IMPLICATIONS FOR ENGINEERING DESIGN EDUCATION AND RESEARCH
Rapid changes in computer-based design tools and the increasingly rapid introduction of new products into a global economy put unprecedented demands on designers and those who educate them. The demands on practicing engineers include completing designs in shorter times and operating with a limited knowledge of functional performance. Engineers must often serve as members of teams whose members are physically dispersed, teams composed of engineers of various disciplines, or teams that may include such professionals as lawyers, psychologists, or artists. Finally, the legal liability associated with designed products Requires not only that the product perform as intended but also that their designs and associated design decisions be documented and traceable.
It follows that tools to facilitate and enhance the productivity of designers and to document the basis for design decisions must advance. Several of the available tools are useful for these ends but have severe limitations. Quality Function Deployment, for example, can be very subjective, and the Analytical Hierarchy process can be applied only in simple decisions. Although they are far from perfect, these tools are widely used because they provide cost-effective solutions for designers.
Several theories of design are not widely used because they are either difficult to apply (Suh, 1990) or restricted in applicability to a narrow set of circumstances (Hazelrigg, 1999). Research to enhance existing theories or develop new ones that could be broadly applied is highly desirable, and the theories must be applicable in the real world, where design time and cost are as critical as product features.
This increasing demand and limited supply of adequate tools warrants research to improve the process of making design decisions. Decision analysis is a particular area for further investigation in engineering design because it has already established applicability in other fields. Further developments in engineering theory established on a credible axiomatic base are desirable as well.
A pervasive need exists for more rigorous design tools that can be easily used and that apply to a wide range of design processes. Progress in this area has been limited. The inability of the engineering design community to constructively critique current theories and to work together to formulate more general tools with wider application is striking.
Recommendation. Constructive dialogue should be encouraged to allow the best aspects of each method and theory for decision making in engineering design to be incorporated into common practice. Sponsored academic research could be used as a mechanism to assist integration of existing theories and development of new ones. Workshops or other cooperative ventures should be held in which experts establish engineering design tools that could evolve into common practice. The ease of use and practicality of these solutions should be paramount.
Much of the research in decision analysis per se has come from such fields as business and economics, where the multiplicity of variables renders intuitive reasoning impractical. Currently, formal decision analysis and decision theory are being confidently applied in such areas as medicine and business, where they are used to facilitate decisions with significant impact. This type of state-of-the-art decision making that takes uncertainty and the cognitive response of decision makers into
account could be a valuable part of engineering design, but no design process tool widely used today in engineering design is mathematically rigorous and universally applicable.
Far greater productivity of design engineers will result from additional advances in computing capability to model, simulate, and visualize products; from the understanding of physical, chemical, and biological systems; and from robust decision analysis tools. The ability to quantitatively describe the attributes of a successful design through modelling will continue to improve design methodologies.
A common framework and knowledge base must be established to use more universally those design or decision-making tools designed for one set of conditions. If simple and attainable input data can be described, and this data readily used by each tool, the formal decision-making tools used successfully in other fields and in other design applications will no doubt increasingly be applied to improving engineering design and the design tools.
Recommendation. More research should be focused on enhancing design tools and methods applicable to all engineering disciplines. The increasing complexity of physical systems, as well as software and biological and medical systems, begs for increasingly quantitative and transparent tools for making and justifying design decisions. Research for designers to develop appropriate knowledge bases is required. The many gaps isolating related tools and theories must be bridged and a taxonomy to facilitate comparisons between approaches must be created.
Underlying any improvement to engineering design processes must be a sound foundation in statistics, probability, and mathematics on the part of the practicing engineer. A brief survey of engineering curricula revealed the first two were given cursory and, in the committee’s view, insufficient treatment. The committee estimates less than a third of newly graduated engineers have even one formal course in statistics! For engineers to make informed decisions, they must have a sound understanding of statistics, mathematics, and probability.
Under current Accreditation Board for Engineering and Technology (ABET) guidelines, mathematics, such as differential-integral calculus and differential equations, is required. Statistics, along with linear algebra, numerical methods, and advanced calculus, is “encouraged.” Only industrial engineering undergraduates have a formal statistics requirement. Just one-fifth of the specialized program requirements give further emphasis to the importance of statistics, but none require statistics.
The mathematics of uncertainty is properly identified as “probability,” and the application of uncertainty, its management, and engineering, is “statistics.” Statistics is the methodology employed when one searches for structure (information) in collections of data possessing error-prone and noise-like variability. Statistics provides for the construction of surveys and experimental plans needed for the production of information-laden data. Gleaning information from data and succinctly expressing that information verbally, graphically, and quantitatively should be a part of every engineer’s education.
Recommendation. Statistics and probability should be required and incorporated into the undergraduate engineering curriculum to emphasize their relevance to engineering design and decision making, process control, and product testing.
Engineering design begins with the development of understanding in the form of a conceptual model of a desired object. Many models may be proposed, each encompassing extrinsic information coupled to the experience and knowledge of the decision maker or team. Uncertainties will certainly abound, and decisions among alternatives will be forced upon the decision team. The theory and practice of making decisions under uncertainty is clearly appropriate to any study of engineering design.
Once a design and its targets have been defined, the design team begins the task of moving from general considerations to specific requirements. In this evolutionary process, decisions are made
repeatedly, balancing costs, sensitivity to manufacturing and environmental variability, product reliability, and customer acceptability. Many tools assist in providing support for these decisions. Tactile physical models may be used or computers may construct three-dimensional visualizations. Decision trees and time path sequences are employed. Matrix displays of weighted attributes that contrast factors versus responses are almost always used. Patterns of experiments (statistical designs) may be employed to estimate sensitivities, make comparisons, and elucidate operable regions. Failure mode and effects analyses are common. These tools provide the basis of knowledge, the frame upon which decisions are made. The art of engineering design couples the use of tools to the ability to make the best decision in the face of uncertainty.
Recommendation. Decision-making tools and decision theory should be included in a required undergraduate design course. Interdisciplinary capstone courses that include legal, social, and economic issues, as well as team building skills, can be particularly useful teaching tools and should be included in this undertaking.
Although the committee tried to reach a consensus—and the entire committee did generally support most of the recommendations—some members held dissenting views that they expressed as follows:
It is not wise to invest more money in developing theoretical foundations for decision making in design, firstly, because, as the report indicates, the developments to date are fragmented and not in any way unified. They are applicable only as adjuncts to the entire complex design process. Secondly, the new challenges facing industry (a.k.a. the changing nature of engineering design) require fundamental changes to the design process, because most of these new products will require “creative” designs (Dym’s terminology) for new, unique, or vastly improved products. Such designs are not amenable to formal methods but require the development of knowledge bases concurrently with the designs.