Roboacs & Computer-Integrated Manufacturmg, Vol 4, No 3/4, pp 517-523, 1988 0736-5845/8853 00 + 0 00

Printed m Great Britain Pergamon Press pie

• Paper

AN INTEGRATED PRODUCTION PLANNING AND SCHEDULING SYSTEM

FOR MANUFACTURING PLANTS

J. B U C K L E Y , A . C H A N , U . G R A E F E , J. N E E L A M K A V I L , M . S E R R E R a n d
V. T H O M S O N
Division of Mechanical Engineering, National Research Council of Canada, Ottawa, Ontario, Canada

1. INTRODUCTION schedules; it utilizes the FAMS system modelling

The National Research Council of Canada (NRCC) capability to verify those schedules.
and ICAM Technologies Corporation (Montreal) The model can simulate in detail the running of a
have designed and built a Manufacturing Planning given schedule to verify that assumptions made by
System called FAMS (Flexible Automated Model- the Scheduler are valid. It also allows the plant
ling and Scheduling). The system is an integrated manager to run through a schedule to identify poten-
scheduling and production management tool for tial trouble spots or to determine strategies in
small- to medium-sized manufacturing companies response to machine breakdowns and other
which operate in a job shop environment; that is, it is unscheduled occurrences. The model can also be
a system designed for use in make-to-order manu- used to evaluate hypothetical scenarios.
facturing enterprises where the workload imposed Shop activity data are stored by the Shop Monitor
on manufacturing resources, as determined by the which updates the manufacturing database.
influx of orders, is largely unpredictable and subject Schedules are updated from job, machine and opera-
to significant variations with time. tion activity information which is supplied from the
The objectives of FAMS are to: Shop Monitor.
1. assist a shop manager in planning requirements The User Interface consists of a menu driven task
for manufacturing resources and their utiliza- initiation system, computer terminal data entry and
tion; reporting, as well as advanced graphics display
2. allow accurate estimation of work order com- routines for the presentation of manufacturing status
pletion dates, which are governed to a large and production information. The Database Interface
degree by shop capacity constraints and job routines allow access to the FAMS manufacturing
mix; database and any other databases supported by the
3. determine optimal allocation of outstanding user.
manufacturing operations to available The remainder of this paper is devoted to a
resources; detailed discussion of the various subsystems of
4. enable a user to experiment with changes in FAMS outlined above.
operating procedures, resource levels or job
mix, to reduce manufacturing costs or increase
productivity. 2. SCHEDULER
A schematic diagram is shown in Fig. 1 which The scheduling and planning module produces
gives an overview of the FAMS system components. master production plans and detailed shop
It is composed of a kernel and a shell. The kernel schedules. User-selectable manufacturing goals,
consists of three modules: the Scheduler, the Model- such as maximizing throughput, minimizing work in
ler and the Shop Monitor. The shell has two inter- process (WIP), finishing jobs according to due date
face modules: a Database Interface and a User and minimizing setup time, can be implemented in
Interface. The scheduling and planning module pro- the optimization process. Additional scheduling fea-
duces master production plans and detailed shop tures are job prioritizing and on-line schedule

517
 518 Robotics and Computer-Integrated Manufacturing • Volume 4, Number 3/4, 1988

ablel O,spatc mgrules

Rule Goal Measurement

T SPT Least
work-m-progress
$ value or total time
mvested on the floor
S/OPN Respect of due date Fraction of jobs tardy
SST Maximum use of Time spent for set-ups
resources
MWKR Maximum throughput $ value of work done/
unit made per period

described above but with modifications to include

external priorities such as customer ranking, emer-
gencies, etc.

2.2. Scheduling system

The scheduling system is used to assist manu-
facturing management to better plan and organize
current and future production by offering the follow-
ing facilities:
• long term planning of resources,
Fig. 1 General overview of the production planning and • optimizing the everyday allocation of resources in
scheduhng system.
the shop,
• release of schedules for the next working period.
The scheduling system proceeds through many
modifications. The scheduler employs the modeller's
steps when processing information for the genera-
simulation capability to verify schedules.
tion of schedules. The steps for schedule generation
are:
2.1. Dispatching rules for the scheduler
• work order entry,
In the scheduling system, job selection decisions
• operation by operation starting time calculation,
are effected by means of dispatching rules which
• capacity planning (rough-cut and detailed),
enable the system to select the job to be processed
• short term dynamic schedule generation,
next on a machine. The job assignment strategy is
• schedule modelling for assumption verification,
based on priorities determined for each of the wait-
• schedule performance statistics,
ing jobs. The decision is taken by weighing a number
• schedule commitment,
of cost factors associated with manufacturing
• schedule delivery to the shop.
activities. A particular dispatching rule can produce
schedules which emphasize different manufacturing
goals. Work order entry
Four dispatching rules namely, Shortest Pro- A work order entry system enables the user to
cessing Time (SPT) rule, Slack per Operation input the sales orders and work statement informa-
(S/OPN) rule, Shortest Setup Time (SST) rule, and tion, viz quantity, item, due date. Using the data
the Most Work Remaining (MWKR) rule, were from the manufacturing database, the system gener-
selected for application from well over 100 rules ates a list of jobs to be performed with their associ-
after a detailed study, as described in Ref. 1. They ated operations. A set of job-operation-machine
were chosen based on their relative effectiveness; (JOM) triplets are built which describe the sequence
the performance measures optimized by each of of work to be performed in order to execute a job
these rules are shown in Table 1. and fulfil a particular work order. The following
To increase system flexibility, the various dis- types of jobs are scheduled: discrete part manu-
patching rules given in Table 1 are combined using facture, assembly, NC programming, fixture manu-
weights which are company-specific and which can facture, partial maintenance and full maintenance.
vary depending upon the corporate manufacturing A list of materials needed to manufacture the
goals. One can also choose just one goal by assigning items in a work order is made from the bill of
all but the relevant weight equal to zero. The final materials and process plans. These material lists are
decision criterion used by the system is the rule available to the user as a checklist within the system
 Planning and schedulingsystem • J. BUCKLEYet al. 519

or are potentially employable as an interface to features of the Dynamic Scheduler are:

material requirements planning (MRP) systems. • the ability to produce a schedule for all, or for a
selected set of machines, departments and jobs in
Capacity planning the shop,
Capacity planning is a medium- to long-range • consideration of several machines to which an
planning procedure which evaluates the capacity operation can be assigned,
requirements of alternative Master Production • consideration of flexible breaks and holidays,
Schedules (MPS), so that management can make • the ability to reserve blocks of time in the
changes in capacity levels and/or shop loads as schedule for some user-specified operation or
necessary. In order to perform the capacity planning, activity,
an operation "back-off" calculation is done first. • the tracking of fixtures so that the user can take
Due dates are computed for all operations belonging advantage of common fixtures and thereby
to a given job by calculating backward from the job's eliminate or reduce setups.
due date. This is done for all jobs which are iden-
tified at work order entry. Latest start times and 3. MODELER
earliest start times are then determined for all JOM Simulation plays a very important role in the
triplets. It is these JOM triplets which are given to FAMS system. It is used to validate the schedule of
the rest of the scheduling system to be set into a operations, to refine resource allocation plans
timetable. associated with imminent production and to deter-
The capacity planning incorporated in the schedul- mine the expected shop behavior under a set of
ing system is done in two stages. In the rough-cut hypothetical conditions, especially when different
planning stage, the total capacity of each critical workloads, operational strategies and resource levels
work center is compared with the total load on that exist. The principal advantage of the simulation
work center for the planning period. The user is model is its ability to account, in detail, for a variety
prompted to take action if the average capacity is of resource availability constraints, shop operating
found to be much lower than the average load. In the rules, shop layouts and singular equipment
second stage, a detailed (i.e. time-phased) capacity characteristics. The scheduling module is generally
plan is generated. Job-operation competition is ana- not able to deal with the impact of such variables.
lyzed at this stage using the well known Project Above all, the model can explicitly recognize
Evaluation and Review Technique (PERT). demands imposed on finite auxiliary resources in the
shop (other than machine tools), such as material
transport devices, human operators, storage areas
Short term (dynamic) scheduling and fixtures.
The dynamic scheduler produces a detailed Simulation techniques also enable a decision
schedule for a time span specified by the user using maker to discover and understand the consequences
the output from the long-term PERT scheduler. It of possible changes to the existing shop. These
uses the discrete event simulation package ANE- changes may be the result of factors beyond normal
VENT. 2'3 The user rates all jobs on two factors: how control, such as the sudden arrival of many new
important the customer is and how urgent the job is. orders or of proposed modifications to resource
These ratings can be modified any time prior to levels or operational strategies (for example, the
producing a schedule. Also, the user is asked to number and types of available machines, manpower
weigh the relative importance of the manufacturing levels, material handler dispatching procedures,
goals. The weights and ratings are then used intern- etc.).
ally to produce a schedule which best achieves the Input to the model consists of two main compo-
user's objective. The performance of the schedule nents: a shop image and a job profile. The shop
with respect to the manufacturing goals is measured image in turn is subdivided into a static and a
and reported. dynamic part, while the job profile consists of a
At any point in time, the user can select a portion time-dependent schedule and a job-dependent bill of
of the optimized schedule to become the schedule to operations.
be issued to the shop floor. Should an event upset The static shop image describes the shop layout,
the released shop schedule, the user can ask for a that is the location of all machines, material handlers
total rescheduling by the Dynamic Scheduler, or and their paths and storage areas. It also specifies
have the Modeler simulate such an event and evalu- the characteristics of these machines, material hand-
ate its consequences prior to rescheduling. Other lers and storage areas. Shop operating characteristics
 520 Robotics and Computer-Integrated Manufacturing • Volume 4, Number 3/4, 1988
such as the start times and durations of the various examined to determine the next location to which
shifts and break periods also form part of this input. this part is to be transferred, and the appropriate
The dynamic part of the shop image provides material handler is charged with this transfer task.
information on the current status of all machines, Conveyors as material handlers are treated some-
material handlers and storage areas, including an what differently in that they can accept parts to be
enumeration of parts which are currently in the moved at any time as long as there is room on the
manufacturing process. The available human conveyor and the conveyor is operational.
resources, their skill levels and their status are also The following operational factors common to
part of the dynamic shop image. most manufacturing facilities are incorporated in the
Time-dependent data for the job profile are given FAMS simulation model.
by the schedule, which provides the number of parts • Material movement: A task for a material hand-
to be made for each job and the start and finish times ling device is initiated or queued whenever a part,
for each scheduled j o b - o p e r a t i o n - m a c h i n e triplet. a part lot or a fixture is to be moved from one area
Apart from regular machining operations, the to another or when an individual part is to be
schedule includes such operational steps as setup, loaded or removed from a machine. Material
teardown and NC program proofing. handlers can be classified according to type (e.g.
The job-dependent bill of operations specifies the robot, forklift, crane, conveyor) to account for
routing through the shop to be followed by the parts differences in their capabilities, speed, range, etc.
of a job during manufacture. It consists of a sequen- They can also be organized independently into
tial list of operations, including material handling "groups", allowing different areas of the shop (or
operations between machines and storage areas. For sets of machines) to be served by a different group of
each machining operation the machine, machining handlers.
time and the probability of the part having to be • Part lots: Parts can be moved from one area to
scrapped are given. For material handling operations another individually or in lots. Lot sizes can vary
the specific handler (if dedicated) or handler type is during the course of a job.
provided. The number of parts to be moved at one • Batch machines and part spawning: The model
time is also specified for material handling oper- can simulate the operation of machines which
ations, allowing material transfers of individual parts process more than one part at a time (e.g. ovens).
or batches of parts with each move. Furthermore, a single part loaded onto a machine
After initialization, the model is ready to be run. can "spawn" more than one output part (in the
At the start of a new shift, or after the completion of case, for example, of a stock cutting operation).
a job operation on a machine, the schedule is • Constraints on human resources: The model rec-
examined one machine at a time, to see if there are ognizes that manpower availability is not un-
further operations such as NC program proofing, limited. The labor force can be subdivided accord-
set-up, machining or teardown to be performed at ing to user-defined skill levels and manpower
this time or at some time in the future. If a new allocated to tasks such as setup, machining and
operation is scheduled to start, all required con- teardown on the basis of specified skill require-
ditions and resources are checked. These include a ments.
check on machine status, availability of an operator • Shifts and breaks: An arbitrary manpower and
or set-up person with the proper skill level and the machine shift structure is permitted, including
availability of a part. If all requirements are met, the provisions for overtime. H u m a n resource breaks
task to load the new part on the machine is issued to (coffee, lunch) are simulated and the attendant
the appropriate material handler or material handler impact on machine utilization and throughput
group. If none of these handlers is available at that represented.
t~me, the task is entered in a "look for work" list The FAMS simulation is based on A N E V E N T , 2'3
from which tasks are later dispensed, one at a time, a general-purpose, interactive, discrete-event simu-
as appropriate material handlers become available. lation package developed by NRC researchers. The
After a part has been loaded on a machine, the interactive capabilities inherited from A N E V E N T
machine will process the part for the duration by the FAMS model permit the user to:
specified for this part and operation in the bill of • Examine in detail the state of the simulated sys-
operations. At the end of the operation cycle, ran- tem at any time during a run so that the nature of
dom number generation is used to determine if the its evolution can be better understood.
current part is to be scrapped. If the part just pro- • Modify shop parameters (e.g. the number of mat-
duced meets specifications, the bill of operations is erial handlers, machine speeds) while a run is in
 Planning and scheduhng system • J BUCKLEYet al 521

progress to determine the impact of such changes the data are entered into the manufacturing data-
on shop performance. base, no updating or correction of the manufacturing
• Introduce a new event (such as a machine break- data is made by the manufacturing planning system.
down) or create a new set of circumstances in the It is left to the user to determine when data must
simulated shop. be modified due to changes in the manufacturing
The simulation software also employs dynamic, environment. Schedules are produced from pro-
color graphics displays that provide an animated duction status data and no inference is made by the
picture of manufacturing activity, and thereby system as to the validity of the data.
• Improve the user's ability to determine how well
the simulated shop is performing. (Displays make
it easier to identify problem areas such as bottle- 5. SHOP M O N I T O R
necks, large w.i.p, inventories, inefficient alloca- Following the above philosophy, production data
tion of material handlers, etc.) are stored in the manufacturing database and used
• Enable the user to understand better the complex by the scheduling system to determine the
interdependence of shop events and the causal completeness and status of j o b - o p e r a t i o n - m a c h i n e
relationships involved therein. triplets. The FAMS system does not do active shop
• Provide a friendly and interesting interface for data collection nor does it perform statistical ana-
p e o p l e - - s u c h as department f o r e m e n - - w h o are lyses of production activities. It is left to the user to
unaccustomed to analyzing large amounts of tabu- analyze production, material and cost data, and to
lar statistical data but who are otherwise ideal update the database, so that "correct" schedules are
users of the tool. produced. Reports of completed work and job status
are available to the user at any time. They only
4. DATABASE INTERFACE reflect shop history as entered into the manu-
A manufacturing planning system requires a great facturing database.
deal of information about production capability and
process procedures. There is also a need to organize
many ancillary activities; this leads to the require- 6. MATERIALS R E Q U I R E M E N T PLANNING
ment of tracking information which is not directly FAMS does not incorporate an active materials
associated with actual material processing (e.g. mat- scheduling system. It does have a passive data inter-
erial requirements, costs). Due to the large amount face through which material information may be
of data needed by a planning system, and the fact passed to and from an M R P system. The Scheduler
that much of this information would already exist in can take into account the availability of materials
databases, special considerations were given to the when developing a schedule; it does this by not
design of the data interface for FAMS. scheduling certain j o b - o p e r a t i o n - m a c h i n e triplets
Since application programs for scheduling and and indicating to the user that these triplets cannot
modelling need access to the database, the user be scheduled due to the unavailability of materials.
needs convenient access to the data in the manu- The material availability information must be sup-
facturing database. A n d since data need to be de- plied to the FAMS system by the user. A simple
livered to and received from external databases, a "flag" mechanism is used to indicate whether mater-
commercial, relational database management system ial is or is not available by a certain date.
was used in the design of the manufacturing data-
base for FAMS. Since the final, commercial product
is to be marketed on several possible types of com- 7. COST ACCOUNTING
puters, a neutral database interface was developed FAMS provides a simple cost determining system.
as well. Direct labor and machine costs can be calculated
The manufacturing database contains reformation from cost structure information in the manufacturing
about the following items: product bill of materials, database entered by the user. In calculating job cost
process plans, processing resources (machines, per- information, the system considers the following
sonnel), material requirements, schedules, pro- types of operations: setup, "processing" (machining)
duction records (Shop Monitor information), and and teardown. In considering these types of oper-
cost accounting. The database is structured so that ations, the system does costing for labor and machine
this information is available at two interfaces--an times; it also takes overhead rates into account. At
application program interface and a user interface. the moment, material handling activity is considered
While range and consistency checks are made when as an overhead.
522 Robotics and Computer-Integrated Manufacturing • Volume 4, Number 3/4, 1988

8. USER I N T E R F A C E Graphic displays are used in the FAMS package to

The user interface for the FAMS package encom- enable the user to monitor ongoing activities during
passes a menu subsystem, a data entry subsystem, a a simulation model run and to convey output infor-
data inquiry and reporting subsystem and a graphical mation by means of graphical representations. Typi-
display subsystem. cally, a graphic display contains a static portion
All FAMS modules can be accessed via the menu which may portray the shop layout including sym-
subsystem. At the highest level, the master menu bolic representations of work-centers, material
contains all the principal selections in the FAMS handlers, queues and storage areas. A dynamic por-
package available to the user. The user can arrive at tion displays the status and statistical information for
a desired destination by stepping up and down these elements using an appropriate combination of
within the multi-level menu structure or by specify- color graphics, bar charts and tabular formats. The
ing the destination directly. A menu selection trig- frequency of updates for graphical displays is con-
gers the display of the next sub-menu when appro- trolled by the user.
priate. A Help facility can be accessed at any level; it
displays all the available selection codes at that level, 9. INSTALLATIONS
and further explanation of each can be requested if The FAMS product was first installed in October
desired by the user. A Self-Teaching facility enables 1986, with International Tools Ltd in Windsor,
the user to become familiar with the different Ontario, acting as a beta-test site. The manu-
options and their functions within the FAMS pack- facturing database of International Tools Ltd was
age. Via this facility, for instance, the user can learn input and a shop floor data collection system was
about the different scheduling criteria offered by connected to the Shop Monitor system in FAMS.
FAMS and evaluate their applicability to his own Capacity planning and scheduling were performed
operations. By means of a Security facility incor- by December 1986. Beta testing was completed by
porated at the various menu entry levels, the FAMS mid-January 1987. The company is now running its
system manager can assign different access privileges conventional schedulmg system and the FAMS
to the users within an orgamzation, thereby control- scheduling software in parallel, in order to deter-
ling the selections available to any specific user. mine any difficulties using FAMS and to put in place
When fully implemented, the menu subsystem is new operating procedures. New installations for
expected to be interfaced to a selection device such FAMS are to commence in March 1987, after testing
as a digitizer or mouse. at International Tools Ltd is complete.
As data are entered, the data entry subsystem
performs on-line checks on data type, range or integ- 10. CONCLUSIONS
rity and reports any detected errors. Application While several other commercial packages 4'5 are
modules which require data to be posted by the user available to perform capacity planning and schedul-
activate either data entry questionnaires or data ing, the production planning and scheduling system
entry forms. Data entry questionnaires typically con- described here has the additional capability to
tain texts followed by a set of questions. Additional schedule and monitor shop activities in great detail
questions may be displayed depending on the ans- while taking into account bottlenecks, breakdowns
wers entered. Data entry forms containing sets of and other unscheduled occurrences. Other functions
data fields may also be used. Limited "smart field" allow the user to tailor the schedules to appropriate
capabdities allow the skipping of non-apphcable criteria which reflect a company's manufacturing
fields. goals. The package provides the plant manager and
Data inquiries are entered via alphanumeric ter- other users of the system with interactive modelling
minals or in graphic form via graphics terminals. which can alert them to critical situations and enable
Typical reports include production reports which them to evaluate recovery strategies after upsets
contain information on production activities for each have occurred in their shop.
machine/job/operation, such as quantity, scheduled Use of this p a c k a g e - - o r many others for that
processing time vs actual, summary of variances, m a t t e r - - m a y create certain initial difficulties m
etc., and status reports, which contain status and small plants, however. It may require certain prere-
utilization information for jobs, machines, material quisites and the observance of discipline that did not
handlers, queues and other resources. In addition, exist in the plant before the package was introduced.
machine schedules are displayed in the form of For instance, use of the package does require a
Gantt charts and reports on machine loading, capac- consistent method of labeling jobs, operations, parts
ity forecasting and job costing are also available. and fixtures. Also required are formal process plans
 Planning and scheduhng system • J BUCKLEYet al. 523

for the manufacture of parts. F u r t h e r m o r e , there ter modeling package based on discrete event simula-
exists a need for large amounts of manufacturing tion. National Research Council of Canada Report
LTR-AN-48, October, 1982.
data to be entered into the c o m p u t e r to build up the
3. Crate, G.F.: ANEVENT--an interactive computer
manufacturing database. modeling package based on discrete event simulation, a
users' guide. National Research Council of Canada
Report LTR-AN-53, October, 1982.
REFERENCES 4. PERA. Production systems--the PERA 4 W capacity
1. Neelamkavil, J., Thomson, V., Rao, D.: Use of dis- planning and scheduling system--a standard and port-
patching rules in manufacturing applications. 5th able package. PERA Report 379, September, 1983.
Canadian CAD/CAM and Robotics Conference, 5. PERA. Production systems--capacity planning and
Toronto, Canada, June 1986. scheduling applied to a group technology manu-
2. Graefe, P.W.U.: A N E V E N T - - a n interactive compu- facturing system. PERA Report 372, December, 1982.