Product Configurators Pave the Way for Mass Customization
Featured Authors - Joseph Neelamkavil & Helen Xie - August 21, 2003
Mass customization' is the buzzword of the current decade. Customers demand products with lower prices, higher quality and faster delivery; yet they also want products customized to match their unique needs. To meet those challenges, manufacturers need to make a paradigm shift and adapt their business model to be compatible with mass customization.
This new paradigm combines mass production's economies of scale with custom manufacturing's flexibility. It allows customers to select, order and receive a custom configured product, often choosing from a multitude of product options. A tool encapsulating this model the configurator is a piece of software that captures customer requirements as input, and then automatically generates custom configured products based on pre-defined product structure and design constraints, while at the same time exactly matching the user's unique needs.
Mass production of identical products the business model of many industries is no longer viable for many firms. Customers demand products with lower prices, higher quality and faster delivery; but they also want products customized to match their unique needs. Accordingly, manufacturers are adapting their business models to mass customization, which enables customers to select, order and receive custom-configured products tailored to their specific needs.
A key technology that enables the implementation of mass customization is product configuration. Mass customization means that customers can select, order and receive a specially configured product, often choosing from among hundreds of product options, to meet their specific needs, yet assuring no increase in price.
To implement mass customization successfully, manufacturers need to overcome several major challenges. The time taken to configure products manually is often prohibitive because of the huge number of combinations of different selections that need to be considered before arriving at a valid configuration. In addition, extensive training and expertise are needed in creating configurations of complex products. Further, there is always the possibility of making errors since the final product may involve consideration of thousands of configurable parts. Errors, obviously, can create major slips in schedule and lead to costly iterations in downstream. A product configurator that enables manufacturers to efficiently deliver customized products by automating product configuration processes is one of the key promising technologies in implementing mass customization.
In simple terms, the product configurator is just a software tool that captures customer's requirements as input, and then automatically generates a configured product exactly matching a customer's specific needs, based on pre-defined design constraints. The configuration task can be defined as designing a specific product using a set of pre-defined component types, while taking into account a set of well-defined restrictions on how the component types can be combined. That is, given customer requirements and built-in product descriptions, the configurator will first search from all possible product options and combinations within the restrictions imposed by design constraints, and then generate a valid product configuration exactly matching the customer's specific needs.
An engineer-to-order type of configuration is an extension of this such that, each component type is also associated with a pre-defined set of parameters, where each parameter has a predefined set of possible values to choose from in order to satisfy all constraints among those parameters. The configurator technology will help manufacturers improve their productivity by shortening lead times, by eliminating the possibility of order errors, and by reducing the need for training costs and expertise of the various design and service personnel.
Configurators How They Evolved
Configuration of technical systems has a long history as an expert system application. Digital's landmark R1/XCON system [McDermott, 1982] is often considered as the very first successful configuration system. Many other organizations followed and developed their own configuration expert systems. These systems use a programming paradigm known widely as the production rules, to provide dynamic and runtime decision-making that are essential for obtaining a valid configuration. They use a uniform mechanism for representing both domain knowledge and control strategy, and embed the knowledge about a single entity over several rules. This makes the knowledge maintenance task for large rule-based systems extremely difficult. To overcome the drawbacks associated with the rule-based systems, a generic, domain independent model for configuration tasks was suggested [Mittal and Frayman 1989] in the form of a constraint satisfaction problem (CSP). The configuration problem is defined by a finite set of variables, with each variable taking only certain values from a domain of finite set of possible values, guided by constraints that restrict the variable combinations and the variable values allowed in such combinations. The configuration task is to find a value for each variable from its domain in such a way that all the constraints are satisfied.
As mentioned earlier, two types of configuration tasks may be conceptualized in accomplishing a customized product: build-to-order and engineer-to-order. The build-to-order configuration typically uses a set of pre-defined component types while taking into account a set of well-defined restrictions on how the component types can be combined. The engineer-to-order configuration extends beyond the build-to-order configuration; here each component type is also associated with a pre-defined set of parameters, where each parameter has a predefined set of possible values. Currently, there is no effective search solution for solving complex engineer-to-order product configurations.
In recent years, with the introduction of e-applications to support business processes and a global trend on mass customization, commercial software vendors are providing configurator solutions, including consulting services, off-the-shelf products, and a combination of the two. Because of the diversity of products offered, user companies really need to understand what they are getting under the banner of configurator. This service for customized product design/manufacturing includes manual product configuration based on expert use of product catalogues and automated product configurators applicable to relatively simple products (example: configuring a computer).
Product catalogues provide a pre-defined but limited number of combinations for choice of products; but they often fail to meet a customer's special needs. Almost all such automated configurators are ineffective to perform technology-intensive configurations. The business models embedded in those systems are not rich and flexible enough to accommodate the many required product and business knowledge data that need to be compiled from product behaviour models, engineering formula, constraints and so on. The current commercial configurators in terms of their underlying technologies are briefly described below:
Procedure-based configurator: They embed business rules of configuration into the software packages and can perform configurations of simple products only. Customers narrow down their selection by choosing from the list of available options, certain options being provided dynamically. A familiar example is the configuration option available from several computer vendors. The selection features are such that the customers won't arrive at any conflicts and the built-in business rules guarantee a valid configuration. For this kind of system, any change in business rules involves software rewriting and testing prior to invoking the system. These procedure-based configurators are most suitable for sales functions with simple business logic. Typically, software consultant companies develop such configurators, often on a project basis.
Rule-based configurator: As described earlier, it consists of a general-purpose (inference) engine and a rule/knowledge base customized to a family of products. Here, the design knowledge representation and the derivation of a feasible design solution are tightly coupled, by allowing the domain knowledge to be intermingled with the control strategy (actions). A few commercial configurators have been developed using the rule-base approach; some even have been integrated with application software such as CAD and PDM. They accept user requirements as features and options, and can generate CAD models based on the selected configuration. Some configurators are also integrated with ERP and CRM packages, with consulting companies providing their own core configuration engine with turnkey solutions for configuring the products.
The Next Generation Configurator
As the engineering beneath a designed product becomes complex and/or when the product gets modified frequently, the creation, modification and maintenance of a rule-base becomes an impossible task. If there is any design change, the rule base needs to be changed accordingly and go through extensive testing. For procedure-based systems, embedded business rules need to be revised extensively. This means that the procedure- and/or rule-base option can't go very far in today's complex and dynamic product development and manufacturing environment. The use of configurators that are built based on a constraint satisfaction problem is most promising in such instances. A feature comparison of various types configurators is given in a table at the end.
A novel engineer-to-order Product Configurator to configure complex engineering products is being developed at the National Research Council Canada's Integrated Manufacturing Technologies Institute (NRC-IMTI) in London, Ontario. It consists of a configuration engine, a product-modeling tool, and a Web presentation framework. The software tool accepts user requirements over the Internet, and based on embedded product definitions (consisting of mathematical formulae, design parameters, constraints, safety requirements, etc.), it generates custom-engineered products automatically to meet user needs. The Configurator is designed to separate the product domain knowledge from program execution strategy, and it incorporates efficient algorithms to find a feasible engineering solution after an extensive search of the pertinent design space. It allows the generation of alternate configurations that may be compared/weighed with respect to certain optimization criteria (example: low cost) before selecting the final configurations by a customer. Our configurator has the following main features:
- It incorporates generic approaches to solving configuration problems so that it can be used for a variety of customizable products; the approaches include a generic configuration engine and a Web presentation framework;
- It embeds effective search strategies and search algorithms suitable for engineer-to-order configuration problems;
- It includes a flexible mechanism for modeling and maintaining product definition, so that individual product definitions can be integrated with the generic configuration engine.
The NRC engineer-to-order configuration tool has progressed through several steps including initial studies, system development, algorithm design, implementation and system evaluation. Good understanding of configuration problems was developed through both theoretical research and an industrial case study. Several constraint-solving methods were explored and some were selected for experiment on target configuration problems. A web-based configuration system was then built, for which a system architecture and web-based presentation framework were designed using multiple tier practice and model-view-control design patterns. Effective, search algorithms were developed, implemented and evaluated for the product configuration problems. When fully developed, the configurator is planned to link dynamically with companywide data repositories to facilitate frequently changing product data. Currently, we are looking for collaborative partnership for further development and for the validation of the configurator via real life industrial data.
|Complexity of product supported
|Can perform automatic search for multiple possible solutions
||Rules for complex products are hard coded
||Cannot backtrack. If constraints are not satisfied at end, human interaction is required |
|Development efforts in terms of its reusability
||Configuration engine reusable
||Interference engine reusable
|Able to support different products
||Able to support different products
||Need to be written for each product|
|Models reflect product constraints involving engineering parameters
||Difficult to extract rules; requires expertise of knowledge engineers
|Need only to modify product's properties and associated constraints
||Need to modify rules and test their integrities to make any product modification
||Need to modify program for product modification of any kind|
About the Authors
Joseph Neelamkavil is a Senior Research Officer at the Integrated Manufacturing Technology Institute (IMTI) of the National Research Council of Canada (NRC). As a member of IMTI's concurrent engineering team, he is actively involved in the creation of tools that support engineering designers during the early stages of product development. His research interests encompass conducting research on techniques for design representations, knowledge capture and designs reuse to accelerate product development process. He is a registered Professional Engineer in the Province of Ontario, Canada. He received his B.Sc. in Mechanical Engineering from India, M.Sc. in Production Engineering from Trinity College, Dublin, Ireland, and M.A.Sc in Mechanical Engineering from the University of Toronto, Canada.
Helen Xie is a Research Officer at the Integrated Manufacturing Technologies Institute, National Research Council of Canada. She has been actively involved in the area of product data representation, product configuration management, and production planning and scheduling. Her current research interests are in application of artificial intelligence and Web technologies on product configuration, conceptual design, and distributed and collaborative design. She received her M.Sc. in Mechanical Engineering from the University of Saskatchewan in Canada, M.Sc. from Beijing University of Aeronautics and Astronautics, and B.Sc. from Tianjin University in China.