Anda di halaman 1dari 12

Advances in Design Automation, ASME 1993 ASME Design Automation Conference Albuquerque, NM, September 1993 pp.

75-86, ASME Press, New York, NY

ISSUES IN PRODUCT LIFE CYCLE ENGINEERING ANALYSIS


David E. Lee and Michel A. Melkanoff Manufacturing Engineering Program University of California Los Angeles, California

ABSTRACT Traditional engineering analysis of product designs has focused primarily on a product's operational performance without considering costs of manufacturing and other stages downstream from design. In contrast, life cycle analysis of a product during its initial development can play a crucial role in determining the product's overall life cycle cost and useful life span. This paper examines product life cycle engineering analysis - measurement of product operational performance in a life cycle context. Life cycle engineering analysis is thus considered both as an extension of traditional engineering analysis methods and as a subset of a total product life cycle analysis. The issues critical to life cycle engineering analysis are defined and include product life cycle data modeling and analysis, analysis tools and their performance regimes, performance tradeoff measurement and problems of life cycle engineering analysis in an organizational context. Recommendations are provided for future research directions into life cycle engineering analysis in the context of integration architectures for concurrent engineering.

1. INTRODUCTION The development of commercial products and military systems involves balancing a series of factors to specify, design, and manufacture products and systems that perform a specific set of operational functions. Listed in Figure 1, these product development factors influence both the overall product definition as well as each stage within

the product development life cycle. For instance, consumer electronic products producers might assess marketing opportunities of different types of music playback equipment (analog tape, compact disc players, digital audio tape, etc.) in order to define what resources (development funding, engineering expertise) to commit to designing and producing a specific mix of equipment both in the short and long term. The functional attributes required for each product in the product line (portability/ weight, programmability, sound fidelity, etc.) affect the use of common subsystems (drive motors, LCD/LED status display panels, power transformers, etc.) and the cost of operating a product (power consumption, service/maintenance, upgrades, and accessories). This assessment of product performance from a total life cycle perspective (i.e. across all life cycle stages) is defined here as product life cycle analysis. In our research, life cycle analysis of a product is viewed as a superset of analysis methods centered about individual life cycle stages. Each of the analyses depicted in Figure 2 seeks to qualitatively and quantitatively measure product performance both at the local life cycle stage as well as across the total product life cycle. For life cycle manufacturing analysis, the primary focus is on determining the optimal sequence of processes to produce a product while concurrently measuring the impact of decisions on other life cycle stages such as fastener selection to allow for future product service and maintenance. In a similar vein, the procedure of measuring the effects of development decisions on a product's Copyright 1993 by ASME

operational performance in an overall life cycle context is defined as life cycle engineering analysis. This is an extension of traditional engineering analysis methods which are applied during detailed design and are used to quantify product operational performance - static and dynamic loading behavior, thermal operational performance, system control response, etc. Life cycle engineering analysis extends current engineering analysis approaches by applying them to other life cycle stages (such as system deformation and thermal behavior analyses under manufacturing processes and burn-in testing) and assessing life cycle performance tradeoffs (assembly versus service, ease of component fabrication versus recyclability, etc.). 1.1 Paper Outline This paper focuses on defining the issues central to a product's life cycle engineering analysis. The product development life cycle is defined in Section 2. Both the types of product development cycles and the general classes of product designs are defined. The overall product life cycle analysis effort is presented in Section 3. The analysis requirements per each individual life cycle stage are summarized and problems of quantifying tradeoff costs and effects across life cycle stages is addressed. Section 4 deals with the specific issues involving engineering analysis in a product development life cycle context. These include product development analysis models, analysis tools and performance regimes, performance tradeoff measurement and problems of engineering analysis in an organizational context. Finally, directions for future research into dealing with product life cycle engineering analysis from a concurrent engineering perspective are discussed in Section 5. 1.2 An Example: Printed Circuit Assembly With the extended use of surface mount technology (SMT) components in digital circuit design because of increased functional requirements and higher number of interconnections between circuit components, electronic circuit designers have opted to use more complex components (Vardaman 1992). An example is the use of quad flat pack (QFP) components in printed circuit assemblies (Lee 1993a, Lau and Erasmus 1992, Lau, Rice, and Erasmus 1992, Lau and Rice 1990). As depicted in Figure 3a, this class of SMT circuit components is the packaging for complex integrated circuit (IC) chips such as microprocessors (Hoffman et. al. 1992) and provides IC protection from environmental contaminants along with serving as the structural and electrical interconnection

Market Opportunities - demand/volume - competitive products - functionality desired Operational Requirements - functionality required - operating environment - reliability - service Cost - development - production/manufacture - operation - maintenance/service - disposal Time - development - production/manufacture - operational life - maintenance/service requirements Organizational Issues - evolutionary/revolutionary product - new/proven technologies - organization experience - development/support infrastructure

Figure 1. Product Development Factors

Figure 2. Product Life Cycle Analysis

framework for such components. Depending on the application and cost concerns, QFP bodies are fabricated from epoxy resin, ceramic, or metal. Further, given the number of interconnects required between the IC chip and the rest of the circuit, the QFP leadframes that serve as the electrical connection framework for the components may have over 300 external leads. Currently, engineering analysis of a printed circuit board (PCB) and its attached components focuses on an overall circuit analysis and usually separate thermal analysis (Lee 1993a). Circuit analysis methods have sought to validate the logical functions of the overall PCB assembly (Riley 1988, Ginsberg 1989). Also, a separate thermal analysis of the same PCB assembly would be performed by simply adding up the thermal output generated from the individual components and modules attached to the board to determine the thermal energy profile of an assembly. With the scale of complexity and cost of individual circuit components such as QFPs and the expanded use of PCB assemblies, problems that were not easily analyzed with these methods have become much more important. These include the effects of differential heating of QFPs and the PCB assembly (Lee 1993a), large differences in thermal coefficients of expansion between the materials used in the package as well as the PCB (see Figure 3b) (Lau and Erasmus 1992), effects of residual stresses from reflow soldering of the QFP to the board (Hwang 1989), effects of vibrational stresses on the PCB and QFP (Engel 1990), and reliability analysis about the effects of thermal and power cycling of the assembly (Lau and Erasmus 1992, Englemaier 1983). These problems encompass the operational behavior of a completed QFP/PCB assembly by requiring performance information from each stage in the product's life cycle. For instance, the large differences of coefficients of thermal expansion used in the PCB assembly affects reliability analysis of the product because of the magnitudes of thermal loads applied to the system at different life cycle stages. During solder reflow, the assembly may be exposed to a thermal load of 250C for a few seconds. In comparison, power cycling of the board during operation may occur for several thousand cycles that raise the temperature from 25C to 100C on a daily basis (Lee 1993a, Lau and Erasmus 1992). In this case, life cycle engineering analysis of a PCB assembly involves integration of several different and historically distinct analytical domains - electrical, thermal, structural, mechanical, reliability/failure analysis. Thus, life cycle engineering analysis requires a more robust and coupled approach to product analysis while

(a) Quad Flat Pack Coefficient of Thermal Expansion 10-6 m/m-C 21 30 2.8 18 70 17

Young's Modulus GPa solder plastic silicon FR4 xy z 10 6 131 22 22 121

copper

(b) Material Properties of the Assembly

(c) PCB Assembly Figure 3. SMT PQFP/PCB Assembly

identifying life cycle costs of design decisions and the effects of these decisions across the product development life cycle.

DISPOSAL
2. PRODUCT DEVELOPMENT LIFE CYCLE The product development life cycle is an explicit model of the stages a product goes through during its life history and the interaction of information and decisions between these stages. The stages in the development life cycle used in this research are design, manufacture, test, operation, maintenance, and disposal. The general properties of each stage are given in Figure 4. Implicitly or explicitly, all companies that develop products have some sort of a product development process (National Research Council 1991). For each stage in the process, a set of models and analysis methods are employed to insure product performance at that stage. For example, models of machining processes and fixtures along with planning algorithms may be used to develop process plans for machined parts in manufacturing process planning (Fuh, Chang, and Melkanoff 1992, Chang and Wysk 1985, Wang and Li 1991). Whether or not departments within a corporation responsible for individual stages provide feedback to preceding stages or feedforward planning information to following stages based on a stage's analysis determines if the development process is a closed or open loop (i.e. a true cycle or not). A further discussion of product development issues in an organizational context is provided in Section 4.4. 2.1 Types of Product Development Cycles Product development cycles can be classified by: * * * * * product type product cost product complexity available development time/cycle previous development experience

DESIGN conceptualize, develop and refine design

dispose and recycle product

MAINTAIN maintain and service product

MANUFACTURE plan, schedule assembly of product

OPERATION use and operation of product

TEST tes product function and quality

Figure 4. Product Development Life Cycle Figure 5. The distinction between the different types of cycles is important from a life cycle analysis perspective because the types of tradeoffs and analytic models that are employed for a large scale development effort might not be as effective for a small scale process. As an example, the length of development cycle iterations, turnover in development staff, cost implications of design decisions, and product complexity for systems such as fighter aircraft or nuclear power plants almost require the development of product development support infrastructures to measure system performance. These include robust simulators of subsystems and system dynamics, performance validation testing suites, and integration test beds.

For instance, the overall development process for a high-performance fighter aircraft (multi-year life span for a multi-million dollar, highly complex military system) differs significantly from the development approach used for a portable cassette player (short life span, inexpensive, consumer product) versus that for a non-invasive, blood oxygen sensor (multi-year life span, thousands of dollars, biotechnology system). From these factors, products and their development cycles can be grouped into broad categories based on the length of the development cycle. These are given in

Large scale - multiple, multi-year, on-going development cycles - decade length operational life - customer product cost/unit $106 - $109 - annual unit production volumes 100 - 102 - multiple complex subsystems - large integration effort - high infrastructure development and maintenance - mid-life upgrades to selected subsystems - continuing sales value - examples: aircraft, buildings, production plants Mid scale - 1-5 year development cycles - 1-5 year operational life - customer product cost/unit $103 - $105 - annual unit production volumes 103 - 105 - critical subsystems - medium scale integration effort - low field infrastructure support costs - replacement by new product models (obsolescence) - examples: computers, cars, electronic instruments Small scale - _ 1 year development cycles - _ 2 year operational life - customer product cost/unit $101 - $102 - annual unit production volumes 106 - 107 - simple subsystems - incremental design changes - little or no field support costs - disposal after failure - examples: consumer electronics, small appliances

* * *

original or new designs transitional or adaptive designs extensional or variant designs

As pointed out by Whitney (1990), the latter two classes are the primary foci for product development staffs - most of their time (up to 85%) is spent in refining existing designs. This is not unreasonable since the product development support infrastructures in most companies have not been geared to designing and producing new and innovative products on a rapid basis either because of the cost of new facilities and equipment and time need to develop models and testing procedures for new manufacturing processes or because of the disruption caused to existing facilities and processes when producing new products. Because of the maturity of existing engineering analysis methods, their use in the detailed design task, and the on-going need to support product development for adaptive and variant product designs, the issues addressed herein center on measuring product functionality tradeoffs in life cycle analysis problems during detailed design as opposed to analysis issues in conceptual design of products. Support of product analysis during the conceptual design task in the area of assembly design analysis has been addressed in Lee and Melkanoff (1991).

Figure 5. Product Development Cycle Types 2.2 Design Approaches While it is not truly possible to separate the synthesis of a product's design from the analysis of the design, the primary focus of this paper is on the analytic tools and methods used in the overall design process. However, the different design problems faced in product development directly impact the types of tools and procedures brought to bear. Given this, product designs (and design problems in general) can be classified into three general types (Finger and Dixon 1989, Whitney 1990):

3. LIFE CYCLE ANALYSIS Life cycle analysis, as mentioned in the Introduction, can be most broadly defined as the assessment of product performance from a life cycle perspective. As viewed from a product life cycle perspective, life cycle analysis is used during the development process to measure product performance in each life cycle stage and provide quantitative feedback about the effects of design decisions among the different stages. This distinction between measurement of total product performance (i.e. performance in each life cycle stage - can it be made, maintained, etc. and how does it work) versus traditional product performance analysis in an operational sense (the product's operating envelope - speed, weight, and other operational features) can be viewed as the underlying philosophy of life cycle analysis. In order to measure overall performance, the product performance criteria must be specified (product specifications) along with any constraints (development time, legal requirements, etc.) for each life cycle stage. In the case of operational specs such as system weight, carrying capacity, power source(s), product useful life, these have been fairly well defined. With respect to

product performance specifications for other life cycle stages such as manufacturing and assembly, testing, and maintenance, these might include minimal numbers of components, single assembly directional motions, access or I/O ports for testing, and reliability definitions for components and subsystems. This is similar to the approaches of the DFx methods (Design for Manufacture (DFM) (Swift 1987), Assembly (DFA) (Boothroyd and Dewhurst 1987), Test (DFT) (Turino 1990), etc.) and the concurrent engineering of products and processes (Nevins and Whitney 1989). 3.1 Product Analysis Methods Within each life cycle stage, specific types of analyses are required to support each of the tasks that occur. For example, the analysis needs of the manufacturing and testing stages can be defined. As shown in Figure 6, each stage has a planning/ development task, a task that focuses on individual components, as well as a task involved in the integration of the complete product. During the production planning and scheduling task, process and assembly plans are generated for the components and subassemblies to be manufactured. This may also include planning and fabricating support tools such as holding fixtures or component molds. In order to validate the product definition in this task, planning and scheduling methods, manufacturing process models, machinability indices, and clamping/work holding force analyses, and machining and assembly plan validation testing may all be used. 3.1.1 Classification of Analysis Methods Based on the types of analyses performed, analytic models used, and specificity/generality of results generated, available analysis approaches can be grouped into the following classes: Physical Implementations - includes prototyping, integration test bed environments, and testing & experimentation on physical systems and system components. Generalized Analytic Methods - includes finite element methods, vibration analysis, heat transfer, circuit analysis, and other analytic methods that rely on analytic models of physical phenomena. State Behavioral - Dynamic - Formal Model Methods - includes petri nets, state charts, controller analysis, discrete event simulation, and other formal system modeling approaches that decompose system

behavior into discrete, well-formed interactions between system elements and components. Deductive and Heuristic Reasoning Approaches includes knowledge-based systems, design for x, feature-based reasoning systems, design and process metrics, and other rule-based approaches to analyzing and modeling system behavior. Probabilistic/Statistical Analytic Methods includes system reliability models, robust or statistical design techniques, monte carlo tolerance analysis, and other probabilistic approaches to analyzing system behavior. Resource Allocation Analysis - includes scheduling algorithms, process planning systems, optimal routing schemes, and other resource allocation approaches. Geometric Reasoning Methods - includes assembly path and motion planning, machining validation approaches, net shape component analysis, and other space partitioning/decomposition and classification analytic methods. In general, many product analysis methods utilize approaches that combine one or more of these classes. For instance, all analysis methods rely upon some sort of physical implementation to validate system behavior/performance. Unfortunately, quite a few companies use this as their primary product development analysis methodology (see Salzberg and Watkins (1990) for a fascinating discussion). Also, while the above classes are fairly broad and distinct, there may be analytic methods that are not fully encapsulated by one or more of these classes. However, the classes are useful in understanding the modeling approaches used by the different methods and the basic analytic underpinning of each. 3.1.2 Stage Analysis Requirements Each life cycle stage currently requires specific types of analyses to assess product definition performance. Moreover, the product definition information provided at each stage varies on the refinement of product data and the variant or adaptive nature of the product design. As identified earlier, manufacturing analysis tools employ planning and scheduling algorithms, production analyses, and geometric reasoning approaches for assembly path planning. From the results of the analyses, feedforward

Product Manufactur e production planning and scheduling component manufacture and procurement

product assembly and integration

Product Test product test planning and validation component and subsystem testing

product integration and testing

Figure 6. Manufacturing and Testing Life Cycle Stages data is transmitted for support tooling fabrication as well as testing conditions for component, subsystem, and system test. Information is generated that identifies problem components and subsystems and then is fed back to design to request engineering changes which are reanalyzed to assess manufacturing feasibility. This constant iterative refinement of product definition data as the product evolves through its life cycle is affected by the coupling between the different analysis methods used in a given life cycle stage as well as data and timing dependencies between stages (Bond and Ricci 1990). Because of this dependence and coupling, approaches to extending analytical tools to be more effective in a life cycle context involve some mechanism to assess tradeoffs between potential decisions during development.

3.2 Life Cycle Tradeoff Measurement Analysis In order to utilize existing product analysis methods such as DFA, DFM, and even general analytic techniques like finite element methods (Zienkiewicz 1977) and computational heat transfer analysis (Jaluria and Torrance 1986) in a life cycle analysis context, their domains of application must be extended to the entire product development life cycle by measuring the effects of stagelocalized decisions. As previously depicted in Figure 2, the life cycle analysis of a product can be partitioned into the applicable analysis methods based on life cycle stages. However, the results of analyzing a product in one stage and the ability of that stage to accomplish its performance specifications and meet applicable constraints may propagate throughout the development cycle. (Issues focusing on life cycle performance tradeoff analysis and constraint propagation from an engineering perspective are also addressed in Section 4.2.) For instance, in the approach of Design for Assembly as defined by Boothroyd and Dewhurst (1987), a product's development is to be completely optimized based on simplifying assembly process costs by minimizing the number of components used within the product, possible directions used to assemble components together, and assessment of the required annual unit production volumes. This approach to product design heavily weighs toward biasing the whole development cycle toward assembly performance criteria and process constraints. By doing so, it negatively impacts on the complexity of individual component fabrication costs (Whitney 1990, page 12), service and maintenance of the product (Bryan, Eubanks, and Ishii 1992) recyclability of some or all components (Burke, Beiter, and Ishii 1992). 3.2.1 Tradeoff Measurement Approaches In order to assess the coupling and dependence between different life cycle stages and analytic methods, there are four general approaches to measuring tradeoffs: Try it and see... - In the context of an overall life cycle development model, this is the de facto approach to assessing the effects of a decision in design that may affect manufacturing as well as vice versa. Impact on global variables & constraints Associated with any product is its overall cost and life time. Other variables, related to physical dimensions and quantities (system velocity profile, weight, etc.) can also be identified and used to measure the effects of global design and local process changes.

Impact on local variables & constraints - At each stage within the development effort, a mapping may be defined between the product's performance requirements at that stage, applicable constraints, and variables considered critical to performance at that stage. These may include component counts, tolerance measurements, and machinability indices in manufacturing. Based on this mapping, measurements of the effects of design changes to production cost and time in manufacturing can be effected. Effect on system behavioral models - With the definition of system performance as a formal dynamic model (petri nets, IDEF, etc.) or a heuristic model (knowledge-based systems, DFx approaches, etc), effects specifically on system performance can be assessed if the change or operation can be mapped into the modeling framework. This provides a macro scale model to identify causal chains. 3.2.2 Tradeoff Measurement Methods To go beyond these general approaches to life cycle analysis tradeoff measurement, understanding the specific quantities that are measured and how they are measured is central to the tradeoff methodology. The classes of variables measured are: * * * * global product variables global physical variables localized life cycle variables logical state variables

framework. (For a more detailed discussion about life cycle analysis, please refer to Lee (1993b).)

4. ISSUES IN PRODUCT LIFE CYCLE ENGINEERING ANALYSIS Life cycle engineering analysis, as defined in this paper, is not simply the traditional analytical methods used to assess system operational performance. Nor is it the broader, overarching analysis that occurs throughout all life cycle stages during the development of a product or system. Specifically, life cycle engineering analysis is the measurement of product design decisions on system performance in its operational stage and propagation of the effects of decisions and processes executed in other life cycle stages into system operational performance. Based on the generalized framework of life cycle analysis, addressing the concerns about analysis models, tradeoff measurement, and understanding the organizational development context is central to performing life cycle engineering analysis in a product development effort. 4.1 Product Development Analysis In an operational context, the issues faced in life cycle engineering analysis revolve around the assumptions built into the analysis methodologies, their limitations and the valid performance regimes for the tools. Because the operational performance of complex systems such as aircraft and buildings affects thousands of people both in the development of these systems as well as those who use them, understanding these issues is critical. 4.1.1 Analysis Assumptions Inherent in any analytic engineering methodology is a set of assumptions that are idealizations and abstractions about the world (Shephard 1990). Whether it is assuming that a linear relationship exists between the stress/strain behavior of a material or a more complex non-elastic one, other concerns about temperature and time dependent creep effects within the materials under analysis may also be of concern (Mendelson 1968, Dowling 1993). Thus, fundamental issues in an engineering analysis deal with the operational environment in which the product or system is to function (temperature, pressure, time, chemical contaminants, etc.) as well as the environment faced by the system in other life cycle stages (e.g. residual thermal stresses from manufacturing processes, subsystem and system environment during shipping and transport, etc.).

From these, specific measures have been adapted to assess changes and impacts upon the underlying models: * * * * * total system cost and life time defined measures & metrics (DFA, DCA, ADEM, etc.) physical phenomena changes (stresses, vibrations, etc.) logical measures (violations of system dynamic behavior) sensitivity/robustness measures (optimization, Taguchi, etc.)

Given this context of different product analysis methods, tradeoff measurement approaches, variables classes, and tradeoff measures of performance, we seek to focus on the engineering analysis methods in a life cycle

Beyond the total operational environment of a product, the limitations of the models used by the methodologies must be clearly understood. The error introduced in defining a mesh over a domain as well as errors generated during computation of a solution to a boundary value problem using finite element methods must be bounded and magnitudes of errors computable (Zienkiewicz 1977, Shephard 1990). Further, the scale of validity of existing analytic methods must be defined. In the case of micromechanical systems, do the bulk material properties measured for "large" components (i.e. @ human scale 10-1 to 100 meters) apply at the micromechanical level ( 10-4 to 10-5 meters) (Wise 1991)? What about heat transfer mechanisms at the micromechanical system level? These concerns about analytical and experimental data limitations must be addressed as part of the analysis process. 4.1.2 Analysis Tools and Performance Regimes In Section 3.1.1, the different types of product analysis tools were classified based on their central analytic approach and the models used. The assumptions and limitations of these tools, as mentioned above, specifies the analysis context. Beyond this, the performance regimes of the engineering analysis methods - the types of analyses that can be generated from the underlying assumptions and models - defines the operational role of these methods. For example, rulebased systems use heuristics and experiential knowledge of design tasks to generate logical inferences about a system. From rules built into such a system about fasteners (screws, bolts, snap fasteners, etc.), a class of assembly problems that focus on combining components together with fasteners can be analyzed and suggestions made for different types of fasteners to use. This defines the performance regime for such a rule-based system. For each of the analysis tool classes, valid performance regimes can be specified. This would indicate the types of analysis problems that can be dealt with using heuristics, metrics, analytic methods, probabilistic approaches, etc. This is similar to the approach of Suri and Shimizu (1989) (design for analysis) where a product design is "designed" for analysis - i.e. can the design be validated. Also, the strengths of different analysis methods (transient vs. steady state, etc.) can be employed. Moreover, issues surrounding life cycle type analyses such measurement of performance across stages and interactions with different analysis domains should be addressed in terms of the types of tradeoffs that can be measured.

4.1.3 Inter-Domain & -Stage Analysis Methods If the assumptions and idealizations inherent in analytic engineering methods can be defined and the valid performance regimes can be specified, then the opportunity exists to integrate analysis methods together that apply to similar domains. In the case of the PCB assembly, analysis of stress, strain, and thermal conditions can be integrated into a single thermomechanical analysis of the PCB system (Lee 1993a, Lau and Erasmus 1992). This allows different constitutive relationships between stress, strain, and temperature to be applied to the elements of system - elastic for the leadframe, orthotropic for the printed circuit board and QFP, and viscoelastic for solder (Lee 1993a). These models could then be extended to electrical analysis of the system. Further, thermal effects in different life cycle stages (manufacture vs. operation) can be fully analyzed. Such an approach to integrating different analytical domains provides an opportunity to expand existing engineering methods into total product life cycle analysis. 4.2 Performance Tradeoff Measurement On the flip side of life cycle engineering analysis is the problem of assessing the effects of design decisions on other life cycle stages as well as measuring the effects of cost drivers from the other stages within the design stage. While realistically the flow of effects is generally from design to the other stages, if a product cannot be manufactured or does not operate in the regimes required, the feedback from these other stages will have a definite and measurable effect upon product design and engineering analysis. How to define and measure these effects in life cycle engineering analysis is the focus of this section. 4.2.1 Performance Measurement In dealing with performance measurement from an engineering analysis perspective, the concern about optimal and robust design performance (i.e. is this the best performing product design) affects the types of the analysis methods selected and thus tradeoff approaches used in conjunction. If there is a high degree of coupling between the product's performance regimes, an analysis approach that allows for coupling between regimes can be applied. As an example, the effects of thermal power cycling of a SMT PQFP/PCB assembly affect the reliability and stress analyses of the assembly (Lee 1993a). In order to deal with the coupling of a thermal analysis, stress analysis, and reliability analysis in an electrical circuit context, a model can be developed that incorporates each of these domains. This would allow for measurement

of system performance in each analytical domain as well as assessments when changes in one domain impact the others. 4.2.2 Tradeoff Measurement As identified in Section 3, there are several issues that affect assessment of tradeoffs across life cycle stages. At the heart of these are the variables selected that indicate the effects of changes imposed by one life cycle stage on the product definition. Globally, across organizations involved in the product development effort, cost and time variables can be easily comprehended. However, these offer little access to the effects of decisions and changes at the local (life cycle stage) level. For instance, localized variables for an aerodynamicist analyzing an aircraft would be surface geometry and lift, drag, and pressure distributions across the aircraft (Bond and Ricci 1990). For the group concerned with weights and loads focus on the mass distribution of the aircraft, aircraft configuration at liftoff, as well as the mass loading distribution of total system. The radar imaging group would focus on the skin materials of the aircraft, its radar cross section profiles and system mission profiles. How these can or could be combined currently involves the efforts of a integrated design team. Beyond identifying the tradeoff variables for the product, the performance at each life cycle stage and specifically in an operational context must be measured. This is partially addressed by the types of performance measurement analysis methods used (Section 4.2.1). Unfortunately, performance analysis localized to the design stage does not provide an effective quantitative measure for other life cycle stages. Also, this identifies the problem of providing a mapping between physical models used within a stage as well as across life cycle stages. 4.3 Organizational Issues In many ways, the issues involved in understanding the performance regimes of engineering analysis tools and establishing tradeoff measures between the distinct life cycle stages may be simpler than dealing with the problems involved in the product development process from an organizational perspective. The problems of tracking and managing changes that occur in product definition data as well as the lack of interaction between departments involved in product development in many companies (Salzberg and Watkins 1990) are only a few of the challenges at the heart of the product development bottleneck.

4.3.1 Change Management Because of the constant and on-going design revision process and personnel changes in the product development staff, a great deal of effort goes into managing changes of data and information used in the development process. Central to this problem is the issue is identifying what product design data is the most up-to-date. The procedures used to deal with this issue are referred to as change management or version control systems (Katz 1990). Assigning version numbers and keeping a heritage graph that tracks the evolution of a design are some of the key concepts of such systems. However the problems of insuring that new versions of a product design can be analyzed and validated, fragmentation and localized revisions of design data, propagation of design changes, and transaction integrity (Eastman and Kutay 1990) directly impact the effectiveness of these systems. 4.3.2 Organizational Concerns In general, the interchange of information during the product development process is not an explicit, formalized affair. Much of the communication occurs in individual one-on-one discussions with a great deal of information as to development decisions (i.e. design rationale) kept by team members as individualized, mental learning experiences. If information is explicitly exchanged within and between stages (through integrated design teams, design reviews, planning sessions, buy-in meetings, etc.), the interaction allows for a sense of ownership in the product under development and shorter iterations between design releases. If the information exchange is implicit (i.e. manufacturing change requests resulting engineering change orders, product recalls, etc.), there is usually a more parochial view of development ("We did our part. It's their problem - they don't know how to .") and much longer and costly design cycles. As a managerial concern, existing cost accounting approaches assess the contributions of an individual or group to a specific task. If the task is the total product development process and the contribution is life cycle engineering analysis, the problem of assigning costs to an identifiable contribution in the analysis may or may not be effective. For team members who serve as the "glue" by integrating different analysis groups and life cycle stages, should they be considered as "overhead"? If not, then to what specific deliverable item should they be charged to? How should this be measured and distributed? Similarly, if a product is developed by more than one corporation, how should these costs be assessed?

4.3.3 Integrating Life Cycle Engineering Analysis In any organization, there are different levels of expertise and states of evolution for groups and departments involved in product design and analysis (Salzberg and Watkins 1990). Analytic methods and computerized tools in airframe structural analysis differ in scope and sophistication than those available for groups involved in aircraft configuration (Bond and Ricci 1990). CAD systems of different vintages and vendors abound. Because of the investments in existing analysis packages and modeling systems, retraining costs, and the isolationist views of many engineering analysis groups (Salzberg and Watkins 1990), the process of integrating groups involved in product development, especially in large organizations, to conduct life cycle engineering analysis will likely extend beyond development of intradomain analytic methods and require large changes in corporate culture.

Especially for variant and adaptive product designs and given the existence of legacy design and analysis systems, a more cost effective and simpler approach may be to interconnect systems to draw upon the strengths of the individual systems.

ACKNOWLEDGEMENTS The authors wish to thank Kos Ishii for his encouragement in the preparation of this paper and the reviewers for the directness of their comments in helping to clarify its focus. This research has been supported by the Institute for Manufacturing and Automation Research (IMAR) through IMAR Project #5, Design for Automated Assembly and Maintenance, and additional gifts from the industrial affiliates of the UCLA Manufacturing Engineering Program. REFERENCES

5. FUTURE RESEARCH DIRECTIONS Based on the issues raised in this paper that are central to life cycle engineering analysis, we suggest the following directions for research: Inter-Domain Analytical Methods: Based on the ever increasing demands on product functional requirements and system complexity, the coupling between different analysis domains is expanding. The ability to measure changes and effects in one analytical domain such as system heat transfer and its impacts on other domains such as system reliability analysis and stress analysis provides greater insight into understanding product operational behavior as well as the effects of other life cycle processes on system behavior. Life Cycle Tradeoff Measures: Given the localized measures to assess tradeoffs in individual life cycle stages, a methodology that allows a broader approach to life cycle tradeoff analysis would extend the capabilities of existing analytic approaches. Interconnected Product Development Systems: Because of the disparate nature of different analytic methods and their underlying models, the current trend toward developing integrated product development environments (Sriram et. al. 1990) may be faced with a limited set of options in attempting to define common product tradeoff measurement schemes for large scale product development efforts.

Bond, A. H., and Ricci, R. J., 1990, "Cooperation in Aircraft Design", Computer-Aided Cooperative Product Development, D. Sriram, et. al., eds., pp. 152-182, Springer-Verlag, Berlin, DM. Boothroyd, G., and Dewhurst, P., 1987, Product Design for Assembly, Boothroyd Dewhurst Inc., Wakefield, RI. Bryan, C., Eubanks, C. F., and Ishii, K., 1992, "Design for Serviceability Expert System", Proceedings of the 1992 ASME Computers In Engineering Conference, v. 1, pp. 91-98. Burke, D. S., Beiter, K., and Ishii, K., 1992, "Lifecycle Design for Recyclability", Proceedings of the 1992 ASME Design Theory and Methodology Conference, pp. 325-332. Chang, T. C., and Wysk, R. A., 1985, An Introduction to Automated Process Planning Systems, Prentice-Hall, Englewood Cliffs, NJ. Dowling, N. E., 1993, Mechanical Behavior of Materials, Prentice-Hall, Englewood Cliffs, NJ. Eastman, C. M., and Kutay, A., 1990, "Transaction Management in Design Databases", Computer-Aided Cooperative Product Development, D. Sriram, et. al., eds., pp. 334-351, Springer-Verlag, Berlin, DM. Engel, P. A., 1990, "Structural Analysis for Circuit Card Systems Subjected to Bending", ASME Journal of Electronic Packaging, v.2, n.1, pp. 2-10. Engelmaier, W., 1983, "Fatigue Life of Leadless Chip Carrier Solder Joints During Power Cycling", IEEE

Transactions on Components, Hybrids, and Manufacturing Technology, v.6, n.3, pp. 232-237. Finger, S., and Dixon, J. R., 1989, "A Review of Research in Mechanical Engineering Design. Part 1: Descriptive, Prescriptive, and Computer-Based Models of Design Processes", Research in Engineering Design, v. 1, n. 1, pp. 51-67. Fuh, J. Y.-H., Chang, C.-H., and Melkanoff, M. A., 1992, "A Logic-Based Integrated Manufacturing Planning System", Proceedings of the 1992 ASME Computers In Engineering Conference, v. 1, pp. 391-400. Ginsberg, Gerald, 1989, Surface Mount and Related Technologies, Marcel Dekker, New York, NY. Hoffman, H. S., et. al., 1992, "Metal Package for a High-Powered Microprocessor Chip", Proceedings of ASME/JSME Electronic Packaging Conference, v. 1, pp. 23-26, ASME Press, New York, NY. Hwang, Jennie S., 1989, Solder Paste in Electronics Packaging, Van Nostrand Reinhold, New York, NY. Jaluria, Y., and Torrance, K. E., 1986, Computational Heat Transfer, Hemisphere Publishing, New York, NY. Katz, R. H., 1990, "Toward a Unified Framework for Version Modeling in Engineering Databases", ACM Computing Surveys, v. 22, n. 4, pp. 376-408. Lau, J., and Erasmus, S., 1992, "Experimental and Analytical Studies of 304-Pin, 0.5 mm Pitch, 40 mm Body Size Plastic Quad Flat Pack (QFP) Solder Joints and Leads Under Bending, Twisting, and Thermal Conditions", 92WA/EEP-33, ASME Winter Annual Meeting, Electrical and Electronics Packaging Sessions. Lau, J. H., and Rice, D. W., 1990, "Thermal Stress/Strain Analyses of Ceramic Quad Flat Pack Packages and Interconnections", Proceedings of the 40th IEEE Electronic Components & Technology Conference, pp. 824-834, IEEE Press, New York, NY. Lau, J., Rice, D., and Erasmus, S., 1992, "Thermal Fatigue Life of 256-pin, 0.4 mm Pitch Plastic Quad Flat Pack (QFP) Solder Joints", Proceedings of ASME/JSME Electronic Packaging Conference, v. 2, pp. 855-863, ASME Press, New York, NY. Lee, D. E., 1993a, "Modeling the Evolution of Thermal Stresses in Surface Mounted IC Packages and Printed Circuit Board Assemblies", UCLA Manufacturing Engineering Program, TR-DEL-01-1993, 1993. Lee, D. E., 1993b, "Issues in Product Development Life Cycle Analysis", UCLA Manufacturing Engineering Program, TR-DEL-02-1993. Lee, D. E., and Melkanoff, M. E., 1991, "Product Design Analysis Using the Assembly Design Evaluation Metric", Proceedings of the 1991 ASME Computers In Engineering Conference, v. 1, pp. 95-102.

Mendelson, A., 1968, Plasticity: Theory and Application, Robert Krieger Publishing, Malabar, FL. National Research Council, 1991, Improving Engineering Design: Designing for Competitive Advantage, Committee on Engineering Design Theory and Methodology & Manufacturing Studies Board, National Academy Press, Washington, D. C. Nevins, J. L., and Whitney, D. E., eds., 1989, Concurrent Design of Products and Processes, McGrawHill, New York, NY. Riley, Frank, ed., 1988, The Electronics Assembly Handbook, IFS Publications, Bedford, United Kingdom. Salzberg, S., and Watkins, M., 1990, "Managing Information for Concurrent Engineering: Challenges and Barriers", Research in Engineering Design, v. 2, n. 1, pp. 35-52. Shephard, M. S., 1990, "Idealization in Engineering Modeling and Design", Research in Engineering Design, v. 1, n. 3/4, pp. 229-238. Sriram, D., et. al., eds., 1990, Computer-Aided Cooperative Product Development, Springer-Verlag, Berlin, DM. Suri, R., and Shimizu, M., 1989, "Design for Analysis: A New Strategy to Improve the Design Process", Research in Engineering Design, v. 1, n. 2, pp. 105-120. Swift, K. G., 1987, Knowledge-Based Design for Manufacture, Prentice-Hall, Englewood Cliffs, NJ. Turino, J., 1990, Design to Test, Van Nostrand Reinhold, New York, NY. Vardaman, J., ed., 1992, Surface Mount Technology, IEEE Press, New York, NY. Wang, H.-P., and Li, J.-K., 1991, Computer-Aided Process Planning, Elsevier, Amsterdam, NL. Whitney, D. E., 1990, "Designing the Design Process", Research in Engineering Design, v. 2, n. 1, pp. 3-13. Wise, K. D., 1991, "Micromechanical Sensors, Actuators, and Systems", Micromechanical Sensors, Actuators, and Systems, pp. 1-14, ASME 1991 Winter Annual Meeting. Zienkiewicz, O. C., 1977, The Finite Element Method, 3rd ed., McGraw-Hill, London, UK.

Anda mungkin juga menyukai