Improving Engineering Design: Designing for Competitive Advantage (1991)

Research is a central ingredient in repairing the national infrastructure in engineering design. It will contribute new knowledge, new ideas, and new people to industry and education and stimulate the creation of new business enterprises. Over time, a well-conceived, sustained program of engineering design research will gradually reduce U.S. companies' reliance on ad hoc design methods and improve their ability to produce higher quality, lower cost products and reduce lead time to market for new or modified products. It must be emphasized that the research must deal not only with the product functional quality aspects of design but also with product cost and time to market.

Research is generally divided into (1) basic research, which creates new knowledge and methods that explain or describe (often formally) natural phenomena or human behavior, and (2) applied research, which extends basic research with an emphasis on producing results directly useful to practitioners. These poles are at the ends of a continuous spectrum of research activity that ranges from the most basic, in which the goal is fundamental knowledge and understanding, to the most applied, in which the objective is to put knowledge to specific, immediate use. This chapter describes a broad topical research agenda for engineering design that ranges from fundamental research to applied topics with broad applicability. Highly directed applied research, usually driven by a specific problem, must remain the responsibility of individual firms.

Though other sections of this report stress the need for industry and engineering education to acquire, use, and teach existing advanced design

methods, much remains to be learned about engineering design processes and the knowledge and strategies needed to perform quality design quickly. Basic research is needed to generate ideas and foundations for new methods, processes, and supporting tools, and to foster continued improvement of practice. As mentioned earlier, competitiveness demands the continual development of new design methods. Using methods developed by a firm's competitors always relegates that firm to a trailing position with regard to quality, cost, and time to market of new and improved products.

Engineering design today is based largely on rather specific ad hoc bits or kernels of knowledge gathered from experience (i.e., heuristics). Though engineering design is clearly a knowledge-based intellectual activity, and its basis in knowledge and strategies should be amenable to acquisition, generation, organization, testing, and evaluation, a foundation of general knowledge, principles, and strategies has not yet been developed.⁶⁰ Once organized and generalized, design knowledge and strategy could be taught, learned, and used more effectively, and gaps in knowledge might be revealed to guide further research. Other roles for basic research include evaluating existing knowledge and strategies and providing formal principles and foundations for new tools and methodologies.

One category of basic research in the agenda presented below consists of studies that investigate the scientific foundations of design. Since skepticism about the possibility of discovering design theories, and about their potential usefulness should they be discovered, still exists both outside and within the engineering design community, it is important that the nature of research into the scientific foundations of design be addressed.

Theories, in any field, are testable, inductively generated statements about relationships among operationally defined variables or abstractions. In the physical and natural sciences, this definition is readily interpretable and well understood. Design, however, is not a physical or natural phenomenon but a complex intellectual and social process that involves many poorly understood variables, abstractions, and possible relationships. This complexity makes difficult, but does not prevent, the formulation of useful theoretical foundations, at least for important aspects of the design process.

In connection with this complexity, the field of engineering design can be viewed as consisting of three independent categories of variables and abstractions: (1) a wide variety of problem types, (2) a wide variety of persons who may be required to solve the problems, and (3) a wide variety of organizations and environments (including tools and available time) in which the persons may be required to function. Attempts to discover crucial variables and abstractions that apply to persons and the environment are likely initially to be either unmanageably complex or else greatly oversimplified. Moreover, research methodology in these categories is cumbersome and difficult to plan and implement. Obstacles faced in the cognitive, so

cial, and environmental aspects of design are much the same as those faced by researchers in such fields as education, sociology, and management. Such design research, having been extremely limited in scope by virtue of very limited funding, has appeared “soft” and quantitatively inconclusive. Even so, it has yielded or confirmed qualitatively useful insights (e.g., most designers tend not to explore alternatives well enough).⁶¹ Further insights that might aid organizations in planning and managing the human aspects of the design process can be expected as research methods are refined and goals are advanced. For example, principles by which concurrent design teams should be organized might be generated or the nature of the tools most helpful to human designers might be discovered.

A less subjective approach to the development of theoretical foundations for design focuses only on engineering aspects of design, omitting human and social environmental issues. Such an approach might begin with well-defined problem types and the associated knowledge and processes required to solve them. Design problems have been categorized as parametric, configuration, and conceptual. A further subdivision can be made on the basis of whether the product to be designed is a single component or an assembly of components and/or subassemblies. ⁶² Once a complete taxonomy of design problems is developed, and means for identifying and formulating each type are known, the search for engineering design methodologies becomes one of finding and explaining generic processes, strategies, and knowledge applicable to each type. Such problem-type-oriented fundamental development has begun and is in fact quite extensively developed for the parametric design of components.⁶³

Work on parametric design of assemblies is progressing, since various optimization approaches are potentially applicable. At the configuration and conceptual design levels, some theoretical development of synthesis and evaluation processes has begun, but nothing so well established or formal as optimization theory or statistical methods has yet evolved for these problem types.

A set of viable, proven, formal processes for solving all types of design problems will constitute a theoretical foundation for engineering design that will evolve as research discovers new or improved methods. ⁶⁴ Though initial progress has been made in recent years, much of it supported by the NSF's Program in Design Theory and Methodology and design-oriented Engineering Research Centers, much more remains to be done.

The committee appraised a large number of potential research areas in engineering design. These areas were weighted according to potential pay-off to industry and education, intellectual interest, probability of success,

and required resources. Ten research topics were found to be crucially important to reforming the practice and teaching of engineering design, and thus worthy of continued expanded effort. These topics were categorized according to objective: (A) developing scientific foundations for design models and methods; (B) creating and improving design support tools; and (C) relating design to the business enterprise. Collectively, they comprise a national research agenda that will serve to guide the NSF, other government agencies, private foundations, industries, and individual researchers in the selection of research priorities and emphasis. That agenda is outlined below.

1. Developing scientific foundations for design models and methods

   1. Computer representations of in-progress designs

   2. Generating, organizing, and generalizing design knowledge

   3. Synthesis: parametric, configuration, and conceptual design

   4. Tolerance synthesis

2. Creating and improving design support tools

   1. Designer-oriented computational prototyping, analysis, and simulation tools

   2. Rapid physical prototyping

   3. Design for ‘X'

3. Relating design to the business enterprise

   1. Quality-cost models

   2. Organization and communication models

   3. Innovation

Each of these 10 research areas is described in more detail below. They are not further prioritized because their value will depend upon the quality of the research performed, and their usefulness will vary from industry to industry.

Research topics in this category deal with the fundamental scientific foundations on which the subsequent development of new design practices and tools will be based. Current design practice and the foundations for many existing design tools have evolved from collections of ad hoc practices and heuristics that are believed to have worked in the past or in other circumstances. Moreover, the knowledge on which most design is based is largely fragmented and unorganized and often highly specific to companies, product types, or technical domains. Formal foundations for new design-oriented CAD and solid modeling systems are needed, and fundamental studies of design models that uncover the knowledge and strategies needed

to perform design will serve as the basis for improved design methodologies in the future.

A representation is a description of a design. Descriptions change during the design process from highly abstract to highly detailed. To support new best-practice product realization processes, especially at early stages and in concurrent design environments, new computer representation methods are required. The need is for a formalism that supports representation of designs at the multiple levels of abstraction and detail appropriate to different stages of the design process.⁶⁵ For example, whereas the representation might focus on functional and manufacturing issues at early stages, much later a detailed specification of dimensional and manufacturing information will be needed. The representation should also support the varied activities involved in the complete design process, including, for example, many types of preliminary and detailed functional analyses and simulations, manufacturing evaluations at many stages, cost and quality estimates, marketing and sales functions, and tool and process design. Current CAD and solid modeling systems, however advanced, are not fully utilized in industry (particularly by smaller and middle-sized firms), because they do not serve these requirements.⁶⁶ Neither do current representations adequately serve preliminary design or early analysis and evaluation processes. They are not transformable into different functional partitions and do not support the different levels of abstraction or incomplete or inconsistent designs that are common in early design stages. Current systems are well founded mathematically (a definite plus), and they support detailed analyses (e.g., finite element methods) reasonably well (though designer interfaces are still awkward). The new representation methods needed to support new product realization processes will owe a great deal to current systems and build on the knowledge and experience gained in their development. ⁶⁷

It is widely believed that the new representation methods will involve feature extraction from existing systems, and work on designing with features has begun, though the effort is small as yet. A formal definition for the term “feature,” though still wanting, is expected to generalize the concept well beyond the original notion of “form features,” i.e., surface elements such as holes, bosses, and fillets. The required definition should probably include information about materials, relationships to other forms, and manufacturing, as well as about form.⁶⁸

Development of a foundation for new representations that support concurrent design, design for “X,” and other aspects of best-practice product realization processes, is an extremely high-priority research need. A new generation of more “intelligent” CAD and solid modeling systems that provide

designers with convenient manufacturing information and powerful, easy-to-use analytical tools simply cannot be built until research resolves the underlying representation issues.^{69 ,70}

Engineering design is a knowledge-based, knowledge-intensive intellectual activity. Designers and others involved in the design of any product or process bring to bear extensive technical knowledge, product knowledge, manufacturing process knowledge, design process knowledge, memories of previous projects, and so forth. Much of this knowledge is presently ad hoc and heuristic, residing implicitly with individuals or within organizations and neither accessible to, nor of a form that is easily accessible by, others within the firm, much less in other firms or disciplines. The handbooks, textbooks, catalogs, trade journals, research journals, and company guidelines in which much of this knowledge has been recorded are generally useful only if close at hand (some say “within reach”) and if they deal specifically with the designer's current problem.⁷⁰ As a data base, this collection is extremely inefficient in terms of accessibility.

A design knowledge base more generally and completely accessible to all engineering designers would be tremendously powerful. For this vision to be realized, existing knowledge must be organized and, where possible, generalized. Once this is done, the knowledge might be made available to designers via CAD systems or computer networks. With the existing knowledge organized, identifiable gaps will serve to guide future research.

A few very small steps have been taken toward improving the design knowledge base. Some knowledge-based expert systems have been developed for specific applications, and some computer-based catalogs and design libraries are becoming available. In the future, these might be incorporated into CAD systems. There are some difficulties (i.e., research opportunities) involved in achieving the desired results here. The volume of information is huge; taxonomies of design knowledge that might serve as organizing principles for the knowledge are still lacking;⁷¹ all the problems of very large data bases are relevant; and there will be problems with some firms' unwillingness to share information considered proprietary. Nevertheless, if the world's best design knowledge can be acquired, organized, generalized, codified, and made available to designers in a convenient fashion, design practices will not only be improved, but also made more efficient. “Reinventing the wheel” can become a phenomenon of the past except, no doubt, where direct competition prevents information sharing. Engineers will be better able to explore alternatives; educators will have a much more teachable knowledge base; and engineers will have the accessible sources of information that are essential for speedy, reliable design practices. With the design

knowledge base better organized, new practices and tools based on proven models and methodologies could be continually developed.

The object of design processes is to synthesize solutions, that is, to combine separate ideas and information into a unified whole. This process of adding and integrating information and knowledge about the design (including its function, shape, size, materials, manufacturing, and so forth) is done almost continuously, from the early, highly abstract and incomplete stages to the later, much more complete and detailed stages. Synthesis at every stage involves generating alternative solutions to the problems at hand, analyzing and evaluating those alternatives, choosing among them, and integrating the information derived into the design so that the design process can proceed to the next step. Because synthesis is so pervasive in design, it is important to understand it on as fundamental a level as possible.

Work to date has generally taken the form of developing models of design processes at various stages and/or for different domains (e.g., linkages, power plants, building structures, and so forth).⁷² The stages usually studied are those mentioned earlier: conceptual design (sometimes referred to as preliminary or embodiment design); configuration design (wherein the basic arrangement of the parts of the design is settled); and parametric or detailed design (wherein the specific values for the different attributes or parameters of the design are determined). By far the most work to date has been done on synthesis at the parametric level. The field of optimization, ⁷³ which applies here, is well developed, but its techniques are not always relevant to realistic design situations; continued work to correct this is needed. Also, optimization methods are not yet available for assemblies of parts that have important interactions or crucial evaluation issues that occur only at the system level (e.g., natural frequency).⁷⁴ Taguchi and other statistical methods for achieving robustness are also parametric design synthesis procedures.⁷⁵ Finally, a number of knowledge-based computer methods for parametric design have been developed with varying degrees of generality and usefulness.

Though most of this work has aimed at developing synthesis models and methods at the parametric stage, a complete science of parametric design has yet to be articulated. Important gaps exist that can be closed by research. At the configuration and conceptual levels, very little has been done even to develop synthesis models and methods. Physical principles, at least qualitatively, are involved in both conceptual and configuration design. To date, the little work that has been done has been limited to narrow, domain-specific studies.

At every stage, synthesis involves the generation of alternative solutions, that is, innovation (discussed in Section C.3 ), evaluation, and decision making. These basic processes as they apply to engineering design need more study and integration into design process models and methods. New synthesis models and methods for various types of design processes can lead to new and improved best practices. We must continue to refine methods for the parametric stage and greatly increase the study of all aspects of synthesis (e.g., innovation, decision making, evaluation methods, knowledge and strategies needed, and so forth) in earlier stages of design.

Although it is an aspect of parametric design, tolerance specification is such a crucial driving factor in product cost and performance that it deserves special attention. ⁷⁶ Tolerances are applied to nominal dimensions of a part or product to indicate allowed divergence from a nominal value. Designers need readily usable information and procedures that support the judicious assignment of tolerances to optimize tradeoffs between product performance and cost. Though tolerances can now be set rigorously in a few highly specific cases, most tolerances are based on experience and company customs that reflect a mostly subjective attempt to balance product performance and manufacturing cost. There often exists a great deal of company-specific data on tolerance-cost relationships, but little general data and even less data that relate tolerances to product performance. Engineering designers are thus usually in the position of assigning tolerances without the benefit of solid data or rigorous theory. Some initial research, if continued, might yield the ability to represent tolerances appropriately in CAD and solid modelers, though this is a difficult task. Deficiencies in current tolerancing methods have been revealed in the process of applying them to mathematically rigorous solid models. Tolerance analysis, an essential aspect of tolerance synthesis and an extremely complex process, especially in three dimensions, is not yet fully developed. Finally, tolerance standards are not always consistent with available and evolving measurement methods.

Research is needed in tolerance analysis, tolerance representations, tolerance-cost relationships, tolerance-performance relationships, and tolerance standards and measurement methods. On the foundations laid by this research it will be possible to build design support tools to aid designers in making optimal tolerance selection decisions.

The introduction of new tools that improve designer productivity or performance is the most direct cause of changes in design practice. Both the

research in the previous section (which will provide foundations for new tools) and the research described here (to result more directly in new tools) are needed to create new tools. There is only a blurred and sometimes arbitrary distinction between research into foundations for new tools and the development or improvement of the tools themselves; this section of the research agenda comprises those subjects closer to tool development.

Analysis and simulation are supporting elements of design processes; they provide data and information about behavior, functional performance, cost, manufacturing, and other issues that are essential to intelligent design decisions. Although many computer-based analysis and simulation methods are available, especially for the detailed stage of design, few are in widespread use, particularly in smaller and middle-sized firms. One reason is that the technologies that employ these methods are not workable in all computer environments. Another is that proper use of these methods and tools requires highly specialized knowledge. Most analysis and simulation methods and tools have been developed for use in the final detailed stage of design. There is a strong need for these in many situations, but as emphasis on decision making shifts to earlier stages (as in concurrent engineering), there is an equally great need to provide new analysis and simulation methods and tools that are applicable before a design is completely specified. ⁷⁷ Research is needed to develop such new methods and make them available to designers. Computational prototyping (i.e., the ability to experiment with the behavior of parts or products using their computer representations) reduces design cycle time by reducing the need for actual physical prototyping. New representations, as discussed in Section A.1 , may be developed that will suppor powerful computational prototyping tools, including on-line handbooks and catalogs to increase efficiency further.

Here, the needed research is (1) to develop new methods of analysis, simulation, and computational prototyping that serve early stages of design and the new concurrent design practices, and (2) to make both existing and new tools readily useful to all designers.

Although analysis, simulation, and computational prototyping aim to shorten design and product development cycles and to improve the quality of the results by doing as much product testing as possible on the computer, ultimately a physical prototype must often be fabricated. Thus, tools are needed that link design and manufacturing quickly and inexpensively for prototype con

struction. Prototypes can serve different purposes at different stages of the design process. One may serve to test the applicability of a new material or process; another may test tolerance issues; some will have multiple purposes. Physical prototyping methods and tools are needed that serve the specific needs of the design process and that enable rapid realization of the desired physical model. The MICON system at Carnegie Mellon University exemplifies the state of the art for electronic systems.⁷⁸ An analogous system from the mechanical domain is Kimura's variant process planning system developed at the University of Tokyo.⁷⁹ First-Cut, a system under development at Stanford University,⁸⁰ bridges the gap between CAD and CAM by supporting simultaneous design of a product and the process used to manufacture it. Commercially available stereolithography techniques can produce complex parts from CAD representations, though material issues are not readily evaluated by this means.

Research is needed to define the various types of prototypes and their purposes, and practical, low-cost, and rapid methods must be developed to meet the needs of each type. For example, for processes such as injection molding, disposable dies may be feasible. Means of reducing the need for physical prototyping can be explored. The goal is to develop methods and tools that enable firms to construct physical prototypes, when necessary, both quickly and inexpensively.

A product must satisfy many objectives: function as perceived by the consumer; ease of assembly; maintainability; testability; safety; disposability; and many others. These are the X's in “design for X.” In best design practice, all are considered at the earliest stages of design as well as continuously throughout the design process. ⁸¹

The first X to receive explicit attention was assembly. Design for assembly (DFA) methods and tools developed by Boothroyd are widely (though not yet fully) disseminated. ⁸² Methods and tools to support design for manufacturing (DFM) in processes such as injection molding and forging have also now been developed. These methods identify, through experimentation, experience, or insight, the crucial features that affect ease or cost of manufacture of parts and assemblies, and then the presence, configuration, or parameters of these critical features are related to assembly or manufacturing time or cost.

This approach can be extended to many more X's and to the entire design and product development process. ⁸³ That is, the critical features that influence each X at each stage of design need to be identified and related specifically to their impact on X throughout the life cycle of the product. At this time, DFA and DFM knowledge relates primarily to the parametric

stage of design, though some aspects apply to the configuration stage. Designing for manufacturing at the conceptual stages and designing for other objectives (i.e., X's) at almost any other stage are not supported by any well-developed techniques. Consequently, consideration of these objectives tends to occur only after major design commitments have been made. Studies are needed that seek to relate the crucial features of a product's early description to its ultimate life cycle quality and cost in terms of each of the many design objectives (X 's).

Research on design in a business context addresses issues related to understanding and supporting design and product development in a companywide context, thus recognizing that functional and manufacturing aspects of a product cannot be considered independently of personnel, marketing, finance, accounting, and other business issues.

Quality and cost models are coarse-grained, but realistic models of relationships between manufacturing costs, time to market, user costs, and quality aspects of a product or process. Useful models are particularly needed at the earliest stages of design to support tradeoff studies and management and engineering decisions.⁸⁴ Accurate accounting for indirect costs and internal transfer costs is also important in these models.⁸⁵ At present, very few cost models and even fewer quality models are available that capture key cost, quality, and time drivers and their relationships at the conceptual design stage. Cost models are available for later stages, but these do not aid early design decision making. Quality models are generally lacking for all stages. Quality-function deployment,⁸⁶ used by some firms at the conceptual stage, is currently highly subjective but could be made more accurate and more widely applicable through research. Taguchi's quality-loss function is used by some firms at the parametric stage. Traditional cost accounting based on unit labor and material costs is usually unrealistic and can lead to inappropriate design and product development decisions. More research, possibly along the lines of activity-based management accounting systems, is needed.

Development of more accurate, tested quality and cost models, especially at the preliminary or conceptual design stages, is essential to support effective concurrent design. Studies are needed that identify early the product features that drive downstream quality and cost and relate these features to ultimate quality and cost, allowing models to be developed that support tradeoff decisions throughout the design process.

Organizational issues in design relate to the planning, organization, and management of product realization processes, including the creation and use of cross-functional or interdisciplinary teams. Communication issues relate to the facilitation and control of information transfer, both internally and externally, in design projects.⁸⁷

Several decades of work in organizational studies conducted by departments of psychology and sociology in universities and by organizational groups in management schools have seen little focus on design or even on engineering. Such work that has dealt with design has produced some useful results in the form of prescriptions for the organization of technical projects, design of facilities, structure of information systems, and organization and management of teams, but current work is weakly focused. Very little has been done, for example, on the flow of information within a design organization or between a manufacturing firm and its suppliers, yet these and other organization and communication studies are relevant to other areas of design research and to the development of better computer-based supporting tools. The goals of research in this area are several: to create and evaluate useful models of how information is and should be exchanged and used in a product realization process; to understand how multidisciplinary teams work in order to improve their performance; to create and evaluate models of product realization processes; and to learn how various supporting tools influence the performance of teams and of a product realization organization. The payoff will be shorter design cycles through improved organization and communication effectiveness. Information about cross-disciplinary teams can be used to support the design education process.

Innovation is the generation and implementation of new, unique solutions to stated problems. It is generally agreed that innovative capability is valuable in product development processes, but there is little agreement on how to stimulate it or on how to evaluate its cost-benefit tradeoffs.

Studies of the innovation process in individuals conducted in a variety of domains have yielded techniques for fostering innovative individuals, but little is known about the effect of innovative individuals on teams and vice versa. There is also little understanding of how to organize and manage groups and design processes so as to encourage innovation; factors that influence innovation in teams, such as team composition, time allotments, group environments, reward policies, and so forth, have not been established. It is conventional wisdom that most industry situations allow too little time

to be spent on innovation, but data to support allocation of more time are not available.

It is likely that knowledge is a key ingredient in innovation, both at the individual and group levels, but this too has been little studied.

Research in innovation is difficult, especially if it is to be credible in and relevant to industrial settings. Studies must be carefully planned and should be directed at establishing the value of applying resources to stimulate innovation and developing and evaluating methods for increasing innovation in groups and organizations.

The benefits to design practice and education that can flow from implementation of this research agenda include:

• a new generation of computer-aided design tools that support preliminary as well as detailed (parametric) design and that provide designers with information needed for manufacturability and life cycle issues;

• prescriptions to improve organization and communication in the product development process;

• useful quality-cost models that can support design and management decision making at the preliminary design stage;

• improved interfaces to engineering analysis and simulation tools;

• better information relating tolerances to cost and performance;

• more complete and implementable methodologies for design problem solving;

• new prescriptions for generating and evaluating configuration and conceptual design alternatives;

• greater availability and accessibility of the knowledge needed by designers to perform all types of design;

• development of design procedures that lead toward integration of all stages of design, from concept through disposal, involving the entire business enterprise, and directly addressing concerns regarding cost, quality, and length of the design cycle.

The importance of the proposed research on these 10 areas to the revitalization of the engineering design infrastructure in the United States and hence to U.S. competitiveness cannot be overemphasized.

Though most of the recommended research is open-ended and should continue for many years, significant and useful intermediate-term (i.e., four to five years) results should be achievable in almost all areas. The best

assignment of resources is to supply for each topic six to nine groups of researchers each consisting of two to four professionals. On average, funding per topic comes to about $2 million annually for four or five years. Total project commitment is thus $20 million annually for four or five years and between 120 and 360 researchers. Because a sufficient number of researchers may not be available, it may not be possible to begin all research immediately.

It is extremely important that all of this research, whether applied or basic, be of the highest quality. Researchers and sponsors must ensure that important research issues and problems are defined and rigorous research methods are followed. To make these research efforts truly responsive to industry needs and to familiarize university researchers with the connections among design and manufacturing, vendors, customers, sales, and service, the research should involve frequent and close interaction between researchers and design engineers in industry.

Both industry and academic communities will have to be willing to reach out and engage in the communication necessary to achieve mutual respect and understanding. Industry representatives will need to value and appreciate, become involved in, and provide support for intermediate- and longer-term research efforts. Engineers in industry need to read and contribute to the research literature related to design. The academic community needs to appreciate that design in industry takes place in the context of highly competitive business enterprises, a fact that has important implications for research. Finally, research results must be disseminated with industrial as well as academic readers in mind.

Engineering design research can yield major advances in engineering design methods, but the research must be related to the problems of industry and must be readily adaptable to the industrial design environment. Though university research efforts in engineering design are frequently long range and their results potentially applicable across a wide spectrum of industries, problems with dissemination of research results have left most engineering designers and engineering design managers believing that current design research has little relevance, so they are unwilling to seek and utilize new research results.

There are a number of paths by which the results of university research in engineering design might be brought into industrial practice. One route is through the development of new engineering design support methods and tools based on the fundamental knowledge generated by basic research.⁸⁸ This process is often slow, however, and can require intensive development work beyond the abilities of university research laboratories. A second

route is through new M.S. and Ph.D. graduates whose education, including their thesis research, has stressed engineering design. These graduates can either enter industry, bringing with them new and advanced engineering design knowledge, or accept faculty positions, through which they can pass their knowledge to a new generation of practicing engineers.⁸⁹ A third route is through design-oriented faculty members who work or consult in industry and engineering designers from industry who spend time in universities.

Although these personal modes of information transfer are important and contribute to awareness of new ideas in industry, effective exploitation of engineering design research demands that research results be put into forms useful to industrial firms. The new methods must be refined and packaged as products (mathematical or statistical computer packages, CAD systems, expert systems, and so on), a task not readily performed either by most universities or by most of the companies that might use the research. A few (generally small) firms have developed some research results into tools and methods usable by industry; other entities that might perform this development activity include government-funded organizations, multiple private design-oriented companies,⁹⁰ industry consortia, start-up firms, or some combination of these. The creation or enhancement of such research transfer paths would especially help small and medium-size companies that are unable to perform the task of development internally.

A mechanism is also needed to perform applied design research, sometimes referred to as precompetitive research, and disseminate the results. ⁹¹ This type of research is directed at industrywide problems that are too large or “too applied” for university laboratories and not amenable to cost-effective resolution by a single company.

Without some sort of organization acting as broker, the results of research on engineering design are not likely to reach the greatest number of potential users. An organization is needed to gather and disseminate information about international best engineering design practice, perform research to improve design methods and tools, and promote design technology transfer. This organization might also help arrange personnel exchanges and arrange privately funded research between universities and industry.

The committee discussed many ideas, methods, and techniques for dealing with the dissemination of research to industry, conducting precompetitive research, and acquiring and disseminating the world's best design practices, as well as the need for greater interaction between universities and industry and for brokering agencies to encourage such interaction. A National Consortium for Engineering Design (NCED) was considered for dealing

with these issues. The following list characterizes, but does not limit, the potential charter of an NCED.

1. Acquire or develop the world's best engineering design practices and processes.

2. Acquire or develop the world's best computer-aided engineering, design, test, materials handling, manufacturing, program planning, and other tools.

3. Document processes, practices, and tools, create training materials, and provide training at several levels (e.g., for trainers, practitioners, and managers).

4. Develop research proof-of-concept software into robust, user-friendly software ready to be used in practice or to be commercially developed for transfer to industry.

5. Conduct engineering design research as described in the research agenda.

6. Provide “hands-on” opportunities to learn by executing new designs.

7. Provide expert support and problem solving capabilities to members.

8. Facilitate collaborative corporate, government, academic, and NCED projects.

9. Establish industry-led engineering design applications projects that provide university faculty, graduate students, and government employees with industrial design experience.

10. Establish university-led engineering design research projects that provide industrial and government people with research experience directed at creating new design practices and tools.

11. Establish government-led engineering design projects to provide industrial and university people with experience in government sourcing processes.

12. Provide on-site courses taught by university, industry, and government people, as appropriate (e.g., a graduate course that can call on government and industry people, as well as professors, to lead lectures and workshops).

13. Host a yearly conference of engineering deans, industrial chief engineers, and human resource directors and government research agencies at which university course content and research directions, industry design and education practices, and personnel exchanges are discussed.

The NCED would be a nonprofit organization, funded by participating industrial, government, and academic organizations. It would, when fully operational, have a board of directors/trustees drawn from industry, academe, and government, as well as from the NCED itself, full-time administrative and technical staffs, and full and part-time representatives from participating organizations. NCED would study and perform research on mechanical structures, opto-electro-mechanical systems, and some widely used materi

als, and it would have enough manufacturing capacity to support concurrent design efforts. The output of NCED would be delivered through documentation, training, consulting, expert participation in member development programs, member visiting appointments to NCED, graduate student programs, sabbaticals, and other appropriate modes.

The NCED environment would lie somewhere between that of a major research center and that of a product development organization. A higher level of support and teamwork would be expected than is traditional within universities or corporate and government research laboratories. Joint projects, carried out under the leadership and at the site of either the member enterprise(s) or NCED, would be encouraged. Intellectual property rights would be negotiated in favor of the sponsoring members. Incrementally funded proprietary projects would be facilitated, and NCED employees would be encouraged to consult for some portion of each month. The primary objectives of the NCED would be to develop and accumulate knowledge of world-class engineering design practice and processes and transfer that knowledge to sponsoring organizations through a variety of formal and informal mechanisms.