Product configuration analysiswith design structure matrix

P.T. HeloLogistics Systems Research Group, University of Vaasa, Vaasa, Finland

Abstract

Purpose – Product configurator is a sales and production-planning tool that helps to transformcustomer requirements into bills-of-materials, lists of features and cost estimations. The purpose ofthis paper is to introduce a method of how to analyse sales configuration models by using a designstructure matrix (DSM) tool. By applying the DSM techniques, the sales configuration managers maysequence the product configuration questions and organize the connection to production.

Design/methodology/approach – First, the paper explains a sales configuration system structurefrom published academic and non-academic works. These sources employ both theoretical andpractical views on the topic of computer-based sales expert systems. Second, the paper demonstratesan application of using DSM for configuration modelling.

Findings – The current sales configuration approaches include constraint-based, rules-based, andobject-oriented approaches. Product description methods vary, but the general problem remains thesame: the configuration process should be designed in such a way that customer selections do notaffect the previous selections. From the user point of view, answering the questions should be smoothand fast. In turn this will lead to the growing importance of building more effective productconfiguration models. DSM offers a systematic way to organise customer interface in salesconfiguration systems.

Research limitations/implications – This paper analyses how DSM could help in planningproduct configuration modelling. Comparison of different sequences is presented. The examples usedare hypothetical, but illustrate the suitability of DSM analysis. Companies are trying to establish easilyconfigured product models, which are fast, flexible and cost-effective for adjustments andmodifications. Use of DSM may help in the roll-out of sales configuration projects. DSM may also beused as a quick view to represent the complexity of product configurability. The future needs forconfiguration tools will be focused towards product model management from the technical limitationsof different data storage approaches.

Practical implications – Configurator software creates product variants, which are logicaldescriptions of physical products. Variants have parameters which describe the customer-madeselections. The parameter selections may have interconnections between the choices. Someselections may affect further selections and some combinations may not be allowed forincompatibility, cost or safety reasons. There are several commercial software packages availablefor creating product configurations. Product description methods vary, but the general problemremains the same: the configuration process should be designed in such a way that customerselections do not affect the previous selections. Answering the questions should be smooth andfast. Configuration of complex products, for instance, airplanes, may include several sub-systemsand have various loops within the quotation process. The use of DSM may help in the roll-out ofsales configuration projects. DSM may also be used as a quick view to represent the complexity ofproduct configurability.

Originality/value – The paper helps both researchers and practitioners to obtain a clearer view onthe development of sales configuration systems and the potential of systematic DSM-based productmodel analysis.

Keywords Sales information, Configuration management, Software tools, Manufacturing systems,Integration, Design

Paper type General review

The current issue and full text archive of this journal is available at

www.emeraldinsight.com/0263-5577.htm

Productconfiguration

analysis

997

Industrial Management & DataSystems

Vol. 106 No. 7, 2006pp. 997-1011

q Emerald Group Publishing Limited0263-5577

DOI 10.1108/02635570610688896

1. IntroductionConfigurators are software tools that allow a user to define a product that meets certaingiven criteria by combining a number of parts, features or functions. In practice, aconfigurator is a software application that is used in industry to enable quotation,marketing, and rapid manufacturing of customised products. Industries are beingforced to move towards increasingly customised and individual products. In order todo this cost effectively, the engineering phase needs to be accelerated.

The use of configurators decreases the engineering work required in the sales phase,but also empowers the product development of platform derivative products (Franke,1998). Configurators may be used for structural analysis, dimensioning or designverification purposes. Manufacturing may use configurators for downloading the rightsoftware on customer tailored embedded systems. After sales may check the soldconfiguration and verify the compatibility of new installations or spare parts with thecurrent system.

Product configurators capture a lot of information related to engineering andproduct pricing (Soininen et al., 1998). The relationships between parameters may bevery complicated and documenting these takes a lot of time and effort. Implementationof project requires mapping the elements and their interactions. Creating a goodconfiguration depends on how good the map is. The map describes the selection routefrom the requirements to the outcomes of configuration.

Configuration of complex products may include several sub-systems and havevarious loops within the quotation process. In this paper, we analyse how designstructure matrix (DSM) could help planning product configuration modelling. Theremainder of this paper is as follows: firstly, basic types of product architectures areintroduced. Secondly, product-modelling techniques for configurations are reviewedand feature-based configuration is presented with some examples. Thereafter, the useof design structure matrices for configuration projects is discussed. Finally, the someconclusions are drawn from DSM applications.

2. Product architecturesProduct architecture is the way that functions are mapped to form a product (Ulrich,1995). There are two generic types of product architectures, which are excluding eachother: these are integrated architecture and modular architecture. In the case ofmodular architecture, each function is connected to a corresponding part, while in thecase of functional architecture; a part may have more than one function. A special caseof integrated architecture is adjustable architecture, which practically means thatthere are controllable parameters that can affect the functionality of the product. Forinstance, a PC computer is a flexible platform for various applications: the softwarepart adjusts the functionality of the machine. Products of modular architecture can becombinations of the following types (Ulrich, 1995):

. Slot. Components are connected with each other with an interface, which is notnecessarily interchangeable with other components. A car radio is a goodexample of this: various radios can be installed in the slot. However, there are noother components that could be mounted in the same slot.

. Bus. Bus architecture is based on a standardized interface, which is common tovarious parts. For example, the PCI bus in personal computers allows theaddition of modems, network cards, video card, etc. on the same type of rack.

IMDS106,7

998

. Sectional. In sectional architecture, all interfaces are similar and there is nocommon part where other parts would connect as in bus architecture. Forinstance, SCSI interfaces in computers are connected to each other with a cable.CD-drives, hard disks and other equipment can be in any sequence.

Modular architectures allow the flexibility for multi-purpose use. The cost efficiencyprovided by a widely used platform is enhanced with modular structures (Meyer andLehnerd, 1997; Meyer and Utterback, 1993). All modular products are combinationsof these structures. However, there are also other ways to classify modular structures(Yu et al., 1998). All modular architectures are connected to variant bill-of-materials(BOM) when considering the modelling part of the process.

From the configuration point of view the product architecture describes how itemsin BOM are connected with each other. A list of materials is a typical outcome of aconfiguration process. The rules are given in the form of allowing and constrainingconnections between items in some product configurators. These connections are verysimilar to product architecture decisions.

3. Product modelling for variant productsMost of the product configurators are based on modelling variable BOM. This genericBOM approach is used in many industries as a practical approach for implementationof mass customization (Jiao et al., 2000). A product variant is a logical description of aphysical product. All derivative products can be described by product variants.Variants have parameters, which make them different from other variants. Forinstance, parameters for a car could include colour or engine type. There are severalcommercial software packages available for creating product configurations. Thetechnical approaches for describing the rules are:

. constraint based approach (Van Veen, 1992; Mailharro, 1998);

. rules based approach; and

. object-oriented approach (Ghatate et al., 2000).

Product description methods vary, but the general problem remains the same: theconfiguration process should be designed in a way that customer selections do notaffect the previous selections. There are various techniques for describing productvariants. In this paper, we will briefly review:

. BOM table;

. logical tree; and

. VAR-variant generator (Van Veen, 1992, p. 82).

3.1 Bill-of-materials tableBOM table is a tool for maintaining modular BOM. End products of the same kind canbe described by individual sequence of parameter values. According to this technique,the table of BOMs consists of three major parameter types:

(1) Basic parameters. The same for all product variants {Lamp}.

(2) Variant parameter. Selection of several choices {Blue shade, black shade, whiteshade}

(3) Option parameters. Choice type – yes/no {Shank}

Productconfiguration

analysis

999

Basic parameters describe what basic structure will be used. Variant parametersdescribe the selections within the basic structure. Option parameters are add-ons toany product. The main difference between variant and option is that in order to specifya product a value must be given for variant parameter, while the option value may benil. Variant parameter and option parameters are not depending on each other.

This type of configuration approach is limited. BOM table is able to support onlyone-level BOM. In other words, there are no configurable sub-sets available. The lowerlevels include a standardised structure, which cannot be effected by the higher-levelparameters. The use of this description method is suitable for the purpose where theproduct is quite simple in terms of structure and the number of parameters is low. Themain benefit of this approach is that it is able to handle dependencies betweenparameters and structural entities. The approach is also rather simple to learn.

3.2 Logical treeLogical BOM-tree is another configuration approach. Logical diagram is a tree-baseddescription of BOM. This approach is beneficial when existing structures can be usedin creating new variations. In this method, logical nodes originating from Booleanalgebra are added into the structure. These nodes can be the following ones:

. Product nodes (parts, components).

. AND-nodes.

. XOR-nodes.

. Empty node (nil).

Product nodes are normal components, parts and items used in all traditional BOMs.AND-nodes show which lower level trees are required for higher levels. In traditionalMRP-II software, pseudo-structures describe the same structure. XOR-nodes describethe structures, where one must select one of the choices. An empty node does not haveany lower structures. It is used in cases where a XOR-node requires a lower levelstructure. Figure 1 shows an example of using logical trees describing a table lamp.

Logical BOM-trees allow only item selection related choices. Use of descriptivefeatures that may affect several items are not allowed. By using this approach it is notpossible to use compatibility parameters such as “colour-red” that would change thecolour of all the parts that can be painted.

3.3 VAR-variant generatorA more advanced configuration approach is the VAR-variant generator. According tothis technique, all products can be described when the following conditions are met:

. all parameters are independent of each other; and

. product variants can be unambiguous if the parameter selection is madeaccordingly.

BOM is described in a way that the main product X has specific lower level structures.The lower level structure is not a standard item, but a set of parts, which have one ormore values. These sets are called clusters. Product X can be described by selecting oneexpression of each cluster. The expression is selected according to a rule, which definesthe lower level parts (e.g. four doors in a car defines many lower level sub-structures).

IMDS106,7

1000

VAR-variant generator is able to handle not only single level BOM, but multi-levelBOMs as well (Van Veen, 1992, p. 88).

VAR-variant generator is based on relational database and can be implementedwith about 15 database tables. The approach is flexible for complex product models,but on the other hand maintaining models takes a longer time to learn. The drawbackof VAR-variant generator is that it is not able to handle complicated connections suchas equations or function based rules (e.g. wire needed ¼ 3*pi*r pwrð2ÞÞ:

4. Feature based configurationMany product configurators are based on a relation database model which restricts theflexibility of use. New object-oriented techniques allow complex relationships whichare not limited to analysis of BOM (Simons et al., 2002). Feature-based configuration isan approach where a distinction between product features (speed, type, size, etc.) anditems (part1, part2, part3) is made. The product architecture is a different issuecompared to the list of features. By using feature-based configurator the user may giveinputs such as:

. the product should cost less than $30,000; and

. the speed should be at least 150 km/h.

These parameters affect many choices and narrow the number of selections. Byfine-tuning some more detailed features the BOM can be created.

4.1 ConceptsThe product variants are defined in feature-based configuration as product families. Eachfamily is an independent data item; it does not share any information with other families.The family is a top-level data item; it includes all information valid for that family.

Figure 1.Table lamp as a logical

tree

Productconfiguration

analysis

1001

The configurable data is arranged for a family in tabs, for example, feature groups. Thetabs include the same type of information that the user would like to see as at the samescreen. If the product is very simple there is need for just one tab. In most of the cases it isreasonable to group the selections in tabs that include 4-6 choices. The following picturegives an example of a product family that is defined to configure a motorbike. Theconfigurator engine shows the structure of the family and the user can change the order oftabs or parameters and organize the data so that it is most usable (Figure 2):

. Product family. A group of similar occurrences is called a product family.A product family is always on the highest level of BOM and thus cannot be acomponent for another product.

. Parameter. Parameter is in relation with only product family, not to anindividual end-product. A parameter may have values: colour, type.

. Parameter value. Value is an occurrence of property in parameters; for instance,values for parameter “colour”: black, blue, white. The value of parametersdescribes the parameter-value combinations that are allowed: colour-black,colour-white, type-standard, type-customised.

. Attributes. Attribute is a property connected to family level, parameter, value,item. An attribute tool can be used for describing compatibility betweenselections (110 V or 2), consumption of resource (e.g. these options will take1,000 W of power, the power supply can feed 1,200 W).

. Rules. Rules define allowed or denied combination of two or more attributes,parameter values, BOM items or variables.

A feature-based product configurator can translate the product model into configuratoruser interface. Figure 3 shows an example of feature-based configuration. Product modeldetails are input by the product model administrator. The user interface can begenerated automatically from features or tailored for custom web design. Finally, salesengineers and the purchaser of the product may see the result in their configurator tools.

4.2 The rules for the familyThe rules define how the values for each parameter are displayed. In a simple case theconfigurator is just used to manage the family structure and there is no need to makecomplicated rules. In that case both of the rules, for features and options, are allowed asdefault. The basic rule types are:

Figure 2.Examples of configurationstructure

IMDS106,7

1002

. deny a combination;

. allow a combination; and

. set value-based on a selection.

The setup of the rules depends on the number of restrictions the family might have.In case most of the features are compatible, allowing combinations is the mostappropriate selection. If there is a situation where there are a lot of rules between theparameters the denied model could be the case. Thus, if there are 200,000 variants andjust a few rules, one should use a deny type of model. And if there are hundreds ofpossible variants and huge number of non-compatible constraints between theparameters, then one should use the allowing type of model.

There may be two types of parameters: features and options. This enables you togroup the parameter to a group where there are a lot of correlations between the rulesand to a group where almost any option is valid. By making this analysis the productmodeller may decrease the number of required rules significantly.

The features are used when there are rules that define that, e.g. there should bespecific values for A and B to enable C to have certain values. In that case, C does not

Figure 3.Example of sales

configurator user interfacegenerated from product

model

Productconfiguration

analysis

1003

have any values before the engine administrator allows the values for C to be displayedfor the specific combination of the rules of A and B. Typically in this case the featuresuse the denied model and only the values that are especially enabled are displayed. Ifthere is a case where parameter C has hundreds of values and only a few of them arepossible on a specific combination of A and B, then it is easier to disable all the choicesand enable only the few that are available. The options are typically parameters thatcan be used for every parameter combination. In case of the options you deny a specificcombination of parameter values.

Figure 4 shows the usage of features and options. All the possible values for the firstfeature A are available. But only after selecting a1, a2, or a3 b1, b3 and b4 are available.Then a combination of {(a1, b1), (a2, b2), . . . ,(a3, b3)} enables the values c1, c2, c3, c4, c5and c6. In the engine we call the blue enables sources and green results targets. So thereis no value for parameter H if no predecessor has allowed values for the parameter.

In case of options the engine allows the denial of specific combinations. Afterselecting value 3c in parameter 3 the user cannot select values 8d or 8j for parameter 8.There can be relations between features and options as well. The above example deniesafter selecting g2 the selection of 7a or 7c.

The product modeller should keep the rules as simple as possible and be sure tocheck out the rule type for the selected parameters. There is no sense to allowcombination 6a ! 9b, because parameter 9 is type option and all the values areallowed as default so the rule is useless. Feature-based configuration is a flexibleplatform for building product models. Building a map of interactions between differentparameters and items takes the majority of time in any product configuration project.

Figure 4.Example of configurationselection paths

IMDS106,7

1004

5. Design structure matrixDSM is an approach used for system analysis of decomposition and integration. DSMis a square matrix showing interactions flow between elements. The elements maypresent components of a product, phases of product development, people inorganisations or design parameters of a product (Browning, 2001; Smith and Eppinger,1997). This method introduced by Steward (1981) has been used in many industrialapplications. The examples mentioned in the literature include design of automobiles,design of jet engines and design of a microprocessor (Ulrich and Eppinger, 1994).

The main idea of DSM is to map the interactions into a matrix, where elements arelisted on the diagonal. Off-diagonal on the lower side represents information flow thatfeeds the following elements; the upper side indicates that the element feeds somethingback upstream. Figure 5 shows three basic types of relationship between configurationelements A and B. The first type is a “parallel” relationship, where elements notdepending on each other may be processed separately. The second type is “sequential”where element B depends from the input fed from element A. The last stereotype,“coupled” refers to a set where elements depend on each other. A change in element Bhas an effect back to A. This cyclic dependency is called “circuits”. In the configurationcontext, the elements may be features describing the products, items that make BOM,or related to variables which allow advanced calculation of configuration outcomes(for example, weight and power consumption).

The use of DSM with configurations is two-fold. In one-level configurations, themain objective is to sequence features in a way that first selection is the mostrestrictive question, and the last questions are not having effect on those parametersalready input. There is a greater need for DSM in multi-level configurations where thechallenge is to capture the possible feedback loops and try to formulizesub-configuration elements. There are several analytical approaches for organising

Figure 5.Stereotypes of system

configurations

Productconfiguration

analysis

1005

matrix. However, in many cases the matrix representation itself helps to re-organisethe feature questions in an applicable sequence. Rules describe the system elementinteractions. The possible combinations include interactions between:

. Features. Parameters and possible values, e.g. colour – green. Feature selectionsmay affect other features. For instance, all colours may not be available in everysize.

. BOM. Items that describe the product architectures.

. Variables. Outcomes of advanced calculations. E.g. weight of product, powerconsumption of configured system, sales price or cost structure.

Figure 6 shows an example of DSM showing the elements and their dependencies onthe configuration model. Complex configuration models that include configuration inseveral levels are challenged by the questions: how should one organise thesub-configurations? Which feature selections are common to all configurations, whichfeatures are connected to only one sub-system? Clustering of the DSM creates groups offeatures, BOMs and variables that should be handled together. This kind of entity is aclear sub-configuration. By using tearing and partitioning, the product modellersmooths the actual configuration process by removing unnecessary feedback loops.

In order to demonstrate how a traditional process flow diagram may be converted toa DSM matrix, consider the example shown in Figure 7. From the user point of view theconfiguration process starts from the top. The sequence is pre-defined and there areseveral incompatibilities between different selections, which could berecommendations or denying. Figure 8 shows the same process described as DSMpresentation. At the upper side of the diagonal, the possible feedback loops in theconfiguration process are visible and the feed-forward components are shown at thelower side. The sequence of process elements could be adjusted manually or by usingadvanced automatic clustering systems provided by many DSM software packages.The result of the analysis can show potential time delays in the order-fulfilment

Figure 6.Example of configurationDSM showing interactionsfor features, BOM andvariables (Problematics.DSM software screenshot)

IMDS106,7

1006

Figure 7.Example of selection

dependencies in computerserver configuration

Productconfiguration

analysis

1007

process from the sales engineering point of view and suggest improvements for theproduct architecture as well.

6. ConclusionsThe configuration projects are very important for today’s companies achievingimproved response. There is an obvious need to integrate the intra-organisational datasystems such as sales and production (Kumar and Meade, 2002), production planningand purchasing (Rantala and Hilmola, 2005) as well as inter-organisational systemssuch as supplier relationships and logistics (Tarn et al., 2002; Helo and Szekely, 2005).The examples from high clock speed industries, such as electronics manufacturing,have shown that the key issue in successful production management is to apply thenew technologies for processes as well (Helo, 2004). Seamless integration betweenproduct engineering and supply chain management may be achieved by usingautomated configuration systems. For this reason, systematic approaches are neededto help managers build configuration processes. This paper has contributed onsuggesting the application of DSM methodology for designing this integration. Thesemethods may be used when companies are moving towards flexible structures andagile virtual environments (Ma and Davidrajuh, 2005).

Feature-based configuration and other object-oriented techniques allow greatflexibility in creating product models. Because of these advances in softwareengineering, the challenge of building models is no more a technical issue but an issueof managing the configuration process. The process may incorporate differentcompanies and logistics operations such as production planning and supply chainmanagement (Helo and Szekely, 2005; Hsu and Chen, 2004). Configuration is typicallyrelated to sales, but it has important links to engineering, too, such as productdesign. The nature of selling technical products is iterative, several reviews arerequired before the final configuration is frozen. The key requirements for a modernconfigurator are:

Figure 8.DSM example ofconfiguration system

IMDS106,7

1008

. flexible modelling features;

. smooth configuration process with minimum number of feedbacks; and

. easy to learn product modelling.

These requirements have a direct connection to strategic objectives such as profitablevariety management. The companies applying sales configuration tools are trying toestablish easily configured product models, which are fast, flexible and cost effectivefor adjustments and modifications. Use of DSM may help in roll out of salesconfiguration projects. DSM may also be used as a quick view to represent thecomplexity of product configurability.

An analytical approach such as DSM should provide a more quantitative basis formass customization and product design strategies. The idea of mass customisation isthat a factory is able to produce one-of-a-kind products with high quality and fastdelivery with a low cost of mass production. The drivers enabling this are newtechnological manufacturing applications and advances in designing procedures.According to Pine (1993):

At its core is a tremendous increase in variety and customisation without a correspondingincrease in costs. At its limit, it is the mass production of individually customized goods andservices.

Use of a product configurator allows maintaining an extensive product variety withoutremarkable reductions in productivity. Since, the beginning of the 1990s various masscustomisation applications have been presented (Anderson, 1997; Pine, 1993).Design-for-manufacturing has become an important research area in productdevelopment as well as the use of product platforms. Despite the innovations,manufacturers are still balancing between time-to-market, product variety and unit-costs(Tseng and Jiao, 1998). Tools for assessing the trade-off between these parameters arerequired. In practice, the challenges are related to the management of the following issues(Tseng and Jiao, 1998):

. measurement and implications of reusability/commonality of parts and products(Jiao and Tseng, 2000);

. designing product platforms for cost-effective customisation; and

. integrated product development for managing the during the productlife-cycle.

Contemporary mass-customisation tools such as product configurators are typicallybuilt based on variable BOM, which are able to deal with technical constraints given inrule tables. Product modelling techniques have been developed in engineering designand industrial software design. The aim in engineering design has been to introducetools for modular architectures. One of the major challenges has been flexible designsfor manufacturing. Typically, this is derived by assumed market information,technological constraints and manufacturing related constraints (Tseng and Jiao, 1998;Yu et al., 1998).

DSM is a powerful approach providing help for product model builders to make amap of interactions between features, BOM items, and design variables. By usingthis kind of configuration sequence analysis, the end-user of the sales configurator

Productconfiguration

analysis

1009

should benefit from less complex user interfaces. Taking the production andlogistics into account in product design is a strategic insight. This results in morepowerful automated sales tools for industry and possibility for true masscustomization.

References

Anderson, D.M. (1997), Agile Product Development for Mass Customization: How to Develop andDeliver Products for Mass Customization, Niche Markets, JIT, Build-to-Order and FlexibleManufacturing, Irwin Professional Publishing, Chicago, IL.

Browning, T.R. (2001), “Applying the design structure matrix to system decomposition andintegration problems: a review and new directions”, IEEE Transactions on EngineeringManagement, Vol. 48 No. 3, pp. 292-306.

Franke, D. (1998), “Configuration research and commercial solutions”, AI EDAM (ArtificialIntelligence for Engineering Design, Analysis and Manufacturing), Special Issue onConfiguration Design, Vol. 12 No. 4, pp. 295-300.

Ghatate, B., Liemandt, J. and Price, A. (2000), “Flash configuration cache”, US Patent US6115547.

Helo, P. (2004), “Managing agility and productivity in the electronics industry”, IndustrialManagement & Data Systems, Vol. 104 No. 7, pp. 567-77.

Helo, P. and Szekely, B. (2005), “Logistics information systems: an analysis of software solutionsfor supply chain co-ordination”, Industrial Management & Data Systems, Vol. 105 No. 1,pp. 5-18.

Hsu, L-L. and Chen, M. (2004), “Impacts of ERP systems on the integrated-interactionperformance of manufacturing and marketing”, Industrial Management & Data Systems,Vol. 104 No. 1, pp. 42-55.

Jiao, J. and Tseng, M.M. (2000), “Understanding product family for mass customization bydeveloping commonality indices”, Journal of Engineering Design, Vol. 11 No. 3, pp. 225-43.

Jiao, J., Tseng, M.M., Ma, Q. and Zou, Y. (2000), “Generic bill of materials and operations forhigh-variety production management”, Concurrent Engineering: Research andApplication, Vol. 8 No. 4, pp. 297-322.

Kumar, S. and Meade, D. (2002), “Has MRP run its course? A review of contemporarydevelopments in planning systems”, Industrial Management & Data Systems, Vol. 102No. 8, pp. 453-62.

Ma, H. and Davidrajuh, R. (2005), “An iterative approach for distribution chain design in agilevirtual environment”, Industrial Management & Data Systems, Vol. 105 No. 6, pp. 815-34.

Mailharro, D. (1998), “A classification and constraint-based framework for configuration”,AI EDAM (Artificial Intelligence for Engineering Design, Analysis and Manufacturing),Special Issue on Configuration Design, Vol. 12 No. 4, pp. 383-97.

Meyer, M. and Lehnerd, A. (1997), The Power of Product Platforms, Building Value and CostLeadership, The Free Press, New York, NY.

Meyer, M. and Utterback, J. (1993), “The product family and the dynamics of core capability”,Sloan Management Review, Vol. 34, pp. 29-47.

Pine, B.J. (1993), Mass Customization, Harvard Business School Press, Boston, MA.

Rantala, L. and Hilmola, O-P. (2005), “From manual to automated purchasing: case: middle-sizedtelecom electronics manufacturing unit”, Industrial Management & Data Systems, Vol. 105No. 8, pp. 1053-69.

IMDS106,7

1010

Simons, P., Niemela, I. and Soininen, T. (2002), “Extending and implementing the stable modelsemantics”, Artificial Intelligence, Vol. 138 No. 2002, pp. 181-234.

Smith, R. and Eppinger, S.D. (1997), “Identifying controlling features of engineering designiteration”, Management Science, Vol. 43 No. 3, pp. 276-93.

Soininen, T., Tiihonen, J., Mannisto, T. and Sulonen, R. (1998), “Towards a general ontology ofconfiguration”, AI EDAM (Artificial Intelligence for Engineering Design, Analysis andManufacturing), Special Issue on Configuration Design, Vol. 12 No. 4, pp. 357-72.

Steward, D.V. (1981), “The design structure system: a method for managing the design ofcomplex systems”, IEEE Transactions on Engineering Management., Vol. 28 No. 3.

Tarn, M.J., Yen, D.C. and Beaumont, M. (2002), “Exploring the rationales for ERP and SCMintegration”, Industrial Management & Data Systems, Vol. 102 No. 1, pp. 26-34.

Tseng, M.M. and Jiao, J. (1998), “Concurrent design for mass customization”, Business ProcessManagement Journal, Vol. 4 No. 1, pp. 10-24.

Ulrich, K. (1995), “The role of product architecture in the manufacturing firm”, Research Policy,Vol. 24 No. 1995, pp. 419-40.

Ulrich, K. and Eppinger, S. (1994), Product Design and Development, McGraw-Hill, New York,NY.

Van Veen, E.A. (1992), Modelling Product Structures by Generic Bill-of-Materials, Elsevier SciencePublishers, Amsterdam.

Yu, J.S., Gonzalez-Zugasti, J.P. and Otto, K.N. (1998), “Product architecture definition based uponcustomer demands”, Proceedings of 1998 DETC. 13-16 September, Atlanta.

Further reading

Davis, S.M. (1987), Future Perfect, Addison-Wesley, Reading, MA.

Krishnan, V., Eppinger, S.D. and Whitney, D.E. (1997), “A model-based framework to overlapproduct development activities”, Management Science, Vol. 43 No. 4, pp. 437-51.

Corresponding authorP.T. Helo can be contacted at: petri.helo@uwasa.fi

Productconfiguration

analysis

1011

To purchase reprints of this article please e-mail: reprints@emeraldinsight.comOr visit our web site for further details: www.emeraldinsight.com/reprints

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.