Notes on 
	Advanced Computer-Based Manufacturing Systems for Defense Needs

Robert H. Anderson
Nake M. Kamrany
ISI/RR-73-10


This study is addressed to the question of how such systems can benefit
DOD procurement.  If these systems were shown to sufficiently benefit
DOD procurement, they might be financed by DOD, independent of broader
support for civilian production.  A precedent is the DOD support of
numerically-controlled machine tools, which could be supported by DOD
since the new technology was essential to procurement plans.

Some observations and conclusions about DOD procurement
1. The purchasing power of DOD procurement is down 26% in the last
	decade.
	Conclusion: need for productivity increase
2. DOD procurement is characterized by small quantities of sophisticated
	elements in batch production.  50% in lots <100.
	Conclusion: concentrate on batch production
3.  DoD procurement is heavily concentrated in a few major categories.
	Conclusion: concentrate effort in aerospace and electronics.
4. DoD proceurement dominated by a few firms.
	Conclusion: DoD has substantial leverage.
7. DoD-related industries have $1 billion yearly inventory costs.
	Conclusion: big savings possible in inventory.

Some further observations:
DoD appetite for electronics is constant volume, not constant function.
When LSI simplifies one function, it will be incorporated into a system
now feasible with more functions.  Thus electronics assembly will not
be eliminated for DoD procurement.

Case Studies
TOW Missile Production, Hughes Aircraft
Western Gear Corporation
Numerical Control Facility, Douglas Aircraft

TOW missile
Over 30,000 missiles have been produced.  Many engineering changes are
made; complicated process of approval.  Unit price at current rate is
$2200 now, formerly above $3500.  $150 assembly labor, or 22% of value-
added.  If total computer-aided manufacturing were adopted, without
being adopted by subcontractors, manufacturing costs would be cut by
34% (estimate), or by 57% if subcontractors were automated.

Western Gear
Simplest gearbox, $2000 in lots of 150 for C-130 (approx 1000 made).
This operates as a job shop: 3000 orders in process.  Management and
support services were 100% of production labor; ie for each worker,
there is another person keeping him working usefully.  Labor costs
are high in the gearbox studied (C-130 gearbox in lots of 150) and
assembly requires human ingenuity to maintain the tolerances.  
We conclude that assembly might not be automatable now.  Information
systems to cut overhead would be valuable.  Such a system exists but
is unusable: changes are difficult, no exceptions are indicated,
output is unusable.


N/C Douglas Aircraft
600 support people for 300 in N/C fabrication.  Manual real-time information

system was instituted when it was impossible to get satisfaction from computer
department.  Manual system brought machine utilization from 25% to 57%.

Recommendations
1.  A manager-oriented Manufacturing Process Control system be evolved.
	In batch production environment, real-time systems for job and
	parts control have a high priority.  At present, software systems
	development seems inadequate in industrial computer service depts.
		too costly
		too inflexible
		not related/not sensitive to needs
2.  Research be supported for programmable automation technology
	programmable assembly
	material transfer
	machine loading



Comments  TOB
While the report did not find significant applications for automated 
assembly, this should not be regarded as a conclusive result showing
that automated assembly is not applicable to industrial production.
1.  The report samples only three systems.  It did not consider
airframe production which is a very large, compact (geographically),
and uniform (high commonality) application for programmable assembly.
2.  The same conclusions do not hold for civilian industry.
3.  The team did not devote a large effort to finding such applications,
and did not make use of talents of experts in programmable automation.