Integrated design of a product family and its assembly system

2.1 Principle On paced assembly lines, the assembled product takes shape gradually, start-ing with one part usually called the base part, the remainstart-ing parts bestart-ing attached a

Job #: 75068

Author name: DeLit

Title of book: Integrated Design of a Product Family

ISBN number: 1402074379

its assembly system

INTEGRATED DESIGN OF A PRODUCT FAMILY AND ITS ASSEMBLY SYSTEM

PIERRE DE LIT

Universite libre de Bruxelles, CAD/CAM Department,

50 av F D Roosevelt, CP165/14, 1050 Brussels, Belgium

ALAIN DELCHAMBRE

Universite libre de Bruxelles, CAD/CAM Department,

50 av F D Roosevelt, CP165/14, 1050 Brussels, Belgium

Springer Science+Business Media, LLC

Copyright © 2003 Springer Science+Business Media New York

Originally published by Kluwer Academic Publishers in 2003

Softcover reprint of the hardcover 1 st edition 2003

2003046112

AlI rights reserved No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording, or otherwise, without the written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work

Permis sion for books published in Europe: permissions@wkap.nl

Permissions for books published in the United States of America: permissions@wkap.eom

Printed an acid-free paper

4.2 Liaison- or command-based representations

4.3 Component- and subset- based representations

69

717172798080

82

2 Concurrent design and "optimisation" problems 872.1 Industrial, multi-objective problems and optimisation 882.2 Solution stability and iterative procedures 892.3 Concurrent development and design sensitivity 902.4 Design choices and solution space pruning 92

1.7 Description of the generic links 120

6 REPRESENTATION OF APS FOR PFS

1 Proposed AP representation

2 Examples

3 Conclusions

7 PRELIMINARY ASSEMBLY PLANNING

I Hypotheses on the APS

2 Proposed preliminary AP tool

2.6 Constructing the PGs for FEns

2.7 Merging the APS of FEns into APs for the PF

2.8 Ap evaluation criteria

3 Case study

5 PF STRUCTURING AND PRELIMINARY DFA

1 Objectives of the OFA

2 Proposed structuring and preliminary OFA issues

2.1 How to obtain variations in the subfunctions of a PF?2.2 Issues for VGCs,oocsand pseudo-varying links

2.3 PF structuring and standardisation

2.4 Preliminary OFA rules

2.5 Summary of the approach

3 Case stud y

3.1 Before the appl ication of the OFA

3.2 Application of the OFA

1.9 FEns and design teams

1.10 Synthesis of the model

Illustrative case study

144 144

146

148

151

151153

155 157

157

158 158

159160

164

168

170183

189

193

3.1 Ap at early design stage 193

Contents IX

Index

References

10 CONCLUSIONS AND FURTHER WORK

I Summary of the results and discussion

254257257257258

259 263

The achievement of this book would not have been possible without thecollaboration of theCAD/CA MDepartment of the Universite libre de Bruxelles(Belgium).

We are also especially indebted to Professors B Raucent (Universite lique de Louvain, Belgium) and Andre Leroy (Faculte poly technique de Mons,Belgium) for their constant willingness to listen to our discussions and for theirinnumerable helpful suggestions

catho-A special vote of thanks must go to Professors P Gaspart and B Leduc versite libre de Bruxelles, Belgium), H Van Brussel (Katholieke UniversiteitLeuven, Belgium), J.-M Henrioud (Laboratoire d'automatique de Besancon,France) and Dr 1.-5 Fan (Cranfield University, United Kingdom), for theirfruitful comments

(Uni-We are also grateful to II' J.Danloy, former project leader with Fabricom,now with Ets Biebuyck s.a , who collaborated with us on the CISAL project.Her pragmatic sense of the industrial problems and her knowledge of assemblysystems have been very precious

Many thanks go to our colleagues Dr B Rekiek, Dr.F.Pellichero and II'.T

L'Eglise who discussed various aspects of this work with us

We also would like to thank Mrs Irene Cluzel, from Fabricom Systernesd'assemblage, who provided us with some of the photographs in this book.Finally, we would like to thank the "Region wallonne" in Belgium for theconfidence she put in us by funding the CISAL project, carried out in collabo-ration with the Universite catholique de Louvain, the Faculte poly technique deMons and the Belgian Research Centre for the Metalworking Industry (CRIF),during which we addressed most of the topics presented in this book

Chapter 1

INTRODUCTION

In this chapter, assembly and assembly lines are first introduced The lems which are inherent to the design of products and their related assemblysystem are described The objectives of this book are set and its scope is pre-sented Section 1.1 presents assembly and the main assembly methods and op-erations Assembly lines are then briefly described in section 1.2 Section 1.3shows that product and system design are interacting processes; changes on thefirst imply modifications on the second and conversely The scope of the book

prob-is depicted in section 1.4 One typical case study that illustrates the authors'words along the book is presented in section 1.5

1 Assembly

1.1 What is assembly?

Assembly can be given two different meanings in the field of ing [131] First it can be seen as "the action of fitting together the componentparts of a machine or other objects " In its second acceptance, an assembly is

engineer-"a unit consisting of components that have been fitted together." For sake ofclarity, in this book assembly will be used in the sense of the action The object

resulting from the assembly is called a product.

The components or parts of a product are the primitive elements it is

com-posed of These components may be individual parts or group of parts ously assembled, and which are considered as a single entity in the production

previ-process Let us also introduce the term constituent, which represents a

compo-nent or a subset of connected compocompo-nents in a given product These terms areformally defined in section 4.1.1

© Kluwer Academic Publishers 2003

1.2 Assembly methods

The assembly method (or assembly mode) characterises the kind of ment used to perform an assembly operation In this book, three possible meth-ods are considered:

equip-• the manual method (which includes the semi-automated, sometimes called

manual assisted mode): this the most flexible one

• the robotic method, also called flexibl e automated mode, which is partially

flexible

• thededicated or hard automated method, which is the most specific method

with the lowest degree of flexibility

The next section describes the different types of assembly operations

The different actions involved in the completion of an assembly operationare generally divided into three primitive actions: feeding, handling and inser-tion

• Feeding a constituent means picking it up from the place it is stored in,

orienting and positioning it before its transfer to its final assembly position

• Handling a constituent consists in transferring it to its final position with

the correct orientation before its insertion

• During the insertion, the constituent is brought together with one or more

constituents thanks to more or less fixed connections (simple insertion,screwing, clipping, welding, adhesive bonding, etc )

Figure 1.1 Manual ing operation assisted with a pneumatic screwdriver (cour- tesy of Tecnimodem Automa- tion).

screw-Introduction 3

The assembly methods used to realise these three suboperations are not essarily the same Figure 1.1 shows an example of dedicated feeding (vibratorybowl feeder), dedicated handling (pneumatic transfer) and semi-automated in-sertion (assisted with a pneumatic screwdriver) Note that two of these sub-operations may be performed continuously When the same resource is used,feeding and handling or handling and insertion are not clearly separated It isthe case for example of a robot fitted with a vision system that picks up parts,orients them and moves them to their insertion place, before a press realisesthe insertion Feeding and handling have then been done continuously

nec-A brief description of these assembly operations, according to the chosenmethod, is given in the next sections

Feeding

During manual feeding, the operator takes the constituent from its storageand places it in the requested position and orientation This operation is most

of the time followed by a manual handling but this is not mandatory

In the robotic mode, an automatic manipulator picks-up the parts and placesthem in their feeding position At the opposite of the manual mode, the initialorientation of the parts cannot be completely unknown; this method is gener-ally used when the parts already have a known orientation Figure 1.2 shows arobot performing a feeding operation

Figure 1.2 Robotic feeding

op-eration (courtesy of Bosch)

Figure 1.3 Vibratory bowl feeder with its

acces-sories (courtesy of Eutomation)

A mechanical bulk feeder is most of the time used in the dedicated mode.There are plenty of different equipment to feed parts from a bulk, the vibratory

bowl being the far most used in the industry Figure 1.3 gives an illustration ofsuch a feeder.

Another feeding method, not cited in section 1.1.2, is the constitution of

ordered sets of parts or the use of kitting The constituents are generally

sep-arated and oriented before being temporarily placed in a magazine, on a parttray when dealing with ordered sets; or even directly on the pallet on whichthe product is assembled in the case of kitting The final insertion occurs later

As the parts have acquired a precise position and orientation, their insertion isgenerally easy to automate

Handling

After their feeding, constituents have to be transferred to their assemblyposition The manual handling is performed by an operator who picks up theparts and places them in their final position before assembly

Robotic handling consists in a manipulator executing a "pick and place"operation to place the constituent s Figure 1.4 shows an example of such anoperation where the components have been fed manually on a conveyor beltand are transferred by a manipulator to their final position before assembly

Parts are

placed here

Parts are picked-up here

Figure 1.4. Robotic handling operation (Menziken Automation).

When the dedicated mode is used, an equipment is especially designed forthe handling operation There is a great diversity of such equipment It can forinstance be a vibrating rail or a tube conveying the components to their finalposition before assembly Some dedicated devices are constituted of numeri-

Introduction 5cally controlled axes and have some similarities with certain kinds of robots;but their flexibility is very low, and their reusability is nearly non-existent.

Figure 1.5 Section of engine assembly line with manually assembled parts (courtesy of MG

He may in some cases use assistance tooling When the check is automated,sensors are used to confirm the correctness of the operation realised

Figure1.6 Robotic insertion with aSCARA

robot (courtesy of Bosch)

Figure 1.7 Dedicated riveting machine

(courtesy of Weber)

2 Assembly lines

A production workshop can be set up following various topologies -line,cells (islands), combination of several lines or pure job shop (isolated work-stations) As this study focuses on the design of flow-lines, only this kind ofproduction system is described hereunder

2.1 Principle

On paced assembly lines, the assembled product takes shape gradually,

start-ing with one part (usually called the base part), the remainstart-ing parts bestart-ing

attached (and other operations being performed) on the various workstations

(wss) the product visits The product is generally assembled on a pallet and the stations are usually connected by a conveyor The concept of assembly line

is illustrated in Figure 1.8 Four WSS are connected by a conveyor A set of

operations is executed at each station After a time called the cycle time Tc, theproduct is moved in the sense of the assembly and is placed in front of the next

WS. Once the last operation has been executed, the finished product leaves theline For each assembly cycle a product is finished and is collected at the end

Introduction 7

of the line and a new one is introduced at its beginning The paced assemblyline is a usual topology for medium and high production volume (cycle timevarying from several seconds to several minutes)

o ! pallet 0 /"conveyor

Figure 1.8 Paced

assem-bly line concept

2.2 Line models

Assembly lines are today often hybrid or heterogeneous, meaning that along

the same line, the production means may be manual, robotic, or dedicated.Figure 1.9 shows such a hybrid line, where manual WSs as well as robots andhard automated machines are used on the same assembly system

The princ iple of an assembly line was illustrated on Figure 1.8 for a single product line The operations performed on a ws are the same for all products

passing through it With the increasingly diversified demand in the tion of white goods and cars, more and more manufacturers devise assembly

produc-lines suited to product families (PF) The term product family is here

intro-duced without explanation Let us simply say that a car model is a variant

of a family of cars PFS are further defined in chapter 4 For this tion an intuitive interpretation is sufficient Two production strategies may be

introduc-applied: mixed-model or batch production A mixed-model assembly line is

a production line capable of producing a variety of different product models,

called product variants (PYs), simultaneously and continuously Stat ions are

then sufficiently flexible to perform their respective tasks on different PYs Forsuch lines, the workload on each station varies with the considered PY Inbatch

production, the same PY is assembled for a certain time and the assembly linesundergoes possible modifications during the batch changes So for a givenbatch, the ideal workload of the wss (not considering operating time varia-tions or jamming problems) is kept constant, but this workload is susceptible

to change from batch to batch

Let us finally mention multiple product lines In this last case, there existsignificant differences in the production processes of the different products

to be assembled (for instance cars and refrigerators), so that they cannot beconsidered asPYS of a given PF

Figure1.9 Example of an hybrid assembly line.

3 Product and assembly line design

3.1 Design decomposition

In an efficient assembly system design, the various tasks, such as productdesign , assembly process planning, assembly system layout, simulation andassembly control , should be integrated Literature (refer for instance to [50]and [67]) generally considers five main interacting modules in the product andassembly line design :

• product design based on the design for assembly (DFA) philosophy;

• assembly planning (AP);

• resource planning (RP) or line balancing (LB) and line layout (LL);

• scheduling;

• simulation

Introduction 9Product structuring andDFA

The purpose ofDFA is to evaluate and improve the design proposals in order

to facilitate the assembly of a product In most mechanical and ical products, this goal may be reached by:

electromechan-• a simplification of the product through elimination or merging of parts ;

• the identification of part characteristics that would cause difficulties forhandling, feeding or insertion, and an adequate modification of the com-ponents

Assembly planning

Stated simply, AP aims at finding out how the product could be assembled

An assembly plan (AP)' specifies a (possibly partial) order of operations thathave to be carried out in order to produce the final product out of its con-stituents

The plans are most of the time based on five types of information, whichcan be grouped into two categories [51]

Component features These are functionalities related to the specific nents of the product One distinguishes the following categories

compo-Functional attributes They allow to define a component as a standard part(screw, rivet, etc.) requiring a particular assembly sequence (AS) , adeformation during the assembly process, etc

Qualitative attributes They make it possible to define a part as fragile,inexpensive, etc., forcing it to be assembled as soon or as late as possi-ble

Relations between components They describe the contacts between nents and are classified as follows

compo-Geometrical relation attributes Such relationships are deduced from thephysical contacts (degrees of freedom) between parts

Technological relation attributes These relations, related to the joiningprocesses between components, complete geometrical ones, and inducerules and constraints on theAP

Assembly movement This information describes the movement needed toestabli sh a contact between two components

Ideally, an AP should be independent from the choice of the production sources, because it allows to uncouple both problems However, in practice theplan and the production mean s may have an important influence one on eachother, especially in the case of automated or robotic assembly

re-Resource planning or line balancing and line layout

The RP is concerned with the selection of production means adequate forperforming all the operations determined by the AP on components specified

by the DFA, while meeting the production volumes set by the marketing Forassembly lines, the definition may be slightly refined The RPaims at:

• Attributing operations to the wss

• Selecting the resources that will perform the different operations or tasks.For manual assembly lines, RPcan be simplified into LB The goal is then toassign tasks to wss

The LL aims to define a precise design of the assembly line It may be

divided into logical layout (RPandLB)and physical layout The latter consists

in deciding where each ofthe stations (with their associate resources) will stand

on the shop floor

Scheduling

Scheduling (or model launching) determines the order in which the PYS areintroduced in a mixed-model line so as to avoid station overflow (the ws hastoo much work to accomplish and causes a bottleneck) and starvation (the sta-tion is waiting for tasks to fulfil)

This scheduling task is part of the line design, because the line layout must

be robust and hold for several production mixes (proportion of each PY in theproduction) Thus , scheduling is considered during the line design and appliedduring the operation phase of the production system

Simulation

The last step in the design process is the simulation of the assembly line toverify the validity of the proposed layout

3.2 Influences and interactions

The various modules of the design process are influenced by each other.DFAinfluences the further AP ;as the assembly plan is a data for the line design ,

it seems clear that AP has a strong impact on RP; finally theLL influences thescheduling

Less obvious influences between modules are the "backward" ones Theseinfluences are seldom taken into account, although they may have a consid-erable impact on the quality of the design Exploiting them allows to adaptthe product design in order to improve the related assembly system These

influences can be subdivided into a posteriori and a priori influences.

An a posteriori influence is actually a feedback It happens when the results

of one module are not satisfactory Suppose for instance that because of the

Introduction 11product design, no good AP can be found Product design modifications arethen proposed, in order to improve theLL.

When a priori influences are taken into account, some choices done

regard-ing the line design impact the product design Suppose for instance that theline designer decides to use an available SCARArobot to perform an operation.This kind of robot only possesses one translation degree of freedom (verticaltranslation) So the use of this resource implies that the considered operationcan be done with a vertical translation

Thus, the line design influences the product design An efficient designapproach shall take both these influences into account

The reader shall note that scheduling may have an influence on the RP, cause the proposed LL must not be too sensitive to changes in the production

be-demand So an a posteriori influence of the scheduling on the RP shall beconsidered for an efficient line design

4 Scope of the book

The last two decades have witnessed important shifts in customers' haviour Manufacturers face fierce competition because of shorter product lifecycles and faster delivery of new products on the market The time-to-markethas become an essential issue, as it enables quick response to changing marketpreferences, and the continuous introduction of innovative technology

be-Companies now need to integrate customers in their strategies and turers have to propose a large variety of products to meet the market's demand

manufac-In the automotive industry for instance, manufacturers strive towards mass tomisation, trying to offer a range of products wide enough to satisfy nearlyeveryone [2] Increasing the range ofPYS attract and retain customers, yet itdramatically augments the cost and complexity of the manufacturing systems

cus-To face the challenge of the time-to-market, designers' effort gave rise to theconcurrent engineering (CE)concept Its main principle is to integrate productand process development to reduce the design lead-time and to improve qual-ity and cost This integrated approach makes it possible to face the problems

of interdependence in designing products and the corresponding productionprocesses However achieving the effective integration is a difficult task Ef-ficient communication and coordination among the diverse disciplines are keyconcepts in ensuring the success ofCE [96]

Itis now widely accepted that assembly is a critically important function in

industry According to surveys presented by Nof et al [127], the assembly of

manufactured goods accounts for over 50% of the total production time , andfor 20% of the production cost; 30% to 50% of the labour costs are involved

in assembly tasks Assembly however fails to attract both the attention and theinvestments it deserves and requires Most assembly work is still undertakenusing methods and tools that have changed little in the recent years Twenty-

years old principles such as the DFA are still not considered in an appreciableamount of companies It stays often believed that the reduction of costs isconditioned by the improvement of the production process While progress

in this area indeed can decrease costs, a much important part of these costs isconditioned by the product design

As many decisions taken at the design stage of the product are decisivefor its entire life [126], it is crucial to integrate the process and the assemblysystem when designing the product As much as 70% of the life cycle cost of aproduct is determined at this stage Changes introduced later greatly increasethe production costs It is therefore very important to make the right decisions

at an early stage of the design process In this respect, an important researcharea nowadays is the development of tools to bridge product and assembly linedesign

The main objective of the research leading to this book was to develop an integrated approach for the design of a product family and its assembly sys- tem The authors restricted their field of investigation to ordinary assembly

(centimetric parts and bigger with submillimetric tolerances) By ordinary, itmust be understood that mini-assembly (millimetre-sized parts with micromet-ric tolerances) and micro-assembly (micrometre-sized parts with submicromet-ric tolerances) [29, 30, 52] are not within the scope of this book, because theconsidered processes are not the same as for classical assembly As there is avast amount of research applicable to assembly systems, there is still a need forcircumscribing the investigated area Thus, the book considers the following :

• Assembly systems follow a flow-line topology This does not only involvelinear lines, but the proposed approach does not hold for job shop or cellularmanufacturing environments, even if some of the results presented in thisbook could be applicable to such environments

• The PFis assembled on heterogeneous mixed-model assembly lines

• The scope is limited to medium- to high- production volumes , with a cycletime varying from a few seconds to several minutes

• A first embodiment design [130] of the considered PF is available Themethodology proposed here may be initiated at very early design stage,but its achievement asks at least for first sketches of the considered PF

Successive refinements of the PFdesign are part of the design process andmay thus occur during the several stages of the proposed approach

• It is supposed that the assembly line can be somehow uncoupled from theremainder of the manufacturing process The connections with such pro-cesses and the implications of the assembly system design on the manufac-turing are supposed to be mastered

Introduction 13

Collaborations with industrials made it possible to test the proposed ology and tools on several case studies, that will illustrate the exposed conceptsthroughout this book One of them will be presented completely, from the earlysketches until the line design : a family of signalling relays whose general as-pect is illustrated in Figures 1.10 and 1.11, and whose functioning is describedhereunder Redesign possibilities on this PF are important; it thus appears to

method-be an interesting didactic example

A relay is composed of thirteen parts described in Table 1.1 (attachmentcomponents [screws] are not detailed in the table) The plastic coil (c1), theiron kernel (k), the iron fixed armature (fa) and the iron mobile armature (rna)constitute a magnetic circuit e. When a current J is applied to the coil, itinduces a magnetic field H in the kernel, which can be determined by thefollowing law (190]:

of the void (4n10-7 Him) and fLthe permittivity of the circuit The coil is

a support for the turns inducing fI and can be neglected in the formulation.Equation (1.2) yields

To get a high magnetic pressure, u, must be high This is why iron is used for

k, fa and rna(u; "'-' 103 )

The contact support (cs) is placed on the mobile armature and the returnspring (rs) The hinge (h) and the stiff plates (spl and sp2) allow the move-ment of rna under the magnetic pressure When the current Jis applied, rna isattracted by the kernel, which thus provokes a displacement of cs Two con-tacting states can thus be achieved on one relay according to the applied current(open, close)

The family includes fou r variants: large and small model, with or withoutspecial contact configuration (the exact nature of this difference is not impor-tant in this example) A docket (d) is glued on the cover for the special contactconfiguration The description of the family :F and its variants (described as a

small (bo.)

large (cl.)

small (e1s) large (kj )

small (ks) large (CI) small (cs) normal (ma)

Component (generic)

Contact support (cs) Contacts (ci)

Return spring (rs) Hinge (h) Stiff Plate I (spl) Stiff Plate 2 (sp2) Docket (d)

Variants

large (csj) small (cse) large normal (ci'n) large special (cils) small normal (cisn) small special (ciss) normal (rs) normal (h) normal (sp I)

normal (sp2) normal (d)

set of components) is the following:

:F = {.1'I' 1'2, 1'3, 1'4 }

.1'i = {fa" rna, cI" k.,C" h, sp 1, sp2, rs,CSi ,ciln,bod

.1'2 = {fas ,rna,cis, ks,cs,h, spl , sp2, rs, cs, cisn,bOs}

.1'3 = {fa" rna, cl., k" C" h, spl, sp2, rs,CSt, cils,bo., d}

.1'4 = {fas,rna, cis,ks,c., h, spl, sp2, rs, cs,, ciss,bo.,a}.

6 Disposition

A flowchart of the book structure is shown in Figure 1.12 and discussed low; the reader shall note that it is not the flowchart of the design methodology,which is explained in section 3.3 and in chapter 10 Dotted lines represent nonessential paths to the appendices

be-2 State of the art The state of the art proposed in chapter 2 covers eral aspects of the product and assembly line design The authors had notthe ambition to propose an exhaustive literature review; they preferred topresent some of them with greater detail An originality in the survey lies

sev-in its systematisation An illustrative example has been chosen, and formerstudies are explained and compared to each other using this example or part

of it It gives a better insight in and comprehension of the references cited

3 PF and assembly line design methodology The methodology proposed for

a concurrent design ofPFSand their assembly system is described in

PF structuring and 5. Representation

preliminary DFA of APS for PFS

9.

Line layout .

tConc lusions and 10.

4 Description of product families The description of a structure mode l for

PFS is presented in chapter 4 This model is capable of supporting the sentation of partial information on the PF at early design stages The FEns,introduced in chapter 3 are forma lly defined in this chapter

repre-Introduction 17

5 PF structuring and preliminary DFA Chapter 5 deal s with early PF design

issues It puts the accent on the importance of an efficient PF structuring,and extends the classical DFA rules to PFs The DFA process follows the PFdecomposition: the FEns are first analysed separately, and the design of thewhole PF is then considered

6 Representation of APs for PFs Chapter 6 presents the proposed AP

rep-resentation for PFS It is based on precedence graphs (PGs), completed topresent several information, such as insertion directions, the nature of thecomponents, or the structure of the PF

7 Preliminary assembly planning The preliminary AP is presented in

chap-ter 7 This stage aims at proposing a set of PGs between components forthe PF For AP too, the decomposition scheme of the PF is integrated in thedesign methodology

8 Detailed DFA and AP In chapter 8, detailed design and planning issues are

addressed It describes the selection of joining processes, the choice ofoperating methods and the selection of potential equipment for each opera-tion

9 Line layout Chapter 9 is dedicated to line layout It covers the

decompo-sition of the line according to the PF structure, the selection of conveyingequipment, LB and RP issues

10 Conclusions and furth er work Finally, chapter 10 presents the conclusions

of this book and suggests possible continuations to the present research.The appendices describe the precedence constraint operator using predi-cate calculus (appendix A), give an overview of the multicriteria decision-aidmethod PROMETHEE [I(appendix B), and finally present a glossary of the book(appendix C)

Notes

1 Both the assembly planning and the assembly plan are denoted by the sameacronym; the context allows to make the distinction

STATE OF THE ART

A survey of previous work in the field of products and their assembly systemdesign is presented and discussed in section 2.1 Section 2.2 gives a review ofrecent work on product family (PF) models, together with an illustrative ex-ample As there are numbers of definitions of a PF, the state of the art on

PF representations reports various definitions of a PF and of a product variant(rv) Previous studies on design for assembly(DFA) is reported in section 2.3

An extensive literature overview on assembly plan (AP) representations is sented in section 2.4 A literature survey on AP is proposed in section 2.5 Astate of the art on line design and material handling equipment(MHE)selection

The first approach sees design as a problem solving approach, hence as atrial and error process The development of the problem and the solution arespiral-like: results of the previous cycles are part of the input for the next one.This basic design cycle [164] is an analysis, synthesis, simulation, evaluationand decision process It is the most fundamental model of design and is part ofeach stage of any design process

The second design process is a sequence of phases starting at a certain straction level, such as concept development, each subsequent stage being afurther elaboration with an increasingly concrete level A very famous exam-ple of such a methodology has been proposed by Pahl and Beitz [130], who

ab-© Kluwer Academic Publishers 2003

20 DESIGN OF A PRODUCT FAMILY AND ITS ASSEMBLY SYSTEM

described the steps to follow in product planning and design process as shown

in Figure 2.1 A characteristic of these models is that they propose a chical approach of the design Itis first performed at a very general level andpartly decomposed into its subfunctions establishing a functional structure ofthe design

hierar-There is finally a development process in which the design activity is notconsidered alone anymore, but integrates production and marketing concerns.Andreasen et al. [6] suggested, in order to reduce costs, to simultaneouslydefine the product and the processes in an integrated product development en-vironment Similar methodologies were also proposed by Pugh [144] or Ulrichand Eppinger [189] Note that these approaches also begin at a certain abstrac-tion level, starting from a concept which is little by little specified and refined.Stoll [181] proposed the process-driven design approach involving a sys-tematic consideration of the downstream processes at early stages of the con-ceptual design Instead of beginning with design optimisation and then consid-ering part fabrication and assembly, this approach considers manufacturabilityand design goals, product and process plan together with ease of assembly andcomponent fabrication, before optimising and refining product performanceand piece-part cost

1.2 Integrated design approaches

Lim et at. [107] and later Zha et at. [206] reviewed relevant literature onthe development of methodologies and systems for integrated intelligent de-sign of assembled products and processes It comes out from these articlesthat although the interaction between product design and AP has been deeplystudied (refer for instance to [9, 10] for recent contributions), little work hasbeen accomplished in the field of concurrent product and assembly line designmethodologies Moreover, the development of a methodology for PFSand theirassembly system design, which is the main concern of this book, had to the au-thors' knowledge not been addressed before the CISAL project [46] So, thestate of the art will focus on works (or groups of works inside an institution)which tackled both the product and line design problem

As explained in section 1.3, many approaches cited in the literature considerthree main interacting modules in the product and assembly line design:

• product structuring and design for assembly (DFA);

• assembly planning (AP);

• resource planning(RP) and line layout(LL).

The main differences between the proposed methodologies are the nature ofthe results of the modules and their interactions one with each other The de-

Task Market company economy

~

Plan and clarify the task

Analyse the marke t and the company situation

, Find aud select product ideas

Formulate product propo sal

I Clarify the task

I Elaborate a require ment list

I Identify esse ntial problem s

r Establish function structures

Search for working principles and working structures

I Comb ine and firm up into concept variants

I Evalua te against technical and economic criteria

I Co ncept

I (Principle solution)

I De velop the construction structure: §

I Preliminary fonn design , material selection and calculerion 1

r- Select best preliminary layouts

I Define the construction structure :

I Prepare the preliminary parts list and production and assembly

I Prepare producti on and operati ng documents:

! Elabora te detail drawing s and parts lists

Complete production assembly transport and oper ating instructions

Oieck all documents

Figure2.1 Steps of the planning and design process according to Pahl and Beitz (from [130]).

©1996 Springer- Verlag London Limited Reproduced with permission.

scribed methodologies start with a conceptual, an embodiment or even with

a first detailed product design (a conceptual design is the generation of ciple solutions that meet design specifications; the embodiment design is the

prin-22 DESIGN OF A PRODUCT FAMILY AND ITS ASSEMBLY SYSTEM

development of the design from the concept in accordance with technical andeconomic criteria; during the detailed design the arrangements, shapes, dimen-sions and surface properties of all individual parts are laid down [130])

Charles Stark Draper Laboratory

In the late eighties, Nevins and Whitney [126] from the Charles Stark DraperLaboratory(csot.jin Cambridge (Massachusetts, USA) proposed a concurrentdesign method for products and manufacturing processes Assembly is con-sidered as the centre point of their vision, due to its implications at the earlyproduct design stage

rr===tI PRODUCT DESIGN AND MODELLINGI

Figure 2.2 Rough fiowsheet of the product and assembly line design approach at theCSDL.

In the nineties, Whitney's team methodology filled out and the team oped software to help designers to devise assembly systems The method isroughly represented in Figure 2.2 It consists in two main stages: assembly se-quence analysis (ASA), a technique yielding feasible assembly sequences (ASS)for a mechanical product, and the choice of the cheapest physical layout Theproduct is modelled by its liaison diagram The precedence constraints (pes)between liaisons are then determined either by questions [42] or cut-set de-composition and questions asked to the user [8] These PCs can then be used

devel-to pinpoint candidate DFA-related redesigns for relaxation of assembly straints so as to ease assembly [41] All ASs are then generated and presented

con-as a liaison-sequence diagram [42] Subcon-assembly partitioning may also be formed [161] The ASS are then edited and studied so as to eliminate undesired

per-ones [1] One or few favoured sequences then remain and may lead, if notsatisfactory, to detail refinement or profound redesign.

The second stage aims at choosing the resources of the assembly system [77]

on the basis of the chosen assembly sequence This RP is suited to deal withmulti-variant products starting from theASfor each of them

Laboratoire d'automatique de Besancon

The Laboratoire d'automatique de Besancon (LAB) in France proposed anintegrated approach for the design of products and assembly processes [180].They considered the design as a succession of refinements from a higher to alower abstraction level Progressions occur in the functional, product, and APdomain, and are integrated to take the dependencies between the three domainsinto account Assembly is viewed as a hierarchical decomposition process,each decomposition being interpreted as a design step

From an initial definition of the functional requirements, partial descriptions

of the product are synthesised Assembly alternatives are then generated fromthis partial product model, and studied Constraints and results of the decisionstaken in the assembly process domain can be returned so as to suggest changes

in the product or constraints in its design refinements A progression in theassembly process design is associated with each progression in the productdesign The team applied this methodology to AP (which was developed forPFS), but the LL is performed once the AP has been completed The mainsteps of the methodology (without taking control into account) are presented

in Figure 2.3 The functional specifications, product analysis andAP [87, 177]are performed concurrently, finally yielding promising assembly trees

Finally, the assembly system is designed Almost all aspects of the linedesign were adapted toPFS [55, 121, 136], but the logical layout determinationfor PFS is limited to line balancing (LB).

Rampersad

Rampersad [145] at Eindhoven University of Technology (The Netherlands)proposed an integrated and simultaneous design of robotic assembly cells Hedescribed the product design as a spiral, in which interactions between the dif-ferent activities (product, assembly process and assembly system design) takeplace with an expected final convergence Rampersad performs the productand assembly process analysis with the so-called "DFAhouse" quality functiondeployment analysis The workstation (ws) design is realised with morpholog-ical charts showing concrete system components able to perform an operation.This approach does not pay much attention to the determination of the mostappropriate AS and LB is not considered However, the method possesses astrong design orientation which influenced later studies

24 DESIGN OF A PRODUCT FAMILY AND ITS ASSEMBLY SYSTEM

ASSEMBLY PROCESS PLANNING

Figure2.3 Logical design progression steps of the product and assembly line design approach

at theLAB

SCOPES project

A global approach to design single product heterogeneous assembly lineswas the result of a collaboration (European ESPRIT III project 6562 SCOPES)between five research laboratories: Cranfield University (UK), CRIF and Uni-versite libre de Bruxelles (ULB, Belgium), University of Stuttgart (Germany)and Ecole polytechnique federale de Lausanne (Switzerland); and two indus-trial companies: Schneider Electric and Dassault Systemes (France) This in-tegrated control model is presented in [51] The off-line part of the softwaredeals with product design, AP,assembly system layout and simulation, produc-tion scheduling, flow control and error recovery The project also integratedon-line production scheduling, flow and quality control, and error recovery.The whole project was integrated in the CATIA CAD system The approach isillustrated in Figure 2.4

A> plays an important role in SCOPES, but the approach can be classified

as "line centred", because the RP is performed on a precedence graph (PG)between operations So the line design is performed on a large set of potentialsequences, and it is the Rp that finally fixes the AS Thus, Rp is a centre point

of the line design

SCOPES had the potential to provide a concurrent sequence generation anddesign environment However, the opportunity has not been fully exploitedand the AI' is still performed once the design has been fully defined

Product design Assembly planning Resource planning Simulation Scheduling Flow control Control MonitoringIanalysis

'"

c

1(5

Figure2.4 SCOPES method forCEof assembled products (from[51 J). © 1996John Wiley

& Sons Limited Reproduced with permission

Unlverslte catholique de Louvain, Faculte polytechnique de Mons

A team at the Universite catholique de Louvain(VCL) and the Faculte technique de Mons(FPMS)(Belgium) described a method for "interactive prod-uct and assembly system design", applicable to single product heterogeneousassembly systems [139, 141, 192] The flowchart of their methodology is given

poly-in Figure 2.5

The approach is composed of three steps First the initial design of the uct is performed, based on the product functionalities and direct DFA (generalalways pertinent DFA rules) This design will be optimised during a secondstage, at which a comparison between different product designs is performed

prod-by computing an economic indicatorCIC (comparable indicative cost) for eachoperation and for different assembly modes (manual, robotic and automated).The result of this stage is an optimised product design and an operating chartwith the corresponding assembly method The third step concerns the LLprob-lem and generally slight product design adaptations are still allowed This ap-proach does not use computer-aided optimisation and selection tools for APand LB; the aim is rather to find a good AS and a satisfactory LB and then

to make the necessary adaptations on the product design A very interestingfeature of these studies is that the authors assumed that there exists a mutualadjustment mechanism, in which redesign advice is deduced from the choice

of assembly equipment

Petit

Later on, Petit [138] proposed an original approach based on a gradualrefinement of the product and assembly line specifications The proposed

26 DESIGN OF A PRODUCT FAMILY AND ITS ASSEMBLY SYSTEM

FEASIBILIT Y DIRECTDFA

OPERATING T IMES RELATIVE COSTS

PRODUCTION VOLUME

Figure 2.5 Flowchart of the interactive methodology developed at the UCL and the FP Ms

(fromr139])

methodology does not focus on the search of an optimal AP or line design, butrather on the adequation between the product and assembly line all along thedesign process His vision of the DFA is structured in two stages [142]: a sim-plification of the product structure through the application of a limited set ofdesign rules, and a progressive adaptation of the product and operation designparameters to the designed assembly system The product design parametersand assembly system feasibility domains are modelled using possibility theoryand fuzzy intervals

The RP is based on "working principles" (wrs), defined as a set of resourcesmaking possible the execution of a given assembly operation [140]; severalWPs may be suitable for a task The WPs are grouped into WSS on the basis

of their affinity The determination of suitable working principles will lead toredesign advice and changes in a first AS

The advantage of the approach is double Firstly, it allows to design theline gradually, some WSS being almost entirely designed as the layout of otherones stays roughly known Secondly, it helps dealing with fully automatedsystems with masked times The method is at the current state of developmentnot suited to deal with mixed-model assembly lines

CISAL project

The CISAL project was a collaboration between three Belgian universities(ULB, UCL, FPMS), initiated to propose an integrated tool to concurrently de-sign products and their production means A description of this project can befound in [46], CISAL deals with heterogeneous assembly lines, but its mainoriginality is that it is applicable to PFs The tool is subdivided into three mainmodules: product analysis, operating modes and assembly techniques, and linelayout, as illustrated in Figure 2.6

The purpose of the product analysis module is to make a first analysis of the

PF and to generate a first set of PCs between assembly operations The secondmodule suggests possible assembly techniques for each operation, proposes alist of related equipment and computes the related operating times and costs.Finally the third module, in charge of the LL, elaborates the logical layout ofthe line (i.e allocating the operations among the WSS along the line), and deter-mines the most efficient product mix and production sequence An importantaspect of the software is the interaction between modules and the possibility

of feedback from one of them on the others An integrated database storesall the information needed concerning the products, the operating modes andtechniques, and the line layout

Discussion

All the mentioned studies tackle the design of heterogeneous assembly lines,except Rampersad's approach, which was especially developed for robotic as-sembly Some of them partially deal with PFs, but in most of them some designsteps are missing

It comes out from the state of the art that there are two main types of designapproaches of product and process The first is more to be used in an iterative

way A product design is evaluated from the assembly point of view, and theresults of this evaluation are used for product redesign; this design loop isapplied repeated satisfactory performances are reached It corresponds to thea

posteriori backward influence mentioned in section 1.3.2 The second is more

an interactive product and assembly line design, defined as follows by Petit:

"The objective of the interactive product and assembly system design can be stated

as being to design, in the shortest time, a product and an assembly system with ahigh degree of compatibility in order to assemble the product at a required productionrate." [138]

In a truly interactive approach, the main assumption is the existence of a mutualadjustment mechanism, in which redesign advice is deduced from the choice

of the assembly equipment It corresponds to the a priori backward influence

cited in section 1.3.2 This implies that the product and assembly line designcharacteristics possess latitudes than can potentially be used to improve thecompatibility of assembly operations and assembly equipment The matching

28 DESIGN OFAPRODUCT FAMILYAND ITS ASSEMBLY SYSTEM

I I I

/ /

/

/ /'

so that the sum of operating times on a ws does not exceed the cycle time (CT)

of the line In the case of a truly interactive procedure (where the line designinfluences the product design from early design stages), AP is considered as a

less crucial problem It shall be noticed yet, that the RPperformed on the basis

of this planning may yield poor results, because of a narrow solution space, andmay thus suggest unnecessary product redesign or lead to equipment overcast

Table 2.1. Summary of the reviewed product and assembly process design approaches.

CSDL LAB Rampersad SCOPES UCL-FPMS Petit CISAL

to design their assembly systems However, this PO is deduced from a set

of eligible AS, thus requiring a former planning This explains why their POgeneration is called "partial"

Rampersad's approach was considered to be suited to single products, cause although the case of variants is evoked in his work, the author gives littleindication on the way to actually proceed to tackle this problem The authorsqualified the approaches of UCL-FPMS and the CISAL project as partially inter-active, because they start with a first product design, which is improved takingthe future line into account

be-2 Product family representation

2.1 Terminology

The terms "product family" (or "family of products") and "product ant" can mean different things, even within the context of manufacturing andassembly As pointed out by Fan and Liu [69], these terms are often used with-out any clear definition This difficulty to establish a common terminology is

vari-a mvari-ain problem in computer vari-aided design vari-and mvari-anufvari-acturing vari-activities [168]

30 DESIGN OF A PRODUCT FAMILY AND ITS ASSEMBLY SYSTEM

re-as PF, but PF is more generally used

Literature gives, among others, the following definitions for a PF

Danloyet al [40] Each product of the PF includes some components which

are common to all products of the family and other ones which are optional.These optional components can be shared by several products of the family.The PFthus contains a basic option for the common components and severaloptions which constitute the different PYsof the family

Dufrene [56], Perrardet al.[136] A PF is a set of products manufactured side a same enterprise and which main functions are identical

in-Erens [57, 59] A PF is a concept created for a market, but that meets the sumers' expectations by introducing variety in a defined product architec-ture and in a defined manufacturing process Example of PFS are cars andwhite goods

con-Falkenauer and Delchambre [66] A PF is defined as a set of distinct ucts called variants, which main functions are preferably similar

prod-Fan and Liu [69] A PF is a group of products based on a specific design cept or from a standard/parent product They are assembled in the so-called

con-"mixed model" or "multi-product" assembly line

Laakko and Mantyla [102] In terms of process planning, PFS are defined onthe basis of similar routings and manufacturing processes All productsinside the family share a common set of individual features

McKayet al [112] A family of products identifies the similarities and

differ-ences between individual products forming a range of products

Meunier[113] A PF is a set of objects responding to one or more criteriadefining a kinship between them

Meyer and Lopez [114, 115] A PF is defined as a set of products sharing asame core technology and addresses a set of related commercial applica-tions

Meyer and Utterback[116] A PF aims at a market segment, as the products

or some product groups inside the family aim at a niche in these segments

Miller and Liberatore [119] A PF represents an intermediate grouping ofend products requiring similar raw materials and manufacturing processes.

An essential characteristic of the family is that all end products share nearidentical features regarding their production This means that any produc-tion line able to produce one of them is essentially able to produce any otherproduct in the family

Rampersad [145] APFcomprises a collection ofPYSwith important ities in their characteristics

similar-Rekieket al.[157] A PF is a set of products presenting common functionsand some features differing from product to product

Sapelkin [167] A PF is a combination of products, assemblies and nents that are linked by collective connections which enable the family toevolve through change of its constituents It is an artificial formation; itsconnections are informational A family is the result of collective work ofgenerations of engineers

compo-Siddique and Rosen [172] APFis defined as a set of products that share mon technology and address a related set of market applications

com-Simpson [174] A PF is a group of products which share common form, tures and functions, targeting one or multiple market niches

fea-Stadzisz and Henrioud [177, 178] A PF is defined as a set of similar ucts whose main functions are identical This means that all these productsare variants of a same functional product The variants are due to optionalparts, differences among the secondary functions or differences between thegeometrical features of the products The products of a family present sim-ilarities in their assembly process; the production volumes for each product

prod-of the family as well as their lifetime do not justify a specific assemblysystem for each of them

Ulrich and Tung [188] A PF is a set of end products made from a muchsmaller set of components

The references above show that definitions presented are either related, putting the accent on a commonality of the constituents of a PF, orprocess related, focusing on the processes and technologies used to producethe products inside the family Some definitions are also based on a commonmarket segment A comparison of the PF with other terms such as "productplatform" and "product range" is given by Erens [57]