You are on page 1of 12

Materials and Design 23 2002.

405 416

Selection of joining methods in mechanical design


C. LeBacq a , Y. Brechet b, H.R. Shercliff a,U , T. Jeggy c , L. Salvo b
a

Engineering Department, Cambridge Uni ersity, Trumpington St, Cambridge, UK b INPG, Domaine Uni ersitaire, 38402 St Martin d Heres, France c CETIM, St Etienne, France Received 1 August 2001; accepted 22 November 2001

Abstract This paper presents a methodology for the selection of joining process in design, implemented in software. The method captures important coupling between the process options, material and design detail, using databases assembled by discussion with experts on mechanical fastening, welding and adhesives. The approach rst eliminates solutions that are not technically feasible, with reasons, and then ranks the remaining solutions on the basis of the degree of agreement with the set of design requirements using a fuzzy logic algorithm.. The software implementation presents the designer with a questionnaire, to assemble the set of requirements in terms of the joint geometry, the materials to be joined, the functions required from the joint, and production conditions. Consideration is also given to other local factors that may inuence the selection, such as the availability of equipment and necessary skills, or current practice. The application of the methods using the software is illustrated by a case study. The selection tool successfully restricts the number of processes to be evaluated further in a given design, for example, before undertaking the more difcult task of detailed cost evaluation. The method is also successful in suggesting modications to the design to facilitate joining, or indicating possible innovative use of unfamiliar processes. 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Materials joining; Selection of material processes; Weighting and ranking factors; Design for assembly; Databases

1. Introduction Industrial products are very rarely monolithic. The problem of joining components is therefore a key issue in the design process. Joining components is a way of allowing simpler forms for the individual components, which are thus, easier and cheaper to process. However, very often joints are weak points from a mechanical or chemical viewpoint: for instance many failures in fatigue or in corrosion occur at welded joints. In order to avoid these problems, special processing conditions or non-destructive testing have to be used, depending
U

Corresponding author. Tel.: q44-1223-332-627; fax: q44-1223332-662. E-mail address: hrs@eng.cam.ac.uk H.R. Shercliff..

on the joining method and these inevitably lead to extra cost in making the product. Joining is also a problem from an efciency viewpoint: it often needs some extra material to be added to the structure such as screws, bolts, or welding ller metal.. And it sometimes leads to local weakening of the mechanical properties of the material of the components for instance in the heat affected zone of a weld.. All these aspects usually lead to the application of safety factors and thus, an increase in mass needed to full a given structural function. In general, the number of joining operations has to be minimised in order to decrease the overall cost of a product. Moreover, recent trends toward recycling may lead the designer to consider disassembling as well as assembling components. Selecting the appropriate joining method, as well as de-

0261-3069r02r$ - see front matter 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 2 6 1 - 3 0 6 9 0 1 . 0 0 0 9 3 - 0

406

C. LeBacq et al. r Materials and Design 23 (2002) 405 416

signing the components while keeping in mind the joining problems, are important aspects of efcient design. This problem can be dealt with at several levels. The design for assembly methodology w1 4x allows a reconception of a product to minimise the number of joints, and an estimation of the time needed in joining for a given product. The methodology we propose here takes the following view of the problem: assuming a joint has to be made, what is the best method to do it? One approach to this problem currently under development is TWIs knowledge-based software JoinIt w5x which, for example, provides answers to frequently asked questions to help identify and avoid common mistakes. Some textbooks systematically summarise the characteristics of a range of joining processes, and provide selection matrices showing the match between each process and a few dominant criteria w1,6x. In recent years, one class of selection tools for mechanical design has been developed around a general methodology of screeningrrankingrfurther information w7x i.e. rst eliminate options which are not viable, then rank the remainder on suitable criteria, and nally point the designer towards sources of detailed information on the most promising options. Following this approach, a general software package for material and process selection has been developed, with a uniform format for the data w8x. At a more specialised level, a task-based analysis appears to be most efcient to capture the specicities of each situation w9 12x. Examples are a selection of polymer composites w13x cast aluminium alloys w14x and cutting methods w10x. The task-based approach recognises that the process characteristics which are truly selective require a treatment which is specic to the given task, so that the questions to be answered for the selection of a joining process are quite distinct from those to select among, say, casting processes w9,10x. We have used a task-based method, rst to eliminate the solutions, which are not technically feasible, and then to rank the remaining solutions with respect to their degree of agreement with the set of requirements. For this ranking, a fuzzy logic algorithm developed for materials selection has been applied w15,16x. Once a joining method is selected, the decisions concerning the most appropriate choice for the process parameters usually require some input from modelling w10x. The present paper is limited to the initial selection of the process, i.e. it is expected to identify for a given application if for example. arc welding is better than epoxy adhesives, but it is not able to specify the arc weld power required. The paper is organised as follows: In Section 2 the set of requirements typical for a joining selection problem is listed. The structure and the content of the databases developed for joining processes are detailed

Fig. 1. Flow chart of the methodology for selection of joining processes.

in Section 3 and the selection algorithm is explained in Section 4. This information enables a technical evaluation to be made of the different joining processes. This technical evaluation is only one aspect of the problem: depending on the availability within a company of the equipment or the required skills, and depending on the current practice in the eld of application, a technically acceptable solution may still not be used. In Section 5 we present a possible method to deal with these criteria which require further local information. The overall structure of the methodology is given in Fig. 1: for a given set of requirements corresponding to assembling two components in a product, the joining processes presented in the database are rst screened in order to eliminate the ones which are technically impossible. They are then ranked according to their technical agreement with the requirements leading to a technical mark between 0 and 100%.. Furthermore, information concerning the local conditions equipment of the company, class of application. then provides a local evaluation for the technically acceptable solutions, to complement the technical evaluation. This methodology has been implemented in proprietary software by CETIM. The application of the method using the software is illustrated by a case study in Section 6.

2. Design requirements for joining One approach to capturing the information needed in order to select a joining method is to use an expert questionnaire. An important aspect of building selec-

C. LeBacq et al. r Materials and Design 23 (2002) 405 416

407

tion tools for the non-expert is the provision of clear guidance as to what the important design issues are. A questionnaire approach deals with this efciently, particularly for secondary processing applied to components such as joining and surface treatments . w9x. A joining questionnaire has been derived from discussions with experts on the three main categories of joining methods: mechanical fastening, welding and adhesives. For selection within each of these domains, say for instance to select between an epoxy and a silicone glue, the questionnaire would need to be more precise, but since we want to compare joining methods from all three classes, the questionnaire has been made relevant to all joining processes. This is a common situation in handling data for selection in design a hierarchy of data is required, with the type of information growing more specic as selection moves from the generic level to the detail level. The important point is that, for the given selection problem, the data must be relevant, discriminating and available w7x. The questionnaire addresses in sequence four types of design issue for joining:

Fig. 3. An example of a specic geometry for an angle joint.

Step Step Step Step

1: 2: 3: 4:

the the the the

geometry of the joint materials to be joined functions required from the joint joining production conditions

The user is free to answer as many of the questions as are considered relevant for the design in hand the benet of the questionnaire is that important criteria are not inadvertently overlooked. The geometry of the joints has been classied as follows: at the most general level, ve classes are listed: plane on plane, aligned cylinder on cylinder, angles between planes, cylinders intersecting cylinders, cylinders intersecting planes. These generic geometries Fig. 2. can be specied further as shown in Fig. 3 for a specic angle shape. However, one purpose of the

software is to allow viable joint details to be identied for each process, if the user only knows the generic joint geometry. Information concerning the size of the components is also required: not all-joining methods are suitable for all sizes, and this can be a selective criterion w6x. The materials to be joined are an important selection criterion: both the specic material types which can be the same or different for the two components. and the surface state roughness, porosity and coatings. should be specied to select the joining process. In the software, the user answers these questions by selecting from standard lists: materials from the material database which covers metals and polymers., roughness and porosity from a list of standard classes, and surface coatings from a list of usual coatings paint, zinc, calamine, grease..... A question concerning the possibility of retaining the coating after the joining process is also proposed. The functions to be fullled by the joint are of course of central importance. The most obvious one is the mechanical transmission of loads. The questionnaire asks the nature of forces and moments acting on the joint and their relative importance Fig. 4.. It is known, for example, that adhesive joints are not good in tension whereas they can be reasonably efcient in shear: this kind of information and the consequent detailed joint geometry. will be of importance in the selection process. The detailed dimensioning of the joint to sustain a given state and magnitude of loading is beyond the scope of this software detailed stress analysis being required for this aspect of joint design.. Other questions are asked concerning the service conditions of the joint: at what temperature will it be

Fig. 2. Generic classes of joining geometries.

Fig. 4. An example of loading conditions specied in the set of requirements transverse tension and transverse bending..

408

C. LeBacq et al. r Materials and Design 23 (2002) 405 416

used, and in what environment dry air, humid air, water, marine environment, acid, alkali, oil, UV etc..? Joints may also need to perform specic non-mechanical functions, for example, the questionnaire includes:

waterproofness degree of precision required for the joint requirement for disassembly hygienic quality electrical conduction thermal conduction vibration damping

The questions discussed above are mainly focussed on the function of the joint, but these are not the only selection criteria. The production conditions to realise the joining may be very restrictive, for instance in the extreme case of a sub-marine assembly problem such as offshore platform maintenance. The questions asked of the user are:

location of the joining operation preparation of the components needed normal economical operating conditions possibility of automating the process safety requirements for production

Furthermore, specic production questions can also be selective, such as the minimum time allowed before moving the joined components after the operation, the possibility to reorient the components before joining, or the accessibility of the two sides of the joint. For instance, bolting is impossible if one side is not accessible and the use of certain adhesives is impossible if the product has to be removed immediately from the assembly site. If non-destructive testing or visual inspection is required after joining, this may be specied too in the questionnaire.

ent metals, or one metal and one polymer.. With the database of materials, a matrix of compatibilities between materials is stored mainly with respect to the problem of galvanic corrosion.. Similarly a matrix of compatibilities between the materials or the material couples and the possible processes is implemented in the software. Some processes may be specic to polymerrpolymer joints for instance, hot plate welding for thermoplastics. or to hybrid joining for instance, anaerobic adhesives require that at least one of the materials to be joined is a metal.. Examples of shapes and modes of loading are given in Figs. 2 4. For each shape, and each loading situation, a qualitative estimate of its efciency for a joining method is provided. For instance, the plane on plane geometry is excellent for resistance welding, provided the combined thickness is between 0.12 and 20 mm. If loaded in tension, adhesives are poor but they are excellent in shear. These criteria on the shape and the loading conditions will be selective to eliminate some processes. Note that these decisions are more subtle than a simple compatibility check between each process and a single requirement it is the combination of process and joint geometry that determines the viable thickness range, or mode of loading w6,10x. The core information of the software is the database on processes. The list of available processes in the database, classied as adhesives, welding and mechanical fastening, is given in Table 1. For each process, numerical and qualitative information are stored, together with a technical sheet which provides as text. information on the production condiTable 1 List of joining processes Adhesives Welding Mechanical fastners Clinching Clips Banding Bolts Rivets Crimping Staples Screws Nails

3. Databases The answers to the questionnaire describe the design requirements for the joining method. Completing the questionnaire makes reference to a database of materials and a database of shapes. To process the information provided by the user to lter out the technically impossible processes, and in order to rank the remainder, a detailed database on processes is needed. The list of materials stored in the database covers both metals and polymers, at the generic level proposed in the CES software w8x. The only material property stored is the maximum utilisation temperature, since we are not dimensioning the joint. The software is concerned with joints between similar materials, and dissimilar materials for instance two differ-

Nitrile rubber Neoprene rubber Polysulde Silicone Polyurethane Thermoplastic Adhesive tapes Cyanoacrylate Anaerobic Epoxy Phenolic Thermostable Inorganic Acrylic

Coated electrode MIGrMAG TIG Plasma-arc Powder submerged-arc Electron beam Laser Diffusion Friction plasticsa rmetals. Gas welding Ultrasonic plasticsa rmetals. Soldering Resistance Hot gas weldinga Radio frequencya Hot plate welding Electromagnetic welding Solvent welding Straticationa

Only for thermoplastics

C. LeBacq et al. r Materials and Design 23 (2002) 405 416 Table 2 A typical set of information stored in the database for joining processes for TIG welding. Criteria Marks for TIG. The best Hygiene criteria during service Electrical conductivity Thermal conductivity Vibration damping Disassembly of the joint Waterproof with a seal. Intrinsically waterproof Size of the components less than 20 cm up to more 100cm. Degree of precision required for the joint between 0.1 and 1 mm. Porosity of materials Surface state before assembly paints, zinc, oil . . . . Surface treatment to be retained after assembly Minimum use temperature Maximum use temperature Service environments :air; acid; alkali; water; sea water; humid and marine environment; oil; UV; solvents Automation of the process Non-destructive control Assembly place: workshop; controlled atmosphere; outside Assembly place: water Excellent Excellent Excellent Impossible Impossible Indifferent Excellent Excellent Excellent The worst Comments

409

Excellent Excellent Impossible Impossible Indifferent Excellent

TIG welding leads to a joint which conducts electricity TIG welding leads to a joint which conducts heat It is impossible to disassemble reversibly a joint obtained with TIG welding No seals used in welded joints

Impossible

Average Impossible Impossible Materials Materials Excellent

Average

If a precision of 1 mm is required TIG is excellent. If a precision under 0.1 mm is required TIG is poor TIG is possible with porous materials but not very good TIG welding is very sensitive to any surface contamination It is impossible to retain a surface treatment after assembly for TIG welding. The minimum use temperature depends on materials irrespective of the joining method The maximum use temperature depends on materials irrespective of the joining method TIG welding has a resistance to the environment mainly controlled by the one of the material TIG welding can be either manual or automatic TIG welding allows non-destructive control

Excellent Excellent Excellent Impossible

Impossible Impossible

Minimum time before 0.0 sec manipulation Roughness of materials between 1 and 600 m Excellent Hygiene criteria and safety during processing Production rate manual. Production rate automation. Maximum processing temperature Accessibility from one side Position of processing Indifferent

0.0 sec

It is impossible to perform TIG welding under water No need to pospone product motion after welding No need to polish before assembling by TIG welding This criterion is not critical for TIG welding, provided care is taken of some toxic smokes and radiation of the arc

Excellent Good None Excellent Indifferent

Good Poor Impossible Indifferent TIG welding allows the components to be welded either from one side or from both sides

tions and the usual application eld. For instance, for an epoxy adhesive, the range of maximum service temperature is given as 260 280 C, while the electrical conductivity is specied as between good and poor depending on the resin specications.. Table 2 gives a typical datasheet from the process database, for TIG welding. The information stored in the data sheet parallel the questions posed in the questionnaire.

4. Selection procedure When the questionnaire is lled, a transcription of the technical requirements is available in exactly the same format as the information stored in the database for each triplet processrmaterialsrjoint geometry.. The agreement with each answer of the questionnaire is then represented by a mark between 0 and 10. The

410

C. LeBacq et al. r Materials and Design 23 (2002) 405 416

rst step is to eliminate the processes which have at least one zero in their evaluation. This conservative procedure ensures that no technically impossible process is retained. Then, among the different processes, which are technically admissible, a global technical evaluation is needed in order to rank them. In order to do so we have to perform a multicriteria evaluation. These kinds of problems always have two facets w7,9x: rstly, one has to decide on the average to be taken of the various performance marks, which are assumed rst to be of equal importance; secondly, one needs to consider the possibility of assigning different importance to selected criteria. In these evaluations, some judgement is involved, i.e. two experts will react differently and will come to different classications. In a similar problem for material selection a fuzzy logic algorithm was implemented, allowing the choice between different averaging procedures to depend on the permissivity of the user, which was evaluated from the reaction to proposed solutions w16x. We have chosen here to limit our range of possible averaging methods to the arithmetic mean which is a permissive average., and the geometric mean which is a restrictive average.. The user can evaluate the effect of permissivity on the ranking between the possible processes, by trying each in turn. The question of allowing criteria to have different importance is dealt with as follows. The mark given for a performance criterion the objective mark since it derives from the objective data stored in the databases. is modied to give an effective mark using a factor to mark up or knock down the value, as required. Various possible curves are available to scale between the two marks: a power law, or a linear law with a threshold, or a hyperbolic tangent law. Each of these laws allows the user to stress, for instance, that an objective mark which is poor should be treated as effectively worse still when that criterion is given the weighting very important. Similarly, a criterion of good can be biased to count more strongly in a positive sense. The parameters of these laws depend on the qualitative importance given to the criterion. For instance if, for a given criterion, a process is given an objective mark 5r10, it might be reduced to a mark 4r10 if the criterion is ranked as unimportant, and 0.5r10 if it is ranked as very important. If all the criteria are of equal importance of course the relation between the effective mark and the objective mark is identical for all the criteria the multiplying factor is unity. This is the default condition in the software. The averaging procedure either permissive or restrictive. is performed on the effective marks. This weighting and averaging method has the drawbacks of all subjective judgement methods, since it depends on the expert using the software. It should not therefore, be given an absolute signicance that it does not pre-

tend to have. It may be considered as a tool to evaluate the stability of the technical ranking with respect to the importance given to the various criteria. The selection procedure rst gives the excluded processes and the reason for their exclusion Fig. 5 gives an example of this software results window.. Then it provides a list of retained processes, together with their ranking depending on the importance applied to each criteria .. Note that presenting the results in this sequence is deliberate, to emphasise that the rst result of the selection is to eliminate impossible methods, and then to rank those which may be possible on the basis of the design information provided by the user. Fig. 6 shows a screen-shot of a sample of the retained process window. The degree of fullment on each criteria is indicated for each technically admissible process, along with the global technical ranking. The individual marks may be presented numerically from 0 10. or as text very poor excellent .. Every mark in the results table may be investigated, to establish the reason for this mark being given.

5. Further information The previous paragraphs describe the technical evaluation of joining processes with a set of requirements described by the answers to a questionnaire. It gives the user a list of processes, which are technically excluded, and a ranked list of processes, which are technically admissible. The real selection procedure doesnt stop there. Of course, an economical evaluation of the different processes could be used to rank them, either following the time evaluation proposed in the DFMA methodology w1,2x, or by more transparent but more complex technical cost evaluations w17 19x. However, a preliminary objective cost evaluation may be made after a much simpler investigation, by considering the local conditions in the company which is going to perform the joining. For instance, for a given process which is technically compatible with a set of requirements, the nal decision may differ widely if the equipment is available or not, or if the relevant skills are present or not. Even if the skills or the equipment exist in the company they may be overcommitted and unavailable. Alternatively, an investment decision in equipment or in personnel may be considered. A second questionnaire addressing these issues is proposed to the user, who gives for each viable process a series of answers concerning the existence, availability or possibility to invest both for the equipment and for the human resources. From this information, a local evaluation is given for each process, represented by a mark between 0 and 10, 10 being the most favourable case where the equipment and the competence are present and easily available, and 0 the worst case where none

C. LeBacq et al. r Materials and Design 23 (2002) 405 416

411

Fig. 5. Example of excluded process list from the software, showing the reasons for process elimination.

of them are present and any investment is out of the question. The tree structure used to compute these marks is given in Fig. 7. The association of a local evaluation mark with a given process gives a picture of the ability of the company under consideration for each possible technique. It may be assumed to be valid without modication for all the products to be assembled by this company. Beside this local evaluation, another criterion which cannot be discounted, in spite of its conservative consequences, is the fact that in given areas of industry some joining processes may be well accepted and familiar, while others are rarely considered. A database in the software stores a list of industrial elds, and an evaluation of the perception in this eld of a given joining process. Each process is assigned one of the following: frequently used, sometimes used, or never used. For instance, the aeronautic industry frequently uses TIG welding but rarely uses resistance welding, while the exact opposite occurs in the automotive industry. This provides a second evaluation for the processes, which are technically admissible: its reputation in the eld of application considered.

6. Case study A case study to illustrate the application of the method and the software is the domestic radiator, illustrated in Fig. 8. The aim is to assemble three corrugated plates in mild steel. The requirements for this assembly problem are shown in Fig. 9. The simplied joint geometry is plane on plane, the size and thickness of the two plates are given, and the materials are specied in the rst instance without surface treatment, but in a second case we will consider painted plates.. Two important functions concern the waterproofness and the ability to conduct heat. The joining has to be permanent, and the service conditions of the product are between 0 and 100 C in a humid environment.

7. Technical evaluation The processes, which are technically excluded, are listed in Table 3, showing the reasons why they were excluded. For instance clinching and nailing are ex-

412

C. LeBacq et al. r Materials and Design 23 (2002) 405 416

Fig. 6. Example of technically admissible process list from the software, showing the evaluation against each of the criteria answered in the questionnaire.

cluded for reasons of waterproofness, whereas bolts and rivets are not excluded for the same reason because it is possible to add a seal. Clipping is excluded

because of the superimposed shapes. Cyanoacrylate adhesives are excluded because of the prescribed working temperature of the radiator, but epoxy adhesives

Fig. 7. Tree structure for the local evaluation of a process.

C. LeBacq et al. r Materials and Design 23 (2002) 405 416

413

Fig. 8. Schematic of the joints in a domestic radiator.

can full this criterion. Various welding processes are excluded because of the materials if the welding methods are specic to polymers., or because of the thickness of plates to be assembled. In fact the thermal conductivity requirement excludes most of the adhesives, and waterproofness and shape of the joining exclude most of the mechanical fasteners, while the dimension restrictions limit the range of possible welding processes. The processes, which are technically admissible, are listed in Table 4. The global evaluation has been performed with the geometric average which is a restrictive averaging., and assuming all the criteria are of equal importance. Although the evaluation has been performed with the complete set of requirements, only the ones which may discriminate between the various technically admissible processes are listed in Table 4. TIG and resistance welding appear to be the best solutions, and indeed they are the ones used in industry. Laser welding is also possible. Some adhesives can

also be considered, but they are the ones for which thermal conductivity is not too bad. Mechanical fasteners are not completely excluded because of the possibility of using a seal, but they are not very good on average due to poor waterproofness. Soldering ranks as very poor because of the thickness of the components, and gas welding because of the difculty to automate the process. It is of particular value in selection problems to examine the details of the comparison with each item of the set of requirements. This information can be used by the designer to indicate and test possible design changes. Selection is by nature an iterative procedure it is straightforward in the software to modify a requirement and to see what effect this has. Now consider the effect of changing the relative importance of the various criteria. The software is constructed such that this information doesnt inuence the processes which are excluded, but it might change the overall ranking in the list of retained processes. Table 5 give an example of such inuence. The requirements are the same, but in case 1 all the criteria are equivalent, while in case 2 thermal conductivity and waterproofness two key issues for a radiator. and plate thickness are given a special importance. For illustration the relation used between effective mark and objective mark is a hyperbolic tangent function. The choice of another function would give similar results as far as classication is concerned. As a result, mechanical fastening methods are now at the bottom of the list, whereas gas welding where the main problems were automation and the size of the components. improves but remains below the other welding methods. Laser welding is excellent, as are TIG or resistance welding, and the selection between these methods will come from economical considerations, which favour resistance welding. If the materials of the plates were aluminium e.g. if mass minimisation was of concern.,

Fig. 9. Case study: the design requirements for the joints in a domestic radiator.

414

C. LeBacq et al. r Materials and Design 23 (2002) 405 416

Table 3 Excluded processes with reasons for elimination, for the joints in a domestic radiator Thermal conduction Polyurethane Adhesive tapes Cyanoacrylate Anaerobic Acrylic Coated electrode MIGrMAG Plasma-arc Powder submerged-arc Electron beam Diffusion Friction metal. Hot gas welding Stratication Solvents Radio frequency Ultrasonic plast.. Friction plast.. Hot plate Electromagnetic Clinching Clips Banding Crimping Staples Nails X X X X X Maximum temperature X X X X X X X X X X X X Environment Size Dimensioning Automation Place of assembly Shape Materials compatibility Produc tion rate Water proof Disassembly

X X X X X X X X X X X

X X X X X X X X X X X X X

X X X X X X X X

Table 4 Technically admissible processes and technical marks, for the joints in a domestic radiator TIG Excellent 1r11 Excellent Excellent Excellent Excellent Excellent Excellent Excellent Excellent Excellent Rivets Poor 6r11 Excellent Excellent Excellent Poor Excellent Excellent Excellent Excellent Excellent Resistance Excellent 1r11 Excellent Excellent Excellent Excellent Excellent Excellent Excellent Excellent Excellent Screws Poor 6r11 Excellent Excellent Excellent Poor Excellent Excellent Excellent Excellent Excellent Laser Good 3r11 Excellent Excellent Excellent Excellent Excellent Excellent Excellent Excellent Good Ultrasonic Poor 6r11 Average Excellent Excellent Excellent Average Excellent Excellent Excellent Excellent Epoxy Average 4r11 Excellent Good Excellent Excellent Good Excellent Excellent Excellent Excellent Soldering Very poor 10r11 Very poor Excellent Good Excellent Excellent Excellent Excellent Excellent Excellent Silicones Poor 5r11 Excellent Good Good Excellent Average Excellent Excellent Excellent Excellent Gas welding Very poor 11r11 Excellent Excellent Excellent Excellent Average Average Excellent Poor Excellent Bolts Poor 6r11 Excellent Excellent Excellent Poor Excellent Excellent Excellent Excellent Excellent

Global mark Ranking Dimensioning Thermal conduction Disassembly Waterproofness Size of components Materials comp. Environment Automation Assembly place

Global mark Ranking Dimensioning Thermal conduction Disassembly Waterproofness Size of components Materials comp. Environment Automation Assembly place

C. LeBacq et al. r Materials and Design 23 (2002) 405 416 Table 5 Illustration of the effect of different emphasis on selected criteria on the ranking of the technically admissible processes TIG Global mark Ranking case 1. Global mark Ranking case 2. Excellent 1r11 Excellent 1r11 Rivets Global mark Ranking case 1. Global mark Ranking case 2. Poor 6r11 Very poor 7r11 Resistance Excellent 1r11 Excellent 1r11 Screws Poor 6r11 Very poor 7r11 Laser Good 3r11 Excellent 1r11 Ultrasonic Very poor 6r11 Very poor 7r11 Epoxy Average 4r11 Poor 4r11 Soldering Very poor 10r11 Very poor 7r11 Silicones Poor 5r11 Poor 5r11 Gas welding Very poor 11r11 Very poor 6r11 Bolts Poor 6r11 Very poor 7r11

415

Case 2, with plate thickness, thermal conductivity and waterproofness given extra emphasis, compared with the ranking in case 1 from Table 4., when all criteria have equal emphasis..

TIG would become favoured. Finally, if the set of requirements is modied to state that the plates are painted before assembling, and that this surface state has to be conserved after joining, all the welding processes disappear from the list of possible solutions, and the adhesives and mechanical fasteners present in the previous selection remain the only solutions.

8. Local evaluation After the technical evaluation, the local evaluation aims to establish the ability of the manufacturer to use the best processes identied. Consider only TIG, resistance welding, laser welding and epoxy adhesives, which were ranked highest in the technical evaluation, and consider possible proles for two companies, A and B. Company A has: no TIG or laser equipment, and no staff in either process; it has trained staff but no equipment in adhesives; its resistance welding equipment and staff are available; and no investment is possible for TIG and laser, but it is possible for the adhesive joining. Company B has no TIG equipment or staff, but is ready to invest in them; it has no laser equipment or staff, and is not ready to invest; it has equipment and staff in resistance welding but these are overcommitted; and it has the equipment for adhesive joining, but the trained staff have left. For the 4 processes, which are ranked on technical grounds as TIG, resistance welding, laser welding and epoxy adhesive, the local evaluation marks are 0, 10, 0, 8.2. for company A, and 5.9, 6, 0, 8.2. for company B. This information, together with the technical evaluation, will help the nal decision for instance, company A has two clear options resistance and adhesives., while

company B should also give TIG welding further consideration. The last evaluation to be considered is the reputation of a process in a eld. For instance, the current process used for radiators in industry is resistance welding. It has to be stressed that the reputation evaluation can be interpreted in two different ways: either as a conservative indicator it is already used so people tend to trust it. or as an incentive to innovate it is technically feasible but nobody does it, so there may be a market opportunity.. The importance of the technical evaluation with respect to the local evaluation and to the reputation is up to the user to judge, but the subjectivity of this interpretation indicates that it would be meaningless to merge all the evaluations in a single mark.

9. Conclusions In this paper a methodology has been presented for selection of joining method. This is an example of task-based process selection: the design issues are largely specic to the manufacturing task of making joints. This methodology has been implemented into a selection software. The selection procedure relies on the transcription of technical requirements for a given joining problem into answers to a predened questionnaire. The technical evaluation compares the answers to this questionnaire with information concerning each triplet processrmaterialsrjoint geometry. stored in databases. The multicriteria evaluation is performed using a fuzzy logic algorithm. The rst output of this software is a list of joining processes, which have to be excluded for technical reasons, which are explained. For a given set of requirements, the user is then given

416

C. LeBacq et al. r Materials and Design 23 (2002) 405 416

a list of processes, which are technically admissible. Each of these processes may be characterised by three evaluations represented by marks between 0 and 10: a. a technical evaluation which represents the degree of agreement with the requirements expressed by the answers to the questionnaire; b. a local evaluation which represents the current ability of the company to use the process; and c. a reputation evaluation which indicates the perception of this process in the industrial eld of the application. These three evaluations, weighted according to the judgement of the user, provide a panel of information for selecting and ranking the possible processes. The next step is of course a detailed economical evaluation of the processes remaining on the top of the list this usually requires consideration of the joining process as one stage within a whole manufacturing operation. This last step is not an easy one, and has not been addressed here. Rather, we have developed a tool, which successfully restricts the number of processes for which the difcult task of detailed cost evaluation is worth doing. The method is also successful in suggesting modications to the design to facilitate joining, or indicating possible innovative use of unfamiliar processes.

References
w1x Swift KG, Booker JD. Process selection from design to manufacture. London: Arnold, 1997. w2x Allen A, Bielby M, Swift K. Int J Advanced Manufacturing Technol, 1991;7:205. w3x Messler RW. Joining of advanced materials. Stoneham, MA: Butterworth-Heinemann, 1993. w4x Tichem MA design coordination approach to design for X, PhD thesis, Delft University, The Netherlands, 1997. w5x TWI, JoinIt software, http:rrwww.twi.co.ukrJoinItr., TWI the Welding Institute ., Granta Park, Cambridge, UK. w6x Houldcroft PT. Which process? Abington, UK: Abington Publishing, 1990. w7x Ashby MF. Material selection in mechanical design. Oxford, UK: Butterworth-Heinemann, 1999. w8x Granta Design Ltd., Cambridge Engineering Selector CES. software v.3.2, Granta Design Ltd, Rustat House, 62 Clifton Rd, Cambridge, CB1 7EG, UK, 2001. w9x Brechet Y, Bassetti D, Landru D, Salvo L. Prog. Mat Sci. 2001;463 4.:407. w10x Shercliff HR, Lovatt AM. Prog. Mat Sci. 2001;463 4.:429. w11x Lovatt AM, Shercliff HR. Materials and Design 1998;19:205. w12x Lovatt AM, Shercliff HR. Materials and Design 1998;19:217. w13x Pechambert P, Bassetti D, Brechet Y, Salvo L, Proc. 7th Int. Conf. Composite Materials ICCM7., London, Institute of Materials, 1996:283. w14x Lovatt AM, Bassetti D, Shercliff HR, Brechet Y. Int. J Cast Metals Res. 1999;124.:211. w15x Dubois D, Prade H. Fuzzy sets and systems, theory and applications Academic Press., 1980. w16x Bassetti D., Aides informatisees a la selection des materiaux, PhD thesis, Institut National Polytechnique de Grenoble, France, 1998. w17x Allen AJ, Swift KG. J. Engineering Manufacture 1990;204:143. w18x Clark JP, Roth R, Field FR. Materials Park, Ohio, USA: ASM International, 1997. w19x Maine E.M.A., Ashby M.F. submitted to Materials and Design 2001.

Acknowledgements The authors wish to acknowledge many helpful discussions with Prof. Michael Ashby, Dr Andrew Lovatt and Dr Didier Landru, and the nancial support of CETIM St. Etienne, France., and the Korber Founda tion.

You might also like