v16 n1 1995 Biemermissile
v16 n1 1995 Biemermissile
v16 n1 1995 Biemermissile
Steven M. Biemer
INTRODUCTION
The Applied Physics Laboratory has been the tech~ Air Defense System (lADS). To include such effects,
nical direction agent for the Tomahawk Weapon Sys~ complex computer simulation models are used as an~
tern (TWS) Program since its inception and is respon~ alytical tools. Before the TWS can be examined at the
sible for estimating and analyzing Tomahawk system operational level, the performance of individual lADS
effectiveness and providing those estimates to Navy components, such as radars and Surface~to~Air Missiles
decision makers. The Laboratory thus ensures that the (SAMs), against a single Tomahawk Land Attack
TWS meets current and future operational require~ Missile (TLAM) (Fig. 1) must be studied through
ments and assists the Navy in developing improve~ detailed detectability and engageability analyses using
ments for new variants of the weapon system. engineering~ level models. The analyst must incorpo~
Tomahawk effectiveness estimates must be devel~ rate the results obtained from these studies into an
oped within an operational context to take into ac~ operational scenario by developing an appropriate set
count the synergistic effects associated with multiple of inputs to the selected force~level models. These
missile strikes against targets defended by an Integrated models are employed as part of an operational~level
Engineering-level analysis
Figure 1. Tomahawk Land Attack Missile in fli ght. Scenario ~ Force-on-force modeling
generation W---"'~ APL MBER
JSAM Planner ~ ADSIM
60 UMBER 1 (1 995)
FORCE-LEVEL EFFECTIVENESS MODELING FOR THE T OMAHAWK
JOHNS HOPKINS APL T ECHN ICAL DIG EST, VOLUME 16, NUMBER 1 (1995) 61
S. M. BIEMER
• SSPk templates for each SAM that are a function of 4. Defensive order of battle (OOB) (the lADS)
intercept range, azimuth angle, and Tomahawk alti~ 5. Defensive system locations and tactics (i.e., their
tude. command and control structure)
This information will be used in an operational anal~ Mission objectives are typically obtained from dis~
ysis, which is the other major phase of the force~level cuss ions with the Tomahawk Program Office (PMA~
effectiveness and survivability analysis process. 280) , the office of the Chief of Naval Operations
(OPNAV~N 8 6) , strike planners at the Naval Strike
Warfare Center in Fallon, Nevada (NSWC/Fallon),
OPERATIONAL .. LEVEL ANALYSIS and APL personnel. Targets that, when destroyed, will
AND TOOLS satisfy those objectives are typically identified by PMA~
280, NSWC/Fallon , OPNAY, APL, and selected oper~
Scenario Generation ational commands. If strike aircraft are involved, the
Naval Air Systems C ommand (NAVAIR) will also
Before Tomahawk effectiveness can be examined at
participate in the mission definition and target selec~
the operational level, the missile system must be studied
tion process.
within an operational scenario. Each scenario has five
Unless the defensive OOB and site locations are
components as follows:
already defined, the analyst must develop an appropri~
1. Tomahawk mission objectives ate database. Determining an OOB is accomplished
2. Strike asset descriptions (including T omahawk and through coordination with the applicable Defense ln~
other strike weapons) telligence A gency (DIA) organ izations for the time in
3. Target descriptions question . Additionally, the Joint Chiefs of Staff have
(a) (b)
~5Yft
4Yft
3Yft
2Yft
Yft
4X .-----.-----,_----~-----.--------,__,
I3X
c
.Q
rou
.2 2X
Q)
0>
C
~
C
~ X
TLAM~ o
-2X -X o X 2X 3X
Cross-range location (nmi)
Figure 4. (a) Closest point of approach (CPA) of a SAM fired against a Tomahawk. (b) A contour plot from IMARS displaying the SAM's miss
distance as a function of the down-range and cross-range location of the SAM's intercept point (X and Y denote any positive numbers).
62 JO HNS HOPKINS APL TEC HN ICAL DIG EST , VOLUME 16, N UMBER 1 (1 995 )
FORCE-LEVEL EFFECTIVENESS MODELING FOR THE T O MAHAWK
JO HNS HOPKINS APL TEC HN ICAL DIG EST, VOLU ME 16, NUMBER 1 (1 995) 63
S. M. BIEMER
to hit the target before it will be declared destroyed. Once the Tomahawk weaponeering data and flight
Otherwise, additional missiles must be assigned. trajectories have been produced, the analyst is ready to
Once the missile assignments have been established, start the operational analysis.
the flight routes must be developed. The flight~route
generation process can be thought of as an optimization Force .. on .. Force Simulations
problem, since various constraints limit where a Tom~ All of the necessary data are now available to per~
ahawk route can be placed. These constraints include form a force~on~force analysis. If survivability informa~
the following: tion is known or assumed and the Tomahawk missile
strike is against strategic, nondefended targets, a spread~
• Navigation constraints sheet may be used to determine the probability of
• Capabilities and locations of threats destroying individual targets, from which other infor~
• Time of arrival constraints mation can be calculated. In most cases, however,
• Launcher constraints survivability information in a many~on~many context
• Target approach route feasibility is not known because multiple TLAMs in a strike cause
To determine the constraints posed by the threats, synergistic effects in overcoming an lADS. Multiple
the DTED tool (described earlier) is used. This software missiles can confuse or even saturate single lADS
tool is not an automated route generator; rather, it is components, contributing to an overall degradation of
a display tool with which the analyst can evaluate performance. Only a many~on~many analysis using a
candidate TLAM routes. The tool can display an over~ sophisticated simulation will be able to represent this
head view of the candidate route and the various threat reality.
sectors perturbed by local terrain. In addition, a vertical The Laboratory uses two force~on~force simulations:
profile of the route and the terrain directly below can APL MBER and ADSIM. The former is an APL~
be displayed. The analyst can then modify the altitude enhanced version of the Multiple Battlefield Engage~
of the TLAM route manually to reflect operational ments and Reactions Model (see the boxed insert on
knowledge of Tomahawk terrain~following capabilities. APL MBER). This simulation models a multiple Tom~
The analyst can rely on a naval operational planner ahawk and aircraft strike against targets defended by a
to assist in developing TLAM routes. Ideally, the routes ground~based IADS. 2 The Air Defense Simulation
should be planned by the operational Cruise Missile (ADSIM) is an APL~developed simulation that models
Support Activities (CMSA), located in Norfolk, Vir~ a Tomahawk~only strike against targets defended by an
ginia, and San Diego, California, since these activities airborne lADS.} Output from both simulations can be
plan operational TLAM routes for the Navy. Security integrated to represent the strike against the airborne
constraints, limited knowledge of future weapon system and ground~based portions of the lADS whenever the
characteristics, and tasking priorities, however, typical~ two types of engagements occur in sequence. When
ly preclude using a CMSA to plan routes for Tomahawk they occur concurrently, an iterative approach is used.
effectiveness studies. Other sources for operational Both models are executed and the results incorporated
planning include NSWC/Fallon, the Washington, into the other at selected times to represent the nec~
D.C., CMSA (a training/testing facility), the Naval essary interaction.
Surface Warfare Center in Dahlgren, Virginia, and
various personnel within APL with expertise in Tom~ Outputs and Measures of Effectiveness
ahawk mission planning. Regardless of who plans the Both graphical and tabular output are provided by
routes, they must consist of a series of turn waypoints APL MBER. The graphical output is obtained from a
and altitude action points between the launch point standard output file containing event records used by
and the target. a graphics program to animate the events. With this
Since TLAM mission plans may not be sufficient for tool, the user can step through a single run of the
the operational analysis, a TLAM trajectory may need simulation. Flight paths of each Tomahawk are drawn,
to be developed. This trajectory would include detailed as well as the locations of SAM sites and targets. When
altitude information along the route, navigation update a SAM engages a TLAM, an engagement line is drawn
points, and actual latitude/longitude/time locations. between the SAM and the current position of the
Depending on the level of fidelity required, the trajec~ TLAM. Symbols represent events such as a TLAM kill
tory can be generated in a variety of ways. If high by a SAM (a red circle), a SAM miss (a red X), and
fidelity in the trajectory is not required, the trajectories a target hit by a TLAM (a red box). Finally, the output
can be derived manually from the routes on the basis can be overlaid on the DTED of the area. Figure 8
of missile climb/dive rates. If more fidelity is required, displays sample graphical output overlaid on a DTED
a 6~degree~of~freedom simulation can be used to gen~ map of the area.
erate trajectories, given the information about the The Air Defense Simulation can also produce graph~
planned route. ical and tabular output. The graphical display is similar
66 JOHNS HOPKINS APL TECHN ICAL DIGEST, VOLUME 16, NUMBER 1 (1995)
FORCE-LEVEL EFFECTIVENESS MODELING FOR THE TOMAHAWK
• Number of aircraft sorties required against a target set are compared, and conclusions are drawn. For example,
when TLAMs are present the number of TLAMs participating in a strike against
• Number of aircraft sorties saved by using TLAMs a common target set would represent a case study. Sup~
pose the MOE is the probability that 80% of a selected
In addition, secondary measures may need to be deter~
target set was destroyed. Figure 9 presents a sample case
mined by decomposing a primary measure into its com~
study result for this situation. Notice that the lowest
ponents. For instance, a measure like "number of
quantity is entirely insufficient to destroy 80% of the
TLAMs required" could consist of an equation involv~
target set. Also observe that, at larger quantities, the
ing two other measures:
marginal increase in probability is reduced consider~
ably. The most cost~ effective quantity would therefore
TLAMrequired = Total TLAMlaunched be either the center value or, possibly, the fourth quan~
tity. Adding missiles would be cost~ineffective (or even
- TLAMreliability failures - TLAMattrited' wasteful).
Regardless of the type of analysis, the analyst
In this case, the number of reliability failures and of must understand that the output of any stochastic
operational~ level simulation is statistics, not truth. Sto~
TLAM attrited are two lower~level measures. The first
can be assumed on the basis of TLAM specification chastic simulations produce probability distributions of
values or testing experience; the second is calculated selected MOEs. It is up to the analyst to determine what
by APL MBER. By successively decomposing MOEs, those distributions mean.
the analyst eventually arrives at measures that can be It is tempting for the analyst to calculate mean
calculated mathematically, estimated by simulation, as~ values for each MOE and use those values as the answer,
sumed, or parameterized. At that point, the decompo~ but this approach can be deceptive. Assuming the
sition is sufficient. In some instances, no decomposition output distributions are normally distributed (typically
is needed. they are when a sufficient number of iterations are
run- but not always), confidence intervals can be used
when comparing and displaying the same MOE for
Analysis of the Results different cases. A common statistical test to determine
Once the input databases have been defined, the whether two mean values are equal when the variances
analyst executes the stochastic models and collects are unknown (but equal) is the Student's~t test with
statistics on the selected MOEs. When this baseline nl + nz - 2 degrees of freedom, where n1 and nz are the
analysis is complete, the analyst has the option of sample sizes of the two cases. 5
performing three additional types of analysis that may Figure 10 is a sample plot of Tomahawk survivability
(and often do) provide vital information in understand~ versus the number of Tomahawk missiles employed.
ing the simulation results. 4 The first type of analysis is Although the mean values differ, the 90% confidence
parameterization, which is the process of selecting a set intervals suggest that the values are equivalent for the
of parameters (usually just one) and varying their values three larger quantities of TLAMs. In fact, the equality
to calculate the boundary conditions on a selected conclusion can be substantiated after statistical tests are
MOE. Parameterization is typically performed when applied to these values.
the value of the parameter is unknown or known with
a level of uncertainty. It is quite useful in examining
1.0
"worst~ case" or "best~case" scenarios as well.
The second approach is sensitivity analysis. The pro~
?t-
a 0.8
cess is the same as parameterization, but the purpose is co
OJ
different. In this type of analysis, the analyst calculates c .....
the selected MOE for varying values of a single input e! 0.6
. - 0)
..... 0)
(f)OJ
parameter to determine how sensitive the resulting 0) ...
'O~
MOE value is in relation to the varying parameter. If eO)
ȣ
0.4
..........
the resultant MOE value fluctuates greatly with small Be
C1l
perturbations to the input parameter, the MOE is said .0 0.2
e
a..
to be sensitive to the input parameter; otherwise, it is
insensitive. This type of analysis can assist in determin~ a
ing which inputs account for their observed values. 1X 2X 3X 4X 5X
Number of Tomahawk missiles
The third method, called case studies, is very com~
mon in Tomahawk effectiveness studies and entails Figure .9. Sample case study result showing the probability of
developing and examining several cases with different destrOYing 80% of a sample target set (X denotes any positive
number).
assumptions independently. The resulting MOE values
1.0
ii -
-
Mean value
90% confidence interval (negative)
The final step in the Tomahawk effectiveness anal,
ysis process involves presenting the results in a format
understandable to the nonmathematician and integrat,
ing documentation collected throughout the process to
formulate a report.
0.8
Cii
>
·5
:;
SUMMARY
CIJ 0.6
'0 Estimating Tomahawk effectiveness within an oper,
~ ational scenario requires the integration of results from
Ei
co 0.4
.0
several analyse. Engineering,level simulations exam,
e
Cl..
ining the detectability, engageability, and vulnerability
of the Tomahawk cruise missile against components of
0.2
an lADS are used to provide inputs to mission planning
and force,level models that estimate TLAM effective,
0 ness in the context of an operational scenario. Through
1X 2X 3X 4X 5X
the effective application of the process described in this
Number of Tomahawk missiles
article, APL can ensure that the TWS meets current
Figure 10. Aggregated probability of survival statistics for a Toma- and future operational requirement and provide the
hawk strike using the quantities listed in a sample scenario (X
denotes any positive number) .
Navy with valuable effectiveness information in devel,
oping improvements for new variants of the weapon
system.
REFERENCES
Once the statistical values are understood, conclu, iMirchell, R. L, and O'Donnell, P. A., Technical Radar Analysis and Modeling
sions can be drawn on the basis of the input data, System, Mark Resources, Inc., Santa Monica, CA (1988) .
2Multiple Battlefield Engagements and Reactions Model User's Guide , TR-12-87,
scenario assumptions, and force,on,force modellimita, Karnak Re earch Corp., San Diego, CA (Sep 1987).
tions. This process is more of an art than a science. 3 ADS 1M Program Description and Usage Documentation, JHUjAPL BAS-88-094
(Dec 1988) .
Although observations can easily be made from graph, 4Law, A. M., and Kelton, D. W., Simulation Modeling and Analysis, 2nd Ed.,
ical presentations of the data, the meaning of the data McGraw-Hill, ew York, pp. 522-601 (1991).
SDunn, O. J., and Clark, A. V., Applied Statistics: Analysis of Variance and
in relation to the study objectives may not be readily Regression, 2nd Ed. , John Wiley & Sons, ew York, pp. 37-51 (1987).
THE AUTHOR
68 JOHNS HOPKI S APL TECHN ICAL DIGEST, VOLUME 16, NUMBER 1 (1995)