Study on life-cycle design for the post mass production paradigm

Transcription

1 Artificial Intelligence for Engineering Design, Analysis and Manufacturing ~2000!, 14, Printed in the USA. Copyright 2000 Cambridge University Press $12.50 Study on life-cycle design for the post mass production paradigm YASUSHI UMEDA, 1 AKIRA NONOMURA, 2 and TETSUO TOMIYAMA 3 1 Department of Mechanical Engineering, Graduate School of Engineering, Tokyo Metropolitan University, Minami-Osawa 1-1, Hachioji, Tokyo , Japan 2 DENSO Corporation, Tokyo, Japan 3 RACE: Research into Artifacts, Center for Engineering, the University of Tokyo, Komaba 4-6-1, Meguro-Ku-Tokyo , Japan (Received March 3, 1999; Revised August 20, 1999; Accepted August 31, 1999! Abstract Environmental issues require a new manufacturing paradigm because the current mass production and mass consumption paradigm inevitably cause them. We have already proposed a new manufacturing paradigm called the Post Mass Production Paradigm ~PMPP! that advocates sustainable production by decoupling economic growth from material and energy consumption. To realize PMPP, appropriate planning of a product life cycle ~design of life cycle! is indispensable in addition to the traditional environmental conscious design methodologies. For supporting the design of a life cycle, this paper proposes a life-cycle simulation system that consists of a life-cycle simulator, an optimizer, a model editor, and knowledge bases. The simulation system evaluates product life cycles from an integrated view of environmental consciousness and economic profitability and optimizes the life cycles. A case study with the simulation system illustrates that the environmental impacts can be reduced drastically without decreasing corporate profits by appropriately combining maintenance, reuse and recycling, and by taking into consideration that optimized modular structures differ according to life-cycle options. Keywords: Life-Cycle Design; Post Mass Production Paradigm; Life-Cycle Simulation; Planning of Product Life Cycle; Modular Structure 1. INTRODUCTION Modern capitalism and industrialism now dominate the global economy. As a result of the competition to develop manufacturing technologies, cheaper products with higher quality have been offered, yielding tremendous social advantages such as high wage. However, there have been disadvantages as well. One of the most serious disadvantages is the environmental issues. Due to the rapid development of technologies, quantity of discarded products is rapidly growing. As a result, mass production and mass consumption have revealed the essential limitation of the earth, such as natural resource availability, energy supplies, and the ability of the biosphere to accept industrially generated waste. This means that we cannot manufacture products as much as we want. Rather, environmental, market, and other Reprint requests to: Yasushi Umeda, Department of Mechanical Engineering, Graduate School of Engineering, Tokyo Metropolitan University, Minami-Osawa 1-1, Hachioji, Tokyo , Japan. umedayasushi@c.metro-u.ac.jp limitations determine acceptable amounts of products. To fundamentally attack and remove these problems, we need to reconsider the current mass production paradigm and to reduce the production and consumption volume of artifacts to an adequate and manageable size balanced with natural and social constraints. We call this idea the Post Mass Production Paradigm ~PMPP! ~Tomiyama, 1997!. In other words, PMPP aims at reducing production volume while maintaining living standards and corporate profits by decoupling economic growth from material and energy consumption. To do so, PMPP includes two approaches ~Tomiyama, 1997!: One is to close a product life-cycle loop by encouraging maintenance, remanufacturing of products, reuse of components, and recycling of materials. The other is dematerialization ; in other words, selling services products perform rather than just selling products. This can be done, for example, by leasing products to consumers and offering various services 149

2 150 Y. Umeda et al. such as upgrading, maintenance, operation supports, and collection of discarded products. As a result, PMPP assumes that an artifact circulates in its life cycle repeatedly and works as a carrier of various services customers need and, therefore, manufacturing industry changes into life-cycle industry that designs and manages the whole product life cycles and offers various services throughout the life cycles. In other words, it is almost impossible without changing business strategies to shift traditional open-loop life cycles under mass production to sustainable closed-loop life cycles. PMPP requires a structural change of design; namely, while designers traditionally design just products, we should design product life cycles as a whole in addition to the product design. This is the central issue of the life-cycle design. The objective of this research is to propose a methodology for the life-cycle design to establish sustainable closedloop product life cycles. To do so, this paper proposes design of life cycle as an indispensable element of the life-cycle design and a decision support tool for the design of life cycle called Life-Cycle Simulation System. The simulation system evaluates product life cycles from an integrated view of environmental consciousness and economic profitability and optimizes the life cycles. This paper also discusses feasibility and advantages of this simulation system by illustrating a case study; namely, modular design of a refrigerator. The rest of the paper is organized as follows. Section 2 clarifies a missing element of existing life-cycle design methodologies and proposes the design of life cycle. To support the design of life cycle, Section 3 describes our method for modeling product life cycles and Section 4 proposes Life- Cycle Simulation System. After illustrating a case study in Section 5, we discuss advantages and issues of the simulation system and the design methodology in Section 6. Section 7 concludes this paper. 2. DESIGN OF LIFE CYCLE We must consider life-cycle issues of products, such as marketing, material acquisition, design, production, logistics, use and operation, maintenance, reclamation, reuse, recycling, and discarding. The present manufacturing industry provides services from marketing through sales; however, reclamation, recycling, and discarding of products are normally outside the domain of manufacturing industries. In contrast, the boundaries and limitations, imposed by the environment, will force the manufacturing industry to gradually provide services in these later stages of the product life cycle. Examples of this movement include the ideas of environmental auditing and the recent Japanese legislation for forced take-back ~e.g., Electric Appliance Recycling Law!. Consequently, for most of the manufacturing industries new areas of value addition and market creation will be in the maintenance, reclamation, recycling, and discarding phases of a product s life. In this sense, manufacturing industry could be redefined as life-cycle industry. The objective of life-cycle design of this research is to design sustainable closed-loop product life cycles. We here define a sustainable closed-loop life cycle as a life cycle that can minimize material and energy consumption, amount of waste, and environmental emissions from the viewpoint of the whole life cycle, while maintaining living standards and corporate profits. For example, if a product is properly designed, corporate profitability can be maintained by supplying various services at later stages. However, if the product is not well designed, these services will be cost factors rather than profit factors. For supporting the life-cycle design, various concepts and Design for X ~DfX! methodologies have been proposed; examples include industrial ecology ~Graedel & Allenby, 1995!, life-cycle design ~Ishii, 1995; Hata et al., 1997!, life-cycle engineering ~Feldmann, 1994!, life-cycle costing, design for disassembly ~Boothroyd & Alting, 1992; Jovane et al., 1993!, design for recyclability ~Krause & Scheller, 1994; Lee et al., 1997!, design for serviceability ~Gershenson & Ishii, 1991!, and end-of-life design ~Ishii, 1999!. However, because these DfX methodologies just focus on certain aspects of design objects ~e.g., disassemblability!, they do not support the whole viewpoint of product life cycles. For example, recycling ~especially material and energy recycling! is in demand to reduce waste and natural resource consumption. Although design for recycling and design for disassembly might increase recyclability of a product, these methodologies cannot tell whether materials used in a product life cycle are balanced with such recycled materials or not. In the worst case, encouraging design for recycling results in creating huge amount of low quality, expensive, and unusable recycled materials ~Umeda, 1999!. To solve this problem, we should model, evaluate, and design flows of material, energy, and money in a product life cycle. Let us call this step of life-cycle design design of life cycle. In other words, design of life cycle is an indispensable element of life-cycle design, where a designer designs balances of the whole life cycle from various viewpoints including material, energy, and money. Life-Cycle Assessment ~LCA! ~e.g., Wenzel & Hauschild, 1994! is a strong tool for evaluating material and energy consumption and emissions of a life cycle. However, it cannot evaluate balances of a life cycle in terms of material, energy, and money, especially when the life cycle has loops such as remanufacturing, reuse, and recycling. To support the design of life cycle, we here take a simulation-based approach; in other words, we develop a decision support tool that simulates flows in a life cycle and optimizes them. 3. MODELING OF PRODUCT LIFE CYCLES This section describes how we model product life cycles. A life-cycle model consists of a product model and a process

3 Study on life-cycle design for PMPP 151 Table 1. Types of life-cycle models Traditional type Recycling type Reuse type Maintenance type PMPP type The traditional product life cycle is today s most common life cycle of consumer electronics in which a customer buys a product and throws the entire product away without executing maintenance when the product becomes faulty. This is similar to the traditional type, but discarded products are recycled. Here we assume that metals are material recycled and plastics are energy recycled for keeping simplicity. This type of life cycles is also practiced in the real world. Current life cycle of automobiles is an example of this type. In this type, modules of collected discarded products are reused if possible, and other modules are recycled. Examples of this type include photocopiers. In this life cycle, after buying a product, a customer continues to use the product by executing maintenance ~i.e., replacing broken modules with new modules! and broken modules just become waste. However, if maintenance cost at one time is too expensive for a customer, he or she replaces it with a new product instead of executing maintenance. Examples of this type include machine tools. This life cycle includes processes of maintenance, modular, and component reuse, and recycling and is appropriately managed ~for instance, products are leased rather than sold and, as a result, collection rate is higher than other types!. Therefore, this type is considered as one of ideal closed-loop life cycles ~see Fig. 1!. network. Here, we assume that a product is modularized to ease tasks such as maintenance, disassembly, and reuse. And we assume consumer electronics like refrigerators and washing machines as an example of the target products. We constructed typical five models of product life cycles shown in Table 1 and compare them in terms of amount of waste, material and energy consumption, and corporate profits. These models are quite simplified but useful for discussing essential advantages and drawbacks of recycling, reuse, and maintenance options. First, we construct the most complex model ~we call it PMPP type!, which includes processes of maintenance, modular, and component reuse. And then we construct other life-cycle models from the PMPPtype model by changing parameter values Model of product We model a product as follows, because we assume modularized products as described above: A product is modeled as a connectivity graph of modules and a module is modeled as a connectivity graph of components that are basic elements in this model. A component is modeled as a set of attributes, such as possibility of recycling and reuse, manufacturing cost, manufacturing energy, lifetime, weight, and material. A component cannot be disassembled nor repaired. In other words, collected used components can be reused as they are, recycled, or just dumped. A module can be repaired by replacing broken components with working ones Process network We represent a product life cycle as a network of processes such as manufacturing, operation, recycling, and remanufacturing ~see Fig. 1!. Important processes in the PMPP type are modeled as follows 1 : Market The size of the market ~market size! is constant and only the target product is sold in this market. After starting to sell the target product, a consumer can only buy the target product and the market is filled with the product in a certain time. In other words, this market model Fig. 1. PMPP-type life-cycle model. 1 Here, main parameters are shown in italic.

4 152 Y. Umeda et al. does not include competition among different products in the market for the sake of simplicity. It is one of our future issues to detail the market model. Operation and maintenance A customer buys a new product for its purchase price. This price does not include maintenance cost nor recycling cost. Each component will break down randomly around its lifetime and the customer repairs the product by replacing a broken module with a working one. The maintenance cost C the customer pays for each maintenance task is calculated with the equation C C m C f, where C m and C f denote module price and maintenance fee, respectively. However, if the maintenance cost C is too high or the product breaks down too often, the customer replaces the entire product with a new product. This customer s judgment is represented as maintenance preference C p ~0 C p 1! and minimum acceptable Mean Time Between Failures ~MTBF!, MTBF min. That is, if the following condition is satisfied, the customer repairs the product, otherwise he or she buys a new product, where T, P, and MTBF~T! denote a certain period of time ~e.g., a year!, product price, and MTBF of the product during the period T, respectively: ( C C p P & ~MTBF~T! MTBF min!. ~1! T Collection and Product Disassembly A certain c ~0 R c 1!# of discarded products is collected to the product disassembly process and the rest are just dumped. The parameter denotes that the collection system of discarded products is not perfect. The product disassembly process disassembles collected products into modules and sends them to the module inspection process. Module Inspection Modules disassembled from discarded products and modules collected from the maintenance process are sent to the module inspection process. The module inspection process classifies collected modules into the following three kinds: 1. If a module is not faulty and reusable, the module is sent to the module warehouse with light reconditioning ~e.g., cleaning!. This is judged by evaluating all components in the module with the following equation; namely, if all components in the module satisfy this equation, the module is sent to the module warehouse. faulty~c! & reuse~c! & ~LT~c! OT~c! GT~c!! ~2! Where, c a component in the module, reuse~c! a flag, which is defined by the user, representing whether the component c is reusable or not, LT~c! nominal lifetime of the component c, OT~c! operation time of the component, and GT~c! a threshold value representing minimum guarantee time of the module. 2. If a module includes unreusable components that do not satisfy Eq. ~2!, the module goes to the rebuild process or the module disassembly process. This judgment is formalized by the following equation; namely, if it is easy to recover the module, it goes to the module rebuild process. If the module does not satisfy this equation, it goes to the module disassembly process. Where, n~c! n~c unreusable! RR min ~3! n~c! C all components included in the module, n~c! number of a components set C, C unreusable a set of components which are unreusable ~C unreusable C!, and RR min a threshold value called minimum rebuild ratio. Rebuild In the module rebuild process, modules are remanufactured by replacing unreusable components with working ones that came from the component warehouse. Rebuilt modules are sent to the module warehouse. Module disassembly and component inspection In the module disassembly process, modules are disassembled into components. And these components are inspected in the component inspection process. Namely, if a component satisfies Eq. ~2!, it is sent to the component warehouse with light reconditioning. Otherwise, the component goes to the recycling process if it is recyclable and, if not, it is just dumped. Recyclability of components are specified by the user in the model of products. Recycling In the recycling process, the components are recycled into material or energy. If material of a component is metal it is recycled into material, if plastics, it

5 Study on life-cycle design for PMPP 153 is energy recycled, and others are just dumped. In the material recycling process, a certain weight recycling ratio (MRR)# of input components is produced as recycled materials that will be used for component manufacturing, and the rest ~1 MRR! is dumped. The energy recycling process generates energy for other processes on one hand, but generates waste on the other hand; namely, a certain weight ratio ~waste ratio of energy recycling (WRER)! of input plastic components becomes waste. Warehouses Modules in the module warehouse are used for maintenance as spare modules and for assembling new products. If the number of modules in the warehouse is smaller than a certain threshold, new modules are manufactured. The component warehouse works in the same manner. 4. LIFE-CYCLE SIMULATION SYSTEM To design closed-loop life cycle that might include multiple reuse of components, it is quite important to evaluate the whole balance of the product life cycle from a technical and economic viewpoint at the early stage of design. In the early stage of design, accurate information about the whole life cycle is not always available. Even in such a case, the balance should be evaluated. Of course, CAD0CAM systems supply precise design information about the product and PDM systems give us detailed bills of products, materials, and processes. With these precise data, the evaluation will be very accurate. However, it is rare that all such information is available at early stage of design and, moreover, additional kinds of information especially about downstream processes, which traditional manufacturing industries do not consider, is important for evaluating the whole life cycle. Examples of such information include practical lifetimes of components, customer behavior, reuse rate, collection rate, and recycling rate. Therefore, this evaluation should be able to be calculated with precise data and without precise data. To solve these problems, we have developed a life-cycle simulation ~LCS! system that supports the design of life cycle described in Section 2 at the planning and conceptual design stages. The system supports a designer to determine life-cycle strategies of a design object by constructing a model of its product life cycle with visual user interface and by executing what-if analyses with life-cycle simulation of the model and its optimization. For example, the designer can determine design requirements ~e.g., reuse rate should be higher than 30%! to achieve global goals ~e.g., to reduce amount of waste into half!. The LCS system consists of the following four tools ~see Fig. 2!: 1. The life-cycle model editor supports a designer to construct a model of life cycle of a design object. Fig. 2. Architecture of life-cycle simulation system. 2. The life-cycle simulator executes simulation of a product life cycle. 3. The life-cycle database manages the data of products, components, and processes. 4. The life-cycle optimizer optimizes parameter values in the target life-cycle model and modular structure of the design object by using the genetic algorithm technique ~Holland, 1975!. We use the genetic algorithm because the life-cycle model has a lot of parameters including modular structure to be optimized and the search space is not continuous. We assume two kinds of users for this system. The first kind are experts who know the product life cycle well and can construct life-cycle models. The other are designers who use these models and execute what-if analyses of the model by using the simulator and the optimizer Model editor The life-cycle model editor consists of two visual editors ~a model editor and a process editor! for constructing and modifying models of product life cycles. As described in Section 3, a model of a product life cycle consists of a product model and a process network. The user can easily construct the life-cycle model by using libraries of products, components, and processes stored in the life-cycle database ~see Section 4.3!. Figure 3 shows a screen hardcopy of the life-cycle model editor. As shown in this figure, a process is described by given parameters, input parameters, output parameters, and a procedure. Given parameters represent attributes of this process and of which values should be given by the user or

6 154 Y. Umeda et al. Fig. 3. Screen hardcopy of the life-cycle model editor. determined by the optimizer and input and output parameters represent parameters coming from0to other processes. The procedure describes behavior of this process by using Smalltalk code Life-cycle simulator The simulator simulates flows of products, materials, money, and information based on a given life-cycle model using discrete event simulation techniques like the Petri Net. For each turn, 3 the simulator executes procedures in the process definitions and propagates parameter values among the 2 This system is implemented on Visualworks0Smalltalk. 3 For example, a turn corresponds to a month. process network according to parametric dependency of the process network. The parametric dependency is automatically derived by the system from the process definitions. As a result, for instance, the number of manufactured products, amount of disposals, and total sales figure in that turn are calculated. The simulator repeats this cycle until it reaches a final turn given by the user Life-cycle database The database manages the following data and supports the user to construct these libraries; a library of products represented by component connectivity graphs, hierarchical catalog of components used in products,

7 Study on life-cycle design for PMPP 155 a library of life-cycle processes used for the process network, a library of evaluation functions used for the optimization ~see Section 4.4!, and a library of upgrade plan of products used for designing upgradable products. This library will be used in future Optimizer The optimizer optimizes the target life-cycle model by using the results of the simulator. Namely, it optimizes values of given parameters in the process network and modular structure of the design object by using the genetic algorithm. The latter modularization is executed by aggregating neighboring components into a module in the component connectivity graph ~see Fig. 4!. Basic algorithm of the optimizer is shown as follows: 1. Initialization After the user specifies given parameters to be optimized in the process network and evaluation functions to be used, a target life-cycle model is sent to the optimizer. 2. Generation of a set of genes The optimizer generates a set of genes each of which represents values of target parameters and a module structure. 3. Simulation For each gene, the simulator executes the life-cycle simulation by using the values the gene represents. 4. Evaluation The optimizer evaluates simulation result of each gene with the given evaluation functions. 5. Operation of genes According to the evaluation results, the optimizer generates a next generation of genes by applying operations such as elite selection, crossover, and mutation. Fig. 4. Optimization of modular structure. 6. Repetition The optimizer repeats the steps 3 5 until the number of generation reaches a certain value given by the user. After the repetition, the optimizer selects the best gene as its result. 5. CASE STUDY: MODULAR DESIGN OF A REFRIGERATOR As an application of the life-cycle simulation system to modular design, we executed the life-cycle simulation and the optimization by taking a refrigerator as an example. Here, we collected data from real products as much as possible Simulation conditions and results To make a refrigerator modular and reusable, we identified the following three major problems of current refrigerator design: Inner parts of the cabinet, which have direct contact with foods, cannot be reused for hygienic reason. The cabinet, in which outer body, inner parts, and pipes are glued with heat insulators, and thus cannot be disassembled maintaining reusability. Gaskets between the body and the doors are most easily deteriorated. Therefore, we here make the following assumptions: Gaskets can be replaced by the maintenance task. Outer parts, which affect attractiveness of the refrigerator, and inner parts of a cabinet will not be reused. A cabinet is decomposed into five independent parts; namely, a structural frame, pipes, an outer part, an inner part, and heat insulators. Figures 5 and 6 show a structural sketch of the refrigerator and its part connectivity graph, respectively. This model of the refrigerator is commonly used in the five types of life-cycle models, but modularized when optimization is executed. For simplicity, we assume that weight and operation energy of the refrigerator is the same even if it is modularized. The parameter lists of the components and the processes are shown in Tables 2, 3, and 4. In these tables, we collected real data such as material, cost, weight of each component and energy consumption of assembly, recycling, components manufacturing processes. However, it was impossible to collect real data of, for example, energy consumption and costs of the rebuild process because reuse of refrigerator is not practiced in the real world. Therefore, we had to assume these values based on the manufacturing processes. We compare five types of life cycles ~the traditional, recycling, reuse, maintenance, and PMPP types! described in Section 3. As shown in Table 4, we execute several simu-

8 156 Y. Umeda et al. Fig. 5. Structure of the refrigerator. Fig. 7. Evaluation functions. lations for each type by changing conditions, including maintenance preference and collection rate, to discuss sensitivity of these conditions. The evaluation functions for the optimizer is shown in Figure 7. These functions are commonly used for the four Fig. 6. Part connectivity graph of the refrigerator. types of models other than the traditional type because the traditional type is the reference model and, therefore, is not optimized. Horizontal and vertical axes in each function in Figure 7 denote relative value to the traditional type and evaluation value ~between 0 and 1 and 1 is the best!, respectively. For example, the revenue function indicates that if total revenue of a model is between 50 and 150% of that of the traditional value, evaluation is good ~namely, 1!. The sum of values of four functions are the evaluation of its product life cycle. Simulation results are shown in Figures 8, 9, and 10. These figures show accumulated values relative to those of the traditional type. Figure 8 shows values of disposal amount, which is in proportional to material consumption, and amount of energy consumption excluding the operation energy. Figure 9 shows corporate revenue and profits. Revenue is the sum of product sales income and income from maintenance. Therefore, revenue also implies total cost for the customers. This means the value of revenue should not be too high for the customer s satisfaction. Profit is also the sum of profits from product sales and that from maintenance. This value should not be too low for the corporate sustainability. And Figure 10 illustrates ratio of revenue; as shown in this figure, revenues from the maintenance section are remarkable in the maintenance type and the PMPP type. Moreover, Figure 11 depicts optimized modular structure of the modular product for each life-cycle type.

9 Study on life-cycle design for PMPP 157 Table 2. Parameter value of components Name Main Material Weight ~g! Reusability Way of Recy. Price ~yen! Mfg. Cost ~yen! Mfg. Energy ~kwh! MTBF ~month! Cabinet frame Fe true M Cabinet Plas false T Cabinet pipe Cu 326 true M Duct Plas false T Fan motor Fe 483 true M Evaporator case Plas. 897 true T Accumulator Fe 177 true M Evaporator Al 532 true M Back grill Fe 986 true M Compressor Fe 7985 true M Sideboard Fe 980 true M Radiator Fe 2669 true M Duct Plas. 633 true T Base Fe 1240 true M Door 1 Fe 2693 true M Door 2 Fe 669 true M Door 3 Fe 1828 true M Door 4 Fe 1834 true M Gasket Plas. 100 false D Door plas. Plas false T SPCB Plas false D MPCB Fe 1564 true D Heater Al 112 true M Tank Plas false T Dryer Cu 111 true M Total T thermal recycling; M material recycling; and D disposal ~not recyclable!. Table 3. Parameters of life-cycle processes Process Energy ~kwh! Cost ~yen! Unit Assembly ~M! component Disassembly ~M! component Assembly ~P! module Disassembly ~P! module Operation month Logistics product Maintenance product Rebuilt component Disposal kg Inspection ~M! module Inspection ~C! component Material P. ~Cu! kg Recycle ~Cu! kg Material P. ~Fe! kg Recycle ~Fe! kg Material P. ~Al! kg Recycle ~Al! kg Material P. ~Plas.! kg Thermal recy. ~Plas.! kg P product; M module; and C component. 6. DISCUSSIONS The results of simulation can be interpreted as follows: 1. While simple recycling decreases amount of waste, energy consumption cannot be decreased so much because additional energy is used for the material recycling process. Economically, the recycling type decreases corporate profits because of additional costs in the recycling process. As a result, the simple recycling might not be a good strategy for realizing PMPP. 2. Simple reuse is more effective for reducing the environmental loads ~namely, reduction of waste and energy consumption! than the simple recycling. However, the energy consumption does not decrease comparing to decrease of waste amount. Economically, profitability ~ratio of profits over revenue! is the second best among all types because in this condition, reuse of components makes their MTBFs shorter and, as a result, replacement of products happens more often. 3. The maintenance type has less effects to the environmental loads than other types because discarded mod-

10 158 Y. Umeda et al. Table 4. Life-cycle parameters Simulation turn 300 ~month!* Maintenance preference 0.33, 0.66, 1.0 Maintenance fee to be optimized between 0 100,000 ~yen! Collection rate 0, 30, 60, 90% Guarantee time 24 ~month! Market size 1000 ~products! MTBF min 6 ~month! material recycling ratio ~MRR! 0.7 waste ratio of energy recycling ~WRER! 0.2 Recent turn for Eq. ~1!# 24 ~month! *To get stable simulation results, simulation turn should be long enough to cause many replacement of products. Simulation turn 300 months might be too long in real world, but it is determined for this reason. ules are just dumped in this type. However, corporate profitability is better than the recycling type because of the additional business opportunity of maintenance. 4. Finally, the PMPP type has the best results for environmental loads and corporate profitability. Advantage of the PMPP type against other types is clear and this result suggests plausibility of PMPP. Based on these results, we can conclude that: Simple recycling is not a good strategy in terms of reduction of energy consumption. The PMPP type is the best strategy for realizing PMPP ~namely, reducing energy and material consumption with maintaining living standards and corporate profits!. In other words, good combination of maintenance, reuse, and recycling is essential. Therefore, the life-cycle simulation tool is quite important for supporting the design of life cycle. Fig. 9. Simulation results ~2!: Economy. It is also remarkable that these objectives are achieved by the structure change of industry from just manufacturing to the life-cycle industry as shown in Figure 10. This result supports the concept of PMPP. Figure 11 can be interpreted as follows: For optimizing the reuse type, larger-sized modules are better because disassembly cost of modules can be saved. For optimizing the maintenance type, smaller-sized modules are better because it can avoid unnecessary replacements of working components. Modularization in the maintenance type is executed based on similarity of lifetimes of components. In the PMPP type, optimized modular structure is between the reuse and maintenance types, because this type has reuse and maintenance processes. Further analysis of the modular structure in this type is one of our future issues. Traditionally, modularization has been a powerful approach to increase concurrency and collaboration of manufacturing for reducing manufacturing costs. The viewpoint Fig. 8. Simulation results ~1!: Environmental loads. Fig. 10. Simulation results ~3!: Industrial structure.

11 Study on life-cycle design for PMPP 159 Fig. 11. Optimized modular structure. of modularization discussed here focuses on efficiency of circulation of modules throughout the product life cycle. While the technique is still simple ~for instance, we only use component connectivity graph so far!, we believe that it offers an indispensable new viewpoint of modularization for the design of life cycle. In this sense, this modularization is complementary to the traditional one. As a result, we can conclude that this life-cycle simulation is useful for modular design because it can give a hint of modular structure in terms of life-cycle loop to a designer by executing the optimization. Other lessons learned from this case study are: The comparison of these five types is executed under the condition shown in Tables 2, 3, and 4. The following observations regarding the relationship between conditions and simulation results are drawn: Obviously, waste amount deeply depends on collection rate. But the PMPP type is always superior to other types under the same collection rate and, moreover, we can expect better collection rate than other types owing to its better life-cycle management because we assume the life-cycle industry rather than just manufacturing industry under the PMPP. Energy consumption depends on ratio of energy consumption of the inverse processes to that of manufacturing processes. If the inverse processes consume more energy, the closed-loop life cycle has no advantage in energy. Our observation is that smaller

12 160 Y. Umeda et al. loop ~i.e., maintenance and reuse! is good strategy for saving energy. Waste amount and profitability of the reuse type depends on lifetimes of components. On one hand, if lifetimes are too short, reuse cannot be executed. On the other hand, if lifetimes are too long, profitability decreases because customers do not buy new products. Waste amount and profitability of the maintenance type and the PMPP type depends on a nonlinear relationship among the maintenance fee, lifetimes of components, and the maintenance preference. For example, if the maintenance preference is too low, maintenance is not executed and, as a result, waste amount increases and profitability decreases because the company cannot get the maintenance fee. Here we assume that profitability of maintenance is higher than that of product sales. Therefore, in the design of life cycle, appropriate setting of the maintenance fee and design of component lifetimes are important; that is, life time should not be so short that too much replacements of products occur and should not be too long to keep enough profits. Therefore, to realize environmentally conscious product life cycles, besides increasing collection rate and developing energy saving technologies, of which importance are always pointed out by many researchers, the simulation results clarified importance of strategic life-cycle design and management, including strategic pricing of purchase price and maintenance fee, and design of component lifetimes and product modularization. We could not collect a part of data needed for executing the simulation because such data do not exist. It is important to create and collect these data by, for instance, executing experiments. Although our simulation models are just examples, these models are the central knowledge for supporting a designer to use this system. That is, these five types of life-cycle models are general enough to support a designer to easily execute simulations of his0her own design objects for different types of life cycles just by adding his0her own data to the models. By comparing with the LCA tools, we can point out three advantages of this simulation tool for supporting the design of life cycle at the conceptual design stage. First, this tool evaluates material and energy balances of the whole life cycles and economic aspect ~corporate profits and customer s costs!, which are indispensable factors for design but not dealt with in LCA. Second, while LCA requires huge amount of basic data, this tool can execute simulations with smaller amount of data that might be ambiguous. Even in such a case, the designer can find out useful information ~e.g., critical points in a product life cycle! by executing what-if analyses. And third, the designer can see and edit visually a product life cycle with this tool and optimize it. These features are indispensable for supporting the design of life cycle. 7. CONCLUSION This paper proposed the idea of design of life cycle and the life-cycle simulation system for supporting it. The results of the life-cycle simulation shown in Section 5 indicate that the combination of the closed-loop life cycle and the maintenance service is essential for realizing PMPP and this system is powerful enough for supporting the design of life cycle. Future works include: to apply this system to various case studies, especially we are planning to study design of upgradable products of a fax with this tool, usage of more detailed cost breakdowns, including nonrecurring and recurring costs, and to integrate this system with other tools such as CAD, PDM, various DfX tools, and LCA tools. REFERENCES Boothroyd, G., & Alting, L. ~1992!. Design for assembly and disassembly. Annals of the CIRP 92 41, Feldmann, K. ~1994!. Life cycle engineering Challenge in the scope of technology, economy and general regulations. RECY 94 (Second Int. Semin. Life Cycle Eng.), pp FAPS, University of Erlangen, Germany. Gershenson, J., & Ishii, K. ~1991!. Life-cycle serviceability design. Proc. Design Theory and Methodol. DTM 91, Vol. DE-31, Graedel, T.E., & Allenby, B.R.~1995!. Industrial ecology. Prentice Hall, Englewood Cliffs, NJ. Hata, T., Kimura, F., & Suzuki, H. ~1997!. Product life cycle design based on deterioration simulation. In Life Cycle Networks, ~Krause, F-L. and Seliger, G., Eds.!, pp Chapman & Hall, London. Holland, J.H. ~1975!. Adaptation in natural and artificial systems. University of Michigan Press, Ann Arbor, MI. Ishii, K. ~1995!. Life-cycle engineering design. ASME J. Mechanical Design 117, Ishii, K. ~1999!. Incorporating end-of-life strategy in product definition. Proc. EcoDesign 99, Jovane, F., Alting, L., Armillotta, A., Eversheim, W., Feldmann, K., Seliger, G., & Roth, N. ~1993!. A key issue in product life cycle: Disassembly. Ann. CIRP 93 42, Krause, D., & Scheller, H. ~1994!. Design for recyclability and economical planning of disassembly based on the Recyclinggraph tool. RECY 94 (Second Int. Semin. Life Cycle Eng.), pp FAPS, University of Erlangen, Germany. Lee, B.L., Rhee, S., & Ishii, K. ~1997!. Robust design for recyclability using demanufacturing complexity metrics. Proc. ASME DTC0CIE. Tomiyama, T. ~1997!. A manufacturing paradigm toward the 21st century. Integrated Comput. Aided Eng. 4, Umeda, Y. ~1999!. Key design elements for the inverse manufacturing. Proc. EcoDesign 99, pp IEEE, Los Alamos, CA. Wenzel, H., & Hauschild, M. ~1994!. Environmental life cycle assessment a tool in product development. RECY 94 (Second Int. Seminar on Life Cycle Engineering), pp FAPS, University of Erlangen, Germany. Westinghouse Corporate Services Council ~1984!. Report on life cycle costs. Westinghouse Productivity and Quality Center, Pittsburgh, PA. Dr. Yasushi Umeda has been Associate Professor at the Department of Mechanical Engineering, Graduate School of En-

13 Study on life-cycle design for PMPP 161 gineering, Tokyo Metropolitan University since He received his Ph.D. in precision machinery engineering from the Graduate School of the University of Tokyo in His research interest include Life-cycle design, inverse manufacturing, maintenance engineering, and functional reasoning. Akira Nonomura has been researcher at DENSO Co. Ltd. since He graduated from the University of Tokyo in His research interests include life-cycle design and computer engineering. Dr. Tetsuo Tomiyama has been Professor at Research into Artifacts, Center for Engineering, the University of Tokyo since He received his Ph.D. in precision machinery engineering from the Graduate School of the University of Tokyo in His research interests include design theory and methodology, knowledge intensive engineering, service engineering, applications of qualitative physics, large scale engineering knowledge bases, soft machines ~selfmaintenance machines and cellular machines!, and manufacturing paradigms.