- Email: [email protected]
QFD in new production technology evaluation
Int. J. Production Economics 67 (2000) 103}112
QFD in new production technology evaluation Antony Lowe *, Keith Ridgway , Helen Atkinson Department of Mechanical Engineering, University of Shezeld, Mappin Street Shezeld S1 3JD, UK Department of Materials Engineering, University of Shezeld, Mappin Street, Shezeld S1 3JD, UK Received 21 September 1998; accepted 12 October 1999
Abstract The large number of new technologies being developed means it is vital for organisations to make appropriate selections that will translate limited capital resources into maximum competitive advantage. However, a thorough economic evaluation of a technology requires considerable time and e!ort. This paper presents a tool developed from the techniques of quality function deployment. This tool allows a rapid evaluation of the feasibility of using the thixoforming process to manufacture products. The paper describes the semi-solid metal processing technology of thixoforming, the relevant quality function deployment techniques and the approach used to develop the tool. 2000 Elsevier Science B.V. All rights reserved. Keywords: Quality function deployment; Technology; Selection; Evaluation; Thixoforming
1. Introduction Computer Aided Design, Computer Aided Manufacture, Computer Integrated Manufacture, rapid prototyping, high-speed machining, hot isostatic pressing and thixoforming are just a small selection of the many innovative technologies being promoted around the world. All of these technologies have the potential to provide manufacturing organisations with a competitive advantage over their competitors. The major di$culty though is in selecting and investing in the most suitable technology. Not only does an organisation have to make
* Corresponding author. Antony Lowe Department of Mechanical Engineering, University of She$eld, Mappin Street, Shef"eld, S1 3JD, UK. Tel.: 0-114-2227767; fax: 0-114-2753671. E-mail address: a.j.lowe@she$eld.ac.uk (A. Lowe)
this selection accurately but also rapidly if it is not to lose the opportunity to an ambitious competitor. Investment in many of these technologies involves a degree of risk, especially for those with few or no existing commercial applications. In these cases the lack of available process expertise means that investment in research and testing will be necessary to develop a stable and reliable process. For such a developed commercial application the bene"ts which have been attributed to the technology by its proponents may not appear as anticipated. There is therefore a need to ensure that this investment and development e!ort is not wasted on inappropriate technologies. Investment appraisal techniques have been re"ned over a number of decades to provide a reasonable "nancial measure of the value of making an investment in capital equipment. Techniques used include Payback Period, Net Present Value and
0925-5273/99/$ - see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 5 - 5 2 7 3 ( 9 9 ) 0 0 1 2 5 - 5
104
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112
Internal Rate of Return. The bene"ts o!ered by these new technologies are not always directly quanti"able (e.g. increased process #exibility) and cannot be included in conventional "nancial appraisal approaches. Primrose [1] has shown how such bene"ts can be accurately quanti"ed, but the level of detail required in such an accurate economic viability analysis requires a signi"cant time and allocation of resources. There is a need to subject these technologies to some kind of "lter before an organisation commits itself to the requirements of an in-depth economic analysis. Researchers in the Department of Engineering Materials at the University of She$eld have been involved in the development of one such innovative manufacturing technology. The technology for the semi-solid processing of metals, also called thixoforming, has been available for some years. Although commercialised in Japan, the US and Europe it has yet to be commercialised in the UK despite the many bene"ts promised. A three year study on the `Exploitation of Semi-Solid Processinga was embarked upon to provide stimulus for this adoption. One of the key components of this work was a software package for potential users of the technology. This software `ThixoCosta was to provide facilities for carrying out a `Cost}Bene"t Analysisa of the technology using a business process perspective, for the application circumstances of any user company. During the development of this software it became obvious that before the managers of a company were able to commit the resources necessary for gathering the cost data for such an analysis, they wanted an `experta opinion on whether their particular products were likely to be suitable. Part of the aim of the project became to analyse the experience held within the Thixoforming Research Group and incorporate it, in a structured way, into a decision making tool within the software, to o!er users such an opinion. Initially, the application of a neural net was considered. This could accept as inputs, measures of relevant attributes that characterise a product. The net could be trained with the details of existing commercial thixoformed products and those of products currently recognised as unsuitable. The
small number of existing commercial users of the technology meant that the availability of such training data was limited and di$cult to obtain, e!ectively preventing the use of this approach. Quality function deployment (QFD) was therefore considered as an evaluation tool. This paper describes the process used to develop this tool and highlights its potential for application to technologies other than thixoforming.
2. Introduction to thixoforming The opportunity for forming processes based upon semi-solid metal alloys was "rst recognised by Spencer et al. [2] while studying the unusual properties of vigorously stirred tin-lead slurries in the early 1970s. The microstructure of the stirred alloy comprises rounded particles of solid surrounded with liquid of a lower melting point, rather than the normal angular and interlocking dendrites (Fig. 1). This microstructure gives the material its thixotropic properties, i.e. when sheared the material #ows but when allowed to stand it thickens. Thixoforming is one member of the family of semi-solid forming processes and possesses characteristics of both casting and forging. The solid feed stock for thixoforming must be pre-treated, so that on heating into the semi-solid state the microstructure is spheroidal rather than dendritic. A general thixoforming process cell incorporates four operations (Fig. 2): 1. A bar of thixoformable raw material is cut into appropriate slug lengths. 2. The slugs are heated in a controlled manner, using either an induction coil or a mu%e furnace into a uniform `mushya state. 3. The heated slug is transferred to the shot sleeve of a suitably modi"ed die casting machine. Initially, the heated billet of material behaves like a solid, holding its shape unsupported and able to withstand the low stresses of handling. When the semi-solid material is subjected to shear stresses during injection into a die, it #ows in a smooth laminar manner and accurately "lls the die cavity.
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112
105
Fig. 1. Dendritic (left) and spheroidal (right) microstructures.
4. The component feeding and gating systems are then removed using a clipping press or band saw. Since the alloy #ows under very low stresses (around 0.1 MPa) the mechanical stresses acting on the die during "lling are small. This property allows softer die materials, such as graphite, easily machinable stainless steels and disposable one-shot nonmetallic dies as utilised in the work carried out by McLelland et al. [3] to be used. This permits the economic application of thixoforming to the small production volumes required by rapid prototyping and mass customised production. Kirkwood [4] describes the attributes of thixoforming as follows: E An energy e$cient process which is easily automated and controlled to achieve consistency. E Production rates that are similar to pressure die casting or better. E Smooth "lling of the die with no air entrapment and low shrinkage porosity giving parts of high
Fig. 2. The thixoforming process.
integrity and allowing application to high strength heat treatable alloys. E Lower processing temperatures reduce the thermal shock of the die, promoting die life and allowing the processing of high melting point alloys (such as tool steels and stellites) that are di$cult to form by other means. E Fine, uniform microstructures giving enhanced component properties.
106
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112
E Improved yield from raw materials and weight savings through optimised component design. The high mechanical, geometric and surface quality of components produced through thixoforming can justify the removal of additional production processes such as machining steps and the need for reinforcing inserts. The resulting simpler more #exible, production process requires fewer resources for control and enables a more rapid response to changing customer requirements. The extent to which these bene"ts can be achieved is dependent upon the design of the component and dies, the optimisation of the processing conditions, the degree of integration into the existing production process and the demands of an organisation's business environment. Chiarmetta [5] and Kenney et al. [6] document a number of drawbacks which currently limit the commercial viability of thixoforming: E The high cost of raw material and the low number of its suppliers. E Considerable research e!ort and expense is required to implement a viable manufacturing process due to the limited available process knowledge. E Die development costs are higher than for conventional forming technologies because of the lack of available process experience and design rules. E The personnel employed to operate and maintain thixoforming plant require a higher level of training than equivalent traditional operators and command higher wages. The number and type of bene"ts and drawbacks attributed to thixoforming are typical of many new manufacturing technologies. For an organisation to make an accurate assessment of the "nancial attractiveness of an investment the potential impact of all these must be considered. This would necessitate a considerable data gathering exercise and the requirement for a detailed understanding of the technology. The proposed QFD-based tool provides a "rst step to prevent the majority of inappropriate assessments and can also speed the eventual "nancial analysis by highlighting the most important bene"ts and drawbacks to be evaluated.
Fig. 3. A generic QFD house of quality [10].
3. Quality function deployment and the prioritisation matrix 3.1. Quality function deployment Quality function deployment (QFD) is a set of powerful product development tools and procedures originated in Japan to take the concepts of quality control from manufacturing and transfer them to the new product development process. Akao [7] "rst began to develop the concepts of QFD in the mid-1960s and this led to the "rst recognised implementation of the approach in 1972 at the Mitsubishi Heavy Industries Kobe Shipyard [8]. The house of quality (HOQ) matrix is the central construct of QFD and is its most recognised form. It is described by Hauser and Clausing [9]: The house of quality is a kind of conceptual map that provides the means for interfunctional planning and communications. There are nearly as many forms of the HOQ as there have been applications and it is this adaptability to the needs of a particular project or user group which is one of its strengths. Fig. 3 illustrates the general format used for the HOQ matrix, which is made up from six major
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112
107
3.2. The prioritisation matrix
Fig. 4. Prioritisation matrix with weighted rows and cells.
components: 1. 2. 3. 4. 5. 6.
Customer requirements (Whats) Technical requirements (Hows) Planning matrix Interrelationship matrix Technical correlation matrix Priorities, benchmarks and targets for technical descriptors
The HOQ matrix translates the needs expressed by a customer into the design targets of a proposed new product. This is e!ected by entering weightings into the central Interrelationship Matrix which represent the scale of the perceived relationship between each customer and technical requirement. These weightings are combined with measures of the relative importance of each customer requirement to calculate a priority for each of the technical requirements in terms of satisfying customers. The remainder of the HOQ contains competitive benchmarking and market analysis information which is also relevant to the setting of the target design values. A simpli"ed form of the HOQ matrix was utilised in this work. The Technical Correlations and Planning matrices were removed and only the prioritised requirements row included at the base. The resulting matrix format is referred to as a prioritisation matrix.
In a prioritisation matrix, as in the HOQ, the interrelationships expressed in individual cells are assigned a value. Numbers or symbols are commonly used in QFD to represent the strength of such an interrelationship. Weightings are also used to illustrate the relative importance of items within the list of customer requirements. An overall value for the `importance of the strengtha of a particular interrelationship can then be numerically calculated. This involves multiplying the weighting in each matrix cell by the relative importance of the row item with which it is associated (see Fig. 4). These values can then summed down a column to give a measure of the `importance of the strengthsa of the overall relationships between a particular column item and the entire list of row items (see Fig. 4). In this manner the matrix allows the items of one list to be ranked based upon their relationships with the items of another list.
4. Building the technology evaluation tool This basic matrix tool was developed in order to focus upon the interrelationships between a product's characteristics (customer attributes in the HOQ) and the characteristics of the thixoforming process (engineering characteristics in the HOQ). The weightings of these interrelationships were combined in a manner similar to that used in a prioritisation matrix to illustrate the relative importance of the characteristics of thixoforming. The sum of these values was used to give a product a score for its overall suitability for thixoforming (see Fig. 5). The "nal tool was referred to as a multi-attribute matrix analysis. Nine characteristics of the thixoforming process were selected based upon the expertise held by the Thixoforming Research Group and drawing on the bene"ts and drawbacks documented in the literature. The eight product characteristics (perhaps better described as product issues) were selected as being relevant to the thixoforming process and su$ciently independent to avoid the repetition of analysis issues. (the chosen characteristics are shown in Fig. 5)
108
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112
3 Medium relationship between the product and thixoforming characteristics 1 Weak relationship between the product and thixoforming characteristics 0 No relationship between the product and thixoforming characteristics (cells generally left blank)
Fig. 5. Proposed multi-attribute matrix analysis technology evaluation tool.
The latter three thixoforming process characteristics (raw material, process development & skills/ wages) are considered to currently be the major drawbacks of the technology. The weightings applied to these characteristics were therefore subtracted rather than added in the calculation of the overall product suitability score. To standardise the characterisation of a product under consideration, each of the eight product characteristics were allocated three settings. For example Critical/Important/Unimportant, High/Small/No, Complex/Medium/Simple and Long/Medium/Short depending upon the nature of the characteristic. Before the matrix relationships could be discussed, it was important to be clear about the de"nition of each characteristic and each of its settings (see Table 1). The next stage was to establish the values for each of the interrelationships that could be displayed within the matrix. These were selected from a four-point scale: 5 Strong relationship between the product and thixoforming characteristics
The issues that were relevant to each combination of product characteristic setting and thixoforming characteristic were generated through a number of repeated discussions and individual reviews carried out by the members of the Thixoforming Research Group. These were documented and then formed the basis for the weighting values to be allocated to each cell. Once again these "gures were subject to several discussions and reviewed to ensure all possible perspectives were included. Table 2 illustrates the transition from a combination of characteristics, to the relevant issues and so into cell weightings. The completed weighting evaluations were then developed into a tool within the ThixoCost software package (illustrated in Fig. 6). This was written in Visual Basic 4.0 and allows a user to rapidly characterise their product using a scrolling menu for each of the product characteristics. As each characteristic setting is chosen the interrelationship values are automatically entered into the cells of the matrix. A second set of scrolling menus provides the user with a facility to enter the relative importance of each product characteristic. These allow entries on a scale of 1}5. To ensure comparable scoring between users (i.e. avoid the situation where a user rates all characteristics with the highest or lowest level of importance) a system for the allocation of importance scores is recommended, as detailed in Table 3 (an application of this importance scoring system is shown in Fig. 6). The system does not have to be strictly adhered to but the use of the full range of importance scores is necessary if the calculated product suitability score is to be analysed correctly by the software. Once the user has selected a setting and importance weighting for each product characteristic, the software calculates a `prioritya for each characteristic of thixoforming in a similar manner to a
Setting 1
Critical: Minimising the weight of the component is a critical issue
Critical: The strength of the component is a critical issue
Complex: The geometry of the component is highly complex. e.g. fuel rail
Critical: The meeting of demanding material property and dimensional tolerances is a critical issue
High: The market will allow a premium price to be charged for a thixoformed product
Long: The lead time between receiving an order and despatching a product is not important and is greater than 3 months
High: A customised product is produced in small batches, unique to each customer
High: In the current process there are more than 3 machining/"nishing/reinforcing operations which could be removed through thixoforming
Product characteristic
Weight
Strength
Geometry
Tolerances
Price premium
Lead Time
Flexibility
Finishing operations
Table 1 De"nitions of product characteristic levels
Medium: There are 1 or 2 machining/"nishing/ reinforcing operations in the current process which could be removed through thixoforming
Medium: A medium sized range of standard products is produced
No: There are no existing process steps which could be removed through thixoforming
Low: A single standard product is produced in large batches
Short: The lead time between receiving an order and despatching a product is critical and is less than 1 month
No: The market will allow no premium to be charged for a thixoformed product
Medium: The market will allow a small premium price to be charged for a thixoformed product Medium: The lead time between receiving an order and despatching a product is between 3 months and 1 month
Unimportant: Meeting dimensional and material property tolerances is not important
Simple: The geometry of the component is basic. e.g. a chisel
Unimportant: The strength of the component is not an important issue
Unimportant: The weight of the component is not important
Setting 3
Important: Material property and dimensional tolerances are important but not critical
Medium: The geometry of a component is of medium complexity
Important: The strength of the component is important but not critical
Important: The weight of the component is important but not critical
Setting 2
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112 109
110
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112
Table 2 Example evaluation of interrelationship values Characteristics
Relevant issues
Lead time & process development
Interrelationship weightings for each product characteristic settings
The time constraints required by short lead times will demand high process development e!ort.
Strength and mechanical The mechanical integrity achieved in thixoformed components is integrity an advantage in high strength applications, in material savings and heat treatments.
Table 3 Recommended relative importance scoring Relative importance score
Application of score
5 4 3 2 1
The The The The The
single most important characteristic next two most important characteristics two characteristics of medium importance next two less important characteristics single least important characteristic
prioritisation matrix. The magnitude of these column totals illustrate those bene"ts and drawbacks which are of the greatest and least impact in the considered potential implementation of the techno-
Long
Medium
Short
1
3
5
Critical
Important
Unimportant
5
3
1
logy. This information is useful in concentrating any "nancial studies on appraising the use of the technology as well as focusing the development necessary in implementing the process (in the Fuel Rail example shown in Fig. 6, the near net shape capability of thixoforming is highlighted as the most important bene"t by its highest characteristic score. Process development is also highlighted as being the most serious drawback by its lowest score). The overall product suitability score (referred to as the `Evaluation Scorea in Fig. 6) is the sum of these `prioritiesa, with the three drawback column totals being attributed negative values in this calculation. The interpretation of this score is made by the software using the scale shown in Table 4. These values were established by the application of the tool to a range of existing commercial
Table 4 Interpretation of product evaluation scores Product evaluation score
Software interpretation
Greater than 100
The product under consideration is suitable for the application of thixoforming technology and a detailed cost bene"t analysis should now be undertaken.
Between 100 and 80
The score allocated to this product by the multi-attribute matrix analysis shows it to be a borderline case. You may wish to repeat this evaluation paying close attention to the characteristic settings and importance weightings chosen before deciding whether to pursue a further cost bene"t analysis.
Less than 80
The product under consideration is presently unsuited to the application of thixoforming.
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112 Table 5 The multi-attribute matrix analysis applied to existing thixoformed products Existing thixoformed product
Multi-attribute matrix analysis evaluation score
Fuel rail (Weber}Italy) Steering link (Buhler}Switzerland) Automobile wheel (Alumax}US) Suspension subassembly (Stampal}Italy)
117 105 109 125
111
a series of tests were carried out. A set of 20 products were selected that were not used in the development of the suitability score interpretation model. These products included commercially produced thixoformed products and also products under consideration for application of the process. The software output was then compared with the opinions of the Thixoforming Group. Table 5 illustrates example values produced by the tool for existing thixoformed products.
6. Further applications thixoformed products, potential products and those product types deemed unsuitable to the process as identi"ed and characterised by the Thixoforming Group.
5. Validation of the evaluation tool In order to verify the validity of the recommendations made by the technology evaluation tool
The validity of the evaluation tool was proven for the range of products with which it was tested. In addition to successfully di!erentiating between those products that are suitable and unsuitable for the process, it e!ectively highlighted the most important process characteristics for consideration in any further analysis and in the "nal development of the process (e.g. the near net shape and process development characteristics in the Fuel Rail
Fig. 6. The multi-attribute matrix analysis screen from ThixoCost.
112
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112
example). Since its development, the multi-attribute matrix analysis has been applied to the products of two UK research partner companies and was successfully demonstrated at a thixoforming conference applied to product characteristics provided by delegates. The development of this tool drew upon a pool of expertise in the technology and a relatively small set of product data used in devising the score interpretation model. It e!ectively utilises such expertise to translate the characteristics of a product into a recommendation for its further consideration for application to a particular technology.
7. Conclusions The multi-attribute matrix analysis tool has been developed from the techniques of QFD and applied to the evaluation of potential products for an innovative metal forming process. This technology evaluation tool does not substitute for a comprehensive economic analysis. The level of subjectivity necessary in allocating characteristic settings, importance weightings and interrelationship scores, if such a tool is to be applied generically, means speci"c accuracy cannot be achieved in each case. Instead the value of such a tool is in the focusing of a subsequent economic analysis or removing the need for it altogether. From this work, it would seem that a similar approach could be as e!ectively applied to the analysis of other new innovative technology applications. High-speed machining for example has both bene"ts (e.g. rapid production cycle, thinner webs, etc.) and drawbacks (e.g. high cost of machines, lubricants and tooling) which the tool could evaluate against relevant product characteristics (e.g. geometry, material, machinability, etc.).
Acknowledgements We are grateful to the Engineering and Physical Sciences Research Council for their sponsorship of the work carried out here under the aegis of the Innovative Manufacturing Initiative (Project No. IMI/RP/01/065).
References [1] P.L. Primrose, Investment in Manufacturing Technology, Chapman and Hall, London, 1991, p. 49. [2] D.B. Spencer, R. Mehrabian, M.C. Flemings, Rheological behaviour of Sn-15 Pct Pb in the crystallization range, Metallurgical Transactions 3 (1972) 1925}1932. [3] A.R.A. McLelland, P.R.G. Anderson, H.V. Atkinson, D.H. Kirkwood, The application of ceramic moulds to semi-solid metal forming, Proceedings of the Fourth International Conference on Semi-Solid Processing of Alloys and Composites, The University of She$eld, 1996, pp. 274}277. [4] D.H. Kirkwood, Semisolid metal processing, International Materials Reviews 39 (5) (1994) 173}189. [5] G. Chiarmetta, Thixoforming of automobile components, Proceedings of the Fourth International Conference on Semi-Solid Processing of Alloys and Composites, The University of She$eld, June 1996, pp. 204-207. [6] M.P. Kenney, J.A. Courtois, R.D. Evans, G.M. Farrior, C.P. Kyonka, A.A. Koch, K.P. Young, Semisolid metal casting and forging, Metals Handbook, 9th Edition, ASM International, Metals Park, OH, 1988, p. 333. [7] M. Kogure, Y. Akao, Quality function deployment and CWQC in Japan, Quality Progress 16 (10) (1993) 25}29. [8] N. Singh, Systems approach to computer-integrated design and manufacturing, Wiley, Chichester, UK, 1996, p. 133. [9] J.R. Hauser, D. Clausing, The house of quality, Harvard Business Review (1988) 63. [10] L. Cohen, Quality function deployment: how to make QFD work for you, Addison-Wesley, Readings MA, 1995, p. 70.
QFD in new production technology evaluation Antony Lowe *, Keith Ridgway , Helen Atkinson Department of Mechanical Engineering, University of Shezeld, Mappin Street Shezeld S1 3JD, UK Department of Materials Engineering, University of Shezeld, Mappin Street, Shezeld S1 3JD, UK Received 21 September 1998; accepted 12 October 1999
Abstract The large number of new technologies being developed means it is vital for organisations to make appropriate selections that will translate limited capital resources into maximum competitive advantage. However, a thorough economic evaluation of a technology requires considerable time and e!ort. This paper presents a tool developed from the techniques of quality function deployment. This tool allows a rapid evaluation of the feasibility of using the thixoforming process to manufacture products. The paper describes the semi-solid metal processing technology of thixoforming, the relevant quality function deployment techniques and the approach used to develop the tool. 2000 Elsevier Science B.V. All rights reserved. Keywords: Quality function deployment; Technology; Selection; Evaluation; Thixoforming
1. Introduction Computer Aided Design, Computer Aided Manufacture, Computer Integrated Manufacture, rapid prototyping, high-speed machining, hot isostatic pressing and thixoforming are just a small selection of the many innovative technologies being promoted around the world. All of these technologies have the potential to provide manufacturing organisations with a competitive advantage over their competitors. The major di$culty though is in selecting and investing in the most suitable technology. Not only does an organisation have to make
* Corresponding author. Antony Lowe Department of Mechanical Engineering, University of She$eld, Mappin Street, Shef"eld, S1 3JD, UK. Tel.: 0-114-2227767; fax: 0-114-2753671. E-mail address: a.j.lowe@she$eld.ac.uk (A. Lowe)
this selection accurately but also rapidly if it is not to lose the opportunity to an ambitious competitor. Investment in many of these technologies involves a degree of risk, especially for those with few or no existing commercial applications. In these cases the lack of available process expertise means that investment in research and testing will be necessary to develop a stable and reliable process. For such a developed commercial application the bene"ts which have been attributed to the technology by its proponents may not appear as anticipated. There is therefore a need to ensure that this investment and development e!ort is not wasted on inappropriate technologies. Investment appraisal techniques have been re"ned over a number of decades to provide a reasonable "nancial measure of the value of making an investment in capital equipment. Techniques used include Payback Period, Net Present Value and
0925-5273/99/$ - see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 5 - 5 2 7 3 ( 9 9 ) 0 0 1 2 5 - 5
104
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112
Internal Rate of Return. The bene"ts o!ered by these new technologies are not always directly quanti"able (e.g. increased process #exibility) and cannot be included in conventional "nancial appraisal approaches. Primrose [1] has shown how such bene"ts can be accurately quanti"ed, but the level of detail required in such an accurate economic viability analysis requires a signi"cant time and allocation of resources. There is a need to subject these technologies to some kind of "lter before an organisation commits itself to the requirements of an in-depth economic analysis. Researchers in the Department of Engineering Materials at the University of She$eld have been involved in the development of one such innovative manufacturing technology. The technology for the semi-solid processing of metals, also called thixoforming, has been available for some years. Although commercialised in Japan, the US and Europe it has yet to be commercialised in the UK despite the many bene"ts promised. A three year study on the `Exploitation of Semi-Solid Processinga was embarked upon to provide stimulus for this adoption. One of the key components of this work was a software package for potential users of the technology. This software `ThixoCosta was to provide facilities for carrying out a `Cost}Bene"t Analysisa of the technology using a business process perspective, for the application circumstances of any user company. During the development of this software it became obvious that before the managers of a company were able to commit the resources necessary for gathering the cost data for such an analysis, they wanted an `experta opinion on whether their particular products were likely to be suitable. Part of the aim of the project became to analyse the experience held within the Thixoforming Research Group and incorporate it, in a structured way, into a decision making tool within the software, to o!er users such an opinion. Initially, the application of a neural net was considered. This could accept as inputs, measures of relevant attributes that characterise a product. The net could be trained with the details of existing commercial thixoformed products and those of products currently recognised as unsuitable. The
small number of existing commercial users of the technology meant that the availability of such training data was limited and di$cult to obtain, e!ectively preventing the use of this approach. Quality function deployment (QFD) was therefore considered as an evaluation tool. This paper describes the process used to develop this tool and highlights its potential for application to technologies other than thixoforming.
2. Introduction to thixoforming The opportunity for forming processes based upon semi-solid metal alloys was "rst recognised by Spencer et al. [2] while studying the unusual properties of vigorously stirred tin-lead slurries in the early 1970s. The microstructure of the stirred alloy comprises rounded particles of solid surrounded with liquid of a lower melting point, rather than the normal angular and interlocking dendrites (Fig. 1). This microstructure gives the material its thixotropic properties, i.e. when sheared the material #ows but when allowed to stand it thickens. Thixoforming is one member of the family of semi-solid forming processes and possesses characteristics of both casting and forging. The solid feed stock for thixoforming must be pre-treated, so that on heating into the semi-solid state the microstructure is spheroidal rather than dendritic. A general thixoforming process cell incorporates four operations (Fig. 2): 1. A bar of thixoformable raw material is cut into appropriate slug lengths. 2. The slugs are heated in a controlled manner, using either an induction coil or a mu%e furnace into a uniform `mushya state. 3. The heated slug is transferred to the shot sleeve of a suitably modi"ed die casting machine. Initially, the heated billet of material behaves like a solid, holding its shape unsupported and able to withstand the low stresses of handling. When the semi-solid material is subjected to shear stresses during injection into a die, it #ows in a smooth laminar manner and accurately "lls the die cavity.
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112
105
Fig. 1. Dendritic (left) and spheroidal (right) microstructures.
4. The component feeding and gating systems are then removed using a clipping press or band saw. Since the alloy #ows under very low stresses (around 0.1 MPa) the mechanical stresses acting on the die during "lling are small. This property allows softer die materials, such as graphite, easily machinable stainless steels and disposable one-shot nonmetallic dies as utilised in the work carried out by McLelland et al. [3] to be used. This permits the economic application of thixoforming to the small production volumes required by rapid prototyping and mass customised production. Kirkwood [4] describes the attributes of thixoforming as follows: E An energy e$cient process which is easily automated and controlled to achieve consistency. E Production rates that are similar to pressure die casting or better. E Smooth "lling of the die with no air entrapment and low shrinkage porosity giving parts of high
Fig. 2. The thixoforming process.
integrity and allowing application to high strength heat treatable alloys. E Lower processing temperatures reduce the thermal shock of the die, promoting die life and allowing the processing of high melting point alloys (such as tool steels and stellites) that are di$cult to form by other means. E Fine, uniform microstructures giving enhanced component properties.
106
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112
E Improved yield from raw materials and weight savings through optimised component design. The high mechanical, geometric and surface quality of components produced through thixoforming can justify the removal of additional production processes such as machining steps and the need for reinforcing inserts. The resulting simpler more #exible, production process requires fewer resources for control and enables a more rapid response to changing customer requirements. The extent to which these bene"ts can be achieved is dependent upon the design of the component and dies, the optimisation of the processing conditions, the degree of integration into the existing production process and the demands of an organisation's business environment. Chiarmetta [5] and Kenney et al. [6] document a number of drawbacks which currently limit the commercial viability of thixoforming: E The high cost of raw material and the low number of its suppliers. E Considerable research e!ort and expense is required to implement a viable manufacturing process due to the limited available process knowledge. E Die development costs are higher than for conventional forming technologies because of the lack of available process experience and design rules. E The personnel employed to operate and maintain thixoforming plant require a higher level of training than equivalent traditional operators and command higher wages. The number and type of bene"ts and drawbacks attributed to thixoforming are typical of many new manufacturing technologies. For an organisation to make an accurate assessment of the "nancial attractiveness of an investment the potential impact of all these must be considered. This would necessitate a considerable data gathering exercise and the requirement for a detailed understanding of the technology. The proposed QFD-based tool provides a "rst step to prevent the majority of inappropriate assessments and can also speed the eventual "nancial analysis by highlighting the most important bene"ts and drawbacks to be evaluated.
Fig. 3. A generic QFD house of quality [10].
3. Quality function deployment and the prioritisation matrix 3.1. Quality function deployment Quality function deployment (QFD) is a set of powerful product development tools and procedures originated in Japan to take the concepts of quality control from manufacturing and transfer them to the new product development process. Akao [7] "rst began to develop the concepts of QFD in the mid-1960s and this led to the "rst recognised implementation of the approach in 1972 at the Mitsubishi Heavy Industries Kobe Shipyard [8]. The house of quality (HOQ) matrix is the central construct of QFD and is its most recognised form. It is described by Hauser and Clausing [9]: The house of quality is a kind of conceptual map that provides the means for interfunctional planning and communications. There are nearly as many forms of the HOQ as there have been applications and it is this adaptability to the needs of a particular project or user group which is one of its strengths. Fig. 3 illustrates the general format used for the HOQ matrix, which is made up from six major
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112
107
3.2. The prioritisation matrix
Fig. 4. Prioritisation matrix with weighted rows and cells.
components: 1. 2. 3. 4. 5. 6.
Customer requirements (Whats) Technical requirements (Hows) Planning matrix Interrelationship matrix Technical correlation matrix Priorities, benchmarks and targets for technical descriptors
The HOQ matrix translates the needs expressed by a customer into the design targets of a proposed new product. This is e!ected by entering weightings into the central Interrelationship Matrix which represent the scale of the perceived relationship between each customer and technical requirement. These weightings are combined with measures of the relative importance of each customer requirement to calculate a priority for each of the technical requirements in terms of satisfying customers. The remainder of the HOQ contains competitive benchmarking and market analysis information which is also relevant to the setting of the target design values. A simpli"ed form of the HOQ matrix was utilised in this work. The Technical Correlations and Planning matrices were removed and only the prioritised requirements row included at the base. The resulting matrix format is referred to as a prioritisation matrix.
In a prioritisation matrix, as in the HOQ, the interrelationships expressed in individual cells are assigned a value. Numbers or symbols are commonly used in QFD to represent the strength of such an interrelationship. Weightings are also used to illustrate the relative importance of items within the list of customer requirements. An overall value for the `importance of the strengtha of a particular interrelationship can then be numerically calculated. This involves multiplying the weighting in each matrix cell by the relative importance of the row item with which it is associated (see Fig. 4). These values can then summed down a column to give a measure of the `importance of the strengthsa of the overall relationships between a particular column item and the entire list of row items (see Fig. 4). In this manner the matrix allows the items of one list to be ranked based upon their relationships with the items of another list.
4. Building the technology evaluation tool This basic matrix tool was developed in order to focus upon the interrelationships between a product's characteristics (customer attributes in the HOQ) and the characteristics of the thixoforming process (engineering characteristics in the HOQ). The weightings of these interrelationships were combined in a manner similar to that used in a prioritisation matrix to illustrate the relative importance of the characteristics of thixoforming. The sum of these values was used to give a product a score for its overall suitability for thixoforming (see Fig. 5). The "nal tool was referred to as a multi-attribute matrix analysis. Nine characteristics of the thixoforming process were selected based upon the expertise held by the Thixoforming Research Group and drawing on the bene"ts and drawbacks documented in the literature. The eight product characteristics (perhaps better described as product issues) were selected as being relevant to the thixoforming process and su$ciently independent to avoid the repetition of analysis issues. (the chosen characteristics are shown in Fig. 5)
108
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112
3 Medium relationship between the product and thixoforming characteristics 1 Weak relationship between the product and thixoforming characteristics 0 No relationship between the product and thixoforming characteristics (cells generally left blank)
Fig. 5. Proposed multi-attribute matrix analysis technology evaluation tool.
The latter three thixoforming process characteristics (raw material, process development & skills/ wages) are considered to currently be the major drawbacks of the technology. The weightings applied to these characteristics were therefore subtracted rather than added in the calculation of the overall product suitability score. To standardise the characterisation of a product under consideration, each of the eight product characteristics were allocated three settings. For example Critical/Important/Unimportant, High/Small/No, Complex/Medium/Simple and Long/Medium/Short depending upon the nature of the characteristic. Before the matrix relationships could be discussed, it was important to be clear about the de"nition of each characteristic and each of its settings (see Table 1). The next stage was to establish the values for each of the interrelationships that could be displayed within the matrix. These were selected from a four-point scale: 5 Strong relationship between the product and thixoforming characteristics
The issues that were relevant to each combination of product characteristic setting and thixoforming characteristic were generated through a number of repeated discussions and individual reviews carried out by the members of the Thixoforming Research Group. These were documented and then formed the basis for the weighting values to be allocated to each cell. Once again these "gures were subject to several discussions and reviewed to ensure all possible perspectives were included. Table 2 illustrates the transition from a combination of characteristics, to the relevant issues and so into cell weightings. The completed weighting evaluations were then developed into a tool within the ThixoCost software package (illustrated in Fig. 6). This was written in Visual Basic 4.0 and allows a user to rapidly characterise their product using a scrolling menu for each of the product characteristics. As each characteristic setting is chosen the interrelationship values are automatically entered into the cells of the matrix. A second set of scrolling menus provides the user with a facility to enter the relative importance of each product characteristic. These allow entries on a scale of 1}5. To ensure comparable scoring between users (i.e. avoid the situation where a user rates all characteristics with the highest or lowest level of importance) a system for the allocation of importance scores is recommended, as detailed in Table 3 (an application of this importance scoring system is shown in Fig. 6). The system does not have to be strictly adhered to but the use of the full range of importance scores is necessary if the calculated product suitability score is to be analysed correctly by the software. Once the user has selected a setting and importance weighting for each product characteristic, the software calculates a `prioritya for each characteristic of thixoforming in a similar manner to a
Setting 1
Critical: Minimising the weight of the component is a critical issue
Critical: The strength of the component is a critical issue
Complex: The geometry of the component is highly complex. e.g. fuel rail
Critical: The meeting of demanding material property and dimensional tolerances is a critical issue
High: The market will allow a premium price to be charged for a thixoformed product
Long: The lead time between receiving an order and despatching a product is not important and is greater than 3 months
High: A customised product is produced in small batches, unique to each customer
High: In the current process there are more than 3 machining/"nishing/reinforcing operations which could be removed through thixoforming
Product characteristic
Weight
Strength
Geometry
Tolerances
Price premium
Lead Time
Flexibility
Finishing operations
Table 1 De"nitions of product characteristic levels
Medium: There are 1 or 2 machining/"nishing/ reinforcing operations in the current process which could be removed through thixoforming
Medium: A medium sized range of standard products is produced
No: There are no existing process steps which could be removed through thixoforming
Low: A single standard product is produced in large batches
Short: The lead time between receiving an order and despatching a product is critical and is less than 1 month
No: The market will allow no premium to be charged for a thixoformed product
Medium: The market will allow a small premium price to be charged for a thixoformed product Medium: The lead time between receiving an order and despatching a product is between 3 months and 1 month
Unimportant: Meeting dimensional and material property tolerances is not important
Simple: The geometry of the component is basic. e.g. a chisel
Unimportant: The strength of the component is not an important issue
Unimportant: The weight of the component is not important
Setting 3
Important: Material property and dimensional tolerances are important but not critical
Medium: The geometry of a component is of medium complexity
Important: The strength of the component is important but not critical
Important: The weight of the component is important but not critical
Setting 2
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112 109
110
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112
Table 2 Example evaluation of interrelationship values Characteristics
Relevant issues
Lead time & process development
Interrelationship weightings for each product characteristic settings
The time constraints required by short lead times will demand high process development e!ort.
Strength and mechanical The mechanical integrity achieved in thixoformed components is integrity an advantage in high strength applications, in material savings and heat treatments.
Table 3 Recommended relative importance scoring Relative importance score
Application of score
5 4 3 2 1
The The The The The
single most important characteristic next two most important characteristics two characteristics of medium importance next two less important characteristics single least important characteristic
prioritisation matrix. The magnitude of these column totals illustrate those bene"ts and drawbacks which are of the greatest and least impact in the considered potential implementation of the techno-
Long
Medium
Short
1
3
5
Critical
Important
Unimportant
5
3
1
logy. This information is useful in concentrating any "nancial studies on appraising the use of the technology as well as focusing the development necessary in implementing the process (in the Fuel Rail example shown in Fig. 6, the near net shape capability of thixoforming is highlighted as the most important bene"t by its highest characteristic score. Process development is also highlighted as being the most serious drawback by its lowest score). The overall product suitability score (referred to as the `Evaluation Scorea in Fig. 6) is the sum of these `prioritiesa, with the three drawback column totals being attributed negative values in this calculation. The interpretation of this score is made by the software using the scale shown in Table 4. These values were established by the application of the tool to a range of existing commercial
Table 4 Interpretation of product evaluation scores Product evaluation score
Software interpretation
Greater than 100
The product under consideration is suitable for the application of thixoforming technology and a detailed cost bene"t analysis should now be undertaken.
Between 100 and 80
The score allocated to this product by the multi-attribute matrix analysis shows it to be a borderline case. You may wish to repeat this evaluation paying close attention to the characteristic settings and importance weightings chosen before deciding whether to pursue a further cost bene"t analysis.
Less than 80
The product under consideration is presently unsuited to the application of thixoforming.
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112 Table 5 The multi-attribute matrix analysis applied to existing thixoformed products Existing thixoformed product
Multi-attribute matrix analysis evaluation score
Fuel rail (Weber}Italy) Steering link (Buhler}Switzerland) Automobile wheel (Alumax}US) Suspension subassembly (Stampal}Italy)
117 105 109 125
111
a series of tests were carried out. A set of 20 products were selected that were not used in the development of the suitability score interpretation model. These products included commercially produced thixoformed products and also products under consideration for application of the process. The software output was then compared with the opinions of the Thixoforming Group. Table 5 illustrates example values produced by the tool for existing thixoformed products.
6. Further applications thixoformed products, potential products and those product types deemed unsuitable to the process as identi"ed and characterised by the Thixoforming Group.
5. Validation of the evaluation tool In order to verify the validity of the recommendations made by the technology evaluation tool
The validity of the evaluation tool was proven for the range of products with which it was tested. In addition to successfully di!erentiating between those products that are suitable and unsuitable for the process, it e!ectively highlighted the most important process characteristics for consideration in any further analysis and in the "nal development of the process (e.g. the near net shape and process development characteristics in the Fuel Rail
Fig. 6. The multi-attribute matrix analysis screen from ThixoCost.
112
A. Lowe et al. / Int. J. Production Economics 67 (2000) 103}112
example). Since its development, the multi-attribute matrix analysis has been applied to the products of two UK research partner companies and was successfully demonstrated at a thixoforming conference applied to product characteristics provided by delegates. The development of this tool drew upon a pool of expertise in the technology and a relatively small set of product data used in devising the score interpretation model. It e!ectively utilises such expertise to translate the characteristics of a product into a recommendation for its further consideration for application to a particular technology.
7. Conclusions The multi-attribute matrix analysis tool has been developed from the techniques of QFD and applied to the evaluation of potential products for an innovative metal forming process. This technology evaluation tool does not substitute for a comprehensive economic analysis. The level of subjectivity necessary in allocating characteristic settings, importance weightings and interrelationship scores, if such a tool is to be applied generically, means speci"c accuracy cannot be achieved in each case. Instead the value of such a tool is in the focusing of a subsequent economic analysis or removing the need for it altogether. From this work, it would seem that a similar approach could be as e!ectively applied to the analysis of other new innovative technology applications. High-speed machining for example has both bene"ts (e.g. rapid production cycle, thinner webs, etc.) and drawbacks (e.g. high cost of machines, lubricants and tooling) which the tool could evaluate against relevant product characteristics (e.g. geometry, material, machinability, etc.).
Acknowledgements We are grateful to the Engineering and Physical Sciences Research Council for their sponsorship of the work carried out here under the aegis of the Innovative Manufacturing Initiative (Project No. IMI/RP/01/065).
References [1] P.L. Primrose, Investment in Manufacturing Technology, Chapman and Hall, London, 1991, p. 49. [2] D.B. Spencer, R. Mehrabian, M.C. Flemings, Rheological behaviour of Sn-15 Pct Pb in the crystallization range, Metallurgical Transactions 3 (1972) 1925}1932. [3] A.R.A. McLelland, P.R.G. Anderson, H.V. Atkinson, D.H. Kirkwood, The application of ceramic moulds to semi-solid metal forming, Proceedings of the Fourth International Conference on Semi-Solid Processing of Alloys and Composites, The University of She$eld, 1996, pp. 274}277. [4] D.H. Kirkwood, Semisolid metal processing, International Materials Reviews 39 (5) (1994) 173}189. [5] G. Chiarmetta, Thixoforming of automobile components, Proceedings of the Fourth International Conference on Semi-Solid Processing of Alloys and Composites, The University of She$eld, June 1996, pp. 204-207. [6] M.P. Kenney, J.A. Courtois, R.D. Evans, G.M. Farrior, C.P. Kyonka, A.A. Koch, K.P. Young, Semisolid metal casting and forging, Metals Handbook, 9th Edition, ASM International, Metals Park, OH, 1988, p. 333. [7] M. Kogure, Y. Akao, Quality function deployment and CWQC in Japan, Quality Progress 16 (10) (1993) 25}29. [8] N. Singh, Systems approach to computer-integrated design and manufacturing, Wiley, Chichester, UK, 1996, p. 133. [9] J.R. Hauser, D. Clausing, The house of quality, Harvard Business Review (1988) 63. [10] L. Cohen, Quality function deployment: how to make QFD work for you, Addison-Wesley, Readings MA, 1995, p. 70.