Supporting design decision-making when applying materials in combination

Supporting design decision-making when applying materials in combination

Materials & Design Materials and Design 28 (2007) 1288–1297 www.elsevier.com/locate/matdes Supporting design decision-making when applying materials ...

166KB Sizes 0 Downloads 55 Views

Materials & Design Materials and Design 28 (2007) 1288–1297 www.elsevier.com/locate/matdes

Supporting design decision-making when applying materials in combination K.L. Edwards a

a,*

, Y.-M. Deng

b

School of Computing and Technology, University of Derby, Kedleston Road, Derby DE22 1GB, UK b Faculty of Engineering, Ningbo University, Ningbo, Zhejiang 315211, PR China Received 27 October 2005; accepted 13 December 2005 Available online 17 February 2006

Abstract The importance of materials selection in engineering design is well recognised. There are already plenty of formalised methods to support selection of individual materials, where the requirements on the materials are, respectively, known. Practical design problems, however, often involve materials in combination – the multiple materials jointly contribute to the performance of the system to produce optimal design metrics; and the selection of materials is often coupled with the determination or design of structural components and their configuration. To address these problems, this paper recommends a multiple-mapping strategy and an inter-level behavioural modelling strategy. The former can simultaneously consider both structural and materials solutions in supporting design decision-making at the early design stage; while the latter can provide a platform for the designer to work out the couplings among the properties of the materials and their corresponding components, thus supporting design decision-making in selecting materials when they are applied in combination, which is generally at the downstream design stage. The paper also discusses the implications and difficulties in providing optimal solutions to the addressed problems; and based on this, to propose some prospects of future work to further improve these design strategies.  2006 Elsevier Ltd. All rights reserved. Keywords: Design decision-making; Materials in combination; Materials selection

1. Introduction The importance of materials selection in engineering design has been well recognized. The design decision-making regarding selecting appropriate materials is dictated by the specific requirements of an application, often the requirements on materials properties. Materials properties can be categorized in a number of ways, but it is conventional to classify them into groups with similar properties, processing route and end use. This generally leads to the following broad classification (but does not imply any scale of performance or ranking): * Corresponding author. Tel.: +44 1332 591729; fax: +44 1332 597741/ 622739. E-mail addresses: [email protected] (K.L. Edwards), [email protected] (Y.-M. Deng).

0261-3069/$ - see front matter  2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.matdes.2005.12.009

     

Metals Ceramics Glasses Polymers Elastomers Composites

The latter class of materials is special in that it provides combinations of properties from the other classifications. Although each category of materials is associated with characteristic properties (mechanical, physical, etc.) there is often overlap and contradiction between categories, resulting in compromise as not all properties may be obtainable. This creates the need for using combinations of materials and the growth in the use of composites. As always, the choice of material or materials to achieve the desired properties ends up being something that satisfies

K.L. Edwards, Y.-M. Deng / Materials and Design 28 (2007) 1288–1297

the most important design requirements. In a normal commercial situation, it is the economics that often has the most significant impact on the final choice of materials. However, technological developments are continually opening up new possibilities all the time. Materials selection plays a significant part in the whole design process for manufactured products, from concept to realisation. The most significant difference is the level of data precision and range of different materials considered as the design progresses. Typically, in the early stages of the design process, the number of materials under consideration is quite large and data on them approximate. As the design progresses the number of materials reduces and the data becomes more refined until eventually a candidate material is selected and used. In this paper, the case of several materials remaining for utilisation is the case and a more normal situation. Utilising several materials simultaneously implies selecting materials for either a single multiple function carrying component and/or several functionally related adjacent components. The selection process is non-trivial but more accurately reflects normal product design practice. Apart from materials properties, other requirements, such as the requirements on materials processing during component manufacturing and those on in-service conditions relating to the materials of components, should also be taken into account in materials selection. Formal process selection is often overlooked in the design process because the attributes that define processes are less specific than in materials selection. Often there are many ways of creating components (shaping, joining, finishing, etc.) but choosing the optimum route is difficult. Normal practice is to adopt ‘tried and trusted’ methods that have worked successfully in the past but this may eliminate new as yet unidentified possibilities. Note this does not imply leading edge technologies but processes new to the designer or design organisation. When considering the processing requirements of several components simultaneously rather than individually, alternatives may arise that would not have necessarily been used for any individual component. The decision-making process rapidly becomes complex, which is exacerbated further when materials selection considerations are introduced. Further, service conditions during product use are also important and need to be factored into the decision-making. Traditional intuitive based approaches can and do break down despite experience underpinning, leading to sub-optimal design solutions. Fortunately, modern computer-based design tools, of which there are now many, support aspects of the design decision-making process, but gaps still exist. In fact, all of the above requirements are used to ensure that the functional requirements of a design can be fulfilled, and a number of design constraints can be satisfied, such as the constraints on available design assets including the available materials and components, the constraints on the size (dimensions), mass and other properties of the components, as well as constraints on the system outputs.

1289

These constraints are also called the required design metrics. The output of design decision-making includes the determination of the physical structure of a design, such as the selection or design of structural components and their configuration, which is often referred to as structural solution; and the selection of materials of these components, i.e. materials selection. Materials selection is often coupled with the determination of structural solution. Hence, they should generally be considered together, in order to satisfy functional requirements and other design constraints or design metrics effectively and efficiently. Because of the existence of multiple design requirements and constraints, there might be various conflicts in one form or another. As such, conflict management must be sought out under various circumstances. Identifying, defining and planning for conflict is a normal aspect of designing, but rarely practised effectively. The role of experience is critical in conflict resolution; knowing what has worked in the past, but is not so helpful to say a novice designer or where new materials/processes are being considered. There is a risk therefore of perpetuating ill-conceived ideas by relying on limited knowledge and an inability to process significant multiple attribute information simultaneously. Even if both these aspects are better facilitated and there is less reliance on intuition, the need to make decisions, although better informed, does not diminish with managing conflict actually becoming a more regular occurrence. This justifies the need for enhanced integration of materials selection within a design process framework, with the appropriate level of support invoked depending on the stage reached. There are already a large number of formalised methods to assist the determination of structural solution and materials selection. Regarding the design models or methodologies for the former design task, some of the well-recognised (and most influential) ones include Pahl and Beitz [1] and Hubker and Eder [2] for systematic design theories, Pugh [3] for integrated design methodology, Suh [4] for axiomatic design theory, and Altshuller [5] for inventive problem solving or TRIZ method, among many others. Regarding those that support the latter design task, Ashby et al. [6] have provided a comprehensive review of the strategies and methods for materials selection, among which their own work is the best known [7,8]. Based on this review, Deng and Edwards [9] have elaborated on some of the representative work categorised into three different strategies, including free searching; checklist or questionnaire; inductive reasoning and analogue procedure [10–13]. These strategies can be applied manually, or in a computer-aided form (be computerised), or both. Dieter [14] presents a unified approach to design methodologies and materials selection. There are ‘expert systems’ available but there is still a significant challenge in selecting materials to match materials characteristics to meet design requirements. There is also a place for materials selection handbooks for reference purposes of which significant examples are by Waterman and Ashby [15] and Dieter [16], although computer-based materials selection packages

1290

K.L. Edwards, Y.-M. Deng / Materials and Design 28 (2007) 1288–1297

now dominate and numerous examples are available of which the Cambridge Engineering Selector (CES) by Granta Design Limited, Cambridge, UK is a prominent example used mostly by educational establishments. Rajan [17] investigates materials structure-property relationships from a materials informatics perspective, which although based on the science of materials is a complementary systematic methodology to the above. Despite the great deal of existing work, there are at least two areas where the support for design decision-making is relatively weak. One area of weakness is related to the simultaneous consideration of structural solution and materials selection, which is generally needed at the early design stage. There are already plenty of theories, models, methodologies, and so forth, in supporting the two ends of the design decision-making task; yet very little work has been oriented to something in-between. Edwards [18,19] has pointed out the importance of integrating materials selection with component design. Raj and his research group [20,21] have presented a systems-based methodology to manage the relationships between design variables of both materials and components. Lu and Deng [22] have proposed a systems modelling methodology to support the integration of materials design/selection with engineering system design. These works, however, are oriented to the downstream design activities, where conceptual solutions involving components and materials are already determined. The other weak area in design decision-making relates to selecting materials in combination. The overview above has shown that the existing methodologies in materials selection concern design situations where only individual materials are to be selected and the property requirements on these materials are known. However, engineering design often relates to an engineered system, where varying number of materials may be applied, thus they need to be determined. Since an overall system output is determined by the joint contribution of the components and their respective materials, rather than by any one single material or component; it is thus necessary of formalised design considerations for selecting optimal (or relatively optimal) combinations of materials. To address these problems, this paper focuses on the strategies for materials selection, both for the simultaneous consideration of structural solution and materials selection, and for selecting materials when they are applied in combination. Design decision-making involving determination or selection of structural components will also be discussed where necessary. Section 2 will specifically elaborate on some design situations involving materials in combination and explain the problems to be addressed in this paper in more detail. Two design strategies tackling the problems, respectively, will be elaborated in Section 3. Section 4 discusses the implications and difficulties in providing optimal solutions to the addressed problems, and proposes some prospects of future work to further improve the design strategies, before concluding remarks in Section 5.

2. Design decision-making problems involving materials selection 2.1. Design situations involving materials in combination There are plenty of examples when materials are applied in combination in engineering design. Below are some of the typical situations: (1) Single components comprising several different materials, which are in discrete and integrated forms, such as a heterogeneous flywheel [23]. (2) Combinations of components, each with similar or dissimilar materials. This is the most common situation in engineering design. One example is the bearing system that includes the housing, the bearing itself (which may contain several different materials if say a roller bearing) and the shaft. All three elements are generally of different materials. (3) Situations involving (1) and (2) together, e.g. the mechanical system that include the above mentioned heterogeneous flywheel. Note that selecting materials in combination discussed in this paper does not include composites or alloys, such as the composite sandwich [24]. For composites and alloys, there are specific considerations in selecting or designing optimal micro or nano materials structure and composition. In addition, Walker and Smith [24] only consider a few candidates of materials combination. Their work in effect focuses on a sequential solution procedure for optimally designing composite sandwich cylindrical shells, which is primarily an optimisation strategy in determining structural parameters such as the skin and core thicknesses, the skin fibre angles. Hence their method for selecting best materials combination is basically ad hoc. 2.2. Design decision-making problems As has been mentioned before, apart from selecting materials in combination, the designers often need to simultaneously seek structural solutions, namely, to determine or design the required components and their configuration. This is especially true for the early stage of engineering design. Therefore, the whole thing can get quite complicated as the materials selection merges more with the design of the components, and the system of these components. The existing design methodologies or design models often take a sequential approach when handling the design task of seeking structural solutions and selecting materials, including materials in combination. For example, regarding the sequence of determining structural components and selecting materials, it is common practice to determine the components first, followed by the determination of the materials of these components. This is understandable because materials are generally regarded as one of the

K.L. Edwards, Y.-M. Deng / Materials and Design 28 (2007) 1288–1297

attributes of components. However, there are several situations where materials may play a same role as that of components in delivering a required function. The authors’ previous publication has elaborated on these situations [9]. Some common ones include: (1) Where specific material properties should be exploited to fulfil a functional requirement; (2) Where specific materials phenomena should be exploited to fulfil a functional requirement; (3) Where materials solution is better than a structural solution. As such, a sequential ‘component to material’ approach may not be feasible for all design situations. Regarding selecting materials in combination, the current practise is primarily based on designers’ experience and expertise. Some common practices include: (1) The designers may first rank the importance of the materials to be selected; or rank the importance of the components bearing the materials. The materials are then selected in a sequence following the order of this rank. (2) The designers may first select the materials whose properties are determinant to those of the other materials. The other materials will be selected in a sequence in accordance with the degree of the dependency of their properties on those of the rest – the greater the dependency, the later will it be selected. (3) Most often than not, the designers may simply compare some candidate combinations, and select one combination out of these candidates. These candidates are chosen out of past similar designs, or out of designers’ own experience and expertise, such as the work by Walker and Smith [24]. Since it is the combination of the materials that jointly contribute to the performance (design metrics) of the system, such sequential approaches or experience-based approaches for selecting materials in combination are not able to guarantee the designed system to achieve an optimal performance. Hence, more suitable strategies should be developed to assist these design decision-making problems. 3. Strategies to assist design decision-making involving materials selection Engineering design, especially design and development for product lifecycle, is a very complex process, involving various stages of design activities, most of which are iterative; various aspects of design information, most of which are coupled with each other; also various kinds of design information, such as design requirements, constraints, conflicts, functions, behaviours, structural components, materials, interactions, enabling principles, knowledge in

1291

various forms, so on and so forth. As a result, it is not possible for any single theory, or model, or methodology, etc. to address all these issues. Existing design theories, models, methodologies, etc. all focused on specific aspects of design, and/or specific stages of design, and/or specific areas of designs, etc. to varying degree of extend. For example, Ljunberg and Edwards [25] have proposed an integrated product materials selection (IPMS) model, specifically for materials selection related product lifecycle design and development. As has been mentioned before, this paper focuses on supporting design decision-making in simultaneous consideration of structural solution and materials selection, as well as in supporting the optimal selection of materials used in combination. The design strategies discussed below are specifically for addressing these problems. 3.1. Multiple-mapping strategy A number of mapping methods were proposed by Deng and Lu [26] in identifying both structural components and materials for satisfying functional requirements, including direct mapping, behaviour-assisted mapping and ‘behavioural process’-assisted mapping. This paper adopts such a multiple-mapping strategy, because this strategy enables structural solution and materials selection to be considered simultaneously during the early stage of engineering design. Briefly, direct mapping is intended for domain-specific or application-specific functional requirements, which may be achieved by commonly used structural components or materials from previous design experience. Sometimes, a function may not just be achieved by a structure or material, but rather by the behaviour exhibited by the structure or material, when it is put under its working conditions. For this kind of functional requirement, behaviour-assisted mapping is necessary, that is to say, designers should first map the required function to behaviour by employing the relevant physical phenomena or effects. A phenomenon is a generalised behaviour or behavioural process, which is used to characterise the working principles necessary for conceptual design. The desired structural components and/or materials are then retrieved from the mapped behaviours, to be exact, from the behaviour actor. ‘Behavioural process’-assisted mapping refers to those design problems where the required functions should be achieved by a ‘behavioural process’ consisting of a number of individual behaviours, rather than a single behaviour. For this kind of design problems, the design must undergo a behavioural process, through which can the required function be delivered eventually. The desired structural components and/or materials are then retrieved from the developed behavioural processes, which shall definitely be more complex than those from simple behaviour-assisted mapping. Table 1 lists the specifics of this multiple-mapping strategy. This strategy for achieving some of the functions can be computerised by developing knowledge libraries to store

1292

K.L. Edwards, Y.-M. Deng / Materials and Design 28 (2007) 1288–1297

Table 1 Specifics of multiple-mapping strategy Type of mapping

Map from

Map via

Map to

Need to retrieve components and materials?

Direct Behaviour-assisted

Functional requirement Functional requirement

Components or materials Behaviour

No Yes

‘‘Behavioural process’’-assisted

Functional requirement

None Phenomena or effects for single behaviour Phenomena or effects for multiple behaviours

Behavioural process

Yes

commonly used or domain specific components/materials, and physical/material phenomena. The functions they achieve are used as retrieval index, so as to assist the direct mapping and behaviour-assisted mapping. For ‘behavioural process’-assisted mapping, the mapped behaviour from a given functional requirement is used as an individual behaviour of the behavioural process being developed. The designers should continue to look for other suitable phenomena, so as to develop other individual behaviours for the entire behavioural process. A backward synthetic search method can be exploited in fulfilling this task [27,28], which is a recursive process in develop the preceding individual behaviours from a succeeding behaviour. The preceding behaviour should have an output that can be used as the input to the succeeding behaviour; that is to say, two connected individual behaviours in a behavioural process are in effect linked by the input–output interactions occurring between them. The process stops when the required inputs to the very first or the first set of behaviours can be provided by the system inputs. As can be seen, by using this multiple-mapping strategy, the designers can equally consider structural components and materials in the direct mapping. They can also equally consider the behaviours delivered by structural components and those delivered by materials in the behaviourassisted or ‘behavioural-process’ assisted mapping, by equally applying the physical phenomena exhibited by structural components and the materials phenomena exhibited by materials. 3.2. Inter-level behavioural modelling strategy The above strategy focuses on identifying components and/or materials to fulfil functional requirements, which is generally necessary at the early stage of engineering design. However, the identified materials may not be very specific. Often, they are only in very rough form, being restricted to some sorts of materials, or materials with some functional properties. In most cases, the materials were not determined at all at conceptual design stage, or after the structural components of the design have been fully determined. All these situations make it necessary for materials selection to be done again at the downstream design stage. The most demanding characteristic of the materials selection problems at the downstream design stage, when the materials are applied in combination, is that there are

considerable couplings among the design requirements on these materials, or among the properties of the materials, as well as on their respective components. Deng and Edwards [9] have summarized two categories of materials selection problems: (1) Materials selection based on the required material properties. (2) Materials selection based on the required design metrics, where the requirements on material properties are coupled with those on the physical structure and the relevant structural properties. The existing materials selection strategies generally concern selecting appropriate materials with given or specified materials properties, that is, they belong to the first category of problems. Hence, the crux of materials selection problems involving materials in combination, which belong to the second category of problems, is to determine the respective properties or attributes of the materials and components, whereby these properties should jointly contribute to the optimal design metrics. To assist design decision-making in tackling this problem, we propose an inter-level behavioural modelling strategy. Behavioural modelling refers to modelling the behavioural process of a system, including determining all the individual behaviours exhibited by the structural components and materials, as well as how these individual behaviours are related with each other. The behavioural modelling is inter-level, because the individual behaviours may be delivered by the structural components (at the component level), or by the materials, including the micro and/ or nano structure of the materials (at the material level). Hence, the generated behavioural process may comprise individual behaviours from both component level and material level. Obviously the behavioural process model is multidisciplinary, because an engineered system may involve behaviours from different domains. In fact, the same components or materials may deliver multiple behaviours relating to different disciplines. For example, a bearing may undergo dimensional distortion because of the load it carries. It may at the same time generate frictional heat. Hence it shall exhibit at least two behaviours belonging to two different physical domains or disciplines. Inter-level behavioural modelling strategy is in effect a generalization of the system modelling methodology

K.L. Edwards, Y.-M. Deng / Materials and Design 28 (2007) 1288–1297

proposed by Lu and Deng [22]. The system modelling methodology was intended for the integration of engineering systems design and materials design (including materials selection), where the behaviours of the system and those of the relevant materials are modelled by a network of inter-connected elemental behaviours. In this paper, the authors extend this behaviour model into a behavioural process model, where the behavioural process consists of behaviours causally connected – a preceding behaviour causes its succeeding behaviour to occur, thus is the reason for the succeeding behaviour. This is to ensure that the behaviours exhibited by materials in combination and the behaviours by those relevant structural elements, once there is coupling between them, can be represented in a single behavioural process. Fig. 1 shows an example of such an inter-level behavioural process model, where there are four individual behaviours exhibited by structural components, B1, B2, B3, B6, i.e. at the component level; another two individual behaviours B4 and B5 are exhibited by the materials, i.e. at the material level. The inter-level coupling is obvious – the material behaviours B4 and B5 require inputs from structural behaviour B1; and the structural behaviour B6 requires inputs from the materials behaviours B4 and B5. In such an inter-level behavioural process model, the behaviour is represented by the various interactions between the behaviour actor and its environment. The

Component or material A

Interface

1293

Component level B3

B2 B1

B6

System input

System output

B4

Material level B5

Fig. 1. An exemplar inter-level behavioural process.

behaviour actor refers to the structural components or the materials that perform the behaviour. The behavioural environment can be the components or materials of the design, or it can be the design environment itself. The interactions from the environment to the behaviour actor are referred to as input actions, while the interactions from the behaviour actor to the environment are referred to as the output actions. The input actions and output actions may be characterised, respectively, by a number of attributes, which are just the commonly talked ‘‘inputs’’ and ‘‘outputs’’. For example, in the piezoelectric behaviour of piezoceramic material, the input action and output action are:

Component or material B

Material properties and structural requirements for ‘A’, ‘B’ and interface

Material properties and structural requirements specific to ‘A’ only

Material properties and structural requirements common to ‘A’ and ‘B’

Material properties and structural requirements specific to ‘B’ only

Fig. 2. Materials properties and structural requirements for adjoining components.

1294

K.L. Edwards, Y.-M. Deng / Materials and Design 28 (2007) 1288–1297

 IA1 = to apply a mechanical stress (input action)  OA1 = to produce an electric field (output action) The attributes of the input action ‘‘IA1’’ might include:  IV1 = force of IA1 (the applied force from the input action ‘‘IA1’’)  IV2 = stress of IA1 (the applied stress from the input action ‘‘IA1’’) And the attributes of the output action ‘‘OA1’’ might include:

be represented by the specified design metrics and the relationships between these design metrics and their relevant design variables, including the properties of the materials, can be derived from the behavioural process model by exploiting the variable dependency graph [22]. Once the properties of all individual materials are known, the existing materials selection methodologies can be applied. Hence, materials selection when applied in combination can be effectively supported. The interface clearly provides the boundary conditions that are transferred to each component/material as shown in Fig. 2.

Specify functional and other design requirements and constraints

 OV1 = field strength of OA1 (the produced field strength from the output action ‘‘OA1’’) Similarly, the behaviour actor may also have its attributes. In this example, the behaviour actor and its relevant attributes are:

Knowledge of components/materials achieving functions

Successful?

Y Multiple-mapping strategy

 C1 = piezoceramic material (behaviour actor)  CV1 = piezoelectric constant of C1 (an attribute of the behaviour actor ‘‘C1’’)

N Behaviour-assisted mapping

Successful? Knowledge of physical/material phenomena achieving functions

Y

N “Behavioural process”assisted mapping

All functions mapped? Backward synthetic search for individual behaviours to form a behavioural process

N

Y Any identified material not specific enough? Any material yet to be determined? Y Develop inter-level behavioural process model

Work out couplings and determine properties of individual materials

Apply materials selection methodologies for individual materials

N

Inter-level behavioural modelling strategy

The attributes of a material are in effect the materials properties, like CV1 in this example. Hence materials properties are incorporated in the behaviour representation, as long as the materials contribute the exhibition of the behaviour. This representation scheme can be used to incorporate all relevant information relating to a behaviour actor in delivering its behaviour, both in the interface (the input/ output action and the corresponding attributes) and the behaviour actor itself (the component or material and the corresponding attributes). By modelling the behavioural process of an engineering system, all the relevant information of the components, materials (including materials in combination) of the system and their interactions will be represented by the individual behaviours of the behavioural process. The relationships between these different aspects of information, which are governed by the relevant physical or material principles, may be characterised by relevant expressions. For example, the relationship between OV1 and AV1, IV1 in the above example is governed by the piezoelectric principle, which can be expressed as ‘‘OV1 = AV1 · IV1’’. As can be seen, such a behavioural process model enables all relevant design information be represented in an integrated manner. The model provides a platform for the designers to work out the couplings between the properties of the materials used in combination, as well as between the properties of the materials and the attributes of the relevant components. Consequently, it facilitates design decision-making in determining the properties of individual material, such that they can collectively achieve optimal design metrics as desired. One method to achieve this goal is by applying a suitable optimisation algorithm, where the objective function may

Direct mapping

Selected materials Fig. 3. Procedure in applying strategies to support design decisionmaking.

K.L. Edwards, Y.-M. Deng / Materials and Design 28 (2007) 1288–1297

Their interdependence is clearly an additional requirement to be satisfied, and may necessitate a dynamic approach if the interface is not static, as in the case of a mechanism, where relative motion exists. However, if adequate information on the system is accessible (relationship between materials and/or components in combination), despite the additional complexity the materials selection process outlined is envisaged robust enough to cope. 3.3. Procedure in applying the design strategies The procedure for applying the two design strategies embedded as an integral part of the design process discussed above is shown in Fig. 3. From this flowchart, it can be seen that the two knowledge libraries provide support for the design activities that apply the design strategies, where the knowledge of commonly used or domain-specific components and materials achieving some functions is used to support direct mapping; while the knowledge of physical phenomena or materials phenomena achieving some functions provides support for the other two types of mapping, as well as for the inter-level behavioural modelling. For both ‘‘behavioural process’’-assisted mapping and inter-level behavioural modelling, the backward synthetic search methods can be used to identify or develop the individual behaviours for the behavioural process. It should be noted that the behavioural process used in the mapping operation is meant for achieving a single functional requirement; while the behavioural process developed for inter-level behavioural modelling operation is used as a platform for the designers to consider the properties of all materials that might be coupled with each other and/ or with some components, hence it might involve all system functions and behaviours, both at the component level and material level. In other word, the former behavioural process can be understood as ‘‘local’’, while the latter behavioural process can be understood as ‘‘global’’. 4. Discussion The issues addressed in this paper are complex, subjective and intellectually demanding and therefore necessitate a range of strategies and methodologies, so as to provide optimal solutions to varying design problems. The design strategies outlined in the previous section only partly engage in the full range of issues but does provide the basis of an integrated approach that in principle is capable with further development of tackling a full spectrum of design problems. The design problems faced are considerable and highly dependent on the application. However, materials can be classified broadly as mechanical, chemical and physical, highlighting desirable specific properties such as strength, stiffness, corrosion resistance, density, thermal conductivity, etc., and link these to the application. Knowledge of the application (e.g. aerospace, automotive, marine, etc.) is vital because it has such a large influence on the appro-

1295

priateness of materials, particularly with regard to operating environment. Therefore, even though real world applications are considered in general terms when developing the approaches outlined in this paper, a validation programme based on actual case studies will help refine and highlight aspects for future development. An additional but regularly occurring problem is the range of materials properties emanating from the classification system outlined at the beginning of this paper that overlap (and contradict), creating complexity and confusion in design decision making. This situation becomes quite complicated when attempting to satisfy the further conflicting requirements of different materials (and processes) and/or components used in combination. Further this is even more difficult if the interface between materials and/or components is not fixed. A crude but effective initial approach is the use of the ‘Venn’ diagram shown in Fig. 2, irrespective of the stage reached in the process outlined in Fig. 3, which helps identify those properties associated with ‘core’ individual materials and/or component requirement as opposed to those properties associated with the interface. Obviously, similar properties can reside in both regions but the interdependence of the two regions has to always be considered. Those properties that are common to each material/component and the interface between them are more tolerant to design change. When all properties are common to all regions irrespective of the material or component, there are design opportunities that can be exploited such as component consolidation, exploited considerably in polymer mouldings. Clearly, if the interface is changing (geometrical, physical, chemical, etc.), then the property requirements will need to be mapped over the anticipated variation to ensure design requirements are met in all desired situations. Any decision support system has to be capable of handling all of these multiple attribute characteristics, which can be considerable in single materials and/or components without the additional complexity of several materials and/or components being considered simultaneously. This leads to even more conflict, greater complexity and confusion, resulting in poor quality decision-making, which at best creates sub-optimal solutions and at worse leads to mistakes. The normal but conservative approach to problem solving is to rely on experience but this is not reliable and can stifle innovation and lead to incremental development only. It is well understood that design problem solving is openended, with many solutions possible to a given problem. Selecting materials is no different and whether the process is occurring at the conceptual design stage or the detail design stage, options will be presented. These options will need to be evaluated and decisions made on the outcome. This is analogous to normal management practice. Key to this critical aspect of decision-making is being aware of the consequences to enable the best decision to be made. Fortunately, unlike normal management practice which is often highly subjective and based on limited information, it is possible in principle to provide a plethora of information

1296

K.L. Edwards, Y.-M. Deng / Materials and Design 28 (2007) 1288–1297

for selecting materials. The main difficulty is managing the large amount of conflicting information in a meaningful and structured way. Also through context dependent reasoning (use of applications as above), courses of action and their implication can be better considered. It is in the weighing up of options that is critical to deciding the best course of action, provided of course the quality of the information (quantitative and qualitative) is both accurate and complete. These aspects are crucial in attempting to simultaneously satisfy structural and materials selection aspects, particularly at the conceptual stage of component design when information is both disparate and vague. The nature of the problem is quite different in the case of selecting materials in combination. This tends to occur more in the latter stages of the design process where considerably more detailed information is available. However, both scenarios call for more structured information management and decision support. 5. Concluding remarks The previous sections have exposed an under researched area regarding materials selection, namely, how to support design decision-making when materials are applied in combination, and when they are coupled with the structural determination or design of components. The fundamental problems involved have been identified and discussed to help facilitate simultaneous consideration of components and materials at mainly the early design stage, as well as to provide a platform for the designer to work out the couplings among the properties of the materials and their corresponding components, so that the respective properties of the materials can be determined. A multiple-mapping strategy and an inter-level behavioural modelling strategy have been proposed to provide respective support to the problems highlighted. However, because of the complexity of the addressed problems, these strategies, their implementation in software where appropriate, and validation through case studies, cannot provide total support to all the issues surrounding the engineering design problems envisaged. There is still the role of design experience that also needs to be properly factored into the process as this underpins most design decision-making. This experience is normally associated with the retention and recall of best practice, and with the supporting rationale it provides vital additional knowledge necessary to help make sense of the sheer breadth and complexity of conflicting information. This is compounded when a combination of materials and components, a more normal situation, is under consideration. For the future, greater consideration needs to be given to mapping selective management techniques, to complement behavioural process modelling, to the solution of technical multiple attribute problems, where subjectivity and conflict are dominant. Also more emphasis needs to be placed on mechanical engineering systems as a whole as well as the individual materials and components that

make up these systems. The interaction of materials and components, under various conditions, is well researched and widely available in the public domain literature but needs to be better built into the materials selection decision-making process. Some of this information is anecdotal in nature, but regularly utilised in design practice, and better ways of handling such information also needs to be developed.

References [1] Pahl G, Beitz W. Engineering design: a systematic approach. second ed. Springer-Verlag; 1996. [2] Hubka V, Eder WE. Theory of technical systems. Springer; 1988. [3] Pugh S. Total design: integrated methods for successful product engineering. Addison-Wesley; 1995. [4] Suh NP. The principles of design. Oxford University Press; 1990. [5] Altshuller G. And suddenly the inventor appeared: TRIZ, the theory of inventive problem solving. Worcester, MA, USA: Technical Innovation Centre Inc.; 1996. [6] Ashby MF, Brechet YJM, Cebon D, Salvo L. Selection strategies for materials and processes. Mater Design 2004;25:51–67. [7] Ashby MF. Materials selection in mechanical design. Pergamon Press; 1992. [8] Ashby MF, Johnson K. Materials and design: the art and science of materials selection in product design. Oxford: Butterworth Heinemann; 2002. [9] Deng Y-M, Edwards KL. The role of materials identification and selection in engineering design. Mater Design 2005;26 [available online]. [10] Edwards KL. Selecting materials for optimum use in engineering components. Mater Design 2005;26:469–73. [11] Giudice F, La Rosa G, Risitano A. Materials selection in the lifecycle design process: a method to integrate mechanical and environmental performances in optimal choice. Mater Design 2005; 26:9–20. [12] Sapuan SM. A knowledge-based system for materials selection in mechanical engineering design. Mater Design 2001;22:687–95. [13] Jee D-H, Kang K-J. A method for optimal material selection aided with decision making theory. Mater Design 2000;21:199–206. [14] Dieter GE. Engineering design: a materials and processing approach. second ed. McGraw-Hill; 1991. [15] Waterman NA, Ashby MF, editors. The materials selector. Chapman & Hall; 1997 [3 Volumes]. [16] Dieter GE (Vol. Ed.). ASM handbook, vol. 20: Materials selection and design. ASM International; 1997. [17] Rajan K. Materials informatics. Mater Today 2005;8(10):38–45. [18] Edwards KL. Linking materials and design: an assessment of purpose and progress. Mater Design 2002;23:255–64. [19] Edwards KL. Designing of engineering components for optimal materials and manufacturing process utilisation. Mater Design 2003; 24:355–66. [20] Raj R. An interdisciplinary framework for the design and life prediction of engineering systems. Trans ASME, J Eng Mater Technol 2000;122:348–54. [21] Subbarayan G, Raj R. A methodology for integrating materials science with system engineering. Mater Design 1999;20:1–12. [22] Lu WF, Deng Y-M. A systems modelling methodology for materials and engineering systems design integration. Mater Design 2004; 25:459–69. [23] Huang J, Fadel GM. Heterogeneous flywheel modelling and optimisation. Mater Design 2000;21:111–25. [24] Walker M, Smith R. A procedure to select the best material combinations and optimally design composite sandwich cylindrical shell for minimum mass. Mater Design 2005 [available online].

K.L. Edwards, Y.-M. Deng / Materials and Design 28 (2007) 1288–1297 [25] Ljungberg LY, Edwards KL. Design, materials selection and marketing of successful products. Mater Design 2003;24:519–29. [26] Deng Y-M, Lu WF. From function to structure and material: a conceptual design framework. In: Proceedings of the fifth international symposium on tools and methods of competitive engineering, 13–17 April, 2004, Lausanne, Switzerland, vol. 1. p. 95–106.

1297

[27] Deng Y-M, Tor SB, Britton GA. A dual-stage functional modelling framework with multi-level design knowledge for conceptual mechanical design. J Eng Design 2000;11(4):347–75. [28] Deng Y-M. Function and behaviour representation in conceptual mechanical design. AI EDAM: Artif Int Eng Design, Anal Manufact 2002;16(5):343–62.