AD-A015 238

ANALYSIS OF AIRCRAFT EVASION STRATEGIES IN

AIR-TO-AIR MISSILE EFFECTIVENESS MObri '
G. Carpenter,

et al

Grumman Aerospace Corporation

Bethpage, New York
August 1975

DISTRIBUTED BY:

National Technical Information Service

U. S. DEPARTMENT OF COMMERCE

MISSILER EFECIEESS MODEL

00
Auus

4a*1'
1975

0W1

I
Grumman Research Department Report RE- 506

I
ANALYSIS OF AIRCRAFT EVASION STRATEGIES
IN AIR-TO-AIR MISSILE EFFECTIVENESS MODELS

by

C. Carpenter and M. Falco

System Sciences

?a

'$-

August 1975

Approved by:

'.

t
L

......

'

.
.(
Charles E. Mack, Jr.
Director of Research

--

UCLLS m
Seruntv Classlfcation

DOCUMENT COtNTROL DATA- R & D

Se urtlY ellss .aheetou of title buod of ibatrotI and ends.
'IsNA

Oft

anno ta mtn -uat b

entered when the overall tep

1tTNGAC ?v I TV ?corpor te author)

2S.

REPORT SECUNITY

t Es CIS*atiet d)

CLASSIrICA TION

Unclassified
;rummn Aerospace Corporation
I

NEPOA'

RoU

N/A

?LE

nalysis of Aircraft Evasion Strategies in Air-to-Air Missile

Effectiveness Models
4 DESCRIP

IVE

*40TS
'Type of repct and incfu.seve dote)

Research Report

Au

.ORSo

(First name. mtddle Initial. lost flrea,,

G. Carpenter and M. Falco

*

.
s.TOTAL

REPORT DATE

August 1975
041

Ia.

CONIRAC T On GRANT NO

II. POROJECT

PAG

ES

No

"ors

ORIGINtTORtS REPORT NUMSERIS)

IPE-506

.40

NIA
N IA

*b.

OTHER SREIONT NOS1

(Any 0fl

numbers Nat me? 60 a ai

None

d.

sc

37

OISTR"RUTION STATEMENT

Approved for Public release; distribution unlimited.

It

SUPPLCkE*NTAN NOTES

12 SPONSORING MILI* ANY ACTIVITY

None

None

This report presents a new methodology with which to quantify

missile effectiveness and aircraft vulnerability. The approach'is a
blend of applications of optimal control theory, stochastic learning
theory, and simulation. This methodology permits an evaluation of aircraft evasive maneuvering and countermeasures deployment strategy as an
integral part of the effectiveness/vulnerability measurements. The
strategy determination is a form of feedback control policy based upon
a discretized set of information thresholds in the relative coordinate
space as would be available to an evading aircraft pilot. The optimization criteria of an evading aircraft is that of maximizing the
survival probability for all relative coordinates. The representative
model chosen for illustration is evasion from a close range air-to-air
IR guided missile. The effectiveness/vulnerability results ss well as
sample trajectories illustrating the optimal evasive maneuvering are
giver for several studies conducted with the model., One study illustrates how optimal maneuvering can degrade IR missile effectiveness by
taking advantage of the close-range, narrow-field-of-view "saturation"
properties of the seeker sensor. Other studies concern the effects of
optimal maneuvering on off-boresight launch effectiveness and on effectiveness sensitivity with warhead lethality variations.

DD Nov 1473

lUNCISTFIE

/securftv Cl,

;tiftcstinn

,I

1i
1 IThis
I !missile

ABSTRACT

report presents a new methodology with which to quantify

effectiveness and aircraft vulnerability. The approach is
a blend of applications of optimal control theory, stochastic learning theory, and simulation.

This methodology permits an evaluation

of aircraft evasive maneuvering and countermeasures deployment

strategy as an integral part of the effectiveness/vulnerability
measurements.

The strategy determination is a form of feedback con-

trol policy based upon a discretized set of information thresholds

in the relative coordinate space as would be available to an evading
aircraft pilot.

The optimization criteria of an evading aircraft is

that of maximizing the survival probability for all relative coordi-

nates.

The representative model :hosen for illustration is evasion

from a close range ai--to-air IR guided missile.

11

The effectiveness/

vulnerability results as well as sample trajectories illustrating

the optimal evasive maneuvering are given for several studies con-

ducted with the model.

One study illustrates how optimal maneuvering

can degrade IR missile effectiveness by taking advantage of the

1J

close-range, narrow-field-of-view "saturation" properties of the

seeker sensor.

Ii

Other studies concern the effects of optimal maneu-

vering on off-boresight launch effectiveness and on effectiveness

sensitivity with warhead lethality variations.

iii

TABLE OF CONTENTS

Section

1.
II.

Int roduc tion ....................................

Air-to-Air Missile Model ............

..

.00...

Aircraft Maneuvering Strategy Development ..

Dynamical Model

.....

..........

.........

Missile Equations of Motion in

Inertial Coordinates ..................

Aircraft Equations of Motion in

.neri.ial Coordinates .................

Missile Guidance Considerations
Missile Seeker Properties

.......

10
12

..................

Missile Warhead Detonation and

II.

15

.........................

Computational Approach
Learning Phase ......

..................

. . .

15

Strategy Initialization ............ ..

15

Trajectory and Outcome Simulation .....

16

Strategy Modification

17

Statistics Phase
IV.

13

. ........

Kill Determination ................

.................

............ ................

18

Computational Results ...........................

21

Summary ....................................

21

Study 1 - Constant Saturatiou Range Seeker .

26

Study 2 - Aspect Dependent Saturation

Range Seeker

V.
VI.
Appendix-

.....................

38

Study 3 - Off-Boresight La'inch Sensitivity

42

Study 4 - Warhead Lethality Sensitivity ....

46

Conclusions
References

.....................................

53

........................

........

56

Computer Program Listing

55

Precdinlg pp ink

LIST OF ILIIJSTRATIONS
4R

"-

Page

Figure

2-1

Aircraft Observable State, R,.

2-2

Elemental Maneuver Chaices ......................

2-3

Coordinate Nomenclacure

2-4

Warhead Lethality ..............................

3-1

Missile Launch Coordinates ..............

4-1

Longitudinal Deceleration Parameter a

21

4-2

Aspect Dependent Saturation Range, Rs

25

4-3

Missile Effectiveness Results, Constant

Saturation Range - 1000 ft (Nonmaneuvering
Versus Optimal)

4-4

.........................

.............

14

...

.......

....

19

26

Missile Effectiveness Results, Constant

Saturation Range - 1000 ft (5g Turn

.......

Versus Optimal) ...................

4-5

Aircraft Vulnerability Results, Constant

Saturation Range 1000 ft ........................

4-6

4-7

4-8

4-9

4-10

4-11

27

29

Evasive Strategies and Trajectories,

Study No. 1, Result 1A (Relative) ................

32

Evasive Strategies and Trajectories,

Study No. 1, Result IA (Absolute) ...............

33

Evasive Strategies and Trajectories,

Study No. 1, Result IB (Relative) ...............

34

Evasive Strategies and Trajectories,

Study No. 1, Result 1B (Absolute) ...............

35

Evasive Strategies and Trajectories,

Study No. 1, Result 1C (Relative) ...............

36

Evasive Strategies and Trajectories,

Study No. 1, Result C (Absolute) ...............

37

vii

Preceding pags blank

LI1ST OF TrABU.S
Pg

Table
4-1

Representative Missile and Aircraft Data

4-2

Missile Effectiveness and Aircraft

Vulnerability Studies

,1-3

Sample Results
Trajectories

...

..

..

..

. ...

..

..

. ...

22

........

. ...

..

. ..

24

Evasive Strategies and

......

.....

ix

.....

. ....

...

......

...

30

4 I
LIST OF SYMBOLS

Aircraft longitudinal -acceleration

CD

Zero-lift drag zoefficient

CD

Induced drig coefficient

%'LIM

Maneuvering

limit due to structural constraint

Autopilot closed loop gain

ap
Guidance gain

g
m

Missile mass

n yc

Yaw acceleration command

PK
P

Kill probability

Decision table element - probability of selecting the

jth control in the ith observed state

Missile-to-aircraft range

RA

Aircraft turn radius

Rmin

Minimum relative range

RS

Seeker saturation rarge

Missile/autopilot yaw rate

Line-of-sight turning rate

line-of-sight turning rate

r sSeeker
r SL!M

Seeker rate limit

Aerodynamic reference area

Thrust

Engine burn time

tDL

Control activation delay time

xi

Prxiliig pap Iaik

I
LC

Total guidance

u.

Elemental mfaneuver controls

v.1

Elcmtntal countermeasure controls

VA

Aircraft velocity

time

Miss ile velocity

VS

:Aircraft

V% V

Relative veto-

X.

Relative states determined by aircraft observable

decompos it ion

xA? Y.

Inertial components of aircraft position

Y1

steady--

te .neuver

velocity

.y components

Inertial components of missile position

Outcome description

Track crossing angle

Aircraft control parameter
Missile inertial heading
Off-boresight angle (gimbal angle)

LIM
.. (:)

P
Seeker gimbal angle
linit
Learning algorithm reinforcement weights
Atmospheric density

ap

\utopilot time 7onstant

Seeker time constant

"

Aircraft inertial heading

Inertial line-of-sight angle
4issile angle-off relative to aircraft velocity vector

xii

I.

INTRODUCTION

Quantifying missile effectiveness and aircraft vulnerability

by using air-to-air and surface-to-air simulation models is of
considerable value in

1) developing more lethal missile systems

and improving their operational deployment and

2) determining

operational tactics to enable an evading aircraft to increase its

survivability.

However,

the use of such simulation models is fre-

quently criticized because:

The missile capability results (which are given

1)

by launch envelope data predicted by simulation

models with limited maneuvering targets) are
overly optimistic in view of real-world statistics
2)

Few approaches exist that can determine an evading

aircraft's best maneuvering and countermeasures
deployment strategies (in terms of maximizing survivability) as a "closed-loop" feedback control
over all relative coordinates

Our approach, described later in this introduction, overcomes these

objections.
References 1 and 2 outline Lhe methodology that has been
adopted to date by many investigators concerned with missile effectiveness.

This methodology is characterized by an extensive

model that simulates in detail missile trajectory, warhead detonation, and associated aircraft structure damage to determine aircraft survivability.

The evading aircraft generally performs pre-

scribed "open loop control" maneuvers such as maximum

"g"

turns

or sinusoidal weaving maneuvers during the trajectory portion of

the simulation.

The missile effectLveness determination is then

made by a statistical sampling over many possible launch conditions

as in Ref. 1, or a selected set of conditions as in Ref. 2. The
methodology in Ref. 3 is a departure from that of Refs, 1 and 2 in
that aircraft evasive maneuvers that maximize the missile miss
distance are computed by gradient methods for selected initial conditions.

This approach moves in the direction of optimizing the

aircraft maneuvering strategy, but has associated with it many

mathematical and computational restrictions.

An impf rtant re-

striction is that the payoff function must be a continuously diff.-rentiable function of the relative coordinates.

This mathematical

diffi-ulty implies a computational problem in solving for the optimal feedback control for the aircraft over all relative coordinates.
Thus, only an open loop control maneuver can be computed for the
aircraft for each launch condition.
The approach taken in this report provides a new methodology
uith which to consider the effectiveness/vulnerability quantification problem.

The approach is basically a blend of applications

of optimal -o- 'ol theory and stochastic learning theory.

application

The first

,.this methodology to one-on-one dogfight game models

with gun we;pons was reported in Ref. 4.

In this setting the best

aircraft m.neuvering strategies (as a discretized feedback control

which maximizes kill probability) for each combatant were computed
by a learning algorithm.

The domains of effectiveness of each sys-

tem were then quancified by simulation.

That same approach -an

easily be extended to the missile combat model considered here,

with the two-sided game model aspect being replaced by a one-sided
control problem for the evading aircraft.

This methodology permits

model realism not generally considered in other optimization approaches:

developing maneuver and counteraeasure strategies based

upon the pilot's available "information thresholds" in the relative

coordinate space an:

using survival probability In place of miss

distance criterir &, an optimization criteria.

-

Thu; the methods

shown here provide one means of resol-ing criticism No

2 and offer

at the same time a means of accounting for the real world vrs
--

omputacional model differences of criticism No. I by

optimal evasive strategy as an integral ;ar

- -

t'iw
-he

of th. effecdiveness

determination.
Sresnt'.1ive

port

evasion mcre! chwsen for study in this re-

-s the close-range air-to-air IR ,;-Aidee mi :ii=

and postlaunch fliu.it phases employed ag, -s3t

formance attributes of current

air- :.ft

'::y fbnt., The pro

in

its

lavach

Oth per-

,,
r.has been

arbitrarily limited to two dimensional flight in the ho izontal

plane with the evading aircraft employing only its maneuver ng con-

trol :apability to increase iLZ surviva'- probability.

The choice

of model was quite arbitrary as the methodology -an easily accommodate radar guided missiles, counterweasures strategy analysis, and
three dimensional flight. In the last case, computer running times

can be expected to increase

f'xt -Ad over the two dimensional case

for corresponding levels or model detai'.

The surfa,.e-to-air missile

evasion problem would be as easily recolved by the methodology as the

air-to-air case.
Four sets o' res,,ILs involving the IR close-range missile model
are obtained to illustrate tne effect of aircraft emfloying optimal
maneuvering versus nonmaneuvering control on missile effectiveness
and aircraft vulnerability measures.

The first result set essen-

tially provides an effectiveness/vulnerability data baseline for a

standardized issile configuration.

The second result set deals

with the saturation of the missi.le seeker field of view for very
close targ(

ranges.

In this case the seeker field of view is satu-

rated and guidance neutralized at relative target ranges determined

I
by the track crossing angle between the missile/aircraft trajectories.

In the first result set the saturation range is a con-

stant independent of track crossing angle.

The third set shows

effectiveness/vulnerability sensitivity to target off-boresight

error at launch.

The last result set shows the sensitivity to im-

provements in warhead lethality.

Sample trajectories depicting

the optimal maneuvering are given along with the effectiveness/

vulnerability results for each of the four cases.
The authors wish to acknov ledge the computational assistance
rendered by Mr. Arthur Kaercher of the Grumman Data Systems Corporation in the course of obtaining the numerical results.

II.

AIR-TO-AIR MISSILE MODEL

This section presents the mathematical models for the evading

aircraft and attacking missile employed in the maneuvering tactics
determination and system capability analyses.

For the models de-

scribed, we have limited the analyses to combat in the horizontal

plane to simplify presentation. The same modeling technique without alteration in solution methodology can be straightforwardly
applied to three dimensional combat problems without reservation.

AIRCRAFT MANEUVERING STRATEGY DEVELOPMENT

The maneuvering strategy construction relies on the specification of which relative coordinates (positions, rates) and what
threshold levels for these coordinates will approximate the aircraft pilot's observable information during the engagement.

The

motivation for the construction and interpretation of the aircraft

observable data in terms of relative coordinate decompositions has
been described in Section 4 of Ref. 1. The decompositions lead to
a finite number of relative coordinate contingencies or regions for
consideration in the strategy development

The horizontal plane

model considered here assumes the aircraft can observe range

and angle-off

(w)

(R)

information with the thresholds shown in the

decomposition of Fig. 2-1.

When crossing a threshold for any of the regions making up the
decomposition, the aircraft will be permitted to select an elemental control as outlined in the hypothetical case below.

In the

vectorgram of Fig. 2-2 the elemental zontrols

could

'I

and

u5

correspond to maximum performanse turns that decelerate Lhe aircraft to a low steady-state velocity;

u2

and

u4

are maximum

performance turns at a high steady-state velocity; and

u3

is a

_7=r,4V'

----

- -

-t-

00

Angle-Off w

0120

1500

2100

Fig. 2-1

Aircraft Observable State, R, w

straight ahead Gash policy.

In the case of countermeasures deploy-

ment, the elemental :ontrol choices might be:

v

- jettison flare decoy

v 2 - illuminate arc-lamp
V3 - jettison and ignite fuel charge
v 4 - no countermeasures deployment
In this illustration the aircraft would select a pair of controls

(u, v) within each region.

The specification of a

(u, v)

pair

for all regions as shown in Fig. 2-1 would constitute a maneuvering/

countermeasures strategy in our terminology.

DYNAMICAL MODEL
The dynamical equations for a representatiw air-to-air infrared guided missile in its postlaunch maneuvering phase is presented
in ihis section.

In addition, the evading aircraft dynamical equa-

tions are also given.

The missile systems sensor characteristics,

guidance constraints, aerodynamic constraints, and warhead properties as incorporated in the model are explained.
The coordinate nomenclature employed with the trajectory dynami, s description is given in Fig. 2-3.

The symbol nomenclature

is presented in the front of this report.

!A

'

F,.

2-3

Coordinate nomenclature

Missile Equations of Motion in Inertial Coordinates

)CM

Im

Cos

=V

VM sin;

YM

=r

-- , .

VM

VCD

SC

2
(ny)
2m
2
ISVWDI
Y

D0

m
T

Aircraft Equations of Motion in Inertial Coordinates

VA sin.

xA
YA

VA cos

RRA
A

" "

.
.

A
=

The values for the control variable

VSS

velocity

and the steady-state

associated with five representative elemental ma-

neu;ver choices are as follows:

u1

u2

Left Turn
Deceleration
to Low V

Maneuver

Left Turn
High
Sustainable

u3

u4

u5

Dash

Right Turn
High
Sustainable

Right Turn
Deceleration
to Low V

Control
VS

-I
VT I

-1

VD

VT

+1
VT

+1
VT

The longitudinal acceleration/deceleration

parameter,

a,

is de-

termined by the foilowing scheme:

1

Determine

VSS

according to the selected maneuver

ui

from the above table.

2)

w-

if
VA > VSS

VA < VSS

VA

a-

VSS

The dynamical capability of the aircraft can be specified then by

the set of parameters

(VTI, VT2 , VD, RTI , RT2, a+ , a-),

given for a representztive aircraft in Table 4-1.

trajectory dynamics also requires the quantities
the strategy development.

which are

The aircraft
R and

a) for

These quantities are made available by

the following computations:

R

R !(
(xA

+ (YA

-)
YM)
Y)2

37r
2

where
)

" YM
nY1 (
tanN(xA - xM)
Missile Guidance Considerations

The equations corresponding to the line of sight geometry as

employed in the missile guidance are as follows:

10

Relative velocity components

V

- VA sin

VM cos

VA cos

VM sin ,

Line of sight yaw rate

r
with

(V x sin V - V

1 -

cos

R
as computed on page 10.

and

The equations corresponding to the seeker and autopilot response and associated constraints are:
Seeker line of sight rate response

(r - rS)
r

with the seeker rate limited by

irs

- r SLI M

Missile/autopilot yaw rate response

r

(Kapnyc

_I

ap
where

Kr VM

r)

M
is th~e commanded acceleration.

The commanded acceleration is restricted in magnitude to currespond to a maximum allowable maneuvering

limit required by

In this case the constraint

structural considerations.

nyc < gLIM

is imposed.
Other guidance constraints are considered in the model:
*

A control a,:tivation delay at launch of

In this case
above for

nyc -0

t > tDL.

11

when

0 < t

tDL;

tDL
nyc

secor'-s.
as

t maximum guidance time of

which-

tG

seconds after

yc = 0

A gimbal angle limit corresponding to seeker line

of sight and boresight axis limitations.
case if

I'd > 4LIM,

tion- nyc = 0
LLols.

In this

then the commanded accelera-

corresponds to- centering the con-

The seeker is modeled to require

the cor-

rect line of sight when excursions lead back to

within the allowable angle range.

Two limits are

P
a lower limit angle value at
LIML
launch, and
a higher value of flight.
LIMF
The thrust (T)- and burn raze (in) are specified :as funcrions
employed:

of time "in the model, and data for the representative problem considered are given in:Section IV. Along with this the aerodynamic
dra coffiiens
drag coefficients

C0
C

adCI C-and

are incorporated as functions of

Mach number and the air density computed from an exponential atmosphere model.

These aerodynamic data for a representative missile

configuration operated at a specific altitude are also given in

Section IV.
MISSILE SEEKER PROPERTIES
It is important to note that the above model developed for
simulation of the seeker response has made use of the assumption
that perfect geometric line of sight information (with a time constant of

Ts)

was made available to the autopilot whenever the

target was in the gimbal angle range of the seeker head.

Line of

sight rate errors due to detector signal/noise properties at various ranges, atmospheric properties, multiple sources, countermeasures, etc. can easily be incorporated in the model in a straight
forward manner without any alteration of the basic solution methodology.

An example showing how a specific type of line of sight

12

rate information error can be modeled is treated in this report.

Briefly, the error modeled deals with the seeker detectors field
of view -being "filled" by the nonzero size source at very-close
target ranges in comparison to the generaly-assumed point source
Vproperty-. This blroming error is-dependent upon the target aircraft track relative to the missile boresight axis as well as relative range.

A more detailed discussion of this error model is

given in Section IV.

MISSILE WARHEAD DETONATION AND KILL DETERMINATION

The- problem of terminating the missile trajectory with detona-

I.
4

tion of the warhead and subsequent damage assessment of the target

aircraft is -tremendously complex-and is generally analyzed: separately from the manuevering problem.

These analyses seek the deter-

mination:of kill t probability given relative -flight conditions sufi

L.

ficient to activate the fusing and detonation-sequence.

This proba-

bility is empirically derived by detailed simulation of the fusing

and detonation procedures and study of the warhead blast and/or

interaction with all known vulnerable areas of the specific

-fragment

Ittarget

aircraft.

Moreover, this :assessment is made for a-ll the

various dynamical aspects, range, -azimuth angle, elevation angle,

velocity magnitude, and orientation at detoxiation.

As this report

is concerned with aircraft evasive tactic metbodology, the kill determination phase is much simplified in scope, but sufficiently
realistic to be representative of -the lethality data typically associated with small close-range infrared-missile warheads detonated
near current fighter aircraft.
t y "kill" one assumes at least to disable the aircraft so- that its
primary mission capability is eliminated, but not necessarily torender catastfophic -destruction.
-

13

The simplification employed- relies-upon- COMPting-the- minimum

relative- range Rmin (from the equation-onpage 10) between thecenters of mass for- any -missile/airc-rat ~aetr
ar
The
minimum range value is ,useJ to de-termine akill--probability -from- a-

Llethalit~y

plot, as shown in Fig.- 2m4. -This -typical :function represents $$average" -data over all missile/aircraft orien-tations-andre!.ative velocities at detonation.
It -is-assumed -that detonationoccurs -at prec isely the minimum relative- -range point.i The -numerical
1Kvalue Is utilized to specify the k*-lln-nkll -outcome -distribution- employed subsequently in the -learning
_ and statistics, phases of
the computational approach.

rh

-14

In--

'U

III.

[-

COMPUTATIONAL APPROACH

The computational method is presented in two phases:

a "learn-

ing" phase, which is associated with the optimal strategy development for the aircraft, and a statistics phase associated with the
determination of the aircraft vulnerability and missile effectiveness measures.

LEARNING PHASE
Strategy Initialization
A stochastic learning algorithm has been developed for strategy
resolution in two-player, one-on-one duels in Ref. 1.

That method-

ology is applied to the aircraft/missile problem in a similar manner.

In this case one player (missile) e.,,ploys a prespecified guidance
policy while the second player (aircraft) has its strategy resolved
by the algorithm in accordance with a survival goal.

Section II

outlines the aircraft maneuvering strategy definitions in terms of

the observable relative coordinate data and elemental controls.

The

countermeasures strategy development, although not described in this

report, would be treated in a manner similar to the maneuvering
case.

One initially begins with the aircraft strategy represented

by the decision table:

15

U.

"ij

x,

where Lhe

1,

...

represent the observable coordinate de-

composition given in Fig. 2-1 and where

u., j= 1, ..., 5

the elemental maneuver choices as shown in Fig. 2-2.

5;

Pij

ith

table begins

0.2

i = 1,

for all

...

represent

Initially Lhe
and

j = I,

that is, the elemental control choices are selected at

random in an equally likely mannLr for each visited coordinate region

x.

Trajectory and Outcome Simulation

With the initial strategy for the aircraft selected as given
previously,

the simulation of a trajectory.

outcome determination,

and strateg:y modification cycle of the learning phase are begun.

The trajectory simulation begins with an initial condition, relative range

R, angle-off

.,

and boresight angle

being chosen

at random in a uniform manner over the allowable space of initial

u. is then selected for the air~raft
J
in accordance -ith the assumed starting strategy for the specific
region x. given by the range (R) and angle-off (w) thresholds.
conditions.

A control :hoice

The missile and aircraft trajectories are then integrated until the
next aircraft threshold is crossed and a new aircraft control deThis process continues until one of the follcwing

cision sele-ted.

events occurs, a min'mum miss distance

R "-48,000 ft,

or

t > tG .

(Rmin)

is obtained,

The region/control sequence employed

by the aircraft meanwhile is temporarily stored for reference.

The outcome of any trajectory sample is determined in the following manner:

For a given

consulting the distribution for

as shown in Fig. 2-4.

If

a kill or miss is obtained by

Rmin
PK

R " 43,000

classified as a miss.

16

given as a function of
or

to,

Rmin'

the outcome is

Strategy Modification
For each of the two outcomes
weightings

it(-T)

(a-) we employ the following

in the strategy modifization portion of the

learning phase.
Outcome
Kill

(-I)

Miss

These weightings are consistent with a survival rationale as a goal

in the combat for the aircraft (see Chapter 4 of Ref. 1).

i

aircraft is in region

using control

then the strategy table is modified from

modifying

I:i

pij

to

pij

ij

uk

and outcome

pij

to

pij

If the
-, occurs,

by first

where
(' )p i j

k
*k
Pij = Pij

The

pij

is then renormalized to form

Pi

pij
i

P
Pij

This process is carried out over all

pairs temporarily

stored for that trajectory resulting in an updated strategy table

for thQ aircraft, as follows:

17

U.

.3
xi

Approximately

"Pij"

100

traje-.'tories with initial conditions se-

lected at random within each region are sini:lated to obtain a converged decision table that represents the "optimized" survival
strategy for the aircraft.

STATISTICS PHASE
The statistics phase of the computat'ons now fixes the converged decision table and computes the aircraft vulnerability and
missile effecti.'eness measures in Monte Carlo fashion.
vulnerability data approximately

100

For the

trajectory computations

with initial conditions in each observable region were obtained.

The quantitative measure of vulnerability is the kill probability
of the missile as a function of all range

(R)

and

angle-off (,)

thresholds relative to the aircraft as outlined in the observable

set description of Fig. 2-1.

The missile effectiveness data are

obtained in a similar way, except here ehe initial conditions are

chosen at random from an arbitrarily chosen set of launch regions
that comprise the totality of all possible launch -:oordinates for
the missile (see Fig. 3-I).

The effectiveness data is given as a

kill probability for target range

tc missile boresight axis

() t

(R)

and target heading relative

for launches within the off-

boresight requirement of the missile seeker head for each region

tWithout any loss in generality we have assumed that

/2

at

launch; otherwise the aIrcraft heading relative to missile boresight axis nomen-!lature on each effectiveness chart should be
1f+'; - w/2.
18

7700

00

1~120

Fir,. 3-1

1*is8sile Launch Coordinates

of the decomposition in Fig. 3-1.

Within each region the

PK

value represents an average of all launches that satisfy the offboresight requirements

-LIM

= t 200.

I
20

IV.

COMPUTATIONAL RESULTS

SUMMARY

T
-

The main ingredient in this methodology development, when com-pared to other missile effectiveness analyses, is the evading. arcraft's employment of i maneuvering and/or countermeasures feedback
control policy that minimizes the kill
over-all launch initial
treated in

this report.)

conditi ,ns.

probability of =the--missile
(Only the maneuvering. case is

These optimized policies canin

_some

cases (depending on the specific missile/target aircraft -systems

design data)- seriously degrade missile system effectiveness: When
compared with system evaluations made with nonmaneuve.ing or openloop maximum

turn policies for the target aircraft-.

The specific missile and target aircraft data given- -in Table 4-1
and Fig. 4-1 and employed to obtain the results pre-sentedoare -representative of current-day IR missile systems deployed for close-.in air

-II

TimeC( ee )
Fig.

h-1 Longitudinal D,.v ele:ation Param~eter a-

TABLE 4-1

REPRESENTATIVE MISSILE AND AIRCRAFT DATA

MISSILE DATA
Aerodynamic Data
Reference Area
Zero-Lift Drag Coefficient

S
CD

(ft2 )

1.3 for M51

C0.0

Induced Drag Coefficient

0.01614
0.8 for M<1
for M< 1

C1I

180.0 - (13.6) (Mfor M > 1

Engine Data

Thrust (15,000 ft altitude)

Lngine Burn Time
Missile Mass at Launch

T
tB
T'o

(ib)
(sec)
(slugs)

Engine Mass Flow Rate

(slugs/sec)

2820.0
5.2
6.211
-0.3584

Seeker Data
Field of View
Gimbal Limits

(deg)
(deg)
(deg)
(deg/sec)

during flight
at launch
Gimbal Rate Limit
Seeker Output Gain

Seeker Time Constant

(1/sec)

2.5
40.0
20.0
20.0
4.0

0.0

Guidance and Control Data

Launch Delay Time

Autopilot Gain
Autopilot Time Constant
Commanded Turning Acceleration
Limit
Guidance Time

tDL
Kap
Tap
nycVM

(sec)

tG

(sec)

(1/sec)

2.0
1.0
0.1
30.0 g's
20.0

AIRCRAFT DATA
Mneuver Data
Dash Velocity

VD

(fps)

900.0

Steady State Turn Velocity

VT 1

(fps)

900.0

Deceleration Turn Limit

Velocity
Steady-State Turn Radius

VT

(fps)

500.0

2
RTI

(ft)

5000.0

Deceleration Turn Radius

RT

(ft)

3555.0

Longitudinal Acceleration
Longitudinal Deceleration

a+
a

(ft/sec 2 )
(ft/sec 2 )

+22.0
see Fig. 4-1

22

i'
f

duels and current-day fighter aircraft maneuvering capability.

representative launch condition selected is at a

The

15,000 ft alti-

tude with the target and launch aircraft flying at the initial speed
of

900 ft/sec.
The results obtained in this report, which illustrate the

levels of degradation achievable by optimal maneuvering '-ontrol,

fall esseLtially into four study categories as outlined in Table 4-2.
Ol

Studies I and 2 fo~us on the blooming error property of narro

of view sensors at close target rangL-s.

field

Study I examines the case

where the sensor field of view becomes saturated at a nominal relative range

Rs = 1000 ft

from the target independent of the missile/

c

target aspect specified by the track crossing angle

In this

case, the missile guidance 0command is neutralized (n

= 0) at the
yc
R s = 1000 ft point until detonation. The missile warhead lethali'.y
model as assumed for study 1 is given in Table 4-2.
the maximum
distance
creases

PK

Rmin
to

value of

0.7

is maintained for a mininum miss

of between zero and

PK = 0

,k ft,

at a distance of

the case where the saturation range

crossing angle

and quadratically de-

20 ft.
Rs

Study 2 considers

is dependent upon the track

The functional dependence on

is shown in Table 4-2.

In this case,

This particular

Rs

versus

for this case

c

relation-

ship is derived in a geometriL way from a simplified afterburner

plume radiance pattern andi sensor field of view limit consideration,
as shown in Fig. 4-2.
250 ft

The

21.

field of view limit gives an

for the tail aspect, inzreasing to

2000 ft

aspect for a typical afterburner plume pattern.

of

The

Rs =

for the beam

2000 ft

value

is maintained constant throughout the beam to head-on asrects

I..

for the sake of simplicity.

'The methodology is not limited to these simplified saturation effect cases as presented in this section, and, in fact, if extensive
model detail in terms of seeker characteristics is available it can
be easily incorporated.

23

cr.
7

s.~

__

24C

e)

'0

--

we'

Study 3 deals with quantifying the off-boresight laun::h angle impact on aircraft vulnerability.
in this study:

Two special cases are considered

the on-boresight c:ase and the

15-20

off-boresight

case representative of launches at the edge of the boresight launch

envelope.

In this study,

the saturation property of the seeker and

warhead lethality model are identical to those ci.ployed in study 2.

*

Study 4 deals with quantifying the -,han-.e in kill effectiveness with

increased warhead Icthality.

In this study, the saturation property

of the seeker is maintained as employed in studies 2 and 3.

The computational results for studies 1 through 4 are presented
with the following format:

In each of the studies the missile ef-

fectiveness results are presented first,

measures second,

the air---raft vulnerability

followed by sample evasion trajectories.

The ef-

fectiveness and vulnerability results are g~iven for both the nonmaneuver and "optimal" learned maneuver cases.

25

F
STUDY NO. 1

SATURATION RANGE SEEKER

-CONSTANT

The missile effectiveness results for study I are presented

in Fig. 4-3.

Three levels of

PK

have been arbitrarily selected

as shown to simplify the results presentation.

The actual

PK

values, however, are available in the computational results for

more detailed appraisals.
fectiveness

in

the launch

The diagram at left shows the kill

ef-

:oordinate space nomenclature of Fig.

3-1,

when the aircraft does not employ evasive maneuvering but is in

straightline flight.

The diagram at the right shows the degradation

in effectiveness made possible with optimal learned maneuvers.

*.. . ...

. ,p

.:
..

. . .,

.,

................................
*

Note

{ ..

.......

. ........

..
*

'.

. ......
. ....

..

*:*"

\.

112.2
"26
.

......
-.

..

. . . . . . ...

...

..
-

L '.'.."

"4-:-..:'.:
..

ll

....
.

::::/

..

...

.........
*..

.
.

..

regions shown in white, in all missile effectiveness, and air-

craft vulnerability results, correspond to levels

26

of

PK < 0.1.

IL
R < 5000 ft

that no missile launches were considered at ranges

in

any of the results reported here. This was an arbitrarily impcsed

2
criterion I.n -light of the missile guidance delay parame'er t
sec

and was implemented only for the purpose of reducing the com-

puter running time by limiting the simulation to fewer initial coin-

--

One can see the degree of degradation in kill effectiveness

ditions.

(0 = 600)

in the tail aspect

240 )

(0 = 120

and head-on aspect

to

Figure 4-4 shows

brought about by the learned maneuvers.

the missile effectiveness comparison between an aircraft employing

_7

LK
t

- :.:.

. - ......"
.. .

-i

. . . .. . . . .

, ..

,.:

..

R,...

K%.'

= 1000
,-Ft
0._
"_ .:/ .r;ZD;,.._.
--

rh Tur :-,
5 r, ._.1_
, ~~~~~~~. ..

..................

tianeuver

;Aircraft

r,Rig'ht Turn

S-5

Fig.

h-h4

'

".

.,Tunv.
Optimal)

.. . .3
0
" . - "..

<

Fi. 4 --,_-issileEffecitivns

. .

.0.....

. .

:
.

a e v
:..::..........::"'

.
.......... Cosat Satuatio

,_[ 0.3 < PL.arne.

0.

'~~03Maneuvering-

Missile Effectiveness Results, Constant Saturation

Rang= 000Fe,_ (5g Turn vs. Optimal )

27

5 g right turn and one employing the learned maneuvers.

5 g

right turn case -represents the kind of "open lo0"

The

maneuvering

usually consider-d in missile effectiveness analyses (see Refs. 1

and 2).
to

For relative headings in thp-right hemisphere,

1800,

the learned maneuvers markedly reduce the

- 00

PK< 0.3

level for all but one region in contrast to the 5 g turn. In

the left hemisphere, the reduction is less marked. The head-on and

tail aspect cases where

been reduced \to levels

PK > 0.3
< 0.3

for the

5g

turn case have

for the maneuvering case.

other hand, some regions now appear in the

learned maneuvers as contrasted with the

> 0.1
.

0.1

On the

range for -the

values in the

5 g

turn 2ase. This particular situation is broLght on by the discrete

levels chosen for plotting; e.g., the 5 g turn case has regions
for which the

PK values are between

0.09

and

0.1,

while the

corresponding regions in the maneuvering case have values between

0.10 and 0.14. Moreover, the sensitivity, of the learning algorithm is such that it :ould not differentiate between maneuver sequences that produced PK levels having those small differences.
At the risk of these small data abnormalities in some of the results, the PK, levels were plotted as computed without adjustment
for solution variability.
The aircraft vulnerability results are presented in Fig. 4-5
for the nonianeuvering versus learned maneuvering cases.

In these

data are given with respect to the aircraft observable stae description in Fig. 2-1. one can see the marked
cases, the

PK

PK

degradation possible in the head-on and tail apsects by employing

the learned maneuvers as contrasted with the straight ahead flight
policy.
We now turn to the illustrations of strategy and trajectory
sample results.

The results were generated from hard copies of

28

,.

".

/...

F
....

'
,

..,- ...

.,

.:...S:

" " " ".

.

..

.. ..-.

: --- ".
A%..;
*,,r

".

0.5 < PI,

0.7

r~ari
uv~ *r n~
0 .3 < PK
n~n!:aeuv_:.in,

0 .5

CO.!

Fi

<

0.3

.; -. ; S"

L a r~
Lanedrh
Mnueip

_. ,,-5
Aircraft Vulnerability Results, Constant Saturation
Ran.-e = 1000 Feet

solutions made available by a computer-driven CRT display.

Two

forms of trajectory plot are illustrated for each trajectory sample.

The first shows motion in coordinates relative to the evading

aircraft and is useful in demonstrating the learned aircraft maneuvering strategy.

The second is the normally drawn plot in absolute

coordinates showing the motion of both vehicles during the engagement.

The optimal learned strategy data, though available, is not

presented for each result in its entirety (viz, with regard all
-,,

regions in Fig. 2-1) because the general characteristics of the derived strategies are more- easily inferred from selected sample trajectories. Table 4-3 supplies
an outline for the S-azole results

21

29

E,

(n

4i U-4~

(nO

ul

CNI3-t
e'i

04

r-

1 %C

Ct*
U)
4 O-CO r-4 14

C4

w.41

-7-~

V)IV)IU
4
0

S4

U)

(n

04

$.04$4a

0w5-

'0
0

t~~~

4.j
0

-r4
V1

-4

tl.(n-

-r

U)ci

Co

0--

l0

'4N4A

0 $4C

00l

0
-r

0)

0-

0j

0 4c

zcJ
V"i
Co4

-4*r4

0)~4C

'S.

0j.'

0
.4LnLllt
-- IZ-4r

30

C:vt

Co

corresponding to studies 1 through 3.

For study 1, the constant

saturation range case, a typical "tail shot" example is given by

result 1A in Figs. 4-6 and 4-7.

Figure 4-6 shows the motion- in

coordinates rela-ive to the aircraft and illustrates the weaving

maneuver (control reversals at sequential range thresholds) developed by the learning algorithm to minimize the kii' ,probabilty
against the constant saturation range type missile seeker.

For

this specific launch initial condition the minimum miss distance

zompar.son is 4
respectively.

and

23 ft

for the straight and learned strategy,

Figure 4-7 shows the absolute coordinate trajectory

omparison for the same initial condition.

Result 1B, shown in

Figs. 4-8 and 4-9, depicts the case "head-on" initial conditions
for the nonmaneuvering and learned maneuvering comparisons.

Fig--

ure 4-8 again shows the weaving maneuver developed by the learning
technique in re.ative coordinates, and Fig. 4-9 shows the maneuvering comparison in absolute :oordinates.
-of a steady

5 g

Result 1C shows the effect

right turn for two counterpart initial conditions

representative of a "beam" launch aspect for the constait saturation

eange seeker.

The dashed trajectory in Figs. 4-10 and 4-11 shows

the turn into the missile approach heading, and the solid trajectory
shows the turn away for the counterpart initial c3ndition.

These

typify the kinds of "open loop" maneuvering control given the evading- aircraft in most effectiveness evaluVtions (see Ref. 1 and 2).
One can see f'om the difference in the miss distance results of
6.7

and

47 ft

for the couiterpart initial conditions -that a

single open loop control policy (whether a maximum

turn or

straight flight) may not furnish the evading aircraft with its best
defensive policy when compared to closed loop control over all
relative coordinates. The effectivene3s results associated with
the open loop cases tend to overestimate missile capability.

31

EVRS I VE STRATEGIES
PARAMETRI C-CASE : >

STUDY NO I1

INITIAL CONDITIONS
Mnw,
110ANGLE-OFFx

SCALE

*ORSIGHT=
3S00 FT.

AND TRAJECTORIES

100.0-

RE'S4LT 1A
Tao.
Ir
PEIE

C.

0.0 DEC.

MEIN z

11ME'IE
0MN: w

Fig. 4-6 Evasive Strategies and TrajecLoii---.

Study No. 1, Result 1A (R,,-az'e.)32

E
A.0aT

E*

. 13 e.
it.? FT.

[i

EVASIVE STRATEGIES AND TRAJEC'TOPTES

PANANETRIC -CASE :->- -STUDY No- I
INITIAL CONDITONS:
ANK
a 11900. 0 FT.
3HEAWING- 2f.0 DEG.
*ONESIGH4Tv
0.0- DEG.
SC". 3000 FT.

RESUJT IA
TIME"I,4.3
'TtME
RMMt~

4A~SC.,

7?.
.9Sc.

It.? FT.

NOW
STRIGH

IA

dolt-

xw

WE

f.

Fig. 4-7 EvsiveSrtge

Std

I3

,Rsl

n rjcois
kAslt)

Vj

EVPSIUE STRATEGIES A~ND TRA~JECTORIES

PMAMITR IC CASE :)>

STUDY NO I

INITIAL CMWOTIONS:
RANGE
m 20000.0 FT.
McO DEG.
VANGLE-OFF=
BL)RESIGHT=
0.0 MEG.
WCALE 3000 FT.

Fig-. 4-8

RESULT 18
TIME
MIN
TIME
RMIN

9 .S SEc.
1..2 FT.
6.9 SEC.
24.9 FT.

Evasive Strategies and Trajectories,

Study No. 1, Result lB (Relativ7e)
34

"MM I I

EVASIVE STRATEGIES AND TRAJECTORIES

PARAfMETR IC CASE :

STUDY NO I

INITIAL CONDITIONS
= 20000.0 FT.
RANGE
170.0 DEG.
HEADING =
0.0 DEG.
BORESIGHT=
SCALE 3000 FT.

RESULT 18
TIME
RMIN
TIME
RMIN

STRAIGHT
MANEUVER

=
a
=
u

6.6
2.3
6. 9
24.9

SEC...
FT.
SEC.
FT.

EVASIVE
IltEUVER

I!
IL

I.. -.. ,.

. .

..

.,,,.,

Fig. 4-9

..

.. .,=

= :. =;

..=

Evasive Strategies and Trajectories,

Study No. 1, Result 1B (Absolute)
35

. .

. .

EVASIVE STRA~TEGIES ANhD TRAJECTORIES

PAIIETR IC CASE : >

STUDY HO t

INITIAL C09VITINS,
RAWG
14000. 0 FT.
255.0 DEC.
ANGLE-OVF=
BWES t 4T r
0.0 DEG.
INITIAL CONDITIONS :TIEa
*14000.0
FT.
RANGE
WA

BMCALE
~
0.0 DEG.
E-FF
FT..uN*105.0

Fig. 4-10

RESULT IC
TIMlE x
WuIN z

10.1 SEC.
47-.0 FT.

TtE=

S
KC
S69C
**

Evasive Strategies and Trajectories,

Study No. 1, Result 1C (Relative)
36

T.

-"

'.

- - --

EUASIE STRATEC-1-ES AND TRAJECTORIES

[.PARA'ETRIC

CASE

:>

STUDY NO I

INITIAL CONCITIONS:
RANGE
= 14000.0 FT.
HEADING =
iORESIGHT=

SCALE

-75.0 DEG.

0.0 DEG.

3000 FT.

RESULT IC
TIME a

10. 1 KC

Rlfr

4?.0 FT.

a
TIME a
WIN

.6
6
B.?

50 RIGHT TIIN
ANW.

4S MIGHT TIN

1,1ont

Fig. 4-11

Evasive Strategies and Trajectories,

Study No. 1, Result 1C (Absolute)

37

o...

STUDY No. 2 - ASPECT DEPENDENT SATURATION RANGE SEEKER

The missile effectiveness results for study 2, the aspect dependent seeker case, are presented in Fig. 4-12. The diagram at
the left shows the kill effectiveness in the launch coordinate
The diagram at the right shows

space for a nonmaneuvering aircraft.

the degradation in effectiveness possible with the optimal learned

maneuvers.

A marked reduction exists in the tail and beam launch

orientations in the
on capability in the

ranges and elimination of the head-

5-16 K ft
to

16

The aircraft vul-

ranges.

48 K ft

nerability results for this study are presented in Fig. 4-13 in

One can see by comparison

coordinates relative to the aircraft.

~A
.

. .... *.. * *...:..

.-

...

".

~Y

...

5 <

PK0.5
.<F
_3
honmaneuvering

03

<

E3o.i < PK<

Fig. h-12

0.5

0.3

Learned
Maneuvers

Missile Effectiveness Results, Aspect Dependent

Saturation Range = 250-2000 Feet

38

UL

-0

..-.

maevrn

fo

in
Fis

std

4-4ad41

se

th

Sml

eut2

ie

lutaephennaevrn/aevr
ai

htinta

aevrn

hard~~~~~~~
obai
a.ba
to.
by

FA

2/seTbe43.

comparisons fo
One~
can

-7

trn evrsa

:~Qar~u~eiT~-

oniin

taeyo

sturtin

icati

otr

.elne7
eaiv

seta
th

tmdumrne

39P

rng

.~

agsfloe

thesoldof

200

re

TRAJECTORIES
EVA SIE STRATEGIES AND
PARAMETRIC CASE :>

STUDY NO 2

INITIAL CONDITIONS
RANGE
15000.0 FT.
ANGLE-OFF=
145.0 DEC.
BORESIGHT=
0.0 DEG.
SCALE 3000 FT.

RESULT 2A
TIME
RMIN
TIME
RMIN

a
x
x
=

11.S

EC-*,mi
0.9 FT.
10.? IEC..
26.5 FT.

Ai

LL

Fig. 4-14

Evasive Strategies and Trajectories,

Study No. 2, Result 2A (Relative)

40

EVASIVE STRATEGIES AND TRAJECTORIES

PARAMflTRIC CASE : >

STUDY NO 2

INITIAL CONDITIONS
RANGE
= 15000.0 FT.
HEROING =
35.0 DEG.
BORESIGHT=
0.0 DEG.
SCALE 3000 FT.

RESULT 2A
TlME
IRMIN
TIME
RIIN

a
=
a
=

11.5
0.1
10.?
26.5

I
JI

/EVASIVE
STIAWIGlT
W4EUJE R

MAEUVER
I

I
II

Fig. 4-15

Evasive Strategies and Trajectories,

Study No. 2, Result 2B (Absolute)
41

StC.,.
FT.
SEC.
FT.

to open the miss distance.

It

is

important to note that in the

interpretation of the optimal learned maneuvers the criterion for

"optimality" is not maximizing the minimum miss distance for the
evader,

but maximizing the survival probability.

This criterion,

for P'-'rple, has the effect of rating control sequences that

achieve miss distances of say

20 ft

and

head lethality as equally desirable.

2500 ft

for this war-

Therefore, the learning tech-

nique may show optimal maneuvering sequences that defy intuition

for certain portions of the relative coordinate space, yet satisfy
the stated criteria for optimality.

Sample result 2B illustrates

the maneuvering comparisons for a close range beam launch initial

condition.

The trajectory result :omparisons are depicted in

Figs. 4-16 and 4-17 for both the nonmaneuvering and learned maneuver cases.

The guidance delay parameter

tDL

is a particularly

important missile characteristic for this launch condition, and

the execution of a proper maneuver sequence by the evading aircraft
can definitively degrade missile effectiveness at these close ranges.
STUDY No. 3 - OFF-BORESIGHT LAUNCH SENSITIVITY
The results for study 3, which is aimed at illustrating the
sensitivity of target aircraft vulnerability to off-boresight launch
conditions, are presented in Fig. 4-18.

The plot on the left shows

the vulnerability for the on-boresight launch case.

pare this plot with the averaged result obtained for

One may com-200

to

+200

boresight error spread as shown in the right hand plot oE Fig. 4-13.
The comparison shows the vulnerability to be basically the same for
tail aspect launches and slightly reduced in the beam and head-on
situations for the on-boresight launch case.
can be compared with data for the

15'

rors at launch are restricted to the

to
+15

These vulnerabilities
200

to

off-boresight er+200

edge of bore-

sight condition; therefore the right half plane depicts results for

42

EVASIVE STRATEGIES AND TRAJECTORIES

PA-w-wIE--TRIC CASE :?

STUDY No 2

RESULT 29

INITIAL CONDITIONS:
--

ANQLE-OFF=
WUESIGNT=

110.6 DEC.
10.0 DEC.

MNa

T Ea

1.

91SC

II-

Fig. 4-16

Evasive Strategies and Trajectories,

Study No. 2, Result 2B (Relative)

43

---

1
471

EVASIVE STRATEGIES AND TRAJECTORIES

PMW*KTR I C CASE : >

STUOY NO 2

INITIAL CONDITIONS :.
MAO

7509.0 FT.

.AEING10.0
IKSI
CHT=

$cM. 3606 FT.

DGWIN
10.0 DEG.

RESULT 20
TIME ',

# STRAIGHTEVSE
MWUER

UER

Evasive? Strategies and Tajectories,

Stud-- 1o. 2, Result 2B (Absolute)

44

2400.0 M

Fig. 4-17

WO-,,mm

11.4
S 1 FT. "KC

TIME

WIN

.4

-NE

. ....---

-R-

I:J
*A,

Borresihh

a~

15'
to

with
resec

tha

sinh

whe

20

LaunchI

totre

copae

ie

ethlfpaea

-ad
1'

fsgt(eeFg

9.

to th/nbrsgtd

launchtcauch Sapl

httela

trjetoie

depicting1

0.45ei~tLuc

2"

la

fet

theevasive

_. J in:r 1,
a

a,- Initial

ji,-.
,-9

>c:d/La,

inlvi

al ?ordiion .o

enlature

maneuvering for the boresight sensitivity analysis are shown in

Figs. 4-20 through 4-23.

Figures 4-20 and 4-21 depict the on-

boresight 'ase for a sample head-on type launch condition in terms

of relative and absolute coordinates, respectively.

Figures 4-22

and 4-23 show the "edge" of boresight launch conditions and subsequent maneuvering representing the lead and lag cases.

STUDY No. 4

-WARHEAD

LETHALITY SENSITIVITY

We now move on to study 4 and show the sensitivity of aircraft

vulnerability with change in warhead lethality.

Table 4-2 describes

the improved warhead model of study 4 as assumed in the comparison

with the baseline warhead of studies 1 through 3.
warhead described as the
baseline warhead called

PK (18,6)

The improved

warhead is as lethal as the

PK (6,20) at three times the minimum range.

Figure 4-24 shows the vulnerability results for the baseline and improved warhead fir both the nonmaneuvering and learned maneuvering

46

EVASIVE STRIATEGIES
PARAiIETR I C CASE

:>

AND TRAJECTORIES

STUDY NO 3

INITIAL CONDITIONS :
s 15000.0 FT.
RANGE
ANGLE-OFF
-40.0 DEG.

SC

9WRESIGHT=
E 3000 FT.
L

RESULT 3A
"I'ME a
R(.,IN =

5.6 SEC.
1.5 FT.

0.0 DEG.
!

'--

Fig. 4-20

Evasive Strategies and Trajectories,

Study No. 3, Result 3A (Relative)

47

EVASIUE STRATEGIES AND TRAJECTORIES

PARRETRIC CAS

:>

STUDY NO 3

INITIAL CONDITIONS :

RANX

a
HEADING
9MqS IGH T"

SCALE

15000.0 FT.
226.0
O. 0OEG.
IDEQ.

RESULT 3A

TMlE s
M1IN w

5.6 WC.
7.5 FT.

3000 FT.

EVASIVE

Fig. 4-21 Evasive Strategies and Trajectories,

Study No. 3, Result 3A (Absolute')
48

EVAS I !'E STR(-ITFGIIES A~ND TRAJECTORIES

PARAMETRIC CASE :>STUOT NO 3
INITIAL CONDITIONS
RANGE
=15000.0
ANGLE-OFFm
-40.0
SORESIGHT=
-1.0
INITAL CONDITIONS:
RANGE
15000.0
A~iLE-OFFx
-40.0
3ORESIGH4T=
10.0
SCALE 3000 FT.

Fig. 4-22

RESULT 30

FT.
DEC.
IDEG;.

TIME
AtIIN

FT.
DEG.

TIME *
RrnN a

DEC.

S.0 SEC.Mmm
23.61 FT.
S.? SEC
?.4 MT

Evasive Strategies and Trajectories,

Study No. 3, Resul.t 3B (Relative)
49

AND TRAJECTORIES

EVASIVE STRATEGIES

STUDY NO 3

PMNIWETRIC CASE :>

INITIAL CONDITIONS :
a 15000.0 FT.
ANGE
202.0 DEG.
HEADING =
-16.0 DEG.
mORESIGHT,
INITIAL CONDITIONS :

rIlE a
MIEN =

TIIE -

a 1500.0 FT.

WINCE

IIN a

233.0 DEC.
1O.0 DEG.

HEADING x
l*0R1SIGHT.
SCALE 3000 FT.

RESULT 3S
5.O SEC.,..u
23.6 FT

5.? KC.
4 fi.

I
T

EYCM I
4-23

Eto
*

L+

W+

t44
50

Study~~~~

:!i

+ ' +.

[.S

LS
-1I

No

ndTaecois
taege
Fi.423Eaiv
~~SNo5,ReutB Ablte

,Rsut3uAboue

~+

III

0.5

<\

-0.

t<P

0.

(38,60) Warhead
onaneuverinLandlaeurin

p?K (6,20) Warhead

Learn05c..c04
Firn42kneuvew

'g0.

0.3cieesReutWrei

Ltafy-td

cases.

The improved warhead model, as the results show, allows

much less of a vulnerability reduction with optimal maneuvering

than did the baseline warhead.

With such an imprcved system, a

tactics analyst would perhaps divert more of his effort into the
countermeasures policy area ratner than rely upon maneuvering to
achieve vulnerability reductions.
This study concludes the computational results portion of this
report.

The studies illustrated were intended to furnish a small

sampling of the type of result that is available to the system designer and tactics analyst in system effectiveness studies using
the described methodology.

52

V.

CONCLUSIONS

The important relationship of aircraft evasive maneuvering

within the analysis of missile effectiveness has been made clear
with the numerical studies conducted here.

The first three r.vsult

sets clearly show marked reductions in missile effectiveness

achieved by the optimal "closed loop" maneuvering over the nontmaneuvering and other open loop control cases for typical close
range IR missile combat.

In the constant seeker saturation range

case, the optimal evasive policy for most launch coordinates is

essentially a weaving maneuver whose control reversal points are

dependent upon the relative range coordinate thresholds. In the
aspect dependent case, the aircraft policy is essentially a maximum rate turn to present a beam aspect to the oncoming missile
followed by a maximum rate turn reversal at the seeker saturation
range threshold.

The off-boresight launch results show that opti-

mal maneuvering can achieve considerable reductions in missile effectiveness at close ranges when launch off-boresight angle lags
the line of sight.

The last of the results configured the improved

missile warhead to be sufficiently lethal so as to obtain negligible

effectiveness red,'ctions with optimal maneuvering.

It indicates

that the major burden of the evasion problem should be placed in the
domain of countermeasures deployment.

The optimal maneuvering data

(particularly in the first three result sets for the baseline missile
warhead lethality) when averaged over all of the launch coordinates
corresponds numerically to those statistical levels achieved in
actual combat firings.

The computational costs are reasonable, re-

quiring one-half hour for the complete solution set for the 2-D problem with projected running times of two hours for comparably detailed
3-D models on IBM 370/168 type computers.

53

The technique as described can be incorporated in existing

simulation programs where the evasive maneuver optimization can be
programmed to become an integral part of the trajectory simulation
process and where the fusing, detonation, and warhead damage assessment portion would be maintained and computed in its original
detail.

On the other hand, the methodology here can be easily

augmented to include summary statistics of the detailed warhead

damage -effects to provide a self-contained program in its own right,
which might offer shorter computer running times per problem than

the former modification.

The methodology also is easily extendable

to countermeasures stm tegy determination and to problems of three

dimensional flight.

If necessary, the methodology can permit analy-

-A

sis of missile effectiveness to be achieved with in-flight alteration of missile guidance or counter-countermeasures policies.

This

is tantamount to reverting back to a two-sided game model for the

combat in place of the one-sided control problem as analyzed here.

--

I--8II

.5

I-

S[

"4

VI.

7-1.

REFERENCES

Drane, R. B., Capt. and Duffey, J.

G. Jr.,

Capt.,

"A Digital

Simulation Model for Evaluating System Effectiveness of Airto-Air Missiles

(U)," Air to Air Combat Analysis and Simula-

tion Symposium Proceedings AFFDL-TR-72-57,

Vol.

II, Air Force

Armament Laboratory, Eglin AFB, Florida, May 1972 (SECRET).

2.

Chandler, W. E., "Survivability of Maneuvering F-4 Aircraft in

Air-to-Air Combat (U),"

LIMC-TP-74-39,

DDC AD-531-557L, Naval

Missile Center, Pt. Mugu, California, September 1974 (SECRET).

3.

Hyde, J. P., Lt. Col (F. J. Seiler Research Laboratory, AFSC)

and Hines, J., Capt. (Department of Aeronautics:, USAF Academy),
"Application of a Computer Simulation and Optimization of a 3-D
Engagement Between a SAM and Tactical Aircraft to Determine
Optimum Evasive Tactics for AC-130 Gunships (U)," Air to Air
Combat Analysis and Simulation Proceedings AFFDL-TR-72-57,
Vol. II, Air Force Armament Laboratory, Eglin AFB, Florida,
May 1972 (SECRET).

4.

Falco, M., Carpenter, G., and Kaeccher, A., "The Analysis of

Tactics and System Capability in Adrial Dogfight Game Models,"
Grumman Research Department Report RE-474,

May 1974 (prepared

for Air Force Office of Scientific Research under Contract

F44620-72-C-0032).

55

Ai

APPENDIX
COMPUTER PROGRAM LISTING

56

S8808888;;;;;c~ ;;0~~

0000000000

W7

cc
IL

)0-

;2
ILco

I-~
(OZO

0 m '-e

UV
-JLJ
>
U.; it

~1~~'
-

it)~w

-0

U .-

*a-

%0

p.

'j

~0OU

~&

0no

4.

0-

C).

U-.

Z~~~
If'%OU. CY
O~

. $1

i
(4

J -

4(YtL
0~~~W

X0

0>~

0
V~U0>00
~
NN.0

fe

N-

qaj I

VwuW

NU
CO--

W
AI~
(~w'
op

0.iO
-

I.

WO

-.1

MO40-OV-?

8
4. J X

W
M

Dh-4

40

6
030-nI

^4n

00000
0000080000

4LLC

Z
q~

On

OJ

0*0000

0iW

Tl3CtZJ

<

)X

;nJ

0000

-1.j

x3
0

0N

z.-

J~- l- l

w
bu

Cc40 81LO
m

c-

wN600.\.N

IS q .)

~
0. ~ It

01IHZJII900I....P.0
44K
( ( 0--

'O

-RJ'..4

Om
Ia30

4 *

(r,-o

4if-.

3E
4aN~~~~~
NZ
0

vaA

-Mwwb0

;.

JjW

0- >Ifl0-.0
-J

0-0

NCw
le

0W
a

Op Z)-~a
'i0:
(/ 0- W

I- w

00

00000080

57

14)U.:

co

00go
0000

I--

&A

v~z

0z
V0*

)ZA

Z0.'

0.

11-

0&Cc

IL.1-

W tL

p-

4c

0ol-

ff

0W

0.

Howw.
XC
of

0
.Y

in

W~N
N

4
1

W'nIA M

1-j4
0440Iwo-OU)V
OMm--w
w
-0-0V

V0\.
k.o

%Id
2

4Z

V-

'I)C;

On

P-10-1

~V

in

(
*'
MM ow

00

0.
C

1-il*
-

14

a-.

'..ns

***

"'0k
-11
1,uj.~

j t-

0.OZ0.IL

owco

%4().

0WC

ir

pa

of

Tw

C-C911N--

Z0.M

-M-Z4

0f

OOM'9--

w'

0000

'

-C

M#Ps
ZW- j O''

0-

on

0.

>

4
.4

*..J0
IL'OIVW

11 xXI

-1

amCIUI

0
*o-9
A

I.
j
& .^4
0

~WW40.

9
Im
4 -. 0-

4" N

W-0

go

YW W )Z

oopflji

-S

imn

s -. A
O*-

4
*~flw
w Ok4
or 1

onsl-XI-0O
C
0..X600)4C
. xN4J
0
1'
-!
ahIH

0.

W4

Coloww-10fl a

ipJmpUI;I'

0.

01

If.J41CMLI *
Ali
MIw0.0.0

ILV

i
N

)I I

rILO'00UC4

0w0-00....s
N 0- -~~
-s - - A

4
U.

4
%%

00

If.0

a-0

49

I-

1-

N
z

0 W40.%

Vj

~ZZ*

w U;

000000000000000000000008

58

o.

oO~oo~oo~0000040000000040000000
000000000000

000000

;*00000oo0ooooooooooc

000000

0000

z
0

En

uj U
T

zz
U.
~
0

r 1A4

I0-

4
Z

0o
oU9- 0
MfI- t,
U.

be\ x-

OC~~~
~ ~

I-

I.--I

)
0

CL414
1z I

cc

O1 _ -Jw-

0-

ori

0000000000

" 0

.-

Z_ aI~ 00d
- a

Zn. 0

: ll

z11 1
U
0Z-

-Z

0a

W04

**D0
-%w).0

a
t~ N N

0i

t-xxxwo
0000

-~)

-4000

4mw Iw

09U,<r .r

woiw

Iy
01

NI-00-0

&Q

j
000

AA-j-

w
-Ii11Zv)

go

va;

-J:*
Jj

oc
000g

. xlx

w
-

00
-Z-

f)%

fat(

1-

9w(.O

-W*
O0z
W114
0

0-

Zon

CO-

z
ZCJ'WU
a

O1

oI-w0

MfE.N44eZ-x

-1
0

0-0z

VW>

@U)U)
0

9-' zi-- N

_$

O' I-

0xw

0-4
4c~ &Z . -

4j

i0w

00000O0000000"we.040

U-

59

00000000000000000600000

00000

a.wps

11

uj

LU

0
Y

0-

I4

C
-

-~

m
0

-0

0-

CLI

a:

008
;0.

0wXV)>aU

00

-]lo

f:P-

>
4l
-

MX11-

0O0
11-

lipif

C
-

""N

C It0

-400*0-00

mxx
O'- I 7Nft
-0 U Z

p-

00
cv
OuWUf

* f 0

&.

00

Of~

--

p"3

N-'X W*XK)-.4

* -w

q**

it1J1

~ll

>

P.
0

>

IfItJog
0

jj

-r

*C0.000

11

0z

jXV)

fi-

IIw
UJmw

-X

I-OO0---I IMZ~0Uc

- -'4 mJU -

ofoWO0
U:ULLAf.+

"-+J(NJX

d;.N

W
g

I0OutuI-

-Z X

Wf

N- 0IZ-1I
ii-&O
),0itif-~
0-0 fill

J>N
U.'

20
0

\0O
.a
j
'OuW
C)

CVJ..J

X im'.

N>O0

ii .
YP.

*I (

P*

04

-f

WU
-"-

^0.C~ +

111-

N3 I()

(14J11

I tX*OWW

.JZ

Dn')D

Ql7j.j

6e0ez

CY

C1

D--O-

oiZ x xv110 -oon

X)N
m4

Nl

(NJ

z
Of 9 NO-(I

.1)DI40%-Yp,
400

NNN
4,40

CY~'

~~~~000
*

t
0; 0

U.

60

###4##O4
10fl

000oo00 Cop

000000000000000000006000000000000@000000000000000
1016

fl00iN1h

cc

cc

*0

we

0k

OOa""mmo"

0O

0000000000@0@00;;;@@@0O@00@00@0@0oo0o00@00000o0000O

I7J

z 0-w

a
I.).04

.j

U,
W
-

IL

0-

4
>0

IJ la
WZ

9-.
4f (Z

ZL
11)
ill
N,

To
't
0 W-6M

P-

40-fl2

0.01-0w
;u

>

z
0
A

tL

"

uj
wm

i+i+I

3*U. tu ow~j

6.
0 Aj

11

WD

n I ec

~t~n
IU 0-W"

j
ewo-

-- 9

ZuI

ONWII
I0.aC

NO

C- 90"No

U9-i
ez

QwLL

510

AOgO
0 "
40

~Z~

4
vo

0Lo

Oft

1- W4

90

II0-

0X)O
0

D- u4x*
L
0 crZa'3L

ku

w
;iG

, .'
f)yf

4'

<-M

fr
x If)4

'"0Z

0IL 0
N

N
i

i#

il-1OUC

U.
1
.
-0-'-O.9-U.
OJ.9.-CO

N cl

-'

4011

oooma'041

O"I.
004

0"L"S.
00000000

z6

00 s "0.
000000000000000

"o

.I

IC

"

1o

11

1
--

0V)

N-i' .~J'

4""0 "o

0p "P

-.

- ..

~ ~
- -

- -

aa
U!~
~z1In

Ai.

>.

P."

:.

04

~UJ

N-

m 4
:0

ujNUN

LL

>

le

N".l

-0

N 0g

.1

Nxw..
W-

1-

0~

F)

'fl04440044w

11
*0C
0

p-~p
9-

Z"

'.~J.JJWS

O-9u.xa
WII*

1.

clJV00 9-

*eX0..
o

-ZI-

1-0.

000000

00000

"

0me

II~..qj..oV13,COw

62t

00000
mO
9.
"- I -

CII

OIIZIt

I- .

If

-JJ.-

--

4O--9

0Z

IV

*Z
-40

JO

P0~-.)

LiIJUIr00

004440049

t- 0. :) 1111if
ax.. ~
-

sowe

- --

44 1-

-.

Z>>>

-^

z4
>

p.

3 W91

000

Qocooc>

O000000,000

t-P

I I'.-

000000000 000000000000000

IIJPR

0.

0.

~o

U)

10.04

00

MD~

N_-

I--

#-~
4AJ

Io

U)

IULU)
V (A
0 C
0

11

--

ZZZ(A(A

w w wII

It

.3

0O)11o.

L4
tL."a g.

w~ V)
j4

0JC
IxCLV

S00~

ILU.J

x000

*
-~~0r

49~l
W A.L0

-.

-II
*

if

Nov.

00
0

0.

00

0..I

~N....-

ow

CA

'*

0.

aJ i

ZU)-

4K--

zJ

P*Z.'Z

Jw
0.
Uf~w
U.111
Z.
,U
9 AiJ

ZNNNNN~~U.
W
m

*e9W
;AU4..m.
N

onPi Of

IIv'X
U.

.Nj
4-9.u
U.1a~

10- i

N~X Z

JN.pn.7

UJI *

ODIC

0O"Uit

& 0Aooft

Q~
Z

"w-.- O

Z9IP" V0.

of,0.0.k

0^

*Z
0.70.

0..

Al

ZWWOCCD

P-I.L

ti.

E.
0.av

~-to

.j

O9-u.x
0.

.aaItLaAV-x~

0
f

0 P

0
0

z
IN

.
00

CI-ilZ~a7
YU
0.I
m
-ir)I
'o

000

0
P.

U.
j

N6JNNNN~N
00000orr

v Q'iu

NN
00

N
0

NNNNNNNNNNNNNNNJN

ANNN

00000000000000

0000

63

NNC% NNNN
-00 0f000-

:.r,-000000000000,400000

oo

C.

C4

co
-

0- f*

00-

"II

I- di

z0

U..ZM.g

K-*
4-*

y
> Irl
*

m
i

01

1- -C*I
ve

CO

I--

Z
w

,O

vee

3N~

W-%
aZ

JI

.0000

fr

lgoi

~~64

000

000000000:oo00

00

0000000000000000

000000000000

0000000000

00000000000

00

-w

C-

a-

04

UJ

x 41.

11mV

o
N
4:

of

%j

4It
w.

u~

**
4 494
~-JXX
tv
0

0- U

09

X4-

CV

a
*-W

I-.

*
O
No *w

l-

-N

z
0
Ii
.P

*
*j

II

Jx
XPIL
Z ..i 1If1

Xi- - - -

p"MI4~
in

4)

0:

0
449

m (~u
X

D4
z V)~4
0Io . 11111111

'.I#-N

w .wwW
Aa449

IIM

OVOOIIM*
o-

-O).

4--

I-N

a-

I
N

-8

0 ?w* ".
0 o- o\
0.
owo W

0.

IN

P-P
*OVO

0+

()

0)-

op-

440j

- -

MX)ak.wZ.
W

1-

O*401
Pf

0o

I Z0

i
*&N sor***

~
If 0

111cIf l

*
i>;J,0
0X~f

IfN0
oi
Jl
U~ C~0
P1

;;4--.
I**ZZ(g

7l
.

*-

0211'-

ii-Z

id-

ZZI

40f

0 O"

3ETZ0

44

00
0

00
008000000000
-

002 00001
U.w4~o~IiIIZ
N

0~-0.00

0W

0000- -. -I'O~0 000000000000000000

0V

00000000000I0000

0000

v#
0.

.00l

W00

~4*I*1
~oxvi~o

-O~.O~u.II--x~.LA

u.Z-u.065U

00

ODI

z
02J
*

w6
lopo

"

401-

'0

4c

IIN
If 9 I'12

IL

ftI

N U.6
-*

U.

oIL

%V

WI-

I-

NCq

8.0

41

J.

eN

N-

It
P-1..

U-

0(

0-

4 V)

II;M

-ec

0-N

~--.

NN
so-."1
DZwolail

11 14
A ,H

00Z

XHOfNN--

MI

I-*00

gjWN

* 4If
* -0'.Jo

Io *

*w.,
0@

1ii

4(

0-011111

4901

0l

-1 J

WHWv~wv. II

IIP- 11 1 1101

-WQIIKII

>V1-

gn.u)4

67

-P~o

SID~
30

W.0' 0-N04
~o*W
e0
0 Cc0 go0
Ud&X69

113EIeI,;

I
o

0-0-444am

us

ax

IA*

11a

#0

*o

w- 0~
X EA

-l0Va
w.J

00

IeH.
I~ I V s

I*J*
U-

f00

eS
W.

4%W
oft-I
ofz*
fl
00 0V
0.~e*,wtx
Iw
o .4
Z A
a( o-V 0

>

aI

I.s

V )

NEKotat

V-)0

4s

*.t
-V I *1

;*0

>a

UfI

3.1
cy

fA

ve a

w0 0-0

ow W
e
U.le

anew-I
>- 0

4~~4in
Z

ODA

0
(Y

IfI
CY 49

0-n
w

if

-.

1 4

Ob*

z~~
49. I

Z0
NOW

00

q-

0
Or
Nw
1-

41t
-

- -

f0-

0*XM
Z
.

X-

0WN**UJ
0"44g
luxxw~o411
~
4c

Z4444If
J.ji

C C

NAW.
+me

.-

NK
Ij04

04
Mm

404o

t*u
11 0

3I U 4c

ww-

t-0

ZAfl

1J4n

~
if lIf

to1 0

I 11N

If 11EX.UI0,.~
11

-J

z
0000000000000

00000000000

U.

68

OOOOOOOOOO0000oeocoooocooooeocoooocooeooOco0oeoo
M" "MMMMMM M"on
M"W MWOO~OeON

M"M"M" WIN M W

IIW

z I.4W
I

a-

4 ..

0
w

A
0*

CC
0ac .

W aaJ
-

IN

tU.
U.~y~a
U. w

0
LU w

I-I

U.

ie
x

LI

fr

.4 *~us
cX 4
a
o-

* .

Ja

A>

U.
z
.

Xw

2
-

3: a-

V)

-ow

Z
0 U.
M* 0.
I-

Uux
21

-J.

x
0

I-1
V)

z.

A4

(3
w

Li

(304
+

1-

0
0
N
0

f
tw

OW.
a s
0
*

4.L
44

1,0

WCuI cn
W
0 1,- NO
.4NI-N 04<U,.
LN
4- a
V)1-f
U~eZ. w
~1--sl
0
V0
Z Ioel
'A
0 0
Ud if If 1- 111111 1I1111uj N 111- 11
1
1
11- V)
I
-4 N 1LZM,1-NLLN.4D4o<
a rI

9 0

=Y

U.

. ..
........
..... 3

1-

ILuI

1-

0
0

U.

4
*

wi
t4
<j

0-.
(31 xII

zi

:) 4

'

0
-W

>

V)

U.Z Z z

.1

01f

_j -A
A x
-- JO

P'-4W
1. 0I-Z o-zw~z
XtL.4

x- Izo

>

u 4K

1-.

I- ct
O
us

w
I
I
ZO0
-usC I)
iI'. P-

-(

01

S
DC
x1:

MU
I3-4
Z(

tu
L-

10O

Z 0 1 0

m'

(L~

0.
0. ul
:) z
OD
il
U
:L

3
a4ax.

LJU

b.

il

IU)
W.

~U.

U~b

II
'
U. Li

69

"P

e""" N

000000000000000000@0000

4c

0
2

0
-10

I
x

0I

oYN

UL00
+:

11
wu

S-

x -LO
-

00

440cx-40
N
0coo

Mo 0

oa
0

I 0

"J0-

.ONNI

0"

fZ- w

t-

MN

o 0 0

0-.ILL.N

0.000000

00000000
U.i~O

-ow

*0
I

o
a

0Op

0)-U

000NNI-

0000

0000

00
eq

00

0000o000000000

4-

0 000 000 000 0000

~.

%C

wz

II

x U., z

M E% 11 Ip 1

Nw

-.

10

0 0
3j

02

LU~

00000

-C.
-J71

f-

tvS~

00

WN.

OUU)

I-1

14

in

Zt

00

IK)

0OOOOQ2
)

coo

IL*

'-OW

k~e72

000000

0000O000000

00000000
00000000000000000

0u

0
0

Wl 0.
0

'

0 *% .44

0-

...

aco 0 o-c

00 1

t.f..
1N 0 0

Y0.

:0I0 - 00
1.-

-O
1.-

--

0
'.4

000
>I

000

00000000

0000000

0
;L 00
w7

0000000000000000000
O.4NMO
Meo

ri

0000000000000000000000000000000

P--

o-

0
14

cc
0

0
41
0

it

0
0

*
0
.IX

CU

_0.

+P.

-,J

P.

~'

0~* 0

w
0

(z.~

ON - o

%D

0O-Oo

00-

a*

w 0
0: 9

040

Il
0

a0

11

1Z_

e... 9

UJIV)0ee
0

C0 w

0-iI)00O *I) 0
()!-00
I"-I0
Q

-0000

.4

+
I

CN
0

00
s.JZ

00

a
0 w

0
.~04

+x

00

MNOMM..0V-00

NO

a
.a

SO
V

C-

Z
0
04

Is

Z04

0 ') 0 a $

7*~.
*'

-.

zcrJ

04ZZ

3u)

00

*0 01,_ oi.4# 4-

OZ
00

0-

000

0-1

0000
0
00
~ 0000000
0

0.~
0 ~

I4

CNINow

NN

0000O0QOO000O00
0O

00

0.

0o0

U.1

47
00

0.

-00

--

'.4---

000000000000000000

000000000000000000

beC

eb

eb b

eb

eb

e eb

eb

eb

EZI~zz~zxazigz

r~

0j

be

i3.

CD-

11
p-9W

b"C
..
P.s -Nz1
Do-00

0Z

ol

CUou(
*-ixo02

M W 11

t.

CI

.-

zz
Nmqi*

A P

000000000000000000be

be00

00000OU00000
*

75