Preference Set-Based Design Method for Sustainable Product Creation Masato Inouea,1, Kai Lindowb, Rainer Starkc, and Haruo Ishikawad a
Assistant Professor, Department of Mechanical Engineering and Intelligent Systems, The University of Electro-Communications (UEC Tokyo), Japan. b Research Assistant, Technische Universität Berlin, Department of Industrial Information Technology, Germany. c Professor, Technische Universität Berlin, Department of Industrial Information Technology, Germany. d Professor, Department of Mechanical Engineering and Intelligent Systems, The University of Electro-Communications (UEC Tokyo), Japan. Abstract. The multi-objective satisfactory design including not only physical performances but also a sustainable aspects from a global environmental protection perspective is necessary. When it comes to a sustainable life cycle, decision-making at the early phases of design is important. The design phase contains multiple sources of uncertainties in describing design. Therefore, handling the uncertainties in the design phase has a great importance. We have proposed a preference set-based design (PSD) method that enables the flexible and robust design under various sources of uncertainties while capturing designer’s preference. As a first step, we investigated different officially-accepted sustainability indicators and identified which indicators are related to the product development process. Thereafter, the proposed method is applied to a multi-objective design problem including technical performances and sustainable issues. This paper discusses the applicability of the PSD approach for sustainable development using the example of an alternator design. Keywords. Sustainable product development, set-based design method, lean product development, early phase of design, product life cycle, multi-objective design
1 Introduction Due to the complexiy of today’s products and incorporated overlapping activities between the lifecycle functions, the multi-objective satisfactory design including not only physical performances but also sustainable aspects from a global environmental protection perspective is necessary. Since essential attributes and 1
Assistant Professor, Department of Mechanical Engineering and Inelligent Systems, The University of Electro-Communications, 1-5-1 Chofugaoka, Chofu-shi, Tokyo 182-8585, JAPAN; Tel: +81 (0) 42 443 5420; Fax: +81 (0) 42 484 3327; Email:
[email protected]; http://www.ds.mce.uec.ac.jp/index-e.html
2
M. Inoue, K. Lindow, R. Stark and H. Ishikawa
characteristics of products are already determined in the stages of product development, it is necessary to integrate sustainable aspects into the product development process. The implementation of the concept of sustainable development requires the use of lean product development method in product creation processes. The product properties which are defined during the process of product creation should support and ensure sustainable development throughout the entire product life cycle [1]. When it comes to a sustainable life cycle, decisionmaking at the early phases of design is important. Moreover, the early phase of design contains multiple sources of uncertainties in describing design. Therefore, handling the uncertainties in the early phase has a great importance. The previous series of our studies have proposed a preference set-based design (PSD) method that enables the flexible and robust design under various sources of uncertainties while incorporating designer's preference structure at the early phase of design [2]. This paper presents an approach to sustainable product creation based on the PSD method and discusses the availability of our proposal for obtaining the multi-objective satisfactory solutions not only about technical performances but also about sustainable issues. Finally, to verify the new approach, the PSD method is applied to an alternator design.
2 Preference Set-Based Design Method (PSD Method) The PSD method consists of four steps, the set representation, set propagation, set modification, and set narrowing, which are described in the following. 2.1 Set Representation To capture the designer’s preference structure on a continuous set, both an interval set and a preference function defined on this set, which is called the “preference number (PN)”, are used. The PN is used to specify the design variables and performance requirements, where any shapes of PN are allowed to model the designer’s preference structure, based on designer's knowledge, experience, or know-how. The interval set at the preference level of 0 is the allowable interval, while the interval set at the preference level of 1 is the target interval that the designer would like to meet. 2.2 Set Propagation and Modification The set propagation method, which combines the decomposed fuzzy arithmetic with the extended interval arithmetic (i.e., Interval Propagation Theorem, IPT [3]), is proposed to calculate the possible performance spaces which are achievable by a given initial design space. Afterwards, if all the performance variable spaces have the common spaces (i.e., acceptable performance space) between the required performance spaces and the possible performance spaces, there is a feasible subspace within the initial design space. Otherwise, the initial design space should be modified in set modification process.
Preference Set-Based Design Method for Sustainable Product Creation
3
2.3 Set Narrowing If the overlapping regions between the possible performance spaces and the required performance spaces exist, there are feasible design subspaces within the initial design space. However, if the possible performance space is not the sub-set of the required performance space, there also exist infeasible subspaces in the initial design space that produce performances outside the performance requirement. Then, the next step is to narrow the initial design space to eliminate inferior or unacceptable design subspaces, thus resulting in feasible design subspaces. To select an optimal design subspace out of those feasible design subspaces, robust design decisions need to be made to make a product’s performance insensitive to various sources of variations. The present method has been also used to define the possible design space by capturing the designer’s preference structure. In addition to the design robustness, we should take into account which one is preferred by the designer. The design preference and robustness are evaluated to eliminate infeasible design subspaces. 2.4 Design Metric for Design Preference and Robustness We have proposed new design metric for design preference and robustness, what is called the preference and robustness index (PRI) [2]. A high design preference means that there are large feasible design subspaces within the designer’s required performance spaces. On the other hand, design robustness includes the accuracy, convergence and stability of design. Since more than one performance variable are commonly considered in the multi-objectives design problem, the PRIs for multiple performances need to be aggregated, what is called aggregated PRI (APRI), to provide the effectiveness of the design alternatives with respect to all performances. A family of parameterized aggregation functions is used for the multi-objective decision making problem, based on the weighted root-mean-power: 1
( PRI1 ) s n ( PRIn ) s s (1) APRI s (( PRI1 , 1 ),, ( PRIn , n )) 1 n By varying the parameter s, the expression Equation 1 produces some wellknown averaging operators: min, harmonic mean, geometric mean, arithmetic mean, quadratic mean, and max. Finally, the set narrowing method first eliminates infeasible or unacceptable design subspaces that produce the performances outside the performance requirement, and then selects an optimal one from a few feasible design subspaces, which are more preferred by the designer and provide better design robustness (i.e., the highest APRI measure). 1
4
M. Inoue, K. Lindow, R. Stark and H. Ishikawa
3 Investigation of Sustainability Indicators 3.1 Sustainability Indicators In order to consider the sustainability of technical products, we investigated different officially-accepted sustainability indicators. Additionally, we identified which indicators are related to the product development process. The National Institute for Environmental Studies (NIES) in Japan has reviewed indicators on sustainable development which were developed by national governments and international organizations, and considered what types of indicators were being used as a database (2006-2008) [4]. This database includes 1,528 issues of sustainability indicators, which can be searched by using a web search engine. Using the database, we selected 32 sustainability indicators which are related to the development process of products. Moreover, we categorized these indicators into three aspects: ecological, economical and social aspects. Furthermore, we selected two sustainability indicators from the selected indicators for further consideration: the CO2 emissions and the mass of products. Both indicators are related to all sustainable aspects and can be calculated quantitatively. The carbon dioxide is one of the green house effect gases and relates to climate change. The goal of sustainable development from an ecological perspective is to not increase the emission amount of CO2. The transportation emission including CO2 is a very important factor, as it is an economical aspect. If the emission amount of CO2 increases, the temperature rises, the amount of production of foods decreases and the price of foods are increasing; finally poor nutrition occurs because of the difficulty for people to gain foods. Therefore, CO 2 emissions are related to health as a social aspect. Weight saving of products is related to the CO2 emissions, the efficiency of transport and the protection of our natural resources. 3.2 Evaluation for Environmental Loads Based on LCA The Japan Vacuum Industry Association (JVIA) has proposed the “JVIA LCA model” which assists designers in figuring out the environmental loads of manufacturing equipment and inventory analysis for environmental loads at the process of product design. The objective of this model is to get information which is needed for LCA of a product. The model can calculate the inclusive sum of environmental loads at each product life cycle process, such as manufacturing process, using process, maintenance process, reuse process, recycle/disposal process. Moreover, this model can also analyze environmental loads at a portion of product life cycle process. In this section, we explain about the evaluation at recycle/disposal process which has no relationship to the performance at the design process. Environmental loads (Aj(a)) [kg], reduced environmental loads (Aj(b)) [kg] and disposal environmental loads (T) [kg] are calculated by the following equations. Aj C j B j (2)
Preference Set-Based Design Method for Sustainable Product Creation
T
A j(a) j
A j (b )
5
(3)
j
Whereas j is the items for environmental loads (disassembly/segregation/cleaning, sub material, transportation, and incineration) or reduced environmental loads (recycle and reuse). Cj is the necessary charge at j(a) or j(b). Bj is the embodied environmental loads intensities at j(a) or j(b).
4 Case Study: Application to an Alternator Design 4.1 Setting of Design Problem In this paper, a design of an alternator structure is chosen to illustrate the effectiveness of the proposed system. The parametric CAD model as shown in Figure 1 was created by defining the design parameters which effect the required performances including physical and sustainable aspects. The purpose of this design is to find the values of two design variables including a height of rotor coil (= height of stator) (X1) and a radius of rotor (X2) as shown in Figure 1. The performance requirements include the considerations on physical performance of the alternator: power (Y1) calculated by Equation 1 and on performance for sustainable aspect: mass (Y2) and CO2 emissions (Y3). nrevolution 2 X1 (4) Y1 I nlayer n pole X2 30 rcoil Whereas nrevolution is the number of revolutions per minute: 2000 rpm, I is current: 70A, μ is magnetic permeability: 2π 10-3, nlayer is the number of layers of rotor coil: 10, npole is the number of poles of rotor: 12, and rcoil is the radius of rotor coil wire 0.25 10-3m. In this study, the values of the parameters except the design variables (X1 and X2) are fixed. The mass of each part can be calculated out of the CAD models using NX as a CAD system.
X1: Height of rotor coil
X1: Height of stator X2: Radius of rotor
Figure 1. CAD model and design variables of the alternator
6
M. Inoue, K. Lindow, R. Stark and H. Ishikawa
4.2 Life Cycle Options for the Alternator In this study, life cycle options for an alternator have been defined. The strategic focus of the options is on the end-of-life (EoL) stage thus the possibilities are remanufacturing, recycling or reuse for each component of the alternator. Information for the development of possible sustainable futures of each component of the alternator can be extracted. The alternator is mainly characterized by standardized parts which are connected with each other by standardized interfaces. This allows that parts could be used in different products and/or product classes. Eventually, standardized interfaces and internationally accepted standards and guidelines allow the manufacturer to globally develop, manufacture and distribute individual parts and assemblies. Once the powertrain reaches EoL it is returned to the manufacturer. Assemblies are partially processed for remanufacturing. This study has an assumption that the components of the stator, rotor coil, rotor, driveshaft, belt fitting, bearing and slip rings are recycled, and the fan and spacer are reused. In general, when it comes to the EoL, the alternators have to be returned to a collection station. The average distance is 800km and the transportation machine is a heavy duty truck (10t, light oil). To specify this data is necessary to calculate CO2 emissions using the PSD method. 4.4 Application of PSD and Results In order to align technical performances with sustainable issues, the PSD method is extended to various sustainable aspects. Due to the high complexity of real sustainable systems, the approach necessitates simplification. We have put the focus on CO2 emissions and mass as a sustainable aspect because both factors have economical, ecological and social impacts on the entire product life cycle. When a design object has some required performances, there are more highly weighted performances or lower weighted performances. To reflect the importance of the required performances in the structure of the alternator, the weight of each required performance is classified: the case of technical performance-orientation (power ω1: mass ω2: CO2 emissions ω3 = 10:1:1) and sustainability-orientation (ω1: ω2: ω3 = 1:10:10) by Equation 1. Figure 2 compares the obtained possible distributions of the power, the CO2 emissions and the mass between the case of performances-orientation and the case of sustainability-orientation. We can confirm that the power performance in case of performance-orientation is better than the one in case of sustainability-orientation. On the other hand, CO2 emissions and the mass in case of sustainability-orientation are lower and smaller than the one in case of performance-orientation. Figure 3 shows the ranged set of solutions of design variables: the height of rotor coil and the radius of rotor. This result indicates that all of the ranged sets of solutions of design variables as shown in solid line are narrowed from the initial preferences of design variables as shown in dotted lines. We can confirm that the design set solutions in case of sustainability-orientation have a narrower rotor coil and rotor because the case of sustainability-orientation fulfills sustainable issues better than the case of performance-orientation. These results show that the multiobjective satisfactory design solutions reflecting the importance of the required
592.2
3.65
592
3.55
Mass [kg]
CO2 emissions [kg]
Preference Set-Based Design Method for Sustainable Product Creation
591.8 591.6
7
Emphasis on performance Emphasis on sustainability
3.45 3.35
591.4
3.25
580
600
620
640
660
580
600
620
640
660
Power [W]
Power [W]
1
1
0.8
0.8
preference
preference
Figure 2. Comparison of possible distributions between performance-orientation and sustainability-orientation
0.6 0.4 0.2
0.6 0.4 0.2
0
0
0.02
0.022
0.024
0.026
0.028
Height of rotor coil [m] Initial preference of design variables
0.03
0.04
0.045
0.05
Radius of rotor [m] Design set solution (A: emphasis on performance) Design set solution (B: emphasis on sustainability)
Figure 3. Comparison of design set solutions between performance-orientation and sustainability-orientation
performances and designer’s intention can be obtained by our proposed PSD method. 4.5 Obtaining a design solution from ranged set of solutions In fact, a designer uses a design value when they draft a design object. The proposed PSD method obtains a ranged set of design solutions. Therefore, we need to define a point solution as a design value from the ranged set of design solutions. This study has an assumption that a designer selects a point solution from the highest value of preference of a design variable as shown in the Figure 3, which means designer’s most preferable region, as a design value. Figure 4 shows the difference of achievable performance value between performance-oriented solutions and sustainability-oriented solutions. The design values by the performance-orientation can achieve high power but heavy performance and high CO2 emissions, meanwhile which by the sustainabilityorientation can achieve light and low CO2 emissions but low power. These results indicate that designer’s preference reflect the alternator.
M. Inoue, K. Lindow, R. Stark and H. Ishikawa
3.44 3.44
591.8 591.8
610 610
3.4 3.40
600 600
3.36 3.36
590 590
3.32 3.32
Emphasis on performance Emphasis on sustainability
CO2 emissions [kg] 11
620 620
CO2 emissions [kg]
591.9
Mass [kg]
3.48 3.48
Mass [kg]
630 630
Power [W]
Power [W]
8
591.7 591.7
591.6 591.6
591.5 591.5
Figure 4. Comparison of achievable performance values between performance-oriented solution and sustainability-oriented solution
5 Conclusions This paper proposed the PSD method that enables the flexible and robust design under various sources of uncertainties while capturing designer’s preference. As a first step, we investigated different officially-accepted sustainability indicators and identified which indicators are related to the product development process. The proposed method was applied to a multi-objective design problem including technical performances and sustainable issues. This presents the applicability of the PSD approach for obtaining the multi-objective satisfactory solutions not only about technical performances but also about sustainable issues. However, further research about product-related sustainability indicators must be carried out. Thereby a major challenge is to quantify the social dimension of sustainability. What has to be achieved, are products that are economically more successful, ecologically more viable and socially more responsible. This research was supported by the Japan Society for Promotion of Science, Grant-in-Aid for Young Scientists (B) No. 21760109. Further research about this topic is supported by the German Research Foundation, DFG PAK 475/1.
References [1] Stark R, Lindow K, Woll R, Kind C. Beitrag der Produktentstehung zur Nachhaltigkeit (Product creation’s contribution to sustainable development), Proc. KT2008; 191–200. [2] Inoue M, Nahm YE, Okawa S and Ishikawa H. Design support system by combination of 3D-CAD and CAE with preference set-based design method, Concurrent Engineering: Research and Applications, in press (February, 2010). [3] Finch W.W, Ward A.C. Quantified relations: a class of predicate logic design constraint among sets of manufacturing, Operating and Other Variations, ASME Design Engineering Technical Conference 1996; 18–22. [4] National Institute for Environmental Studies (NIES). Sustainable development indicators database. Available at: < http://www.nies.go.jp/sdi-db/search_e.php >. Accessed on: Jan. 22nd 2010.
