Professional Documents
Culture Documents
1. Introduction
Flexibility is a major consideration in the design of manufacturing systems.
Flexible manufacturing systems (FMSs) have been developed over the last two
decades to help manufacturing industry move towards the goal of flexibility. An
F M S comprises three principal elements: computer controlled machine tools; a n
automated transport system and a computer control system. An F M S combines high
levels of flexibility with high productivity and low levels of work-in-process inventory.
It may also allow unsupervised production. In order to achieve these desirable
benefits the control system must be capable of exercising intelligent supervisory
management. Scheduling is at the heart of the control system of an FMS. The
development of effective and efficient F M S scheduling strategies remains an impor-
tant and active research area.
However, unlike classical scheduling research, a common language of commu-
nication for FMS scheduling research has not been properly defined. For conven-
tional machine shop scheduling problems, a classification scheme was developed and
has been serving as a systematic framework in classical scheduling theory. Although it
is limited in describing real scheduling problems, within this framework the type of
problems addressed in a research work can be stated concisely. Different methods can
be evaluated and compared for the same type of problems. For F M S scheduling, such
a framework has not been developed. Each research paper addresses a specific FMS
scheduling environment. When a new model or method is developed, its scope is often
not clear. Comparison of different methods is also difficult. In order to analyse and
compare scheduling strategies in F M S and develop more effective and efficient
approaches, it is necessary to have a sound framework within which to define
precisely a specific problem type from the wide range of problems which can occur.
This paper presents a comprehensive classification for F M S scheduling problems
which allows the specification and identification of problem types and problem
instances. It utilizes a classification scheme for F M S configurations recently described
by MacCarthy and Liu (1993). Unlike the scheme for classical machine shops, it
provides a comprehensive description of real F M S scheduling problems because it
takes into account all the significant factors which impact on scheduling decisions.
The paper first discusses terminology and clarifies a number of terms relevant to
FMS scheduling. The nature of scheduling decisions in FMS is then described. This
background discussion provides the basis to identify the major elements affecting
FMS scheduling problems. Finally, a classification scheme using five descriptors is
developed and explained using examples.
2.3. Machine set up, tool changing, loading and part routeing
These concepts are closely related and the following discussion clarifies the view
taken here:
Machine set up is defined here as the process or decision of assigning tools on a
machine, in order to perform the next operation(s), from its initial state o r a working
state arising from a previous operation. A working state for a machine here means a
state under which it can perform certain operations, i.e. tool setting has been
completed for these operations.
Tool chunging has a similar meaning but often implies the change from one
Classifying FMS scheduling problems 649
working state to another rather than from an initial state of the machine. Two types
of tool changing may be identified in FMS. The first type is choosing the next working
tool from a tool magazine. This is usually done automatically and takes very little
time. This type of tool changing will not be considered here. The second type is
changing the set up of the tool magazine. This type of tool changing has the same
meaning as machine set up. Sometimes it is not necessary to change all the tools in a
tool magazine because some of the tools on the magazine for previous operation(s)
may also be needed by the next operation(s). The time needed for changing may
therefore depend on the number of tools to be changed, i.e., the operation(s) which
will be performed next on the machine. In this case the set up times are said to be
sequence-dependent.
Machine loading is defined here as the process or decision of allocating tools to
machines before the start of a production period with the assumption that the
loaded tools will stay on the machine during the production period. This decision
must consider which operations may be performed on each machine during the
period. Therefore, machine loading also includes the allocation of operations to
machines.
Part routeing is defined here as the process of determining the machines on which
each operation for a part is to be performed, i.e. determining the route or sequence of
machines for each part passing through the system. Machine loading decisions may
result in a unique machine for each operation. It may also result in a situation in
which one operation can be performed optionally on two o r more machines, i.e., all
the tools required for the operation are assigned on each of these machines.
I t can be seen that all the above concepts are concerned with two types of
decisions-assignment of tools and allocation of operations. These two decisions
are inter-dependent. Loading considers both of them. Machine set u p o r tool
changing concentrates on the tool assignment decision either made before the start
of the production period o r during this period, assuming the allocation of
operations is known in advance. Part routeing concentrates o n the operation
allocation decisions, leaving the tool changing decision t o be made later o r
assuming tools are already assigned to the machines.
type
FMS /
constraints
Capacity
/ Job
description
/
Production
environment
Scheduling
criteria
The first four descriptors describe the relevant aspects of the system and the
environment. The last descriptor indicates the scheduling objective. For brevity these
descriptors may be written as A/B/C/D/E. Possible values for each of these descrip-
tors are discussed below.
'A' indicates the type of system to be scheduled. Valid types are SFM, FMC,
MMFMS, MCFMS.
'B' indicates types of resource constraints involved in the scheduling decision and
654 J . Liu and B. L. MacCarfhy
the constraint levels of these resources. The constraint level of a resource type can be
noted using the combination of the level name and the resource name, e.g., 'limited:
pallets'. Therefore this descriptor can be presented as a collection of such combina-
tions. When infinite capacity is assumed for a type of resource it will not appear in the
description because it does not set constraints for the scheduling problem. For
consistency the following abbreviations may be useful.
Constraint levels:
0: zero capacity lim: limited capacity
Resource types:
M: machines HD: material handling devices
SB: storage buffers T D : tool changing devices
PL: pallets FX: fixtures
'C' describes the complexity and the routeing flexibility of jobs. The following
abbreviations may be used.
JCI: only one operation for each job
JC+: two o r more operations for some o r all jobs
RFI: only one machine for each operation
RF+: two o r more machines for some or all operations
'D' indicates the requests from a higher production management level. The
following abbreviation may be used.
dd: there are due date requests
pr: orders are handled periodically
cn: orders are handled continuously
Ipt: one part per type
+pt: more than one part per type
rr: there are ratio requests
bs: there are batch size requests
'E' indicates the scheduling criterion. Possible values include: mean completion
time (c),
makespan (C,,,), and maximum tardiness (T,,,), etc.
This classification scheme attempts for the first time to identify and describe all the
major factors which affect the modelling of, and the solution to, F M S scheduling
problems. Different modelling frameworks and solution approaches can be devel-
oped, evaluated and compared within this scheme.
6 . Examples
Two recent problems chosen from the literature are described below using the
framework proposed in this chapter and classified using the proposed classification
scheme.
(I) Lee and lwata (1991) describe in detail an F M S scheduling problem. It can be
stated clearly as follows using the classification attributes introduced in this
paper.
FMS Type: FMC
Capacity constraints: machines limited number (4)
MHS l AGV
storage infinite
Classifying FMS scheduling problems
(2) Mukhopadhyay et al. (1991) describe another problem in detail. It can also be
easily stated with our classification attributes:
FMS Type: MMFMS
Capacity machines limited number (4)
constraints: MHS limited number of devices (2 AGVs)
storage limited capacity
tool drum limited capacity
pallets limited number
Job description: more operations for a job
more machines for an operation
Production more than one part per type
environment: product mix given
lot-size given
orders handled continuously
Scheduling criteria: mean flow time.
Using the proposed classification scheme this problem can be classified as
these limitations. It is much more comprehensive and takes into account all the major
factors which may impact on a real FMS scheduling environment. For this reason we
believe it has significant potential in the analysis and design of real FMS control
systems.
7. Conclusions
In this paper we have proposed a classification scheme for FMS scheduling
problems based on the analysis and discussion of the scheduling decisions in FMS and
the factors related to these decisions. This classification scheme provides a systematic
framework for description and analysis of FMS scheduling problems and for
development, evaluation and comparison of FMS scheduling approaches. The
usefulness of the scheme has been demonstrated through examples. The authors
intend to present shortly the results of research in important classes of FMS
scheduling problems, particularly for FMC and MMFMS environment. The nature
of the problems will be described within the framework proposed here. The generali-
zation of the solution approaches will also be discussed within the context of this
classification scheme.
The approach proposed here does not attempt to be fully comprehensive but does
address the major scheduling problems in a wide range of real systems.
References
BAKER, K. R., 1974, lnrroduction ro Sequencing and Scheduling (New York: Wiley).
LEE,Y. H., and IWATA, K., 1991, Part ordering through simulation-optimization in an FMS.
Inrernarional Journal of Producrion Research, 7 , 1309-1323.
LIU,J., 1993, A general structured approach to scheduling problems in flexible manufacturing
systems. PhD thesis, University of Nottingham, UK.
MACCARTHY, B. L., and LIU,J., 1993, A new classification scheme for flexible manufacturing
systems. Inrernarional Journal of Production Research, 31, 229-309.
MUKHOPAOHYAY, S. K., MAITI,B., and GARG,S., 1991. Heuristic solution to the scheduling
problems in flexible manufacturing systems. Internarional Journal of Production
Researc/~,10, 2003-2024.
NAKAMURA, N., and SHINGU, T., 1987, Scheduling of flexible manufacturing systems. In Toward
the Facrory of rlte Future, H.-J. ~ u l l i n ~ e r , a nH.
d J. Warnecke (eds), bp. 147-152.
SHANKER, K., and TZEN,Y. J., 1985, A loading and dispatching problem in a random flexible
manufacturing system. Inrernarional Journal of Production Research, 23, 579-595.