Go to the table of contents.

Previous ...



8 REFLECTIONS


8.1 Related work

In this section a small number of references have been selected that show similarity with the proposed concept of configurable components as a whole or with some mechanisms or features. The first two references (Arndt et al., 2000 and Hami-Nobari & Blessing, 2005) show many similarities with the configurable component concept as a whole. The following two references (Wang & Roy, 2005 and Kormeier & Rudolph, 2005) highlight and emphasize a graph oriented approach to product representation. Though this may not have been clearly stated in the description of the configurable component concept above, a graph oriented approach is required to establish and manage the different mechanisms and elements of the configurable component concept. Last in this section are two examples of agent-oriented and self-organizing features that can be expected for product definition. These two examples are represented by the three references Wellman (1995), Kormeier & Rudolph (2005), and Kannengiesser & Gero, 2005.

Arndt et al. (2000) present software tool developed by DaimlerChrysler to provide support for vendors of transportation systems and their subsystems in early phases of the product design. The software tool is called system design for reusability (SDR). SDR serves as a design assistant. A block diagram editor allows describing systems, ports, and interfaces. Constraint propagation and constraint solving is seamlessly integrated in the modeling of the system structure. According to the authors SDR supports a reusable and modular product design by different modeling features: systems with ports; taxonomy with system types; interfaces with types; and modular constraints between parameters of systems. SDR stresses the principles of modular system design and requires a thorough and consistent segmentation of the system at all levels with a defined set of interfaces between modules. Modeling a product with SDR involves three tasks: (1) modeling the properties of a product and its sub-system, (2) modeling the product structure, and (3) modeling the design logic behind the product. The SDR system shows several similarities both on conceptual level and in terms of mechanisms and modeling elements with the configurable component concept described in this thesis. The SDR system and the configurable component concept have been developed separately and most likely from different starting points and different business objectives in mind. A more detailed comparison between the two would be needed in order to fully understand the similarities and differences.

Another work that shows similarities with the concept of configurable components is the proposed approach of effects-oriented description of variant-rich products (Hami-Nobari & Blessing, 2005). They state that the variety of products and parts is constantly changing due to a changing market situation. According to Hami-Nobari & Blessing (2005) the current approach based on a code-based description of variant rich products comes to its limitations. The complexity of describing variant rich products in the technical product documentation can hardly be managed any longer. This view of the limited capabilities of current product descriptions is the same as has been presented in this research. Hami-Nobari & Blessing (ibid.) present an effect oriented description approach as a means to achieve an efficient and understandable description of variant rich products and parts. In their approach the interdependencies between parts are expressed explicitly with regard to the effects and properties between the parts. The starting point for the effect-oriented description of variant-rich products is the definition of components and parts as technical objects. Technical objects characterize individual, clearly identifiable, material parts with clear physical boundaries. The physical boundaries of the parts are characterized by their geometries. At the boundaries parts can have interactions with their environments. The environment represents any physical surrounding of parts. Thus, at the boundaries of the parts there can be some exchange of energy, materials, or information. Parts and components can be identified by description of their attributes in detail. The attributes of parts mark their inherent and distinctive properties. In practice, attributes are often represented by object parameters. The attributes of parts can be distinguished in different categories (e.g., technical, commercial, geometrical, disposition, and organizational). Geometric attributes characterize the boundaries of parts and can therefore be relevant for the description of interdependencies between parts. Interdependencies between parts are based on technological, geometrical, sales, and legal restrictions. For an effect-oriented description of these restrictions the technical and geometrical effects (or reasons) of the restrictions are described and represented explicitly and in general as conditions or equations between the parts. Therefore the attributes of the parts are linked with each other by the equations. The described equations are integrated into the hierarchy of a developed class structure. The described equations can be inherited to lower relevant levels of the hierarchy. This way the equations are valid for all relevant part variants. A case study at Daimler-Chrysler is presented to illustrate and validate the proposed approach. The outcome of the case study is that, through the use of an effect oriented description approach, the efforts and the complexity in the technical product documentation can be significantly reduced. The approach to identify variant, design, and performance parameters is similar to the concept of configurable components. Another similarity is the approach to defined conditions, equations, and constraints that exercise control over the relationships among the defined parameters. An important difference seems to be that the concept of configurable component as defined in this thesis does not include any categorization of components to a class hierarchy with inheritance. The approach to build the product structure is also a notable difference.

Wang & Roy (2005) present a graph grammar based mechanical assembly family model. The generation of an assembly in the same family is modeled as the manipulation to the graph that represents the base assembly by applying graph production rules. The assembly variant generated is a graph with components as nodes and joints between components as edges.

Another example on the theme of graph grammars is provided by Kormeier & Rudolph (2005) who use graph representations of the design to identify recurrent patterns at various levels of abstraction. The nodes represent the basic entities, the building blocks of the design. The nodes act as placeholders for objects, functions, or concepts and contain the semantic information of a design. The nodes therefore represent the geometry, the physics and the functionality of the design parts. By connecting the nodes in the design graph through edges, the nodes are related to their context. The topological structure of the design is thereby captured through the graph structure. The parametric aspects of the design are coded into the nodes of the graph. An additional purpose of the edges is to exchange and relay information in the form of variables between the nodes in the graph. According to Kormeier & Rudolph (ibid.) the abstract design graph representation has proven to be a very flexible and transparent method to store and modify design information. Both the topology and parametrics of the design can efficiently be varied through an application of design rules.

Wellman (1995) writes that the current dramatic expansion in network technology, infrastructure, including the explosive growth of the internet, will lead to an inevitable proliferation of automated, distributed information agents. In my own view, trends towards automated solutions have, in many cases, historically proved to be too optimistic. However, the viewpoint provided is interesting and we could view information agents from a more human- and team-oriented perspective. Wellman (1995) continues and states that, with these advances, opportunities and incentives for the decentralization of design activities are rapidly emerging. As a consequence, computational support for distributed design collaboration presents a variety of new opportunities and challenges. In particular, the specialization of design expertise suggests a future where teams form ad hoc collaborations dynamically and flexibly, according to the most opportunistic connections.

An approach to product modeling using computational agents to represent product data is proposed by Gero & Kannengiesser (2003). The agents are situated which means that they produce all data representations in response to the specific need in a certain situation. The proposed agents share some similarity with configurable components. For example in terms of ability to adapt to changes in the environment and in terms of being rather self-sufficient with capable internal constructs. The agents are proposed to have a formal representation schema (Kannengiesser & Gero, 2005) based on function (F), behavior (B), and structure (S). The function describes the role of the agent for the observer (i.e. what the agent is here for). The behavior describes attributes derived or expected to be derived from the agentís structure, which includes how the agent acts under specified conditions (i.e. how the agent fulfils its function). The structure describes the agentís components and its relationships, which includes the agentís capabilities and knowledge (i.e. what the agent consists of). The FBS view provides a comprehensive set of constructs to model different kinds of agents and to allow the agent to treat the world uniformly at different levels in a conceptual hierarchy.


8.2 Future research


8.2.1 Functional requirements as a source for variation

In the context of the function-means tree approach (see section 4.2.2), one way to deal with variant functional requirements could be to introduce the concepts of mandatory versus optional requirements and complementary versus alternative requirements.

Mandatory functional requirements must be present in the design solution. Design solutions derived from an optional functional requirement may or may not be included in the product depending only on the choice made by the company itself regarding what product offerings to give the market or on the choice of the customer (i.e. there should be no design or production constraints preventing the choice of the option). Alternative functional requirements are never expected to be present simultaneously in a design solution. That is, the design solution must implement either one or the other FR; it is not a requirement to implement both. However, there is nothing that prevents a design solution from implementing alternative FRs, provided that they do not negatively impact each other and that other constraints and optimizations are not compromised. Complementary FRs are however expected to be possible to be present simultaneously in a design solution. One possible approach to illustrating this is given in Figure 78.

Figure 78: Variant functional requirements.


8.2.2 Mechanisms for interfaces and interactions and function representation

An area that in this phase of the research has been left out of the present scope is the mechanisms necessary for defining and managing interfaces and interactions between configurable components. Much research exists on different approaches to represent and deal with interfaces and interactions. The remaining task is to select an approach and adapt it to the requirements derived from the need to deal with un-configured and parameterized design solutions in the context of the configurable component framework. In doing this, it may be necessary also to select, adapt, and introduce mechanisms to define and describe functions and functionality.


8.2.3 Bridging the gap between engineering disciplines

The configurable component concept shows many similarities in rationale, purpose, and general functionality with the framework (see section 4.5.2) described by Zhang & Jarzabek (2003). Using the terminology of Zhang & Jarzabek a systems structure is similar to an x-framework and a configurable component is similar to a meta-component (or a x-frame) that is used to generate a variant specific component (in our case a part definition) given a certain variant specification. Zhang & Jarzabek (2003), however, treat the root x-frame (or specification x-frame) especially, since they define this to carry the variant specification. The work of Zhang & Jarzabek has its base in the software engineering discipline. This work and several other contributions from the software engineering community regarding product lines of software products could potentially be leveraged and moved towards more common representations and methods between the mechanical, electrical, and software engineering disciplines.


8.2.4 Boundary spanning communication in the supply chain

The configurable component concept is defined to enable self-sufficient sub-sets of design information with almost arbitrary levels of information hiding. It should therefore be an adequate approach to facilitate more extensive design data to be communicated through several tier levels in the development and manufacturing supply chains. This has, however, not been included within the scope of the current research and is therefore an interesting topic for future research.


8.2.5 Design tool implementation

The concept of configurable components was implemented as replacement of a core product description system within the scope of this research work. The concept has however not been implemented with the specific purpose of being a design support tool and especially so for the early phases of product development. An IT implementation of the concept as a design support tool that also includes the requirement of providing seamless support at different tier levels in the supply chain is an interesting topic for future research and further validation of the proposed concepts.


8.3 A reflection inspired by an open systems dynamics analogy

Hitchins (2003) highlights that open systems interactions present a concept of continual change over time. The mean level of disorder is increased by energy, and the rate of change from order to disorder to order is increased by energy. Energy does not only create disorder. For open systems, it also creates order from that disorder. Hitchins (2003) formulates a hypothesis, or empirical law, of open systems dynamics: open, interacting systemsí entropy cycles continually at rates and levels determined by the available energy. This can perhaps be compared to the business systems. An active customer base coupled with a highly competitive automotive industry competing globally could be viewed as a rather high level of available energy. Using the hypothesis of open systems dynamics this high level of available energy would imply a high rate of change from order to disorder and again to order within and between the business systems on the marketplace. That is, the increasing market dynamics with more customized sales and more and smaller product segments imply that more forming and breakdown of structures in the business systems will occur. Therefore, the need for robust (non-rigid, adaptable, and/or configurable) structures would be increasingly important.

With a systems oriented view and for illustrative purposes we could use a perspective in which we can view the changing taste of customers on the market (including the effects of competition) as a kind of disturbance (or noise) forced upon our system (Figure 79). We can then either be robust towards that disturbance or be affected in some way by it. Being affected by a disturbance is equal to not being able to provide the desired output. In the case of our business operations being affected by the disturbances, this will to some extent translate to not being able to reach the expected sales volumes with the expected contribution margins.

Being able to cope with disturbances and changing market conditions requires some flexibility and robustness in the producing system (compare with flexible manufacturing systems). Besides the basic need for flexibility in the entire production system Ė a major issue must be how and where to create this flexibility. The assumption must be that creating flexible solutions requires greater investments than non-flexible solutions. Therefore, if flexibility is created where it is not needed, it is a waste of resources. A critical issue must therefore be to identify where flexibility is needed and then identify how this flexibility is best created. The proposition to deploy configurable product and manufacturing process models based on configurable components presented in this thesis is therefore believed to be a step in the right direction in order to facilitate both the identification of the appropriate flexibility and the management of the operational processes that, on a daily basis, must manage and adapt to changing conditions utilizing the available flexibility in the business system that has been created. Being able to cope with disturbances and changing market conditions requires some flexibility and robustness in the producing system (compare with flexible manufacturing systems). Besides the basic need for flexibility in the entire production system Ė a major issue must be how and where to create this flexibility. The assumption must be that creating flexible solutions requires greater investments than non-flexible solutions. Therefore, if flexibility is created where it is not needed, it is a waste of resources. A critical issue must therefore be to identify where flexibility is needed and then identify how this flexibility is best created. The proposition to deploy configurable product and manufacturing process models based on configurable components presented in this thesis is therefore believed to be a step in the right direction in order to facilitate both the identification of the appropriate flexibility and the management of the operational processes that, on a daily basis, must manage and adapt to changing conditions utilizing the available flexibility in the business system that has been created.

Figure 79: Cybernetic and open loop control models (Hitchins, 2003).



Go to table of contents.