Professional Documents
Culture Documents
x
+
G
y
+
H
(1)
here
Q =
u
v
w
E
, F =
u
u
2
+ p
uv
uw
Hu
G =
v
uv
v
2
+ p
vw
Hv
, H =
w
uw
vw
w
2
+ p
Hw
QdVol +
N ndS
1
Re
N
v
ndS =0 (2)
with
N =Fi +Gj +Hk
N
v
=F
v
i +G
v
j +H
v
k
Fig. 1. Gas boundary of imagination for a cross section.
Then we apply Eq. (2) to the given computational mesh system.
To solve these equations, the nite volume scheme with explicit
multi-stage RungeKutta time stepping formulated by Jameson et
al. [22] is used here. In order to accelerate the computations, the
implicit residual smoothing technique with local variable parame-
ters [16] is applied in the present paper.
2.3. Boundary condition
(1) At the solid surface boundary: no-slip and no thermal conduc-
tion
u
w
= v
w
= w
w
=0,
T
n
w
=0
(2) At the outside boundary: the characteristic variable boundary
conditions [16] are used here.
2.4. Turbulent model
The BaldwinLomax eddy viscosity model with the modica-
tions suggested by Degani and Schiff [11] was used here. The B-L
model [4] is simplistic and it is known that the model can perform
well for attached or little separated ows, but it leads to large
inaccuracies for high angle of attack and largely separated ows.
Previous studies have examined the effects of turbulence models
on moving boundaries see, for example, Barakos and Drikakis [6].
For unsteady ow computation, the CPU time per step doubles
when a two-equation model is employed instead of an algebraic
one on oscillating and ramping aerofoils in the study [6]. Here the
B-L model was used for two reasons, one is that the angle of at-
tack range at which the missile oscillate in pitching and the mean
angle of attack are small in the paper, the other is that it will not
lead to long computing times.
2.5. Grid generation
An OH grid is generated by combining a series of two-
dimensional meshes. Each mesh is generated previously by confor-
mal mapping technique in planes normal to the longitudinal axis
of the missile. In two-dimensional grid generation processes, the
leading edge of the wing is stretched outward to be a boundary of
imagination, called gas boundary, shown in Fig. 1, in order to get a
high quality of the grid. The gas boundary has a thickness of zero
in the computational process.
3. The results and analysis
3.1. The body of revolution
To validate the computational results, the surface pressure of
a body of revolution at different angles of attack is studied rst,
and the results are compared with the experimental data [26].
The body had an ogival nose 3 diameters long (3D), tangent to a
cylindrical afterbody 7.7 diameter long (7.7D). The dimensions and
details of the model are shown in Ref. [26]. The origin of Cartesian
coordinates is at the nose vertex of the body, and x axis coincident
with the body axis, and =0
. No
corrections were applied to the data to account for this deection.
Wind-tunnel wall corrections were considered negligible because
the model was small compared to the test section.
A grid independence study was performed in order to ensure
that the mesh was rened suciently. An OH grid of 86 120
90 in the axial, circumferential, and radial directions respectively,
was used as the coarsest mesh. The grid was generated for a half-
body. The distance from the wall to the rst cell center was less
than 2 10
6
D. In addition, the grid was clustered in the vicinity
of the body nose in attempt to accurately capture solution varia-
tions in that region. The second grid was generated by adding 40
nodes in the circumferential direction to the coarsest mesh. Finally,
the nest mesh used for the study had 200 nodes in the circumfer-
ential direction. No additional grid points were added in the axial
direction for this mesh because of its accuracy, as well as our ex-
perience in using it for aerodynamic loads computation in similar
case before this study. The grid independence study was performed
with the body of revolution at =15
, 15
, and 20
=0.3).
132 L. Yang et al. / Aerospace Science and Technology 15 (2011) 129136
Fig. 2. (continued)
Fig. 3. Normal-force distributions along the body length.
Fig. 4. Normal-force coecients of the body vs .
Table 2
Difference between the mesh independence calculations.
C
A
C
N
86 160 90 cells 0.31% 1.30%
86 240 90 cells 0.22% 0.19%
Fig. 5. The surface grid of the missile.
Fig. 6. Crosswise section grid at the wing.
Fig. 7. Crosswise section grid at the tail.
L. Yang et al. / Aerospace Science and Technology 15 (2011) 129136 133
Fig. 8. Lift, drag and pitching moment coecient vs for different reduced fre-
quency.
3.2. The cruise missile
In this case, a generic cruise missile oscillates in pitching with
(t) =
m
m
cos()
where
m
is the mean angle of attack, which was varied to ob-
tain results for three ranges of angle of attack, 0
, 0
, and 0
, that is
m
equal to 0.5
, 1.5
, and 2.5
,
respectively. is the phase angle of oscillation
=t, =2 f , f =
r
k
Here r
k
is the reduced frequency. This pitching function was used
since it produces a motion without any discontinuities in acceler-
ation at the beginning and end of the motion, thus being easier to
implement in a CFD programme.
The missile oscillates in pitching about three longitudinal loca-
tions on the conguration, x/L =1.6, x/L =1.73 (about 1/4 chord)
and x/L =1.912, here L is the half span of the missile (shown in
Fig. 5). Three values of reduced frequency are used: 0.336, 0.5 and
0.628. These three reduced frequency are chosen for two reasons,
one is that the three number are the magnitude of general reduced
frequency countered in the real ight, the other is that compar-
isons between the results of this study and the study before can
be made. Here the free stream Mach number and Reynolds num-
ber are 0.6 and 1.61 10
6
, respectively.
A grid independence study was also performed with the missile
at =1
.
The static aerodynamic coecients results are also presented for
reference. Notice the hysteresis for 0
in
steady case and pitch up case (r
k
= 0.628) are shown in Fig. 9(a)
and (b), respectively.
Fig. 10 shows the variation of lift, drag, and pitching moment
coecients with about three longitudinal locations on the con-
guration, x/L = 1.6, x/L = 1.73 and x/L = 1.912. The reduced
frequency r
k
= 0.628. Notice that each of these cases has some-
what different characteristics at the same reduced frequency. The
results for C
D
and C
M
only show slight differences when compared
with C
L
results.
The lift coecients of the missile which oscillates in pitching
at three ranges of angle of attack 0
, 0
, and
0
L
, was calculated, shown in Fig. 13. It shows that
while the angle of attack ranges at which the missile oscillates in
134 L. Yang et al. / Aerospace Science and Technology 15 (2011) 129136
Fig. 9. Pressure contours over leeward side of the wing.
pitching are different, the derivative C
L
is the same for the same
phase of oscillating under the same reduced frequency condi-
tion. The variations of the derivative C
L
vs are different from
each other for different reduced frequency 0.336 and 0.628.
Here the derivative C
L
is calculated for the rst cycle, since the
lift coecients for the second cycle are nearly the same as those
in the rst cycle. So the phase of oscillation is between 0 and
2 in Fig. 13. The variation of the derivative C
L
vs in pitch-
up and pitch-down are shown in Fig. 13(a) and (b), respectively.
For comparison purpose, the derivatives C
L
where is near 0,
and 2 are not given because the value of the derivative there is
very big or even innite comparing with the most value of the
derivatives.
Considering the reason why the derivative C
L
is the same
for the same phase of oscillating under the same reduced fre-
quency condition at different ranges of angle of attack shown in
Fig. 13. In this case, take the lift coecients as the function of
t, x, , ,
...
, . . . . That is
C
L
=C
L
(t, x, , ,
...
, . . .)
here x indicates some information of oweld, for example, the
location of ow separation. The missile oscillates in pitching with
(t) =
m
m
cos()
the derivative of the angle of attack with respect to time
(t) =
m
sin(t)
the second order derivative of the angle of attack with respect to
time
(t) =
m
2
cos(t), . . .
That is
C
L
=C
L
(t, x, ,
m
)
Here take x as the same information, because the angles of attack
for the missile which oscillates in pitching at the three ranges of
angle of attack are small, and the information of the oweld dy-
namic system are similar.
Therefore
C
L
=C
L
(t, ,
m
)
The derivative
C
L
=
dC
L
d
=
dC
L
/dt
d/dt
=
dC
L
/dt
m
sin(t)
Here the function (dC
L
/dt)/
m
must be the same for the three
ranges of angle of attack because of the same derivative C
L
for the
same phase of oscillating , that is (t).
The variation of (dC
L
/dt)/(2
m
) vs time is shown in Fig. 14.
It shows that the function (dC
L
/dt)/
m
is the same for the three
ranges of angle of attack. From the above reasoning, we may reach
a conclusion that the derivative C
L
is the same for the same phase
of oscillation (t) when the oweld going into the same dynamic
systems.
4. Conclusion
Three-dimensional Reynolds averaged NavierStokes numerical
simulations were carried out to predict the oweld characteristics
of a pitching winged missile based on the nite volume method.
The Baldwin-Lomax eddy viscosity model with the modications
suggested by Degani and Schiff was used here. In order to validate
the computational results, the aerodynamic loads of a slender rev-
olution body investigated numerically are also given, and agreed
well with the experiment data.
The characteristics of the unsteady aerodynamics about a
winged missile, which is oscillated in pitching with different fre-
quencies, with different center of rotation and at different range of
angle of attack are shown in this paper. It is found that from the
computational results, the unsteady hysteresis of the lift, drag and
pitch moment coecients vary not only with the frequency, but
also with the position of the oscillating axis. Although the angle of
attack range at which the missile oscillate in pitching are different,
the derivative of lift coecients with respect to the angle of attack
is the same for the same phase of oscillating under the same re-
duced frequency condition when the oweld of the missile going
into the same dynamic systems.
L. Yang et al. / Aerospace Science and Technology 15 (2011) 129136 135
Fig. 10. Lift, drag and pitching moment coecient vs about different longitudinal
locations.
Fig. 11. Lift coecient for different ranges of angle of attack (r
k
=0.336).
Fig. 12. Lift coecient for different ranges of angle of attack (r
k
=0.628).
Fig. 13. The derivative C
L
vs in pitching-up and pitching-down.
Fig. 14. The function (dC
L
/dt)/(2
m
) vs time.
136 L. Yang et al. / Aerospace Science and Technology 15 (2011) 129136
References
[1] J.D. Anderson, Governing equations of uid dynamics, in: J.F. Wendt (Ed.), In-
troduction to Computational Fluid Dynamics, Springer-Verlag, 1992, pp. 1551,
Chapter 2.
[2] M.T. Arthur, F. Brandsma, N. Ceresola, W. Kordulla, Time accurate Euler calcula-
tions of vortical ow on a delta wing in pitching motion, AIAA Paper 99-3110,
1999.
[3] K.J. Badcock, F. Cantariti, I. Hawkins, M. Woodgate, L. Dubuc, B.E. Richards, Sim-
ulation of unsteady turbulent ows around moving aerofoils using the pseudo-
time method, International Journal for Numerical Methods in Fluids 32 (5)
(2000) 585604.
[4] B.S. Baldwin, H. Lomax, Thin layer approximation and algebraic model for sep-
arated turbulent ows, AIAA Paper 78-257, 1978.
[5] G. Barakos, D. Drikakis, An implicit unfactored method for unsteady turbu-
lent compressible ows with moving boundaries, Computers and Fluids 28 (8)
(1999) 899922.
[6] G. Barakos, D. Drikakis, Computational study of unsteady turbulent ows
around oscillating and ramping aerofoils, International Journal for Numerical
Methods in Fluids 42 (2) (2003) 163186.
[7] L.W. Carr, Progress in analysis and prediction of dynamic stall, Journal of Air-
craft 25 (1) (1988) 617.
[8] L.W. Carr, M.S. Chandrasekhara, M.C. Wilder, Effect of compressibility on sup-
pression of dynamic stall using a slotted airfoil, Journal of Aircraft 38 (2) (2001)
296309.
[9] F.N. Coton, R.A. Galbraith, An experimental study of dynamic stall on a nite
wing, Aeronautical Journal 103 (May 1999) 229236.
[10] R.M. Cummings, S.A. Morton, S.G. Siegel, S. Bosscher, Numerical prediction and
wind tunnel experiment for a pitching unmanned combat air vehical, AIAA
Paper 2003-0417, 2003.
[11] D. Degani, L.B. Schiff, Computation of turbulent supersonic ows around
pointed bodies having crossow separation, Journal of Computational
Physics 66 (1986) 173196.
[12] D. Drikakis, M. Hahn, A. Mosedale, B. Thornber, Large eddy simulation us-
ing high-resolution and high-order methods, Philosophical Transactions of the
Royal Society A 367 (2009) 29852997.
[13] P.A. Durbin, A perspective on recent developments in RANS modeling, in:
W. Rodi, N. Fueyo (Eds.), Engineering Turbulence Modeling and Experiments,
vol. 5, Elsevier, Amsterdam, 2002, pp. 316.
[14] J.A. Ekaterinaris, M.F. Platzer, Computational prediction of airfoil dynamic stall,
Progress in the Aerospace Science 33 (1997) 759846.
[15] L.E. Ericsson, Complex angular amplitude and frequency effects on delta wing
unsteady aerodynamics, AIAA Paper 99-4009, 1999.
[16] Z. Gao, Research on the hysteresis properties of unsteady aerodynamics
about the oscillating wings, Applied Mathematics and Mechanics (English Edi-
tion) 20 (8) (1999) 895907.
[17] D. Greenblatt, D. Neuberger, I. Wygnanski, Dynamic stall control by intermit-
tent periodic excitation, Journal of Aircraft 38 (1) (2001) 188190.
[18] E. Guilmineau, P. Queutey, Numerical study of dynamic stall on several airfoil
sections, AIAA Journal 37 (1) (1999) 128130.
[19] M. Hahn, D. Drikakis, Implicit large-eddy simulation of swept-wing ow using
high-resolution methods, AIAA Journal 47 (3) (2009) 618629.
[20] J. Henkner, Phenomena of dynamic stall on swept wings, in: 22nd International
Congress of the Aeronautical Sciences, ICAS Paper 2000-2.9.2, Aug.Sep. 2000.
[21] I. Heron, R.Y. Myose, Delta wing vortex burst behavior under dynamic
freestream, Part 1 Fast pitch-up during deceleration, AIAA Paper 2007-6725,
Aug. 2007.
[22] A. Jameson, W. Schmidt, E. Turkel, Numerical solutions of the Euler equations
by nite volume methods using RungeKutta time stepping scheme, AIAA Pa-
per 81-1259, 1981.
[23] P.E. Morgan, M.R. Visbal, Simulation of unsteady three-dimensional separation
on a pitching wing, AIAA Paper 2001-2709, June 2001.
[24] H. Ranke, Unsteady separation in two-dimensional turbulent ows, in: 19th
Congress of the International Council of the Aeronautical Sciences, ICAS Paper
94-4.5.1, Sep. 1994.
[25] S. Subchan, R. Zbikowski, J.R. Cleminson, Optimal trajectory for the terminal
bunt problem: an analysis by the indirect method, AIAA Paper 2003-5791,
2003.
[26] B.E. Tinling, C.Q. Allen, An investigation of the normal-force and vortex-wake
characteristics of an ogive-cylinder, NASA-TND-1297.