at the front of a blunt
Significant attention was directed to body against a supersonic flow by comparison
ex?erimenta! and numerics! tests ef differe~t with the kzditicnsl one Eee-J!e Uonynyl"; CgUtC;InVm'I .
oscillatory regimes of the flow at the spike- Experimental tests were conducted using an
nosed blunt bodies that were observed in some axisymmetic cylindrical model with a flat
range of needle length values. forward part, Le. butt-end, of diameter,
D=80mm. Two types of models were tested: a)
The main cause of this phenomenon is with 1 needle (n=l) and b) with 5 needles
connected with flow separation at the needle (n=5). In both cases, one central needle was
or body points of some essential curvature mounted at the center of the circular front flat
changes. In the common case, the flow is not part of radius, R=40mm. For the second series
always steady, and may be periodically of models, another 4 needles were placed
unsteady. Two distinct modes of instability symmetrically around the central needle. They
have been observed. In the "pulsationyym ode, were placed at the angular interval of 90" and
the conical separation bubble formed on the at the radial distance, ~ 0 . 5R . Non-
concave part of the body periodically inflates dimensional needle lengths, L/D, were chosen
and expands in the radial direction, taking a based on constructive opportunities of
hemispherical shape. In the "oscillationy' application, and on known referenced data in
mode, a conical foreshock envelops the this field. These data show that maximum drag
separation bubble and the accompanying shear reduction using needles at supersonic speeds
layer oscillates laterally. Its shape changes can be achieved in the range of needle lengths,
periodically from concave to convex. The L/D=1.5-2.0, so that three needle lengths were
pulsation mode was first observed by Mair [2] manufactured and tested: L/D=1.5, 1.0, and
and the oscillation mode was observed by 0.5. For this needle length range, the end of the
Bogdonoff and Vas [3]. The terminology is needle at one limit, L/D =1.5, is located farther
due to Kabelitz (1971). A comparison shows from the butt-end than the detached bow shock
.that the oscillating flow around a concave wave which occurred in the case of flow
body is a sub-case of the self-sustained without a needle at the body front. For the
oscillations of impinging free shear layers other limit, L/D= 0.5, the needle is located
reviewed by Rockwell and Naudascher [4]. completely inside the compression shock
layer, i.e. closer to the butt-end than the
In the next sections, we will present the main corresponding detached bow shock wave.
experimental tests and numerical simulation Before the experimental tests, we intended to
results recently obtained by the authors in this manufacture only one needle set with needle
area with the purpose of improving the thickness (diameter), d=O.lD. However,
efficiency of some propulsion systems. during the first experiments, it became
Experimental tests were conducted in the obvious, that in this case, n=5, the needle-butt-
IM/MSU aerodynamic wind tunnel, A-7 [5], end model too thick was and creates the single
3
shock wave in front the spikes. On the other The pressure gauges IKD-27 were employed
hand, needle geometry (thickness, length, for measurement of the free stream total
shape) can play the main role in the pressure. Static and total pressures were also
organization of a flow circuit at the body. It measured. The relative root-mean-square error
was assumed that when the forward circuit of pressure measurement <p> was -0.0 1 at the
separation zone forms at the single needle, the upper measurement limit. Automatic
shock wave has conical or ogival shape. For processing of these measurements was carried
several needles, another shock wave geometry out by the computer complex and printed. The
should appear, similar to the case of Mach number in the test area of the wind
supersonic flow around star-shaped bodies. tunnel is close to three, M=2.97, and Reynolds
hI 1.ou-Pll l~~wyi,th ?dach type or regul~rif iterac:ion number, Re=! .6x!06, whzre Reyiiolds number
of shock waves, similar to a flow around is calculated on the free stream parameters and
several forward separation zones, or to model diameter. Two types of visualizations of
formation of a uniform forward separation the stream were used: a) with large exposure
zone. As a result of this analysis, the following time, t= O.Olsec, that considerably exceeds the
needle geometric parameters were chosen: DL duration of a pulsation cycle. These depict
= 0.05; 0.81; 0.1, i.e. for the given model size, average flow field features in time. And b)
the needle diameters are: d=4; 6.5; and 8mm. with small exposure time, t=lO%ec, that
allows the observation of flow fluctuations and
A supersonic wind tunnel with permeable test shock wave movement.
section wall was employed for the
experimental tests. This wind tunnel has the III. EXPERIMENTAL RESULTS
transonic speed range with continuous 3.1 It is known that for supersonic gas flow
transition through to supersonic speeds. The with Mach number, M=3, around the
wind tunnel with test section size, 0.6 x 0.6m, cylindrical butt-end, the pulsation regime that
is equipped a pressure and suction ejector first was observed by Mair [2], occurs in a
allowing the conduction of experimental tests range of needle lengths about L/D=O.4-1.7
in a wide range of Mach number, M=0,4-4,O (Maul1 [6]). In this case, pulsations of "the
and Reynolds number, Re=1x105-2.5~10~~ second sort" (Antonov, Shalaev [7]) take
where Reynolds number is calculated relative place. The body shape is the determining
to lm. The aerodynamic forces acting on the factor as to which pulsations occur. Therefore,
model are measured on a four-component for the butt-end model with only one needle,
tenzometric scale of the 19mm diameter TV- such pulsations could be expected in the
128 type. A tenzoamplifier device is a tracking present experiments. In Mair's exP eriments,
.system controlling a voltage value from the the pulsation frequency was 6x10 sec-' for
measuring diagonal tenzometric scale that is Mach number, M=1.96, and in Maull's
cemented on the elastic elements of the model. experiments, for Mach number, M=6.8, it was
Estimation of errors has shown that the total 2x104 sec-'. In the present methodical set of
root-mean-square measurement error for the experiments the pulsation regimes were also
aerodynamic coefficients are <C, >=0.03 for observed: different pulsation phases were seen
the axial force coefficient (if a=O" then on the shlieren photos of the flow field and
C,=CD), for the normal force coefficient, measured forces had oscillatory character as
<Cn>=0.02, and for the longitudinal moment shown by the tenzoscales. A consecutive
coefficient, <m =0.02. registration of forces at the specific
z>
frequencies was impossible because the
measuring instrument was too inertial.
4
However, some estimations of the fluctuation and vs needle diameter for different needle
ranges for various force components were lengths is shown in Figure le. In the last
obtained. figure, experimental data obtained by Hunt for
a single needle are also plotted. Comparison
3.2 All experimental test results presented in of this coefficient for single needle and five
this paper were obtained in the IM/MSU wind needle models vs needle length for the needle
tunnel for fixed free stream Mach number, diameter, d/D=0.082 is illustrated in Figure If.
M=3. In Figures 1a -f, the drag coefficient, CD, Average values for different force coefficients,
vs different geometric parameters of the drag-CD, normal-CN, and longitudinal
needles are shown for a zero attack angle, momentum relative to the point, x=l, vs angle
-m=n ou . Ifi Figure la, the CD coefficient is of ZPtCk, G,E ire Showii iii Figires 22-c fGr the
plotted for the case M=3, n=l, d/D=0.082, single needle model, and in Figures 2d-f, for
L/D=l , and EO.' These values were measured the 5-needle model.
during one experiment with the equal time
interval (k is a registration number). One can 3.3 Several additional experimental tests were
see that there is an oscillatory change of CD conducted with the purpose to determine the
coefficient that reflects the existing pulsation influence of multi-needle (n>5) application to
range qualitatively. The tenzoscales employed the force coefficients. Figures 3a-d. The limit
can not register each fluctuation cycle (the case for infinite needle number is illustrated in
scales own fluctuation frequency is near 100 Figure 3a,b. The drag coefficient for the solid
Hz). Therefore it does not show exact values shell mounted to the front butt-end by the 5
of forces in the extrema1 points of the needles is plotted vs shell length (a) and angle
fluctuations. Nevertheless, when there are not of attack (c). In Figure 3b, this coefficient is
regular fluctuations as in the case of a shown for large finite needle number, n=53
cylindrical butt-end without p y needle, for and n=103. These needles are mounted
example, the measured CD values during one symmetrically along the coincident circles
experiment change not more than -<0.03>, i.e. located from the circle center on the radial
within the scales error of measurement. distances: R=0;5;. .. 35mm. Recall that the
Comparison of fluctuation results for two butt-end circle diameter is D=SOmm. A drag
models with 1 and 5 needles shows that in the coefficient comparison for all considered
last case (n=5), the CD deviations from its cases, n=l; 5; 53 and 103, is shown in Figure
average value are much less than for a single 3d that illustrates a preferable application for
needle (n=l). drag reduction of single needles.
In Figure lb the average values of the CD 3.4 Optical methods of flow visualization were
coefficient (black symbols) and its oscillatory employed for better understanding of this
values (white symbols) are shown for the phenomenon and for explanation of the force
model with 5 needles (n=5) with different aerodynamic characteristics changes.
lengths, L/D=O.O; 0.5; 1.0; and 1.5 and They include a traditional shlieren method and
different needle diameters, d/D =0.05, 0.082, high resolution video filming of the unsteady
and 0.1. For the same needle diameters, this pulsation regimes. In Figure 4a-I, the main
coefficient value vs needle number is models tested in supersonic flow with Mach
illustrated in Figure IC for the needle length, number, M=3, at zero angle of attack are
D/L=l. Comparison of CD coefficient for shown: a)-butt-end without needles; b),c) -with
single needle (n=l) and five needle (n=5) single needle for two different instants of
models vs needle length is shown in Figure Id, pulsation flow regime; d)-f) -for 5 thin needles
5
model of thickness, d/D=O.Od, and different calculated with a space-centered scheme. The
needle lengths, L/D=0.5;1.0 and 1.5; and g)-i) boundary conditions on the body surface and
-for 5 thick needles model of thickness, on the shock wave are approximated in the
d/D=O.l with the same as previous needle context of the general conception of a finite-
lengths. volume method with the second order of
accuracy. A steady solution of the difference
IV. NUMERICAL SIMULATION equations is found by means of an iterative
RESULTS procedure based on solving pseudotime-
4.1 A finite-volume implicit numerical method dependent VSL equations with an implicit
was developed to solve the two-dimensional scheme. The flowfield calculation at every
aiid axisymmetric time-dependent Xavier- iiine step is perfomled by the Gaiiss-Seidel
Stokes equations (NSE) in the conservative line relaxation numerical technique. Based on
form. It was assumed that the flux vector could this approach, the numerical code was
be split into 'viscous' and 'inviscid' parts. The developed by V. G. Gromov with colleagues
solution of the Riemann problem of heat- [8,9] at the IM/MSU for the MS platform and
conducting gas with frozen chemical used for several research projects. Note that
composition has been obtained for a this approach is similar to the approach
computation of 'inviscid' fluxes. Spatial employed in the NASA numerical code
derivatives in the 'viscous' terms are VULCAN.
approximated with second-order accuracy.
Piecewise-parabolic distributions of the 4.2 The purpose of numerical simulation of
physical variables over the network cells and this problem is to check some experimental
'minmod' limiters lead to a TVD-scheme of test results and to obtain more exact
second-order accuracy. The finite-difference aerodynamic characteristics for pulsation flow
equation set is resolved by the sweep method regimes. Several numerical simulation tests
along the lines normal to the surface and the were conducted for axisymmetric supersonic
Gauss-Seidel iteration procedure. Coordinate- flow at the butt-end with the single needle (1-
oriented differences are introduced in the model). Three examples of such test results are
implicit part of the finite-difference operator. shown in Figures 5a-f where Mach (a-c) and
These are chosen in accordance with signs of pressure (d-f) has been plotted for the
the eigenvalues of Jacobi matrices in the relatively thick needle, d/D=O. 1, and different
convective terms. A new finite-difference needle lengths, L/D=O.9; 1.0; and 1.6. These
scheme for solution of the viscous shock layer results were based on NSE simulation with the
(VSL) equations with multicomponent k-w turbulence model. In this case, steady
diffusion and nonequilibrium chemical solutions were obtained. One can see the
reactions in gas and on the body surface is presence of two separation zones at the corner
developed due to the finite-volume approach. points of the needle and main butt- end, as
The flux-difference splitting method based on well as the vortical flow at the needle lateral
the linearized solution of the Riemann problem surface. The drag coefficient calculated from
in isoenergetic approximation is applied for the numerical simulation for the case tested
calculations of the inviscid fluxes through the experimentally, d/D=O.l; L/D=l .O, is equal to
interfaces of cells. Introducing into the CD= 0.52 that is very close to the experimental
expression for the fluxes limited anti-diffusion value.
corrections provides realization of the TVD
conditions and second order of accuracy of the However, these numerical simulation results
smooth solution. The viscous fluxes are are very sensitive to changes of numerical
scheme parameters or applied models. For
6
example, for the same conditions in inviscid like to thank Dr. Jay C. Hardin for his attention,
flow (Euler approximation) or for a laminar interest in our research, review and useful
viscous flow, numerical simulation results suggestions.
show periodic oscillatory variation of the
flow. An analysis of such solution leads to the VII. REFERENCES
conclusion that mass flow rate pulsation takes 1. Shvets A.I., Aerodynamic of Supersonic Shapes,
place. Such pulsatile flow at the needle is 1987, Moscow State University Press, Moscow.
2. Mair, W., "Experiments on Separated Boundary Layers
shown in Figure 6a-n for the free stream Mach
on Probes in Front of Blunt Nosed Bodies in a Supersonic
number, M =3. In these pictures, Mach
Air Stream," Philosophy Magazine, Vol. 43, 1952, pp.
contours are plotted for different instants and 695-71 6.
show different shock wave positions and flow J2. RYnVnbAunmVn.l VSff. M., Vzs 1. E. ?:e!imhq investigatiens
structure during one oscillation cycle. of spiked bodies at hypersonic speeds.- J. Aerospace Sci.,
1959, V. 26, Ne 2, p. 65-74.
4. Rockwell, D. and Naudasehar, E., "Review of Self-
V. CONCLUSIONS
Sustaining Oscillations of Flow Past Cavities,"
As a result of these experimental and ASME Journal of Fluids Engineering. 100. 1978, pp.
numerical simulation tests, the main 152165.
conclusions were made. 1) Application of 5 5. Aerodynamic Facilities at the Institute of Mechanics
MSU, 1985, MSU Publishing House.
needles instead of a single needle mounted at
6. Maull, D J , "Hypersonic Flow Over Axially
the front butt-end leads to growth of all force
Symmetric Spike Bodies," Journal of Fluid Mechanics,
coefficients at zero angle of attack of the tested Vol. 8, Pt 4, Aug 1960, pp 584-594,
model as well as at almost all angles of attack, 7. Antonov A. N. , Shalaev N. E. Some features of non-
stationary separated currents on bodies with the
a 45 ". Especially, such growth is observed for
established ahead needle. The mechanics of Fluid and
the relatively small needle lengths (L/D=0.5)
Gas, 1979, N 1, c. 97.
at zero angle of attack. 2) Increase of the 8. Levin V.A., Afonina N.E., Gromov V.G., Navier-
needle number leads to some flow Stokes Analysis of Supersonic Flow with Local Energy
stabilization, i.e. oscillation amplitude Deposition", AIAA-99-4967, Nov. 1999
9. Karlovskii, V.N., and Sakharov, V.I. , 1986,
decreases for the 5-needle model by
Numerical investigation of supersonic flow past blunt
comparison with the single needle model. 3)
bodies with protruding spikes, 1986, Fluid Dynamics (Izv.
Normal force coefficient, CN, for the 1 needle Akad. Nauk SSSR - Mekh. Zhidk. Gaza), V01.21, No.3,
model is essentially less than for the 5-needle pp.437-445
model, and this value grows almost
monotonically with increase of angle of attack
for the 5-model, and non- monotonically for 1-
model with maximum value at angle of attack,
a=1 0 ".
,
VI. ACKNOWLEDGEMENTS
We would like to acknowledge the NASA
Glenn and Langley Research Centers,
especially Dr. Dennis M. Bushnell, Charles W.
McClinton, David W. Lam, and Curt Snyder
from the Naval Air System Team for interest
and support of our research. This research was
partially conducted under the NASA grants:
NAG-3-2422 and 2495, and under the
sumortinp CRDF grant REI-2068. We would
Be.1.U M-1.8.IJD-a 7
IM (A-8). M-3. UD-G
1.5
---n ---
CD 0.082
1. o
0.5
D
0.0
0 2 4 6 8 k 10 0.0 0.5 1. o L/D 1.5
1.5 1 1.6
CD C D
1.3
1. o 0.8
n=l n=S
0.0
0.0
0.00 0.05 d/D 0.10
1 .o
C D
C D
C C 0.8
0.6
0.4
0.0 0.5 1.0 1.5
Fig.1 Drag coefficient, CD,f or 1- and 5-needles butt-end models vs needle length,
L/D, and diameter, d/D; D is circle diameter of front butt-end.
1.6 I 8
C D
1.2
1
d)
0.8
r
i f-----Y I
a - o-082 I I j l
0.4 0.6
0 5 10 15 "." 0, 5 10 15
0.09
C N
0 - 0.0s I 1
-
A 0.082 I I
0.06
0.03
0.00
0 5 10
0 5 10 15
-
1.5 0.9
C m C m
(x = 1) (x = 1)
1. o 0.6 -
0.5 :I
//
O0 --? 0.05 ","I I
1
1
-
0.0 0.0
ao
0 5 0 5 10 a0 15
lo l5
Fig.2 Aerodynamic characteristics of 1- and 5-needle butt-end tested at the IM/MSU Wind Tunnel A-7.
M=3, a=O, L/D=0.5 M=3, d/D=0.025, L/D=0.438
n= 53 R=0,5,15,25,35 mm
n=103 R=0,5,10,15,20,25,30,35 mm
1.50
CD
1.35
1.20
1. OS
0.90
I .45 1 1
0.75
l/D 0 5 10 a 0 15
0.0 0.1 0.2 0.3 0.4 0.5
a)
M=3, d/D=0.475
1/D L/D n d/D L/D
0 -0 -0 o 103 0.025 0.438
A -0.125 -0.5 A 53 0.025 0.438
V -0.25 -0.5 V 1 0.05 0.5
0 -0.5 -0.5 5 0.05 0.5
1.6
0.0 2.5 5.0 7.5 10.0 12.5a o1 5.0 0 5 10 ao1 5
a
c)