modular production systems: a new manufacturing paradigm

Modular production systems:

a new manufacturing paradigm

G. G. ROGERS1* and L. BOTTACI2

1Department of Mechanical Engineering, Curtin University of Technology, GPO Box 1987,

Perth 6102, Western Australia2Department of Computer Science, University of Hull, Hull HU67RX, UK

Received August 1995 and accepted July 1996

It is well recognized that manufacturers of consumer goods throughout the world are facing

major new demands, including shorter product life-cycles and increasing competition. In

response, companies are restructuring and moving away from traditional process-centred

work practices in favour of concurrent engineering methods. In particular, design for man-

ufacture has gained widespread recognition as a means of reducing production costs and lead

times. However, optimal design for manufacture is di�cult to achieve using current-day work

organization and business structures. An underlying problem is the lack of a scienti®c

framework for production. To address this need, this paper proposes a radical and far-

reaching new manufacturing paradigm based upon on building production systems from

standardized modular machines. The manufacturing concept, termed modular production

systems (MPS), is aimed speci®cally at `hard' low- to medium-technology consumer products,

as typi®ed by goods such as children's toys and kitchen appliances. The rationale for MPS as a

means of enabling concurrent product and production system design is put forward, and the

long-term implications and work required to establish the concept are discussed.

Keywords: New manufacturing paradigm, modular manufacturing virtual manufacturing

1. Introduction

Global competition in manufacturing and changing con-sumer demand are resulting in a trend towards greaterproduct variety and innovation, shorter product life-cycles,lower unit costs and higher product quality (Ralston andMunton, 1987; Cohen, 1988). As a result, manufacturers ofboth low- and high-technology products are experiencingsigni®cant new demands and challenges to remain com-petitive. In particular, in addition to reducing costs, it isbecoming strategically important for manufacturers to:

(1) Shorten the `design to market' lead time. This is re-sulting in the need to design `right ®rst time', because thereis not enough time to correct design errors, or to re-engi-neer products for lower cost or higher quality (Shina,1991);

(2) Ensure that goods are produced to a high and con-sistent quality;

(3) Forecast production costs and lead-times in order tohelp assess the market potential of a product prior to sig-ni®cant investment;

(4) Alter production capacity in response to changingdemand without incurring signi®cant costs or productionlead-times;

(5) Introduce new products frequently to retain or gainmarket share.

The application of computers and the adoption of con-current engineering work practices have signi®cantly im-proved the e�ectiveness of manufacturers in meeting thesenew objectives. However, as outlined in the following sec-tions, it is becoming evident that further progress is limitedby the fundamental structures and methods of manufac-turing business. To overcome these problems, this paperputs forward a radically new business model for the man-ufacturing of low- to medium-technology consumer goods,which is based upon building `¯exible' production systemsfrom standardized modular subsystems. Moreover, thisapproach seeks to provide a systematic link betweenproduct design and production system design. The belief isthat establishing such a formal link, combined with ¯exi-*Author to whom all correspondence should be addressed.

Journal of Intelligent Manufacturing (1997) 8, 147 ± 156

0956-5515 Ó 1997 Chapman & Hall

bility by con®guration, will enable the new manufacturingdemands to be met more e�ectively. Before detailing thisnew manufacturing concept further, the paper starts byhighlighting the underlying reasons why CIM conceptsbased upon present-day production techniques are limited.

2. Robotics and CIM

In the last 30 years computers have had a major impact onmanufacturing e�ectiveness, and have been applied tofunctions as diverse as payroll, stock control, parts pro-duction, process planning and component testing. Morerecently, the combined use of CAD, CAM and CNC ma-chinery has resulted in major improvements in the cost-e�ectiveness, quality and ease with which a wide variety ofcomponent parts can be produced.

Moreover, the desire to automate all aspects of manu-facturing has helped to establish the concept of computerintegrated manufacturing (CIM). However, there is nostrong agreement on the scope of CIM, and there are few± if any ± generic models for CIM (Lawrence, 1986).Furthermore, despite considerable research, signi®canttechnical problems remain unsolved. For example, a cen-tral problem limiting CIM is how to achieve cost-e�ectiveautomated ¯exible assembly. This is central to CIMbecause:

(1) In Western countries assembly can often consume upto 50% of the production costs and 50% of the totalmanufacturing labour force (Swift, 1987);

(2) Once designed and constructed, automated assemblysystems generally produce goods of a higher quality andconsistency and at a lower unit cost than those producedusing manual methods;

(3) Reduction in manual labour reduces not only directsalary costs but also the need for company support func-tions, such as personnel, recruitment and sta� training.

Anthropomorphic-type robotics have generally been seenas playing a central role in ¯exible assembly. However, inpractice, such automation remains limited to highly re-petitive specialized work, and is not useful for multi-product, small-quantity manufacturing (Tsukune et al.,1993). Noted causes of such limitations include the fol-lowing:

(1) Robots share many of the constraints associated withdedicated machines, such as the need for a highly `struc-tured' work environment with special-purpose jigs, ®xtures,grippers, feeders etc. (Redford and Lo, 1986);

(2) The motion paths and mechanical properties of therobot arm are often poorly optimized for any particulartask. Consequently, the cycle times and positioning accu-racy of robotic automation are often considerably less thanthose of dedicated machines (Weston, et al., 1989);

(3) Robotic automation tends to be complex, and typi-®ed by excessively complex mechanisms and software(Canny and Goldberg, 1992). The custom software alonecan often represent four to ®ve times the cost of the robot(Carlisle, 1989).

These problems have led Carlisle (1991) to conclude thatthe factory ¯oor is not interested in and cannot supportcomplex robotic technology. Indeed, applying high tech-nology to existing work structures and practices is in-creasingly coming to be questioned, because it onlypartially increases the competitiveness and e�ciency ofmanufacturing (Ralston and Munton, 1987; Syan, 1994a).A principal problem, as Davis (1991) notes, is that pro-duction machinery is designed and implemented with theobjective of seeking technical excellence in machine func-tion and performance at the expense of the needs of theoverall manufacturing system.

3. Concurrent engineering

Concurrent engineering is concerned with ensuring thatproduct life-cycle requirements, such as service, quality,production and time-to-market, are considered at theproduct design stage. The bene®ts of this approach can beconsiderable. In particular, designing products speci®callyfor ease of manufacture can result in signi®cant reductionsin manufacturing costs and product lead-times, often withreductions in the region of 50% compared with traditional(functionally) based work methods (Syan, 1994a).

Central to concurrent engineering methods is the needfor good communications and teamwork between all thepersonnel involved. However, with `traditional' (function-al) manufacturing business structures this objective is oftenhindered because:

(1) Information may be dispersed between many loca-tions, and is therefore di�cult to access;

(2) Existing (functional) work structures can be barriersto free information ¯ow between groups ± in particular,between product design and process design groups (Salz-berg and Watkins, 1990);

(3) Work attitudes in Western manufacturing nations donot value teamwork and the sharing of information ashighly as the more traditional attributes of technical com-petence and creativity (Shina, 1991);

(4) Design for manufacture (DFM) requires consider-able detailed manufacturing process knowledge. Conse-quently, rarely are all the design alternatives and theirmanufacturing implications carefully evaluated.

Problems caused by access and communication of infor-mation are being addressed by new work structures andmanagement practices, such as the formation of product-centred multidisciplinary teams (Watson and Pullen, 1989;Pawar, 1994; Syan, 1994b).

148 Rogers and Bottaci

However, despite the introduction of concurrent engi-neering teams, signi®cant problems still remain with regardto achieving truly `optimal' design for manufacture. Thisstems from the fact that what constitutes a good design formanufacture is often not clear, even to production per-sonnel. Whitney (1990) illustrated this point by noting thesurprise experienced by a chief of a concurrent engineeringteam who found that the manufacturing department didnot know what a good design for assembly should be. Tohelp alleviate some of these problems, DFM tools havebeen developed that analyse the manufacturing cost im-plications of a design (Dewhurst and Boothroyd, 1983;Swift, 1987). However, although such tools undoubtedlyhelp to further DFM, they are still limited because:

(1) The analysis is based largely upon `rule of thumb'information, and is able to provide only an approximationof the relative di�erences in the manufacturing costs be-tween alternative designs;

(2) Such tools are totally dependent on the skills and thewillingness of the user to conceive cost-e�ective designs;they cannot directly guide a designer towards an e�ectivedesign.

It has been noted that the di�culties with DFM and as-sociated tools are deep rooted, and stem from the lack ofan underlying scienti®c and explicit representation of pro-duction processes and assembly methods (Swift, 1987).Indeed, DFM lacks a precise systematic method because,as Voelker and Requicha (1977) noted, `the discrete goodsindustry runs on centuries of experience, human skills andknow-how; there is no underlying base of science'.

4. A way forward

From the above discussion it is evident that althoughcomputers and the adoption of concurrent engineeringmethods have signi®cant bene®ts, both of these approachesare ultimately limited because:

(1) Fully automated ¯exible assembly is often technicallyimpossible or not cost-e�ective;

(2) Optimal design for manufacture (DFM) requires aformal and explicit link between product design and pro-duction. Currently this is virtually impossible to achievebecause of the complexities of present day manufacturingmethods and the underlying lack of a scienti®c basis toproduction.

It may be noted that the cause of these di�culties stemsfrom the working principles of today's manufacturing in-dustry, which are based largely on the concepts of spe-cialization and functional division of tasks. However, thesemethods, which resulted from the mass production era,were never intended for the modern-day demands charac-

terized by high variety and short-cycle manufacturing(Roobeek and Abbing, 1988).

We conclude that the underlying manufacturing princi-ples, methods and associated business structures used to-day must change if we are to meet the demands of the new`agile' era. Indeed, on a wider point, the highly specializedand non-transferable nature of manufacturing knowledgeassociated with the current manufacturing business para-digm has been noted as a principal cause of the unevendistribution of wealth in the world (Yoshikawa, 1994). Tosee what the basis of this new manufacturing paradigmshould be, the following section reviews the bene®ts asso-ciated with the modular approach to constructing pro-duction systems.

5. Modular manufacturing concepts

The building of artefacts from standardized modules,subsystems or components has been common for centuries.In the context of engineering product design, Stoll (1986)has noted that modular construction permits `standardizeddiversity' by using di�erent combinations of standardcomponents. He also notes that modular design resistsobsolescence, shortens redesign, enables new designs to berealized by using existing modules, reduces costs, and easesmaintenance. Moreover, where modular constructionmethods have become widely established, as in electronicsfabrication using standard components (e.g. integratedcircuits resistors and capacitors), the design process isgenerally assisted by sophisticated design and veri®cationtools.

With regard to manufacturing machinery construction,the modular concept has been used for many years. Forexample, many machine tool manufacturers produce cus-tomized machine tools largely by con®guring their existingmachine subsystems (Hu et al., 1993). Automation equip-ment suppliers, such as Festo, SMC and Parker, supplyproprietary modular hardware, such as actuators units andgrippers, for building modular work-handling systems.Moreover, although traditionally associated with `hardautomation', this modular manipulator approach is gain-ing recognition as an alternative means of achieving thepromised ¯exibility of anthropomorphic robots, albeit bycon®guration (Weston et al., 1989; Canny and Goldberg,1992).

Modular concepts are also utilized in the construction ofentire production systems. Noted bene®ts of this approachinclude the following:

(1) Providing greater scope in the way production isorganized, and the opportunity to readily recon®gureproduction to meet both short- and long-term objectives(Merchant, 1985);

(2) Enabling the use of the simplest integrated combi-nation of processes, machine systems, tooling, people, or-

Modular production systems 149

ganizational structures, information ¯ows, control andcomputer systems necessary to perform a given task (Da-vis, 1991);

(3) Helping to eliminate islands of automation andfurther the reuse of machinery (Tsukune, et al., 1993).

However, Tsukune et al., (1993) also note that there aresigni®cant problems currently limiting the progress ofmodular manufacturing:

(1) The large number of manufacturing machine ele-ments currently in use makes modular production systemdesign and control di�cult;

(2) The design, production and control processes arebased on completely di�erent models; this results in com-plex transformations between the `idealized world' in whichdesign tools operate and the `real-world' in which manu-facturing occurs. These problems stem principally from thefact that there are no standards for modular machinery.Moreover, there is no agreement on what the building el-ements should be. For example, a `machine module' cur-rently encompasses everything from a complete machinetool, such as a robot with an integrated controller, to amachine building element, such as an actuator unit, motoror transmission system. Indeed, such diversity of manu-facturing machinery and hardware for constructing ma-chinery exists principally as a result of the wide diversity ofproduction requirements.

6. Modular production systems:

a new manufacturing paradigm

The modular production systems (MPS) concept has beenproposed as a way of overcoming the limitations resultingfrom a lack of modular machine standards (Rogers, 1990).Moreover, MPS seeks to provide a radical new manufac-turing business framework suitable for the `agile' manu-facturing era. The module standards are based upon auni®ed `reduced' set of `primitive' production elements,which are at a level of modularity lower than hitherto. Themodule categories comprise just four classes: process ma-chine primitives, motion units, modular ®xturing andcon®gurable control systems (Table 1). The belief is thatappropriate selection of modules from these categories willenable a diverse range of e�cient, automated and inte-grated production systems to be built.

To help clarify the signi®cant di�erences in this conceptfrom existing modular manufacturing methods it is usefulto consider the following analogy.

6.1. An analogy from electronics

For many years the universally adopted method for small-to medium-volume digital electronics implementation hasbeen to integrate logic devices, such as ICs, using printedcircuit interconnection. However, despite the popularity ofthis method, it has its limitations. In particular, electronic,mechanical and manufacturing design con¯icts can arisewith component positioning, track layout and mountingmethods. Furthermore, the functionality of the integratedcircuit devices is rarely fully utilized.

Table 1. The principal modules used for building an MPS

Process machine `primitives' (PMPs)

These are the principal material processing sub systems that operate upon and change the state of materials. For many processes, such as

pressing, injection moulding and drilling, the modules are functionally similar to existing process machinery. However, they now

conform to precise, prede®ned, performance, dimensional and control standards.

Modular actuator elements (MAEs)

These are for the provision of motion tasks and perform two central roles:

(1) In association with modular tooling and jigging, MAEs are used to build simple (e.g. 1 or 2 DOF) material and component transfer

systems;

(2) When appropriate PMPs do not exist, MAEs are used with modular tooling and jigging to form the basis of new special-purpose

process machinery (e.g. material cutting, glue laying). For example, Fig. 1 illustrates how a number of modular actuator elements might

be con®gured with a drill unit to perform a speci®c process operation, i.e. bore a number of holes in a workpiece.

Modular tooling and jigging

Tooling and jigging hardware for `tailoring' PMPs and MAE units to perform speci®c functions such as those de®ned above.

Con®gurable control systems

These provide the communication network for programming, commanding and synchronizing the various MPS subsystems.

Note: Neither `low-level' sensors, such as binary switches and optical encoders, nor `high-level' sensors, such as vision, form a distinct category of module.

The reason for this is twofold:

(1) Low-level sensory data needed for motion control purposes forms an integral part of the associated PMP and motion modules;

(2) `High-level' real-time sensory feedback information needed normally to support component orientation, handling and feeding tasks is largely

eliminated by virtue of integrated component production and assembly.


Until recently these issues were not of major concern todesigners. However, with a continually growing demandfor increased functionality, reduced package sizes, fasteroperating speeds, lower power consumption, reduced costsand higher reliability, such methods of construction havecome under increased strain.

Today, gate-array technology is a superior approachover the use of `traditional' IC subsystems, such as coun-ters and timers, for building medium- to high-volumedigital functionality. The key di�erence with this new ap-proach is that speci®c logic requirements are met by in-terconnecting very basic logic elements such as NAND orNOR gates. In particular, the gate array approach elimi-nates redundant functionality and simpli®es manufacturingwork by virtually eliminating component positioning, tracklayout and component-mounting tasks. Moreover, al-though designing functionality using the gate-array ap-proach is considerably more complex (for humans), thedesign process is fundamentally more systematic, and thuslends itself to design automation by computer tools. As aresult, computer tools are a central and essential feature ofthis new approach to system design, performance veri®ca-tion, implementation and testing.

It is evident that the catalysts for the recent strikingimprovements in digital electronics fabrication have comefrom:

(1) Changing the building blocks from `complex' sub-systems, i.e. ICs, to primitive' elements, i.e. logic gates; and

(2) The computerization of many aspects of the (gatearray) circuit design process.

6.2. The working principles of MPS

Returning to MPS, the aim is to establish, in a similarmanner, a new manufacturing method based on the con-®guration of a reduced set of standardized `primitive'production elements. In particular, building on the analogy

with electronic fabrication, the `primitive' MPS modulesreplace the `complex' subsystems of `traditional' produc-tion machinery in a way similar to gate arrays replacing ICdevices such as counters and timers. Moreover, the belief isthat similar signi®cant bene®ts, such as the formalizationand subsequent automation of the (production) designprocess, would result.

The modules identi®ed by the MPS concept comprisejust four classes or categories: process machine primitives,motion units, ®xturing and con®gurable control systemsTable 1. Based on these module categories, the belief is thatit will be possible to build entire automated productionsystems using a `core' of process machine primitives `inte-grated' by motion modules, ®xturing and control systems.

To see how and why adopting such methods of pro-duction system construction might be of bene®t it is ®rstnecessary to clarify the two new manufacturing principlesthat MPS seeks to promote: just-in-place parts productionand automated production system design.

6.2.1. Just-in-place parts production

A central objective of the MPS concept is to simplify au-tomated assembly by eliminating the principal assemblytasks. In particular, by building integrated `just-in-place'component production and assembly systems the belief isthat the `assembly bottleneck' caused by component ori-entation, transfer and feeding can be virtually eliminated.Furthermore, work in progress and transportation betweencomponent supplier and manufacturer are signi®cantlyreduced or virtually eliminated. These savings alone can beconsiderable, because component transportation can oftenrepresent up to 20% of component costs with present-dayjust-in-time production methods.

Thus by considerably reducing the need for either as-sembly labour or special-purpose automated assemblymachinery, the result will generally considerable cost sav-ings, even though component costs might increase (becausethe utilization of individual process machinery such as in-jection moulders, press and stamping machinery will re-duce compared with current methods).

6.2.2. Automated production system design

Con®guration design tasks are generally amenable to for-mal design procedures (Dym, 1994). The implementationof design procedures using knowledge engineering pro-gramming methods has led to numerous engineering designtools. Examples include BUDS for modelling mechanicalengineering sub-assemblies (Theobald et al., 1993), PRIDE(Mittal et al., 1986) for designing paper-feed systems andmodular ®xture design systems (Ngo and Leow, 1994).Equally, because MPS design is principally a hardwareselection and con®guration task, it is envisaged that formalMPS design procedures and tools could be produced.Moreover, it is envisaged that such tools will also enable

Fig. 1. An assembly of modular units for boring a workpiece

(O'Meara, 1992).


manufacturing implications, such as production feasibility,costs and return on investment, to be accurately deter-mined at the product design stage.

7. Bene®ts resulting from MPS

Building on these new principles of integrated manufac-turing and systematic production system design the fol-lowing sections summarize how, through adoption of theMPS approach, the new manufacturing objectives identi-®ed in Section 1. would be satis®ed.

7.1. Shorten the `design to market' lead-time

Designing products explicitly for ease of manufacture andassembly is recognized as one of the principal means ofreducing `time to market'. Because the MPS method per-mits the development of systematic MPS design procedures(albeit computationally complex in nature), automation ofthe design process is feasible. This automation, combinedwith simpli®ed production system construction resultingfrom standardized modularity, will help to shorten pro-duction lead-times signi®cantly by enabling an MPS to bespeci®cally designed and con®gured for a new product.

7.2. Ensure that all goods are produced with highand consistent quality

MPS exploits the systems principle that combining units ofknown performance and reliability results in systems ofequally predictable performance and reliability. As a result,provided appropriate standards are chosen and the MPSdesign algorithms are correct, then the MPS methods willensure that products are produced with consistent andknown quality.

7.3. Forecast production costs and lead-times

MPS supports virtual manufacturing by enabling themodelling of all aspects of production. As a consequence,because MPS design is computer oriented, the belief is thatproduction system costs and construction times can bepredicted prior to investment in hardware.

7.4. Alter production capacity without signi®cant costs orproduction lead-times

By having precise computer models of MPS hardware andsystematic procedures for MPS design the belief is that itwill be feasible to readily alter the elements and structure ofa MPS to match the production rates throughout life-cyclemarket demands. As a result, the production system can beòptimally' tuned to maximize utilization of the modularmachinery.

7.5. Frequently introduce new products to retain or gainmarket share

Because MPS design is automated and the hardware ismodular it becomes feasible to `rapidly' con®gure and re-con®gure production systems without signi®cant costs andtime. Moreover, the redundant modular hardware can bereused at a later date for another MPS.

8. Implications and work required to establish MPS

Clearly, such a new paradigm for manufacturing cannot beachieved òvernight', and much research work and invest-ment will be required before such a concept could be im-plemented. In particular, central to implementation of theMPS concept will be the need to:

(1) Establish òpen' standards de®ning machine moduleperformance, dimensions and control;

(2) Change the emphasis on product design and manu-facturing process selection to meet the constraints imposedby the MPS approach (Table 2). In particular, all pro-duction operations must be performed by a prescribedrange of `process modules'. Furthermore, those processeswith low accuracy or reliability will lose favour, becausethere is little, if any, redundancy in the production ma-chinery. Equally, processes with long cycle times, such ascasting, are less desirable, because the MPS philosophyaims to maximize utilization of modules and thus ensure ahigh return on capital investment;

(3) Produce advanced design and analysis tools to sup-port designers in the rapid concurrent design of productsand associated MPSs. In particular, the functionality ad-dressed by such support tools must include:

(i) Ensuring that the product design is within theprocess constraints imposed by the MPSmethod;

(ii) Assisting in the generation of the structurallayout of the overall MPS;

(iii) Supporting the design of any special-purposematerial processing and component transfermachinery built from modular actuator units,modular jigging, ®xturing etc.;

(iv) The generation and veri®cation of the controllogic necessary for the sequencing of the entireproduction system.

9. Design tools for the structural layout design of an MPS

To progress the MPS philosophy, ongoing research is fo-cused on producing generic systematic design proceduresfor determining the structural layout of the overall MPS. Inparticular, the following subsections outline a prototypeMPS design procedure that has been developed to assist in


the selection and arrangement of PMP modules for man-ufacturing an MPS-oriented product at a speci®ed rate(Enright, 1994). The procedure developed comprises threemajor stages: product analysis, product representation forMPS design, and MPS synthesis.

9.1. Product analysis

MPS design starts by `extracting' information from theproduct design pertinent to the design of the associatedMPS. In particular, the product analysis aims to:

(1) Identify groups of components that can be producedand assembled independently of each other to form anMPS subassembly;

(2) Determine the process modules and order of as-sembly for each component forming a given MPS subas-sembly;

(3) De®ne the order in which all subassemblies and re-maining components are to be amalgamated to constructthe ®nal product.

Currently, a product analysis procedure using a morpho-logical chart approach, similar to current generation designfor assembly and manufacture systems (Dewhurst andBoothroyd, 1983), has been developed for cylindrical-basedproducts, as typi®ed by products such as a torch or a bi-cycle pump. The method is based on the concept whereby adesigner ®rst identi®es key components, termed basecomponents, upon which the construction sequence of thesubassembly can be based.

9.2. Product representation for MPS design

Product construction analysis tables (PCAT) (Fig. 2) aresubsequently produced, which represent the results of theproduct analysis. These tables include information de®ning:

(1) The members of the subassembly;(2) The base component;(3) An explicit order for constructing the subassembly;(4) The process machine primitives (PMPs) for manu-

facturing each component;(5) The manufacturing cycle time based on one PMP per

component.

In addition, information de®ning how subassembles are tobe interconnected to form the ®nal product is stored in aproduct construction analysis graph (PCAG). These tablesare the basis for subsequent synthesis of the MPS.

9.3. MPS Synthesis

The ®rst stage of the MPS synthesis procedure is to de-termine the number of PMP modules needed to manufac-ture the product at the desired production volume. Thisrequires identifying which PMPs, if any, result in a sub-assembly cycle time lower than that required by the pro-duction output rate. If the cycle time of a PMP is too lowthen that PMP unit is increased in number until the re-quired cycle time is achieved. Alternatively, the designermay wish to reconsider the product design and select adi�erent process. This `balancing' of process cycle timeleads to modi®ed PCATs and a modi®ed PCAG. From the

Table 2. MPS product design and process selection constraints

Process selection

All components must be capable of being produced from process machine primitives (PMPs). It can be assumed that PMPs will exist for

all commonly available processes except:

(1) Those processes where large time periods or manual labour are required, e.g. casting of metal and forging. Such processes are

deemed to be inappropriate to MPS philosophy, and wherever possible such methods of component production should be eliminated;

(2) Standardized components (e.g. screws, bolts), as these will nearly always be best produced in the normal manner and fed using an

appropriate feeder (e.g. bowl feeder, bandolier)

Cycle time requirements

Wherever possible, all components are made in situ exactly as and when required, i.e. just in time (JIT) and just in place (JIP). Processes

that have cycle times an order of magnitude slower than the required production rate need to be reconsidered.

Note: Slow processes can e�ectively be speeded up by using two or more identical PMP units in parallel, however, eventually the

capital costs of the units may become prohibitive. Equally, PMPs with minimum rates of production that exceed the required production

rate should generally be avoided.

Quality and reliability

All operations and processes must be selected with a view to having a reliability as close to 100% as possible. The objective is for every

operation and component to be produced right-®rst-time to the required speci®cation. Any process that fails, even momentarily, will

cause an interruption of the whole production system.

Manipulation and assembleability

All components are to be handled and assembled using simple pick-and-place units built from the MAE units. By not losing the

orientation and positioning of components the objective is to keep the amount of on-line sensing to a minimum.

Material costs

Clearly, in common with normal manufacturing practice, material costs that are unduly expensive must be rejected.


modi®ed PCAT and PCAG charts a schematic represen-tation, called an MPS design graph, is produced (Fig. 3).This graph de®nes the number and order sequence of thePMP modules derived from the product design (i.e. theassembly order), assembly processes and the MPS cycletime requirements.

The MPS synthesis procedure developed to date is stillvery much in an embryonic stage of development, andconsiderable further work is required before e�ective MPSdesign tools can be produced. In particular, further work isto be focused at:

(1) Extending and enhancing the range of morphologicalcharts for product analysis;

(2) Extending the MPS synthesis procedure such that thegeometrical con®guration of the PMP units is determined;

(3) Formulating standards for the PMP modules. Inparticular, we are currently investigating the factors thatwould in¯uence standards for plastics injection-mouldingPMP units (Oey, 1993).

10. Discussion and conclusions

This paper has outlined a new approach to consumer sectormanufacturing based upon the use of a limited range ofstandardized `primitive' machine elements. To implement

Fig. 2. PCAT: a tabular representation of a sub-assembly. BM = base member; C = component.

Fig. 3. An example of an MPS design graph.


this modular production system (MPS) concept will requirethe development of machine module standards and newdesign techniques and tools. Provided long-term invest-ment is forthcoming, it is envisaged that the MPS conceptwill provide the quantum leap needed to overcome currentproblems hindering cost-e�ective ¯exible manufacturing.In particular, it is anticipated that the MPS approach willprovide the framework for a major epoch in consumermanufacturing industry, which will be characterized by:

(1) A standard and `universal' model for productionsystem design and operation suitable for a wide variety oflow- to medium-technology consumer products;

(2) Flexible production by module con®guration, pos-sibly enabling production systems to be designed and builtin a matter of weeks or even days;

(3) The ability to accurately predict production param-eters, such as operating costs, at the product design stage;

(4) Signi®cant reductions in manufacturing machinerycosts resulting from competition to supply machineryconforming to an `open' MPS module standard;

(5) Reuse of machine modules once production of aparticular product is ®nished;

(6) Minimal need for specially designed ®xturing, tool-ing and software;

(7) Much smaller production systems, because onlymachinery speci®c to the product requirements is installedon the shop¯oor.

Moreover, MPS manufacturing will result in major struc-tural changes in manufacturing business. In particular, asdepicted in Fig. 4, the construction of an MPS will prob-ably best be undertaken by specialist system builders whoare solely responsible for the leasing of module hardwareand the building of such production systems. `Manufac-turers' would then be more concerned with product design,marketing and sales. Ultimately this would give rise togeographically distributed `generic' factories, which areleased to `manufacturers' who seek their production ca-pabilities. Thus many goods could be produced on a morelocal basis to the markets in which they are required,

thereby making savings on transportation costs and asso-ciated environmental e�ects, reduced inventory stocks anda decreased `production to sales' lead time.

References

Canny, J. F. and Goldberg, K. Y. (1992) A `RISC' paradigm for

industrial robots, Technical report RAMP 92-3, University

of California, Berkeley CA.

Carlisle, B. (1989) The changing nature of assembly automation,

in Proceedings of the 10th International Conference on As-

sembly Automation, IFS publications, Bedford UK, pp. 31±

37.

Carlisle, B. (1991) Is US robotics research of any use to US in-

dustry? IEEE Robotics and Automation Society Newsletter,

July.

Cohen, S. (1988) The myth of the post-industrial economy. Sie-

mens Review, March/April 4±19.

Davis, R. K. (1991) A systems approach to machinery design and

implementation, in Proceedings of Eurotech Direct, Machine

Systems Vol., I. Mech E., 2±4 July, pp.19±24.

Dewhurst, P and Boothroyd, G. (1983) Computer aided design

for assembly. Assembly Engineering, February, 18±22.

Dym C. L. (1994) Engineering Design: A Synthesis of Views,

Cambridge University Press.

Enright, R. (1994) Automated design of modular production

systems, Internal Report, Dept of Electronic Engineering,

University of Hull, UK.

Hu, W., Kong, Z., Zhu, G., and Yu, J. (1993) Modules for

modular design of machine tools, in Proceeding of Interna-

tional Conference on Engineering Design, The Hague, 17±19

August, Heunsta Zurich pp. 1287±1294.

Lawrence, A. (1986) Taking o� slowly. Industrial Computing,

April, 19±22.

Merchant, E. (1985) The importance of ¯exible manufacturing

systems to the realization of full computer integrated man-

ufacturing in Flexible Manufacturing Systems, Warnecke, H.

J. (ed), IFS (Publications) Ltd. and Springer-Verlag.

Mittal, S., Dym, C. L. and Marjana, M. (1986) PRIDE: An expert

system for the design of paper handling systems. IEEE

Computer, 19(7), 21±41.

Fig. 4. The MPS business paradigm (Oey, 1993).


Ngo, B. K. and Leow, G.L. (1994) Modular ®xture design: a

designer's assistant. International Journal of Production Re-

search, 32, 2083±2104.

O'Meara (1992) A knowledge-based CAP system for the auto-

mated selection of actuators, MSc thesis, University of Hull,

UK.

Oey, V. (1993) Speci®cation and modelling of plastics moulding

machines for modular production systems, MSc Thesis,

University of Hull, UK.

Pawar, K. S. (1994) Organisation and managerial issues, in

Concurrent Engineering, Concepts, Implementation and

Practice, Syan, C. S. and Menon, U. (ed), Chapman & Hall,

London, pp. 49±74.

Ralston, D. and Munton, T. (1987) Computer integrated manu-

facturing. Computer-Aided Engineering Journal, August 167±

174.

Redford, A. and Lo, E. (1986) Robots for Assembly, Open Uni-

versity Press, Milton Keynes, UK.

Roobeek, A. and Abbing, M. R. (1988) The international impli-

cation of CIM. International Journal of CIM, 1(1), 3±12.

Rogers, G. G. (1990). Modular production systems: a control

scheme for actuators, PhD Thesis, Loughborough Universi-

ty, UK.

Salzberg, S. and Watkins, M. (1990) Managing information for

concurrent engineering: challenges and barriers. Research in

Engineering Design, 2, 35±52.

Shina, G. S. (1991) Special report: concurrent engineering, IEEE

Spectrum, July 22±37.

Stoll, H. S. (1986) Design for manufacture: an overview. ASME

Applied Mechanics Reviews, 39(9), 1356±1364.

Swift, K. G. (1987) Knowledge-Based Design for Manufacture,

Kogan-Page, London, pp. 11±18.

Syan, C. S. (1994a) Introduction to concurrent engineering, in

Concurrent Engineering, Concepts, Implementation and

Practice, London, Syan, C. S. and Menon, U. (ed), Chapman

and Hall, London, pp. 2±23.

Syan, C. S. (1994b) Introduction, in Concurrent Engineering,

Concepts, Implementation and Practice, Syan, C. S. and

Menon, U. (ed), Chapman & Hall, London.

Theobald, G., Culley, S. J., Ajdenan, N. J. S. and Vogwell, J.

(1993) The modelling of engineering subassemblies based on

standard components, in Proceedings of International Con-

ference on Engineering Design, The Hague, 17±19 August,

Heunsta, Zurich.

Tsukune, H., Tsukamoto, M., Matsushita, T., Tomita, F., Okeda,

K., Ogasawara, T., Takese, K. and Yuba, T. (1993) Modular

manufacturing. Journal of Intelligent Manufacturing, 4, 163±

181.

Voelker, H. and Requicha, A. (1977) Geometric modelling of

mechanical parts and processes. Computer 10(12), 48±57.

Watson, P. and Pullen, J. (1989) Engineering for competive ad-

vantage. Proceeding of the Institution of Mechanical Engi-

neers Part B: Journal of Manufacturing, 203, 69±74.

Weston, R. H., Harrison, R., Bootts, A. H. and Moore, P.R.

(1989). Universal machine control system primitives for

modular distributor manipulator systems. International

Journal of Production Research, 27(3).

Whitney, D. (1990) Designing the design process. Research in

Engineering Design, 2, 3±13.

Yoshikawa, H. (1994). Manufacturing in the future ± some topics

of IMS, in Distributed Automonous Robotic Systems Asama, A.

et al. (eds), Springer-Verlag, Berlin.


modular production systems: a new manufacturing paradigm

Documents