ABSTRACT
This paper
examines current approaches to the design of cellular manufacturing systems. A
number of significant limitations are identified. These include their focused
nature, their inadequate consideration of performance evaluation, their partial
consideration of manufacturing systems and their lack of integration in the
strategic decision making process. The creation of a Whole Business Simulator
(WBS) and its integration with an appropriate design methodology is proposed as
an alternative holistic approach. WBS is based on the use of a mixture of real company systems and specialist
simulation elements, capable of comprehensively evaluating manufacturing
systems design decisions.
1. INTRODUCTION:
The
manufacturing function is key to the competitive challenge that many businesses
face in the industrial market places of today.It is stated that the way for
manufacturing systems to provide the competitive advantage necessary, is for
them to be designed to meet market requirements. All too often manufacturing
facilities have not been designed but have developed in an incremental
piecemeal fashion, which has led to them being highly complex, fragmented and
difficult to manage. It should also be remembered that decisions regarding the
design of manufacturing systems influence the total operating costs incurred by
the manufacturing facility during its life.
Many have
argued for some time that manufacturing decisions directly affect corporate
performance. They also assert that a competitive manufacturing operation is
more than the achievement of high efficiency and low costs
: ‘A company’s manufacturing
function is typically either a competitive weapon or a corporate millstone. It
is seldom neutral.’
Some also
point out that the ‘secret weapon’ of the worlds best competition is a superior
manufacturing system. Therefore, it is important for many businesses to examine
their manufacturing systems. In order to compete effectively with the worlds
best competition, and improve on poor
performance, manufacturing systems need to be redesigned to meet the
competitive challenge that now faces them.
2:
CURRENT APPROACHES TO THE DESIGN OF CELLULAR MANUFACTURING SYSTEMS
Approaches
to the design of cellular manufacturing systems appear to fall into one of four
broad categories.
Category 1: Design Techniques & Procedures
Considerable research has been
undertaken in the area of manufacturing cell formation. The process of defining
cells involves determining what separate facilities are required to manufacture
a specific range of components or products. The techniques and procedures may
be classified into different types as discussed below.
Machine Grouping Techniques: These techniques group machines together into cells. Parts then have to
be assigned to the machine groups that have been defined. The techniques that
fall into this type are typically based on the use of similarity coefficients.
The focus of much of the research in this area has generally
been limited to the mathematics of the approach.
Part
Family Grouping Techniques:
These techniques are concerned
with grouping parts into families.
Machines then have to be allocated to the manufacture of particular
families of parts. Classification and
Coding is most widely used.
Machine-Part Grouping Techniques:
The group of techniques that
fall into this type attempt to form groups of machines and families of parts
simultaneously. Burbidge was the pioneer of the work with the development of
Production Flow Analysis (PFA). Other research in this area includes with the
development of Rank Order Clustering (ROC).
The area of cell formation has received
a significant amount of attention as discussed above. The key point to note
however, is that the attention is of a narrow nature, stopping well short of
what might be described as attempts to address the design of cellular
manufacturing systems.
Category
2: Application of Systematic Design Approaches
Recognising
that the techniques discussed above are not in themselves an adequate approach
to the design of cellular manufacturing systems, some authors have published
approaches that attempt to be more complete. One of the earliest papers
outlining a systematic approach to the design of cellular manufacturing systems
was by Thornley. The paper outlined an approach that started with data
collection followed by the
definition of manufacturing cells, which were
then analysed for load. The layout of the facility was then determined and an
assessment of economic savings and other benefits made. Others who have
proposed systematic design approaches include Scott .
Although
the approaches published are systematic, they are deficient in a number of
respects. For example, design involving the use of static load calculations
only is the norm with limited study of dynamic behavior. Material control
systems are usually only treated in a tangential manner rather than being
treated as an integral part of the design process. In addition, none of the
approaches are, or profess to be, general methodologies appropriate for the
design of cellular manufacturing systems.
Category 3:
Integrated Modelling
This approach to the design of cellular
manufacturing systems is based on the integration of four types of software,
used sequentially for manufacturing systems modelling. They are:
1] Lotus
1-2-3 (spreadsheet)
2]
Analytical modelling
3] Discrete
event simulation and
4] Cinema
(animation) , as the set of computer tools for the modelling and analysis of
manufacturing systems. Although much effort appears to have been spent on
ensuring that data can be passed from one modelling tool to another (e.g. the
development of ‘Simstarter’ to allow the conversion of analytical models into
simulation code), the tools have not been integrated into a design methodology
for manufacturing systems. Little indication of the manufacturing system design methodology
necessary to effectively utilise these computer tools is explored.
It was to
some extent tried to remedy the lack of an explicit documented design
methodology, by specifying design activities that should be undertaken when
using each computer tool in sequence. So for example, Lotus 1-2-3 is specified
for use in the ‘Initial Design’ (basic system parameter design) and Manuplan II
(analytical modelling) for ‘Rough Cut
Design’ (initial analysis of system dynamics).
This
approach has the strength of providing a means for both a static and dynamic
analysis of a manufacturing system, and providing sonic guidance as to how
computer modelling tools may be used in the design of manufacturing systems.
However, a number of weaknesses can be identified. The application of the
approach is acknowledged as being
limited in the context of manufacturing systems design, focussing on the selection and layout of the direct
production equipment together with its associated operating parameters.
In addition, the design methodology
that is specified is limited in its detail and application. Finally, the
objectives against which manufacturing systems designed are judged are somewhat
limited when viewed in a business context. For instance, with respect to a
maximum implementation cost, maximum
machine uti1isation and a maximum lead time - there is no consideration of either profitability or return on capital
employed.
Category 4: Design Methodology Based on Systems Engineering :
The primary contribution in developing an approach to the design of
manufacturing systems based on systems engineering has come from Parnaby with
support Love & Bridge. The approach advocated by Parnaby takes the format,
at its most basic, of figure 1, and is detailed below.
Data Collection & Analysis: Data
concerning markets, volume and variety of products, factory processes and
component routings is collected. In addition, it is important that business and
manufacturing strategies are examined and understood at this stage.
Cell Definition: The objective is to define a
manufacturing architecture that matches the needs and requirements of the
market and manufacturing strategy.
Usually, manufacturing cells
are grouped around material flows or part similarities. In effect a ‘Concept
Design of the manufacturing facility is generated.
Steady State Design: This stage of the approach is
concerned with designing the system to meet AVERAGE requirements.
The term steady state is used to indicate that
nothing changes with time. Thus,
average demand is used to
establish the first estimates of the type and quantity of resources (people,
machines, materials, tooling etc.) required to meet the demand, utilising
average performances (e.g. for cycle times and change-over times).
Dynamic Design: The design established at the
steady state design step of the methodology is tested against variations from
the average values assumed above. Both the requirements of the manufacturing
system in terms of product variety and volume and internal performance
parameters such as breakdowns are varied to assess their effect and determine
whether any changes to the design should be made to make its performance more
robust.
Control Systems Design: Shop-floor control systems are designed as are mechanisms to put in place
a robust production plan. It is still widespread design practice to develop the
production control procedures after implementing the engineering technological
project. The consequences of this can be very severe. For example, it was found
that most difficulties with the implementation of cellular manufacturing
systems have been because of the poor design of the production planning and
control systems used. The approach
taken has been to first re-organise the shop layout and then use the existing
control system.
Job Design: All tasks necessary for the new
manufacturing architecture to function adequately are determined through an analysis
of the redesigned manufacturing system. Although the design methodology
outlined above is by far the most comprehensive attempt that has been published
to date, there are a number of areas where it could be described as limited.
For example, there is no use of optimisation the process being ‘satisficing’ in
nature rather than ‘optimising’ in nature. In addition, the methodology in
breaking down the design of a manufacturing system into discrete steps ignores
important inter-relationships within the system at any given time -
relationships and their effect on performance are only evaluated at the end of
the design process.
3.
LIMITATIONS TO CURRENT APPROACHES FOR THE DESIGN OF CELLULAR MANUFACTURING
SYSTEMS
Some more
fundamental limitations can however be identified and these are discussed
below.
Scope Of Approaches To Manufacturing Systems
Design: It is clear that most work has been directed towards solving the cell
formation problem rather than focusing on the total design of manufacturing
systems. There has been an over-emphasis on techniques that are used to define
cell structures rather than methodologies for the design of whole systems. This
point has been emphasised by Parnaby who defines two types of systems design
problem:
- The macro problem: Concerned with
large systems, integrating machines, processes, control systems, etc.
- The
micro problem: Concerned with for example small electro-mechanical mechanisms.
The point
is that much work has treated the manufacturing system design problem as a
micro problem rather than a macro problem. Much work has focussed on the use of
simulation, but other than little attempt has been made to put simulation in
the context of a complete manufacturing systems design methodology.
It is widely accepted that methodologies which try and solve complex real-world
problems within a systems context should have two key features:
Systematic: An orderly and well
disciplined way of getting things done.
Systemic:
A form of thinking based on ‘wholes’ and their properties. There should be a
focus on ‘holistic’ rather than reductionist thinking. This is best illustrated
with the concept of ‘emergence which is concerned with properties that exist at
one level in a hierarchy that cannot be explained by the properties of lower
levels in the hierarchy. Crudely, they are the difference between the system
properties’ and the sum of the systems component properties’. In terms of a
manufacturing systems design the emergent property could be the overall
economic criterion (such as return on capital) that the system must satisfy.
System
Objectives & Evaluation
Cellular manufacturing systems
are neither designed or evaluated in terms of the overall emergent properties
that are required of them. When a manufacturing systems design is undertaken it
should start with objectives that are related to business goals such as
improved return on capital employed. For
example, cells are generated with respect to the minimisation of inter-cell
moves. As a result, cell assessment or evaluation, with respect to the desired
emergent properties, must take place independently of cell formation.
The models that are often used for
the evaluation of cellular manufacturing systems are often inadequate the whole
system or the ‘synergistic’ benefits that often result as a consequence of
cellular manufacture. For example some simulation studies indicate better
performance for process layouts rather than cellular layouts. Such quantitative
evaluation does not reflect much of the physical evidence that has been
obtained . For example, some have compared process layouts and cellular layouts
on the basis of first-in first-out despatch rules at work stations, ignoring
the fact that a cellular layout facilitates the introduction of JIT. In
addition, the ‘cells’ used in the evaluation had significant inter-cell part
movement indicating that they were not very well specified.
Manufacturing
Systems Boundaries
It is common practice not to
consider the performance of the whole manufacturing system when introducing
cellular manufacture. With most cellular implementations either a temporary or
permanent ‘residual’ or ‘remainder’ uncellularised machine shop is operated in
conjunction with the newly created cells. A temporary residual is created by cellular
implementations taking place in small steps or in a time phased manner (i.e.
cells are implemented over a number of months or even years). A permanent
residual is created when it is decided that certain components or machine-tools
are not suited to operation in a cellular structure .
Typically, the effect of
cellularisation on this residual, which could be quite large, is not
investigated and improvements demonstrated for cells might be obtained at the
expense of the remaining manufacturing system. This is an example of local
optimisation with boundaries not being adequately addressed rather than system
optimisation. An evaluation of this situation could be very important for a
business considering the move to cellular manufacturing.. Thus a system boundary
is drawn around the manufacturing cell in question rather than the whole
manufacturing system. If the transient behaviour (performance during the period
from starting cellularisation to completing it) of the system is not
understood, implementation might be abandoned because of poor performance. It
is therefore common practice not to consider the performance of the whole
system when introducing cellular manufacture.
4.WHOLE BUSINESS SIMULATION:
Whole Business Simulation
(WBS) , and its integration with an appropriate design methodology is a concept
that is well suited to overcoming the weaknesses of current approaches to
manufacturing systems design that have been discussed above.
Whole Business Simulation Overview
This section will discuss the
core elements of WBS. Typically, WBS includes the following functions:
a customer model / demand generator
the design function
production engineering (process
planning)
material requirements planning
supplier(s)
manufacturing operations
ancillary cost generators (to
generate overhead costs not covered elsewhere)
and an accounting system
The above would represent a
‘minimum’ system, but the architecture could be extended readily to include any
other function. The elements of the system are linked by the same kind of
transactions that occur in the real world. The basic process can be illustrated
by the following example that relates to figure 2.
A group of elements covers all
the basic operations of the factory: sales, materials planning and control,
manufacturing operations and purchasing. Additional elements are needed to
represent the activities of external companies, e.g. suppliers and customers. A
sales demand triggers sales orders to be passed to the materials requirements
planning (MRP) system. Works orders and purchase orders are generated by the
MRP system, using the usual algorithm. These orders are passed to the factory
simulator and. supplier model as appropriate. Local planning or scheduling
rules would be applied in the factory module that simulates production and
warehousing activities. Stock movements are posted to the MRP system, as are
works orders completions, shipments to customers and deliveries to suppliers.
The system is self-contained requiring no external order or demand data
streams.
Standard accounting transactions are
generated from events that occur in the operations group of elements. For
example, sales orders and deliveries lead to invoices being issued to the
company’s customers’. Following an appropriate delay invoices are paid and the
‘books’ updated. Purchased items are dealt with in a similar fashion. Where
transactions cannot be related to driver activities in the core elements of the
model, an ancillary generator is used to produce them. This approach may be
used to cover the cost of general overhead expenses. At the end of each
accounting period, the accounting system can produce a complete set of accounts
for the company, including the profit & loss statement, balance sheet and
funds flow statement.
The model is able to evaluate
changes in any one of the system elements to assess its impact on the financial
situation of the company. The model can be run and analysed with and without
the change and the effect compared. Changes in manufacturing, planning and
control policies, overtime policies, change-over times or even accounting
practise could be evaluated by the system.
A demonstration WBS system has been built at
Aston University in order to illustrate the viability of the concept utilising
object orientated design techniques. A number of commercially available packages
have been used on the system, which is able to demonstrate how the design,
manufacturing system and operations decisions described above would be
evaluated in a fictitious manufacturing company.
5.Whole
Business Simulation & Manufacturing Systems Design:
WBS can be
used to compare different manufacturing systems designs. Providing the
simulator used is appropriately flexible, the system described above could be
used to compare different designs. The accounting transactions associated
(including both initial investments and operating costs) with each alternative
design would be catered for in the same way as theywould occur in the ‘real
world’. Savings associated with the manufacturing designs such as reduced
current assets (in for example, the form of work-in-process) or reduced
expenses would show automatically in the accounts. Each design variant would be
tested and the alternative that had the ‘best’ impact on for example, return on
capital employed would be selected. The impact of delays in implementation (the transient) or
the effect of excessive or inadequate demand could be assessed in terms of its
financial impact on the company. In terms of evaluation, decisions would be
made on the basis of the firm’s
accounting systems rather than its costing systems.
ADVANTAGES OF WBS.
· Evaluation of total system performance. As WBS can model the whole
business inappropriate system boundaries would not be drawn around individual
cells, lea ding to the neglect of the sometimes large so-called ‘residual’
or‘remainder’.
· The design and evaluation of manufacturingsystem with respect to overall
desired emergentproperties based on business objectives.
· The integration of manufacturing systems designinto the strategic
decision making process. AsWBS can model the effect of changes in
themanufacturing system on the performance of thewhole business, the profile of
manufacturing will be raised in companies and information of interest to senior
management produced, rather than details on local performance.
6. CONCLUSIONS:
This paper has
presented a categorisation of current approaches to the design of cellular
manufacturing systems. Each category has been discussed and a number of
limitations in current approaches to the design of cellular manufacturing
systems identified. In essence, although the identified techniques and
methodologies are systematic they are not systcmic or holistic in nature. The
concept of a Whole Business Simulator has been presented as a means for the
evaluation of decisions in manufacturing companies. The Whole Business
Simulator, when embedded within an appropriate methodology for the design of
manufacturing systems will overcome many of the limitations of current
approaches.
EmoticonEmoticon