Abstract—We present an approach for the design, evaluation

and implementation of remanufacturing processes in a given facility. Based on the description of the market situation and involved actors, a planning method is derived. Data acquisition procedures for product, process and facility are described. Two mixed integer programs are developed; the first one optimizes the remanufacturing process and evaluates its economic viability and the second sequences the remanufacturing tasks. Additionally, the paper describes the technical implementation into a software and the exemplary application on the product category Flat Screen Monitors.

Index Terms— Remanufacturing, process design, process planning, mixed-integer programming

I. INTRODUCTION E-USE and remanufacturing of Waste Electrical and Electronic Equipment (WEEE) are a matter of current

concern, driven by economic, ecologic, social and legislative factors. The potential that lies in the reuse and remanufacturing of IT-Equipment is yet to be fully exploited. Only a few specific products are considered – e.g. mobile telephones [1], [2], [3], [4] - and not all treatment opportunities available are applied. Moreover the financial uncertainties concerning product quantities, types and conditions that are long associated with remanufacturing processes limit the entrance of new actors into the sector. Although planning decisions highly influence the efficiency of a production system carrying out remanufacturing operations, decision support and planning tools, which are standard in assembly industries, are seldom available and applied. The need for such tools is evident. Based on a market analysis and the identification of actors that are aiming to extend their operations by establishing remanufacturing processes as value adding processes, an integrated planning system is presented in this paper.

This project is funded by the German Research Foundation (Deutsche

Forschungsgemeinschaft, DFG) within the industry transfer project 64, E8 “Planning of a Remanufacturing System for LCD-Flat Screen Monitors” in cooperation with Flection Int’l, Location Germany. Stylianos Chiotellis and Sebastian Kernbaum are Research Assistants and Günther Seliger is Professor at the Technical University Berlin, Institute for Machine Tools and Factory Management, Department of Assembly Technology and Factory Management, PTZ 2, Pascalstr. 8-9, DE 10587 Berlin.

II. ACTORS INVOLVED AND TASK DEFINITION Actors involved in closed-loop-economy and their relation

are widely discussed in the literature [5], [6]. The following are identified as capable to extend their operations with remanufacturing oriented value adding processes: OEM’s, suppliers, maintenance shops, existing remanufacturing companies, and partly recycling companies.

These actors have in common the access to functional or non-functional products and components that are valuable enough to justify their upgrade, through some remanufacturing treatment, to a marketable state or condition.

Different are the actors’ knowledge on the product itself and the product’s condition as well as their operational capability to bring it to a valuable condition. They also differ in available equipment, processes and capacities, i.e. in remanufacturing capabilities. This raises the following questions for each specific actor:

• To what extend do I incorporate remanufacturing activities in my existing facility/facilities?

• Which specific remanufacturing processes am I able to carry out?

• How much investment is needed in order to offer high quality remanufacturing?

• What are my specific costs and revenues once the remanufacturing processes are operational?

To answer these questions in a suitable manner, a four step planning approach has been developed, implemented in a software tool and exemplarily applied on the example of flat screen monitors remanufacturing.

III. PLANNING METHOD The proposed approach involves four steps: the data

analysis, the process design phase, the remanufacturing process optimization and the scheduling/sequencing of remanufacturing tasks phase. A. Data Acquisition

The product and facility relevant data that are required for the proposed planning approach are determined and acquired. Data regarding the product need to include the product structure, i.e. components and component groups, joining elements and techniques, disassembly sequences, material ratios etc. and end-of-life data (disassembly times and costs and recycling quota). For ease of system integration, product models are developed using the commercially available software system ProdTect [7]. ProdTect is a software tool which supports the development of ecologically sound

Remanufacturing process planning for IT equipment

Stylianos Chiotellis, Sebastian Kernbaum, Günther Seliger

R

products by providing information related to a product’s treatment and recycling at an early product development stage [8]. In the ProdTect product model input module, the product structural information is composed of: • Parts information, such as material composition,

disassembly movement, dimension, shape, accessibility. • Connection information composed of the different joining

elements in the product. • Priorities information, which gives an order of the parts

inside the product. A part is prior to another if it needs to be dismantled to get access to the other.

ProdTect calculates technical, economic and ecological parameters. The resulting data, such as disassembly times and sequence, can then be utilized for the planning of the end-of-life processes for a product [9]. An overview of the software tool ProdTect is given in Fig. 1.

Fig. 1. ProdTect Overview Product accompanying information were analyzed and

prepared for being used for the proposed planning approach. The next step involves acquiring data regarding the resource

and capacity planning of the given facility. Such data include the available installed capacity, the actual job schedule and the inventory levels and can be acquired from the ERP system of the facility. Facility accompanying data were analyzed and prepared for application in the proposed planning approach in accordance to the VDI 3633 guideline ([10], Fig. 2). B. Process Design

The second planning step is the development of the process model. The process model design determines all possible sequences and types of remanufacturing steps that are required in order to process a batch of incoming products, e.g. testing, cleaning, disassembly, reassembly from a specific product type, e.g. flat screen monitors. The sequence is derived by conducting trial runs on a number of sample products. A graphical user interface (GUI) has been developed in order to assist the planner to visualize the developed remanufacturing process in the form of a network. This network acts as the

interface of the database that contains the product, process and facility information.

Fig. 2. ERP System Data Structure

Fig. 3 gives an overview of the structure in the example of a typical remanufacturing process. Products entering are first examined according to product relevant testing criteria. Based on these tests, a list of failures can be documented for each specific product. These lists of failures need to be treated to bring the product back to a valuable condition – they therefore create specific “remanufacturing paths” for each single product in the overall designed process network. Based on the test results, treatment decisions are made: if the amount of failures exceeds the amount of failures that are treatable in the designed process, the product is send to material recovery processes. In case the product is in a condition to be sold to customers, only cleaning and packaging operations need to be carried out. If failures are worth to be treated, the product is send to a first disassembly operation. Information available from ProdTect (e.g. disassembly times and costs) is connected to this operation and further information needed for the planning approach can be added by means of the GUI, e.g. station assignments.

The disassembled component enters the next operation, where treatment operations for possible failures are executed. Additionally, testing and cleaning operations can be added to treat the component in order to raise its value.

The rest of the product enters the next disassembly step; the disassembled component is treated to correct its failures and increase its value in the same manner as the previous component.

The complete product follows this structure until all failures are treated and a functional product can be reassembled: all treated components and the remaining product parts are sent to the reassembly station. After the reassembly step, a final functional test needs to be carried out before remanufactured products enter the resale step.

Products are sold in different quality classes depending on the condition during the first general test or depending on the amount of failures that could be treated (failure handling report,) during the remanufacturing processes. Quality classes for flat screen monitor were already presented in [11].

Facility Structure

•Manufacturing and Transportation Equipment, Material Flow, …

Equipment

•Capacity, Technical Restrictions, Availability, Maintenance Interval, …

Work Schedule

•Batch Size, Work Precedence, Manufacturing Equipment, …

Organisation

•Shift control, Sequencing, …

System Loads

•Production Jobs, …

Costs

•Energy, Maintenance, …

Fig. 3. Procedure for Process Design (simplified)

C. Remanufacturing process optimization

Objective of the third planning step is to determine whether a batch of incoming products is to be commissioned for remanufacturing, i.e. whether the remanufacturing of a certain number of products of a certain type is of economical interest for a given facility and to what extent the remanufacturing processes will be carried out. This implies that there are a variety of options to be examined; fully refurbish, partly refurbish, recycle etc. A mixed integer optimization program (MIP) has been developed to assess the economic feasibility of processing the batch of the incoming products under the given actual capacity situation of the facility.

The objective of the optimization is to maximize the profit of remanufacturing the incoming batch under constraints of capacity availability (workers and machines), operational costs, product flow, product condition, inventory levels, transportation costs and additional capacity investments that may be required. All relevant data for the optimization model are provided from the product model, the process model, the ERP system of the facility and the input of the planner regarding cost related parameters. Especially for the condition of the incoming products, the planner can select between using available historical data or an estimate based on his own experience. A strategic decision of the available budget to be allocated for additional capacity investments needs to be taken beforehand. Based on the solution of the model, the planner can decide whether the available batch of products is to be acquired or not and determine the optimal number of steps of the remanufacturing process.

The remanufacturing process has been modelled in remanufacturing steps which are arranged in seven categories Type(1,…,7) respectively: test steps, reassembling, disassembling, steps that require additional parts to be purchased and access to the warehouse, sale steps after which products can be sold to generate income, the entrance steps (product entrance) and finally general remanufacturing steps. This categorization is executed by the planner (the planner defines which steps belong to each category) at the beginning of the optimization and is required due to special product flow constraints that accompany the different types of steps. In addition, it should be mentioned that our model does not differentiate between a product to be remanufactured, a part or a component.

The following set notation is used in the remainder of the paper: I 1,…,i , : remanufacturing " " 1 This set includes all remanufacturing steps. The various step types (and thus corresponding subsets) are defined with the use of the binary parameter:

1, … ,7 1 1, … ,70

, , 2

W 1,…,w , w N : workstations 3 The decision variables that are used are shortly explained:

, , , respectively denote the number of products that are between two remanufacturing steps , the number of products at step before performing the

remanufacturing task and the number of products at step after the remanufacturing task is performed. stand for

Product withn Components

General Tests: 

visual, mechanical, electrical, logical, …

List of Failures:F1, … Fm

Monitor with zero faults, no treatment required

Packing accessories:‐Cables‐Handbook‐Drivers CD‐Warranty Sticker  

Disassembly

Components 1…n

Q4: Material Recovery 

Monitor with massive faults, treatment impossible 

Sale Monitor in Quality Classes acc. to condition:Q1: best qualityQ2: minor qualityQ3: lowest quality

x%

y %

100‐(x+y) %

Failure Treatment

Components 1…n

Optional Cleaning

Components 1…n

Optional Testing

Components 1…n

Component

Rest of Product

All Failurestreated?

as long as all Failures m could be treated

Evaluation of Faults and Treatment Decision

no

Reassembly

yes

Packaging for Sale

Failure Handling Reports

Functional Test

Components from:‐Storage‐Procurement

Q4: Material Recovery 

the number of parts required from storage for step where : 4 1 , for the number of parts

returned to the storage after step and variables denote the final storage level after step , where : 41 . denote the number of additional parts/components that need to be purchased for step : 4 1 . Variables illustrate the number of additional workers to be assigned to workstation while the number of new workstations to be installed. Finally, denote the number of batches of parts/components to be purchased and

are used for modelling convenience and stand for the number of main components to be reassembled in a reassembly step. All decision variables are restricted to take values from the set of non-negative integers.

The following parameters have been used in the modelling: ,

1 0

,  , , (4)

1 ∑ ,⁄ , : 2 1 ,∑ , , : 3 1 ,1,

(5) Parameter (4) indicates the allowed step sequences, i.e. whether the transition from remanufacturing step to remanufacturing step j is meaningful, while parameter (5) is used to model the different flow conservation behaviours of the various remanufacturing steps types. Some steps result in a change in the number of product, i.e. a disassembly step would result in an increase in the number of products while a disassembly would result in its decrease thus different step types need to be addressed explicitly in the model. The objective function (6) maximizes profit, i.e. maximizes the sum of income from the process minus the sum of relevant costs. is the sale price of each product after completing step where , the cost of purchasing a new part, the

labour cost per worker per hour, the transportation cost per batch of purchased parts for each step that requires parts to be purchased, number of batches of purchased parts for each step and the cost of a new workstation. The MIP model follows (6)-(23):

max1

, 21

, 1 21

1

1 1

1

, , ,

(6)

Subject to:

, , , : 7 0 ,

,

(7) , , (8)

, ,

, : 4 1 5 0

, : 5 0,

, (9)

, ,

: 4 1 , , (10) , , (11)

, , , , , , (12) , , ,

, : 1 1 , (13) , , ,

, : 2 1 , (14) , ,

, : 2 1 : 1 , (15) , , ⁄ ,

, : 3 1 , (16)

, 1 2 ,

1 2,

, , (17) ,

, (18) ,

: 4 1 , (19) , : 4 1 , (20)

, : 4 1 , (21)

,

(22) , , , , , , ,

, , , (23)

The first group of constraints (7)-(16) addresses the flow of material through the process. Constraints (7) are flow conservation constraints that hold for all type of steps and ensure that the number of products in step before performing the task is equal to the sum of all products transferred there from all possible previous steps k. Especially for the entrance step (Type6) of the process it is assumed that the number of products is equal to the initial warehouse level. Constraints (8) ensure that the flow of products is conserved according to the

type of remanufacturing step. Constraints (9) guarantee that the number of parts needed for step is provided by predecessing steps, parts in storage or purchased parts. Constraints (10) are valid for steps that are able to purchase parts and have access to the warehouse (Type4) and determine the amount of products/parts to be returned to storage after the execution of the step. In addition, (11) limit the number of parts that can be purchased per Type4 steps to ; since we are dealing with obsolete IT equipment as remanufacturing products, it is not rare that the supply of parts and components is limited. Equations (12) determine the allowed transitions from step to step j. Constraints (13) address test stations (Type1) and define their product output, i.e. number of products that have past or failed the test, according to the parameter , which denotes the percentage of products to be transferred form step to step j where step is a testing step. In the case of reassembly steps (Type2), the number of products that are identified as the main component ( ) needs to be matched by all other required components that are coming into the step. This is ensured by constraint (15) which identifies the main part of the reassembly operation and constraint (14). Parameter 1 is used to identify the product or part that will be considered as the main part in the reassembly step (Type2). In the case of disassembly steps (Type3), the number of products on the exit of the step increases since we obtain more components. Equations (16) apply the corresponding flow constraints.

The next group of constraints addresses capacity restrictions. More than one step can be executed in one workstation; the binary parameter , (see constraints 17) determines whether step can be executed in workstation . Constraints (17) ensure that the sum of the time required in order to carry out all steps assigned to a workstation , does not exceed the capacity of the station. It also tackles capacity extensions by hiring additional workers or by investing in new machines for the workstations. Here denotes the number of existing workers assigned to the station and

each workers capacity in pieces per hour. Constraints (18) impose a bound ( ) on the total number of workers that can be assigned to a workstation. This expresses possible limitations of space or lack of necessary tools per workstation.

Constraints related to the warehouse include equations (19) where the storage level after completing step is forced to be consistent with all product flow regarding the warehouse, i.e. flow that originates or is directed to steps that have access to the warehouse (Type4). Constraints (20) ensure that the number of parts drawn from the warehouse is at most equal to the actual warehouse level.

The set of equations (21) imposes that the number of parts/components to be purchased is limited to the transportation capacity. Here is the maximum number of parts or components that can be transported per order. Note that each Type4 step is related to specific parts/components which have various properties and dimensions thus different maximum quantities that can be transported per batch. Finally, constraints (22) restrict the available investment budget for the installation of additional workstations to a maximum .

IV. REMANUFACTURING SEQUENCING The last step in the planning process is the generation of the

remanufacturing sequences for the given batch of products and facility; in the previous step tasks have been assigned to machine groups regarding technological and economical restrictions and now these tasks need to be sequenced.

A mixed integer program that minimizes the make-span of the remanufacturing operations has been developed. An optimal schedule is derived on the basis of the remanufacturing capacity and tasks already deployed in the shop-floor. A convenient factory layout has been assumed on the basis of the previous planning step. Equipment (machine or workplaces) are arranged into workstations of identical equipment; testing workplaces are arranged within the testing workstation, cleaning workplaces form the cleaning workstation and so on.

The following set notation is used: 1, … , , , tasks to be scheduled (24)

1, … , , , workstations, (25) 1, … , , , workplaces in each workstation (26)

1, … , , , sequence number of a job, i.e. the place of the job in the queue of a workstation (27) Also let , , be the processing time of job in workplace

of workstation . The decision variables are the binary:

, ,

1, ,

0,

, , , (28) and,

, , , , , , (29) the starting time at which a task in the sequence position is processed at workplace in workstation .

The objective function (30) minimizes the total make span, i.e. the time required to complete all tasks. Here, , , is the start time of the task in the last sequence position in the last workplace of the last workstation of the operations and the term ∑ , , , , expresses the duration of the operations from there on. The MIP program follows (30)-(34):

min , , , , , ,

(30) Subject to:

, , 1, ,

(31)

, , 1, ,

(32)

, , , , , , , , ,

, , (33)

, , , , , , , , ,

, , (34)

Constraints (31) ensure that each task is assigned exactly one sequence position within each workstation, while (32) that each task schedule position is assigned to exactly one job in a workstation. Equations (33) and (34) are sequencing constraints.

The model communicates with the previous planning step in the sense that the tasks to be scheduled and the number of existing workstations and workplaces (workers) are derived from the results of the process optimization.

V. SOFTWARE IMPLEMENTATION

The presented approach was implemented in a software tool whose structure is depicted in Fig. 4. The software is hosted on a Microsoft Windows Server 2003 and utilizes Microsoft SQL Server 2005 for data storage. The graphical user interface was generated in a .NET framework and includes sections for data management, process design and results visualization.

Fig. 4: Software structure.

Active Server Pages allow online functionality and platform independent usage of the software; the software can be accessed from any location with access to the Web through a variety of browsers. The mathematical models are programmed in the external software AIMMS 3.7 using its native programming language. CPLEX 10.1 from ILOG is used to solve the MIP models. The .NET framework enables for seamless integration of AIMMS into the user interface. Optimal results for the process optimization part have been generated within seconds for the instances tested so far. The scheduling part requires significantly more computational effort and thus a limitation of 30 minutes running time has been imposed to the software for the cases when optimal results are hard to reach.

VI. EXEMPLARY APPLICATION The presented approach has been tested with a variety of small instances and is currently being tested with real size problem instances generated in cooperation with an industry partner. The concept has being applied on the example of flat screen monitors. This product type has been favored for various reasons; the massive production volumes, increasing demand and continuous technological innovations that introduce new generations of products create a steady stream of obsolete products that require remanufacturing. The software can be easily adapted to cover different product types. In addition, a scheduling heuristic is to be implemented as an alternative to the scheduling algorithm in order to address bigger problem instances without running time compromises.

REFERENCES [1] Kharif, O., “Where Recycled Cell Phones Ring True”, BusinessWeek

Online, 07-26-2002. [2] Quinn, J., et al., “Cellular 2002, A Study of the Worldwide Cellular

Telephone Market”, Phoenix, Arizona, USA, 2002. [3] Seliger, S.; Franke, C.; Ciupek, M.; Basdere, B.: Process and Facility

Planning for Mobile Phone Remanufacturing, Annals of the CIRP Vol. 53, Issue 1, 2004.

[4] Franke, C.; Basdere, B.; Ciupek, M.; Seliger, G.:Remanufacturing of mobile phones – capacity, program and facility adaptation planning. omega – The International Journal of Management Science, Vol. 33, 2005

[5] Franke, C.: Beitrag zur Steigerung der Nutzenprodukti-vität von Ressourcen durch Anpassungs-programmplanung am Beispiel von Flachbildschirmen, Dissertation Technische Universität Berlin, 2006

[6] Takata, S.; Kimura, F.; van Houten, F.J.A.M.; Westkamper,E.; Shpitalni, M.; Ceglarek D.; Lee, J.: Maintenance: Changing Role in Life Cycle Management, CIRP Annals - Manufacturing Technology, Volume 53, Issue 2, 2004, Pages 643-655.

[7] Herrmann C., Frad A., Revnic I., Luger T.: Product Architect, a new approach for transparency and controlling of the End-of-Life performance. In: Proceedings of the EcoDesign2005, Tokyo, 2005

[8] www.prodtect.com, last access: 13.11.2007 [9] Hesselbach J., Herrmann C., v. Westernhagen K.: Datenbanken als

Informationsträger von der Produktentstehung bis zur Demontageplanung. In: Proceedings of „Umweltinformatik zwischen Theorie und Industrieanwendung“, 13th International Symposium, Magdeburg, 1999

[10] VDI Guideline 3633: Simulation of systems in materials handling, logistics and production – Fundamentals. Beuth Publishing House, 2000.

[11] Franke, C.; Kernbaum, S.; Seliger, G.: Remanufacturing of Flat Screen Monitors – Economical and Technological Challenges until 2010. In: Proceedings of the 38th CIRP International Seminar on Manufacturing Systems (ISMS), Florianopolis, Brazil, May 2005

InternetInformation

Services(IIS)

Query Results

ServerMS Windows Server 2003

Response

Initialize

(MS SQL Server)

GUI für data management,process design, and software control

ProdTect ERP System(e.g. SAP R/3)

ReadingProdtect XML

Reading ERP-System

data

AnwendungsserverASP.NET 2.0: Active Server Pages

Database

Datasets(.xsd XML Schema File)

AIMMS Project

Request

Webbrowser

Clientlocation- and platform

independentPlaner

Results Retrieval

AIMMS COM Object