Copyright© 1997, American Institute of Aeronautics and Astronautics, Inc.
AIAA-97-3725 OPTIMAL EXTERNAL CONFIGURATION DESIGN OF UNGUIDED MISSILES Omer Tannkulu* & Veysi Ercant
A97-37313
Defense Industries Research and Development Institute (TUBITAK-SAGE) Mechanics and Systems Engineering Research Group (MSMG) PK16, Mamak 06261, Ankara, Turkey
Abstract In this paper a simple optimal external
respectively and two constraint functions related to stability are considered in the case study.
configuration design method is proposed that can be used in conceptual and preliminary design stages of
Nomenclature
an unguided missile development project. Cost and constraint functions are derived from the results of
Drag force coefficient. C,
linear time-invariant aeroballistic theory (constant
Roll damping moment stability derivative.
roll rate and forward speed, de-coupled axial and
Roll moment due to fin cant stability
transverse dynamics); therefore different phases of derivative.
flight are examined separately. Curve-fitting is used to reduce number of trial cases and hence work
c.
Static moment stability derivative.
required to obtain aerodynamic and inertial data.
Transverse damping moment stability
Optimal configurations are determined by using a
derivative.
modified steepest descent algorithm. A case study is presented in which external configuration of an unguided light assault missile is optimized for free
cm
Transverse damping moment stability derivative. Magnus moment stability derivative.
flight at low subsonic speed. Three cost functions Static force stability derivative. related to stability, range and warhead performance * Coordinator. Member AIAA. t Research Engineer, Flight Mechanics Section. Copyright © 1997 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.
700 American Institute of Aeronautics and Astronautics
Copyright© 1997, American Institute of Aeronautics and Astronautics, Inc.
d
Gradient vector of F(x).
/.(*)
Cost function.
A
:
Reference length, rocket motor diameter.
p^
:
Free stream density.
H, v
:
Penalty constants.
f
:
Transverse damping factor.
Modified cost function. Inequality constraint function. *,(*)
Equality constraint function.
Hessian matrix of F(X) .
Axial moment of inertia.
Introduction
The main objectives in external configuration
Transverse moment of inertia.
design of unguided missiles are to obtain adequate
Non-dimensional axial radius of
stability in all phases of flight, short minimum range, long maximum range, low dispersion and
gyration. Non-dimensional transverse radius of
large payload mass. In practice it is difficult to
achieve these objectives due to complicated nature of gyration.
unguided missiles as nonlinear, time-varying and m
Total mass. random systems. Significant advances have been Warhead mass. made in analysis and system identification aspects of
Pdyn
Magnus instability roll rate limit.
Yaw-pitch-roll resonance roll rate.
flight dynamics of unguided missiles since World War II. A number of range, dispersion and stability
Dynamic stability factor.
criteria have been determined by using analytical,
Static stability factor.
numerical and experimental techniques. On the other
S
Reference area, ;r/l2/4.
hand, very few studies have been published on
V
Speed of missile. Vector of scaled variable geometrical
external configuration design of unguided missiles1.
The almost untouched synthesis problem is challenging due to three reasons: Firstly, design
parameters. Lower limit of x .
criteria are functions of aerodynamic and inertial parameters which are in turn complicated functions
Upper limit of x . of free stream flow conditions, missile geometry and
a
Adaptive step size of iterations. mass distribution. Secondly, design criteria are often
701 American Institute of Aeronautics and Astronautics
Copyright© 1997, American Institute of Aeronautics and Astronautics, Inc.
contradictory. Thirdly, design criteria are different
flight (subsonic, transonic, supersonic). If M
for different flight phases of a given type of unguided
different
missile. (Naturally, design criteria are different for
aerodynamic data has to be determined MX N3 times.
different types of unguided missiles. Artillery shell,
Separate flight dynamics analysis of all of these
artillery missiles, high kinetic energy projectiles,
configurations is necessary which is difficult and
light assault missiles, anti-tank missiles, sounding
costly. Moreover, such an analysis gives information
rockets and re-entry vehicles all have to be optimized
about configurations with the selected discrete values
with different cost and constraint functions.)
of (/, c, s)only.
flight
speeds are considered, then
Nature of the problem will be discussed by a
Consider an external configuration design
simple example. Consider the unguided artillery
analysis where only s is changed. As s is increased
missile configuration with cruciform tail fins
m, Ia ,
presented in Figure 1:
, C,r, Cls and CD all
increase. An increase in s has both good and bad results in terms of range, dispersion and stability performance. Good results are increase of static and
Warhead
damping dynamic Rocket Motor
Figure 1. Unguided artillery missile configuration.
stability, and
decrease of
dispersion due to thrust misalignment. Bad results
are decrease of range and increase of dispersion due to wind. Increase of s can have good or bad results
c and s are chord and span of tail fins
in terms of Magnus dynamic stability since Ia , It,
respectively while / denotes length of mid-section
Cm
case. These are the geometrical parameters that are
easiest to modify once rocket motor and warhead
good or bad results in terms of the likelihood of yawpitch-roll resonance since Ia , It, Cm^ , C, and Clg
properties are fixed. If N different values of each of
these parameters are considered, then total number of candidate configurations to be examined becomes N*. Change of a single geometric parameter changes all aerodynamic and inertial data. Flight speed of an unguided artillery missile varies significantly during
all increase.
One must also note that cost and constraint functions are different for boost and coast phases of
flight of an unguided artillery missile. (As an example, most of the dispersion takes place during boost phase. Hence, dispersion criteria may not be
702 American Institute of Aeronautics and Astronautics
Copyright© 1997, American Institute of Aeronautics and Astronautics, Inc.
considered among cost and constraint functions in a
data by using curve fitting. This approach reduces
first order analysis of coast phase.)
number of trial cases significantly and it also creates
In the next section, details of a method that
the possibility of analysing configurations with x
can be used in optimal external configuration design
data other than the original sparse set. Aerodynamic
of unguided missiles will be presented This will be
data at a given Mach number and inertial data are
followed by a case study about a light assault
usually smooth functions of geometrical parameters;
unguided missile.
hence, polynomial functions give satisfactory results in terms of curve fitting.
Optimal External Configuration Design
The proposed method for optimal external
• Step#4:
Determination
ft (x)(i = 1,2,3, ...,Nf),
of
cost
inequality
functions
constraint
configuration design has four basic steps:
functions
gl(x^i = l,2,3,...,Ng^
and
equality
• Step#l: Investigation of characteristics of the
unguided missile that is to be optimized.
constraint functions hl(x%i = 12,3,...,Nh) by using
Determination of cost and constraint functions for
the approximate aerodynamic and inertial functions
different phases of flight. Determination of lower and
that were obtained in Step-#3. The optimization
upper limits for variable geometrical parameters with
problem is formulated as follows:
special
emphasis
given
to
producibility
considerations.
• Step#2: Determination of aerodynamic data (at
Minimize:
//(*)
Subject to:
g,(x)^0,(i = 1,2,3,...,Ng)
several Mach numbers depending on flight speed range of the unguided missile) and inertial data for a sparse set of variable geometrical data by using
ht(x) = 0,(i = 1,2,3,...,Nh) x,<x<xu
experimental, numerical and theoretical methods. • Step#3: Determination of approximate functional
In above expressions x, and xv are lower and upper
relationships between aerodynamic data at different
limits of x that are usually dictated by producibility.
Mach numbers and geometric data by using curve
Optimal configurations for different cost and
fitting. Determination of approximate functional
constraint functions are determined by using the
relationships between inertial data and geometric
703 American Institute of Aeronautics and Astronautics
Copyright© 1997, American Institute of Aeronautics and Astronautics, Inc.
following
iterative
modified
steepest
descent
algorithm2:
iteration, a is the adaptive step size of iterations wmle
j
is the
gradient vector and H is the Hessian
matrix:
a = vr * d.I ~ ——— s+
ak = _ T
-
(2)
2F ?x, dx.
(3)
Case Study
9
(5)
(6)
The subject matter of this representative case study is optimal external configuration design of an unguided short range light assault missile, schematic (*/),.'/(**+>),
drawing of which is shown in Figure 2. The one man
portable missile is shoulder launched and there is no
rocket motor propulsion once the missile leaves its tube launcher. The original anti-armour shaped (A: = 1,2,3,...).
charge warhead of the missile is to be replaced by another warhead as a part of a product improvement
In above expressions F is the modified cost function which is obtained by combining cost and
constraint functions
to turn the constrained
optimization problem into an unconstrained one. x is the vector of variable geometric parameters which has a dimension of n. // > 0 and v > 0 are penalty
constants of very large magnitude for inequality and
equality constraints respectively. Magnitudes of //
project. No modifications will be performed in terms
of solid propellant rocket motor since mass of the new warhead is constrained to be less than or equal to mass of the old one. Free flight Mach number of
the missile is 0.32 with the old warhead. The missile will have a low subsonic flight speed with the new warhead as well. The missile has six planar tail fins with no dihedral which means that it has six-gonal
and v are increased in every couple of steps of 704 American Institute of Aeronautics and Astronautics
Copyright© 1997, American Institute of Aeronautics and Astronautics, Inc.
rotational symmetry and six planes of mirror
section. Lower and upper limits of these parameters
symmetry3.
that were determined by taking producibility and warhead mass constraints into account are presented in Table 1. Five equally spaced values of each parameter were used to obtain a sparse set of 125 configurations.
39 215
50
50
Table 1. Upper and lower limits of variable
geometrical data.
Figure 2. Unguided light assault missile. All
dimensions are in millimeters.
Lower Limit
Upper Limit
/ [mm]
30
60
There are three reasons while such a missile
c [mm]
10
15
was selected as the subject of this case study: Firstly,
s [mm]
50
100
there is only one flight phase to be considered (short
duration free flight during which the missile has a
Closed
form
expressions
that
relate
relatively straight mean trajectory). Secondly, the
aerodynamic stability derivatives to geometry were
flight takes place at low subsonic Mach numbers
obtained by using Bryson's method which is
where aerodynamic characteristics are almost
restricted to incompressible potential flow. Drag
constant, and hence only one set of aerodynamic data
coefficient data of the 125 configurations were
has to be determined. Thirdly, it is possible to derive
determined by Missile DATCOM data base for a
approximate closed form expressions for a large
Mach number of 0.32. CD was assumed to be related
number
of
aerodynamic
stability
derivatives
to x by the following functional relationship:
3
(including Cm !) by using the method of Bryson for
flight at low subsonic Mach numbers.
CD(x) =
Three geometrical parameters of the light assault missile were selected to be variable (7)
parameters of the optimization problem: Span s and chord c of tail fins and length / of warhead mid-
705 American Institute of Aeronautics and Astronautics
Copyright© 1997, American Institute of Aeronautics and Astronautics, Inc.
x data were scaled by using xl = l/lmax, x2 = c/cmax
asymmetries.) Closed form expressions for
and x3 - s/smm to improve quality of curve-fitting.
pres are given below:
and
c,(/= 0,1,...,9) coefficients for CD were determined by using curve-fitting utility of Sigma Plot software.
M
2V I,
(8)
A Turbo Pascal program was prepared to
determine inertia! and aerodynamic data of the light V
assault missile based on a simplified solid model,
(9)
results of Bryson's method for aerodynamic stability derivatives and curve fitted Missile DATCOM CD
where,
data. Values of the following cost and inequality constraint functions were calculated for the 125 kt2
configurations by using the program: •
(10)
f i : The first cost function is related to Magnus
instability and roll-pitch-yaw resonance. Linear time-
_ 2T
(11)
invariant aeroballistic theory has two important results related to roll rate p of a statically stable (12)
slightly asymmetric unguided missile4. Firstly, if
magnitude of p exceeds a certain limit denoted by pdyn during flight, then a Magnus instability takes
1 T = ——=
-cmpp'-cz;-cD\
(13)
place. Secondly, if magnitude of p coincides with a
certain value denoted by pres then a yaw-pitch-roll resonance takes place. (Non-linear aeroballistic
(14)
ka , =
theory predicts that a linear resonance can be
followed by lock-in of p to pns which in turn can be followed by a severe instability known as catastrophic yaw due to nonlinear induced aerodynamic
moments
and
configurational
706 American Institute of Aeronautics and Astronautics
2m
(15)
Copyright© 1997, American Institute of Aeronautics and Astronautics, Inc.
Magnitude of piyn is usually three to five
• /3: The third cost function is related to warhead
times larger than magnitude of pres. In external
performance. A very simple criterion in terms of
configuration design, it is desirable to have a large
wea
difference between
*"***
tries to meet
Pdyn
P°n Wstem effectiveness is the ratio of warhead
and Pres since one usually
Pns
condition by using tail
m
«
to total mass
Hence
-
>
the thirf cost
functionis selectedas
'
fin cant. Hence, the first cost function is selected as, /s=—-•
m
/,=-
2V
(20)
(16)
•
g[: The first inequality constraint function is
related to the static stability factor sf which is • / 2 : The second cost function is related to range defined as, performance. Linear time-invariant aeroballistic theory predicts variation of speed of an unguided s. =
missile in free straight flight to be,
*
(17)
Cz« '
(21)
c
Magnitude of ss should be larger than a certain limiting value (which is usually taken as 1) for
where 5 is the non-dimensional arclength:
adequate static stability. In case of the light assault missile a survey of s3 data of 125 configurations (18)
showed that there were no problems in terms of static stability, ~ 4.5 < ss <~ 5.5. Nevertheless, the first inequality constraint function was selected as,
Hence, the second cost function is selected as, 81 = ^D
m
(22)
(19)
The reason for this selection is related to dispersion of the light assault missile which is due to aiming 707
American Institute of Aeronautics and Astronautics
Copyright© 1997, American Institute of Aeronautics and Astronautics, Inc.
errors and configurational asymmetries. Nothing can
of cost and constraint functions with 5 and c at the
be done to reduce dispersion due to aiming errors in
maximum value of / are shown in Figure 3.
terms of external configuration design. On the other
Optimal external configurations for the three
hand, dispersion due to configurational asymmetries
cost functions were determined by using a Turbo
can be reduced by increasing s,.
Pascal program that implemented the modifed
•
steepest descent algorithm discussed in the previous
g 2 : The second inequality constraint is related to
section. Optimal (scaled) x values are presented in
transverse damping factor which is defined as,
Table 2:
H
(23)
Table 2. Optimal x values for ft(i = 1,2,3). /I
ft
/3
*1
0.997
0.996
0.997
X2
0.993
0.813
0.667
*3
0.715
0.711
0.707
Magnitude of £ should be larger than a certain
limiting value for adequate dynamic stability. In case of the light assault missile a survey of ^ data of 125
configurations showed that there were no problems in terms of dynamic stability, ~0.12<£<~0.19. Nevertheless, the second inequality constraint
Values of the cost functions at the optimal x values are
function was selected as,
/, = -0.82108,
/2 = 1.59146
and
/3 = -1.43706 respectively. (ft values are scaled in (23)
such a way that the difference between maximum and minimum values and mean are both equal to
The reason for this selection is also related to the
one.)
desire for keeping dispersion due to configurational Discussion
asymmetries below a certain level. Polynomial functional relationships similar to (7) were assumed to exist between each /, g and x , coefficients of which were determined by using curve fitting utility of Sigma Plot software. Variation
• An examination of Figure 3 shows that the correct optimal value were obtained in each case.
• In all cases xl =~ 1, / =~ lmm for the optimal configurations.
708 American Institute of Aeronautics and Astronautics
Copyright© 1997, American Institute of Aeronautics and Astronautics, Inc.
• In all cases
x2 =~ 0.7
for the optimal
configurations. Value of optimal span is specifed by
mission, rather than the best configurations for
different phases of the mission.
the damping factor inequality constraint g 2 .
References
• The static stability inequality constraint g, is overruled by the damping factor inequality constraint
^ayzac,
R.
and
Carette,
E.,
"Parametric
Aerodynamic Design of Spinning Finned Projectiles g2 fOTXl=l,l
= lmaic.
Using a Matrix Interpolation Method," AIAA
• No convergence problems were observed in Journal of Spacecraft and Rockets, Vol. 29, No. 1,
iterations to find the optimal x in all cases. January-February 1992. Conclusion
2
Leblebicioglu, K., EE-553 Optimization Course
There are a large number of problems that Lecture Notes, Electrical Engineering Department,
remain to be investigated in terms of external Middle East Technical University, 1994. configuration design of unguided missiles:
• Extension of the current method so that a number
3
Nielsen, J. N., Missile Aerodynamics, Nielsen
of cost functions can be optimized at the same time.
Engineering & Research, Inc., Mountain View, (Multiple objective constrained optimization.) California, 1988. • Optimal external configuration design analysis of
other types of unguided missiles (artillery shell, artillery missiles, high kinetic energy projectiles, anti-tank missiles,
sounding rockets,
4
Murphy, C. H., "Free Flight Motion of Symmetric
Missiles", Ballistic Research Laboratories, Aberdeen
re-entry
Proving Ground, Report No. 1216, July 1963.
vehicles). • Optimal external configuration design analysis
with a much larger number of variable parameters. • Optimal external configuration design analysis of cases where nonlinear aeroballistic phenomena are observed. • Development of a method that can determine the
best configuration for the whole of a specified 709 American Institute of Aeronautics and Astronautics
Copyright© 1997, American Institute of Aeronautics and Astronautics, Inc.
f, (Magnus Instability-Resonance)
9, (Static Stability)
,.-0.8-''
0.70
0.75 0.80
0.85 0.90 0.95 1.00
x, (Chord)
f2 (Range) -—0.3-
—-Q.2—
1.0 -T————1.8-
0.70
0.75 0.80 0.85 0.90
0.95 1.00
X2 (Chord)
g2 (Damping Dynamic Stability)
o.°S
0.70
0.75 0.80
0.85 0.90 0.95 1.00
i, (Chord)
o.*f, (Warhead)
-o-*
0.70
0.75 0.80 0.85
0.90
x, (Chord)
I x"
0.70
0.75 0.80
0.85 0.90 0.95 1.00
Xj (Chord)
Figure 3. Variation of cost and constraint functions with c and s for / 710 American Institute of Aeronautics and Astronautics
.
0.95 1.00