Professional Documents
Culture Documents
Production Planning and Control in Textible Industry A Case Study
Production Planning and Control in Textible Industry A Case Study
Abstract
This paper presents an interactive model based system for the management of production in textile production systems
focusing on the Master Production Scheduling problem. Because of the special characteristics of the industry, that is mainly
the multi-phase process with multiple units per phase, different planning horizons and different production requirements for
each phase, the scheduling of these systems becomes quite complex. Apart from a comprehensive presentation of the set of
the modules the system is composed of, together with their interrelationships, the above characteristics are analyzed, and
their impact on the production control system is explained. The system is also related to two well-known production control
systems, namely MRP-II and Optimised Production Technology. The system’s attributes are presented with the aid of data
structure diagrams, while the complete algorithm concerning the Master Production Scheduling module, in a pseudo-code
form, and the corresponding part of the database are illustrated in Appendix A.
Keywords: Master production scheduling; Decision support systems; Production planning; MRP-II; Textile industry
0166.3615/96/$15.00 &pyright 0 1996 Published by Elsevier Science B.V.. Published by Elsevier Science B.V.
PZI SOl66-3615(96)00038-3
128 N.I. Karacapilidis, C. P. Pappis / Computers in Industry 30 ( 1996) 127- 144
in that it takes into account the special characteristics two clothes being processed in sequence. Partial
of the textile industry. These characteristics are dis- changeover times take place between two succes-
cussed below. sive related jobs and are much smaller than the
The textile industry imposes a variety of con- total ones, which refer to unrelated jobs (see
straints concerning the integration of an overall Appendix A for the definition of relation be-
scheduling procedure. Typically, a textile production tween jobs). Minimization of the total setup times
unit is characterized by a multi-phase manufacturing is among the most significant objectives in the
process with multiple production units per phase scheduling of a textile industrial unit.
(i.e., parallel machines). The mixed character of a Throughout the set of phases, jobs can be split
textile production system, which lies between job- and processed in parallel. Nevertheless, job split-
shop and flow-shop, makes production management ting has to be weighed up with the advantageous
quite complex. In addition, there are sequence de- results, mainly in terms of quality of constant
pendent operations, and different planning horizons processing of a particular job in the same ma-
and production characteristics for each phase. Conse- chine.
quently, different production planning algorithms for Simultaneous setting up of the weaving machines
each phase are required. For example, the weaving which have been charged with the parallel pro-
process is characterized by long planning horizons cessing of a particular job should be avoided. It is
and relatively slow speed of machines, very long worth mentioning that the setting-up of a weaving
setup times, very large production batches, and mixed machine usually requires more than two workers,
order and stock-based production. On the contrary, while only one worker can attend to the normal
the warp making process is characterized by short operation of about ten of them.
planning horizons and high speed of machines, short Textile production systems may be treated as a
setup times, small production batches and only or- succession of local problems, one per each pro-
ders-based production. The above phases pose the duction phase. The coherence of these local prob-
most complex production scheduling problems. lems should be taken into account by “material
Additional special characteristics of the textile requirements planning” or “just-in-time” ap-
industry, that have been taken into account in the proaches [2].
development of the YFADI production control sys- The rest of the paper is organized as follows: A
tem, are the following: literature review is given in the next section. The
- Most textile companies are ageing while the tech- system architecture is presented in Section 3; the set
nology changes rapidly. These companies own of modules the system is composed of and their
machines of different ages and production charac- coordination are also discussed. The MPS problem
teristics, such as processing speed, changeover and the related algorithms that have been developed
possibilities and facilities, etc. for the system are comprehensively described in
- The changeover (i.e., setup) times of the ma- Section 4 (a pseudo-code form of these algorithms,
chines are dependent on the sequence of jobs on and the corresponding part of the database are illus-
the machines. Usually, there are two types of trated in Appendix A), together with an application
changeover times in the weaving phase, the total example. The relation of YFADI with two well-
and the partial ones, depending on the types of established production control systems, MRP-II and
Optimised Production Technology (OPT), is illus- system characterized by linear input-output relations.
trated in Section 5. Finally, concluding remarks are In addition to the development of decision support
given in Section 6. systems, various expert systems have been also de-
veloped for production scheduling and planning (see
for example [16]). Studies on the evaluation of these
2. Literature review approaches reveal that they have been developed
mainly to perform certain scheduling/planning func-
Production/operations management has been the tions just as good as humans do, with considerably
focus of a wide literature covering all aspects of greater speed and less human effort [17]. These
planning and control of industrial processes [3-71. systems can generally demonstrate greater consis-
Most of the related work has been based on mathe- tency, which is certainly a worthy objective. How-
matical analysis and traditional Operational Research ever, in order to be really helpful in real production
methodology. The advent of the information technol- applications, they must have the ability to adjust to
ogy has given rise to new approaches based on direct new problem environments and improve their knowl-
involvement and interaction of the user when apply- edge state. In order to implement advanced systems
ing respective decision aids in the form of software in the area, key factors seem to be the ability of
tools. Thus, in recent years, a lot of work has been enumerating alternatives before changing the prob-
done in the area of Decision Support and Knowledge lem description, and the successful employment of
Based Management Support systems with applica- embedded algorithmic knowledge. Classical OR ap-
tions in the scheduling of medium and large scale proaches have a lot to contribute to this last point
Make-To-Stock (MTS) and Make-To-Order (MTO) (see for example [IS]).
companies [8-121. Most of the research in the area Some of the special problems of the textile indus-
has certainly been aimed at the first category. This is try discussed above have been addressed by special-
due to the fact tha.t systems developed for MTS ized algorithms, based on graph theory [2]. However,
environments are usually reckoned to be also appli- they strictly distinguish MT0 and MTS environ-
cable to the MT0 ones. Differences in the require- ments, and are not applicable to a hybrid case. In
ments between the two categories are extensively addition, their application to a multi-phase produc-
presented in [13], focusing on the application areas tion line is not reported. As it is made clear in [2],
of production scheduling, capacity control and set- these algorithms allow sequencing of jobs in the
ting of delivery dates, and discussing the issue machines only if the jobs succession characteristics
whether the availablla research can meet the needs of are not complex, and are rather inefficient in terms
the MT0 sector. However, the distinction made in of computation time and data size. In order to reduce
[13] between the above two categories of companies the complexity of the scheduling problem, multi-
is sharp. As argued. above, this is not the case in phase production systems are often decomposed in
textile production systems. separate production units, and different types of
DICTUM is an interactive model-based decision control are introduced [19,20]. For example, [19]
support system, worth mentioning for the analysis distinguishes between goodsflow and production
and synthesis of large-scale industrial systems unit control, which concern planning and control
[14,15]. The system has been developed in order to decisions on the factory and the production unit
primarily meet the needs of chemical production level, respectively. Coordination of the production
systems. Besides elements such as an information process, through the production units mentioned
system consisting of data banks and database man- above, is the basic problem in these systems; it may
agement systems, it also includes a flexible model refer to different production phases, specific types of
generating system flar formulating system models, a jobs, inventory levels, etc.
simulation and muhicriteria optimization system for Work on the development of an on-line environ-
evaluation of consumed resources, a sophisticated ment for the manager in the textile industry has been
user interface and modules for report generation. It is reported in 121,221. Again, it addresses only one
argued that DICTUM is applicable for any complex phase of the production line, namely, the weaving
130 li.1. Karacapilidis, C.P. Pappis/ Computers in Industry 30 (1996) 127-144
phase. In this work, previous methods of work allo- * User-friendliness: Keeping in mind that the tex-
cation in weaving have been investigated and im- tile industry in Europe mainly consists of small
provements are suggested. The main motivation be- and medium size enterprises and employ mostly
hind the development of such an environment is that persons with limited computer education, the sys-
a computer simulation of the “weave room” opera- tem requires limited, in time and cost, training of
tions can lead to an efficient real-time decision mak- them.
ing system. The software of the system consists of three parts:
the Database Management System (DBMS), the
Model Based Management System (MBMS) and the
3. The YFADI DSS
User Interface. The commercially available Oracle
YFADI, meaning weft in the Greek language, is a RDBMS has been used in our implementation. Its
decision support system that has been developed for main advantages, in comparison with a third genera-
the production planning and scheduling of a Greek tion language, are the easiness of model develop-
textile industrial plant. In its first release it covers the ment, the modular design and implementation, inter-
operations of warp making, starching and weaving operability within the widely applied operation sys-
(the shaded area in Fig. l), but a future release will tems (i.e., DOS, UNIX, etc.), the unlimited number
cover all the production phases of the industry, that of records, the possibility of definition of variable
is, from yarn spinning to the final sewing of clothes. length fields that results in lower system memory
Its main objective is to provide the manager with requirements, and the encouraging results of previ-
efficient production management tools, applicable to ous systems developed on it. The MBMS has been
the multi-phase production of a variety of products. developed by the research team involved in the
The alternative production plans, provided by the project and includes all the algorithms and models
system, help the user to make decisions about the needed. It is written in mixed C and SQL program-
production rates for each product [23-251. YFADI is ming languages, with the aid of the Pro * C tool of
characterized by: Oracle RDBMS. The system requires CPU capable
. “Openness”: The system has been developed of supporting Oracle (i.e., IBM 80386, HP, VAX,
taking into consideration the characteristics of the etc.). The User Interface, in the first release, has
textile industry, with its mixed manufacturing been designed using exclusively Oracle tools. Fur-
type of process, and the variety of the type of the thermore, YFADI may either be a stand-alone sys-
final product (cloth, textile or warp). Several tex- tem or consist of a server connected via a network
tile enterprises have been contacted, and the re- (Ethernet has been selected) with simple terminals,
lated users’ requirements have been identified depending on the needs of the specific user enter-
before the application of the system to the spe- prise.
cific textile enterprise. It was among our main In order to better analyze the user needs and
objectives during the development of the system coherently develop the system, the MBMS was parti-
to keep it “open” for further applications and tioned into the following eight modules (see Fig. 2
extensions. This has been achieved via the proper illustrating the “backbone’ ’ of their interdependen-
identification of the “objects” involved (e.g., ties), interrelated via the Oracle DBMS:
machines, products, shifts, setups, personnel, - Forecasting
planning horizons, etc.) and the appropriate de- - Orders Processing
sign of both the Database and the Model Based - Aggregate Production Planning
Management System described explicitly in the * Master Production Scheduling
sequel. Special attention has been also paid to the - Material Requirements Planning
modelling of the multi-phase production. * Inventory Control
- Rapid and eficient data interchange in order to - Purchasing
cope with frequent changes in production sched- - Work in Progress
ules and the remoteness of the sites where the In particular, the Forecasting module makes the
various operations often take place. short, medium and long term forecasts and measures
N.I. Karacapilidis, C.P. Pappis/Computers in Industry 30 (1996) 127-144 131
available in the right time. Operation Sheets and ties (EOQS) are determined and the size of lots for
BilJs of Materials are employed. The module is batch production is evaluated. The module is updated
closely collaborating with the MPS module as it is about the “reserved” inventory by the MPS, and
made clear in the next section. Using a backward calculates the actual inventory status by obtaining
procedure, the MRP module defines the requirements daily (or, periodically, upon user’s wish) data from
in intermediate products and, finally, in raw materi- the Work in Progress and Purchasing modules. The
als in order to fulfil the production schedules. By main outputs here are spreadsheets concerning the
aggregating the material requirements for each pro- safety stocks, re-order points, EOQs for the “made-
duction order, MRP derives analytical schedules of to-stock” products, available inventories of raw ma-
what is needed (both quantities and due dates). The terials and finished products, in-progress and on-order
main inputs of the module are: the Production inventories. Such reports can be easily classified per
Schedules produced by MPS; the Bills of Materials, kind of product, supplier, size, quality requirements,
that are available in the Database and, the available place of storage, usage, date of entry, availability,
stocks that are provided by the Inventory Control etc.
module. The MRP module also collaborates with the The Purchasing module deals with the evaluation
Inventory Control and Purchasing modules providing of alternative schedules for the supply of the neces-
information concerning quantities of materials al- sary materials, considering various cost elements and
ready available and ordered, respectively. quality requirements. The module is fed with data
The Znventoly Control (IC) module deals with the from the MRP, concerning the scheduled materials
management of the inventory of each product. Atten- requirements; the IC module, concerning the avail-
tion is given both to MTS and MT0 products. Safety able stocks and the Database, concerning costs and
stocks, re-order points and Economic Order Quanti- lead times for alternative suppliers. The module
specifies the best placement of orders. Algorithms for weaving. These orders feed the MPS module,
have been developed for the appraisal of various where the scheduling of the above phases takes
suppliers, combined with possible quantity discounts. place. The module primarily takes into account the
After the orders have been placed, the module moni- available capacity determined by the APP module.
tors their progress. Every time an order arrives, the The user may specify the desired capacity levels and
IC module is updated. Reports about orders in preferences about the set of orders. He can also
progress, orders received or delayed, and order cost consider alternative scenarios, produced by the MPS
and quantity are solme of those produced by the module, and relate them to the available capacity,
module. determined by the APP module for different policies
The Work in Progress module, based on data on regarding subcontracting, number of shifts, etc. The
the current production situation concerning each work production orders are converted into purchasing or-
center, reports on the progress regarding the imple- ders corresponding to the requirements for yarn, both
mentation of schedules. The module receives infor- for warp and weft, via the MRP module. More
mation about the production status of each work explicitly, in each phase the MPS procedure pro-
center and the progress of each order, and compares duces a schedule that feeds the MRP module, which
them with the scheduled ones (usually on a daily in turn, feeds the MPS module of the previous,
basis). As a next step, the IC and the MPS modules according to the sequence of phases a cloth is con-
are informed about the actual production situation structed, phase.
and any eventual deviations. The user may alter Working first in a “backwards scheduling” way
previous production schedules, through the MPS the scheduling procedure starts from the weaving
module, taking these deviations into account. Two phase. After retrieving the set of the orders concem-
main reports are available: one about the progress of ing weaving from the database, the system interacts
the orders and another about the progress of each with the user in order to define priorities of the jobs,
work center (including machines and personnel). Ag- the scheduling policy and the rules of sequencing
gregated data are also produced, concerning devia- (Fig. 4). As described in Appendix A, the allowable
tions for long term production periods, aiming at values for the status of a job are:
adjusting the corresponding parameters of the MPS Scheduled but not in-progress;
module, in order for the latter to be more effective in Unscheduled;
the future. In-progress, and
Finished.
The system asks the user to specify whether he
4. The master production scheduling module wishes to schedule only the unscheduled jobs, i.e.,
jobs that are considered for scheduling for the first
The MPS module is at the core of the MBMS. For
a predefined time horizon, usually between 3 and 6
months, it helps the manager to determine the exact
Priorities and Scheduling Policies Specification
quantities to be daily produced and the correspond-
ing jobs’ sequencing and machine loading. As men- A. Consider:
1. Only the unscheduled jobs
tioned above, the variety in the form of customers’ 2. Every job that its processing has not
orders (clothes, textiles or warps) and the multitude started yet
of phases in the textile industry make scheduling not E. Priorities specification to be made:
1. By the user
an easy task. Fig. 3 illustrates the scheduling proce-
2. By the system
dure implemented in our system. It refers to the
C. For the jobs with the same priority consider
weaving, starching and warp making phases. as tie-braker:
The customers’ orders concerning textiles requir- 1. Their due date
2. Their release date
ing weaving, from the OP module, and the corre-
sponding estimates for the demand, from the Fore-
casting module, are cumulated into production orders Fig. 4. Specification of priorities and scheduling policies.
134 NJ. Karacapilidis, C.P. Pappis/Computers in Industry 30 (1996) 127-144
machine job job job trans. machine machine slack job production scheduled
code orocess setuo time available code time process volume qty
WMl 300 10 3 23/06/94 12:oo WMl 35
(a) (b)
Cc) Cd)
qty(m), job process(in’hr), job setup(hrs), job transportation time (hrs). slack time (hrs). prcduction volume (m).
WMI 23/06/96 12:OO 25/06/96 12~30 10,500 NM1 23/06/96 12:00 25/06/96 12:OO 10,500
WM3 20/06/96 12:00 22/06/96 09:OO 4,100 NM3 20/06/96 12:OO 23/06/96 1600 10,100
WM4 21/06/96 17:OO 25/06/96 12:OO 15,400 NM4 21/06/96 17:OO 25/06/96 12:00 15,400
WMl 0 16/06/96 0900 2.5/06/96 12:oO 40,000 NM10 16/06/96 09:OO 24/06/96 12:00 34,000
(a) before schedule improvement procedure (b) after schedule improvement procedure
that the machine is available and the due date) is number of rolls from the warp making machines:
calculated (see Fig. 5(b)). The algorithm takes both
starching job start date:
setup and transportation times into account. In our
example, the total quantity that may successively be warp making jobs due date:
I I
scheduled to the related jobs is 65,900 meters and,
therefore, 4,100 meters remain unscheduled. The Fig. 8. Example data for the scheduling of the warp making phase.
system identifies frorn the database the machines that
can process the job A.10.023.00, and retrieves the Finally, Fig. 8 gives an example of the “creation”
appropriate data (Fig. 5(c)). WM3 and WM5 are the of the appropriate jobs in the warp making phase. In
only candidate machines and their maximum produc- our case, the maximum allowable number of cones
tion volume is also calculated (Fig. 5(d)). WM3 is was 672. Thus, for the production of a warp consist-
selected by the system for the production of the ing of 6,476 threads, six rolls of 648 threads each
remaining 4,100 meters, as it can produce the maxi- plus four rolls of 647 threads each have to be
mum volume. The set of decisions (i.e., production scheduled in the warp making machines. The length
orders) concerning the scheduling of the job are of each roll from the warp making phase is equal to
demonstrated in Fig. 6(a). As one can see, the finish the length of the roll required in the starching ma-
times of three out of four job parts coincide. A basic chine. Note that the due date of all ten rolls of the
constraint in our case study was that no more than warp making machines is equal to the start date of
two job parts were alllowed to finish on the same day the corresponding job in the starching machine. In
(in order to avoid co-occurring changeover times). this case, job setup and transportation times are
This is achieved by the application of the schedule considered to be very small and, consequently, negli-
improvement procedure, which produces the outputs gible.
illustrated in Fig. 6(b).
Fig. 7 summarizes the results of the scheduling of 4.2. Schedule procedure
the starching phase for the last production order in
the weaving phase (concerning WMlO). Since the The MPS module of the system tries to consecu-
maximum length of the warp in WMlO is 10,000 tively schedule related jobs in order to achieve
meters the required number of starched rolls is 4. minimization of the sum of their changeover times.
The machine does not require all of them simultane- Jobs belonging to the same category have identical
ously, thus four jobs, corresponding to four starched relationship indices and require only partial
rolls of 8,500 meters each, are “created”. These changeover times, while jobs from different cate-
jobs have then to be scheduled in the starching gories require total changeover times, between their
machine. consecutive processing. The current status of the
database is traced in order to retrieve the relationship
finish indices of both in-progress and scheduled jobs, and
time
the corresponding machines that these jobs have
WMIO 16/06/96 09 00 24/06/96 12:OO
t----- I been already allocated to (WMl, WM4, and WMlO
max warp: jobs created: in the example above). For each machine in which a
job process: Part No. due date
job J has been scheduled or is in-progress, the
production volume for the period between the sched-
number of roll 1 16/06/96 09:oo
2 20/06/96 06:OO uled finish time of J and the due date of the job the
roll length: 3 21/06/96 16:00
4 23/06/96 0200
system considers for scheduling is calculated. Addi-
I I I tionally, the unscheduled jobs are grouped according
qty(m),Productionvolume(I@,maxwarp(m),job process(nvlv),roll lengm(m). to their relationship indices. The consecutive
Fig. 7. Example data fca the scheduling of the starching phase. scheduling of jobs in each of these sets is cost
136 N.I. Karacapilidis, C.P. Pappis/Computers in Industg 30 (1996) 127-144
effective, but may lead to violations of customers’ starched warp is fed with both the orders derived
requirements (i.e., due dates). This can be balanced from the requirements of the weaving machines and
through interactions with the customers (influencing the (usual) customer orders. In our case, due to the
the OP and APP modules) before finalizing the horizontal integration of the enterprise, jobs defined
production plans. The algorithms for the minimiza- by the warp scheduling procedure are characterized
tion of the changeover times are analytically de- by maximum priority. Consequently, during the
scribed in Step 4 of the MPS procedure in Appendix scheduling of the starching and warp making ma-
A. The machine loading is done in such a way that chines, jobs that have been ordered by the customers
the number of the machines required is minimized, are less favoured than the ones determined by the
while the current due dates’ requests are attained. above procedure. The scheduling of the starching
After the above matching procedure of jobs, some phase is done taking into account the priority of the
parts of them may have not been scheduled yet. This jobs, primarily, and the jobs’ due date, secondarily
case was made clear with the example above. In this (see again Fig. 7 in the example above).
case, the set of machines capable to process the job Each warp making machine is characterized by a
is considered and the unscheduled part is scheduled maximum allowable number of cones (or bobbins) of
following the machine loading procedure for the yarn in its corresponding bobbin stand. The jobs
non-related jobs (Step 6 in Appendix A). The ma- concerning warp making are characterized by a vari-
chine loading of the non-related jobs, for which there able called warp density, that is the number of the
are no related jobs in-progress or scheduled, is quite threads in a warp, and by the maximum number of
similar with the above procedure. The basic differ- cones (6,476 and 672, respectively, in the example
ence is that, in this case, job setup stands for the total above). A special procedure has been developed for
changeover time. the specification of the feeding of the starching
Because of the large changeover times during the machine. This is essential for the minimization of the
weaving phase, avoiding too many set-ups during the setup times. As in the case between the phases of
same day is a special requirement in the textile weaving and starching, the part of the system
industry. In our case study, no more than two database concerning orders for unstarched warp is
changeovers during the same day were allowed. The fed with orders derived from the requirements of the
schedule improvement procedure takes the schedule starching machine (which keep the same priority
determined so far as input and, by reallocating work index that they had in the starching phase) and usual
loads among the appropriate machines, leads to a customer orders. The starching machine processes
schedule that conforms with the above constraint the rolls produced by the warp making machines in
(see Fig. 6(b) for the example above). The corre- parallel. Thus, the due date of all jobs in a warp
sponding algorithm is analytically presented in Step making machine that correspond to a job J in the
7 of the MPS procedure in Appendix A. starching machine, is set equal to the estimated start
As made clear in the introduction, a weaving date of J (10/06/96 15:OO in the example above).
machine is fed by rolls of starched warps and the Finally, due to the fact that, in our case, each job for
necessary weft, that crosses the yarns of the warp. unstarched warp can be processed at a specific warp
Due to the fact that the machine does not require all making machine, the scheduling of these jobs is done
the rolls simultaneously, a warp scheduling proce- similarly to those of the starching phase (the system
dure has been developed in order to specify a sched- is able to consider the general case though).
ule for feeding the weaving machines. This schedule As it has been made clear from the above, the
will be “translated” into job orders for the starching MPS and MRP modules in YFADI are tightly inter-
and the warp making phases. The maximum allow- related (see Fig. 3). We should mention here that the
able length of a warp (10,000 meters for the WMlO distinction between them as two separate modules is
in the example above) is defined according to the only conceptual. The only motivation for that is the
dimensions of the roll that a weaving machine may adoption of the MRP concept as a “standard” proce-
accommodate and the kind of the associated yarn. dure, independently of the type of the production
The part of the database concerning orders for system.
N.I. Karacapilidis, C.P. Pappis/Computers in Industry 30 (1996) 127-144 137
5. Relation to MRP-II and OPT production con- available into account. This kind of iterations be-
trol systems tween backward and forward scheduling, combined
with the interactions of APP and MPS modules
This section aims at providing an overview of the discussed above, help the user to consider alternative
relation of the YFADI DSS with the production scenarios and, consequently, to refine his plans and
scheduling concepts of two well-known computer- make the final decisions.
ized production control systems, MRP-II (Manufac- MRP-II systems do not take capacity constraints
turing Resources Planning) [26,27] and OPT (Opti- into account when processing the requirements from
mised Production Technology) [26-291. Their archi- the master schedule. Thus, it is quite possible that
tecture is illustrated in Fig. 9 and Fig. 10, respec- the capacity requirements exceed capacity availabil-
tively. ity on a resource for some periods of time. It may be
MRP-II systems generally use backward schedul- possible to increase capacity in some way by em-
ing. OPT, on the other hand, uses a combination of ploying more people, buying new machines or sub-
forward and backward scheduling techniques (Fig. contracting, but in the short term this may not be
11). Non-critical pre-bottleneck operations of an or- possible. In this case MRP-II usually attempts to
der are backscheduled, from the scheduled start time replan the master schedule to ensure that there are no
of the bottleneck operation, using a procedure similar capacity overloads. This process is often described as
to MRP-II. YFADI makes no distinction between loading to infinite capacity or as capacity require-
critical and non-critical resources. Its scheduling ments planning. OPT, on the other hand, deals with
technique is primarily based on a backward schedul- this problem by avoiding the scheduling of resources
ing technique, since the due dates of the jobs are other than bottlenecks. In a finite loading procedure
taken into account. Following the scheduling proce- capacity is given and the schedule is treated as a
dure dictated by the special characteristics of the variable. YFADI, in its first version, has been specif-
industry, after the specification of the work loads in ically designed for a textile industrial unit, the policy
each machine, a forward scheduling technique is of which is to accept every customer order because
adopted, taking the earliest date that each machine is subcontracting is easily available and quite prof-
itable. This may be interpreted as scheduling with a fairly well defined bottleneck for its principles to
infinite capacity. As discussed in the previous sec- be valid. In many plants the bottleneck is not clearly
tion, the MPS module of YFADI schedules consider- defined, and manufacturing contingencies may cause
ing finite capacity, as it has been determined via the it to wander within the timescale of an OPT sched-
APP module. However, what has been adopted in ule. The second objection is concerned with the
our implementation is that all orders are primarily tightness of the OPT schedules, which must be pre-
considered for a draft production plan from the MPS cisely adhered to if the plan is to retain its integrity,
module. Successive interactions of the latter with the and make little allowance for any random interrup-
APP module refine the final plans, incorporating tions. OPT gives no guidance on schedule recovery.
adjustments of the number of machines in operation These disadvantage does not exist in YFADI, be-
and/or work shifts for a specific period, as well as cause the schedules produced by it are not tight. The
about the volume to be produced by subcontractors. system enables the user to interact at various points,
At this point, YFADI works similarly to MRP-II in order to achieve fine tuning of schedules. The user
systems. may simulate “what-if” questions and create alter-
The OPT system, via its Bruin module, comprises native selections. Additionally, the Work in Progress
a non-interactive algorithm that determines process module has been designed in order to report the
batch sizes, production sequences and buffer stocks progress regarding the implementation of schedules.
for the critical resources, based on their capacity The user may consider possible deviations from
limitations, to maximize throughput. On the con- goals, and rerun the remaining plan through the MPS
trary, the scheduling procedure of YFADI interacts and APP modules.
with the user in order to specify priorities and YFADI makes, as in OPT, an implicit distinction
scheduling policies at each production phase. YFADI between transfer and process lot sizes, allowing both
also avoids the two main criticisms that OPT faces of them to be variable. This is described in the
so far. The first is that OPT relies on the existence of scheduling procedure of the weaving machines,
where not all the starched rolls are required simulta-
neously, as well as in the scheduling procedure of
the starching machine, where the complete set of the
corresponding unstarched rolls from the warp mak-
ing machines has to be provided before the starching
procedure starts.
Forward
scheduling
Another specific characteristic of YFADI is the
Backward way that its MPS and MRP modules operate and
scheduling collaborate. Due mainly to the special characteristics
+
Backward of the industry, the analysis of the components of the
scheduling
final product, i.e., the cloth in our case, that have to
Component 1 Component 2 be produced by the enterprise is effected through the
Fig. 11. Scheduling in OPT (from [27]). MPS module following a capacitated scheduling pro-
NJ. Karacapilidis, C.P. Pappis/ Computers in Industry 30 (1996) 127-144 139
cedure. Further analysis of the requirements of a ments, such as the chemical industry. Implementa-
cloth in terms of requirements for starched rolls, and tion issues such as the interrelations of its modules,
of the latter in terms of requirements for unstarched the existence of multiple machines per phase, various
rolls of warp, are typical examples of such a cooper- planning horizons and production requirements, can
ation. MRP in YFADI uses the same bill of materials be adapted to any characteristic and/or scheduling
structure as in the MRP-II and OPT systems. Gener- policy of such a company. There are two basic
ally speaking, the scheduling procedure in YFADI reasons that advocate it. First, the coordination of the
has been developed being oriented to the textile system’s modules through the aid of the database;
industry and, therefore, is superior to other, general- Oracle plays a central role here, supporting a proper
purpose, systems in the way that it treats the existing modular design and implementation environment.
specific characteristics mentioned above. Secondly, the object-oriented approach during the
implementation of the system; the set of objects
identified and used for the specific case study can be
6. Conclusion easily modified in order to cover the needs of a
similar firm.
The objective of this paper has been to present
work done on designing a DSS for the production
management in the textile industry. After a short Acknowledgements
discussion of the model based management system,
the paper gives a comprehensive analysis and syn- The authors thank the anonymous referees for
thesis of the MPS procedure in such an environment. their useful suggestions and comments on the struc-
This procedure has been developed by taking into ture and contents of earlier versions of this paper.
account the specific features of the industry as well
as some particular methods and heuristics that man-
agement adopts. A particular example has been pre- Appendix A
sented, which covers most of the scheduling parame-
ters. Finally, the system has been related to two A.1. The system database
well-known production management systems, MRP-
II and OPT.
The database of the MPS module is composed of
YFADI has been developed aiming at inventory
the following tables:
reduction, increased -productivity, improved customer
(i> The table mpsjobs, with the fields:
service and control a’f the business in a textile indus-
trial unit. The phasles covered (weaving, starching job-code The code of a job; it is of al-
and warp making) are the most difficult ones in phanumeric type.
terms of scheduling. The system has been integrated product_code The code of the specific product;
in a structured form, oriented by the textile manufac- it is of alphanumeric type and
turing process phases. Two particular features of characterizes the operation type
YFADI are that: (i.e., weaving, starching or warp
- a production order can be split up into a set of making).
jobs which is then assigned to multiple parallel stv The quantity of the specific job
machines; (in meters).
* all customer orders are accepted and the available job-in The date, on which the order re-
capacity is adapted accordingly, basically due to lated to the specific job was re-
the ease of subcontracting. ceived.
Both features may characterize production units in job_dd The due date of the job.
the textile and other industries as well. The system is job-start-date The date, on which the process-
certainly applicable to them, and especially to those ing of the specific job has been
characterized by multi-phase production environ- scheduled to start.
140 N.I. Karacapilidis, C.P. Pappis/Computers in Industry 30 (1996) 127-144
job-end-date The date, on which the process- jobgrocess The speed of the machine
ing of the specific job has been (“picks” or meters per hour). It
scheduled to be completed. varies depending on the machine
priority The priority index of the job; it is and the kind of the product.
of numeric type. job_trans The transportation time of a job
job-status The status of a job; the allowable from the specific processing cen-
values are: S (scheduled, but not ter to the next one.
in-progress), U (unscheduled), I prod It indicates the quantity of the
(in-progress) and F (finished). specific product that can be pro-
relation A number indicating the related duced at the specific machine for
jobs, that is jobs that the sched- the time interval between the date
uler wishes to be processed that it is free and the due date of
(mainly during weaving) consecu- the job.
tively in order to minimize the
setup time. All related jobs are (iii) The table machine-status, with the fields:
assigned the same number. machine-code As above.
status The status of the machine. Al-
lowed values are B (busy), F
(ii) The table job-machine, which relates the jobs (free), 0 (out of order) and M
to the various processing centers (i.e., machines ca- (maintenance).
pable to process the specific job). It consists of the product_code As above. In the case that the
fields: field status is F, it is assigned
null.
job-code As above. job-code As above.
machine-code The code of a machine capable to start_mach_stat The date that the specific status
process the specific job. Usually, starts.
there are more than one records end_mach_stat The date that the specific status
with the same job-code indicat- ends.
ing the alternative processing
abilities of the system. It is of A.2. The MPS procedure
alphanumeric type and character-
izes the type of the machine with This section describes the MPS module of the
the first two letters (i.e., WE for system, in pseudocode. It should be noted that the
weaving machine, ST for starch- first version of YFADI covers the operations of
ing machine and WA for warp weaving, starching and warp making.
making machine).
Step I: Retrieve all orders concerning weaving from
job-setup The setup time for the specific
database.
job and the specific machine (in
hours). It is used when the previ- Step 2: Define:
ous job processed at the specific - priorities of the jobs,
machine has not the same rela- 9 jobs to be scheduled, (i.e., only unscheduled
tion with the specific job. jobs or both scheduled, but not in progress, and
job_setup_rel The setup time for the specific unscheduled jobs),
job and the specific machine (in * rules of sequencing (i.e., due date or FIFO).
hours) when the previous job pro-
cessed at the specific machine has Step 3: For every job with code job-code, quan-
the same relation with the spe- tity qty and relationship index relation find from
cific job. tables machine-status and mpsjobs
NJ. Karacapilidis, C.P. Pappis/ Computers in Industry 30 (1996) 127-144 141
/ * The function ceil(x) produces the next integer Step 10: Order jobs concerning starching by:
which is greater than x. */ first key the jobgr (increasing), and
Update the tables machine_stutus and mpsjobs. second key the job_dd (increasing).
From the table machine-status find the maximum
Step 8: For all jobs concerning weaving end_mach_stat,
( for every machine scheduled to process it let free, of the starching machine, where status
{ find = ‘B’.
(i) job-start-date; / * The maximum end-much-stat, where status
(ii> qty; = ‘B’, is identical with the date that the machine
(iii) ma-warp, and is free. */
(iv) jobgrocess. For every job {
/ * The mu._warp is a variable indicating the update tables mpsjobs and machine-status with
maximum allowable length of a warp and it is { job-start-date = free;
defined according to the dimensions of the roll job-end-date = free + (job-setup + (qty /
that a weaving machine may accept and the kind jobgrocess) + job_trans)/24;
of the yam. Thus, this variable defines the jobs job-status = ‘S’ }
concerning starching and warp making, since a set
weaving machine does not require all rolls simul- free = free + (job-setup + (qty /job-process)
taneously. * / + job_truns)/24 }.
Compute:
numb = ceil( qty / mu._wurp); Step I I: For all jobs concerning starching
posot = qty / numb. { find
For k = 1 to k I numb, with Step 1 (i> job-start-date;
{ create records concerning starching in the (ii) qty;
table mpsjobs, with (iii) warp-dens, and
qty = posot; (iv> jobgrocess.
job-status = ‘U’; Compute
if k = 1 then job_dd = job-start-date number = ceil( warp-dens / numcone);
else job_dd = job-start-date +
+ (k - l)( posot / jobgrocess)/24 + / * The field numcone indicates the maximum
job_trans, + job-setup; allowable number of cones of yam in a warp
job_pr = 0 ). making machine. In this application, numcone =
/ * The jobs created in this step are characterized 672. */
by jobgr = 0, since they are of maximum prior-
ity. Consequently, the jobs concerning warp mak- amount = ceil( warp-dens / number);
ing and starching that have been ordered from the
various customers are characterized by jobgr / * The jobs concerning warn making are defined
greater than zero. by warp-dens, that is the number of the threads at
A variable x/24 is used for the addition of hours a warp, and by numcone. For example, for a warp
to a date type (DD/MM/YY HH: MI, i.e., consisting of 6,476 threads and if numcone = 672,
day/month/year hour : minute) variable. * / the manager has to schedule 6 rolls with 648
I I. threads and 4 rolls with 647 threads for the warp
making machines. This is essential for the mini-
Step 9: Define: mization of the setup times of the above machines.
* priorities of the jobs on the starching machine; */
- jobs to be scheduled on the starching machine
(see Step 2). ol= amount * number;
dtf = 01 - warp-dens.
N.I. Karacapilidis, C.P. Pappis/ Computers in Industry 30 (19961 127- I44 143
For k = 1 to k 5 dif, with Step 1 [3] ES. Buffa and R.K. Sarin, Modern Production/Operations
{ create records concerning warp making in Management, John Wiley and Sons, New York, 1987.
[4] K.N. Dervitsiotis, Operations Management, McGraw-Hill,
the table mpsjobs, with:
New York, 1981.
qty = qty of the corresponding starching [5] M.A.H. Dempster, J.K. Lenstra and A.H.G. Rinnoy Kan,
job; Deterministic and Stochastic Scheduling, D. Reidel Publ.
job-status = ‘U’; Co., London, 1982.
job_dd = job-start-date of the corre- [6] C.C. Holt, F. Modigliani, J.F. Muth and H.A. Simon, Plan-
ning Production, Inuentories and Work Force, Prentice-Hall,
sponding starching job;
Englewood Cliffs, NJ, 1960.
jobgr = joibgr of the corresponding [7] S. French, Sequencing and Scheduling, Ellis Horwood, New
starching job, and York, 1982.
numthreads := amount-l }. [8] G. Doukidis, Fr. Land and G. Miller, eds., Knowledge Based
Management Support Systems, John Wiley and Sons, 1989.
[9] V. Lotfi and C.C. Pegels, Decision Support Systems for
Production and Operations Management, Irwin, Homewood,
/ * The field numthreads indicates the number of IL, 1986.
threads in a roll of ,a warp making machine. */ [lo] T.E. Morton and D.W. Pentico, Heuristic Scheduling Sys-
For m = 1 to m :Z number-dif, with Step 1 tems, John Wiley and Sons, New York, 1993.
[ll] T.E. Vollmann, W.L. Berry and D.C. Whybark, Manufactur-
{ create records concerning warp making in
ing Planning and Control Systems, Irwin, Homewood, 1988.
the table mpsjobs, with: [12] D.E. Brown and W.T. Scherer, eds., Intelligent Scheduling
qty = qty of’ the corresponding starching Systems, Kluwer Academic Publishers, London, 1995.
job; [13] L.C. Henry and B.G. Kingsman, “Production planning sys-
job-status = ‘U’; tems and their applicability to make-to-order companies”,
European Journal of Operational Research, Vol. 40, 1989,
job_dd = job-start-date of the corre-
pp. l-15.
sponding starching job; [14] A. Barnikow, U. Behrendt, K. Hartmann and M. Scharni,
jobgr = jo,bgr of the corresponding “DICTUM: Decision support system for analysis and syn-
starching job, and thesis of large-scale industrial systems. Part I: Components”,
numthreads = amount } Computers in Industry, Vol. 18, 1992, pp. 135-144.
[15] L. Dietzsch, U. Kroeller and K. Hartmann, “DICTUM:
I.
Decision support system for analysis and synthesis of large-
/ * Since the starching machine processes the scale industrial systems. Part II: Databases and industrial
rolls of the warp m,aking machines in parallel, the applications”, Computers in Industry, Vol. 18, 1992, pp.
due date of a job in a warp making machine is the 145-153.
job-start-date of the corresponding starching job. [16] M.S. Fox and S.F. Smith, “ISIS - A knowledge-based
system for factory scheduling”, Expert Systems, Vol. 1, No.
*/ 1, 1984, pp. 25-49.
[17] J.J. Kanet and H.H. Adelsberger, “Expert systems in produc-
Step 12: Keeping in mind that the table job-machine tion scheduling”, European Journal of Operational Re-
has only one record for each job concerning warp search, Vol. 29, 1987, pp. 51-59.
[18] R.M. O’Keefe, V. Belton and T. Ball, “Experiences with
making, that is, each job is processed at only one
using expert systems in O.R.“, Journal of the Operational
warp making machine, the scheduling of these jobs Research Society, Vol. 37, No. 7, 1986, pp. 657-668.
is similar with those concerning starching (Step 10). [19] J.W.M. Bertrand, J.C. Wortmann and J. Wijngaard, Produc-
tion Control: A Structural and Design Oriented Approach,
Elsevier Science Publishers B.V., Amsterdam, 1990.
[20] V.S. Tanaev, Y.N. Sotskov and V.A. Strusevich, eds.,
References Scheduling Theory: Multistage Systems, Kluwer Academic
Publishers, London, 1994.
[l] R.P. Olsen, The Textile Industry: An Industry Analysis Ap- [21] S. Jayaraman and R. Malhotra, “Weave room of the future.
proach to Operations Management, Lexington Books, Part I: Team approach to operations - A simulation study”,
Toronto, 1978. Textile Research Journal, April 1989, pp. 237-243.
[2] A. Guinet, “Textile production systems: A succession of [22] S. Jayaraman and R. Malhotra, “Weave room of the future.
non-identical parallel processor shops”, Journal of the Oper- Part II: Monitored data for real time resource allocation”,
ational Research Society, Vol. 42, No. 8, 1991, pp. 655-671. Textile Research Journal, May 1989, pp. 271-274.
144 N. I. Karacapilidis, C. P. Pappis / Computers in Industry 30 (1996) 127-144