+ All Categories
Home > Documents > Review of modelling techniques and tools for decision making in manufacturing management

Review of modelling techniques and tools for decision making in manufacturing management

Date post: 20-Sep-2016
Category:
Upload: kv
View: 213 times
Download: 0 times
Share this document with a friend
7
Review of modelling techniques and tools for decision making in manufacturing management K.V. Pandya Indexing terms. Business process re-engineering, CIMOSA, IDEF, input-output analysis, Modelling tools, Petri-nets, SSADM, SADT Technology asset management Abstract: The author reviews methodologies and modelling tools which have been developed for use in manufacturing systems. The strengths and weaknesses of the different tools are discussed, and examples of applications are given. An under- standing of general systems theory as identified by Checkland (1981) is assumed. The methodologies and modelling techniques discussed make the common assumption that systems concepts, such as organised interconnectedness and emergent properties, apply in the context of the problem domain. The modelling tools may be used to model business processes including: all the plan- ning and control processes, process technology and management and administrative. Following the review of the methodologies and modelling techniques a practical guide in the form of a table is proposed for business processes. 1 Systems methodologiesand modelling tools This paper assumes an understanding of general systems theory [l]. A ‘methodology’ is a set of principles of method which, when applied to the particular situation, guide the user to develop a method uniquely suited to the problem. Within manufacturing management these meth- odologies include SSADM, SADT, GRAI and are often used with formal methods, Austin [2]. A ‘modelling tool’ is a communication device used to aid the generation and classification of ideas, and/or to analyse the quality of a design. The modelling tools employed in manufacturing management include DFDs, IDEF, input-output analysis (tOA) and Petri nets. 1.1 Need for and advantages of methodologies and modelling tools A system, which maybe a manufacturing cell, a sched- uling system, or an order-entry system, is complex. It may be viewed as a hierarchy of subsystems and com- ponents which interact to define the uniqueness of the system. In analysing the system, consideration has to be given to 0 the position which the subsystem and components takes within the hierarchy 0 IEE, 1995 Paper 2095A (IZ), first received 13th February and in revised form 9th June 1995 The author is with the University of Strathclyde, Department of Design Manufacture and Engineering Management, James Weir Building, 75 Montrose Street, Glasgow G1 IXJ, United Kingdom IEE Proc.-Sci. Meas. Technol., Vol. 142, No. 5, September 1995 0 how the changes will affect the emergent properties exhibited by the system 0 how changes to a subsystem will affect the other sub- systems in the hierarchy 0 identification of the true underlying needs of the users, and to the nature of the problem itself 0 integrating the subsystems to give a global view. This arises from the current business ‘pressures’ such as shorter lead times, shortages of raw material and lower order quantities 0 getting it right first time. As asset technology becomes more expensive, the cost of making errors in the design of a system increases. The main problems associated with system projects include: excessive time and costs, and the failure to address the underlying needs of the systems users, Nicholls [3]. Systems methodologies address these prob- lems by establishing a framework which guides the analyst through the problem solving process. This process may be product design, managing asset tech- nology, or planning and control. Division of the process into clear steps enables more accurate costing and sched- uling of projects. Methodologies such as SSADM and SADT address the need for highly integrated systems by forcing a multidimensional view of the system. This requires a top-down approach to problem solving, forcing the system into the context of the hierarchy, and the problem into the context of wider problems. Modelling tools such as IDEFO, logic data structures (LDS) and data flow diagrams (DFDs) provide a stand- ard means of describing and analysing systems. This facilitates communication between developer and user, and between developers. It also simplifies understanding, modification and maintenance of the systems, ensuring good discipline. This is essential for development of the system in the current business requirements such as wide product variety, flexibility, decreasing life cycles and total customer satisfaction. Wu [4] identifies the primary func- tions of a systems development methodology as (i) to improve the efficiency of the systems develop- ment process (ii) to provide a means of measuring the quality of a systems design, such that informed decisions may be made and resources allocated accordingly The criteria used for assessing modelling techniques include user friendliness, probability, rigour, speed of development, and comprehensives, Martin [5]. However, the selection of a particular methodology and modelling technique is of less importance than an understanding and acceptance of, and adherence to the principles of general systems, Swann [6]. 37 I
Transcript
Page 1: Review of modelling techniques and tools for decision making in manufacturing management

Review of modelling techniques and tools for decision making in manufacturing management

K.V. Pandya

Indexing terms. Business process re-engineering, CIMOSA, IDEF, input-output analysis, Modelling tools, Petri-nets, SSADM, SADT Technology asset management

Abstract: The author reviews methodologies and modelling tools which have been developed for use in manufacturing systems. The strengths and weaknesses of the different tools are discussed, and examples of applications are given. An under- standing of general systems theory as identified by Checkland (1981) is assumed. The methodologies and modelling techniques discussed make the common assumption that systems concepts, such as organised interconnectedness and emergent properties, apply in the context of the problem domain. The modelling tools may be used to model business processes including: all the plan- ning and control processes, process technology and management and administrative. Following the review of the methodologies and modelling techniques a practical guide in the form of a table is proposed for business processes.

1 Systems methodologies and modelling tools

This paper assumes an understanding of general systems theory [l]. A ‘methodology’ is a set of principles of method which, when applied to the particular situation, guide the user to develop a method uniquely suited to the problem. Within manufacturing management these meth- odologies include SSADM, SADT, GRAI and are often used with formal methods, Austin [2]. A ‘modelling tool’ is a communication device used to aid the generation and classification of ideas, and/or to analyse the quality of a design. The modelling tools employed in manufacturing management include DFDs, IDEF, input-output analysis (tOA) and Petri nets.

1 .1 Need for and advantages of methodologies and modelling tools

A system, which maybe a manufacturing cell, a sched- uling system, or an order-entry system, is complex. It may be viewed as a hierarchy of subsystems and com- ponents which interact to define the uniqueness of the system. In analysing the system, consideration has to be given to 0 the position which the subsystem and components

takes within the hierarchy

0 IEE, 1995 Paper 2095A (IZ), first received 13th February and in revised form 9th June 1995 The author is with the University of Strathclyde, Department of Design Manufacture and Engineering Management, James Weir Building, 75 Montrose Street, Glasgow G1 IXJ, United Kingdom

I E E Proc.-Sci. Meas. Technol., Vol. 142, N o . 5 , September 1995

0 how the changes will affect the emergent properties exhibited by the system

0 how changes to a subsystem will affect the other sub- systems in the hierarchy

0 identification of the true underlying needs of the users, and to the nature of the problem itself

0 integrating the subsystems to give a global view. This arises from the current business ‘pressures’ such as shorter lead times, shortages of raw material and lower order quantities

0 getting it right first time. As asset technology becomes more expensive, the cost of making errors in the design of a system increases.

The main problems associated with system projects include: excessive time and costs, and the failure to address the underlying needs of the systems users, Nicholls [3]. Systems methodologies address these prob- lems by establishing a framework which guides the analyst through the problem solving process. This process may be product design, managing asset tech- nology, or planning and control. Division of the process into clear steps enables more accurate costing and sched- uling of projects. Methodologies such as SSADM and SADT address the need for highly integrated systems by forcing a multidimensional view of the system. This requires a top-down approach to problem solving, forcing the system into the context of the hierarchy, and the problem into the context of wider problems.

Modelling tools such as IDEFO, logic data structures (LDS) and data flow diagrams (DFDs) provide a stand- ard means of describing and analysing systems. This facilitates communication between developer and user, and between developers. It also simplifies understanding, modification and maintenance of the systems, ensuring good discipline. This is essential for development of the system in the current business requirements such as wide product variety, flexibility, decreasing life cycles and total customer satisfaction. Wu [4] identifies the primary func- tions of a systems development methodology as

(i) to improve the efficiency of the systems develop- ment process

(ii) to provide a means of measuring the quality of a systems design, such that informed decisions may be made and resources allocated accordingly

The criteria used for assessing modelling techniques include user friendliness, probability, rigour, speed of development, and comprehensives, Martin [5 ] . However, the selection of a particular methodology and modelling technique is of less importance than an understanding and acceptance of, and adherence to the principles of general systems, Swann [6] .

37 I

Page 2: Review of modelling techniques and tools for decision making in manufacturing management

2 Methodologies (a) DFDs are easy to use and therefore powerful com- 2.1 Structured sysrem analysis and design

methodology (SSADM ) SSADM is a procedural framework which was developed specifically for use in computer system development pro- jects. The structure and principles adopted by the meth- odology are of interest. This is because

(i) SSADM is a structured approach and provides a framework with which analysts can work. The develop- ment process is broken down into six stages as shown in Fig. 1, preceded by a feasibility study. Each stage is

stage 1 investigate current

system

stage 2 specif !cat ion of required

system -- stam 3 specification of technical

U--

Fig. 1 Sir stages o f S S A D M

divided into a number of steps, which are in turn com- posed of a number of tasks. Clear division of the develop- ment process, as adopted by SSADM, enables analysts to spend more time analysing, thus facilitating more accur- ate scheduling and costing of projects.

(ii) The methodology requires quality reviews to be conducted at the end of each stage. This provides a means of controlling the quality of the final design.

(iii) The methodology also requires analysts to consult the users at critical stages in the development process. This ensures that the final design will adequately address the needs of the users.

(iv) The methodology explicitly distinguishes between logical design, which is concerned with ‘what’, and phys- ical design which is concerned with ‘how’. This requires the ‘what’ to be completed before the ‘how’.

(v) The methodology requires analysts to consider function, data and event-oriented perspectives of the system, thereby encouraging the development of highly integrated systems.

SSADM makes use of three modelling tools: data flow diagrams (DFDs), logical data structures (LDSs). and entity life histories (ELHs) to provide function, data and event views of the system, respectively.

DFDs provide a static structured representation of the functions within a system, identifying the processes to be performed and the data/object interfaces between those processes. The notation used in DFDs, as popularised by DeMarco [7], and the constructs proposed by Cane [SI, are shown in Fig. 2. Advantages associated with DFDs include [3]

372

munication devices

problem (b) DFDs support top-down decomposition of a

dela? 7 publisher ,,/ purchase

address Orders orders.

orders batched credit status & & customers pending order

Fig. 2 Scenario of DFD 0 sourceidestination 0 function C datastore - dataflow ~~

(c) Levelling of DFDs help to control the size of dia- grams making them easier to understand

(d) DFDs are easy to produce, rough version can be drawn quickly and rapidly modified

DFDs are a less rigorous method of modelling than IDEFO, with no explicit rules governing either the number of process boxes which may be represented at a particular level, or the cross-checking of data/object flows across the levels of a diagram. Martin [SI suggest that DFDs are deceptively simple, and that their use can produce models that look right, but which on further and more critical examination are likely to be full of inconsis- tencies. DFDs should only be used to give an overview of the document/material flows within a system.

A logical data structure (LDS) diagram shows the entities within a system and the relationships between those entities and represents the simplest way in which the data required to perform the system processes identi- fied in the associated DFDs may be organised. LDSs are a useful tool for structuring databases, Wu [4]. Entity life history (ELH) diagrams show how the entity types associated with a system change over time. Such dia- grams can also become rather ‘messy’ in practice.

2.2 Structured analysis and design technique (SADTI

SADT is systems analysis and design methodology which makes use of a number of graphical and textual tools including activity diagrams, data diagrams, node lists and data dictionaries to represent the structure of the system being addressed. The noteworthy features of the method- ology include

(i) Top-down decomposition is used to enable the gradual introduction of detail, thereby controlling com- plexity, Banerjee [SI.

(ii) The methodology requires analysts to consider activity, or function, and data views of the system being addressed, thereby encouraging the creation of integrated systems.

(iii) The methodology requires a functional (or logical) model be created before considering the physical design (or implementation model) thereby forcing ‘what’ to take precedence over ‘how’.

(iv) The methodology also requires all decisions and comments to be recorded in written form, and includes rules governing the creation of such records.

I E E Proc.-Sci. Meas. Technol., Vol. 142, No. 5. September 1995

Page 3: Review of modelling techniques and tools for decision making in manufacturing management

Activity diagrams have an identical structure to those produced using the IDEFO modelling tool, with the exception that a subdiagram is included in the top-right hand corner of an activity diagram as a means of showing the position taken up by the subject of the diagram within the systems hierarchy. Data diagrams are the opposites of activity diagrams; a data diagram con- sists of data boxes linked by arrows representing activ- ities. A data dictionary is a formal definition of the data appearing within a system.

A node listing is a record of the node numbers and titles of the activity/data diagrams used to represent the structure of the subject system. Diagrams are assigned node numbers according to the node number of their parent diagram and the number of the activity box within the parent diagram which they represent. SADT has the advantage of being a tried and tested formula, Wu [4]. The methodology also has computer support in the form of software packages AUTOIDEFO and SPECIFX. The use of computer packages is beneficial in ensuring conformance to establish rules, and simplifies maintenance and modification of models.

2.3 Graphe a resultatis er acitities intercies (GRAI) The GRAI methodology was developed for the purpose of addressing the design of the management or decision making subsystem of manufacturing systems. The meth- odology considers the analysis and design phases of the systems development process, and in doing so employs two graphical tools, GRAIgrids and GRAInets, Chen vol .

GRAIgrids are used to identify the relative position of the various decision centres within the decision making system hierarchy, Carrie [ I t ] . A decision centre is char- acterised by three attributes: the function or task per- formed, the control horizon over which the decisions extend, and the frequency with which decisions are made.

A GRAlgrid consists of columns headed by standard functions, and rows headed by desired time horizons and periods. Decision centres are recorded within the relevant squares, and significant flows of information between the decision centres are shown by arrows. The transmission of decision frames between decision centres is shown in Figs. 3 and 4. GRAInets are used to examine the physical

H = 1) 2 pafi customer product P:Z3 months ’ I infwmation t kelcpmnt plan I lk HdJ5 P.1 day months i i n f o r m a i i m y customer ~ ~ ~ ‘ ‘ f ~ prig t

I

H.112 weeks shop P=1/2 hours schedul

Fig. 3 Example of usage oJGRAlgrid

and management activities involved in the decision making processes and the information required to make particular decisions. Decisional and process activities are represented by vertical and horizontal arrows, respect-

IEE Proc.-Sci. Meas. Technol., Vol. 142, No. 5 , September 1995

ively, with resource and information requirements being shown in the rectangles of the diagram.

The GRAI methodology involves two major phases. Analysis of the existing decision making system is per- formed using GRAInets and GRAIgrids to highlight

resource

to decide

r e d t 5

resources to do

results

Fig. 4 Example of usage ofCRAInets

inconsistencies and omissions. The design phase involves creating conceptual and structural models of the ‘to be’ system, with grids and nets being used to validate the latter. Finally, a functional specification is created using the structural model for guidance.

Wu [4] criticises the GRAI methodology for failing to require analysts to consider the technological, economic, or financial feasibility of their designs, and suggests that the standard functions shown in the GRAIgrids are not representative of the decision making activities performed by many businesses. It is also suggested that the GRAI methodology is geared towards the creation of central- ised systems which, given the need for rapid, timely deci- sion making, are undesirable.

3 Modelling tools

3.1 Integrated computer aided definition (IDEF) IDEF is a systems definition method which was devel- oped under sponsorship from the US Airforce by Soft- Tech Inc. [12]. The three main modelling tools provide function, information and dynamic models of the system. The tools are complimentary. A machine, for example, maybe viewed as being a resource in the dynamics model, a mechanism to perform an activity in the function model, and an entity to which data is attributed in an information model.

One of the major drawbacks of IDEF is the lack of cohesion between the types of models produced [13]. The system developers have to maintain such relationships in an informal manner, a task which becomes increasingly difficult as the size and complexity of the model domain increases. The IDEF family extends to IDEFl5. Some of these members are still under development. IDEF is thus a powerful tool to analyse, specify and design integrated manufacturing systems. It is suitable for modelling exist- ing as well as new systems. In this paper only IDEFO and IDEFIX are reviewed.

3.2 IDEFO IDEFO provides a static, structured representation of the functions within a system and describes the interactions between these functions. It is regarded that SADT is a general tool whereas IDEFO, based on SADT, is applied

313

Page 4: Review of modelling techniques and tools for decision making in manufacturing management

mainly in manufacturing. SADT was developed first and IDEFO is a derivative of it. The basic construction used in an IDEFO model is the function block linked with other blocks by inputs, outputs, the mechanism and con- trols. Links between the blocks may be either physical

WTL file order comDonent

part program I batch tools file raw material/

details update WTLfile

inform SS

download

make 2 of

part program

local scheduling database program

Fig. 5 diagram for decisions

Decisions by cell supervisor on receipt of new order: parent

3.3 IDEFIX IDEFlX provides information models of the system. Information models define the structure of the informa- tion needed to support the functions identified using the IDEFO. IDEFlX is based on entity relationships. An entity is something about which data is stored, such as a part of raw material or a customer order. Each entity has an associated set of attributes, IDEFlX [16]. The relationships between entity types maybe one-to-one, one-to-many and many-to-many. IDEFlX model of a cell controller is shown in Fig. 7 [14]. The IDEFlX mod- elling technique has drawbacks, which may be. the reason for it not been widely used in industry. These drawbacks include

(i) IDEFlX does not support composite entity types, and requires attributes to be single values of simple data types such as strings and numbers.

maintenance WTL component details file details

inform shop due dote cannot be met

make 2 of

file Fig. 6 Decisions by cell supervisor on receipt of new order: details of decisions

objects such as material (components or subassemblies) or information flow. A function is activated when an input, for example a new order request arrives, to the cell supervisor, Figs. 5 and 6 [14].

IDEFO is also a portable modelling tool, with the decomposition of functions into subfunctions. This pro- vides the opportunity to describe a system at any desired level of detail. Another feature of the IDEFO modelling technique is the explicit attention given to context, viewpoint, and purpose. ‘Context’ defines the position that the subject of the model takes up in the systems hierarchy, as with SADT. ‘Viewpoint’ concerns the perspective which the model adopts, and the ‘purpo5e’ establishes why the model was created.

IDEFO has been widely used in industry, with the resulting benefit that a wide user-base exists. Frank [IS] in describing the use of IDEFO to develop a generic, func- tional model of a cell controller, notes the following benefits of the tool:

(i) user friendliness (ii) computer support (iii) rigour and conciseness, with the existence of well

documented rules and procedures making the tool well suited to applications involving discussion and analysis sessions.

The static nature of the models produced using IDEFO may be the greatest failing of the modelling tool, requiring significant manual effort and interpretation to be conducted to identify those functions that should process a particular input and subsequently to verify their consistency.

314

(ii) Domain constraints describe the semantics of attributes associated with entity types. Constraints such that the salary of an employee cannot exceed that of his supervisor cannot be expressed using IDEFlX.

(iii) Using attributes to identify instances of entity types requires an attribute value to be assigned before an instance of an entity type can exist. In the real world records, such as purcase orders, can be prepared before they have their primary attributes, for example a pur- chase order number, assigned to them.

3.4 Inputloutput analysis This method generates the necessary ingredients required to give results. This method is considered to be simpler than IDEFO. A process is defined first, the outputs ident- ified and then all the inputs [17]. The outputs are classed as the needs which the system is trying to satisfy and the inputs are those things necessary for the process to gener- ate outputs. Two subsystems can be matched together by connecting the inputs of one subsystem to the outputs of the other. In a similar fashion a whole enterprise can be modelled as integrated subsystems, Parnaby [18, 281. This is a very powerful tool. IOA may be used as a design aid to clarify a problem and to analyse ‘before and after’ systems. Examples of its application include a plan- ning department’s decision making process, shown in Fig. 8 (Towill [19]) and analysis of production control func- tion (Berry [20]).

Benefits of the IOA include (i) the conciseness and completeness of system modelled, and (ii) structured analysis of complex systems being possible by linking simple subsystems.

I E E Proc.-Sci. Meas. Technol., Vol. 142, No. 5, September 1995

Page 5: Review of modelling techniques and tools for decision making in manufacturing management

Limitations of misinterpretation, included, and (iii)

Fig. 7 I D E F I X

the IOA include (i) the possibility of (ii) details of the process not being if feedback and parallel activities are

from an understanding of the biological nervous system. Nets are built on a large number of simple adaptable processing units which simulate the function of a neuron

work - to-I is1

part identification operation number operation time start time

scheduleslschedule

part

loadslworks to wor kcentre

A part identification WIC identification (Wr t name and Part manufactured (WIC name and WIC number) bylmanufactures number)

setup times reference number - quantity change over times

WIC before WIC after

-

i s made by sequence of requirement tested I 1 I

1 tools inspection I uses

part number inspection station

inspection type

W/C number part number

tool type tool l i fe tool location --

model for cell information

W/C number par t number

inspection required setup details part (or robot 1

needed then the analysis may be very detailed and in the brain. The units are interconnected in a manner complex. that can store experimental knowledge through examples.

They have the ability to take in information from the 3.5 Neural nets outside world (input layer), process it without a set of Apart from the symbolic modelling tools, business pro- rules and gve the desired outputs. cesses are increasing modelled by analytical tools, for Wu [21] has identified two main subsystems for example neural networks. Neural networks originated neural net applications: machine level and operational

pzzx--demand Annual forecast demand \ I Stock held

held between distributor and including factory)

Availability targets Ability to Supply orders on time

Recommended factory order Recommended stock target Recommended forecast order

\ Averaae demand /

n Perceived trend in demand Previously scheduled customer \ orders Minimum batch quantity constraints (tor when low order quantities placed ) Supplier performance

m a n chedulei

Factory liason via telephone

(and I memos

factory

produdion 3months hence

Forecast

orders lor 4 or 5 months ahwd to

foctory

Fig. 8

I E E Proc.-Sci. Meas. Technol., Vol. 142% No. 5, Septemher 1995

IOA for planning departments decision making process [ I91 375

Page 6: Review of modelling techniques and tools for decision making in manufacturing management

Table 1 : Suitability of modelling tools and methodology in manufacturing management and business processes

Methodology/ tools

SSADM SADT IDEF. IDEFO

IDEFlX GRAl: grid

nets Neural nets Petri nets Inputfoutput

analysis Jackson’s Warnier-Orr

Applications in manufacturing/business

processes

A B C D E

*** *** ** * * * ** ** * * * *.* *** *** ** * tt *ti t i l *t*

** *** *.* ** t

I* I** *.. ** * * * * tt. * * ** * ** * *. I. *** *** **

* * * * *I*

* * * f. ***

Key: *** most suitable ** may be used * not recommended

A -Traditional corporate-level planning and control (Business pro- cesses with planning horizons of > 1 month, e.g. technology acquisition, forecasting, marketing functions, long-term planning)

B -Traditional factory-level planning and control (Business pro- cesses with planning horizons of 1 week to 1 year, e.g. purchas- ing functions, production functions)

C -Traditional shop-level planning and control (Business processes with planning horizons of 1 day to 1 month, e.g. tools manage- ment, inventory control, scheduled maintenance, quality control)

D -Traditional cell-level planning and control (Business processes with planning horizons of 1 day to 2 weeks, e.g. short-term scheduling (dynamic), process monitoring, unscheduled maintenance)

E - Development of data for manufacturing/business process

level. It is considered that most applications were at machine level. It is only in recent years that decision making functions are being considered. Benefits of neural nets include their applicability to real-time environment and their ability to mimic human brain. Their short- comings include the requirements of analytical know- ledge and development.

3.6 Petri nets Petri nets model the static properties of discrete-event systems, concentrating on two basic concepts: conditions and transitions. A condition may, for example, be the presence or absence of jobs in the queue for a work- centre, or empty spaces in the output buffer of a machine. The fact that a condition holds can give rise to the ‘firing’ of a transition, corresponding to the occurrence of an event,which in turn may change the state of the system causing certain conditions to start holding and other to stop. A Petri net model consists of place, transitions and directed arcs. Places are shown as circles and correspond to conditions. Transitions, as suggested, correspond to events and are shown as bars. Directed arcs define the input and output relationships between places and trans- itions, respectively. Tokens, shown as filled-in circles, indicate when a particular condition holds. All conditions leading into a transition must hold before the transition can be fired. Petri nets strike a balance between the speed and simplicity offered by mathematical programming on the one hand, and the flexibility provided by general simulation packages such as SIMAN and SLAM on the other. Petri nets also enable the description of a system to be organised in a hierarchical manner, and can be used to model conditions of conflict, concurrency, and dead- lock. Deadlock occurs when a system is not able to change its state.

376

Petri nets have been shown to be consistent with the contingency response decision support requirements of flexible manufacturing systems and have also been used to analyse the qualitative features of a FMS. One of the noted benefits arising from their use is the facility to clearly portray partial execution strings, corresponding to possible sequences of events in response to a disturbance such as a resource failure or lack of raw material.

Disadvantages associated with Petri nets [22] include (i) lack of a general methodology for constructing

(ii) diagrams can become cluttered when modelling

(iii) lack of general software to support the computer

Petri net models from a systems specification,

complicated systems, and

coding of Petri net models.

4

Structured design, Jackson structured programming and Warnier-Orr design methodology are, like SSADM, methodologies intended specifically for use in projects involving computer systems. Unlike SSADM, the meth- odologies noted previously are not thought to be of general interest. Martin [SI provides the following criti- cisms:

Structured design: The strategies used by the method- ology, as proposed by Yourdon [23], are difficult to apply in practice, the notation used is confusing, and the measures of quality, i.e. coupling and cohesion, deterior- ate into rules of thumb in practice.

Jackson structured programming: The methodology, pro- posed by Jackson [24], is limited to the design of a rare breed of simple programs, with complex programs having to be viewed as composites of simple programs. It is not suitable for use in the development of online or database systems, or both.

Warnier-Orr design methodology: The methodology, a composite of logical construction of programs (Warnier) and structure program design, Orr [25], makes the dubious assumption that the structure of a system and its input can be determined from the structure of the output of the system. Like Jackson structured programming, Warnier-Orr design methodology is not suitable for use in projects involving the development of database systems.

The methodologies reviewed thus far may be classified as belonging to the systems approach [l] and as such make the common assumption that the problem to be addressed, the systems associated with the problem, and the objectives of those systems may be clearly defined.

Methodologies and tools for use in computer system projects

5 Conclusions

Various methodologies and modelling tools have been used in manufacturing. The widely used structured ones have been reviewed. The use of analytical techniques such neural nets is on the increase and has thus been included. In a real environment a combination of these tools may be used to model a complete business environment. For example, the IDEFO may be used for the functional modelling of shop-level business functions (e.g. tools management). In the same enterprise neural nets could be used for process control of each cell within the shop (e.g.

I E E Proc.-Sci. Meas. Technol., Vol. 142, N o . 5, September 1995

Page 7: Review of modelling techniques and tools for decision making in manufacturing management

process monitoring) and the IOA could be used for the material Bow in that factory.

All the tools and techniques discussed may be used for modelling a business environment in general. However, some tools are better suited for modelling one particular aspect. It is left to the developers to decide which of these tools are best suited for a particular business function. The developers should group the business functions in ‘natural cells’, based on information or material flow (Johnson [26]). It is believed that the enables tools to structure their decision making around these natural groups. The most appropriate modelling tools for a business process in today’s business processes is shown in Table 1 which also illustrates the tools and techniques in relation to the timing horizons of the business processes, and to the conventional planning and control levels.

The tools and techniques may be used for ‘hard’ as well as ‘soft’ systems, as described by Towill [27]. It is considered by this research that hard systems may be better modelled by tools that are appropriate for short time horizons, e.g. neural nets. The soft systems, i.e. man- agement functions (timing horizon for such business pro- cesses is medium to long term), may be better modelled by tools such as SSADM. Tools such as IDEFO and IOA are quite powerful because in addition to modelling the processes, they also easily illustrate interconnection between subsystems. This paper has reviewed these tools and techniques and proposed a simplistic practical guide in the form of a Table.

6 Acknowledgments

The author wishes to thank Jillian MacBryde, David Cairnie and William White.

7 References

1 CHECKLAND, P.: ‘Systems thinking systems practice’ (Wiley,

2 AUSTIN, S., and PARKIN, G I . : ‘Formal methods: a survey’.

3 NICHOLLS, D.: ‘Introducing SSADM - the NCC guide’ (NCC

4 WU, B.: ‘Manufacturing systems design and analysis’ (Chapman &

5 MARTIN, J., and McCLURE, C.: ‘Structured techniques the basis

1981)

National Physical Laboratory, 1993

Publications, 1987)

Hall, 1992)

for CASE (Prentice-Hall, 1988)

6 SWANN, P.E.: ’Execution is the key to success of a system for manufacturing material flow control‘, Ind. Eng., October, 1984

7 DEMARCO, T.: ‘Structured analysis and system specification’ (Prentice-Hall, 1979)

8 GANE, G., and SARSON, T.: ‘Structured systems analysis: tools and techniques’ (IST Inc., 1977)

9 BANNERJEE, S.K.: ’Application of structured design techniques in advanced manufacturing technology’. Proceedings of 2nd mter- national conference on CAPE, Edinburgh, 1987, pp. 245-253

IO CHEN, D., MARCOTTE, F., and DOUMEINGTS, G.: ‘Review of existing tools’. GRAI Laboratory, ESPRIT Project, FOF Work- package 3

I I CARRIE, A., and MACINTOSH, R.: ‘A structured approach to process redesign’. Proceedings of the 29th annual conference of the British Production and Inventory Control Society, 1994, pp. 153- 167

12 ‘Architecture’s manual, ICAM definition method, IDEFO’ (CAM-I Inc., Arlington, Texas, USA, 1980)

13 MALAHOTRA, R., and JAYARAMAN, S.: ‘An integrated frame- work for enterprise modelling’, J. Mamf. Syst., 1992, 11, (6). pp. 26-440

14 PANDYA, K.V.: ‘Model for production planning and control deci- sions at cell level: a case study’, Comput. Inteqr. Manu/: Syst., 1994, . . 7, (2), pp. 75-92

control’. lnt. J. Prod. Res.. 1990.28. (9). DD. 1623-1633 15 FRANK, I.T., LOFTUS, M., and WOOD, N.T.A.: ‘Discrete cell

16 ‘Information modelling manual, IDEFI’L extended’. ICAM project priority 6201, 1985

17 ‘Miniguides - Lucas manufacturing systems engineering hand- book‘. Lucas Engineering and Systems (Institution of Production Engineers, 1989)

Res.. 1979. 17. DD. 123-135 18 PARNABY, J.: ‘Concept of manufacturing system’, In t . J. Prod.

19 TOWILL: D.R.; ‘The dynamic analysis approach to manufacturing systems design’, Adu. Manuf: Eng., 1989, 1, pp. 131-140.

20 BERRY, D., and NAIM, M.: ‘A systems engineering analysis of information and material flows in a manufacturing company’. Pro- ceedings of Factory 2000,1994, pp. 138-145

21 WU, B.: ’An introduction to neural networks and their applications in manufacturing’, J. Intelli. Monuf., 1993,3, pp. 391-403

22 CECIL, J.A., SRIHARI, K., and EMERSON, C.R.: ‘A review of Petri net applications in manufacturing’, Int. J. AMT, 1992, 7, pp. 168-77

23 YOURDON, E.: ‘Modern structured analysis’ (Prentice-Hall, 1989) 24 JACKSON, M.A.: ‘Princides of Dromam design’ (Academic Press. . -

New York, 1975)

New York. 19771. on. 13-22 25 ORR, K.: ‘Structured systems development’ (Yourdon Press Inc.,

26 JOHNSON, P::’EKective office support for redesigned manufac- turing system’. Proceedings of international conference ~ Sunder- land Advanced Manufacturing Technology, 1989

27 TOWILL, D.R.: ‘Engineering change; or is it change engineering? A personal perspective‘, Proc. IEE, A, 1991,138, pp. 11-21.

28 PARNABY, J.: ‘The design of competitive manufacturing systems’, Int. J. Technol. Manage., 1986, 1, pp. 385-396

IEE Proc.-Sci. Meas. Technol.. Vol. 142, No. 5, September I995 377


Recommended