lUL 6 fi92
Coupling Logistics to
Operations to Meet
the Threat (CLOUT)
I. K. Cohen, John B. Abeli, Thomas F. Lippiatt
Appr^rved for public nimmuf
Tlie iMdflurch reported here wae sponsored by the United States
Air Force under Contract F49620-9X-C-0003. Further information
may be dirtained from the Long Range P lan ni n g and Doctrine
Division, Directorate of Plans, Hq USAF.
Library of Congress Catido^ng in PubUcatkn Data
Cohen, I. K.
Coupling logistics to operations to meet uncertainty and the threat /
I.K. C<*en. John B. Abell, Thomas F. Ui^iatt.
Prepared for the United States Air Force.
Includes biUiogrqthical refeiences.
I. United Stiues. Air Force—Equipment—^Maintenance and rq>air
2. United States. Air Force—Inventory control. 3. United States.
Air Force—Combat sustainaWlity. J. Abell, John B. II. Lippiatt,
III. United States. AirFbrce, IV. Title.
Tha RAND Pbhfintkm Saite fUt JfapQit li Ha
pabtkatkm d o e um a n ti ng aad tmmtOmg
rsMarch findings and final m a aiclt liati^
laiMKlt othar outgata of sgoamd
dMbatioiL P i d»lk s itioB » of lUiai db
Coupling Logistics to
Operations to Meet
the Threat (CLOUT)
I. K. Cohen, John B. Abell, Thomas F. Lippiatt
A Project AIR FORCE Report
Prepared for the United States Air F^^rce
Approved for public release; distribution unlimited
This report provides an overview of a RAND project entitled
“Enhancing the Integration and Responsiveness of the Logistics
Support System to Meet Peacetime and Wartime Uncertainties,”
popularly known as the “Uncertainty Project.” The Uncertainty
Project was concerned primarily with quantifying the magnitude and
pervasiveness of uncertainty in the demands for resources in
peacetime and war, and developing and evaluating initiatives to cope
with it. It pursued the concept that uncertainties could typically best
be met by designing a responsive support system—one that, by taking
management actions, could respond to unanticipated events in “near
At RAND there has been increasing interest in state-of-the-world
uncertainties as distinct from statistical uncertainties. For example,
in RAND’s Arroyo Center, which supports the U.S. Army, a study and
depot demonstration effort has been undertaken covering much of the
material discussed here. Projects dealing with the effect of
uncertainties on munitions requirements and distribution have also
This report should be of general interest to policymakers, policy ana¬
lysts, resource managers, and those persons engaged in the further
development and implementation of the Air Force’s new Logistics
Concept of Operations.
The Uncertainty Project was sponsored by Headquarters, USAF
(AF/LEX and AF/LEY) and Headquarters, AFLC (XP). It was
undertaken within the Research Management Program of RAND’s
Project AIR FORCE, a U.S. Air Force federally funded research and
development center established to undertake policy analysis.
The research described in this report was designed to:
• Understand better the magnitude and extent of variability in
peacetime demands for logistics resources.
• Develop and evaluate initiatives for increasing the flexibility,
robustness, and responsiveness of the logistics system at both
the base/theater and depot levels to increase readiness and sus¬
tainability in the face of the unpredictability of peacetime and
wartime resource demands.
• Define the need for a means by which the infrastructure should
be tasked to support the “real-time” needs of the combat forces
TAKING EXPLICIT ACCOUNT OF UNCERTAINTY
Uncertainty is of essentially two kinds: (a) statistical uncertainty,
defined as variability observed in repeatable phenomena, and (b)
state-of-the-world uncertainty, defined as uncertainty about phenom¬
ena that are not repeatable, not observed or observable, or both.^
Planning for wartime is fraught with uncertainty of the latter kind; in
fact, state-of-the-world uncertainty can fairly be said to dominate the
wartime scenario. It affects peacetime planning as well. State-of-the-
world uncertainty is fundamentally different from the statistical type;
the variability involved is literally impossible to predict. Yet in prob¬
lems involving state-of-the-world uncertainty, analysts have tradi¬
tionally retreated to analytic methods intended to deal with statisti¬
cal uncertainty, the methods with which they are competent and
comfortable. As a result, planning efforts may ignore a wide range of
Analysis of the demand for avionics spare parts demonstrates that
this is true for Air Force logistics planning. In both planning and ex¬
ecution, support systems implicitly assume a level of predictability in
resource demands that does not often exist in peacetime and almost
certainly will not exist in war.^ In peacetime, the actual range of
demands experienced for certain spares is several times as large as
^Hodges and Pyles, pp. 8-14.
^See Crawford, Gordon B., Variability in the Demands for Aircraft Spare Parts: Its
Magnitude and Implications, RAND, R-3318-AF, January 1988.
the range our planning procedures assume—so large that buying suf¬
ficient spares to meet the potential demand is prohibitively expensive.
Such uncertainty about the future is often ignored in our planning for
wartime as well. The very use of a single wartime planning scenario
implicitly denies that we have great uncertainty about such impor¬
tant factors as the level of activity; the impact of these activities on
the demands for resources; force beddown; mission mix; damage to
spares, repair, and personnel resources in base attacks; the system
disruptions that seem almost inevitable in wartime; and all of the
other surprises the enemy is likely to induce.® During execution,
however, circumstances in wartime may shift from state-of-the-world
uncertainty to statistical uncertainty. In such a case, it may be pos¬
sible to forecast demands over short periods of time.
RESPONDING TO UNCERTAINTY
This situation suggests a general class of support strategies in plan¬
ning both for peacetime and, particularly, for war—strategies that ac¬
count for uncertainty. One central consideration is response time.
Response time can mean the sum of any of repair times, processing
and handling times, order-and-ship times, transportation times, or
even communication times in particular applications. The longer the
response time, the more decisionmaking depends on forecasting fu¬
ture events. At the present time, the anticipated scenarios we need to
use in planning are not likely to eventuate as predicted; thus, the
more quickly the system can respond to urgent, unanticipated needs,
the less vulnerable it will be to uncertainty and the more robust its
performance will be in the face of it.
Even when an advanced priority system is used in peacetime depot-
level repair, long response times (from 20 to 50 days, where the
longest times are associated with bases with wartime deployment
tasking and war reserve spares kits) force the system to rely to an
undesirable extent on forecasts, which in turn are vulnerable to un¬
certainty. Thus it is desirable to reduce response times to a few days,
a dramatic reduction to be sure, and difficult to achieve, but desirable
given the problems in predicting events over long planning horizons.
In fact, a very responsive system may be able to operate satisfactorily
in a reactive mode so that the need for being proactive is reduced.
®See StockHsch, J. A., Linking Logistics and Operations: A Case Study of World
War U Air Power, RAND, N-3200-AF, 1991.
There are some logistics planning problems whose implicitly long
planning horizons may not be tractable to change. The problem of es¬
timating the best mix of spares procurements is one such example.
Because of the relatively long planning horizon, alternative strategies
for hedging against incorrect forecasts are needed.'*
Finally, the magnitude and pervasiveness of uncertainty also suggest
the need in both capability assessment models® and spares require¬
ments estimation® to take explicit account of more realistic levels of
peacetime and wartime uncertainty and also of the support system’s
ability to cope with some of this uncertainty through widely used
management adaptations. This need exists in every stage of the
planning, programming, budgeting, and execution system.
THE CLOUT INITIATIVES
The general approach to the uncertainty problem and the set of spe¬
cific initiatives that have emerged from this work are called CLOUT
(Coupling Logistics to Operations to meet Uncertainty and the
Threat). The CLOUT initiatives are examples from a taxonomy of
more generic strategies for coping with uncertainty.'^ They are in¬
tended to enhance the flexibility and responsiveness of the logistics
system in the theater, at the Air Logistics Centers, and in command
and control systems, thus enhancing the robustness of system per¬
formance in the face of uncertainty. Although this report describes
the CLOUT initiatives with emphasis on their application to the
Tactical Air Forces, our expectation is that this initial set of initia¬
tives will be augmented by other initiatives that extensions to this
work will identify. In fact, some of these initiatives and some exten¬
sions to them are reflected in the new Logistics Concept of Operations
and the Logistics Strategic Plan.® Moreover, a study is currently
being formulated at RAND that is intended to develop and demon-
*See the discussion in Lippiatt, Thomas F., Variability in the Budget Forecasts for
Depot-Level Component Repair, RAND, N-2930-P&L, 1991.
®A capability assessment model was developed to support the analyses reported
here. It is described in Isaacson, Karen E., and Patricia Boren, Dyna-METRIC Version
5: A Capability Assessment Model Including Constrained Repair and Management
Adaptations, RAND, R-3612-AF, August 1988.
®Another RAND project has been established as an outgrowth to this work under
the auspices of the Resource Management and System Acquisition Program of Project
AIR FORCE to develop improved spares and repair estimation methods that take ac¬
count of realistic levels of uncertainty and the effects of management adaptations.
'^See the discussion in Hodges and Pyles, pp. 20-24.
®l/S Air Force Logistics Strategic Plan, Department of the Air Force, DCS/Logistics
and Engineering (AF/LEYX), Revised October 1990.
strate a multiechelon spares and repair and financial management
system for the Air Force that will explore additional means for im¬
plementing these initiatives.
Essentially, the CLOUT initiatives generally place less reliance on a
richness of spares and take greater advantage of more flexible re¬
sources, such as maintenance and transportation. That strategy de¬
rives logically from the difficulty and cost of a “buyout” strategy that
would attempt to provide ample quantities of spares, for example, to
allow the system to cope with the levels of uncertainty in demand
that it might face. Though the difficulty of a buyout strategy for
wartime may be self-evident, such a strategy is also troublesome for
peacetime. Even in peacetime, buyouts are not likely to be economi¬
cal, and they can be especially problematic in a constrained funding
environment. At the theater level, there are significant payoffs to be
gained from alternative operating policies for theater distribution
systems that take fuller advantage of responsive lateral resupply and
lateral repair options. Capability assessments also suggest that
closer coupling of the depot repair system to the combat forces has
significant payoff in aircraft availability. The thrust of the thinking
underlying CLOUT is to rely less on an ampleness of goods and more
on management adaptations.® That thinking has important impli¬
cations for system design as well as management. Many characteris¬
tics of the current system need to be changed to achieve the kind of
relevant, timely, and robust performance needed to cope with unan¬
ticipated, urgent demands for resources.
The theater-level CLOUT initiatives are as follows:
• Lateral repair by bases with repair capability to support bases
that would otherwise have to send their repairable assets back
to the depot for repair.
• Forward stockage, with emphasis on intermediate-level test
equipment parts, shop replaceable units (SRUs), and lower-de¬
mand parts that do not qualify as War Reserve Materiel (WRM).
• Responsive theater transportation to support lateral resupply,
forward stockage, lateral and theater (rearward) repair, retro¬
grade, and distribution of assets coming into the theater.
• Improved operating policies and decision rules in prioritizing the
repair of assets and allocating them to bases with the objective
®Hodge8 and Pyles, p. 23. Hodges describes these two approaches as examples of
passive and active strategies, respectively, to cope with uncertainty.
of achieving dynamic aircraft availability goals at each of the
Clearly, several of the CLOUT initiatives are already used in some
form by logistics managers in the current system to overcome urgent,
unanticipated demands for resources. Cannibalization and lateral
supply are examples of adaptive management that can help overcome
asset shortages. Logistics managers often resort to these and other
adaptations. The distinction between such adaptive behaviors and
the CLOUT initiatives is that the CLOUT initiatives are meant to be
systemic, implemented with full visibility of the theaterwide or
worldwide asset position, and oriented explicitly toward the achieve¬
ment of specified aircraft availability goals.Thus they cannot al¬
ways be implemented by local managers. In addition to extending the
manager’s ability to adapt, CLOUT is intended to facilitate and im¬
prove the adaptations that are already available.
There is a need, of course, to consider the vulnerability of transporta¬
tion and communications systems and assets in wartime. A system
built on the assumption of worldwide visibility of the asset position
and available transportation must be carefully thought through in
terms of its viability in combat. It should be noted, too, that systemic
implementation of the kinds of initiatives described here would re¬
quire policy changes that would make them operationally accepted.
For example, when a base receives a redistribution order directing the
reallocation of an asset, it would need to respond promptly with the
shipment (unless, of course, there was reason to believe that an error
had occurred). Similarly, unit commanders would need to subscribe
to the concept of mutual base support in a practical sense. It doesn’t
seem that such an idea is unreasonable. Commanders might have
good reason to believe that they might gain as much from such mu¬
tual support as they would give up to it.
The “wholesale” and depot-level CLOUT initiatives include the
• Responsive, assured intertheater transportation.
Air Force reporting system currently in use provides visibility of the world¬
wide asset position of recoverable items. Called AFRAMS (Air Force Recoverable Asset
Management System), it provides data to an AFLC system used by item managers and
others for various item management functions.
^^We use the term "assured” to mean that transportation is always available. It is
more of an assumption than a fact because it assumes that higher-priority require¬
ments do not interfere with the transportation of aircraft spare parts. We do not dis¬
cuss here the management systems or other resources required to "assvue” transporta¬
• Enhanced flexibility and responsiveness of the depot’s repair
process to increase the timeliness of repair actions and their rel¬
evance to the current needs of the combat force.
• Distribution of serviceable assets that explicitly accounts for
mission urgency and the current asset position worldwide.
THE POTENTIAL PAYOFF
Assessments of the CLOUT initiatives suggest the magnitude of the
effects of particular uncertainties on the performance of the logistics
system and the payoffs that might be achieved through a more rele¬
vant, timely, and robust support system. They show that responsive
lateral supply and lateral repair improve system performance sub¬
stantially, and that theater-level priority repair helps the system cope
with demand uncertainties even when individual repair facilities are
saturated with workload. Responsive depot repair also pays off. It
helps mitigate the effects of additional uncertainty in demand for
NrtS' 2 assets and provides crucial test equipment repair parts to the
theater. The CLOUT initiatives also support alternative force
beddowns better than the current system and help reduce the disrup¬
tive effects of base attacks.
Although particular aspects of logistics system management are em¬
phasized in this report, its underlying thinking applies to other areas
of management. For example, responsive depot support depends
heavily on assured, responsive transportation and timely handling
and processing of assets in transit from the depot to the base and in
retrograde shipment. We do not address many of the important is¬
sues involved in making these parts of the logistics system more re¬
sponsive; yet, they are only a few of the many aspects of logistics
management that need to be modified in terms of goals, incentive
structures, operating policies, management systems, etc., before the
total system is genuinely responsive.
PLANNING FOR UNCERTAINTY IN A CHANGING WORLD
Between the time this research was done and the time of this publica¬
tion, the world has changed dramatically. The scenario that provided
the context for this work was a high-intensity conventional conflict
between NATO forces and those of the Warsaw Pact. This scenario
was based on prescribed national poliqr. The Uncertainty Project ob-
^^Not repairable this station. Defective assets beyond the repair capability of a
base, typically returned to an Air Logistics Center for repair.
served that there was likely to he considerable variation assoc'ated
with many aspects of this scenario—variation not taken into account
in forecasting the demand for support. The new environment faced
by military planners is quite different in some ways, but quite similar
in others. The high-intensity European conflict provided a prescribed
scenario for military planners. Many felt that if resources were ade¬
quate for the European scenario, they would be adequate for other
scenarios which could be expected to absorb a smaller proportion of
the force. Given the changes in the world in recent months, the
European scenario no longer has the same significance in planning
and resource allocation decisionmaking. The Air Force now faces a
planning problem in which there is great uncertainty about the range
of scenarios it is likely to face and the variation that is likely to exist
within each scenario.
This work points out the important fact that, in any scenario, units
may face shortages that need to be overcome in very short response
times. That is likely to be the case no matter what specific combat
scenario evolves. Thus the need for flexible, adaptive management
and mechanisms for supporting it persists for military planners. In
general, management adaptations like the CLOUT initiatives help
the logistics system achieve needed levels of responsiveness, enhanc¬
ing its robustness not only for a European scenario, but in the face of
uncertainty about what scenario will evolve, what mix of weapons
will be needed, when, where, etc. Specific initiatives, however, may
not have the magnitude of benefit in particular scenarios that they
might have had in a high-intensity NATO scenario. Intratheater lat¬
eral supply, for example, depends on having more than one base with
a specific weapon system within reasonable proximity in a theater.
Management adaptations yet to be defined and evaluated may turn
out to be important in any particular future scenario. While .such
initiatives obviously need to be understood in terms of their applica¬
tions to specific situations, their implications for combat support are
still important, despite the fact that the traditional threat seems to
have changed. They may become even more important, given our
greater uncertainty about the kinds of future threats we will face. At
the time of this writing. Operation Desert Shield/Storm has ended,
but final evaluations have not been completed. However, that experi¬
ence has tended to confirm the principal assumptions and outcomes of
this study. Prediction is indeed hazardous; adaptations are extremely
important; and quick response is what the forces in the field expect.
In any event, this report discusses the NATO scenario and not re¬
gional contingencies. We trust that this emphasis will not interfere
with the communication of the basic CLOUT views.
The implications of this approach for logistics management, logistics
policy analysis, and the design of logistics management systems are
fundamentally important. The implications of uncei tainty in particu¬
lar applications are sometimes both difficult and unfamiliar for many
persons. To illustrate some of the issues involved and to suggest that
there are feasible approaches for dealing with them, in Sec. 5 we dis¬
cuss two examples of applications, one involving the prioritization of
component repair and the other the estimation of spares and repair
Perhaps the most important central message of this work is the need
to take more exjilicit account of uncertainty, particularly state-of-the-
world uncertainty, in formulating policies and designing systems, and
to take explic t steps to ensure that the performance of those policies
and systems is robust in the face of those uncertainties. This doesn’t
mean that the uncertainties actually faced will indeed be matched by
the existing robustness of the system. It does suggest, however, that
it is more realistic to emphasize ^bustness than to behave unrealisti¬
cally, as if demands were known, designing systems that are opti¬
mized to meet those questionable demands.
The CLOUT initiatives are examples of management adaptations
that enhance the performance of the logistics system in peacetime
and wartime. As we have shown, they help mitigate the effects of un¬
certainties. To the extent that we ignore statistical and, especially,
state-of-the-world uncertainties in logistics planning, particularly for
wartime, we are vulnerable to events unfolding in ways that defeat
specific solutions. To the extent that we take explicit and realistic ac¬
count in planning of our uncertainties and the effects of management
adaptations in overcoming them, we will be better able to develop so¬
lutions whose performance is robust in the face of uncertain futures.
This is the fundamental message of this work. It is a message that
applies to broad categories of management decisionmaking and policy
analysis. It is an important message for Air Force logisticians and for
those involved in logistics management system design.
It is difficult to imagine how this work could have been undertaken
without the special relationship that exists between the Air Force and
RAND under Project AIR FORCE. For that relationship we are
grateful. Within that context many individuals in the Air Force and
at RAND deserve credit for the work discussed in this report. Their
names and contributions would fill several pages. If we were to single
out one senior person in the Air Force and another at RAND who
participated in and supported this work in extraordinary ways, we
would mention Major General Edward R. Bracken, USAF, and
Michael D. Rich.
PREFACE . iii
ACPCNOWLEDGMENTS . xiii
FIGURES AND TABLE. xvii
1. INTRODUCTION. 1
Project Results and Report Organization. 2
2. UNCERTAINTY IN RESOURCE DEMANDS. 4
Variation in Spare Parts Demand. 4
Assumptions of Predictability. 7
The Current System. 11
3. AN ALTERNATIVE APPROACH. 15
Focus of the Analysis. 15
Key Principles . 16
Required Infrastructure. 19
The CLOUT Initiatives. 20
4. AN ASSESSMENT OF CLOUT. 27
Capability Assessment Tools. 27
Details of the CLOUT Initiatives. 28
Scenario and Scope of Analysis. 29
CLOUT Payoffs with Wartime Demand
Uncertainty and No Damage. 32
CLOUT Support in the Face of Battle Damage. 35
CLOUT Support of Alternative Basing Options .... 39
A Summary of CLOUT Assessments. 41
5. SOME EXTENSIONS OF THE CLOUT LOGIC TO
OTHER APPLICATIONS. 43
Example 1: Prioritizing Depot Repair and
Allocating Assets to Bases .. 43
Example 2: A Policy Study of Spares and Repair
2.1. Variability in Demand Rates for FlOO Engine Unified
Fuel Control. 5
2.2. Variability in Demand Rates for F-15 Converter
2.3. How Variability Affects a Squadron’s Wartime
2.4. Why Buying Out Is Not the Answer. 9
2.5. The Current System. 11
2.6. The Current System with Uncertainties. 13
3.1. Scope of the Analysis. 16
3.2. CLOUT: Enhancing Theater Responsiveness. 21
3.3. CLOUT: Enhancing Depot Responsiveness. 22
3.4. CLOUT: Command and Control System. 24
3.5. CLOUT: The Larger Context. 26
4.1. Effects of Uncertainty on F-16 Aircraft Availability .... 33
4.2. Responsive Support Pays Off in a “Benign”
Environment . 34
4.3. Range of Airbase Attack Damage. 36
4.4. Outcomes of Two Particular Replications. 37
4.5. CLOUT Payoffs with Base Damage. 38
4.6. Payoffs of CLOUT in Dispersed Operation. 41
5.1. Graphic Portrayal of Study Design. 49
4.1. F-15 and F-16 Depot Repair Airlift Requirements. 35
Aircraft Availability Model. A computational al¬
gorithm developed by the Logistics Management
Institute to allocate stock levels of recoverable
assets in a way that approximately maximizes
aircraft availability subject to a budget con¬
straint. It is used by AFLC to compute procure¬
ment requirements for recoverable aircraft
Air Force Logistics Command.
Air Force Recoverable Asset Management
System. A system of asset reporting that pro¬
vides central visibility of recoverable assets
Avionics intermediate shop. The maintenance
shop that repairs avionics LRUs. Shops with
this name are located at bases as well as the
Bench check serviceable. The result of fault di¬
agnosis that fails to confirm a defect.
Base level self-sufficiency stock. Spares that are
authorized for units that are expected to fight in
place. In the context of this research, BLSS is
authorized for main operating bases.
Command, control, and communications.
Collocated operating base. Allied bases that are
expected to host the deployment of USAF units
Continental United States.
Civil Reserve Airlift Fleet. The commercial air¬
craft that are planned to be used to enhance our
military airlift capacity in wartime.
Central secondary item stratification.
AFLC’s central system for allocating stock levels
for recoverable spares to the depot and bases.
AFLC’s system for computing requirements for
recoverable spares and depot-level repair.
AFLC’s system for collection, processing, and
analysis of worldwide maintenance data.
AFLC’s system that provides central visibility of
the worldwide asset position for recoverable as¬
sets. It operates with data from the AFRAMS.
Distribution and Repair In Variable Environ¬
ments. A prototype algorithm for prioritizing the
repair of recoverable assets at the depot and al¬
locating the serviceable assets that emerge from
repair to locations to maximize the probability of
achieving specified aircraft availability goals.
Dyna-METRIC Dynamic Multi-Echelon Technique for Recov¬
erable Item Control. RAND has developed a se¬
ries of capability assessment models to support
policy analytic studies of the logistics system.
Dyna-METRIC Version 4, an analytic model, is
incorporated in AFLC’s Weapon System Manage¬
ment Information System (WSMIS). Version 5, a
simulation model, was used in this research.
Version 6, an advanced, hybrid analytic-
simulation model, the latest version of the Dyna-
METRIC series, extends Version 5 to incorporate
the indenture relationships among LRUs and
SRUs, and adds more explicit representation of
Dyna-SCORE Dynamic Simulation of Constrained REpair. A
discrete-event, Monte Carlo simulation model of
repair shops similar in repair process to the
avionics integrated shop. Dyna-SCORE was de¬
veloped to explore the payoffs of certain man¬
agement adaptations in repair activities.
Monte Carlo trial
European Distribution System. The intratheater
airlift system of the United States Air Forces,
The posture of the combat force in terms of num¬
bers of aircraft of each t5T3e at each location. The
force beddown could also be specified by aircraft
Intermediate level. The term used to describe
the maintenance activities that repair assets for
return to aircraft or to base stocks; the level of
maintenance between the organizational and de¬
Line replaceable unit. Components that are re¬
moved from aircraft when a discrepancy is sus¬
pected. In the indentured relationships among
component parts of an aircraft, for example, they
are typically thought of as component parts of
Mission capability. A term used to describe a
condition such that an aircraft is not mission ca¬
pable for lack of a component part. The requisi¬
tion in the supply system for that component
part is called a MICAP requisition.
Main operating base. A base that has the infra¬
structure associated with peacetime support of
an Air Force unit, typically a wing, that is
expected to fight in place in the event of a war.
Replication of an experiment to estimate experi¬
mental error in which outcomes are determined
purely by chance.
North Atlantic Treaty Organization.
Not fully mission capable. The status of an air¬
craft that is flyable, but whose capability to per¬
form its assigned mission is in some sense de¬
graded, constrained, or inhibited.
Not repairable this station. The status of a
recoverable asset that cannot be repaired at in¬
termediate level and must be returned to the de¬
pot for repair.
PDS Pacific Distribution System. The proposed in¬
tratheater airlift system of the Pacific Air Forces.
POS Primary operating stock (formerly peacetime op¬
erating stock). Spare parts authorized to bases
to support peacetime operations but which may
also be used in wartime.
PPBES Planning, programming, budgeting, and execu¬
RR WRSK Remove and replace war reserve spares kit. A
WRSK computed using the assumption that no
intermediate repair capability exists.
RRR WRSK Remove, repair, and replace war reserve spares
kit. A WRSK computed using the assumption
that intermediate repair capability will be avail¬
able in wartime.
SRU Shop replaceable unit. A subcomponent of an
LRU which is typically removed and replaced
during intermediate-level repair.
UK United Kingdom.
VTMR Variance-to-mean ratio. The unbiased estimator
of the variance divided by the mean of a process.
WMP War Mobilization Plan.
WRSK War reserve spares kit. A set of spare parts that
is authorized to a unit to help support its combat
operations during the early days of wartime.
Air Force logistics planning for wartime is based on several trouble¬
some assumptions. The assumption that underlies the Air Force’s
computations of war reserve spares requirements implies a level of
variability in demands substantially less than the level actually ob¬
served in peacetime. Moreover, the uncertainty in peacetime resource
demands will be compounded in wartime by system disruptions, re¬
source losses, and the inevitable surprises of combat. Another trou¬
blesome assumption is that no spares need to be procured to provide
for the pipeline of components removed from aircraft in the belief that
they are defective but subsequently diagnosed as serviceable. Such
assumptions, coupled with the levels of uncertainty that pervade the
wartime environment, tend to induce an unwarranted level of opti¬
mism in logistics planning.
This research was undertaken to understand the implications of these
uncertainties better and to identify and evaluate initiatives to over¬
come them. This “Uncertainty Project” addressed these issues, espe¬
cially as they apply to the Tactical Air Forces. Its objectives were to:
• Improve the Air Force’s understanding of the magnitude and
extent of the variability in peacetime demands for logistics re¬
• Develop and evaluate alternative methods for increasing the
flexibility, robustness, and responsiveness of the logistics system
at both the base/theater and depot levels to increase readiness
and sustainability in the face of the uncertainty in peacetime
demand that would undoubtedly increase in wartime.
• Define and evaluate the logistics command, control, and com¬
munications (log C^) systems and other mechanisms by which
the infrastructure should be tasked to support the “real-time”
needs of the combat forces.
The set of initiatives that has emerged from this work is called
CLOUT (Coupling Logistics to Operations to meet Uncertainty and
the Threat). The CLOUT initiatives are designed to enhance the flex¬
ibility, robustness, and responsiveness of the logistics system in the
theater, at the Air Logistics Centers, and in command and control
The project consisted of six major tasks:
1. Assess the magnitude and pervasiveness of variability in de¬
mand for logistics resources and its impact on logistics readiness
and sustainability measures.
2. Identify and evaluate responsive base/theater alternatives, lo¬
gistics support structures and strategies, and improved in¬
3. Identify and evaluate responsive depot alternatives. These al¬
ternatives include flexible, adaptive repair and the development
of scheduling algorithms to ensure that spare parts and other
resources are repaired and distributed to meet the immediate
needs of combat forces.
4. Define and evaluate logistics command and control systems at
Air Force Logistics Command (AFLC) and in the theater to as¬
sess the current needs of the forces, and translate those needs
into action by providing directions to various repair and distri¬
bution elements of the logistics infrastructure.
5. Develop models for use in evaluating the repair capacity of se¬
lected base-level or depot-level shops.
6. Develop an approach to estimating spares and repair require¬
ments in light of demand uncertainties and management adap¬
PROJECT RESULTS AND REPORT ORGANIZATION
The Air Force has embraced the ideas discussed in this report and
has included them in a new Logistics Concept of Operations, the im¬
plementation of which is now being thought through by the Air Staff
and the Major Commands. It emphasizes the “fog and friction” of
war, and it appeals to the use of CLOUT-like management adapta¬
tions to help the logistics system cope with the uncertainties of com¬
bat scenarios. The thinking underlying the ideas discussed in this
report has implications for the other Military Departments as well.
The results of the first task were published and are available in a
previous RAND report; they showed unanticipated levels of variabil¬
ity in the peacetime demands for aircraft spare parts. ^ Some of those
findings and their implications are discussed in Sec. 2 of this report.
^Crawford, Gordon B., Variability in the Demand* for Aircraft Spare Parts; Its
Magnitude a/id Implications, RAND, R-3318-AF, January 1988.
Section 3 discusses an alternative approach to logistics system design,
intended to make its performance more robust in the face of uncer¬
tainties. The results of the evaluations of some of the CLOUT initia¬
tives are discussed in Sec. 4. The capability assessment model devel¬
oped and used for these evaluations is also described in a previous
RAND report.* In Sec. 5, we describe very briefly the problem of pri¬
oritizing depot-level component repair and the research currently un¬
der way in estimating spares and repair requirements as practical
examples of the kind of thinking that underlies the initiatives dis¬
cussed in this report. We offer some concluding remarks and recom¬
mendations in Sec. 6.
Our work in this project in developing and demonstrating a mecha¬
nism to prioritize depot component repair to make the depot more re¬
sponsive to the current needs of the combat force is described in two
companion reports.® RAND’s research in logistics command and con¬
trol systems was undertaken in a separate project, and was also de¬
scribed in a separate report.* The model used in this work to explore
the policy alternatives affecting depot maintenance shops, called
Dyna-SCORE (Dynamic Simulation of Constrained Repair), was de¬
scribed in an earlier report.® Finally, the research to incorporate ex¬
plicit consideration of uncertainty and management adaptations in
spares and repair requirements estimation is still ongoing in a sepa¬
rate, follow-on project, and is planned for publication in 1992.
*Isaacson, Karen E., and Patricia Boren, Dyna-METRIC Version 5; A Capability
Assessment Model Including Constrained Repair and Management Adaptations, RAND,
R-3612-AF, August 1988.
®Abell, John B., et al., DRIVE (Distribution and Repair In Variable Environments):
Enhancing the Responsiveness of Depot Repair, RAND, R-3888-AF (forthcoming and
Miller, Louis W., and John B. Abell, DRIVE (Distribution and Repair In Variable
Environments); Design and Operation of the Ogden Prototype, RAND, R'4158-AF
*Gustafson, H. Wayne, Combat Support Command, Control, and Communications
(CSC^): Robust Methods to Mitigate Communications Disruptions, RAND, R-3942-AF,
®Tsai, Christopher L., Dyna-SCORE: Dynamic Simulation of Constrained Repair,
RAND, R-3637-AF, July 1989.
2. UNCERTAINTY IN RESOURCE DEMANDS
In peacetime as well as war, demands for support are impossible to
predict. Even if we accept the flying programs in formal planning sce¬
narios and assume no deviation, we soon discover that we cannot reli¬
ably anticipate demand in the real world. The problems in predicting
spare part demands in peacetime have been observed over a long pe¬
riod. An early major report on this unpredictability was written at
RAND in 1957 by Brown. ^ Over the years, similar studies have been
published. 2 More recently, Crawford quantified the magnitude and
pervasiveness of variability, and thus unpredictability, in the de¬
mands for aircraft recoverable spare parts, and pointed out that such
variability extends to the numbers of assets in resupply pipelines.®
Crawford’s work was the latest in a long history of RAND research
into the problem of forecasting the demand for aircraft spare parts
that began in the 1950s and is still being carried on. The lesson
emerging from the current work is that although improvements in
forecasting may be achievable, parts demand processes have such
large inherent variability that they tend to swamp out such improve¬
ments; therefore, a logistics system with enhanced flexibility and re¬
sponsiveness seems to be the best approach to coping with uncertain¬
ties in resource demands. Although this work focuses on avionics line
replaceable units (LRUs), it is safe to assume that the same logic ap¬
plies to other resources.
VARIATION IN SPARE PARTS DEMAND
Figure 2.1 illustrates the kind of variability that we sometimes ob¬
serve in peacetime spare parts demands. For each of three bases, the
graph shows demand rates^ over several years for the FIDO engine
unified fuel control. The striking message is the high variation of
demand from quarter to quarter and from base to base. This varia¬
tion is so great that, no matter where one imagines himself in time, it
^Brown, Bernice B., Characteristics of Demand for Aircraft Space Parts, RAND, R-
292, July 1966.
®Crawford, op. cit.
^Removala for apparent cause per 1000 flying hours.
Meas.iTe of variability
VTMR = 3.9
1980 1981 1982 1983
Year and quarter
Fig. 2.1—Variability in Demand Rates for FlOO Engine
Unified Fuel Control
is very difiicult to predict the next quarter’s demand. This variation
can be expressed as a single number: the variance-to-mean ratio
(VTMR).® For this example, the VTMR is nearly four.
Figure 2.2 tells a similar story for another component: the F-15 con¬
verter programmer. Again we see variation both within a base and
across bases, but in this case the variability is even larger; the VTMR
®The variance-to-mean ratio is computed using the unbiased estimator of the
variance divided by the mean demand rate. It has the form [(n/(n-l)][£(X^) •
E^(x)]/E(X), where X represents the observed demands and n denotes the number of
individual obsei-vations. The period of observation as well as the partitioning of the
observed data into intervals affects the numerical value of the estimator. In this
example, we used quarterly observation^ as used AFLC in its estimations of demand
variability for purposes of computing requirements, not necessarily the ‘^est”
partitioning in any statistical estimating sense, but one with which logisticians are
moat familiar. Regardless of the data partitioning, the estimates shown here seem
large when contrasted with those that would result from a Poisson demand process
which is assumed in the Air Force's computations of its war reserve spares
1980 1981 1982
Measure of variability
VTMR = 9.0
Year and quarter
Fig. 2.2—Variability in Demand Rates for F-15
Faced with these kinds of observations, one may conclude that this
variability is caused by some special phenomenon. In an attempt to
formulate some reasonable hypotheses about the factors causing
unanticip ited and variable demand for the same component, we in-
t jrviewed a number of maintenance personnel who have experienced
these changing demand levels in the field. We found as many differ-
snt responses as there were respondents; explanations included vari¬
ations in: climate, scheduled maintenance practices, operational use
>f aircraft, component wear and tear characteristics, crews’ malfunc-
ion reporting, new and modified components used, maintenance
kills, and types of mission flown. Even if these explanations are cor¬
set, they are oftentimes unknowable in advance; moreover, they are
3 varied that no suitable hypothesis can be drawn. We concluded
lat whatever is causing the variation cannot be predicted.
his large peacetime demand uncertainty is unlikely to diminish dur-
g war. Even in a benign wartime environment—i.e., one where
dng hours increase but no losses occur—it seems reasonable to be-
ve that demand rates will differ from those seen in peacetime. In
Bcetime, one might even argue that, over time, inroads in pre-
dictability will occur, at least for those observations made for re¬
peated conditions. In the wartime case, however, such inroads are
less likely for many reasons. The sortie conditions studied and un¬
derstood in peacetime may not occur in wartime, for example, and the
actual conditions that might be faced in wartime may bring some
unanticipated outcomes. In a threatening and demanding wartime
environment, aircrew and maintenance responses to equipment
degradation are likely to be different from their responses in peace¬
time. Moreover, the specific actions that might be taken by the en¬
emy, and thus the resulting demands for resources, are likely to be
unknowable. To take just one example, Rich et al. have pointed out
that U.S. Air Forces have almost always enjoyed air superiority over
their own bases and facilities.® That superiority, although enjoyed in
the recent Gulf War, is no longer assured. Contingencies could evolve
in which demand could be dramatically intensified by enemy attacks
against airbases and other parts of the support infrastructure that
disrupt the combat support system and destroy critical resources.
The Air Force continues to pursue initiatives that are expected to
inhibit and reduce the impact of enemy attacks against its bases.
Mitigating the effects of much of the damage suffered in such attacks
might appropriately be managed by logistics. Thus, in peacetime and
benign wartime environments—to say nothing of hostile wartime
conditions—demands can be expected to arise in unanticipated ways.
ASSUMPTIONS OF PREDICTABILITY
But even though demand is unpredictable, formal support systems
tend to act as i/that weren’t true. Predictability and stability are the
premises on which these systems were developed. In each stage of
the Planning, Programming, Budgeting, and Execution System
(PPBES), formal resourcing mechanisms typically do not make realis¬
tic assumptions about uncertainty. The decisionmaking processes
underlying the allocation of logistics resources implicitly assume sta¬
ble, predictable, benign environments.
These assumptions are especially clear, and especially dangerous, in
planning for war. For example, the models used by the Air Force for
defining a War Reserve Spares Kit (WRSK) assume a VTMR of one.'^
®Rich, Michael, William Stanley, and Susan Anderson, Improving U.S. Air Force
Readiness and Sustainability, RAND, R-3113/1-AF, April 1984.
^The use of a VTMR of 1.0 is associated witn the assumption that demand follows a
Poisson distribution. The Poisson assumption underlies the Air Force’s computation of
its WRSK requirements. Other probability distributions describing parts demands are
Yet we have just observed actual VTMRs for certain components of
four and nine—in peacetime. What does this variation imply for a
squadron’s performance? Figure 2.3 shows how one measure, the
percent of non-fully-mission-capable aircraft (NFMC) during the first
30 days of a postulated NATO wartime scenario, is affected by several
different VTMRs when no special action is taken to mitigate the effect
of variation. We used a RAND model called Dyna-METRIC to do this
A VTMR of 1.0 results in about 25 percent non-fully-mission-capable
aircraft after 30 days. (This outcome emulates the one assumed in
defining the WRSK.) If the actual VTMR is 2.0, that percentage rises
to about 40, and so on for higher VTMRs. It is not likely that every
component in the WRSK will have a VTMR of 2.0 or more, but neither
is it necessary for this to happen before we need to worry about seri¬
ously degraded squadron performance. Even if only the critical
Day of the war
Fig. 2.3—How Variability Affects a Squadron’s Wartime Capability
also used by the Air Force in spares requirements computations, but they, too, are
members of the Poisson family because they increase the tractability of the problem.
®Isaacson, Karen E., et al., Dyna-METRIC Version 4: Modeling Worldwide Logistics
Support of Aircraft Components, RAND, R-3389-AF, May 1988.
(driving) components have a VTMR larger than one, capability might
be degraded. And, clearly, if the real-world VTMR is four, then the
WRSK will provide a very different performance from what we had
anticipated, even in a benign environment. Of course, special actions
by the combat unit could moderate this variability, if they are avail¬
able and if they are used. Although such management adaptations
are common, formal resourcing mechanisms typically ignore them,
just as they often ignore uncertainty.
Clearly, the range of variation in real-world demand, even under be¬
nign wartime circumstances, presents a problem. One response
might be to buy “sufficient” resources to mitigate the effects of unpre¬
dictability with inventory. And, indeed, in those few cases where cur¬
rent formal systems address uncertainty at all, that tends to be the
approach. Unfortunately, as Fig. 2.4 shows, investing more money
may not solve the problem. Even if we were willing to more than
double our WRSK investment to accommodate the actual VTMRs, we
still might not buy the correct mix of spares because the VTMRs are
not stable. As a result, as Fig. 2.4 suggests, for the converter pro¬
grammer, “buying out” could still result in a shortage of components
in the face of a rising VTMR; for the unified fuel control, many parts
might go unused in the face of a declining VTMR. While the time pe¬
riods used in this illustration are relatively short compared to pro¬
curement lead times for aircraft spare parts, they do serve to suggest
the nature of the problem: instability in observed VTMRs.
VTMRs of worldwide quarterly demands
per 100 flying hours
July 1980-June 1981
July 1981-Sept 1982
FI 00 Unified
Fig. 2.4—^Why Buying Out Is Not the Answer
Moreover, the “buyout” strategy—even if practical for peacetime—
does not address the unpredictability of the wartime environment.
We do not know how wartime demands will differ from peacetime ex¬
perience. Conceivably, the adversary could induce the need for un¬
planned, unanticipated mixes of missions or force beddown changes;
damage the support infrastructure; disrupt support systems; or de¬
stroy resources. These possibilities make the resourcing problem ex¬
traordinarily difficult for the conventional models used in formal sys¬
Formal systems deal with the difficulty these risks present by ignor¬
ing them. Air Force planning for wartime is based on the use of
“planning scenarios.” For example, the computation of expected de¬
mands for such resources as spare parts, repair, and transportation
uses the War Mobilization Plan’s (WMP) specified flying programs.
Many resources are allocated by using peacetime flying history,
merely scaling up peacetime observations (i.e., mix of components to
be repaired, repair hours) to reflect the difference between peacetime
and planned wartime flying hours. But deviations from these scenar¬
ios are inevitable. Yet, typically, capability assessment models are
not used to examine the implications of such deviations to warfighting
capability. And no attempt is made to predict the demand created by
a less-than-benign wartime environment, or to assess how support
systems will respond.
For some contingencies, such assumptions of predictability are not
prudent; for a high-intensity conflict, such as the NATO scenario im¬
plied, they are inappropriately optimistic. Given current resourcing
techniques and what the wartime environment might turn out to be,
it is virtually certain that some resource constraints will develop at
This is not simply a problem of prediction. It seems reasonable to as¬
sume that wartime will not turn out to be merely a scaled-up peace¬
time operation. But beyond that, assumptions become simply
guesses. We cannot realistically anticipate the demands generated by
actual wartime flying programs, much less those created by enemy
attacks on infrastructure. In a very real sense, all we can say is that
the demands of war are likely to be uncertain and largely unpre¬
dictable. Thus, even if we made progress in predicting peacetime de¬
mands, the wartime uncertainty would remain. It is likely that some
undesirable level of uncertainty will prevail lor peacetime, as well.
Yet the formal resourcing system consistently assumes the opposite.
THE CURRENT SYSTEM
To understand how thoroughly current planning depends on pre¬
dictable demands, consider the existing support systems. The left
portion of Fig. 2.5 provides a schematic of airbases in the theater.
The main operating bases (MOBs) have primary operating stock
(POS) and an increment of stock to take care of the increased flying in
wartime called base level self-sufficiency stocks (BLSS). The MOBs
have full intermediate-level repair capabilities. Some of the repair
facilities are hardened. However, the collocated operating bases
(COBs) may have some weapon systems—for example, the F-16—^that
are deployed with only the WRSK, and others—^for example, the F-
15—that have a WRSK and limited intermediate-level repair capa¬
bility, but no hardening. However, in the longer run, hardening of
some of the repair facilities was planned in the NATO scenario.
Since we intend to provide resources to these organizations to be self-
sufficient for the first 30 days of war, there is an implicit assumption
Command and control
— Command and control
• POS + BLSS
• Full 1-Level repair
• Some hardening
• Limited or no repair
• No hardening
Air Logistics Centers
Fig. 2.5— ^The Current System
that there will be no resupply from the depot during that time. There
may, in fact, be some limited resupply. However, the self-sufficiency
orientation may, in fact, detract from the system’s concern with de¬
veloping effective resupply and lateral supply systems. Figure 2.6
adds to Fig. 2.5 the uncertainties that may cause the resources pro¬
vided to be insufficient to meet the needs of the combat forces. The
arrow on the left focuses on the uncertainties that the theater might
face. As indicated, demands might occur on the flight line that are
unanticipated even in “benign” environments. These unanticipated
demands may result from the dynamic flying programmed by the
Command and Control System. Moreover, as the peacetime data in
Figs. 2.1 and 2.2 demonstrated above, unanticipated demands may
result in ways that underscore the fact that we simply do not under¬
stand the demand-generation process. Furthermore, in the kind of
high-intensity war anticipated in the NATO environment, for exam¬
ple, many attacks may cause resource losses or system disruption. It
may be that as a result of these attacks, the numbers of surviving
personnel, spares, repair resources, and the like within bases will be
imbalanced so that the sortie potential of the surviving aircraft will
be seriously impaired. Where repair facilities and aircraft shelters
are hardened, the extent of this imbalance in surviving resources is
likely to be reduced. In anticipation of an attack, elements of the
combat forces may need to disperse aircraft and other resources tem¬
porarily. After an attack, aircraft may be pinned in or pinned out so
that combat operations cannot proceed as planned.
Such eventualities are largely ignored in resourcing combat organi¬
zations for wartime. Despite the fact that these kinds of eventualities
could dominate the combat scenario, units have been resourced as if
demands were predictable. Furthermore, despite critical shortages
that might develop, plans for using the depots in such circumstances
remain, at best, in doubt.
In effect, the prevailing view is that resources are provided for 30
days, and hence no resupply needs to take place for this period of
time. This view is reinforced by instructions to organizations deploy¬
ing without maintenance to delay the repair of removed components
until maintenance arrives at the unit during the last week of the 30-
day period. Of course, when the need arises, it is likely that resupply
will indeed take place. However, the uncertainty orientation would
suggest that the need for resupply is inevitable and that aggressive
management systems for retrograde® and resupply should be planned
for and should be in place.
The right side of Fig. 2.6 presents the depot side of the operation in
peacetime. The peacetime system is described because the plans for
wartime operation are unclear. In wartime, overrides to the standard
system probably will be required. Yet, these overrides are not made
explicit. Here the problems stem from the unpredictability of demand
and the assumption of a stable, steady-state system. We examined
unpredictability in Figs. 2.1 through 2.4; we now turn our attention to
the assumption of a stable, steady-state system.
The problem of shortages is not limited to unanticipated demands.
More often than not, logistics system managers recognize that short-
®The evacuation of repairable aaaets from the theater of operations to Air Logistics
Centers in the continent^ United States (CONUS).
ages do and will exist and that management systems need to be de¬
signed to respond to these shortages. Thus there has been a continu¬
ing interest in providing differential support among units, missions,
theaters, etc. The importance of setting goals for combat units so
that resource allocations can be made to “make the best of’ available
resources in the face of differences in unit or mission priorities is
receiving increasing attention.
Within existing priority groupings, the current system distributes as¬
sets in the order in which they were requisitioned. The oldest requi¬
sition is the first filled. This scheme may be appropriate in an opera¬
tion where demand is both predictable and stable; in the Air Force re¬
coverable spares management system, it is neither. As a result, what
is distributed to a base may bear little relationship to need.
On the repair side, the system is similarly hampered by the mis¬
guided assumption that demand is predictable. Repair contracts are
established using old asset position and demand data to predict what
the demand will be for the next several quarters. If demand rises
above a quarter’s prediction, the tendency is to defer that demand
until the next quarter. The hazard is a repair system that may be
neither sufficiently timely nor sufficiently relevant. Overrides to the
system will obviously occur, and those overrides will give emphasis to
emerging needs rather than predicted needs. Without specific plan¬
ning and detailing of a wartime system and without a “fine-tuning” of
near-real-time system needs during peacetime operations, it’s less
likely that the depot will be as timely and as relevant as it needs to
The exception to this unhappy situation is a MICAP.^® When an
airplane is down for lack of a part, that part is handled with dispatch
in the depot repair system. It could perhaps be handled with more
dispatch, but at least the criterion is appropriate; the airplane,
rather than the part. It is this portion of the system that could make
the system more timely and relevant in wartime. This is likely to be
especially so if repairable assets are transported to the depot
promptly. What is obviously required is more explicit planning and
implementation procedures regarding extraordinary depot respon¬
siveness in wartime. This need was reinforced by Operation Desert
Shield/Storm, during which ad hoc procedures were often invented to
cope with particular urgent shortages.
parts shortage aiTecting Mission CAPability.
3. AN ALTERNATIVE APPROACH
Although formal support systems underestimate the variation in
peacetime demand, the logistics system itself continues to function,
meeting unanticipated demands much of the time. Informal man¬
agement adaptations, ignored by the formal resourcing system, often
solve the problem. These actions provide clues to the institutional ar¬
rangements and formal management adaptations the Air Force
should use to extend management’s ability to address uncertainty—
especially the greater amounts of uncertainty that might exist in
wartime. Basically, they suggest a very responsive and adaptive sup¬
port system—one that that will react quickly and positively to meet
unanticipated demands. This report represents an attempt to devise
such changes, integrating Logistics and Operations so that the logis¬
tics system can do a better job of meeting resource demands, despite
FOCUS OF THE ANALYSIS
At this stage, the initiatives are narrowly focused on a logistics opera¬
tions or execution system. Figure 3.1 shows the scope of functions
studied to date. Although the issues we have discussed may be rele¬
vant to many other facets of logistics, our suggestions currently focus
on one critical resource: aircraft recoverable spare parts. They en¬
compass the processes of supply and repair, both at the depot and at
the base; flight line removal and replacement; and transportation and
The functions displayed in Fig. 3.1 eire parts of the execution stage of
the PPBES (Planning, Programming, Budgeting and Execution
System). Within this stage, repair, workloading, and distribution
(location) decisions are emphasized. Other functions in the stage,
such as procurement, are not discussed. Although the current analy¬
sis emphasizes this portion of the PPBES, the ingredients of a pre¬
ferred execution system need to be reflected in the other stages of the
PPBES. Section 4 provides examples. For the present context, suffice
it to say that functions within a stage and the four stages of the
PPBES must be integrated and act in concert with one another.
Airbases Air Logistics Centers
Fig. 3.1—Scope of the Analysis
To make the execution stage in the PPBES less vulnerable to the
large errors in decisionmaking that might result from uncertainty, we
suggest reducing, where possible, dependence on long-term prediction
of demands. ‘ The first principle behind this strategy is that logistics
operations be based on demands as they become known in real time
and as they are predicted more reliably over very short horizons.
Although even these predictions will often be wrong, especially in
wartime, hedging strategies that make and revisit allocations over
such short periods are likely to be useful.
Misallocations are inevitable, of course, but one powerful tool for
dealing with them is to reduce response times. When unanticipated
demands occur, a support system with very short response times can
mitigate the effects of such misallocations through lateral supply,
very timely depot replenishment, or other management actions.
Shorter response times also reduce pipelines, thereby reducing safety
stock requirements and total spares investment costs.
^Clearly, this is not always possible, especially in decisionmaking about capital in¬
vestments, repair capacity, contract repair, and similar decision contexts in which one
is constrained to longer lead times.
The second principle of this approach is that Logistics must be linked
very closely with Operations. System performance goals provide an
important part of that link. It is suggested that the role of the
Logistics community is to provide Operations with the necessary
number of weapon systems appropriately configured to meet mission
needs as judged by Operations. Of course, support is inherently a col¬
laborative process. There are often various ways to meet a given op¬
erational need, and the differences may be important to Logistics.
Also, because of “inevitable” resource constraints, appropriate com¬
promises may need to be worked out by Operations and Logistics per¬
sonnel. But if analysis of the support system at this level is to be
feasible, the judgments by Operations cannot be at issue.
To make the proper decisions that will provide each unit with the
necessary resources, Logistics needs appropriate goal measures. To
ensure that these measures reflect operational needs tempered by the
feasibility of attainment, they should be set by Operations in concert
with Logistics. Goals related to weapon systems performance, unlike
those that focus on components or commodities, closely reflect opera¬
tional and mission urgencies. They also promote a common under¬
standing by Operations and Logistics personnel. The system perfor¬
mance strategy, then, is this: Decisions reached with current asset
status information should use weapon system needs as the objective
function. In this report, aircraft availability is used as the goal mea¬
With its focus on very short planning horizons, the system must be
especially sensitive to rapidly changing needs. Especially in high-in¬
tensity warfare, it seems vital to give those directing wartime opera¬
tions the flexibility to employ units in accordance with dynamic oper¬
ational urgencies. Operations may alter its assessment of the relative
mission importance of some units, for example. As a result, aircraft
availability goals might be changed. New allocation and reallocation
of assets will be required. Even if unit availability goals stay rela¬
tively constant, allocation/reallocation of unit resources may often be
needed to help Operations achieve its goals. Repair might be con¬
strained, for example, by inadequate capacity, shortages of bits and
pieces, or damage from enemy attacks.
When resources are constrained, the logistics system must use what¬
ever resources are available to satisfy the most critical needs of the
combat forces. This may mean the reallocation of resources across
units as well as the allocation of incoming resources from the
Continental United States (CONUS) or elsewhere. In addition to
maintaining a focus on units or bases, we need to consider the totality
of resources, unit, theater, CONUS, and worldwide, and to determine
how those resources can be allocated or reallocated to provide support
where it is most needed. This suggests the need for an effective prior¬
ity system, one that discriminates among alternatives in a way
consistent with operational urgencies. But the common approach,
merely saying that one weapon system is more important than an¬
other, provides little help. TVpically, priorities are useful only where
there is a means for indicating the relative importance of weapon sys¬
tems by the quantitative use of goal statements. Thus, weapon avail¬
ability goals by unit potentially provide a more operationally useful
means for allocating resources. Aircraft availability goals specified by
unit will permit statements of differential needs across bases (even
within the same weapon system, if desired) that can be changed over
If the formal support system is to provide resources to bases in a way
that is consistent with operational needs over the short run, it will
often have to deal with the unexpected demands that, as we have
seen, are likely to occur in the face of uncertainties. Under the cur¬
rent system, when local shortages or maldistributions across units
develop, informal management adaptations such as redistribution,
lateral supply, and lateral repair are often used to respond; however,
such adaptive responses need to be more proactive (unless it can be
shown that enhanced responses to events as they occur are sufficient
to meet operational needs) and systematic than they now are. Such
tactics, applied systematically and on a wider scale, can help the for¬
mal support system meet Operations’ changing needs. This approach
is based on the idea that the system needs to be flexible enough to
provide the decisionmaker with levers to use in innovative ways.
Thus, it is important in many circumstances, especially in wartime, to
have a variety of well understood, well practiced management adap¬
tations available. Redistribution, lateral supply, and lateral repair
need to become well developed and well practiced mechanisms sup¬
ported by visibility and decision support that make them systematic.
These techniques are particularly important in solving the alloca¬
tion/reallocation problems just discussed. The idea here is not simply
to respond to MICAPs, but rather to be proactive in resource alloca¬
tion, precluding MICAPs before they occur. Given the unpredictabili¬
ties that have been discussed, attempting to be proactive in an envi¬
ronment of long planning horizons and long response times would
make no sense. But where response times are very short, then a
proactive approach may be justified because the cost of recovering
quickly from a wrong decision may be small. In particular circum-
stances there may not be a need to be proactive, for example when re¬
sponse times are very short or the scenario especially dynamic.
Orchestrating the short-horizons approach and responsive adapta¬
tions described above requires an advanced system for combat sup¬
port command and control. The specification of goals, to take just one
component, presents a significant challenge. The C® system must
elicit requirements over the short run from Operations. If the re¬
quirements cannot be met, the next best options suitable to
Operations need to be worked out. Likewise, if the infrastructure is
to respond to near-real-time needs, it must have good visibility of re¬
sources. When making decisions about the components to be repaired
in support of the combat forces, for example, very current information
is required about the worldwide asset position as well as the current
status of aircraft. Such visibility must be updated frequently 2 And
the system must support the unit-theater-CONUS-worldwide per¬
This element of the system, like other elements, needs to be able to
respond appropriately to unanticipated events. It is inevitable that
the system will need to act in degraded modes. Delayed and other¬
wise degraded information is likely to be a way of life in high inten¬
sity conflict involving enemy attacks against the infrastructure and
its control system. Backup modes may be needed to sustain opera¬
tions. Understanding the effects of particular kinds of information
degradation on the quality of decisions is £dso important in the design
of an effective and robust command and control system.®
Command and control systems are touched upon here; the means for
establishing their goals, for example, is the subject of future reports.
But in broad outlines, such a system would be especially concerned
with eliciting aircraft availability goals from Operations, tasking the
infrastructure appropriately, and assuring that corrective actions are
taken when responses to the tasking are ineffective.
^This attribute has special significance for repair organizations that are not collo¬
cated with combat units. Typically, collocated intermediate-level maintenance does not
operate on the basis of long-run forecssts of the repair needs of supported organizations
because they have first-hand, current knowledge of the state of the force. Issues of
providing special visibility become important for noncollocated repair organizations.
This situation is the case for depots. In such cases, the use of near-real-time data for
repair decisions is critical.
®This issue is explored in Gustafson, op. cit.
If the infrastructure is to respond to near-real-time needs, it must
also have the capability to respond to unanticipated, urgent demands,
sometimes of unusual size. It might achieve such robustness by hav¬
ing resources that are able to respond to wide ranges of demand.
Priority systems, for example, imply common resources, including
both transportation and repair. If the repair resource, people or
equipment, is stovepiped —i.e., dedicated to only one kind of asset—a
priority system has no leverage. In such cases, there may be no way
for the repair system to take care of unanticipated demands except to
buy more repair capability or more stock. Yet, as we saw earlier, such
attempts to buy out may result in considerable sums of money being
spent to take care of unanticipated events that may never occur. An
option may be to enhance the scope of repaid by making repair re¬
sources common to a number of components. This can be achieved
through investments in test equipment and other repair resources
that can be used to repair many different kinds of components, for ex¬
ample, or by cross training personnel so that they are skilled in the
repair of many components.* More or fewer of these common re¬
sources can be applied to satisfy urgent needs depending on how de¬
mands eventuate. It may be more cost-effective to invest in enhanc¬
ing scope of repair than in trying to buy “sufficient” repair or spares
to hedge against all of the uncertainty in demand.
THE CLOUT INITIATIVES
In pursuing the directions just outlined, we have developed a number
of integrated initiatives that we call CLOUT: Coupling Logistics to
Operations to meet Uncertainty and the Threat. The purpose of
CLOUT is to integrate Logistics and Operations and to provide a
number of management adaptations so that we can meet support de¬
mands responsively, despite many of these uncertainties.
Given the foregoing backdrop, a schema for the workings of a respon¬
sive infrastructure is provided in Figures 3.2 and 3.3. In the theater,
for example (see Fig. 3.2), one might want to provide a MOB with the
ability to support some COBs with lateral repair. In that case, the
MOB would have to be sized, resourced, and managed so that it could
respond to the intermediate-level repair needs of its assigned COBs.
To reallocate assets in response to operational goals, a proactive lat¬
eral resupply system is needed that reallocates assets in very short
*Gotz, G. A., and R. E. Stanton, Modeling the Contribution of Maintenance
Manpower to Readiness and Sustainability, RAND, R-3200-FMP, January 1986.
Command ar.d control
Command and control
Effects of unpredictable
• Need to
Fig. 3.2—CLOUT: Enhancing Theater Responsivenegg
times. Such characteristics imply a more advanced European Dis¬
tribution System (EDS) and a more advanced Pacific Distribution
System (PDS).® In this context, for example, a more advanced EDS is
one that not only responds quickly to MICAPs but attempts to pre¬
clude them, and it would recognize that operational needs at the
bases may be dynamic, so quantities of resources at the bases may
have to be adjusted quickly in accordance with this need. Those deci-
^The EDS was the intratheater transportation system of the U.S. Air Forces
Europe; the PDS was the proposed intratheater transportation system of the Pacific Air
Command and control
Airbases Theater Air Logistics
Fig. 3.3—CLOUT: Enhancing Depot Responsiveness
sions need to be in the hands of the operations and combat support
controllers. Their wisdom will be required to avoid excessive turbu¬
lence in the system that could conceivably result from rapidly chang¬
ing goals and priorities.
To enhance the depot’s responsiveness (see Fig. 3.3), assured and re¬
sponsive intertheater lift must be available, not just for moving ser-
viceables to the theater, but for returning parts to be repaired. Both
supply and distribution will be based on theater needs, as defined by
In addition to these initiatives, there might be payoff from establish¬
ing an intermediate-level repair facility somewhere in the rear: for
example, in the UK or Portugal.® A kind of “Queen Bee” concept—af¬
ter the fashion of consolidated jet engine field maintenance arrange¬
ments of the past, where one base might support 120 or so aircraft—^is
also a promising alternative. Locating a limited depot capability in
the theater to complement the CONUS depot capability may also
Figure 3.4 places this infrastructure into the context of a command
and control system. If the theater command and control system has
near-real-time asset information and can specify availability goals for
each base and change those goals from time to time to reflect opera¬
tional urgencies as they unfold, we can develop algorithms to allocate
resources or assign priorities over short planning horizons to maxi¬
mize the chances of achieving the goals. For example, the algorithm
could advise the MOB how to sequence repairs of components to pro¬
vide relevant, timely support to the COBs as well as to the MOB it¬
self, thus contributing most effectively to meeting the operational
goals. The same logic applies to theater distribution. The allocation
and reallocation of components should also be based on maximizing
the probability of achieving the aircraft availability goals.
The depot problem has substantial similarity to the theater problem.
Information about availability goals by base, and information about
the near-term flying program as well as the worldwide asset position,
can provide the needed priorities. In this case, we have actueilly writ¬
ten a protot 3 q}e assignment algorithm. Called DRIVE (Distribution
and Repair In Variable Environments), it determines the priorities
for the component repair system and for the distribution system so
that maximum support is provided according to the availability crite¬
ria.’^ More is said about the DRIVE prototype in Sec. 5.
Damage and disruption are likely to involve the need for allocating
and reallocating spares and other resources. The advanced command
and control system needs to be able to support the decisionmaking
that will be required in such circumstances. Another anticipated sit¬
uation that the command and control system should be able to help
manage is that in which aircraft land at bases other than their home
bases. Aircraft may be pinned out of their bases as a result of enemy
^During this study there were particular reasons for considering Portugal.
^Abell et al. and Miller and Abell, op. dt.
Command and control Command and control
attacks. Under such circumstances, it may be a matter of consider¬
able urgency to be able to launch combat sorties from locations other
than home base. There may also be other reasons for Operations to
have the flexibility to operate away from home base for short periods.
In such situations too, the allocation and reallocation of spares and
repair are likely to need to be extended to other resources.
Figure 3.5 reminds us that CLOUT is not just combat support opera¬
tions. We need to include notions of uncertainty and the system’s
adaptations to uncertainty in each stage of the PPBES. We have se¬
lected the execution phase for initial review and change, in part be¬
cause it might be argued that planning, programming, and budgetary
systems need to reflect the essence of the execution system. Certainly
if each stage is to be integrated with the others, explicit consideration
of uncertainty and management adaptations should be common to all
RAND is also exploring ways to incorporate explicit consideration of
uncertainties and management adaptations in requirements systems.
With an upgraded requirements system, the programming and bud¬
geting stages of the PPBES will become consistent with the suggested
execution and planning stages. As mentioned earlier, a study is also
being formulated to develop and demonstrate multiechelon spares
and repair and financial management systems for the Air Force that
are intended to explore additional means for implementing lateral re¬
pair, lateral supply, and priority repair.
Secs. 4 and 5 provide more concrete views of the CLOUT initiatives.
Evaluations of some of the CLOUT initiatives are also reported. It
should be emphasized that even in the spares, repair, and trans¬
portation areas, CLOUT needs extensions. The representation of un¬
certainties that the logistics system may face in wartime also needs
improvement in those decisionmaking processes involving irreducibly
long lead times. Additional initiatives need to be identified for poten¬
tial use in making the logistics system more robust in the face of
peacetime and wartime uncertainties and the dynamic support needs
of Operations. Moreover, CLOUTs orientation needs to be extended
to resources beyond spares, repair, and transportation. For example,
munitions management is clearly a logical extension to the thinking
underlying the CLOUT initiatives. Munitions availability could be
improved through systematic reallocation when demands evolve in
unanticipated ways owing to changes in planned targets or
unexpected rates of expenditure. A veiy responsive industrial base
could be another important management adaptation for dealing with
uncertainty in this context.
Command and control Command and control
4. AN ASSESSMENT OF CLOUT
In this section we explore, under a variety of wartime scenarios, some
of the potential payoffs of the responsive and robust CLOUT initia¬
tives discussed in the preceding section.
These scenarios reflect wartime resource demand uncertainties, some
postulated effects of battle damage, and the potential need for more
flexible wartime basing options. We focus primarily on the theater
CLOUT initiatives, but we also consider how a responsive CONUS
depot repair system with a comparably responsive intertheater trans¬
portation system might also contribute to enhanced system perfor¬
Our principal findings are as follows:
• Current planning for wartime combat support does not take
sufficient account of the uncertainties of potential wartime
requirements and scenarios.
• The CLOUT initiatives can help absorb and mitigate the effects
of such uncertainties and the effects of battle damage as well.
• CLOUT enhances the ability of the logistics system to support
alternative basing strategies and dispersed operations.
CAPABIUTY ASSESSMENT TOOLS
Capability assessment tools are central to the planning process.
Enhanced tools (in light of CLOUT) are in development and in proto-
t 3 q)e use at RAND. In the past few years, there have been advances
in analytic evaluation models. They use more appropriate criteria,
such as available aircraft or sorties, and they incorporate a wide
range of wartime dynamics in terms of changes in flying programs or
how the force is phased into the war.
However, there are serious shortcomings in the way most capability
assessment models handle repair, and as we have seen, repair is criti¬
cal to CLOUT. Essentially, these models allow repair time to be in¬
dependent of workload in a shop. Effectively, this provides uncon¬
strained repair capacity, a very inappropriate assumption for some
critical shops. The current models do not use the more advanced
kinds of priority systems. Moreover, they do not deal with priority
distribution. Nor do they represent the lateral supply or lateral re¬
pair capabilities that we have been discussing.
RAND’s development in this area is intended to provide initial at¬
tempts to overcome these limitations. Indeed, we must have them if
we are to evaluate the CLOUT initiatives (management adaptations)
against different levels of uncertainty. To date, we have focused our
efforts on avionics repair. These protot3T3e developments need to be
evaluated for their applicability to the repair of other aircraft sys¬
The RAND models cover two assessment scenarios: theaterwide as¬
sessments and repair shop assessments. This section incorporates
some of the results from the first model, called Dyna-METRIC
Version 5.^ Another model, Dyna-SCORE, was used in exploring
management initiatives in depot-level component repair.^
DETAILS OF THE CLOUT INITIATIVES
In the paragraphs that follow we discuss our approach to evaluating
the CLOUT initiatives and clarify our assumptions.
Reactive and Proactive Lateral Resupply
The Air Force has established a goal of 1.5 days as the time required
for the European Distribution System (EDS) and the proposed Pacific
Distribution System to move an asset from one base in theater to an¬
other. Although the goal is the same in wartime and peacetime, for
evaluative purposes we used a two-day in-transit time. Currently the
EDS causes a lateral resupply action to occur only when an aircraft is
not fully mission capable (NFMC) at one base and another base has
the needed part—i.e., the system is reactive. CLOUT envisions proac¬
tive lateral resupply made feasible by near-real-time visibility of the
asset position at all bases in the theater, coupled with very short re¬
sponse times and a more sophisticated command and control system
as suggested in Sec. 2. In particular, when possible, we ship a part
from one base to another to preclude an airplane being kept down for
lack of the part rather than waiting for the shortage to occur before
^Isaacson and Boren, op. cit.
^Tsai, op. cit.
simple zero-balance rule was used: if one base has no serviceable asset and one
or more other bases can give up an asset (and more than one serviceable is on hand),
we ship from the “richest” base. In this analysis we treat all bases with the same
Mission Design Series (MDS) with equal priority. The DRIVE algorithm, briefly dis¬
cussed in Sec. 3, has the potential to provide even better decision support, and to deal
explicitly with diflerential priorities across bases as discussed in Sec. 1.
The intermediate-level priority repair and distribution rules we used
are based on a theater criterion: maximizing the number of aircraft
in the theater that are fully mission capable.^
Responsive Depot Support
The analysis includes responsive depot resupply of test equipment
parts as well as aircraft components, i.e., line replaceable units
(LRUs). Test equipment parts are critical in keeping intermediate
repair facilities in the theater operational, and, t 5 T)ically, the depot is
the only source of supply when local stocks are depleted. Because of
modeling constraints we do not model the depot resupply of LRU re¬
pair parts; therefore, we underestimate the potential payoff of a re¬
sponsive depot system. We model the depot only in terms of resupply
time because we cannot, at the present time, model priority depot re¬
pair and distribution.®
SCENARIO AND SCOPE OF ANALYSIS
Our scenario is basically a NATO wartime scenario. The measure of
effect is the percent of aircraft not fully mission-capable (NFMC) at
the end of the first 30 days of conflict. We chose this performance
measure because it is more demanding than a measure of partial
mission capability and somewhat more sensitive to inventory system
performance. We will examine three cases: (1) one in which there is
a more realistic level of demand uncertainty but no battle damage; (2)
one in which there is differential damage to spares and repair facili¬
ties across bases due to airbase attack; and (3) one involving an alter¬
native basing strategy but no damage.
The analyses focus exclusively on F-15 and F-16 avionics LRUs that
are included in the range of the WRSK. Avionics components tend to
be both cost drivers and the primary cause of NFMC aircraft.
^The model used in these analyses would not reallocate an asset to a base through
lateral supply if an asset were already enroute to the base from the depot. The priority
repair rule used was rather sophisticated. If lateral supply was not in effect, it priori¬
tized the repair of assets to alleviate shortages first at the base with the hipest pro¬
portion of its aircraft NFMC. With lateral supply, it prioritized the repair of the assets
with the most shortages theaterwide. If no shortages existed, it prioritized the repair
of assets with the earliest anticipated shortages. Lateral repair did not involve a cen¬
tralized intermediate repair facility, only a main operating base.
®Our models are currently being changed to model depot repair and distribution
Assessments with Dyna-METRIC show that more than 75 percent of
the aircraft down are NFMC because of avionics.® Avionics compo¬
nents constitute more than 80 percent of the cost of a typical WRSK.'^
The item demand data were extracted from WRSK listings main¬
tained by Headquarters, AFLC. Bench check serviceable (BCS) rates
were estimated from data extracted from the maintenance data col¬
lection system. Wartime fl 5 dng programs are the same as those used
in computation of the WRSK requirement by Headquarters, AFLC.
We have represented demand uncertainty in four ways: First, the
base case represents the standard Air Force wartime planning as¬
sumptions with a VTMR of 1.0 for all components. Second, we have
also included what we call fault isolation uncertainty. Frequently, a
maintenance technician removes what he believes is a failed compo¬
nent from an aircraft on the flightline and sends it to the intermedi¬
ate-level maintenance shop to be repaired. In the shop the component
is placed on the test stand but no malfunction is found. These BCS
removals are not counted in the supply system; therefore, they are not
counted in the standard estimation of component demand rates used
in computing peacetime and wartime spare parts requirements. In
peacetime, BCS actions constitute about 30 percent of all avionics
removals worldwide, but the rate varies greatly among bases and
components.® Third, we used actual peacetime VTMRs for major
components.® They range from 0.75 to 5.0 for individual components,
but equivalent performance (percent NFMC on day 30) is obtained if
the VTMR for all components is set to about 3.0. Fourth and finally,
we considered the case where the VTMR of every component is set
equal to 4.0 to reflect the additional uncertainty in the demand for
spare parts that may be experienced in wartime.
Avionics intermediate-level maintenance is available at all F-15
bases, along with a test equipment spares kit (separate from the
®Dyna-METRlC Version 5, the capability assessment model used in these evalua¬
tions, is described in Isaacson and Boren, op. cit.
^Simple examination of WRSK listings on the AFLC CREATE computer system
®Source: AFLC D056 data system using data from 1985 and 1986. In wartime and
in peacetime exercises simulating wartime, flight line mechanics may be more careful
in the face of supply shortages or the lack of an intermediate-level repair facility. They
may swap what they believe are good components in and out of the aircraft before
declaring that a particular component has failed. Such adaptive behavior is difficult to
model or predict, so we assumed peacetime BCS experience for this analysis.
®The source for the VTMR estimates was also D056 data from 1985 and 1986.
WRSK or BLSS).^° In the F-16 case, intermediate-level avionics
maintenance is available only at main operating bases (MOBs), with
test equipment speures support available only from standard POS and
BLSS stocks. In this analysis we assume that there is no intermedi¬
ate-level repair taking place at the F-16 collocated operating bases
(COBs), and that they have to rely solely on their A^SKs.^^ It is
assumed that all WRSKs are full at the start of the war.
Version 5 of the RAND Dyna-METRIC model used in this analysis in¬
corporates the lateral resupply rules and repair priority rules we have
already described. The model assumes full cannibalization of LRUs
at the flight line to minimize the number of NFMC aircraft due to
supply shortages—i.e., it assumes that every component of the air¬
craft can be removed and installed on another aircraft. The model
also reflects test equipment failures and the resupply of test equip¬
ment parts from the depot (if available). If the right part is not avail¬
able to fix the test equipment, that test stand becomes partially capa¬
ble in that it can repair only that subset of the LRUs not affected by
the missing part (or parts). If two or more test stands of the same
type are at a particular location, the model allows cannibalization of
parts from the stands so that the greatest range of LRUs can be re¬
paired on at least one of the stands.
No less than 50 replications of a 30-day war underlie each of the re¬
sults portrayed in the graphs that follow. The results shown are ac¬
tually means of the 50 or more experimental replications. In those
cases where the NFMC rate is high, i.e., 40 or 50 percent, the experi¬
mental error is higher than when the outcome is, say, 15 or 20 per¬
cent. To gain some intuition about the role of experimental error in
this analysis, consider the fact that with 50 replications, an NMCS
rate of 30 percent can be expected to vary between about 28.4 and
^*^Special Purpose Recoverables Authorized to Maintenance (SPRAM), a kit of re¬
pair parts valued at about $8 million, is separate from the WRSK. This kit has greatly
improved repair performance at bases where there is only a single avionics test stand;
with two test stands but without such a kit, cannibalization is typically required to
keep at least one test stand fully operational. A SPRAM kit was us^ in Coronet
Warrior I. Further discussion of this topic can be found in Pipp, Capt. D., USAF,
“Coronet Warrior—A WRSK Flyout,” Air Force Journal ofLogietica, Smnmer 1988.
^^This is consistent with Air Force planning for wartime in the F-16 case. As dis¬
cussed earlier, flightline mechanics at COBs may, in wartime, become more inventive
in the face of supply shortages and use an airplane that is already down for some other
reason as a hot mockup to perform limited intermediate-level maintenance on failed
components. Such behavior has been observed in F-16 peacetime exercises simulating
wartime. Again, such adaptive behavior, called “shade tree maintenance” by some, is
difficult to predict and to model. See Viccellio, H., Mqj. Gen., USAF “Coronet Warrior
II,” briefing given to the Air Force Logistics Conference, 1988.
about 31.6 percent roughly 95 percent of the time. This result de¬
pends on the number of aircraft in the scenario as well as the mean
proportion of aircraft NMCS; nevertheless, it is safe to say that the
experimental error associated with these analyses is sufficiently
small that it alone does not account for the magnitude of any of the
differences in system performance portrayed in this section.
Many different cases were examined during the course of this work.
We have selected for inclusion here only a few that we felt were illus¬
trative of the performance gains that can be achieved with the
CLOUT initiatives, not necessarily those that had the most improved
performance. It is important to note that there was no case in which
performance did not improve as a result of the CLOUT initiatives, al¬
though, clearly, one can postulate scenarios that tend to defeat them.
For example, if there were only one base having aircraft of a
particular type in the theater, the value of lateral supply or lateral
repair would be sharply reduced, since it only applies to items that
are common to aircraft at more than one location. Ignoring such
pathological exceptions, it is reasonable to say that these adaptations
always pay off in terms of improved system performance.
CLOUT PAYOFFS WITH WARTIME DEMAND UNCERTAINTY
AND NO DAMAGE
We begin our description of the CLOUT payoffs by considering the ef¬
fects of realistic levels of demand uncertainty on aircreift availability.
The scenario involves 144 F-16s, including one MOB with a wing of
72 aircraft, and three COBs, each with a squadron of 24 aircraft. The
MOB has two sets of avionics test stands, the normal allocation of
primary operating stocks, plus the wartime increment (BLSS). The
COBs have no intermediate-level repair capability, but each has a
remove-and-replace WRSK. Consistent with standard Air Force
planning for wartime, there is no depot resupply for the first 30 days
of the war. Figure 4.1 shows the effects on system performance of
explicitly accounting for fault isolation uncertainty and more realistic
levels of demand variability.
Given the standard Air Force planning assumptions (demand rates
observed in peacetime and a VTMR of 1.0), about 14 percent of the
144 aircraft will be down on day 30.^^ If we take account of fault
the time of this analysis (1986) the design support objective (DSO) for the
WRSK was 4 out of 24 aircraft Nf^C (or about 17 percent) for all components (not just
Standard Fault Fault Fault
demands isolation isolation isolation
VTMR = 1 uncertainty uncertainty uncertainty
VTMR = 1 VTMR = 30 VTMR = 4.0
• One 72-aircraft
-Two test sets
- POS + BLSS
• Three 24-aircraft
- No repair
Fig. 4.1—Effects of Uncertainty on F-16 Aircraft Availability
isolation uncertainty, the NFMC rate rises to almost 25 percent. If
we include the actual observed peacetime VTMRs (roughly 3) in addi¬
tion to the fault isolation uncertainty, more than 30 percent of the
aircraft are down. If we use a VTMR of 4.0 to reflect some of the ad¬
ditional uncertainties of wartime, 40 percent of the aircraft are not
fully mission capable due to avionics by day 30. As we suggested in
Sec. 1, the effects of demand uncertainty can be signiflcant.
Figure 4.2 shows how CLOUT counters the effects of demand uncer¬
tainty. For easy reference, the left side of the figure reproduces the
data in Fig. 4.1. The responsive, proactive lateral resupply system
previously described substantially reduces the NFMC percentage.
The expected number decreases from almost 40 to just over 25 per¬
cent. If the MOB also does intermediate-level priority repair for the
COBs as well as for its own aircraft, the percent of aircraft NFMC is
reduced to about 20 percent.
avionics) on day 30 of the conflict. Since that time, the DSO has been increased to 6
out of 24 (or 26 percent) NFMC.
Uncertainty Effects CLOUT Initiatives
standard Fault Fault Fault Lateral Plus Plus
demands isolation Isolation isolation resupply lateral responsive
VTMR - 1 uncertainty uncertainty uncertainty repair depot
VTMR = 1 VTMR - 3.0 VTMR = 4.0 resupply
Fig. 4.2—Responsive Support Pays Off in a “Benign” Environment
Note that the repair capability at the MOB was intended to support
only 72 aircraft with its two sets of test stands under standard Air
Force planning assumptions. With the lateral repair initiative in
place, it is now supporting 144. Although MOB repair is saturated,
because the base is using priority repair—repairing the most impor¬
tant items—its repair capability still makes a significant contribu¬
tion. In addition to these gains, if the depot and intertheater
transportation can respond with 10-day resupply times on average,
they can reduce the number of aircraft NFMC (for avionics) to approx¬
imately 14 percent, roughly the original planning objective. The de¬
pot contributes to this reduction by absorbing the uncertainty in de¬
mand for the components that are beyond base-level repair capability
repair capacity of the MOB was not assumed to change in this illustration, to
show the payoff of later^ repair with existing numbers of test stands.
as well as by supplying the test equipment parts needed to keep the
mob’s maintenance activity fully operational.^*
Concern has been expressed about the availability of airlift to meet
the needs of a responsive depot system, especially during the early
days of a NATO scenario. Table 4.1 shows the airlift requirement in
pounds of cargo per day each way to transport all LRUs beyond nor¬
mal base repair capability (excluding engines) for all F-15 and F-16
units in a NATO conflict. The estimates assume wartime flying-hour
rates and peacetime NRTS rates. Although these estimates are sub¬
ject to uncertainty, they are reasonable approximations. The tonnage
shown requires less than two C-141-equivalent sorties per day during
the early surge period of the scenario, and less than one during the
following sustainability period. Since the vast msyority of critical
components can be loaded aboard narrow-doored aircraft, unmodified
Civil Reserve Airlift Fleet (CRAF) aircraft might be likely candidates
to fill this airlift requirement. The transportation requirement to
couple the depot more closely to the operational force is modest, and
the potential payoff is significant in terms of combat capability.
These results show that the CLOUT initiatives help mitigate the ef¬
fects of significant levels of demand uncertainties in wartime. They
also enhance system performance in the face of battle damage.
CLOUT SUPPORT IN THE FACE OF BATTLE DAMAGE
Another major source of wartime uncertainty derives from enemy air
attacks against our bases. Past simulation studies of such attacks
suggest that collateral damage of critical logistics resources, e.g..
F-15 and F-16 Depot Repair Airlift
(pounds per day each way)
^*Recall that we are underestimating the depot’s potential contribution in this
analysis because we are not representing the demand for SRUs for which the depot is
the primary supplier.
spare parts and avionics intermediate shops (AISs), is likely. Typical
Warsaw Pact regimental air attacks against three F-15 bases—
Bitburg, a MOB with 72 aircraft and two sets of test stands; Lahr, a
COB with 48 aircraft and two sets of test stands; and Sollingen, a
COB with 24 aircraft and a single set of test stands—were modeled to
assess their effects on system performance with and without the
CLOUT initiatives. The attacks were primarily targeted against
runways and aircraft in the open and in shelters. All bases were
modeled with aircraft shelters. Critical support resources such as
spares and avionics test equipment were dispersed in shelters or in
their own hardened facilities.^® Figure 4.3 shows the range of ex¬
pected losses for aircraft, spares, and AISs at each base for 10 Monte
Carlo trials of the same attack size and profile. The small square su¬
perimposed on each of the vertical lines represents the average for the
10 trials. These results suggest that the nature and extent of dam-
Fig. 4.3—Range of Airbase Attack Damage
(10 Monte Carlo trials: same attack size and profile)
^^Dispersal of support resources followed patterns developed in USAFE’s Salty
Demo Exercise, an airbase survivability test held at Spangdahlem AB, Germany, in
age induced by airbase attack, even when the attack size, profile, and
targets (i.e., aiming points) are specified, are very unpredictable.
Figure 4.4 shows the same data as 4.3 but also shows the outcome of
one particular replication. Note that at Lahr there was a high proba¬
bility that the AISs would be damaged, while there was a low proba¬
bility that spares would be lost.’^® Sollingen, on the other hand, had a
low probability of losing its AIS, and a high probability of losing
spares. The implications are important. Although the logistics sys¬
tem attempts to provide a balanced mix of resources among bases at
the beginning of a conflict, after such an attack there is likely to be an
imbalance of resources across bases—that is, some bases may be rela¬
tively rich in a particular resource while others may face a paucity of
Each dashed line represents a single replication.
Fig. 4.4—Outcomes of Two Particular Replications
^^For aircraft and spares, losses in a single replication either occur or do not occur.
The number that occur is represented as a proportion of the total and is viewed as an
estimate of the probability of loss of individual assets. The AISs were treated dif¬
ferently here. They were enclosed in shelters, and if a bomb fell within the shelter, it
was assigned a probability of destroying each AIS according to the locus of the hit.
it. Such imbalances diminish system performance and reduce sus¬
Figure 4.5 shows the performance of the current system and the
CLOUT initiatives with no battle damage in one case and, in the
other, with the expected damage from the attack series shown in Fig.
4.4. The graph portrays the results for the loss of two AISs, one at
Lahr and one at Sollingen, and results for the loss of one AIS, at
Lahr. In the case where two AISs are lost and the CLOUT initiatives
are in place, Sollingen is supported by lateral repair at Bitburg, while
Lahr must live with its one surviving AIS. In the case of one AIS loss
at Lahr, it must live with its one remaining AIS. CLOUT provides for
lateral resupply. No lost resources are replaced from CONUS for the
first 30 days. The leftmost bar in Fig. 4.5 shows that with no attack
and wartime demand uncertainties, more than 20 percent of the air¬
craft will be NFMC due to shortages of avionic parts. Note that F-15s
Current System CLOUT Initiatives
No With No With
attack airbase attack airbase
Fig. 4.5—CLOUT Payoffs with Base Damage
(F-15s: 72-aircraft MOB, 48-aircraft COB, 24-aircraft COB)
perform better than the F-16s because they have priority intermedi¬
ate-level repair available to absorb the demand uncertainties. The
second bar shows the effect of battle damage with no CLOUT initia¬
tives in place: almost 40 percent of the aircraft are down by day 30 if
two of the five AISs are lost.
With the CLOUT initiatives there is significant improvement even in
the no-damage case, a reduction from more than 20 percent to about
15 percent NFMC without depot support. With responsive depot sup¬
port we would expect only 5 percent down. The depot makes more of
a difference in the F-15 case because the F-15 has a higher not-re-
pairable-this-station (NETS) rate than the F-16. F-15 units are also
more dependent in wartime on the AISs for repair of LRUs because
every unit deploys with at least one AIS and a remove-repair-and-re-
place (RRR) WRSK, which has fewer LRUs and more repair parts
than the F-16 WRSK. As a result, the depot resupply of critical test
equipment parts has a larger payoff in the F-15 case.
With the CLOUT initiatives in place in the face of battle damage, the
expected NFMC rate drops from almost 40 percent to about 30 per¬
cent if two AISs are lost. If a responsive depot system is in place as
well, the NFMC rate on day 30 is reduced further to about 20 percent
in +he case of two AISs lost. This is about the best the system can be
expected to do with three surviving AISs, because the repair capacity
is totally saturated. As in the case of the F-16 MOB, priority repair
makes the most of the surviving AISs. If only one AIS is lost, respon¬
sive depot support will put this three-base complex in a better posi¬
tion even with battle damage than it would have been in with no
damage but no depot support in the first 30 days. Clearly, the
CLOUT initiatives can significantly improve logistics support in the
face of battle damage to support resources.
CLOUT SUPPORT OF ALTERNATIVE BASING OPTIONS
In wartime it may be desirable or necessary for aircraft to operate
from bases other than their home bases. This is another source of un¬
certainty in combat, and the logistics system must be sufficiently
adaptive to provide effective, continuing support to such aircraft.
This adaptability also gives operations the flexibility to exercise al¬
ternative basing options.
There are several reasons why such flexibility might be required.
Examples include reaction to airbase attack, dispersal in anticipation
of attack, better staging for particular missions, the need to form
composite wings, or even political considerations. Aircraft that are
Eurbome when an attack occurs may be forced to recover away from
their home bases. If the aircraft are then unable to return to their
home bases quickly, it might be important for them to be able to fly
combat sorties from the alternative locations. In other cases, run¬
ways could be so severely damaged that aircraft would have to change
operating locations for extended periods even though the support fa¬
cilities at their home bases were still operational.
The requirement to disperse in high-threat environments like NATO
is frequently discussed. The primary motivation is to spread the force
or to move rearward to reduce vulnerability. Such dispersals may be
for relatively short periods or for longer times, with aircraft returning
to home base only for mE^or maintenance. An example of the latter
case would be to disperse the F-15 air defense force by deploying two-
aircraft or four-aircrEift flights to a large number of bases. Temporary
or more permanent basing changes could also be motivated by mis¬
sion requirements such as range-payload requirements, turn rate
considerations, or a threat concentration in a different region.
Temporary or permanent composite wings may also be a desirable
basing option. Again, they could be used to spread the force (of a
particular type of aircraft) to reduce vulnerability or to facilitate at¬
tack packaging (more than one MDS), easing command and control
and coordination problems. An example of the latter that has been
implemented is the F-4G/F-16 operation at Spangdahlem AB in
Germany. There may be other examples in the deep strike area.
The point here is that as the wartime situation unfolds, it may be
necessary for a variety of reasons to support aircraft flying combat
sorties from locations other than their home bases. The logistics sys¬
tem should be flexible and responsive enough to provide Operations
as much basing flexibility as possible. The CLOUT initiatives have
the potential to provide such flexibility.
Figure 4.6 shows a possible dispersal option for the same configura¬
tion of F-16 bases shown in Fig. 4.2 (a 72-aircraft MOB and three 24-
aircraft COBs). In this case, each COB disperses a 12-aircraft unit
with half the spares in its WRSK. The MOB is now supporting seven
bases. The bars on the left reflect the performance of the current sys¬
tem with the original MOB and three COBs; with fault isolation and
wartime demand uncertainty (no battle damage), 40 percent of the
aircraft are NFMC on day 30. With dispersal, almost 50 percent of
the aircrEift are NFMC.
Without CLOUT With CLOUT Initiatives
Fig. 4.6—Payoffs of CLOUT in Dispersed Operation
(F>16s: 72-aircraft MOB, six 12-aircraft dispersed operating bases)
The right-hand side of the figure shows the same cases with the
CLOUT initiatives in place, lateral resupply and repair, and respon¬
sive depot support. The left bar in each pair reflects the CLOUT per¬
formance without dispersal; the right bar shows the CLOUT perfor¬
mance under dispersed operation. Note that dispersal under CLOUT
degrades system performance very little. These data suggest that the
performance of the system with the CLOUT initiatives is very ro¬
bust—i.e., it is not particularly sensitive to basing options.
A SUMMARY OF THE CLOUT ASSESSMENTS
These assessments suggest the magnitude of the effects of wartime
uncertainties on logistics system performance and demonstrate the
potential payoff of a responsive and robust support system repre¬
sented by the CLOUT initiatives. They clearly demonstrate the po¬
tential payoff of CLOUT in making the most of available resources
worldwide in the face of the shortages and maldistribution that are
virtually inevitable in wartime. Although the payoffs of the CLOUT
initiatives have been evaluated and discussed in a wartime context
here, they also help mitigate the effects of asset shortages that arise
These assessments have not been exhaustive, nor were they intended
to be. For example, we did not show the effects of our inability to
forecast wartime flying activities, which are likely to be far more dy¬
namic and unpredictable than current planning assumptions. We as¬
sumed that units deployed with full WRSKs. In reality, they may not
be full for a variety of unpredictable reasons, such as funding con¬
straints, transportation constraints, or longer procurement lead times
than anticipated. In each of these cases we believe that CLOUT
would again show significant payoff.
We summarize the results of the CLOUT assessments as follows:
• Lateral resupply and lateral repair pay off, and theater priority
repair helps mitigate the effects of demand uncertainty even
when repair is saturated.
• Responsive depot resupply pays off. It, too, helps mitigate the
effects of uncertainty in demand for items that are beyond base
repair capability, and it supplies needed test equipment parts.
Indeed, we underestimated the depot payoff in this analysis be¬
cause we did not include responsive resupply of repair parts or
the depot’s ability to absorb base repair overflow and respond to
• CLOUT initiatives better support alternative basing strategies
and reduce the effects of damage and disruption.
discussion of the effectiveness of some of the CLOUT initiatives in mitigating
the effects of asset shortages in peacetime is contained in Abell, John B., and Thomas
F. Lippiatt, Effective Logistics Support in the Face of Peacetime Resource Constraints,
RAND, N-2921-AF, June 1990.
5. SOME EXTENSIONS OF THE CLOUT LOGIC TO
As suggested throughout this report, in decisionmaking about re¬
source allocations or support strategies, the more dependent we are
on accurate forecasting, the more vulnerable our solutions tend to be
to the future evolving in ways that we did not predict. In problems
involving relatively long planning horizons, forecasting tends to dom¬
inate the solution more than in problems involving relatively short
planning horizons. The shorter the planning horizon, the more the
solution is dominated by current circumstances and the less it is dom¬
inated by the need to forecast. When faced with state-of-the-world
uncertainty, or even substantial statistical uncertainty,^ the shorter
we can make the planning horizon, the less vulnerable we are to
events that defeat specific solutions. However, the length of the
planning horizon must be a function of the system’s responsiveness
In order to illustrate how these ideas might be applied, and to raise
additional issues related to their implementation, we discuss in the
paragraphs that follow a specific logistics management problem and a
specific policy analytic study we have undertaken in RAND’s
Resource Management and Systems Acquisition Program. The first
example involves the prioritization of depot component repair and al¬
location of the serviceable assets to bases; the second involves the
formulation of a policy study of the Air Force’s system for estimating
spares and repair requirements.
EXAMPLE 1; PRIORITIZING DEPOT REPAIR AND
ALLOCATING ASSETS TO BASES
In AFLC's current component repair workloading system, the world¬
wide asset position that exists at the end of any particular fiscal
quarter is used by the repair requirements computational system to
estimate the quarterly repair requirement for the fiscal quarter that
begins six months later. Thus, when the repairs actually occur, the
data that were used to estimate the requirements for those repairs
are six to nine months old. Moreover, the computed repair require¬
ment for any particular quarter may be modified by negotiations be-
^Hodges and Pyles, op. cit. The discussion throughout this section draws heavily
from this source.
tween the materiel management and maintenance organizations at
the Air Logistics Center. The product of these negotiations is a set of
quarterly repair goals to which the depot maintenance activity com¬
mits itself, subject to renegotiation during the quarter. The goals are
frequently adjusted, most often for lack of repairable carcasses or lack
of repair parts. But this reflects nothing more than the uncertainty
in the system; reparable generations and demands for repair parts
seldom eventuate as forecast. Thus it is seldom sensible to stay tied
to negotiated quarterly repair goals when we are trying to achieve
specified aircraft availability goals in the face of uncertainty.
Rather than quarterly repair goals, what is needed is a set of esti¬
mates of repair demands that may occur during the quarter, along
with a measured evaluation of shortages that exist in the system at
the start of the quarter, to be used to lay in consumables and allocate
depot repair resources. But they should be viewed simply as esti¬
mates, not goals. The mix of serviceable assets that actually emerges
from the depot as a result of the policies in the current system may be
quite different from what would be most responsive to the current
needs of the force at the time the repairs are being done.
Given the levels of unpredictability that pervade the system, even in
peacetime, the use of such goals seldom makes sense because the as¬
set position evolves so unpredictably. The structure of the process of
determining repair requirements implicitly assumes predictability
that simply doesn’t exist.
The Air Force has the means to implement a much more responsive
system of component repair. It operates a supply transaction report¬
ing system called the Air Force Recoverable Asset Management
System (AFRAMS). AFRAMS supplies transaction data to a standard
AFLC data system designated as D143. Transactions are transmitted
daily from base supply computers worldwide into the central system;
thus, D143 has the capability to provide a very current snapshot of
the worldwide asset position.® D143 makes it feasible to prioritize the
repair of components using data that are current virtually at the time
the repairs are being made. Such an approach helps mitigate the
effects of uncertainty in the evolution of the asset position.
®Aa a practical matter, problems of inaccuracy have been found in the D143 system,
apparently due to human errors in data transmission and, perhaps, to other problems.
In principle, thou^, the system is intended to support item managers and odiers with
a very current and accurate view of the worldwide asset position. This discussion
assumes that the system operates as intended.
In an attempt to take explicit recognition of uncertainty in repair de¬
mands, one might take a somewhat different view of the problem of
prioritizing repairs and allocating the serviceable assets to locations
worldwide. Consider this problem given the objective of achieving the
highest probability of meeting specified aircraft availability goals at
the end of a base-specific planning horizon.® For those bases that
have wartime deployment tasking, we wish to provide spares—in
addition to those required to support peacetime fl 3 dng operations—^to
carry the unit through the first 30 days of war without depot replen¬
ishment. Recall that we pointed out earlier that such a strategy may
not be a desirable one simply because of the difficulty in predicting
what assets will actually be needed in wartime. The Air Force cur¬
rently computes its war reserve spares requirements with such a pol¬
icy. That policy should be reexamined; but if, for the present, we ac¬
cept that policy as a constraint, we must also accept the consequences
of a longer planning horizon and its associated forecasting problems.
The length of the planning horizon in this decision problem equals the
age of the data, an induction lead time at the depot, an average repair
time, and a base-specific order-and-ship time. For those bases with
wartime deployment tasking, the planning horizon will be lengthened
by an additional 30 days to provide for the spares required for
This is precisely the planning problem addressed by DRIVE, the algo¬
rithm mentioned earlier in this report, a prototype of which was im¬
plemented for demonstration purposes at the Ogden Air Logistics
Center; it has been used there to demonstrate the feasibility of priori¬
tizing the repair and allocation of F-16 avionics components using as¬
set data from D143.^ As a practical matter, the peacetime planning
horizon used in the DRIVE prototype at Ogden is 18 days plus a base-
specific order-and-ship time, plus 30 additional days for bases with
wartime deplo 3 Tnent tasking. This results in planning horizons be¬
tween 20 and 50 days long. Thus the DRIVE prototype is undesirably
more vulnerable to uncertainty than it would be if lead times were
dramatically shorter and if very responsive corrective action could be
taken when urgent, unanticipated demands arose.
In determining its asset allocations, DRIVE estimates the expected
number of NRTS actions during the total planning horizon for each
®In the simple illustration used here, we assume that the aircraft availability goals
are 100 percent at all bases. This implies the need to meet all demands during the
^Discussions of this demonstration can be fo\md in Abell et al., and Miller and
Abell, op. cit.
base, using peacetime NRTS rates for the peacetime portion of the
planning horizon and 100 percent NRTS rates for the wartime plan¬
ning horizon (because the F-16 deploys without intermediate-level
avionics maintenance capability). The expected NRTS actions are
then pooled over the entire planning horizon at each base, and
DRIVE prioritizes repairs and allocates assets against the pooled ex¬
pectations using D143 asset data and probability distributions of de¬
mands inferred from peacetime demand data.®
The robustness of DRIVE’S solution to the combat support problem
would be enhanced by substantially shortening the planning horizon
by making the replenishment system dramatically more responsive.
Such an approach would result in specific allocations that would be
less vulnerable to uncertainty simply because, when an unanticipated
urgent demand arose, the system could respond to it promptly. The
need to forecast over long horizons would be mitigated, and the spe¬
cific allocation solution would be more robust in the face of uncertain
Another part of the original problem that we have not discussed is
that of estimating repair requirements. The forecasting problem
again raises its ugly head. In order to posture itself adequately by
making the proper capital investments, procuring the right mix of
skills, provisioning itself with consumables, and planning for future
workload, the depot repair activity must forecast. Unfortunately, the
lead times involved in many of the decisions faced by depot manage¬
ment imply substantial planning horizons. The planning, program¬
ming, and budgeting system, for example, requires estimates of re¬
source requirements several years into the future. Thus the need to
forecast is often unavoidable. The lesson to be learned from this re¬
search is to couple such forecasts with an execution system that is rele¬
vant, timely, and robust, because we know that forecasts wrongly used
may commit us to specific solutions that are vulnerable to the future
eventuating in ways that our forecasts never suggested. Our hope is
that we will learn to evaluate management alternatives as much in
terms of their flexibility and robustness as we do in terms of their at¬
tractiveness given our forecasts of the future.
^Formulations of the depot repair prioritization problem and asset allocation prob¬
lem were considered in earlier RAND research. See Miller, B., A Real Time METRIC
for the Distribution of Serviceable Assets, RAND, RM-6687-PR, October 1968. See also
Buchanan, A. L., et al., Determining Depot Repair Priorities: Some Informal Notes,
RAND. RM-7904-PR, July 1972.
EXAMPLE 2; A POLICY STUDY OF SPARES AND REPAIR
One importeint study that emerged from the Uncertainty Project is in¬
tended, among other things, to develop an improved approach to es¬
timating spares and depot repair requirements that explicitly recog¬
nizes (a) the role of uncertainty in shaping resource demands and (b)
management as a resource that shapes system performance.® We
discuss the study here because it illustrates a practical application in
policy analysis of the ideas articulated earlier in this report. Its scope
includes primaty operating stock (POS, formerly peacetime operating
stock) as well as war reserve spares kits (WRSK).
The Air Force’s current system for estimating spares and repair re¬
quirements models demands for aircraft spare parts as a steady-state
process. This leads to serious estimating errors, especially in outyear
requirements estimations. Moreover, procurement actions taken in
response to changing values of item pipelines over time tend to induce
long supply, i.e., an overabundance of assets in the system. The diffi¬
culty of estimating outyear requirements correctly is compounded by
the long planning horizons involved, often a few years in the case of
spares procurement actions, as well as the fact that there is instabil¬
ity in our estimates of item characteristics and our perceptions of
what particular kinds of assets will be in the inventory system in the
future. The database that supports the Air Force’s estimates of
spares and repair requirements reflects this instability; it changes
substantially from year to year. We call the sum total of all of these
changes churn. This variability is simply another manifestation of
uncertainty. Churn costs the Air Force money because it induces the
need for additional investments to maintain a specified level of sys¬
tem performance. Thus two of the principal thrusts of the study are
to develop and demonstrate (a) improved approaches to modeling de¬
mands and (b) effective strategies to hedge against the effects of
Another important characteristic of the current system is that it ig¬
nores the contributions of several management adaptations to system
performance in peacetime and wartime. Such adaptations include
cannibalization, expedited transportation, priority repair, with¬
drawals of assets from war reserves during peacetime, and similar
®The project, “Enhancing the Logistics Requirements Estimation Process,” RCN
3738, et\joy8 the joint sponsorship of AF/LEX, AFLC/MM, and AFLC/XP. The Director
of Maintenance Policy, Office of the Assistant Secretary of Defense (Production and
Logistics), provided additional funding for the study because of its implications for the
requirements estimation process in the other Military Departments.
actions taken in response to urgent, unanticipated demands. The
third principal thrust of this policy study, the one we will discuss at
greater length here, is to develop an improved approach to estimating
spares and repair requirements that will (a) account explicitly for re¬
alistic levels of statistical and state-of-the-world uncertainty in shap¬
ing spares demands, and (b) model the effects of management adapta¬
tions on system performance.
The current requirements system makes many assumptions in trying
to represent the logistics system and the environment in which it will
operate in the future, assumptions of the variety enumerated earlier.
Often, these assumptions are clearly inconsistent with behavior in the
logistics management system. Most of the assumptions are conserva¬
tive in the sense that they induce the procurement of more spares
than might otherwise be procured. Moreover, the requirements sys¬
tem ignores several management adaptations that act to improve lo¬
gistics system performance in the face of uncertainty:
• Consolidating asset shortages into the least number of next
higher assemblies (cannibalization).
• Priority repair or expedited repair.
• Expedited transportation and handling.
• Withdrawals of assets from WRSKs.
• The availability of POS assets for use in wartime deployments.
• Lateral supply.
On the other hand, the system also makes assumptions that are
counterconservative. An example is the assumption that the only
bases to which stock levels need to be allocated are those that have
experienced two or more demands in the previous 12 months It is
impossible to model accurately a system as complex as the -''ar Force
logistics management system. Thus many of its features are rot ex¬
plicitly modeled in the requirements computation, or not modeled re¬
alistically. An important task in this study is to determine what
management adaptations should be incorporated in the requirements
model, how they should be modeled, and to what extent they should
shape the determination of requirements.
The design of this study is intended to replicate the Air Force’s spares
and repair requirements computation (1)041), central stock leveling
system (D028), and WRSK/BLSS requirements computation and
provide the resulting asset position to a capability assessment model,
Dyna-METRIC Version 6. Dyna-METRIC will evaluate the peace¬
time and wartime performance of the stockage posture in terms of the
aircraft availability it delivers. Based on this performance, we will
change the assumptions about management adaptations that are
made in the current system, and evaluate the new stockage posture
that was computed after the changes.
Figure 5.1 illustrates this strategy graphically. Dyna-METRIC
Version 6 is a simulation model that incorporates management adap¬
tations and their effects on system performance. Feedback from the
evaluations will help us estimate the effects of including specific
combinations of management adaptations in the requirements sys¬
tem. Our estimates of system performance coupled with the bud¬
getary implications of alternative models of demand uncertainty and
management adaptations will suggest which combination seems to
'AFLC's Weapon System Management Information System
Fig. 6.1—Graphic Portrayal of Study Design
yield the least-cost investment mix for an acceptable level of system
But what about uncertainty? In an effort to explicate our uncertain¬
ties about peacetime and wartime scenarios, the design specifies the
use of several alternative scenarios in the evaluations. Moreover, we
will subject the asset position to alternative demand streams even in
peacetime. We hope to achievo sufficient richness in scenario varia¬
tion and demand-stream characteristics so that we will be able to
evaluate the robustness of the stockage posture as well as its perfor¬
mance in a specific scenario. For each management adaptation we
build into the requirements model, we will specify a scenario that is
intended to defeat the payoffs of that adaptation. For example, we
will evaluate the stockage posture computed with a cannibalization
assumption with a scenario in which the force is dispersed, thus in¬
hibiting the payoff from cannibalization. We will evaluate the stock-
age posture based on a responsive transportation system with a sce¬
nario in which transportation times are lengthened owing to enemy
actions and system disruptions. In evaluating the stockage posture in
peacetime, we will use demand data from a database three years
away from the database used to compute the requirements.
We are hopeful that this approach takes a more realistic view of both
uncertainties and management adaptations than does the current
system. Its design is an attempt to apply the thinking underlying the
CLOUT initiatives to a practical policy study.
6. CONCLUDING REMARKS AND
The CLOUT initiatives are examples of management adaptations
that enhance the performance of the logistics system in peacetime
and wartime. As we have shown, they help mitigate the effects of
uncertainties. To the extent that we ignore the statistical and, espe¬
cially, state-of-the-world uncertainties in logistics planning, par¬
ticularly for wartime, we are vulnerable to events unfolding in ways
that defeat specific solutions. To the extent that we take explicit and
realistic account in planning of our uncertainties and the effects of
management adaptations in overcoming them, we will be better able to
develop solutions whose performance is robust in the face of uncertain
futures. This is the fundamental message of this work. It is a
message that applies to broad categories of management deci¬
sionmaking and policy analysis. It is an important message for Air
Force logisticians and for those involved in logistics management
The Air Force has already incorporated a set of CLOUT-like
initiatives in its new Logistics Concept of Operations. The Major
Commands and the Air Staff are thinking through the imple¬
mentation of such initiatives. In general, the evaluations discussed in
this report and the logic underlying the CLOUT initiatives suggest
the pursuit of additional research and exploration of management
initiatives that will make the logistics system more flexible and
responsive in the face of uncertainty. Depot material support policy,
depot contract repair policy, material processing and handling during
each segment of the depot and base repair pipelines, and exploration
of the tradeoffs between investments that shorten item pipelines and
investments in aircraft recoverable spare parts are topics that are
consistent with the spirit of this work and that seem worthy of
investigation by logistics ar.d financial analysts and managers. The
payoffs of a responsive logistics system are clear. The problem is how
to achieve required levels of performance at reduced costs. That, we
feel, should be the central focus of Air Force logistics research in the
Abell, John B., et al., DRIVE (Distribution and Repair In Variable
Environments): Enhancing the Responsiveness of Depot Repair,
RAND, R-3888-AF (forthcoming).
Abell, John B., and Thomas F. Lippiatt, Effective Logistics Support in
the Face of Peacetime Resource Constraints, RAND, N-2921-AF,
Air Force Logistics Command Regulation 57-18, Operational Re¬
quirements, Management, and Computation of War Reserve
Materiel (WRM), Department of the Air Force, Wright-Patterson
Air Force Base, Ohio, 22 April 1979.
Air Force Regulation 66-267, Maintenance Data Collection System
(MDCS), Department of the Air Force, Washington, D.C.,
August 1984, changed December 1988.
Astrachan, M., and A. S. Cahn, Proceedings of Rand’s Demand
Prediction Conference, January 25-26, 1962, RAND, RM-3358-
PR, January 1963.
Berman, Morton B., et al.. Evaluating the Combat Payoff of
Alternative Logistics Structures for High-Technology Sub¬
systems, RAND, R-3673-A, October 1988.
Brown, B. B., Characteristics of Demand for Aircraft Spare Parts,
RAND, R-292, July 1956.
Crawford, Gordon B., Variability in the Demands for Aircraft Spare
Parts: Its Magnitude and Implications, RAND, R-3318-AF,
Drezner, S. M., and R. J. Hillestad, Logistics Models: Evolutions and
Future Trends, RAND, P-6748, March 1982.
Emerson, D. E., An Introduction to the TSAR Simulation Program:
Model Features and Logic, RAND, R-1284-AF, February 1982.
Emerson, D. E,, TSARINA — User’s Guide to a Computer Model for
Damage Assessment of Complex Airbase Targets, RAND,
N-1460-AF, July 1980.
Emerson, D. E., USAFE Airbase Operations in a Wartime Envi¬
ronment, RAND, P-6810, October 1982.
Hitch, C., Uncertainties in Operations Research, RAND, P-1959, 25
Hodges, James S., Modeling the Demand for Spare Parts: Estimating
the Variance-to-Mean Ratio and Other Issues, RAND, N-2086-
AF, May 1985.
Hodges, James S., with Raymond A. Pyles, Onward Through the Fog:
Uncertainty and Management Adaptations in Systems Analysis
and Design, RAND, R-3760-AF/A/OSD, July 1990.
Isaacson, Karen E., and Patricia Boren, Dyna-METRIC Version 5: A
Capability Assessment Model Including Constrained Repair and
Management Adaptations, RAND, R-3613-AF, August 1988.
Lippiatt, Thomas F., Variability in the Budget Forecasts for Depot-
Level Component Repair, RAND, N-2930-P&L, 1991.
Miller, Louis W., and John B. Abell, DRIVE (Distribution and Repair
In Variable Environments): Design and Operation of the Ogden
Prototype, RAND, R-4158-AF (forthcoming).
Nelson, H. W., and J. W. Petersen, Integrated Supply-Support
Policies: The LP-III Experience, RAND, RM-2839-PR, January
Petersen, J. W., H. W. Nelson, and R. M. Paulson, The Costs and
Benefits of Responsive Support Operations, RAND, RM-2871-PR,
Pipp, Captain D., USAF, “Coronet Warrior—A WRSK Flyout,” Air
Force Journal of Logistics, Summer 1988.
Pyles, R. A., The Dyna-METRIC Readiness Assessment Model:
Motivation, Capabilities, and Use, RAND, R-2886-AF, July 1984.
Pyles, R. A., and Z. Lansdowne, Demand Predictability for Naval
Aviation: Implications for Spare Parts and Depot Capacity,
RAND, R-3757-A (forthcoming).
Rich, M. D., I. K. Cohen, and R. A. Pyles, Recent Progress in Assessing
the Readiness an-’ Sustainability of Combat Forces, RAND,
R.3475-AF, October 1987.
Rich, M. D., W. Stanley, and S. Anderson, Improving U.S. Air Force
Readiness and Sustainability, RAND, R-3113/1-AF, April 1984.
Stockfisch, J. A., Linking Logistics and Operations: A Case Study of
"World War II Air Power, RAND, N-3200-AF, 1991.
Trainor, Colonel Richard F., CPL, USAF, ‘The Evolution of an Air
Force Logistics Concept of Operations,” Air Force Journal of
Logistics, Vol. XII, No. 1,1988, pp. 1-4.
Tripp, Robert S., Morton B. Berman, and Christopher L. Tsai, The
Concept of Operations for a U.S. Army Combat-Oriented
Logistics Execution System with VISION (Visibility of Support
Options), RAND, R-3702-A, March 1990.
Tsai, Christopher L., Dyna-SCORE: Dynamic Simulation of Con¬
strained Repair, RAND, R-3637, July 1989.
Viccellio, H., Maj. Gen., USAF, “Coronet Warrior II,” briefing given to
the Air Force Logistics Conference, 1988.