Академический Документы
Профессиональный Документы
Культура Документы
NALLAN C.SURESH*
JACK R. MEREDITH*
EXECUTIVE SUMMARY
This article illustrates the use of Group Technology (CT) principles for integrating the various
elements of Computer Integrated Manufacturing (CIM). Several observations are made regarding
the current state of batch manufacturing in the U.S. in light of these principles. Following a brief
historical background, the various elements of CT are described as they occur in the manufacturing
cycle. Since the main prerequisite to CT is part family identification, the various classification and
coding systems for identifying part families, and also non-codification systems, are briefly described.
The benefits of design rationalization and variety reduction are then explained in the context of
CAD, CAM, and CAPP. Next, cellular manufacturing, currently the major problem area in the
U.S. regarding GT, is discussed. The discussion includes new production technologies and concepts
such as economies of scope. Materials management and operation scheduling are discussed next,
highlighting the GT/MRP interface. The effect of CT and other new technologies on quality is
then addressed and the significant impacts here are noted. Next, the effects on related areas such
as personnel and accounting are described, including worker satisfaction, incentive schemes, and
cost tracking.
INTRODUCTION
Figure 1 gives a schematic of these areas, which are discussed in detail in the following
sections.
152 APES
FIGURE 1
COMPONENT 1
CLIENT 2
CDlPONENT 3
COMPONENT 4
CLONIC 5
COMPONENT 6
COMPONENT 7
CORONET 8
COMPONENT 9 t I
UMPONENT 10 I 1
COMPONENT 1
COMPONENT 3
C~PONE~ 7
COMPONENT 2
ATONES 6 PART FAMILY 2 ECONWI ES OF SCOPE
CCWONENT 8
COMPONENT 4
C~PONENT 5
COMPONENT 9
COMPONENT 10
and coding methods currently in use. Numerous coding systems exist: Brisch (U.K.),
Opitz (West Germany), MICLASS (TNO, Netherlands), CODE (MD%, U.S.), and SAGT,
to name a few. These coding systems are often classified into “monocodes” (hierarchicaf),
or “polycodes” (faceted). But this classification is only of academic interest since almost
all codes are a combination of the two. The US Air Force is currently attempting to
develop a generic classification and coding system, called GTCC.
Codes can also be classified as emphasizing either design aspects or process features.
But codes such as MICLASS/MULTICLASS, which is probably the most popular in the
154 APICS
U.S., attempt to incorporate both design and manufacturing aspects and for this reason
are quite long, running more than 30 digits. This code, developed by TN0 of Holland,
is now marketed by the Organization for Industrial Research, located in Boston.
Classification and coding involve a lot of preliminary groundwork and, for this reason,
have not been very popular. But this investment can be partially recovered when a
company adds a CAD system which requires the data to be keyed into the data base in
any case. Non-codification methods, such as PFA [3], involve a matrix manipulation for
forming cells of machines and the part families to be machined in them. The major
consideration behind the choice of the method for identifying part families is: How fast
must the payoff be? This assumes, of course, that the technical factors, such as the length
of the code, its manufacturing applicability, its ease of computerization, and so on, have
all been addressed.
Errors during codification are a major problem and semiautomated approaches are
currently being developed. In many ways, this is similar to the creation and maintenance
of data dictionaries in EDP installations.
Process Planning
Group technology principles are utilized in Computer Aided Process Planning (CAPP)
under two types of implementation: the Generative Method and the Variant Method. In
the Generative Method, the process planning logic is stored in the system as a knowledge
base, together with the various algorithms that define the technical decisions. This is
similar to a Decision Support System. Then, taking into account the production processes
available within the firm, every new drawing is analyzed from its fundamentals in order
to arrive at a process plan.
Obviously, this is a very difficult system to realize. Computer Aided Manufacturing-
International (CAM-I) has developed CAPP software that has been implemented at firms
such as Lockheed-Georgia. One well-known system, GENPLAN, is not a truly Generative
system but comes closer than most other systems.
The Variant Method, on the other hand, is easier to implement and more popular. It
involves using the code to retrieve the process plan of a similar part designed earlier and
then making a minimal amount of modification to it.
The benefits of CAPP, identified in Figure 4, include lower process planning lead
NUMBER OF
SHAF’ES
0 1 2 3 4 5 6 7 o
Number of Heeks of Engineering Design With Ciy
times, greater accuracy and consistency, routing standardization, and above all, considerable
savings in tooling investments. The same CAPP principles are used in tool design and
tooling standardization. The impact of CAPP is also felt on the shop floor when the
routings become more standardized. Figure 5 illustrates these advantages for the case of
General Dynamics, where the routings became considerably simpler after CAPP imple-
mentation.
Figure 6 depicts GT as the interface between the design engineering and process
planning stages, a potential solution to the CAD/CAM interface problem.
Cellular Manufacturing
The benefits of cellular manufacturing are the major reasons for adopting GT. This
aspect, therefore, was the first to be implemented in practice. The “group layout,” as it
came to be called, offers tremendous advantages. In job shops, the layout of the machines
is traditionally on the basis of process specialization, i.e., similar machines are kept
together. But in a group layout it is based on the products, better known as “part-family”
specialization. As noted by Burbidge [5], this is perhaps the most radical change in batch
manufacturing since the days of the Industrial Revolution. This difference is illustrated
in Figure 7.
At first sight, part-family layouts might seem inappropriate in intermittent and one-of-
a-kind situations. But on reflection, it will be found that true, one-of-a-kind situations
156 APICS
FIGURE 4
Impact of CAPP
-2 -1 0 1 2
PRODUCTIONLEAD TIME
PROCESS PLANNING
LEADTIME
I%~H~NE L~II_IZATION
PRODUCTQUALITY
DIRECTLAWRUTILIZATION
UNIFORMITYOF PROCESSPLAN:
COSTESTIMATING PROCEDURES
kw~/Bu~ DECISIONS 1
PRODWTSTANDARDIZATION
CRITICAL/NOR SKILLS
~TERIAL STANDARDIZATION
PRODUCIBILITVOF PARTS
PLANT LAYOUT
MATERIALS HANDLING
PROWTI~N SCHEDULING
CAPACITY
PLANNING
-
are extremely rare. The use of group layouts has been found to be applicable even in the
shipbuilding industry in the U.K. [8].
The difficulty of redesigning existing job shops for cellular manufacturing seems to be
stalling the implementation of this aspect of GT in the U.S. In many ways this is similar
to the predicament faced by many British companies during the 1960s. Almost all the
techniques for designing cells have come from the UK, but the creation of cells has
always been more of an art. There is a noticeable lack of consulting expertise in the U.S.
in regard to this, when one compares the situation in England where pioneers like
Burbidge and several others assisted numerous companies in changing their layouts.
158 APICS
FIGURE 6
/G / MANUFACTURING
FIGURE 7
‘-I---
l
A
on
I II_
TURNING II GRINDING
a FINE BORING
160 APES
FIGURE 8
Throughput Time with Functional Layout: 8 Weeks
No, CENTER 1 2 3 4 5 6 7 8
I
1 TURNING Q IOIT]
H
2 MILLING 1 Q 101 T1
3 DRILLING I Q lolq
4 DEEIJRR ,lIIqlql
L I I I I I I
E 3
4
2
GROUP NO,
only marginal but also difficult to measure. Simply by creating a cell that produces some
part families at a faster rate while the rest of the shop follows conventional job shop
procedures creates imbalances of marginal benefit.
The justification problems should thus not be too surprising. On the other hand,
creation of semiautomated cells, even without a formal codification system, has in many
Given the short cycle operations found in group layouts, what type of materials
management systems should be employed? In the U.K., one of the first effects to be
noted was the inapplicability of classical methods such as reorder points (“stock control”
in British terminology). This led to the development of MRP-type methods involving the
explosion of a master schedule using the bills of material. This roughly coincided with
the MRP movement spearheaded by the computer industry and APICS (American
Production and Inventory Control Society) in the U.S.
In the U.K. this method was referred to as the “flow control” method. Later, this was
refined to the “period batch control” (PBC) method. It has been shown [ 15, 171 that the
PBC is simply an MRP system, but with small bucket sizes and lot-for-lot ordering. Thus,
short cycle flow control was advocated as the most suitable method for managing
materials in this kind of just-in-time manufacture. Companies such as Serck Audco in
the U.K. naturally evolved to this state after implementing GT, but the results in England
never reached what was predicted by the proponents of PBC, such as John Burbidge.
However, the full implementation of these practices did occur in Japan. It is worthwhile
considering whether this has anything to do with the basic management tenets of the
Western world.
There is currently a concern among many American companies as to whether MRP
concepts have, with the advent of GT and the just-in-time philosophy, perhaps become
outdated. But it has been shown [ 15, 171 that an MRP system is not only compatible but
ideal for cellular manufacturing. Also, it is now realized [ 12, 171 that the failure rate of
MRP in functional layouts is essentially due to the inherent complexities of this
environment.
The implementation of MRP in a GT context is generally much easier because of the
following factors [ 17, p. 831:
l Manufacturing lead times are much shorter and more predictable.
l The Period Batch Control (i.e., lot-for-lot) form of ordering, as the Japanese have
adopted, is more appropriate.
l Work flow is more ordered and streamlined than in process layouts, resulting in
easier control with less documentation and expediting requirements.
l Reduced WIP and other inventories contributes to greater accuracy of inventory
records, still a problem with functional layout based MRP systems.
The basic requirements of MRP such as BOM structuring, the realism of the master
schedule, etc. are unaffected. Also, production smoothing plays a minor role in master
scheduling because of reduced changeover costs. And the absence of independent
component lot sizing in the product structure makes capacity requirements planning
much easier.
There is another major difference as well. Priority planning subsequent to order issue
is obviated in group layouts. In functional layouts, with their distinction between operation
priority versus job priority and all their other attendant complexity, it is difficult to
control the operations and the tendency has thus been to computerize this confusion
162 APES
through “shop floor control” (SFC) systems. Once again, before rushing headlong into
computerized approaches, it is wise to get back to basics. In group layouts, SFC is easier
because almost all the operations are carried out in a single cell. Also, there is no
distinction between operation priority and job priority-they are one and the same, with
all the machining done just before actual need. Figure 9 shows the schematic of a GT-
based MRP system.
Economic lot sizing persists in many firms. The pitfalls of economic lot sizing however,
have come to be known very widely:
l In conventional EOQ-type models, as well as those used for dependent demand in
an MRP framework, the setup cost is assumed to be fixed. This is simply invalid in
a GT context, and cellular manufacturing in general. Also, new production technologies
have enabled the quick downloading of programs and tool changeovers which has
drastically reduced setup times. Therefore, EOQs of 1 are not an unrealizable dream
anymore.
l Even with conventional models, it has always been realized that the cost function is
insensitive near the EOQ. Even reducing the quantity by half results in very marginal
cost increases.
l The basic philosophy of item-by-item lot sizing is based on suboptimization. By
calculating the economic quantities on an item-by-item basis imbalances in the
system are created that result in inacceptably high opportunity and other costs.
l The excess inventories created add to quality, storage, obsolescence, and other costs.
Quality Control
In conventional job shops, costing systems are geared to the work center approach.
Every cost center has an overhead rate that is charged to a particular component
separately from the labor rate. But in GT, and even conventional settings, the proportion
of direct labor has decreased due to increasing automation, invalidating the allocations
of overhead based on direct labor. Also, since in cellular manufacturing the cells are the
//
Raw
Imp. Material
Stores EIEI c
F
Suppliers
GT Cells
El
4
\ --- Purchase
-----------7
“Master I
(Source:Reference17)
cost centers, a different costing and management control system is required. This has
been clearly observed in many British firms changing over to GT.
GT also impacts the Personnel/Human Resources area. In GT, operations are largely
conducted in a single cell, which leads to a skill requirement of breadth, rather than
depth. Also, the part-family approach apparently leads to greater job ~tisfa~tion. This
aspect has been noted in Japanese factories by many U.S. writers but, strangely, has not
been traced to cellular manufacturing and the part-family concept. Similarly, for incentive
schemes, group incentives have been found to be preferable to individual incentives,
which should not be surprising.
The U.S. is currently in the peculiar situation of being one of the major bastions of
the functional mode of manufacturing, and at the same time, the center of virtually all
of the newly introduced technologies rooted in part family-oriented manufacturing such
as CAD, CAM, CAPP, and FMS. These new manufacturing technologies are increasingly
pointing out the need for total-systems approaches. ~anufactu~ng cannot be viewed as
a set of isolated subsystems anymore but must be treated as a single sequential set of
integrated activities. For the sake of convenience, using a divide and conquer approach,
we have organizationally and academically fragmented these activities. Even outside the
firm there has been institutional divisionalization. For example the Society of Manufacturing
Engineers focuses on upstream manufacturing functions, while APICS focuses on such
areas as production and inventory functions. This has led to the deveiopment of myopic
principles, practices, “crusades,” and, of course, numerous acronyms. Unfortunately,
most manufacturing firms lack the managerial insight and background to foster practices
based on synthesis and a total systems viewpoint.
Upon reviewing the recent industry crusades, a significant departure from such systems
as MRP and MRP II may be noticed. Current emphasis (e.g., see [lo]) is on zero
invento~es, just-in-time, flexible automation, and so on. These newly endorsed manufac-
turing practices are essentially derived from Japanese manufacturing methods. Unfortu-
nately, they seem to have been somewhat hastily compiled and constitute a perpetuation
of the subsystem approach of the past. They can be very confusing to manufacturing
managers, for several reasons.
For one thing, having been indoctrinated into MRP principles for so long, this apparent
shift in focus is disturbing. Of course, current wisdom states that the classic p~n~iples
are not outdated; the new principles simply have to be added on. Yet there really has
not been a coherent linkage between these new concepts and the ones endorsed up until
now. Also, many of these newly advocated practices do not constitute rational objectives
unless and until we depart from the present functional mode of operation.
Consider, for instance, the directive “develop flexible, multipurpose workers.” This
essentially stems from the fact that this has been observed in Japanese companies. As we
mentioned earlier, in cellular manufacturing scheduling flexibility requires that workers
have a wide breadth of skill because all operations are confined to the cell and there are
invariably more machines than workers. This is a spontaneous result of switching over
to a cellular mode of operation. To practitioners operating in a functional mode, this
directive really does not constitute a rational objective. Similarly, it can be seen that
166 APICS
When considering new technology, managers are advised to step back in time and
perspective, and take a long-range view before plunging into the “automate or evaporate”
impulsiveness so currently in vogue. By adopting simplicity of operation as a goal, and
group technology principles as the philosophy to attain it, factory automation will proceed
more naturally while its benefits accrue to the firm.
REFERENCES
1 APICS Dictionary, American Production and In- Technology,” Uperatians Management Review,
ventory Control Society, 1983. Winter, 1984.
2. Buffa, ES., “Research In Operations Management,” 12. Krajewski, L.J., B.E. King, L.P. Ritzman, and
Jomnal of‘ Operat~uns Managemenf, Vol. 1. No. D.S. Wong, “A Viable U.S. Manufacturing Strategy:
1, 1981. Reshaping the Production Environment,” Upera-
lions Management Review, Spring, 1984.
3 Burbidge, J.L., “Production Flow Analysis,” The
13. Kriegler, A.M., “GT Improves Flow, Cuts Costs,”
Production Engineer, April, 1911.
American Machini.~t, March. 1984.
4. Burbidge, J.L., The Intr~~i~ct~an af‘ Group Tech- 14. Mitrafanov, S.P., “Scientific Principles of Group
noiagy, Heinemann, London, 1975. Technology,” Boston Spa: National Lending Li-
5. Burbidge, J.L.. Introduction to Ist I~~erna~io~~l brary (translated from Nauchnye Osnovy Gruppavui
Conference on Group Technology, 1l.0, Turin, Tekhnolagii, Leningrad, Linizdat, 1966).
Italy, 1969. 15 New, CC., “MRP and GT: A New Strategy for
6. Burbidge. J.L., “The Case Against the Economic ~om~nent Pr~uction,” Prod~ctjan and I~~e~~ory
Batch Quantity,” The Manager, January, 1964. Management. Third Quarter, 1977.
7. Burbidge, J.L., S. Eilon, and W.E. Duckworth, 16 Schonberger, R.J., Japanese Man~~cturing Tech-
“Dragons in pursuit of the EBQ,” Operational niques: Nine Hidden Lessons in simplicity, The
Research Quarterly, Vol. 15, No. 4, 1964. Free Press, 1982.
17. Suresh, NC., “Optimizing Intermittent Production
8. Gallagher, C.C., SK. Banerjee, and G. Southern,
Systems Through Group Technology and an MRP
“Group Technology in the Ship Building Industry,”
System,” Production and Inventary Management,
International Journal &Production Research, Vol.
12, 1974. Fourth Quarter, 1979.
18. Suresh, NC. and J.R. Meredith, “A Generic Ap-
9. Goldhar, J.D. and M. Jelinek, “Plan for Economies
proach to Justifying Flexible Manufacturing Sys-
of Scope,” Harvard Business Review, Nov-Dee,
tems.” Proceedings of ORSA/TIMS First Special
1983.
Interest Conference on FMS, Ann Arbor. August,
IO. Hall, R.W., “Zero Inventory Crusade-Much More 1984.
than Materials Management,” Production and In- 19, Voss, C. A., “New Technology Versus the Lot Size
ventory Management, Third Quarter, 1983. Aigo~~rn industry,” Operations Marzagement Re-
Il. Hyer, N.L., “Management’s Guide to Group view, Summer, 1984.