Int J Adv Manuf Technol (2007) 31: 1251–1259 DOI 10.1007/s00170-005-0292-6
ORIGINA L ARTI CLE
Kai Jin . Hong C. Zhang
A decision support model based on a reference point method for end-of-life electronic product management
Received: 26 May 2004 / Accepted: 5 July 2005 / Published online: 30 June 2006 # Springer-Verlag London Limited 2006
Abstract Although a decision support system is an efficient tool that has been widely used in management areas, it still encounters some problems when it is implemented in an environmentally conscious decision-making problem. Those problems include lack of sufficient and reliable data on environmental impact of end-of-life product treatment, material composition analysis, proper expression of the users’ aspiration levels, system architecture, and the proper optimization method. This paper proposes a systematic decision-making tool for end-of-life (EOL) electronic product recycling scenario analysis and selection. This model can help original equipment manufacturers (OEMs) specify the most satisfactory recyclers based on the buyers’ prices and environmental impacts; it can also assist recyclers decide the recycling scenarios for a batch of end-of-life electronic products based on the EOL value, environmental impacts, and processing time. The decision-making process includes the following steps: selection and definition of attributes, quantifying impacts, scaling of attributes, aggregation, and an interactive optimization process. A reference point method is used in the optimization process. Keywords Multiple criteria analysis . Decision support systems . Electronic product recovery . Reference point method
1 Introduction Currently, increasing attention has been paid to the recycling of electronic equipment and computers, particularly due to the increasing consumption of electronics. K. Jin (*) Department of Mechanical and Industrial Engineering, Texas A&M University Kingsville, Kingsville, TX 78363, USA e-mail:
[email protected] H. C. Zhang Department of Industrial Engineering, Texas Tech University, Lubbock, TX 79404, USA
With growth in digital and wireless technologies, electronics will continue to become obsolete at a dizzying pace in the coming years. Manufacturing industries, especially the electronic industries, are facing increasing pressure that requires environmental responsibility in regard to their products. The end-of-life take-back of electronics products offers both challenges and opportunities for manufacturers and recyclers. This paper focuses on the decision-making problem in end-of-life electronic product management. From a recent report of the EPA, 24 million computers in the U.S. became obsolete in 1999. Only about 14 % (or 3.3 million) of these were recycled or donated. The rest, more than 20 million computers, were dumped, incinerated, shipped as waste exports or put into temporary storage. For the 3 years between 1997 and 1999, it is estimated that some 50 million U.S. computer towers will have been dumped, burned, shipped abroad or stored to await eventual disposal. With the decreasing of the lifespan of computers, computer junking is happening at a faster rate. In 1997 the average lifespan of a computer tower was 4–6 years and 6–7 years for computer monitors. This is expected to fall to 2 years before 2005, and by the year 2005, one computer will become obsolete for every new one put on the market. How will we deal with so many obsolete electronics? Disposal of electronic products in landfills is discouraged due to the limitations of landfill space and the presence of leachable lead, an element that is recognized by both the United State Environmental Protection Agency and the Basel Convention as a hazardous waste [3]. Particularly affected are printed circuit boards with tin/lead solders and cathode ray tubes (CRT’s). In addition, electronic products also contain precious metals and plastics as well as ceramic and glass materials that may be reused if proper material separation and recycling is available. Common electronics recycling problems include: metal recycling, plastics recycling, cathode ray tubes (CRT) glass recycling and printed wired board (PWB) recycling problems. Various recycling scenarios exist for electronics scraps based on their material contents. Although some material recycling processes such as plastics recycling and PWB recycling have been researched individually through
1252
industrial or academic efforts, a generic methodology for finding an appropriate recycling solution for a whole endof-life electronic product is still missing. The objective of this paper is to develop a generic decision-making methodology for material recycling and disposition of electronic products, and to furthermore build a decision support system to help analysts and decision-makers think systematically about the selection of recycling scenarios. The DSS will generate a most satisfying recycling scenario for general end-of-life electronic products based on the combination of environmental impact, intensity of resource recovery, and recycling feasibility. Original equipment manufacturers (OEMs) are forced primarily by the government and customers to take more responsibility for their products and to recycle these EOL products. An intangible benefit to the OEMs, arising from recycling, is the “green” image. As environmental consciousness increases, many customers would like to buy green products that have less environmental effects including solid waste, materials used, energy requirement, and recyclability. Many governments now have official ecology label schemes which inform customers of environmentally friendly products. Usually, an OEM makes contracts with recyclers to take care of their end-of-life products. In order to build their “green” image, OEMs consider not only the profit but also environmental impacts when they select recyclers.
2 Related works and methodologies In 1995, Hoshino et al. provided an optimization analysis for recycling-oriented manufacturing systems. These manufacturing systems include production division and disposal division. In Hoshino’s model, it maximizes two measures of performance—total profit and recycling rate— which are solved by goal programming. A web-based decision support system for waste disposal and recycling was developed by Bhargava and Tettelbach [1]. The goal of this system is to give consumers better access to information and maximize their economic incentives for recycling. The model determines the transfer stations that are to be visited, items that are to be dropped at each station visited, and the travel route between stations, and makes the best trade-off between the monetary pay-off and travel times. DSS users access the system via standard Web browsers, and can query the database or use the decision models. Providers of recycling services can update data about their site through the Internet. The system utilizes a mathematical model schema and integer programming algorithms to determine the optimum recycling plan. In this DSS, two decision criteria, payoff and travel cost, are considered. Another important criterion, environmental impact, is ignored. The environmental problem in the recycling scenario selection is considered by Legarth [9]. A general method for finding new recycling alternatives in the metals producing industry is presented and tested on two printed wire-board scrap cases. It tried to find the recycling process which has the least environmental problems and the largest resource
recovery. This method can be utilized by EPAs wishing to advocate environmentally superior recycling alternatives, or by recycling enterprises wishing to assess the best ecological solutions for new recycling systems. Unfortunately, the OEMs have not been helped by this method.
3 Treatment options and environmental impact assessment Basically, the treatment options of end-of-life products include reuse/resale, repair, refurbishing, remanufacturing, cannibalization, recycling, incineration, and landfill [11]. Among them, reuse and recycling are the methods usually used for electronic product recovery. Resale is defined by Low et al. [10] as an operation whereby the existing product is recovered and sold, with minimum intervention, to another customer requiring a similar product function. This may be in the same geographical location or may be in another distant second market. The products may have to match different infrastructures in the second market, for example, different power supplies [10]. Yan and Gu [13] defined reuse as the further use of a waste product in its original form such as the refilling of a previously discarded container. Reuse can also be the process in which the disassembled components are directly used in another application. Recycling is the reformation or reprocessing of a recovered material. The EPA defines recycling as, “the series of activities, including collection, separation, and processing, by which products or other materials are recovered from or otherwise diverted from the solid waste stream for use in the form of raw materials in the manufacture of new products other than fuel [2].” Many designers, policy makers, and consumers believe recycling is the best solution to a wide range of environmental problems. Recycling does divert discarded material from landfills and reduces the use of natural resources, but the recycling process also causes some other environmental impacts. Efficient recycling not only helps solve problems such as overfull landfills, depletion of scarce resources, and increase in poisonous substances made from incinerators and landfills, but it also has commercial implications. Recycling stations do provide cash incentives for participants who redeem certain valued recyclable products, but given the variety of recycling services, OEMs or consumers cannot easily exploit these incentives in the best way. The decision support system for recycling can solve this problem. The multiple objective decision making algorithms in the decision support system can help users find the most satisfying recycling plan based on the cost, environmental impacts and processing time. Some useful information on each recycling station includes location, distance, payoff rates, environmental impacts and contact information provided for users. Environmental impact assessment is a systematic process that examines the environmental consequences of development actions in advance. One of its important purposes is as an aid to decision-making. Encountering an environmental
1253
decision-making problem, decision makers should consider the environmental impact assessment along with other documentation related to the planned activity. In the past two decades, many environment experts and organizations have paid attention to the environmental impact assessment models and estimators such as the Environmental Priority Strategy (EPS) developed by the Swedish Environmental Research Institute, Eco indicator ’95 [5], and ’99 [6] developed by the Pre Consultants company in The Netherlands, and Environmental Development of Industrial Products (EDIP) developed by the Institute for Product Development in Denmark. When environmental assessment methods are utilized in a recycling scenario, they will provide environmental impacts assessment for the end-ofproduct treatment. Environmentally weighted recycling quotes (EWRQ) presented in 2000 by the design for sustainability research group in Delft University of Technology [8] is used in the decision support system. All calculations of the EWRQ are based on the Eco indicator ’99.
– – –
4 Mathematical models
Objectives:
A complete decision support system should include a userfriendly interface, a powerful database management, and an integrated multiple objective optimization algorithm for helping the decision maker find decisions that are the best for attaining goals specified by the user. From some points of view, we can say the optimization algorithm is the core part of a decision support system. Considering the different requirements of the original equipment manufacturers (OEMs) and the recycling service providers or recyclers, two mathematical models for the optimization problem are discussed in the proposed decision support system. One is the general model, which may be useful for both recyclers and OEMs. y1 ¼
n P
xi1 þ xi2 xi3 xi4 xi5 xi6 xi7 xi8 xi9
i¼1
y2 ¼ z1 þ z2 y3 ¼ EWRQ Objectives: – – – – – – – – – – – – –
Max y1 and y3 Min y2 y1: End of life value (EOL value) y2: Processing time y3: Environmentally weighted recycling quotes (EWRQ) xi1: Revenue from reuse components xi2: Revenue from recovered material xi3: Collection costs xi4: Transportation cost xi5: Inspection costs xi6: Sorting costs xi7: Shredding costs xi8: Recovery process costs
xi9: Waste disposal costs z1: Transportation time z2: Recovery time
We need to maximize the EOL value, the EWRQ, and the negative process time simultaneously but usually these objectives are in conflict with each other. Therefore, they represent a multiple objective decision making problem. Considering that usually the OEMs aren’t concerned with the processing time, and only consider the buyer price and the environmental impacts of their end-of-life products when they choose recyclers to make contracts, another model will be provided for OEMs for making decisions when they make a contract with recyclers. Special model for OEMs: y1 ¼
n P
xi1 þ xi2 xi3 xi4 xi5 xi6 xi7 xi8
i¼1
y2 ¼ EWRQ
– – – – – – – – – – – – –
Max y1 and y2 y1: Collection costs or buyer price y2: Environmentally weighted recycling quotes (EWRQ) xi1: Revenue from reusing components xi2: Revenue from recovered material xi3: Transportation costs xi4: Inspection costs xi5: Sorting costs xi6: Shredding costs xi7: Recovery process costs xi8: End of life value z1: Transportation time z2: Recovery time
When the OEMs have some experiences with the estimation of the EOL value, they can choose the recycler by maximizing the buyer price and maximizing the EWRQ simultaneously. There are basically three types of constraints in the proposed models. There are non-negative constraints, potential requirement constraints and potential policy constraints. Negative constraint is easy. All attributes in the model are greater than or equal to zero. The potential requirement constraints include the lower bounds of the end of life value (EOL Value) and EWRQ, and the upper bounds of the processing time. They are usually specified by the decision maker. This range is represented by R3. The solution range Dε is also bounded by the trade-off coefficient Dε. ( ) 3 X 3 D" ¼ q 2 R : min qi þ " qi 0 1i3
i¼1
q is the objective vector. The feasible solutions of this second model are a set of recycler options. Those options are also subjected to some
1254
policy constraints. Those constraints include the EPA regulations and the user’s preferences.
mented Chebyshev norm case, the weight coefficient αi and the reference level qi have the following relations:
5 Decision support model
qi ¼ qi;up
The methodology used in the DSS is an interactive multiple objective decision-making method integrating Chebyshev norms and the reference point method. In order to find the order-consistent achievement function in the reference point method, let’s start from the Chebyshev norm function. It has been shown that an augmented Chebyshev norm results in proper efficient solutions. The natural solution is: ! k qi;up qi qi;up qi X þ" b qneu ¼ arg min max 1ik qi;up qi;lo i¼1 qi;up qi;lo (5.1) The weighted compromise solution is: ! k qi;up qi qi;up qi X þ" b qα¼ arg min max αi αi 1ik q qi;up qi;lo q i;up i;lo i¼1 (5.2) The reason for using the reference points as the main parameters controlling the selection of Pareto-optimal points in the reference point method is because reference points usually provide much better interpretations of the decision makers’ preference than weighting coefficients. However, we need to find the relationship between weight coefficient and reference point. Fortunately, in the aug qi;up qi
k qi;up qi X þ" max qi;up qi;lo ¼ 1ik qi;up qi;lo i¼1
k X qi qi qi qi þ" 1ik qi;up qi q qi i¼1 i;up
qi;up qi;lo αi ¼ qi;up qi
η¼
,
(5.3)
k X qj;up qj;lo j¼1
qj;up qj
k X qj;up qj;lo j¼1
(5.4)
(5.5)
qj;up qj
If we know the reference point of each decision objective, we can get the corresponding weighting of each objective from (5.4). However, we can’t solve the reference points qi from (5.4) if we know the weighting vector. In order to make sure the qi is between qi,up and qi,lo, ηαi >¼ 1. Therefore, η α1i 8i ¼ 1; . . . k: With any sufficiently large η, for each weighting coefficient αi, we can assign a reference point value to it based on Eq. (5.3). Since qi;up qi qi;up qi qi qi αi η¼ 1 η ¼ qi;up qi;lo qi;up qi qi;up qi Since qi;lo qi qi;up ; i¼1;. . .;k
(5.6)
(5.6)
k X qi qi qi qi " þ 1 þ "k min 1ik qi;up qi q qi i¼1 i;up
Hence, minimizing the weighted distance from the upper bound point in the Chebyshev norm is equivalent to maximizing the following order-consistent achievement function: σðq; qÞ ¼ min
qi;up qi;lo ηαi
!,
(5.7)
Similarly, the neutral solution b qneu can also be obtained from an achievement function. qi qi;mid q þq qi;up qi 1 If we define qi;mid ¼ i;up 2 i;lo ; qi;up qi;lo ¼ 2 qi;up qi;lo Therefore:
(5.8)
k k qi;up qi qi;up qi X X q qi;mid qi qi;mid 1 þ" ¼ min i max " þ ð1 þ "k Þ 1ik qi;up qi;lo 1ik qi;up qi;lo q qi;lo 2 i¼1 qi;up qi;lo i¼1 i;up
(5.9)
1255
The achievement function for the neutral solution is: k X qi qi;mid qi qi;mid þ" 1ik qi;up qi;lo q qi;lo i¼1 i;up
σðq; qmid Þ ¼ min
The optimization process in the proposed DSS can be summarized in the flow chart in Fig. 1.
(5.10)
DSS Determine Ideal and Nadir Values
DM Input
Pairwise Comparison Matrix R
Aspiration Level
Use LSF to Calculate w0
Solve a set of equations for reference points qa0 based on (5.3) and (5.5)
Ignore Input
Use the q0mid = 1/2 (qup + qlo) to find the optimal solution to maximize the achievement function (5.10) Find the optimal solution to maximize the achievement function (5.8)
Calculate k other optimal solutions by maximizing the achievement function (5.8) with perturbed reference points
qha = qh-1e
h = h+1
Provide optimal solutions {qeh} to DM
Satisfied?
Yes
Get decision making solution
Fig. 1 Flow chart for the proposed algorithm in the DSS
No
Store qe
DM input new qha
1256
6 System development and an application example In order to realize environmentally conscious products and efficient, high-quality recycling, a web-based electronics product evaluation and recycling system was developed in the Advanced Manufacturing Lab, Texas Tech University to support user collaboration based on sharing product information over the Internet. This system presented in Fig. 2 consists of six functions: (1) product Disassembly, (2) product recycling, (3) material assessment (4) environmental impact assessment, (5) product evaluation, and (6) product and material information management. It generates disassembly cost and disassembly process planning, efficient recycling strategies, recycling costs, recoverable materials, material compatibility, environmental impact information, and product design evaluation. The decision support-model is an important and necessary module in the system for generating efficient recycling strategies.
The proposed decision model in Fig. 3 includes four phases. In the first phase, the environmental impact of an end-of-life product is evaluated using the EWRQ based on Eco-Indicator ’99. In the second phase, the three attributes— cost, time and recycling quotes—are rescaled into dimension-free units based on the calculation of the ideal and nadir points. In the third phase, three objectives—end of life value, EWRQ, and processing time—are aggregated into an achievement function. Finally, an interactive decision process helps users find the best satisfactory recycling plan. The graphical user interface (GUI) in the ECDS model has five functional blocks: 1. The initial selection block. The user is asked to select the product, input the amount of the products, and select their decision objective.
Top Tier Client's Desktop
Designers
Customers
Applet
Web Server
Five Functions
Middle Tier
Product ProductDisassembly Disassembly Product ProductRecycling Recycling Material Material Assessment Assessment
Management
XML+XSL
Environmental Environmental Impact Impact Assessment Assessment Product Product Evaluation Evaluation
Product and Material Information Management XML Documents
XML Encapsulation API
Bottom Tier Databases,main frames,legacy systems
PDM
JDBC-ODBC BRIDGE
CAD/CAM
Fig. 2 Web-based electronic product evaluations and recycling system
Relational DB
Servlet Containers
1257 Fig. 3 The decision support module
Environmental Metrics, Guidelines and rules
Model Contents
Phase I Quantify Environmental Impact
CAD/CAE Data Product Data
Phase II Database Management
Re-scaling
Chebchevnorm function
Weight Coefficients
Phase III Aggregation
Achievement Function
Interactive Decision Making Process
Reference Point Method
Phase IV
End-of-Life Product Management Scenario
2. The user aspiration expression block. The user can input reference points and pairwise comparison matrices or use the default natural reference points. 3. The candidate scenario presentation block. The user can preview each candidate scenario by choosing a scenario from a list and interact with the decision making process to determine the most satisfactory scenario. 4. The optimization process block. This can generate utopia and nadir points for each objective, derive weighting vector from comparison matrix, assign reference points based on the weighting vector, and generate the most satisfied recycling scenario based on the reference points or pairwise comparison matrix. 5. The report process block. The user can generate, save, open or print the decision report. The structure of the GUI is shown in Fig. 4. The ECDS software has been developed in the Microsoft Visual Basic 6.0 environment. Microsoft Access was chosen as the tool for the database design and implementation. Desktop computers were chosen as the recycling product during the system testing. Taken 100 DELL Optiplex GL 5100 computers as an example, the EOL value and the EWRQ of eight recycling options are shown in Fig. 5.
If we use the maximization of a linear weighted combination of the EWRQ and EOL value, we obtain only extreme cases of 7765 (EOL Value) and 2750 (EOL Value). The more balanced cases of 4955 (EOL value), 3500 (EOL value), and 2280 (EOL value) will never be selected no matter how the weighting coefficients are changed. However, in the ECDS model, even when the user does not input the reference points, the option with the 4955 EOL value is selected in the first iteration, and the option with the 3500 EOL value is selected in the second iteration.
7 Conclusion This decision support system is encouraged by integrating recycling analysis, environmental impact analysis, multiple objective decision making and database management. The recycling of EOL electronic products is a complex issue because of the combination of environmental impact, cost, and other factors. The critical task is to determine the proper recycling strategies. In this research, an innovative multiple decision making approach will help the decision maker find the optimal recycling plan for various electronic scraps. The research result can be extended to the recycling process of all kinds of electromechanical scraps. Potentially, the methodology can be adopted for assessing
1258
Initial Selection 1. Select product 2. Input amount 3. Select decision objective
Optimization Process 1. Generate utopia and nadir points 2. Derive weighting vector 3. Assign reference points 4. Generate most satisfied recycling scenario
User Aspiration Expression 1. Input reference points 2. Input CM 3. Use default natural reference points
Report Process 1. Generate report 2. Save, open or print report
Candidate scenario representation 1. Preview candidate scenario 2. Interactive decision making
Fig. 4 GUI structure
various recycling alternatives in order to provide an environmentally friendly recycling solution vs. minimum recycling costs. The complete environmental analysis of the EOL products and decision analysis method can also
help the product designers evaluate their products from the environmental point-of-view while in the process of product design, and, furthermore, this will support the design for recycling.
Recycling options for 100 DELL Optiplex GL 5100 Computers
EOL Value
Fig. 5 Recycling options for 100 DELL Optiplex GL 5100 computers
9000 8000 7000 6000 5000 4000 3000 2000 1000 0 0.00%
7765 5388
4955 3500 1650
5.00%
2280 1850
2750
10.00% 15.00% 20.00% 25.00% 30.00% 35.00% EWRQ
1259
References 1. Bhargava HK, Tettelbach C (1997) A web-based decision support system for waste disposal and recycling, Working paper, Naval Postgraduate School, Monterey, CA 2. EPA (1991) Guidance for the use of the terms “recycled” and “recyclable” and the recycling emblem in environmental marketing claims. Compact Disc Fed Regist 56(191):49992–50000 3. EPA (1996) The U.S. EPA’s 25th anniversary report: 1970– 1995. EPA, Washington, DC 4. EPA (1997) 1995 RCRA biennial report, EPA, Washington, DC 5. Goedkoop M (1998) The Eco-Indicator 95 Final Report, Pre Consultants, Amersfoort, The Netherlands 6. Goedkoop M, Spriensma R (2000) The eco-indicator 99, a damage oriented method for a life-cycle impact-assessment methodology report, 2nd edn., Pre Consultants, Amersfoort, The Netherlands 7. Hoshino T, Yura K, Hitomi K (1995) Optimization analysis for recycle-oriented manufacturing systems. Int J Prod Res 33 (8):2069–2078
8. Huisman J, Boks C, Stevels A (2000) Environmentally weighted recycling quotes-better justifiable and environmentally more correct. Proceedings of the International Symposium on Electronics and the Environment, May 2000, San Francisco, California, pp 105–111 9. Legarth JB (1997) Environmental decision making for recycling options. Resour Conserv Recycl 19:109–135 10. Low MK, Williams D, Dixon C (1996) Choice of end-of-life product management strategy: a case study in alternative telephone concepts. 1996 IEEE Int Symp Electron Environ, May 1996, Dallas, TX, pp 112–117 11. Thierry M, Salomon M, van Nunen J, Van Wassenhove L (1995) Strategic issues in product recovery management. Calif Manage Rev 37(2):PP114–PP135 12. Wierzbicki AP, Makowski M, Wessels J (2000) Model-based decision support methodology with environmental applications, Kluwer, Dordrecht, NL 13. Yan X, Gu P (1995) Assembly sequence planning for life-cycle cost estimation, Manufacturing Science and Engineering, ASME, New York, pp 935–957