~160-
Defense Program Lifecycle Management:
A Dynamic Model for Policy Analysis
by
Jack B. Homer! and Ivan Somers2
1University of Southern California
2Hughes Aircraft Company
-l61-
Defense Program Lifecycle Management:
A Dynamic Model for Policy Analysis
ABSTRACT
The System Dynamics Lab at the University of Southern
California's (USC) has worked with Hughes Aircraft
Company's Electro-Optical and Data Systems Group to
develop a system dynamics model for analyzing alternative
policies available to a defense contractor for managing the
production program lifecycle. Program lifecycle
management is of prime importance to firms, like Hughes,
that design, manufacture, and maintain complex military
equipment. These firms have come under increasing
government scrutiny and control, Particularly with regard
to cost and schedule risks.
The USC-Hughes model addresses the notion that cost and
schedule risks can be substantially reduced through
improved program management, even in the face of
possible hurdles thrown up by customers and suppliers.
The model suggests, for example, that overruns,
particularly cost overruns, may be significantly’reduced--
without adversely affecting product quality--by carefully
limiting the number and type of discretionary mid-
production design improvements.
This presentation outlines the background and basic
structural elements of the USC-Hughes model,
demonstrates the model's ability to track historical data
from two different cases, and highlights some of the policy
findings that have emerged from the model.
-162-
Background
Client Motivation
¢ Project initiated in 1985 to analyze over-time impact of
design changes on cost and schedule risk.
¢ Concerns over cost amplified in an environment of
reduced Congressional budgets.
¢ Schedule slippages a chronic problem hurting customer
relations and jeopardizing contracts.
¢ Overruns recognized as a dynamic problem: result of
inefficiencies accumulating over entire program
lifecycle of engineering design, manufacturing, and
integrated logistics support (ILS).
Model Scope
¢ Model focuses on manufacturing and ILS phases of
lifecycle. In particular, what happens when design
imperfections are discovered mid-production?
¢ Significant design imperfections common in concurrent
production programs: Government often requires
manufacturing phase begin before design fully debugged
and tested.
¢ Model focuses on flows of parts, assembled units, and
engineering changes. Other factors (labor, equipment,
cash, etc.) assumed available and non-constraining.
~163-
Overview of Model Structure
Hughes/EDSG >
|eore
ngineeri
ECRs(3) ECRs(4
| -Design maturity et Customer
-Design changes
drawings
ECRs(2) 3S deviation/
waiver
requests
Manufacturing
deviations/
-Scheduling waivers
-Incorporation «
of changes l
-Part procurement
& fabrication part
-Unit assembly, ee
test & rework Vendors
<—____-_____—
parts
spare part units,
orders spare parts
ILS
-Field repair
& readiness
Sources of ECRs (Engineering Change Requests)
(1) Engineering analysis revealing design problems
(2) Test yield and producibility problems
(3) Field reliability problems
(4) New customer performance requirements
-164-
Case Studies and Policy Issues
Two Case Studies
¢ Model initially developed and calibrated to represent one
particularly troubled program still in production.
* Model "revalidation" involved recalibration to represent
a different production program already completed.
Central Policy Issues
¢ Mid-production engineering change requests (ECRs):
Most ECRs are internally generated rather than
customer-directed. Should such discretionary ECRs be
terminated altogether at some point during production?
Should each be accepted only if its expected impact
exceeds some minimum amount? Should they be
grouped together and released in blocks rather than in a
continuous stream?
¢ Disposition of old-version parts:
When new-version parts arrive to the factory, it is a
common practice to "purge" (discard) old-version parts
from raw inventory and work-in-process. Indeed, not to
do so may be considered a "deviation" from plans to roll
new-version parts into production quickly. Should this
practice of purging be modified? ,
¢ Ordering of parts:
Because of uncertainty in vendor delivery times, it is a
common practice to order parts so that most arrive well
before they are needed for assembly. This is known as
"front-loading" the parts delivery schedule. Should this
practice be modified?
Production Units
Production Units
-165-
CUMULATIVE SHIPMENTS, CASE 1
2500
2000 4
1500 ~
1000 4
Simulated
Historical
2.3 Month Avg. Delinquency
24 36 48 60 72 84 96
CUMULATIVE SHIPMENTS, CASE 2
2500
2000 +
1500 4
1000 4
500-7
Historical
24 36 48 60 72 84 96
Drawing Changes
Drawing Changes
~166-
CUM. ENGINEERING CHANGE REQUESTS, CASE 1
6000
5000 +
4000 4
Historical
Simulated
6.5 ECRs/Drawing on Avg.
Planned
T 7 T T
12 24 36 48 60 72 84 96
Month
3000
2500 4
2000 4
1500 4
1000
Historical c 4.3 ECRs/Drawing on Avg.
Planned
12 24 36 48 60 72 84 96
Constant Dollars
Constant Dollars
-167-
CUMULATIVE PROGRAM COSTS, CASE 1
4,00+8
3.0€+8 7
2.0e+8 4
1.0648
39% Overrun
Simulated
\
Planned
12 24 36 48 60 72 84 96
CUMULATIVE PROGRAM COSTS, CASE 2
2.0e+8
1.5e+8 5
1.0e+8 4
5.0e+7 5
18% Overrun
Simulated
\
Planned
-168-
Causal Structure Behind Problems
of Cost and Schedule
Customer
“we - OX.
Maturity Nw’ ON S ‘pesion
of Design * Changes
. |
a Purging of
Aacepiabilty ——_ 01¢-version
Parts
Unit Part Familiarity
Rework & Shortages with Parts
Repair
vA
+ Assembly Labor
Productivi
Parts “* ”
Ordered
| -
COST nell
i SCHEDULE
OVERRUN Retesting >> SLIPPAGE
-169~
Results and Recommendations
¢ Purging of old-version parts can generate severe parts
shortages largely responsible for cost and schedule
problems. It also disrupts assembly. Purging actually
slows the rate at which acceptable units are produced,
the opposite of its intended effect.
Should eliminate part purging, except where part
changes are customer-directed or "must-do" (i.e., non-
discretionary). Reach understanding with customers on
this policy so that use of old-version parts is not
generally considered a deviation.
* Marginal benefit of ECRs decreases as design improves,
but marginal cost remains the same. Also, marginal net
benefit is greater early during manufacturing phase,
before assembly comes up to speed. So, early
termination of discretionary ECRs can be significant
cost saver, also may reduce schedule slippage.
Filtering-out of less significant ECRs followed by
outright termination of all discretionary ECRs can cut
costs more than termination alone, but the additional
savings are relatively small, and appropriate
implementation is tricky. The benefits of filtering may
not be worth the extra effort and uncertainty involved.
Block release of mid-production design changes delays
their benefits and increases the disruption they cause.
They should be released without blocking.
-170-
Results and Recommendations (continued)
* Front-loading of the parts delivery schedule builds up
raw parts stocks and disguises the extra attrition caused
by unexpected rework and repair (plus any purging).
This can lead to the factory being caught short of parts
at end-of-contract should the production program be
temporarily or permanently discontinued. This may
result in much additional schedule slippage. Should
eliminate the practice of front-loading.
* When tested independently, "no part purging" and "ECR
termination" policies each led to major theoretical
reductions in cost and schedule overruns for the two
historical cases.
But "no part purging” had the greater impact, and
combination of the two policies led to relatively minor
additional cost improvement (and no additional schedule
improvement) over "no part purging" alone. Indeed, the
beneficial impacts of "ECR termination" taken alone are
largely due to its indirect elimination of much purging.
« Whether tested independently or together with other
policies, "no front-loading" policy led to sizeable
theoretical reductions in schedule slippage for the two
historical cases.
CC BY-NC-SA 4.0