Multi-objective reverse logistics model for integrated computer waste management

Download Multi-objective reverse logistics model for integrated computer waste management

Post on 02-Apr-2017




5 download

Embed Size (px)


  • Management & Research online version of this article can be found at:

    DOI: 10.1177/0734242X06067252

    2006 24: 514Waste Manag ResPoonam Khanijo Ahluwalia and Arvind K. Nema

    Multi-objective reverse logistics model for integrated computer waste management

    Published by:

    On behalf of:

    International Solid Waste Association

    can be found at:Waste Management & ResearchAdditional services and information for Alerts:

    What is This?

    - Dec 12, 2006Version of Record >>

    at UNIV OF UTAH SALT LAKE CITY on November 27, 2014wmr.sagepub.comDownloaded from at UNIV OF UTAH SALT LAKE CITY on November 27, 2014wmr.sagepub.comDownloaded from

  • 514 Waste Management & Research

    Waste Manage Res 2006: 24: 514527Printed in UK all right reserved

    Copyright ISWA 2006Waste Management & Research

    ISSN 0734242X

    Multi-objective reverse logistics model for integrated computer waste management

    This study aimed to address the issues involved in the plan-ning and design of a computer waste management system inan integrated manner. A decision-support tool is presentedfor selecting an optimum configuration of computer waste man-agement facilities (segregation, storage, treatment/processing,reuse/recycle and disposal) and allocation of waste to thesefacilities. The model is based on an integer linear program-ming method with the objectives of minimizing environmen-tal risk as well as cost. The issue of uncertainty in the estimatedwaste quantities from multiple sources is addressed using theMonte Carlo simulation technique. An illustrated example ofcomputer waste management in Delhi, India is presented todemonstrate the usefulness of the proposed model and tostudy tradeoffs between cost and risk. The results of the exam-ple problem show that it is possible to reduce the environ-mental risk significantly by a marginal increase in the availa-ble cost. The proposed model can serve as a powerful tool toaddress the environmental problems associated with expo-nentially growing quantities of computer waste which arepresently being managed using rudimentary methods of reuse,recovery and disposal by various small-scale vendors.

    Poonam Khanijo AhluwaliaArvind K. NemaDepartment of Civil Engineering, I.I.T. Delhi, India

    Keywords: Computer waste, integrated waste management, multi-objective optimization, reverse logistics model, Monte Carlo simulation, wmr 9041

    Corresponding author: Arvind K. Nema, Department of CivilEngineering, I.I.T. Delhi, New Delhi-110016, India.Fax: ++91 11 26581117 e-mail:

    DOI: 10.1177/0734242X06067252

    Received 5 September 2005; accepted in revised form 25 April2006

    Figures 14 appear in color online:


    Solid waste management, which is already a mammoth taskin India, has become more complicated by the arrival of com-puter waste, particularly personal computers, printers andother computer peripherals. It has been estimated that thetotal number of obsolete personal computers emanating frombusiness and individual households in India would be around1.38 million in 2003 (Toxics Link 2003). A recent publica-tion estimated obsolete personal computers to be around2.25 million units in India in 2005, and projected it to toucha figure of 8 million obsolete units by the year 2010, at anaverage annual growth rate of approximately 51% (Boralkar2005). Considering an average weight of 27.18 kg (ToxicsLink 2003) for a desktop/personal computer then approxi-

    mately 61 155 tonnes of obsolete computer waste would havebeen generated in India in 2005 and, at the projected growthrate, this would increase to about 217 440 tonnes by the year2010. An effective waste management system should includewaste collection and transportation, resource recovery throughsorting and recycling, resource recovery through waste process-ing, waste transformation and disposal (CPHEEO 2000). Thesemanagement steps are aptly applicable even for computer-waste. A comprehensive model should not only incorporatethe above-mentioned management steps but should also beable to suggest the location of such facilities, transportationroutes and allocation of different wastes to the facilities. Fur-thermore, for hazardous waste streams such as that of compu-

    at UNIV OF UTAH SALT LAKE CITY on November 27, 2014wmr.sagepub.comDownloaded from

  • Multi-objective reverse logistics model for integrated computer waste management

    Waste Management & Research 515

    ter waste, in which recovery and reuse are important concernsand the need exists to minimize both environmental andhealth risks and maximize returns, reverse logistics can be usedas a powerful tool to integrate all of the above aspects into acomplete systems framework.

    The objectives of this paper are to: (1) review existingoptimization models/techniques for solid hazardous waste man-agement; (2) to present a decision-support tool using reverselogistics for computer waste management, which would helpselect optimum configuration of transportation routes andfacilities considering environmental risk as well as cost; and(3) to address uncertainty in the estimated waste quantitieswith respect to different time steps.

    Review of existing models for hazardous waste management

    In the past many researchers and environmental engineershave attempted to address the problem of solid waste manage-ment with the aid of various mathematical models. Althoughthe focus of this paper is the management of computer wasteusing the systems approach, very scanty literature is availableon the relevant subject. Hence, we cover literature pertainingto hazardous waste management and computer waste.

    Peirce & Davidson (1982) applied a linear programmingtechnique to identify a cost-effective configuration of trans-portation routes, transfer stations, processing facilities andsecure long-term storage impoundments. Jennings & Sholar(1984) formulated the regional hazardous waste managementsystem as a transportation routing problem with sources gen-erating multiple types of wastes. Zografos & Davis (1989)suggested a multi-objective formulation of hazardous wasterouting problem using a goal programming approach toaddress population at risk; risk imposed on special populationcategories, travel time and property damages. Zografos &Samara (1990) proposed a combined location-routing modelexamining trade-offs between hazardous waste transportationand disposal risks, routing risk and travel time. The modeldetermined the location of hazardous waste disposal facilitiesand the routes from given hazardous waste generation sites tothe selected disposal facilities.

    Lund (1990) proposed a linear programming method thatboth evaluates and schedules adoption of each of several pos-sible recycling efforts, minimizing total present value costand considering the effect of recycling on landfill exhaustionand future costs. The method also suggested a least-cost life-time for the landfill, considering the recycling costs of defer-ring landfill closure and the benefits of deferring landfill clo-sure and future replacement costs.

    List et al. (1991) surveyed methodological research onhazardous materials transportation in the areas of risk analy-

    sis, routing/scheduling and facility location. The review tracedthe evolution of models from single-criterion optimizationsto multi-objective analyses. ReVelle et al. (1991) suggesteda model based on the method of shortest path, a zero-onemathematical programme for sighting and the weightingmethod of multi-objective programming for simultaneous sit-ing and routing the disposal of hazardous waste. Stowers &Palekar (1993) proposed a model that simultaneously consid-ered the risk posed by location and transportation risks whilesearching for an optimal location of a single obnoxious facil-ity on a network.

    Jacobs & Warmerdam (1994) presented a linear program-ming model to aid decision-makers in the simultaneous rout-ing and siting of hazardous waste transport, storage and dis-posal operations. Boffey and Karkazis (1995) addressed theproblem of safe transport of hazardous waste with the aid of alinear and a non-linear model. They further derived a condi-tion, which, if satisfied, ensures that the linear model andthe non-linear model generate the same optimal solutionpath, and, if not satisfied, provides a strategy for obtainingthe optimal solution to the non-linear problem.

    Mirchandani et al. (1995) described a model based onheuristics for optimally locating a number of inspection sta-tions along a road network plying trucks carrying hazardouswaste, with the objective of intercepting the maximum numberof trucks to prevent hazardous material (HAZMAT) viola-tions.

    Ferrer (1997) addressed the complexity of personal com-puter (PC) manufacturing and the difficulties in developingan adequate recovery process. He also proposed and evalu-ated a recovery process. Fleischmann et al. (1997) did a sys-tematic overview of the issues arising in context of reverselogistics. They discussed the implications of the emergingreuse efforts and reviewed the mathematical models pro-posed in the literature. Giannikos (1998) presented a multi-objective model for locating disposal or treatment facilitiesand transporting hazardous waste along the links of a trans-portation network. Four objectives were considered: (1) min-imization of total operating cost; (2) minimization of totalperceived risk; (3) equitable distribution of risk among popu-lation centres; and (4) equitable distribution of the disutilitycaused by the operation of the treatment facilities. A goal-programming model was proposed to solve the problem.Nema & Gupta (1999) proposed a model based on a multi-objective integer programming approach to suggest the opti-mal configuration of facilities for transportation, treatmentand disposal with minimum cost and minimum risk to theenvironment. Hu et al. (2002) presented a cost-minimizationmodel for multi-time-step, multiple-type hazardous wastereverse logistics system. The model addressed the classicalhazardous waste treatment problem with a systematic man-

    at UNIV OF UTAH SALT LAKE CITY on November 27, 2014wmr.sagepub.comDownloaded from

  • P.K. Ahluwalia and A.K. Nema

    516 Waste Management & Research

    agement strategy rather than with waste treatment technolo-gies as conventionally employed.

    Shih & Lin (2003) presented a multiple criteria optimiza-tion approach that considered minimization of the cost, riskand workload for collection system planning for infectiousmedical waste. A compromise programming method was usedto integrate the three objectives, and an example of a collec-tion of infectious waste in Taiwan City was presented. Thelocation of medical institutions, an actual road map and pop-ulation density were provided, using a geographic informa-tion system.

    Nema & Gupta (2003) improved upon their suggestedmodel based on a utility function approach by basing the modelon integer goal programming technique. The model was ableto address practical issues such as multiple objectives, com-patibility between waste types, compatibility between wasteand waste technologies and the waste residue generation asso-ciated with treatment technologies.

    White et al. (2003), with the help of a case study, describedthe recovery of computers as a step-by-step process and alsoframed an environmental research agenda for recovery man-agement.

    As can be seen, none of the above mathematical formulationsaddress all components of a complete solid waste managementsystem, and most of them do not address the reverse flow ofwaste, which is necessary for addressing special waste streamssuch as computer waste. As is demonstrated in the present study,recovered cost is substantial in developing countries such asIndia (Toxics Link 2003). In addition, the generation of wasteat any source node varies with time and so it is advisable to havea multi-time step model that takes into account such variations.Moreover, the above models do not also take into account theuncertainty associated with the data related to waste genera-tion. Furthermore, risk, as addressed in most of the models, isonly considered for the transportation of hazardous waste. Itneeds to be accounted for at every stage, namely storage, seg-regation, treatment and disposal facilities. Integration of allthe above components would make a complete system ofcomputer waste management that can serve as a tool for theconcerned managers.

    Proposed model formulation

    Any regional network of waste management consists of sourcenodes, a set of transportation routes, and facilities, such assegregation facilities, storage facilities, treatment/processingfacilities, reuse/recycle facilities and disposal options. Moreo-ver, the generation of waste at any source node varies withtime and therefore a multi-time step model that takes intoaccount time variations in the generation of waste is pro-posed. Each activity, whether it is transportation of waste,

    processing, storage or disposal, has a certain cost and risk fac-tor associated with it. Thus, the main objective of any solidwaste management programme is to select each activity suchthat the cost and risk factors are minimized.

    The following two objectives are addressed in the presentmathematical formulation.

    1. Minimization of total cost, which includes transportationcost, segregation cost, storage cost, treatment/processing cost,disposal cost and cost recovered from the reuse and recy-cle of waste. For each facility operating costs, as well ascapital costs are included.

    2. Minimization of total risk, which includes transportationrisk, as well as site risk.

    Each of these objectives can be minimized individually to obtaincost and risk for the minimum cost and minimum risk scenar-ios. However, because the two objectives have different units,combining both of them poses a problem. This has beenaddressed by Nema & Gupta (1999) by proposing a compositecostrisk utility function.

    Objective = Minimize (U)U = Weighting to risk (risk/minimum achievable risk) +

    Weighting to cost (cost/minimum achievable cost).

    The decision-maker can assign the different weightings to costand risk and the model can...


View more >