001 17页，A Review of Research in Mechanical Engineering Design. Part II Representations,...

Res Eng Des (1989) 1:121-137 Research in

E gsineedng lgn 1989 Springer-Verlag New York Inc.

A Review of Research in Mechanical Engineering Design. Part II: Representations, Analysis, and Design for the Life Cycle

Susan Finger 1,* and John R. Dixon 2 1Robotics Institute, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA; 2Department of Mechanical Engineering, University of Massachusetts, Amherst, Massachusetts, USA

Abstract. This is the second of a two-part paper summa- rizing and reviewing research in mechanical engineering design theory and methodology. Part I included 1) descriptive models; 2) prescriptive models; and 3) computer-based models of design processes. Part II includes: 4) languages, representations, and environments for design; 5) analysis in support of design; and 6) design for manufacture and the life cycle. For each area, we discuss the current topics of research and the state of the art, emphasizing recent significant advances. A final section is included that summarizes the six major areas and lists open research issues.

Introduction

This two-part paper, the first in a series of reviews to be published in Research in Engineering Design, summarizes and reviews the state of research in engineering design theory and methodology, con- centrating on mechanical engineering design. Sub- sequent reviews will concentrate on other areas of engineering design or on special sub-topics. The goal of the series is to inform the community at large of advances in the developments in engineering design research. We also hope that it will enable researchers to place their work in context and thus guide continuing work. The series of papers is also intended to be an efficient starting place for those who wish to become familiar with the engineering design literature relevant to their interests.

There are, of necessity, limits to the nature and scope of this review. First, the review is not intended to be a substitute for reading complete papers; it is intended only as a brief summary of, and guide to, the literature. Although we have made every reasonable effort to be complete, omissions are inevitable. There can also be errors of commis- sion caused by misinterpretation or lack of full understanding on our part of papers included in the

* Reprint requests: Robotics Institute, Carnegie Mellon Uni- versity, Pittsburgh, PA 15213, USA

review. We apologize to both readers and researchers for these errors.

The scope is limited in several ways. We intend only to include research in engineering design, and then only that portion of engineering design broadly called "mechanical," which includes products, machines, structures, and the like. Research in geometric modeling, architectural design, manufacturing, expert systems, and optimization are included only when the research is directly relevant to design of mechanical systems. We have also not attempted to cover the many new, commercial computer- aided design (CAD) systems which have begun to incorporate the research ideas discussed in this review.

The research discussed in this review paper has been conducted primarily in the United States. Work outside the U.S. has not been excluded, but is not covered systematically. Finally, research on mechanical design in very specific technical domains (e.g., mechanisms and heat exchangers) is not covered unless it is clearly extendible to other mechanical design domains.

This review is organized into six sections based on our current view of the active design theory and methodology research areas. These six areas are:

1. Descriptive models of design processes 2. Prescriptive models for design 3. Computer-based models of design processes 4. Languages, representations, and environments

for design 5. Analysis to support design decisions 6. Design for manufacturing and other life cycle is-

sues such as reliability, serviceability, etc.

These six categories are certainly not mutually ex- clusive, and some research overlaps two or more areas. In such cases, we have done our best to inform readers where research projects have been placed. In Part I, the first three of the above six topics were reviewed. In Part II, we review the last

122 Finger & Dixon: Research in Mechanical Engineering Design

three, beginning with languages, representations, and environment for design.

5 Languages, Representations, and Env~onments

In some areas of engineering design, such as circuit design, formal representations exist for the artifacts being designed which capture their important physical, functional, and logical attributes. A fundamental concern in mechanical engineering design research is that complete representations do not exist for mechanical artifacts. Intensive effort over the last fifteen years has resulted in the creation of valid, robust computer-based models for the geometry of mechanical designs. However, except in limited domains such as kinematic linkage design, no formal representation exists for the physical and functional attributes of mechanical designs. This section discusses research in mechanical engineering design that has begun to address this concern.

Another related topic is the environment within which the designer works and within which the design evolves. Currently, many of the tools used to create designs, whether computer- or paper-based, are incompatible with one another, so a design may be transformed from one representation to another many times as it evolves. In addition, even if the design tools all used a common representation, the coordination and interaction of the tools with the designer is still an open research issue.

5.1 Representation of Form

The representation of the geometric form of a mechanical design has received much attention, largely through the emergence of computer-aided design systems. We discuss two different, but con- verging approaches to the representation of form. The first approach is geometric modeling, either boundary representation (b-rep) or constructive solid geometry (CSG) in which the objective is to create a valid, computer-based representation of a solid object. The other approach is shape grammars, and their extensions, in which the goal is to create geometric rules (a grammar) by which a class of objects can be generated or described.

5.1.1 Solid geometric models. Requicha and Voelcker [112] cover the progression from the early CAD systems, which merely duplicated the lines that would have been drawn on a blueprint, through wire-frame models, through to solid modetlers, in which complete, valid solid objects are represented. This progression is of interest to those in design

research, because the same need--that of increas- ing the expressiveness of the representation-- drives much of the research in design representation. Voelcker [145] also discusses the limitations of the current geometric models as design systems because their purpose is to represent the geometry of a completed geometric object, rather than an evolv- ing one. A discussion along similar lines can be found in Nielsen [94].

One approach to creating geometric modeling systems for design is to use variational geometry. Gossard [50, 77, 81] combines CSG and boundary models in an object graph so that changes in dimensions result in changes in geometry and topology. Variational geometry is most useful for redesign and tolerance analysis and synthesis.

Recently, non-manifold geometric modeling systems have been created by Weiler [146, 147] and by Prinz et al. [56]. These non-manifold systems are promising as the underlying geometric modellers for design systems because one-dimensional, two-dimensional, and three-dimensional geometric entities can be represented in a uniform fashion. In addition, these models contain topological information that enables high-level descriptions of features. (See Section 5.3.)

5.1.2 Shape grammars. In 1975, Stiny [126] created shape grammars based on the formalisms of computational linguistics [28]. Using a formal grammar, instances of a class of objects can be generated based on a sequence of production rules. Architects in particular have been interested in shape grammars, using them to generate a family of floor plans or ornamentation. For example, Flemming [48] has used a variant of shape grammars to generate fa- cades and floor plans for new buildings so they would blend into a historic district. Tutorials on shape grammars can be found in both Earl [40] and Stiny [127]. The textbook, An Introduction to For- mal Language Theory [89], which unites formal language theory with an introduction to computational linguistics, is a good starting point for design researchers interested in formal languages.

Researchers from several different areas have become interested in using the formalism of grammars to describe, generate, and parse designs. For example, Woodbury [156] has created a structure grammar that extends shape grammars to structures in space, and he is now working on a three-dimensional grammar for solids [60]. Stiny [129] has written about possible extensions to his work that would use grammars to generate design attributes other than simply shape.

Fitzhorn [47] shows the formal relationships between language theory and solid modeling systems.

Finger & Dixon: Research in Mechanical Engineering Design 123

He proves that a two-dimensional grammar that is a variant of a graph grammar can produce three-dimensional solids. He creates three grammars, one of which generates the constructive solid geometry representation, the second of which generates the boundary representation, and the third of which generates plane models.

Based on Fitzhorn's work, Pinilla [102] has created a grammar that can be used to parse the geometric features of a design. He uses a non-manifold topological representation of a design which enables a general, but formal, representation of form features. His work is discussed in greater detail in Section 5.3.

5.2 Representation of Behavior

The formal representation of the function and behavior a of mechanical designs has been explored by, among others, Pahl [95], Crossley [31, 32] and Lai [76]. Each takes a distinctly different approach to the problem. Crossley has developed a graphical system for laying out the mechanical functions of a design. In his system, functions such as "dump" or "orient" are each assigned graphical icons. The icons can then be arranged in a graph to represent the overall function of the design. Crossley suggests that each icon might have associated with it a list of possible mechanisms that would provide the required function. Because the icons do not have any deeper structure, the functionality of the design layout cannot be checked. In addition, he does not address the problem of integrating functions in the physical components. In contrast to Crossley's graphical system, Lai has created a formal, English language-based system called FDL for representing the function and structure of mechanical designs. In FDL, nouns and verbs are used to create sentences that represent the function of a design, and design rules operate directly on the nouns and verbs in the sentence. Allowable verbs (for example "fasten") do not have physical or mathematical representa-

Mechanical engineers tend to use the words function and behavior interchangeably. Qualitative physicists make a distinction between these words; that is, the design's function is what it is used for, while its behavior is what it does. For example, two bolts may each have been designed to function as fasteners, be made of the same material, and have the same geometry, but if one is cast and the other machined, they will have different behaviors. Another example is that a motor may be designed to function as a power transformer, but it can also function as a door stop because it has additional behaviors due to its mass. Because function and behavior are used interchangeably in mechanical engineering, we will not distinguish between them. Un- less otherwise noted, function is used in the sense of the behavior of the design.

tions and so their meaning is determined by the rules that use them.

Ishida et al. [66] describe a system for detecting unanticipated functions of machines, such as leak- age or the impossibility of disassembly based on the Takase's Feature Description language [136]. Their goal is to create a computer simulation based on a human designer's problem-solving activity.

Fenves and Baker [45] present a spatial and functional representation language for structural designs. They use operators that execute a grammar (like the grammars described in Section 5.1.2) to generate architectural layouts as well as structural and functional configurations; however, they must assume that the layout and structure are independent if they are generated sequentially.

Ulrich and Seering [140] use a formal representation of function based on bond graphs [98]. Using a strategy of design and debug, they transform each component in the graph that represents the design requirements directly to functionally independent physical components. Reconfiguration for function sharing is performed "after the components have been selected. Ulrich and Seering have extended the approach above to the conceptual design of dy- namic systems [139, 141]. A system has been developed that prepares a schematic description of a system of functional components to meet a given behavioral specification. From the schematic, an initial physical system is developed by substituting devices for each function. Finally, iterative redesign (they call it debugging in this case) is used to improve on the initial design. Bond graphs are em- ployed to represent the design. In [113, 114], Rind- erie also uses a representation for function based on bond graphs; however, his focus is on how the function graph can be transformed and then mapped into different physical systems. Of primary concern is that physical components always exhibit behavior in addition to the behavior for which they were selected. For example, in addition to providing power reduction, a gear pair has a mass and a geometric configuration.

Joskowicz [70, 71] presents a method for designing kinematic mechanisms based on functional specifications. Using configuration spaces, he has created a method which enables explicit reasoning about the relationship between the structure and the function of the objects. While the domain is limited to kinematic linkages, this system begins to address one of the major open questions in design; that is, the relationship between the desired functionality for a design and it's final shape.

Green and Brown [51] present a qualitative model for reasoning about the shape and fit during the design process. They are concerned with how


surface features of a design are grouped, oriented, and matched until the designer can attempt to con- firm a fit. Bacon and Brown [11] present a top-down approach to reasoning about the behavior of mechanical devices that uses analogy and knowledge about the behavior of already understood devices. Their goal is to model, using a computer, the process by which a human engineer would discover the behavior of a device given some formal description of its structure.

5.3 Feature-Based Representations

While there is no consensus on a precise definition of a.feature, most researchers working in the area agree that a feature is an abstraction of lower-level design information. Abstractions of design information are becoming of greater importance as design systems evolve. The research in feature-based design systems has been motivated by the realization that geometric models represent the design in greater detail than is useful for designers, process planners, assembly planners, or for rule-based systems that emulate these activities. The concept of features began with form features. Form features are associated with the surface of parts, especially machined parts and include holes, bosses, and ribs. In recent work the concept has been made much more comprehensive.

An early paper by Wesley et al. [148] discusses the need for a higher-level language for describing assemblies, tools, and assemblers. In another paper, Pratt [109] discusses the role of solid modeling as the interface between design and manufacturing. In his paper he presents feature-based process planning systems in which form features are the bridge between the geometry created by the designer and the process plan. Pratt and Wilson [110] give a detailed discussion of the requirements for a solid modeling system to support form features. In a later paper, Pratt [11 l] makes specific recommendations for the attributes that a geometric modeller should have to be feature-based.

Dixon [33] has defined a feature as "any geometric form or entity that is used in reasoning in one or more design or manufacturing activities," and more recently [38] as "an entity with both form and function." A similar definition emerged from a recent workshop on Features in Design and Manufacturing [128]. There, a feature was defined to be "a relationship among a set of elements of a design."

Thus, features are not limited to being geometric entities nor are they limited only to design and manufacturing, although most of the research to date has been on geometric features for design and manufacturing. Feature-based representation can be ob-

tained by feature extraction (see Section 5.3.2), from an existing CSG or boundary representation, or by designing with features from the outset.

5.3.1 Feature-based design systems. Dixon et al. [38] have developed feature-based design systems in which the designer is provided with a set of design-with features. These features arise from the combination of activity and process, Ibr example design (activity) of castings (process) gives rise to a set of primitive features such as hollow box, slab, corner, and boss or hole. The systems developed are described in more detail in Section 7.5. A tenta- tive taxonomy of design-with-features and a discussion of the origin of features is described in [33], and an architecture for a design-with-features system for components is also presented in [37].

Cutkosky and Tenenbaum [34,351 have created a system called FIRST-CUT in which a product and its production process are designed simultaneously. This system is a feature-based system, and the part is created by applying machining operations that create manufacturing features, such as slot or hole, in the part. The process is essentially one of "de- structive solid geometry" since the part is created by removing material.

5.3.2 Feature extraction. Most of the research in feature extraction has been for process planning, although some research has been done on features for other types of analysis such as the work by Woo [151] for finite element analysis. In either case, the focus of the work described in this section is on extracting manufacturing form-features from a previously defined geometric model. Once the features have been extracted, the design can be analyzed for manufacturability, and previously compiled plans can be retrieved to create the required features. A review of current feature-based process planning systems can be found in [142]. Among these feature-based process planners are Henderson, [61, 62], Choi, [27], Kumar et al. [75], and Hayes [59],

The Quick Turnaround Cell (QTC) at Purdue [23] connects a feature-based design system, an automatic process planner, and a manufacturing cell. In this system, the features are manufacturing form- features, and the emphasis is on rapid prototyping of parts, rather than on the design process itself. Roy and Liu [116] present a feature-based representation that is a hybrid CSG/B-Rep data structure to represent dimensioning and tolerancing. Again, the model is constructed from form-features.

Sakurai and Gossard [117] present a procedure for recognizing shape features in 3D solid models~ They use a feature graph that is a b-rep subgraph and what they call facts which possess characteris-


tic combinations of topology and geometry. They use graph matching to find features; however, their feature graphs are not given by a grammar, but by instance enumeration.

The feature recognition system described by Pinilla [102] is currently being extended to enable feature-based designs to be generated, represented, and parsed. This extension is possible because the underlying representation of a feature is based on elements of a well-defined grammar; however, com- binatorial explosion in the generation and search presents a major obstacle to practical applications.

In all the feature extraction models, feature interaction is a difficult problem; that is, even if the system is capable of recognizing a hole and recognizing a slot, it may not be capable of recognizing a hole in a slot. Some of the work being done in graph-based topological grammars may solve this problem in theory, but practical solutions are not close at hand.

5.4 Product Models

In 1981, Eastman [41] pointed out that computers were no longer just a vehicle for the analysis of designs, but had become a medium for the representation of designs. He predicted that computers would eventually replace traditional media such as paper and pencil, and he discussed the superiority of computers for geometric modeling, semantic in- tegrity, and abstraction hierarchies. This paper was among the first to discuss the idea of an integrated product model, as opposed to a CAD database, for mechanical designs.

Since the early eighties, researchers have worked to create integrated models that combine representations of geometry, semantic knowledge, and engineering models in what have come to be called engineering databases or product models. Among those working in this area are Maryanski [100], Shaw [120], Spooner [124], Su [131], and Su- zuki et al. [135].

The Product Data Exchange Specification (PDES/STEP) is a new international standard for exchanging product information. PDES/STEP is a major extension beyond IGES (Initial Graphic Ex- change Specifications). Whereas the IGES standard is concerned with exchange of information intended for human interpretation (e.g., drawings and wire- frame), the PDES/STEP standard is concerned with exchange of a complete product model intended for use by CAD/CAM systems (e.g., process planners, NC path generators, and others). Because this standard is being coordinated with international standards groups and is likely to be adopted internation- ally by industry, the PDES/STEP development is of interest to designers and design researchers. Plan-

ning is well along for the standards for mechanical product models and printed wiring board data. A first version, including some consideration of form features, will be available in 1989 [99].

5.5 Environments

The problem of creating an environment within which designers can work is not limited to computer-based systems. Much of the work on prescriptive models of the design process, discussed in Part I, Section 3, is directed toward organizing the information available to designers as well as controlling and coordinating the methods and tools used by them. The environment becomes more important when the design system is computer-based. Even if the design tools all use a common representation and data base, the coordination and interaction of the tools with each other and with the designer is still an open research issue.

Shah and Wilson [119] discuss the mismatch between current CAD tools and the needs of designers. They state that designers need multiple levels of abstraction, generalizations of geometry, product definition models, and better visualization tools. In a similar paper, Logan [82] cites the same types of mismatches and requirements for architectural CAD systems.

Habraken [57] has created a design environment based on the analogy that design is like a game. Using this analogy, Habraken creates a constrained, but rich, universe in which design concepts can be explored. The idea of a game provides a conceptual framework that can be used to study how designers interact with the design problem, with their environment, and with each other. In related work, Gross et al. [53-55] have created a Con- straint Manager design environment that is based on the model that design is search within a constraint space. The environment enables the designer to navigate through the constraints on the design.

Arbab [4-7] is working towards an intelligent CAD system in which a tool box of automated problem solving aids allow designers to conceive, evolve, and document their designs. Arab has focused on the explicit representation and manipula- tion of geometric knowledge. Papers and abstracts from researchers working in the area of CAD environments can be found in the proceedings from meetings of IFIP Working Group 5.2, particularly the series of workshops on Intelligent CAD [63-65].

Researchers from the field of artificial intelligence, interested in the field of design research, have begun to explore system architectures for design. For example, Fox [46] and Millington [86] addressed the issue of integrating design represen-


tations and design tools in a unified architecture. The environments associated with distributed design problems are discussed in Part l, Section 4.4.

5.6 Summary

The representation of the geometry of mechanical designs is highly developed and systems are widely available, although there are still questions of which system or combination of systems are appropriate for different design tasks [94]. However, if the design task requires more than low-level geometry of an object; that is, if it requires knowledge of how features are connected, or how the design was intended to behave, or how it does behave, or how material properties affect behavior, there are no tools at hand to aid the designer.

Both Dixon's and Cutkosky's systems are true design-with-features systems in that the designer can compose and edit the design based on the feature representation. However, in both systems the features are, for the most part, based on manufacturing processes. There are still open issues whether designers can create designs using manufacturing features and whether designs composed from manufacturing features can be used by other models that address assembly, maintenance, and other concerns.

The systems created by Fenves and Barker, Ul- rich and Seering, and Rinderle each have a underlying formal grammar, whether implicit or explicit, that enables the designer to represent the behavior of the design. However, many aspects of the behavior of mechanical designs cannot be modeled except in large analytical programs. In addition, the transi- tion from desired behavior to design description can be made in only a few domains such as mechanism design.

The preliminary design-with-features systems enable designers to compose designs from higher- level entities; however, there are still many open issues. For example, it is unclear whether a general framework based on features will enable designs to be interpreted from many different points of view, or whether features can be used in design systems to capture the behavioral attributes of a design.

6 Analysis in Support of Design

Analysis is an important element of design; without analysis to provide accurate evaluations of ex- pected design performance, designs would be based on, at best, guesses and heuristics. Traditionally, the distinction between design and analysis has been blurred, and analysis often subsumes design. To be sure, trial designs must be evaluated, and

engineering analysis procedures provide one of the most important means for evaluation. Analysis yields quantitative information about the performance of a design that can guide design or redesign decisions. However, it is now more widely recog- nized that analysis supports design, and not the re- verse.

Much attention is currently being focused on the realization that design and redesign decisions must take into account issues of manufacturability and life cycles concerns such as reliability, maintainabil- ity, disposability, and other so-called ,ilities." In this paper, design for manufacturing and other life cycle issues is reviewed in Section 7. Here in Sec- tion 6 we consider research more specifically related to the design-analysis interface where "analysis" means engineering analysis for predicting results such as stresses, deflections, heat flow, mo- tions, fatigue, efficiency, and the like. Interfaces and access to optimization methods and finite element programs are included here, while analysis methods for assembly are included in Section 7.

6.1 Interfaces to Optimization Methods

As noted in Part i, Section 4.1 the development of an appropriate criterion function is often an impedi- ment to the use of optimization methods for design. This has led to research that attempts to provide more designer-oriented interfaces to existing optimization procedures.

The research on design optimization interfaces at Brigham Young University is embodied in a program named OPTDES.BYU [13-16, 42, 96]. The program provides a powerful knowledge-based interface that assists designers in formulating optimization problems and interpreting the results. An- other approach has been developed by Mistree et al. [72, 87, 88, 90]. They have developed a decision support problem technique "that includes expert systems to assist students in formulating problems for their adaptive linear programming methods." Many specific examples have been demonstrated.

Research applying symbolic computation to reduce the complexity of optimal design problems has been done by Agogino et al. [1, 2]. In a program called SYMON [29], monotonicity analysis is used to reason qualitatively about the nature of constraints and their influence on design solutions. Results in effect reduce the size of the search space. Output from SYMON can be used as in put to another program, called SYMFUNE, that reasons with the constraint equations to further confine the search.

Chieng and Hoeltzel [25, 26] have designed and


implemented a design and analysis tool for mechanical components and assemblies called OPTDEX (Optimal Design Expert). Design cells are created that support the design of various elements such as bearings or speed reducers. The concept in this research is to provide an environment that integrates AI, mechanical design knowledge, and optimization methods.

Another interface for mechanical designers to optimization is described by Ishii and Barkan [67]. They propose a rule-based sensitivity analysis methodology that uses a table of production rule relationships between design variables and performance parameters. The approach provides interactive advice about critical constraints during the parametric iterative redesign process and about formulating problems for optimization.

Other work that provides assistance to designers using optimization methods is found in Balachan- dran and Gero [12]. In this work, knowledge-based systems are described to assist formulation and selection of optimization algorithms. Diaz [36] describes and illustrates an approach based on fuzzy set theory that enables a richer, more flexible definition of the criterion function than traditional optimization methods. Additional references that provide designer interfaces to optimization include [22, 85, 103, 115]. Haftka [58] gives a review of structural shape optimization methods. Some possible dangers of structural optimization techniques are discussed by Thompson [137]. A good review of optimization methods for large-scale systems can be found in [8].

Finally, methods for using optimization methods in the presence of the complex concerns of design such as cost and delivery time are discussed by Na- kazawa [92] and by Mackenzie [84]. Nakazawa's work is interesting because he uses as his objective function the minimization of information required in manufacturing [132].

6.2 Interfaces to Finite Element Analysis

Designers need convenient and timely access to appropriate analytical procedures. For those procedures that are too complex, sophisticated, or new for designers to perform themselves, convenient or even automated interfaces are required. In many companies, this has been provided by creating a group of analysis specialists, often called the Engi- neering Department. The interface in this case is a human one, and we do not know of research that has studied designer-analyst interaction. There are, however, efforts to develop computer-based interfaces to the more complex analysis computer pro-

grams. Success with these efforts can lead to new, practical tools for designers that will make access to reliable analytical results easier and hence more readily usable for early design decisions.

Shephard [121] reviews the state of automatic generation of finite element meshes in 1983. More recently, Kela [73] describes an experimental system to generate 2-D meshes from CAD data bases and to redesign the mesh automatically until a satis- factory analysis is obtainable. Both these papers review the other literature related to automated finite element mesh generation.

6.3 Analysis at Early Design Stages

Most engineering analysis procedures require a complete description of the design to be analyzed. This makes them applicable only during the parametric design phase. How, then, do we evaluate designs at the earlier stages of design?

Wood and Antonsson [152-155] make use of fuzzy set theory to aid preliminary design decisions with analysis tools developed for computations on imprecise parameters. Examples applying the approach to beam design and brake design are presented in [153]. Rinderle's work [113], see Part I, Section 4.3, incorporates analysis integrally into the configuration design process. Gelsey [49] describes two programs related to automatic analysis of mechanisms that recognize and simulate kinematic parts automatically from a CAD data base.

Other papers have addressed the issue of preliminary design analysis. Libardi [80] describes the requirements for a system to support analysis of incomplete and abstract designers and analysis in different functional domains. Cline [30] discusses a system under construction that will support analysis of in-progress designs by providing designers with a number of convenient options for creating and using analytical models. Dym [39] describe an environment, currently being implemented, that assists structural designers in choosing analysis procedures at various stages of the design process.

The development and maintenance of a symbolic representation of the design is critical to this approach. Shephard [122] presents a discussion of the issues involved in analysis for design at early stages of the process. Jones [69] has developed a small system that selects and applies analytical models, such as cantilever beams, and thin plates, automatically. The system uses a feature-based representation of the design and considers the accuracy and purpose of the analysis in making a selection. How- ever, this work is just a beginning to the research required in this area.


6.4 Summary

Once a design has been carried to the detailed design stage, analysis procedures are available to predict or simulate the performance of the design along many different dimensions. Better interfaces to these procedures are necessary to make them more accessible to designers and to enable them to be used properly. However, a much greater need exists for better analytical tools in the early stages of design when critical decisions are made based on qualitative information. Tools and methods are needed to enable designers to explore alternatives fully and efficiently. Designs must be evaluated and analyzed at every stage from conceptual to detailed design. At the moment, little is known about how to do this, although the work noted above is an en- couraging beginning.

7 Design for Manufacturing and the Life-Cycle

Until recently, designers have been perceived to be concerned primarily with function and fit. Other issues were of lesser concern. In particular, the design implications of manufacturing, that is, ease of manufacture, process planning, and inspectability as well as other life-cycle issues such as serviceability, disposability, were Considered only after important design decisions and commitments were made. This practice has led to many less than optimal designs when the entire life of a product--from con- ception to disposal--is considered. Awareness of the economic cost associated with this practice has now led to growing interest into what is variously called "design for manufacture," "concurrent design," "simtdtaneous engineering," or "design-for- X," where X can stand for any or all of the life cycle issues that are relevant to the total life cycle value of an artifact.

7.1 Concurrent Design

Traditionally, the decisions that are made between the time a new product is conceived until the time it is shipped have been sequenced and compartmen- talized. One reason for this is simply that so much knowledge is required to design for all life-cycle issues that no one person or small group can know everything required. The traditional design sequence has now hardened into institutional structures, accompanied by all the organizational and human inertia that this implies. Thus, research into designing for the life cycle has the potential for pro- ducing major changes in the practice of engineering

design. It is possible to view research in life-cycle design from two, not totally independent, perspectives: 1) studies related to knowledge, and 2) studies related to process. The first perspective focuses on acquiring, organizing, and utilizing knowledge of life-cycle issues that relate to early design decisions 2. The second perspective focuses on organizing and controlling the design processes to enable early, concurrent consideration of life-cycle issues.

Finger et al. [46] describe a system called Design Fusion which is based on three underlying concepts: integrating life-cycle concerns through the use of views from multiple perspectives, where each perspective represents a different life-cycle concern such as manufacture, distribution, maintenance, etc.; representing the design space at different levels of abstraction and granularity through the use of features, where features are the attributes that characterize a design from the viewpoint of any perspective; and using constraints to guide the design.

A comprehensive view of concurrent design is presented by Whitney et al. in "The Strategic Ap- proach to Product Design" [149]. The authors propose a method of organizing the design process that focuses on assembly as the integrating activity, which can serve to bring all the various life cycle issues into communication and interaction. Ex- amples of cases are also presented in which the manufacturing "process is the design" or in which manufacturing process decisions precede many functional design decisions.

One concept for concurrent design is to design products (or parts) and their manufacturing processes simultaneously. Pioneering work in this approach is reported by Cutkosky and Tenenbaum [34, 35]. In the first of these papers, a system called First-Cut is described that enables designers to work in manufacturing modes in which manufacturing operations are specified as a means to design the desired part. In the second paper, the role of features in concurrent design is explored, with the con- clusion that "the combination of features and a process representation is the right foundation upon which to build a complete end-to-end design tool for addressing [functional, geometric, and manufacturing] constraints". Though the First-Cut implemen- tation of these ideas is limited to machining, the authors are also beginning to apply the concepts to injection molding.

z Knowledge of how to modify an almost complete, detailed design for some life-cycle issue is not necessarily the same as the knowledge needed at the conceptual or configuration level.


One method of implementing life-cycle design that combines these two perspectives is organizational change. All the various specialists, instead of acting separately and sequentially, are from the outset brought together to perform the design. This plan brings the knowledge possessed by all the life cycle experts to the same place at the time design decisions are being made. Research in organizational change and behavior is beyond the scope of engineering design research, but several reports and discussions have appeared in the engineering literature [20, 93]. Another smaller, less formal example of concurrent design is reported in [118].

It should be noted that bringing together experts on life-cycle issues does not insure that knowledge about making design decisions and compromises will also be available. We must distinguish between the specialist's knowledge of a life cycle issue and knowledge about creating and modifying early design concepts so that the life-cycle concerns are re- solved. Whitney et al. [149] argue that, by relating all decisions to assembly concerns, including the function of the assembly, the needed focus will emerge. However, it is not certain that a team of specialists will have, for example, the knowledge to set machined, molded, or cast tolerances to opti- mize a part considering function, reliability, serviceability, manufacturability, etc. Explicit knowledge of the relationships of life-cycle issues to early design decisions is needed to perform life-cycle design. Again, this relates directly to the question of the evaluation and analysis of designs at the configuration and conceptual stages.

7.2 Design for Manufacturing

Boothroyd and Dewhurst [17-19] have performed pioneering research on the accumulation and organization of knowledge of handling and assembly directly related to design. This work is based on the hypothesis that a small number of abstract features of the components in an assembly can be used to predict, with useful accuracy, the time required for assembly. Both manual and automatic assembly are considered. The features include specified aspects of part size and symmetry. The predictions of handling and assembly times can be used to point to needed design changes from the viewpoint of assembly.

In other design for assembly research, Poli and his colleagues [105-107] have developed a spreadsheet approach to rating designs on the basis of their ease of automatic assembly. The results point to part and product features that tend to increase assembly costs.

The systems described above require that the designers compute and enter manually the required data on size, symmetry, and other features. Myers [91] describes an algorithm which, when an assembly is designed in a geometric solid modeller, automatically computes the manual handling times of the various components using Boothroyd's theory and data. In this work, the features needed are extracted from the solid modeller boundary representations. This automation of manual handling analysis related to design has not yet been extended to automatic handling or to insertion times.

In other design for manufacturing work, Poll [74, 104] has compiled and organized knowledge on design for forging. As with Boothroyd's work in assembly, analyses of forging relative cost and diffi- culty are based on identification of selected design features, and the results point to potential design problems or improvements from the viewpoint of the forging process. Work by these researchers is in progress on design for injection molding [108].

Heuristic information is available from firms and industry associations related to design for manufacturability. For example, for casting there is [21], for extrusion [3], tbr forging [104], and for injection molding [108]. However, this type of knowledge is not yet embedded in CAD and solid modeling systems in a way that makes it available to designers using these systems.

In work that is similar in spirit to Suh's axiomatic approach to design, Ayers [I0] discusses manufacturing as the concentration of information in matter. While he does not discuss design per se, Ayers sees the optimal design and manufacturing process as the one that maximizes the economic value by mini- mizing the information required to describe and produce a product. An overview in design for manufacture is given by Stoll in [130].

7.3 Tolerances

Although tolerances are critical to both functional performance and manufacturing cost, tolerances have received very little theoretical treatment. There are three areas for research: 1) the relationships between tolerances and cost, 2) the relationships between tolerances and functional performance, and 3) the representation of tolerances in computer-based design systems.

Published data on the relationships between tolerances and costs are almost non-existent. Chase [24] has fit cost-tolerance curves to data published by Jamieson [68]. Work is in progress to analyze and publish more data that can perhaps provide the


basis for theory or, at the least, some quantitative generalizations.

A few researchers [78, 101, 123, 125, 134, 150] have studied how to synthesize tolerances in order to minimize manufacturing cost based on various assumed models for the tolerance cost relationship. These approaches employ optimization methods to minimize an assumed cost function.

Research into the effects of tolerances on functional performance is even more limited. Evans [43, 44] describes a possible theoretical approach to the problem, but the theory is not developed.

The assignment of tolerances can be viewed as one of assigning values to attributes; that is, as the parametric design problem. As such, it is necessary to be able to analyze the effects of tolerance stack- up in complex assemblies. There are several methods for doing this analysis as described by Green- wood [52] and also by Turner [138].

7.4 Design for Other Life Cycle Issues

Design for manufacturing (as well as for function, of course) is the most active design-for-X research field; research results for other X's are scarce. Suri [133] has proposed and is working on Design for Analysis, that is, designing products and manufacturing systems so that they can be easily analyzed. His argument is that analysis is another process, just like manufacturing or assembly, that a design must undergo. Therefore, just as one designs for manufacture or designs for assembly, one should design for analysis.

A detailed architecture for a "unified life-cycle engineering" (ULCE) environment is presented by Brei et al. [20]. This report also recommends research and development on the following life-cycle issues: (a) human interactions in design, (b) theory, methodology, and tools for design, (c) data-base management for design, (d) user interfaces, and (e) automatic management of detail design changes.

Research on design for other life-cycle concerns, such as design for reliability, testability, maintain- ability, is much further advanced in fields such as electronics and software design than it is in mechanical design. Ayers [9] in an interesting position paper, discusses the relationships among complexity, reliability, and manufacturing. His thesis is that the manufacturing of mechanical products must evolve toward creating integrated, multi-purpose mono- liths, similar to integrated computer chips, if mechanical products are to reach the same levels of reliability and reproducibility. Fiering and Villa- marin [144] have studied designs that have failed in surprising ways to uncover the factors that lead to

unreliable designs. Koen et al. [97] have investi- gated techniques such as fault tree analysis to develop tools to aid designers in designing large complex systems.

7.5 CAD Advisors

CAD systems with embedded knowledge to provide designers with early on-line advice about manufacturing and life-cycle issues have been proposed and experimented with. All such advisory systems require a representation of the in-progress design in terms of features, whether obtained by feature extraction or by designing with features. Henderson [62] describes a feature extraction system for machined parts that provides intbrmation relevant to process planning.

Experimental designing with features systems have been developed at the University of Massa- chusetts by Dixon et al. In [143], rotationally sym- metrical parts are designed from features like disks, cones, and cylinders; the system provides automatic print-review level manufacturabitity advice. In [79], extruded parts are designed from wall and intersection features; the system provides an automatic interface to finite element beam analysis. In [83], cast parts are designed from macro-features like box, L-bracket, U-channel, and slab; the system provides advice about manufacturing limits, hot spots, and filling problems. Dixon [38] has proposed a general architecture for design-with features systems to provide manufacturing and life-cycle advice and redesign suggestions to designers during the design process. Many of the systems discussed in Section 7.1, such as the one by Finger et at., use similar architectures to integrate feature- based design and manufacturing advisors.

Turner and Anderson [138] have developed a feature-based design system for machined parts that couples fixturing, process planning, NC code generation. The system is used to produce parts quickly with very little operator intervention. An important aspect of this work is the inclusion of tolerance information with the feature representation.

7.6 Summary

To date, design-for-X has meant primarily design for manufacturing. Research in design for manufacturing is extensive, especially for assembly and machining. Research effort in knowledge acquisition and organization is still needed, as well as in practical ways to get the information to designers in a useful and timely fashion. In contrast, a fundamental understanding of tolerances is still lacking, al-


though research interest in this area is growing. One common thread in all of the work in life-cycle design is the need for better underlying representations of mechanical designs. A clear dependence exists between the research on features-based representations and the research on life-cycle design.

8 Summary

A research review should not only point to what has been done, but also to what remains undone. We summarize here by listing the accomplishments and the outstanding research issues, as we see them, organized by the six topic areas of Parts I and II in the review.

Descriptive Models

State o f the Art

1. Understanding of how mechanical designers create designs has increased through the body of data collected from protocol studies. The results of these studies will enable new design tools that to support designers. (Section 2.1).

2. Preliminary hypotheses on the strategies used by designers have been generated, and from these, cognitive models of some of the skills used by designers have been created. (Section 2.2).

3. Understanding of how teams of designers work and interact has increased. Research in computer-supported cooperative work and on distributed problem solving complement the work in this area. (Sections 2.3 and 4.4).

Outstanding Research Issues

1. Hypotheses concerning the strategies used in design must be tested, validated, and integrated into design systems.

2. Cognitive models of design strategies must continue to be developed to increase our understanding of how designers design and as a basis for tools in conceptual designs.

3. Most product designs are created by teams of designers, and yet we know little about how design teams work or how to decompose a design problem to be solved by a team.

Prescriptive Models for Design

State o f the Art

1. Prescriptive models of the design process are used widely in teaching design and have been

successful in helping designers to organize the stages of preliminary design. (Section 3.1).

2. Morphological analysis has been successfully used for many years in configuration design. (Section 3.2).

3. The prescriptive models of both Taguchi and Suh are being applied in practice and have resulted in less expensive and more robust designs. (Section 3.3).


1. The prescriptive models of the design process make intuitive sense to many designers, but more research is needed to validate the methods and to integrate them with computer-based methods.

2. The mapping between the requirements of a design and the attributes of the artifact is not understood. Because the goal of designing is to create artifacts that meet the functional requirements, more fundamental research is needed on relating the attributes of designs to functional requirements, that is, on prescribing the artifact.

Computer-Based Models o f the Design Process

State o f the Art

1. Successful models for parametric design have been demonstrated. Progress has been made in understanding the crucial role for knowledge about dependencies between design variables and performance parameters. (Section 4.1).

2. Successful initial models for configuration design have been demonstrated pointing out the key role of features at this level. (Section 4.2).

3. The foundation has been laid for tools to support computer-aided design of mechanical assemblies. (Section 4.2.1).

4. Preliminary successes have been reported in some domains in designing from functional requirements. (Section 4.3).


1. The models and methods for parametric design are highly domain dependent. Research on a unifying parametric design paradigm, which must include both numeric and non-numeric methods, is needed.

2. Research is needed to enable evaluation and redesign of configurations without the need for instantiation at the parametric level.

3. The utility of strategies for distributed problem solving in design must be explored.


4. The role of physical principles in relating form and function is not yet fully understood.

Languages, Representations, and Environments

State of the Art

1. Geometric modeling is well advanced; robust constructive solid geometry and boundary-representation models are widely available. (Sec- tion 5.1.1).

2. New, geometric modeling paradigms based on non-manifold topologies, more suitable for designing, have been developed. (Section 5.1.1).

3. Formal representations of behavior for classes of mechanical designs have been created. (Sec- tion 5.2).

4. Research in feature-based representations has advanced rapidly in the last few years, and several feature-based design systems have been developed. (Section 5.3L

5. Integrated product models have progressed to the point where standards can be written for exchanging product data as opposed to graphical representations of engineering drawings. (Section 5.4).


1. A major research area common to all design problems is the representation of mechanical designs. The research issues include: representation of incomplete designs; representation of the evolution of a design, including design changes and version and configuration control; representation of non-geometric attributes of designs such as behavior and design intent; linkages and dependencies between representations of different attributes of designs; and integration of features, or high level abstractions, into the design representation. The role of formal grammars and languages in design representation must be explored further.

2. Much research remains before feature-based design systems can be used in practice.

3. An important area that has received little attention to date is the creation of design environments that integrate available tools into a con- sistent system to support the designer.

Analysis in Support of Design

State of the Art

1. Interfaces to optimization procedures have been created to make these powerful methods

useful and tractable in design systems. (Sec- tion 6.1).

2. Research in automatic finite element analysis has reached a stage where it is now practical to create interfaces between these powerful ana- lytic tools and design systems. In addition, studies are beginning to shed light on analysis for early design stages. (Sections 6.2 and 6.3).


1. A major research issue is the analysis and evaluation of designs at the early and intermediate stages of design. Research is needed on the generation and evaluation of alternatively concepts, embodiments, and configurations to complement the observed tendency of designers to pursue a single-concept design.

2. Another open research question is how to provide designers with the ability to design and analyze, not only at different levels of abstraction, but also from various functional viewpoints, for example from a kinematic, structural, or thermal viewpoint.

3. Research is needed to create CAD systems that support conceptual design stages by enabling designers to design, modify, and analyze at multiple levels of abstraction and in multiple viewpoints.

4. More work is required to complete and dissem- inate automated interfaces for parametric design and optimization as well as for detail design and finite element analyses.

Design for Manufacturing and the Life-Cycle

State of the Art

1. Concurrent design is under investigation on a number of fronts. Research is progressing on enabling multiple players to view, criticize, and modify a design, on enabling concurrent product and process design through the paradigm of process planning, and on enabling concurrent design through organizational change. (Section 7.1).

2. Much of the knowledge required to support design for manufacturability has been organized and is being disseminated. Design for assembly is especially mature. (Section 7.2).

3. Experimental manufacturability advisory systems on feature representations, have been integrated into CAD systems. (Section 7.5).


I. No theory or methodology exists to decompose a design into manageable design problems


and then to recompose and assemble the re- suiting designs into a product.

2. The organization and communication proto- cols necessary for concurrent design are not understood.

3. Continued acquisition and organization of manufacturing knowledge in forms useful to designers is needed.

4. Fundamental and applied work on tolerances, especially relating cost to performance, is es- sential.

5. More design-for-X studies are needed if concurrent design for life-cycle performance is to become a reality.

6. CAD advisory system must be able to deal with more complex geometry and with combinations of features.

The mechanical engineering design research community has made major advances over the last few years. Preparing this review was a much longer and harder task than we had anticipated. The research community in mechanical engineering design has made significant progress not only in advancing our understanding of design, but also in clarifying the research methods necessary to study design. The progress being made toward a better understanding of design, and hence toward better design tools, is remarkable.

References

1. Agogino, A.M. and Almgren, A.S., "Symbolic Computa- tion in Computer-Aided Optimal Design," Expert Systems in Computer-Aided Design, Gero, J.S., ed., North-Hol- land, Amsterdam, t987, pp. 267-284

2. Agogino, A.M. and Almgren, A.S., "Techniques for Inte- grating Qualitative Reasoning and Symbolic Computation in Engineering Optimization," Engineering Optimization, Vol. 12, No. 2, Sept/Oct 1987, pp. 117-135

3. A Guide to Aluminum Extrusions The Aluminum Associa- tion, Washington, DC, 1976

4. Arbab, F., Cantor, D.G., Lichten, L. and Melkanoff, M.A., "The MARS CAM-Oriented Modeling System," Proceedings of the Conference on CAD~CAM Technology in Mechanical Engineering, IFIP, Cambridge, MA, 1982, pp. 281-288

5. Arbab, F. and Wing, J., "Geometric Reasoning: A New Paradigm for Processing Geometric Information," IEEE In- ternational Symposium on New Directions in Computing, August 1985

6. Arbab, F., "A Paradigm for Intelligent CAD," Intelligent CAD Systems 1: Theoretical and Methodological Aspects, Springer-Verlag, 1987

7. Arbab, F., "An Environment for Geometric Reasoning in Intelligent CAD," 1FIP WG 5.2 Workshop on Intelligent CAD Systems, Gosard, D., ed., IFIP, Cambridge, MA, Oc- tober 1987

8. Arora, J.S. and Thandedar, P.B., "Computational Methods for Optimum Design of Large Complex Systems," Compu- tational Mechanics, Vol. 1, 1986, pp. 221-242

9. Ayers, R.U., 'Complexity, Reliability and Design: The Coming Monolithic Revolution in Manufacturing," Work- ing Paper WP-86-48, International Institute for Applied Systems Analysis, 1986

10. Ayers, R.U., "Manufacturing and Human Labor as Infor- mation Processes," Research Paper RR-87-19, Interna- tional Institute for Applied Systems Analysis, November 1987

11. Bacon, S.D. and Brown, D.C., "Reasoning about Mechani- cal Devices: A Top-Down Approach to Deriving Behavior from Structure," Computers in Engineering, 1988, Ameri- can Society of Mechanical Engineers, San Francisco, CA, August 1988, pp. 467---472

12. Balachandran, M. and Gero, J.S., "A Knowledge-Based Approach to Mathematical Design Modeling and Optimiza- tion," Engineering Optimization, Vol. 12, 1987, pp. 91- 115

13. Bailing, R.J., Parkinson, A.R. and Free, J.C., "OPTDES. BYU: An Interactive Optimization Package with 2DI3D Graphics," Technical report, Brigham Young University, 1982

14. Balling, R.J., Pister, K.S. and Polak, E., "DELIGHT- STRUCT: An Optimization-Based Computer-Aided Design Environment for Structural Engineering," Computer Meth- ods in Applied Mechanics and Engineering, No. 38, 1983, pp. 237-251

15. Balling, R.J., Parkinson, A.R. and Free, J.C., "Methods for Interfacing Analysis Software to Optimization Soft- ware," Computers & Structures, Vol. 22, No. 1, 1986, pp. 87-98

16. Bailing, R.J., Free, J.C. and Parkinson, A.R., "Consider- ation of Worst-Case Manufacturing Tolerances in Design Optimization," Journal of Mechanisms, Transmissions and Automation in Design, Vol. 108, December 1986

17. Boothroyd, G., Poll, C. and March L., "'Handbook of Feeding and Orienting Techniques for Small Parts," Tech- nical report, Mechanical Engineering Department, Univer- sity of Massachusetts, 1978

18. Boothroyd, G., Poll, C. and March, L., Automatic Assem- bly, Marcel Dekker, New York, 1980

19. Boothroyd, G. and Dewhurst, P., "Design for Assembly-- A Designers Handbook," Technical report, Department of Mechanical Engineering, University of Massachusetts, 1983

20. Brei, M.L., Cralley, W.E. Dierolf, D, Owen, D.J., Rich- ter, K.J. and Rogen, E., "Architecture and Integration Re- quirements for an ULCE Design Environment," Paper 17-2063, Institute for Defense Analysis, April 1988

21. Casting Design Handbook, American Society for Metals, Metals Park, OH, 1966

22. Cha, J.Z. and Mayne, R.W., "Optimization with Discrete Variables via recursive Quadratic Programming," Ad- vances in Design Automation--1987, American Society of Mechanical Engineers, September 1987

23. Chang, T.C., Anderson, D.C. and Mitchell, O.R., "QTC-- An Integrated Design/Manufacturing/Inspection System for Prismatic Parts," Computers in Engineering 1988, American Society of Mechanical Engineers, San Francisco, CA, August 1988, pp. 417-426

24. Chase, K.W., "Design Issues in Mechanical Tolerance Analysis," Manufacturing Review, Vol. 1, No. 1, March 1988, pp. 50-59

25. Chieng, W.-H. and Hoetzel, D.A., "An Interactive Hybrid


(Symbolic-Numeric) System Approach to Near Optimal Design of Mechanical Components," Engineering Optimi- zation, Grodon and Breach Science Publishers, Great Brit- ain, 1987, pp. 369-383

26. Chieng, W.-H. and Hoetzel, D.A., "A Generic Planning Model for Large Scale Engineering Design Optimization with a Power Transmission Design Case Study," Proceed- ings 1987 ASME Computers in Engineering Conference, American Society of Mechanical Engineers, New York, August 1987, pp. 113-124

27. Choi, B.K,, Barash, M.M. and Anderson, D.C~ "Auto- mated Recognition of Machined Surfaces from a 3-d Solid Modeler," Computer Aided Design, Vol. 16, No. 2, 1984

28. Chomsky, N., Syntactic Structures, Humanities Press, At- lantic Highlands, NJ, 1957

29. Choy, J.K. and Agogino, A.M., "SYMON: Automated SYMbolic MONotonicity Analysis System for Qualitative Design Optimization," Proceedings, ASME 1986 Interna- tional Computers in Engineering Conference, American Society of Mechanical Engineers, Chicago, IL, July 24-26, 1986, pp. 305-310

30. Cline, T., Abelson, H., Harris, W., Tou, I. and Fong, W., "Symbolic Computing in Engineering Design," Third AI in Design Workshop, AAA1, Minneapolis, MN, May 13, 1988

31. Crossley, E., "Make Science a Partner," Machine Design, April 24, 1980

32. Crossley, E., "A "Shorthand" Route to Design Creativ- ity," Machine Design, April 1980, pp. 96-100

33. Cunningham, J.J. and Dixon, J.R., "Designing with Fea- tures: The Origin of Features," Proceedings of the ASME Computers in Engineering Conference, American Society of Mechanical Engineers, San Francisco, CA, July 3 I-Au- gust 3, 1988

34. Cutkosky, M.R. and Tenenbaum, J.M., "CAD/CAM Inte- gration Through Concurrent Process and Product Design," Intelligent and Integrated Manufacturing Analysis and Synthesis, American Society of Mechanical Engineers, New York, 1987, pp. 1-10

35. Cutkosky, M.R., Tenenbaum, J.M. and Muller, D., "Fea- tures in Process-Based Design," Proceedings of the Inter- national Computers in Engineering Conference, American Society of Mechanical Engineers, San Francisco, CA, July 1988

36. Diaz, A.R., "A Fuzzy Set Based Environment for Design Optimization," Design Theory "88: Proceedings of the t988 NSF Grantee Workshop on Design Theory and Methodol- ogy, Springer-Verlag, New York, 1989

37. Dixon, J.R., Cunningham, J.J. and Simmons, M.K., "Re- search in Designing With Features," IF1P WG 5.2 Work- shop on Intelligent CAD Systems, Gossard, D., ed., IFIP, Cambridge, MA, October 6-7, 1987

38. Dixon, J.R., "Designing with Features: Building Manufac- turing Knowledge into More Intelligent CAD Systems," Proceedings of ASME Manufacturing International-88, At- lanta, Georgia, April 17-20, 1988

39. Dym, C.L and Salata, S.E., "Representation of Strategic Choices in Structural Modeling," Proceedings of the 1989 ASCE Congress, American Society of Civil Engineering, May 1989

40. Earl, C.F., "Shape Grammars and the Generation of De- signs," Principles of Computer-Aided Design, Pitman Pub- lishing, 1987

41. Eastman, C.M., "Recent Developments in Representation in the Science of Design," Proceedings of the 18th Design Automation Conference, ACM, tEEE, June 1981, pp 13-21

42. Ellsworth, R., Parkinson, A. and Cain, F., "The Comple-

mentary Roles of Symbolic Computing and Numerical Op- timization in Engineering Design Software," Advances in Design Automation--1987, Design Methods, Computer Graphics and Expert Systems, vol. I, Design Technology Conferences, Boston, MA, September 1987

43. Evans, D.H., "Statistical Tolerancing: The State of the Art, Part 1 ," Journal of Quality Technology, Vol. 7, No. 1, January 1975

44. Evans, D.H., "Statistical Tolerancing: The State of the Art, Part 2," Journal of Quality Technology, Vol. 7, No. 1, January 1975

45. Fenves, S.J. and Baker, N.C., "Spatial and Functional Representation Language for Structural Design," Expert Systems in Computer-Aided Design, Elsevier Science (North-Holland), IFIP 5.2 1987

46. Finger, S., Fox, M.S., Navinchandra, D., Prinz, F.B. and Rinderle, J.R., "Design Fusion: A Product Life-Cycle View for Engineering Designs," Second IFIP WG 5.2 Workshop on Intelligent CAD, IFIP, Cambridge, UK, 19- 22 September 1988

47. Fitzhorn, P., Formal Languages, PhD dissertation, Colo- rado State University, 1985

48. Flemming, U., "The Role of Shape Grammars in the Anal- ysis and Creation of Designs," 1986 SUNYBuffalo Sympo- sium on CAD; The Computability of Design, SUNY Buf- falo, NY, December 1986

49. Gelsey, A. and McDermott, D., "Spatial Reasoning About Mechanisms," 7'0 appear in Advances in Spatial Reason- ing, Ablex. 1989

50. Gossard, D.C., Zuffante, R.P. and Sakurai, H., "Repre- senting Dimensions, Tolerances, and Features in MCAE Systems," IEEE Computer Graphics and Applications, March 1988, pp. 51-59

51. Green, D.S. and Brown, D.C., "Qualitative Reasoning during Design about Shape and Fit: A Preliminary Report," Expert Systems in Computer-Aided Design, Elsevier Sci- ence (North-Holland), IFIP 5.2 1987, pp. 93-117

52. Greenwood, W.H. and Chase, K.W., "'A New Tolerance Analysis Method for Designers and Manufacturers," Jour- nal of Engineering for Industry, Transaetions of ASME, Vol. 109, May 1987, pp. 112-116

53. Gross, M. and Fleisher, A., "Design as the Exploration of Constraints," Design Studies, No. 3, 1984

54. Gross, M.D., Design as Exploring Constraints, PhD dissertation, M.I.T., 1986

55. Gross, M., Ervin, S., Anderson, J. and Fleisher, A., "Con- straints: Knowledge Representation in Design," Design Studies, Vol. 9, No. 3, July 1988

56. Gursoz, E.L. and Prinz, F.B, "Corner-Based Representa- tion of Non-manifold Surface Boundaries in Geometric Modeling," Technical Report, Engineering Design Re- search Center, Carnegie Mellon University, 1989

57. Habraken, N.J. and Gross, M.D., "Concept Design Games: Games for Experimentation in Design Theory and Methodology," Proceedings from the NSF Workshop on the Design Process, Waldron, M.B., ed., Ohio State Uni- versity, Oakland, CA, February 8-10, 1987, pp. 13-42

58. Haftka, R.T. and Grandhi, R.V., "Structural Shape Opti- mization--A Survey," Computer Methods in Applied Me- chanics and Engineering, Vol. 57, 1986, pp. 91-I06

59. Hayes, C.C. and Wright, P.K,, "Automating Process Plan- ning: Using Feature Interactions to Guide Search," Tire Journal of Manufacturing Systems, Vol. 8, No. 1, January 1989

60. Heisserman, J. and Woodbury, R., "'Solid Grammars for Generative Geometric Design," Submitted to the NSF En-


gineering Design Research Conference, University of Mas- sachusetts, Amherst, Amherst, MA, June 11-14, 1989

61. Henderson, M.R. and Anderson, D.C., "Computer Recog- nition and Extraction of Form Features: A CAD/CAM Link," Computers in Industry, Vol. 6, No. 4, 1984, pp. 315-325

62. Henderson, M.R. and Chang, G.J., "FRAPP: Automated Feature Recognition and Process Planning from Solid Model Data," Computers in Engineering 1988, American Society of Mechanical Engineers, San Francisco, CA, Au- gust 1988, pp. 529-536

63. IFIP WG 5.2 Workshop on Intelligent CAD Systems, Gos- sard, D., ed. IFIP, Cambridge, MA, October, 1987

64. Second IFIP WG 5.2 Workshop on Intelligent CAD Sys- tems, Yoshikawa, H. and Holden, T., ed., IFIP, Cam- bridge, UK, September, 1988

65. Third IFIP WG 5.2 Workshop on Intelligent CAD Systems, Yoshikawa, H., ed., IFIP, Osaka, Japan, to be held Sep- tember, 1989

66. lshida, T., Minowa, H. and Nakajima, N., "Detection of Unanticipated Functions of Machines," Design and Syn- thesis, Elsevier Science (North-Holland), 1985, pp. 23- 28

67. lshii, K. and Barkan, P., "Rule Based Sensitivity Analy- sis--A Framework for Expert Systems in Mechanical De- sign--," Proceedings of the IFIP W.G. 5.2 Working Con- ference on Expert Systems in Computer-Aided Design, IFIP, Sydney, Australia, 16-20 February 1987

68. Jamieson, A., Introduction to Quality Control, Reston Pub- lishing, 1982

69. Jones, J.C., "Continuous Design and Redesign," Design Studies, No. 1, 1983

70. Joskowicz, L. and Addanki, S., "From Kinematics to Shape: An Approach to Innovative Design" 7th National Conference on Artificial Intelligence, AAAI-88, Minneapo- lis, MN, August 21-26, 1988

71. Joskowicz, L., "Reasoning about Shape and Kinematic Function in Mechanical Devices," Robotics Research Technical Report 402/173, New York University Courant Institute of Mathematical Sciences, September 1988

72. Kamal, S.Z., Karandikar, H.M., Mistree, F. and Muster, D., "Knowledge Representation for Discipline-Indepen- dent Decision Making," IFIP 87, IFIP, Australia, 1987

73. Keta, A., Perruchio, R. and Voelcker, H., "Toward Auto- matic Finite Element Analysis," Computers" in Mechanical Engineering, July 1986

74. Knight, W.A. and Poli, C., "'A Systematic Approach to Forging Design," Machine Design, January 24, 1985

75. Kumar, B., Anand, D.K. and Kirk, J.A., "Knowledge Representation Scheme for an Intelligent Feature Extrac- tor," Computers in Engineering 1988, American Society of Mechanical Engineers, San Francisco, CA, August 1988, pp. 529-536

76. Lai, K. and Wilson, W.R.D., "FDL: A Language for Func- tion Description and Rationalization in Mechanical De- sign," Computers in Engineering, ASME, New York, 1987, pp. 87-94

77. Lee, K. and Gossard, D., "A Hierarchical Data Structure for Representing Assemblies: Part 1 ," Computer-Aided De- sign, Vol. 17, No. 1, 1986, pp. 16-19

78. Lee, W.J. and Woo, T.C., "Tolerancing: Its Distribution, Analysis and Synthesis," Technical Report 86-30, Depart- ment of Industrial and Operations Engineering, University of Michigan, 1987

79. Libardi, E.C., Dixon, J.R. and Simmons, M.K., "Design- ing with Features: Design and Analysis of Extrusions as an

Example," Spring National Design Engineering Confer- ence, American Society of Mechanical Engineers, Chicago, IL, March 24-27, 1986

80. Libardi, E.C., Dixon, J.R. and Simmons, M.K., "Com- puter Environments for the Design of Mechanical Assem- blies: A Research Review," Engineering with Computers, 1988, pp. 121-136

81. Light, R. and Gossard, D., "Modification of Geometric Models Through Variational Geometry," CAD--Com- puter-Aided Design, Vol. 14, No. 4, July 1982, pp. 209- 214

82. Logan, B.S., "Representing Design Problems: Formal and Informal Models in Design," Second IF1P WG 5.2 Work- shop on Intelligent CAD, IFIP, Cambridge, UK, 19-22 September 1988

83. Luby, S.C., Dixon, J.R. and Simmons, M.K., "Designing with Features: Creating and Using a Features Data Base for Evaluation of Manufacturability of Castings," Proceed- ings, ASME 1986 International Computers in Engineering Conference, American Society of Mechanical Engineers, Chicago, IL, July, t986, pp. 285-292

84. MacKenzie, C.A. and Gero, J.S., "Learning Design Rules from Decisions and Performances," Artificial Intelligence, Vol. 2, No. 1, 1987

85. Mehta, S.I. and Korde, U.P., "An Expert System to Choose the Right Optimization Strategy," Computers' in Engineering 1988, American Society of Mechanical Engi- neers, San Francisco, CA, July 31-August 4, 1988, pp. 483-487

86. Millington, K., Robertson, A. and Smithers, T., "Architec- ture for AI-Based Design: A Foundation Made Concrete," Preprints of the Second IFIP WG 5.2 Workshop on Intelli- gent CAD, IFIP, Cambridge, UK, September 1988, pp. 299-311

87. Mistree, F., Hughes, O.F. and Phuoc, H.B., "An Optimi- zation Method for the Design of Large, Highly Constrained Complex Systems," Engineering Optimization, Vol. 5, 1981

88. Mistree, F. and Muster, D., "Design Harmonization: A Computer-Based Approach for Design in the Systems Age," Optimization in Computer-Aided Design, Elsevier Science Publishers B. V. (North-Holland), 1985

89. Moll, R.N., Arbib, M.A. and Kfoury, A.J., An Introduc- tion to Formal Language Theory, Springer-Verlag, New York, 1988

90. Muster, D. and Mistree, F., "A Philosophy, Approach and Method for Negotiating Solutions to Design Problems Us- ing the Decision Support Problem Technique," Technical report, Department of Mechanical Engineering, University of Houston, 1988

91. Myers, W.L., Dixon, J.R. and Simmons, M.K., "Com- puter Analysis of Mechanical Assembly from a CAD Data Base: Manual Handling Times," Proceedings of the 1987 ASME Computers in Engineering Conference, American Society of Mechanical Engineers, New York, August 9-13, 1987

92. Nakazawa, H., "Information Integration Method," Design and Synthesis, Elsevier Science Publishers B. V (North- Holland), 1985

93. Newman, J. and Kranowski, M., "Case Study--Matrix Printer: No Pulleys, Belts, or Screws," IEEE Spectrum, Vol. 24, No. 5, May 1987, pp. 50-51

94. Nielsen, E.H., Dixon, J.R. and Simmons, M.K., "How Shall We Represent the Geometry of Designed Objects?," Technical Report 6-87, Mechanical Design Automation Laboratory, University of Massachusetts, 1987


95. Pahl, G. and Beitz, W., Engineering Design, The Design Council, Springer-Verlag, London, 1984

96. Parkinson, A.R., Bailing, R.J. and Free, J.C., "OPTDES. BYU: A Software System for Optimal Engineering De- sign," Proceedings, 1984 ASME Computers in Engineering Conference, American Society of Mechanical Engineers, Las Vegas, NV, August 12-16, 1984

97. Patterson-Hines, F.A. and Koen, B., "Direct Evaluation of Fault Trees using Object-Oriented Programming Tech- niques," IEEE Transactions on Retiabiticy, To appear in 1989

98. Paynter, H.M., Analysis and Design of Engineering Sys- tems, MIT Press, Cambridge, MA, 1961

99. PDES/STEP Standard National Institute of Standards and Technology, available through NTIS as no. NIST-IR-88- 4004, 1989

100. Peckham, J. and Maryanski, F., "Semantic Data Models," ACM Computer Surveys, Vol. 20, No. 3, September 1988, pp. 153-189

101. Peters, J., "Tolerancing the Components of an Assembly for Minimum Cost," Transactions of the ASME Journal of Engineering for Industry, August 1980

102. Pinilla, J.M., Finger, S. and Prinz, F.B., "Shape Feature Description and Recognition Using an Augmented Topol- ogy Graph Grammar," Submitted to the 1989 NSF Engi- neering Design Research ConJ~rence, University of Massa- chusetts, Amherst, June 11-14, 1989

103. Polak, E., Siegel, P., Wau, T., Nye, W.T. and Mayne, D.Q., "DELIGHT.MIMO: An Interactive Optimization Based Multivariable Control System Design Package,'" IEEE Control System Magazine, December 1987

104. Poli C. and Knight, W.A., "Design for Forging Hand- book," Technical report, Mechanical Engineering Depart- ment, University of Massachusetts, 1984

105. Poli, C., "A Design for Assembly Spreadsheet," Design and Synthesis, Elsevier Science Publishers B. V (North- Holland), 1985

106. Poli, C. and Knight, W.A., "Rating Products for Ease of Assembly," Machine Design, August 21, 1986

107. Poli, C. and Fenoglio, F., "Designing Parts for Automatic Assembly," Machine Design, December 10, 1987

108. Poli, C., Escudero, J. and Fernandez, R., "How Part De- sign Affects Injection Molding Tool Costs," Machine De- sign, November 24, 1988

109. Pratt, M.J., "Solid Modeling and the Interface between Design and Manufacture," IEEE Computer Graphics and Applications, July 1984, pp. 52-59

ll0. Pratt, M.J. and Wilson, P.H., "Requirements for Support of Form Features in a Solid Modeling System," Report R- 85-ASPP-01, CAM-l, June 1985

111~ Pratt, M.L~ "Synthesis of an Optimal Approach to Form Feature Modeling," Computers in Engineering 1988, American Society of Mechanical Engineers, San Francisco, CA, August 1988, pp. 263-274

112. Requicha, A.A.G. and Voelcker, H.B., "'Solid Modeling: A Historical Summary and Contemporary Assessment," 1EEE Computer Graphics & Applications, March t982, pp. 9-24

1 I3. Rindede, J.R., "Function and Form Relationships: A Basis for Preliminary Design," Proceedings from the NSF Work- shop on the Design Process, Waldron, M.B., ed., Ohio State University, Oakland, CA, February 8-I0, 1987, pp. 295-312

114. Rinderle, J.R., Colburn, E.R., Hoover, S.P., Paz-Soldan, J.P. and Watton, J.D., "Form-Function Characteristics of Electro-Mechanical Designs," Design Theory "88, Pro-

ceedings of the 1988 NSF Grantee Workshop on Design Theory and Methodology, Springer-Verlag, New York, 1989

115. Rogers, J.L. and Barthelemy, J-F. M., "An Expert System for Choosing the Best Combinations of Options in a Gen- eral Purpose Program for Automated Design Synthesis," Proceedings, 1985 ASME Computers in Engineering Con- ference, American Society of Mechanical Engineers, Bos- ton, MA, August 4-8, 1985

116. Roy, U. and Liu, C.R,, "Feature-Based Representational Scheme of a Solid Modeler for Providing Dimensioning and Tolerancing Information," Robotics and Computer-Inte- grated Manufacturing, Vol. 4, No. 3/4, 1988, pp. 335-345

117. Sakurai, H. and Gossard, D.C., "Recognizing Shape Fea- tures in Solid Models," Submitted to IEEE Computer Graphics and Applications, 1989

118. Seader, R. and Tobin, B., "Meeting the 'Impossible' Schedule," Plastics Design Forum, November/December 1987

119. Shah, J.J. and Wilson, P.R., "Analysis of Knowledge Ab- straction, Representation and Interaction Requirements for Computer-Aided Engineering," Computers in Engineering 1988, American Society of Mechanical Engineers, San Francisco, CA, August 1988, pp. 17-24

120. Shaw, N.K., Bloor, M.S. and de Pennington, A., "Product Data Models," Research in Engineering Design, Vol. 1, No. 1, 1989

121. Shephard, M.S. and Yerry, M.A., "Approaching the Auto- matic Generation of Finite Element Meshes," Computers in Mechanical Engineering, April 1983

122. Shephard, M.S., Korngold, E.V. and Wentorf, R., "Ideal- ized Models in Engineering Analysis," Technical report, Rensselaer Design Research Center, RPI, 1988

123. Speckhart, F.H., "Calculation of Tolerance Based on a Minimum Cost Approach," Journal of Engineering for In- dustry, Transactions of ASME, Vol. 94, May 1972, pp. 447-453

124. Spooner et al, "Modeling Mechanical CAD Data with Data Abstraction and Object.Oriented Techniques," Proceed- ings of the IEEE Data Engineering Conference, IEEE, Los Angeles, 1986, pp. 416-424

125. Spotts, M.F., "Allocation of Tolerances to Minimize Cost of Assembly," Journal of Engineering for Industry, Trans- actions of the ASME, Vol. 95, August 1973, pp. 762-764

126. Stiny, G., Pictorial and Formal Aspects of Shape and Shape Grammars, Birkhauser, Basel, 1975

127. Stiny, G., "Introduction to Shape and Shape Grammars," Environment and Planning B, Vol. 7, July 1980, pp. 343- 351

128. Report on the Workshop on Features in Design and Manu- facturing, Stiny, G., ed. University of California, Los Angeles, Los Angeles, CA, February 26-28, 1988

129. Stiny, G., "'Formal Devices for Design," Design Theory "88: Proceedings of the 1988 NSF Grantee Workshop on Design Theory and Methodology, Springer-Vedag, New York, 1989

t30. StoU, H.W., "Design for Manufacture: An Overview," Ap- plied Mechanics Reviews, Vol. 39, No. 9, September 1986, pp. 1356-1364

131. Su, S.Y.W., "Modeling Integrated Manufacturing Data with SAM*," IEEE Computer, Vol. 19, No. 1, 1986, pp. 34-39

132. Suh, N.P., Bell, A.C. and Gossard, D.C., "On an Axioma- tric Approach to Manufacturing and Manufacturing Sys- tems," ASME Journal of Engineering for Industry, Vol. 100, No. 2, May 1978, pp. 127-130


133. Suri, R., "A New Perspective on Manufacturing Systems Analysis," Design and Analysis of lntegrated Manufactur- ing Systems, National Academy Press, Washington, DC, 1988, pp. 118-133

134. Sutherland, G.H. and Roth, B., "Mechanism Design: Ac- counting for Manufacturing Tolerances and Costs in Func- tion Generating Problems," Transactions of the ASME Journal of Engineering for Industry, February 1975

135. Suzuki, H., Inui, M., Kimura, F. and Sata, T., "A Product Modeling System for Constructing Intelligent CAD and CAM Systems," Robotics and Computer-Integrated Man- ufacturing, Vol. 4, No. 3/4, 1988, pp. 483-489

136. Takase, H. and Nakajima, N., "A Language for Describing Assembled Machines," Design and Synthesis, Elsevier Science Publishers B. V (North-Holland), 1985, pp. 471- 476

137. Thompson, J.M.T. and Hunt, G.W., "Dangers of Struc- tural Optimization," Engineering Optimization, Vol. t, 1974, pp. 99-110

138. Turner, G.P. and Anderson, D.C., "An Object-Oriented Approach to Interactive Feature-Based Design for Quick Turnaround Manufacturing," Proceedings of 1988 ASME Computers in Engineering Conference, American Society of Mechanical Engineers, San Francisco, CA, August 1988

139. Ulrich, K. and Seering, W.P., "Conceptual Design: Syn- thesis of Systems Components," Intelligent and Integrated Manufacturing Analysis and Synthesis, American Society of Mechanical Engineers, New York, 1987, pp. 57-66

140. Ulrich, K.T. and Seering, W.P., "Function Sharing in Me- chanical Design," 7th National Conference on Artificial Intelligence, AAAI-88, Minneapolis, MN, August 21-26, 1988

141. Ulrich, K.T. and Seering, W.P., "Synthesis of Schematic Descriptions in Mechanical Design," Research in Engi- neering Design, Vol. 1, No. 1, 1989

142. Unger, M.B. and Ray, S.R., "Feature-Based Process Plan- ning at the AMRF," Computers in Engineering 1988, American Society of Mechanical Engineers, San Francisco, CA, August 1988, pp. 563-569

143. Vaghul, M., Dixon, J.R., Zinsmeister, G.E. and Simmons, M.K., "Expert Systems in a CAD Environment: Injection Molding Part Design as an Example," Proceedings, 1985 ASME Computers in Engineering Conference, American Society of Mechanical Engineers, Boston, MA, August 4- 8, 1985

144. Villamarin, B., Empirical Assessment of Propensity for Surprising Failures, PhD dissertation, Harvard University, 1988

145. Voelcker, H.B., "Modeling in the Design Process," Design and Analysis of Integrated Manufacturing Systems, Na- tional Academy Press, Washington, DC, 1988, pp. 167-199

146. Weiler, K.J., "Edge-Based Data Structures for Solid Mod- eling in Curved Surface Environments," IEEE Computer Graphics and Applications, January 1985

147. Weiler, K.J., Topological Structures fbr Geometric Model- ing, PhD dissertation, Rensselaer Polytechnic Institute, 1986

148. Wesley, M.A., Lozano-Perez, T., Lieberman, L.I., Lavin, M.A. and Grossman, D.D., "A Geometric Modeling Sys- tem for Automated Mechani

001 17页，A Review of Research in Mechanical Engineering Design. Part II Representations,...

Documents

Transcript of 001 17页，A Review of Research in Mechanical Engineering Design. Part II Representations,...