Скачать 94.91 Kb.
Proceedings of the TMCE 2004, April 12-16, 2004, Lausanne, Switzerland, Edited by Horváth and Xirouchakis
2004 Millpress, Rotterdam, ISBN
development of a Product model to support engineering change management
Engineering Design Centre
University of Cambridge
P. John Clarkson
Engineering Design Centre
University of Cambridge
Engineering change is a topic of increasing interest as firms strive to quicken their development processes to better meet customer needs. Making an alteration to a design is not always simple – the execution of a change can lead to unexpected consequences as propagation may occur causing other parts or systems to be affected, some of which may not even be directly connected to the initially altered component. Predicting if an alteration will spread, especially in complex products made up of many highly connected parts, can be very difficult. Engineers need more support with the evaluation of proposed engineering changes. This paper describes the creation of a diesel engine product model, which highlights the key linkages within the engine and rates the likelihood and impact of changes propagating. Validation of the model indicates that it could provide support for the evaluation of engineering change.
engineering change management, product models
Over the past few decades, companies that design and manufacture products have witnessed dramatic changes to the environment in which they operate. Intense global competition, dramatic advances in technology and the creation of fragmented markets populated by sophisticated customers have lead to a situation of rapidly shortening product life cycles and the demand for customised offerings (Clark & Fujimoto, 1991). In order to survive, firms must improve the efficiency of their product design processes and cut development times (Smith & Reinersten, 1998).
One of the key procedures connected with product design and development is the engineering change process. Although, employees often regarded making alterations to parts, drawings and software as a nuisance (Acar et al., 1998), there is a growing realisation amongst businesses that a well-managed engineering change process is essential for businesses to compete in the future (Terwiesch & Loch, 1999). Effective engineering change management assists a company in matching the technological innovations of rivals and can help to develop a competitive advantage (DiPrima, 1982).
This paper reports upon ongoing work being carried out with a UK-based diesel engine company to develop models and tools to support the engineering change process. Engines designed and built by the company power in excess of 5,000 applications for more than 1,000 customers. Thus, engineering change is a topic of major interest to the firm.
The next section provides general background information on the subject of engineering change, support tools and product modelling. Section 3 describes the motivation and basis for the model whilst section 4 describes the model elicitation. Findings are discussed in section 5 along with possible further uses of such product models.
It is important to distinguish engineering change from the general concept of change in a business/ organisational context. Change management is a term that is common in management literature. It refers to the administration and supervision of organisational transformation – be it the results of merging two firms or implementing a new business process. Engineering Change Management refers to the process of making modifications and alterations to parts, software and drawing that have already been released at any point in the product lifecycle (Terwiesch & Loch, 1999). In this paper, any mention of change refers to engineering change.
The topic of engineering change has largely been ignored in academic literature. An extensive review of engineering change management literature published between 1980 and 1995 discovered only 15 ‘core’ papers (Wright, 1997); other authors have also mentioned this scarcity (Huang & Mak, 1999).
Close attention is starting to be paid to engineering change processes and this is, in part, being driven by the needs of companies to comply with Configuration Management and Quality Management standards (e.g. ISO10007 (ISO, 1995) and ISO9001 (ISO, 2000)), which demand clearly documented processes for all key business activities. In conjunction with this, the trend towards the outsourcing of large aspects of a design to suppliers and contractors has meant that rigorous and explicit management procedures are required.
Configuration management (CM) is a formal discipline, which allows complex products to be designed and produced concurrently by several businesses separated by thousands of miles (Lyon, 2001). One of the key aspects of CM is the control of engineering changes. Although originally developed for electro-mechanical goods, most recent literature on CM has focused on software products (Huang & Mak, 1998). The focus of CM is on document control and the administration of product options; the more technical issues involved in making changes are either ignored or covered in little depth.
Companies approach change management quite differently lower down their organisations due to the specific requirements of their product ranges, but at a macro level they have broadly similar processes and procedures (Pikosz & Malmqvist, 1998). The terminology may differ from firm to firm, but the intent of the phases is similar. Figure 1 shows a generic high-level engineering change process with 5 steps based upon ISO 10007 (ISO, 1995).
Perhaps the most critical phase in any engineering change process is the evaluation of the possible impacts of a proposed change (step 2 in figure 1). Two important aspects that must be considered are the impact upon the product itself and the affect on the development process (budgetary, organisation and schedule considerations)
Impact on the product
At a fundamental level the exact characteristics of each product significantly affect the possible impacts of any change. The level of connectivity between parts and systems is vitally important as changes can propagate from component to component, and system to system, so that one change leads to many.
One of the key factors, which influences whether a change will spread, is the product architecture. Modular designs can be adapted much more easily than highly integrated devices, but only if the interfaces between the modules remain unchanged. This has been termed local change, which just involves one component or system (Lindemann et al., 1998). Once the interfaces need to be altered, the complexity of the issue increases markedly. This is interface-overlapping change, which is especially common in complex products with high connectivity between parts. However, it is important to note that sometimes changes can propagate without causing interface alterations, for example vibration and electro-magnetic interference effects.
Impact on the process
Exactly when an engineering change occurs during product development can have a dramatic impact upon the schedule and cost of the project (Lindemann & Reichwald, 1998). The later an alteration is implemented the higher the cost: alterations that arise in the design phase are much cheaper to deal with than those that occur during production ramp-up; these in turn are less costly than a change that causes production tooling to be altered (Terwiesch & Loch, 1999). Once production has started the impacts spread further into many other business processes. How change affects the manufacturing process is an area that has received much attention, especially the impact on the bill of materials and on Material Resource Planning (MRP) systems (e.g. Ho, 1994). Engineering changes lead to an increase in the amount of product data that must be handled. Ensuring that only current data is available is one of the major challenges (Wright, 1997).
Change propagation can have a huge impact on the organisation, budget and schedule. At a macro level, change behaviour can be split into several patterns based upon the amount of effort required to successfully implement the change and the time-scale that the alteration is carried out in. The most extreme situation has been termed a change “avalanche” (Eckert et al., in press), where the initial change spreads requiring huge amounts of extra effort and causing schedules to overrun; other authors have termed this the “snowball effect” (Terwiesch & Loch, 1999). In such situations implementation of the original change can either be abandoned or more effort and expense must be invested.
There is agreement that computer based tools are essential to support engineering change management within firms (Huang & Mak, 1999; Lindemann et al., 1998). Paper based systems are generally inefficient and become more so as the volume of alterations rises. Most companies ultimately depend upon staff remembering changes, but the cognitive overload on senior engineers is enormous because they need to mentally track changes over long periods of time. This tacit knowledge is extremely difficult to pass on between experts or for novice designers to learn.
Many different packages to support the engineering change process are commercially available (often as part of Enterprise Resource Planning (ERP), Product Lifecycle Management (PLM) or Product Data Management software (PDM)) or developed in house. These systems are mainly used to record and track changes and provide no support for the impact analysis phase. Huang and Mak (1998) provide a good introduction to this area.
Surveys undertaken towards the end of the 1990s have shown that relatively few companies seem to be using such systems (Huang & Mak, 1998; Yee et al., 2000) or making full use of the potential of computer tools (Pikosz & Malamqvist, 1998). Although dramatic advances have been made and more companies are embracing such software packages, modern PDM systems still mainly provide support for workflow and automate the paper flow. Engineering change information relative to the product is still not modelled in current PDM systems (Riviere et al., 2003)
Impact evaluation tools
Modern CAD software, such as Catia1, can predict the impact of changing a component by analysing the product geometrically and calculating the mismatch to neighbouring components, but not more complex interactions (e.g. vibration). Modelling tools such as Modelica2 describe the energy exchange between different subsystems in a dynamic fashion. Global (systems level) behaviour can be predicted to a certain extent, but not effects at the component level.
There is no known commercial package that helps to predict the propagation of a change, although some work is being carried out in academic institutions. For example a tool called RedesignIT has been reported upon (Ollinger & Stahovich, 2001), which uses a product model consisting of the relevant physical quantities and the causal relationships between them, to generate and evaluate proposals for redesign plans. The technique has been used to evaluate possible modifications that could improve the output torque of a turbocharged diesel engine.
C-FAR (Change FAvourable Representation) is a technique that has been proposed to facilitate change representation and propagation mechanisms (Cohen & Fulton, 1998). The technique is built upon the EXPRESS information model, which was created to define engineering products and support the management of engineering data. Linkage values are a key part of C-FAR; these connect the attributes of one entity to the attributes of another. Experts rate the linkage values and matrices built from this data enable change propagation to be estimated.
Another tool, called the Change Prediction Method (CPM) is under development and it aims to assist in the understanding of how changes spread through a product (Clarkson et al., 2001). This approach uses a Design Structure Matrix (DSM) as the basis of the product model. A DSM is a square matrix that displays the relationships between the components of a system or product in a compact form. The row and column labels are the same and an off diagonal mark shows that there is a relationship between the two elements. A link is marked with a cross in simple binary DSMs, but can be replaced by numerical values to show the weighting of relationships. A comprehensive review of DSMs is provided by Browning (Browning, 2001).
The CPM tool uses a simple model of risk, where the likelihood of a change propagating is differentiated from the impact of such an occurrence: risk is defined as the product of likelihood and impact. A product model consisting of two numerical DSMs is created, which show the likelihood and impact of change propagation occurring between directly connected components. A route counting algorithm is then used to calculate the combined risk of change propagation, which is the sum of the direct risk and indirect risk (change spreading via intermediate parts). The DSM product modelling technique used for the CPM tool is used as the basis for the product models developed in this paper.
As is clear from section 2.3, the key issue in any proposed method or tool to support engineering change management is the development of an accurate and appropriate product model. A model is an abstraction of a real system that gives approximations of complex physical phenomena as part of physical systems. During the conceptual design phase, a great deal of product modelling occurs as different ideas are analysed and evaluated. For example, Quality Function Deployment (QFD) is a technique that models the match of customer requirements to the functional requirements of the product (Hauser & Clausing, 1988). Models can be used to examine various aspects of the device from the performance of specific systems (or modules) to the architecture of the product. One aspect of product modelling that has been covered in depth is the modelling of product function. Pahl and Beitz (1996) have lead the way in developing structured methods to establish function structure. The product is modelled as a series of black boxes with three types of input and output: energy, material and information.
The modelling of product architectures has been an important research topic for the past two decades. For example, Pimmler and Eppinger (1994) used static DSMs to reveal and examine alternative product architectures. The method investigated four different possible interactions between components/ assemblies: spatial, energy, information and material. These were related to the functional modelling concepts that were proposed by Pahl and Beitz (1996) and Suh (1990).
One of the most important issues involved in product modelling is the level of granularity chosen. For architectural models, the possible product breakdowns range from ones containing a few large parts and assemblies to ones that include every single part in the bill of materials (BOM). A balance needs to be struck between the amount of effort required to elicit and maintain the model and the level of accuracy required.
The motivation for the development of a component linkage product model came from two sources. A review of the available literature, some of which has been described in section 2, showed that there was a general industry need for tools and techniques to assist engineers evaluate and implement changes. This was supported by an empirical study conducted within the company. As mentioned above, the topic of engineering change is of vital interest to the firm.
The authors carried out twenty semi-structured interviews with engineers and designers. The subjects came from a wide spectrum of roles within the firm and had a range of experience within the industry. Each person was questioned on key topics related to engineering change management. In parallel to the interviews, several Design Change Meetings were observed and a senior engineer was shadowed for a week.
The key issues highlighted in the interviews were:
The interviewees gave many examples of past changes to the engines that they had been party to, most of which described situations where the implementation of a change had not gone smoothly. One instance, which was mentioned by several designers, concerned an outlet pipe change. A project was undertaken to replace the metal water outlet pipes in the engine with plastic ones to reduce overall engine weight. The water outlet pipes contain a temperature sensor and this was transferred from the old design of pipe to the new one. It was only once the redesigned engine was in production that it was noticed that the sensor no longer functioned: it used to earth through the pipe, which was no longer possible due to the material change. The solution was to redesign the sensor with a return wire. As a senior project manager commented “Nobody thought about it when they introduced plastic pipes. It was very embarrassing and very expensive.”
The key issue that arose was that engineers and designers found it hard to appreciate fully the complexity of linkages between parts that could cause changes to propagate. An engineer summed up this situation well, “We miss the other things that the components are doing because most components are doing several jobs.” One experienced design manager commented that certain connectivities were only considered because the firm had “been burnt by them” in the past. To support new engine development, a good overview of the linkages between components is seen as vital.
Before describing the elicitation exercise, it is essential to define what is meant by a component linkage: it is a direct relationship or connection that exists between two pieces of a product. Depending upon the level of granularity chosen, the pieces can be individual parts, sub-assemblies or modules. The linkage can represent any important relationship that would connect the two pieces, from a connection, which, if broken, causes the device to cease to operate (e.g. electrical flow) to an association, which, if violated, does not affect the product's primary performance (e.g. a vibration effect). Linkages can either be symmetrical (acts in both directions i.e. A → B and B → A) or directional (there is a flow or transfer from one component to the other i.e. A → B only).
Obviously change can only spread from one component or sub-system to another if they are connected in some way. Even a simple product has a complex network of linkages between its parts. As the complexity of the product examined increases, the number of potential types of linkage rises to include issues such as heat, vibration, electricity, etc.
Linkage analysis differs from functional analysis in a number of ways. Firstly, linkage analysis examines the connections between components and assemblies whereas functional analysis identifies the functionality of the individual components themselves. Added to this, linkage analysis looks at a product from a change perspective and as such has to fully embrace negative aspects (e.g. heat dissipation, electro-magnetic interference, vibration, etc.) and could include aesthetic/ form issues as well as the functional aspects. Also linkage analysis could be used to examine how the product fits into the complex web of processes that create it and support it in the market place. For example parts and sub-systems can be linked due to similar manufacturing processes or due to supply chain issues such as being sourced from the same supplier.
The approach described in this paper has similarities to that of Pimmler and Eppinger (1994), which was outlined in section 2. One key difference is that this work focuses upon making product models to assist the alteration of already established products (i.e. the architecture is already established), whereas the Pimmler and Eppinger work focused upon the development of different architecture concepts.
Given the experience gained from the development of the CPM tool (Clarkson et al., 2001), which was outlined in section 2, it was decided that DSMs were the most suitable basis for constructing the product model.
This section describes the method used to build the diesel engine model; it is very similar to DSM creation methods as described by Browning (2001). The method used to elicit the model has 4 steps: (1) selection of the product to be examined; (2) creation of a suitable product breakdown and identification of key linkages; (3) priming of the product DSM by a product expert; (4) elicitation of the complete model with a team of engineers.
Towards the end of the empirical study, the firm launched a new product, the 1104 engine, which is shown in figure 2. This was the product chosen for the modelling exercise because the company had a need for support with its continuing development and it was a project that was still fresh in most minds. Clear explanations of diesel engine design and function can be found in many books (e.g. Maleev, 1954).
In order to identify the key linkages and a suitable level of granularity for the model, it was decided to run an exercise with a small group of experienced engineers. An initial engine model was created and then four designers were asked to identify the key linkages present and discuss the model granularity.
There are approximately 550 parts in an 1104 diesel engine, but of these only about 100 can be regarded as critical, non-standard components. However, as the number of possible direct linkages increases as a squared function of the number of components in the product breakdown, it was decided to choose a coarse granularity. Thus, the engine was broken down into 26 major assemblies, which were based upon the options offered in the company’s engine reference manual and this model was put into DSM form. An engineer in charge of new engine development and the authors discussed the important linkages. It was felt that mechanical, spatial, thermal and electrical connections must be included to develop an accurate and relevant engine model.
The four engineers selected were a design team leader, two conceptual designers and an analysis specialist. They were chosen because all four had many years of experience with diesels and a good overview of the engine. Each was asked to use his knowledge to fill in a blank DSM, marking all the key linkages. As a guide the designers were asked to consider mechanical, spatial, thermal and electrical linkages, but were given free reign to add in extra linkage types or amend the model if they felt that important connections could not be represented. Whilst filling in the matrix the subjects were encouraged to “think aloud” (as described by Ericsson & Simon, 1993) and explain their reasoning behind each identified linkage. The blank matrices were printed on A1 paper and then transferred to Excel spreadsheets for analysis.
The subjects largely agreed on geometric connections (spatial and mechanical) between parts (besides some divergent interpretations of component assemblies). However, dynamic links, electrical and thermal links were interpreted quite differently. The filling in of the matrix was strongly influenced by what the engineers knew about the design and what their daily activities were. For example, the analytical designer knew most about thermal links, because they fell into his area of expertise.
The exercise showed that a larger scale breakdown was required to accurately model the engine and there was also a need for more clearly defined linkages to focus the engineers' minds. It was apparent that it would be highly unlikely to gain a complete picture of the engine linkages from one person, even from an engineer with a good overview of the product and many years of experience.
Improved engine model and linkage definitions
As it is hard to define what a typical 1104 diesel engine is (as the options can vary widely between customer applications), it was decided to base the team exercise around a specific production engine. A turbocharged, electronically controlled engine was chosen with a gear driven compressor and a belt-driven alternator. An engine model made up of 41 parts and assemblies was decided upon, which is grouped into two categories ‘core’ and ‘non-core’.
|A process Model of Successful Supplier Integration into New Product/Process/Service Development||Decision support and disease management: a Logic Engineering approach John Fox and Richard Thomson|
|Fluent in both business and technology. 10 years of product management preceded by 10 years of software engineering. Expertise in information security. Masters||Statistical Support and Web Development for a Web-based Master Sample Management System for Integrating Aquatic Ecosystem Status and Trend Monitoring|
|Enabling Decision Support and Costing of Product Designs by using Visual Metaphors||Product/System Development Propulsion (Earth Based)|
|Market Orientation and Partnership Learning in Product Development and Design||Product news plan b sales double, Glamour poll shows strong support for Rx-to-otc switch, wcc welcomes new staff member|
|Fema emergency Management Institute (emi) Emergency Management Higher Education Project Development of a Course Treatment||Welcome to the course Management of change|