Recovery-channel selection in a hybrid manufacturing-remanufacturing production model with RFID and product quality

International Journal of Production Economics 219 (2020) 360–374

Contents lists available at ScienceDirect

International Journal of Production Economics journal homepage: www.elsevier.com/locate/ijpe

Recovery-channel selection in a hybrid manufacturing-remanufacturing production model with RFID and product quality

T

Mehran Ullaha, Biswajit Sarkarb,* a b

Department of Industrial & Management Engineering, Hanyang University, Ansan Gyeonggi-do, 15588, South Korea Department of Industrial Engineering, Yonsei University, 50 Yonsei-ro, Sinchon-dong, Seodaemun-gu, Seoul, 03722, South Korea

A R T I C LE I N FO

A B S T R A C T

Keywords: Product recovery channels RFID (EOL/EOU) management Return rate Product quality

Due to the accelerated expansion of technology and improved living standards, the use of electronic products is increasing exponentially. Unfortunately, the improper disposal of used electronic products, such as cell phones, has resulted in rapid degradation of the environment and is, therefore, posing a great threat to human health. Meanwhile, proper recycling is also likely to generate high profits; thus, recycling of electronic products is a necessity of the time. However, the profitability of recycling and remanufacturing depends upon the return rate, which is very low in the cell phones industry. In this paper, we first identify the root causes of low return rate and, then, develop a novel Radio Frequency Identification (RFID) based return channel to increase the recycling rate. A dual-recovery-channel hybrid manufacturing-remanufacturing production model is proposed, which procure used products of different quality from both the traditional market-driven recovery channel as well as the new RFID based channel. A mathematical model is developed considering the cost of implementation and the design of the proposed RFID based recovery channel. Recovery-channel selection is studied, and results show that a hybrid collection strategy with 85% share of channel-1 and 15% of channel-2 is the optimal one. Moreover, the collection from the proposed RFID based channel increases as the demand increases. For the proposed RFID based system, reader sensing power is found more significant compared to the cost of readers. A numerical example is given with three different cases and impacts of different input parameters are studied to draw important results. Managerial insights are given to assist the designer of the system in some critical decisions.

1. Introduction With the world currently focused on sustainability, remanufacturing and closed-loop supply chain management have received unwavering attention of both the industrial and the academic world (Panda et al., 2017; Maiti and Giri, 2017; Habib et al., 2019; Ullah et al., 2019). Remanufacturing and recycling are even more important in the cell phones industry because some components of electronic products are usually hazardous, and improper disposal causes severe environmental and health threats. However, remanufacturing is enabled by product take-back, and unfortunately, the return rate in electronic product industry is very low. According to World Bank statistics, the number of cell phone subscribers globally exceeded 7 billion in 2015. This increase indicates that there is a huge number of retired cell phones. According to available data, nearly 400 million wasted cell phones are generated each year in the world (Xu et al., 2016). Of which, only one percent are recycled properly; and even in the developed world, for example, the

*

US, the collection rate is below 20%. Traditionally, the return rate is considered exogenous in nature, we reject this assumption and suggest a novel recovery system to control the return rate of obsolete cell phones. This paper develops an RFID based recovery channel to improve the return rate in cell phone remanufacturing. A hybrid manufacturing-remanufacturing model is developed with two different recovery channels: the traditional market-driven channel and the proposed RFID based recovery channel. It is pertinent to mention that the profitability of any remanufacturing industry depends upon the quality of the collected products, which is highly random in nature. In the case of two recovery channels, return product quality and recycling cost defers from each other. In such scenarios, production coordination and recovery-channel selection are two of the most important lot-size independent production decisions. Therefore, this study investigates the return rate and channel selection in a dual-recovery-channels setting, considering stochastic quality of used product with RFID based recovery channel.

Corresponding author. E-mail address: [email protected] (B. Sarkar).

https://doi.org/10.1016/j.ijpe.2019.07.017 Received 6 September 2018; Received in revised form 1 June 2019; Accepted 15 July 2019 Available online 18 July 2019 0925-5273/ © 2019 Elsevier B.V. All rights reserved.

International Journal of Production Economics 219 (2020) 360–374

M. Ullah and B. Sarkar

on two questions: (i) Why it is important? And (ii) what is the current situation of cell phone recycling in the world? The third research area studies the use of RFID technology in a supply chain, especially in terms of reverse logistics and a closed-loop supply chain. Because this paper contributes to all these research areas, they are reviewed briefly in the following paragraphs.

1.1. Return rate in the cell phones industry In a very short period, cell phones have become an integral part of our life and have emerged as a ubiquitous electronic product (Li et al., 2015a, 2015b). According to a World Bank report, the total active cellular subscribers reached 7.68 billion, with a total population of 7.53 billion in 2017 (Mobile cellular subscriptions, 2018). This has dramatically increased the global sale of cell phones; more than 1.6 billion cell phones were sold globally in 2010 alone, even greater numbers were sold in the ensuing years (Welfens et al., 2016). Out of which, only 10 to 20 percent are returned and disposed of properly, while the rest of cell phones end up in municipal wastes, drawers, cupboards, and other storages in houses and offices (Silveira and Chang, 2010). The cell phone example is just a tip of the iceberg; this happens with all types of electronic products. People even keep the end of life/end of use (EOL/ EOU) computers, monitors, tablets, and other electronic waste in their storages. Therefore, the return rate in electronic products is always low. The main reason behind this huge sale is the rapid replacement of cell phones by consumers, which is driven by technological innovation. Cell phone manufacturers are rapidly introducing new models by adding different features into their cell phones, which drives the consumers to replace their old phones with the latest models. This increases the replacement frequency, and, as a result, the electronic waste from cell phones increases exponentially. In fact, cell phones have the shortest life cycle of all electronic products. Research shows that more than 50% of the phones have a life cycle between three to six months, and almost 90% of the phones have a lifecycle less than a year (Chan and Kai Chan, 2008; Zeng and Hou, 2018). The technical lifetime of a cell phone, as believed by the manufacturers, is 10 years; while the average estimated replacement time is 1–2 years (Geyer and Blass, 2010). Thus, the majority of cell phones that enter its EOL/EOU stage may still have value (in terms of performance and durability). They can be reused, recycled, refurbished, or remanufactured if properly collected and recovered. Much of the previous research, concerned with the cell phone supply chain, focus on waste generation rate, existing collection systems, recycling processes, and associated risks (Silveira and Chang, 2010; Jang and Kim, 2010; Welfens et al., 2016). Majority of the published papers studied the existing return systems and only pointed out the problem of low returns in the cell phone supply chain, including works by Sullivan (2006); Ongondo and Williams (2009); Geyer and Blass (2010); Wilhelm et al. (2015); and Welfens et al. (2016). However, no published research explicitly considered solutions for the low return rate problem. The main restrictive assumption so far in this regard is the belief that cell phone return is an exogenous process and is, therefore, outside the direct control of the manufacturer or the collection firm. This research rejects this assumption and enables the manufacturers and the collectors to control cell phone collection quantity. We proposed an advanced collection system as an innovative recovery program for obsolete cell phones. The system utilizes the internet of things (IoT) technology (viz. RFID tags, readers, and sensors) to increase the return rate of EOL/EOU cell phones. The remainder of this paper is structured as follows: Section 2 provides a brief review of the related literature; The problem is defined in Section 3; Section 4 explains the proposed system; Section 5 provides a mathematical model with solution methodology; Numerical examples, important insights for managers, and cost justification of the proposed system are given in Section 6; while Section 7 concludes the paper.

2.1. Used product quality in hybrid manufacturing-remanufacturing systems Even though remanufacturing is considered a new terminology, research in this field is dated back to 1960s, with the first work reported from Schrady (1967). He developed an inventory model, named as (R, I) policy, for products that can be repaired with single manufacturing batch followed by many repair batches. He ignored disposal cost and considered infinite and instantaneous repair rate. Later on, Nahmiasj and Rivera (1979) considered finite storage and repair rate and extended Schrady (1967) work. Taking remanufacturing literature towards a new dimension, Richter (1996a) and Richter (1996b) extended Schrady's model and proposed a hybrid system that considered both production and repair with disposal. Richter assumed that demand can be fulfilled by producing new products from raw material and by repairing used/collected items from the end customer. Extending his own work, Richter (1997) showed that a pure recovery or pure disposal is the optimal policy. With similar findings, this work was further extended by Richter and Dobos (1999); Dobos and Richter (2004); and Dobos and Richter (2006). For a comprehensive review, readers are referred to study Govindan and Soleimani (2017) and Diallo et al. (2017). Dobos and Richter (2006) studied used product quality and assumed that all returned products are not suitable for remanufacturing. They found that in such cases, a mixed strategy economically outperforms pure strategies. Aras et al. (2006) studied a hybrid system and assumed used product dependent remanufacturing cost and lead time. Hwang et al. (2009) assumed a minimum quality level for used products, and only products having quality above that level are bought back from consumers. They also considered that the quality of used products is a random variable, and the collection rate depends on the minimum quality level and the incentives paid to the consumer. In a very similar manner, El Saadany and Jaber (2010) also studied price and minimum acceptable-quality dependent return rate. Their results suggested that a mixed policy is the optimal one. Korugan et al. (2013) studied the effects of used product quality on the remanufacturing process. In Guo and Ya (2015), the return rate, buy-back price, and remanufacturing cost depends upon the used product quality level, which is exponentially distributed.Li et al., 2015b and Giri and Sharma (2016) studied remanufacturing yield, which is the percentage of used products that meet the quality criterion for remanufacturing. Recently, Moshtagh and Taleizadeh (2017) used quality based return rate and modeled a CLSC with three different distributions for the return rate; while Tian and Zhang (2018) studied pricing and disassembly scheduling of returned products with price dependent yield. Maiti and Giri (2017) studied two-way recovery, however, they did not consider used product quality. In fact, they only focused on finished product quality. Heydari et al. (2018) considered stochastic remanufacturing capacity. Ahmed et al. (2018) studied the repair of imperfect quality items with trade credit policy; and Sarkar et al. (2018) discussed products quality with variable production rate. Sarkar (2019) studied product quality in terms of defective products in multi-stage production system, and Dey et al. (2019) considered quality improvement with variable production rate. All these authors considered a single recovery channel, with all products having the same quality. Furthermore, the impacts of dualrecovery channel selection have never been studied in the literature. Therefore, there is a need to revisit used product quality in dual recovery channel settings.

2. Literature review Three main areas of research are of particular relevance to this study. The first research area studies returned product quality in hybrid manufacturing-remanufacturing models. The second research area relates to product return in the cell phone supply chain, which is focused 361

International Journal of Production Economics 219 (2020) 360–374

M. Ullah and B. Sarkar

fee, which the consumer has to pay, and the return rate for cell phones in 2009 was reported to be 20% (Silveira and Chang, 2010). The Ministry of Environment South Korea initiated a waste deposit program in 1992. This program required manufacturers to deposit a cost in proportion to their production and retrieve the amount in proportion to the recycled products. Unfortunately, the program was not successful, and the government had to abolish it because the manufacturers simply deposited the charges and did not make any effort to recycle (Ministry of Environment, 2016). The government of California enacted a law regarding cell phone recycling (California Cell Phone Recycling Act of 2004), obliging the retailers to accept every cell phone from their customers for recycling. The return percentages were 17%, 25%, 27%, 21%, 15%, 13% for the years 2007–2012, respectively (DTSC, 2006). The main reason for this failure was the lack of action by the manufacturers because a take-back system is complex, expensive and increases management complexities. Besides, the economic viability of reuse and take-back activities is also unclear. Therefore, manufacturers are usually against the implementation of these types of policies and laws. Traditionally, the return rate is considered an exogenous variable and out of the direct control of the manufacturer. This assumption provides a way for the manufacturers to skip their responsibilities by calling it an “out of control process”. If the quantity of the return products is considered a controllable variable, then the manufacturers can be forced to collect all products sold previously. Thus, we will focus on developing a system in which the return rate is directly under the manufacturer's control. We reject the idea that firms must passively accept product returns, and devise a system in which return rate is under control. In this system, EOL/EOU products can be traced and retrieved with the help of RFID technology.

2.2. Recycling in cell phone supply chain Recycling and reuse is becoming one of the most important parts of supply chain and production (Iqbal and Sarkar, 2019). The main aim of EOL/EOU cell phone management is to prevent them from ending up in municipal waste in order to reduce environmental deterioration and health risks. There are two main reasons for managing these wastes independent of municipal wastes (Khetriwal et al., 2009). First, they contain hazardous elements, which can contaminate the environment, if released into soil or water. Second, refurbishing and recycling programs can lead to the retrieval of precious raw materials and investment tied to these cell phones (Huisman, 2003). Cell phone wastes carry several metals representing a real reserve of metal resources, which should be recovered. Although some may argue that the metal content per cell phone is quite small, however, as billions of phones are in use globally, which collectively represent tremendous quantities of different metals (Silveira and Chang, 2010). Ideally, the significant increase in sales should correspond to an increase in the collection and recycling of cell phones. However, the collection rate is very low and we do not even know the fate of a huge number of cell phones. Welfens et al. (2016) reported that in the US alone, more than 300,000 cell phones are discarded every day. Silveira and Chang (2010) reported that the collection rate of EOL/EOU cell phones in the United States is less than 10%. Geyer and Blass (2010) cited a report from the US Environmental Protection Agency, which estimates that the collection rate in the US is below 20%. According to a US Geological Survey report (Sullivan, 2006), almost 500 million EOL/ EOU cell phones would have accumulated in different storage areas waiting for proper disposal or recycling in the US by 2005. The report further stated that less than 1% of the total obsolete phones are discarded and recycled properly. Sullivan (2006), Geyer and Blass (2010), and Wilhelm et al. (2015) suggest two main reasons that contributed to low returns in the cell phone supply chain, 1) discarding into municipal waste and 2) hibernation period. The small size is the main reason for both problems. Due to their smaller size, people often do not take care of the retired cell phones and throw them away in municipal waste. Even environmentally conscious consumers also neglect this problem, because they do not realize risks associated with an improperly disposed cell phone. This work focuses on the possible solutions to the previously mentioned two problems, and introduces a smart and novel recovery system that minimizes the hibernation period and facilitates the separation of discarded cell phones from municipal waste.

2.2.2. Risks related to improper disposal of cell phones For unused cell phones, hibernation is one of the two main reasons responsible for lowering the return rates. Usually, the unused cell phones are kept in drawers, storerooms, and other areas once the consumer stop using them. This is a huge loss of financial resources. According to INFORM, Inc. (a non-profit organization), an estimated 500 million discarded phones accumulated in customer storages by 2005. The total investment tied to these 500 million phones is estimated to be $314 million. According to Ongondo and Williams (2009), 50–90 million cell phones are spoiled in drawers and cupboards in the UK. Apart from the monetary losses, these cell phones pose a great threat to human health and the environment. A cell phone may contain more than 40 elements including antimony, silver, barium, copper, cobalt, chromium, gold, iron, lead, nickel, palladium, tin, aluminum, and zinc. Out of which, more than 12 are identified as potentially harmful for human health and the environment. In terms of its weight, almost 35–40% of the total weight of a cell phone contains metals (Silveira and Chang, 2010; Wu et al., 2008). These elements are persistent bio-accumulative toxins (PBTs) and are carcinogenic, therefore, can cause reproductive, neurological, central nervous system, kidney problems, immune system dysfunction, and endocrine problems (Silveira and Chang, 2010). Landfilling of electronic product disposal has also some serious consequences. For instance, high concentrations of lead and cadmium were found in rice from a waste recycling area in China, and the soil was found to be severely contaminated by copper, cadmium, and mercury (Fu et al., 2008). Not only this, but the improperly disposed of phones are also causing pollution by contaminating household wastes, soil, wildlife, water, and marine life. For example, cadmium leakage from a single cell phone battery could contaminate 600,000 L of water (Manivannan, 2016). Therefore, it is important to manage EOL/EOU cell phones in an environmentally friendly manner. Serious actions are required to improve recycling in this sector, and the hurdles should be removed by approaching the problem in novel scientific ways.

2.2.1. Existing collection methods in the cell phone industry and their failure Many countries are operating different e-wastes management techniques, and have adopted legislation and economic tools to implement these systems. According to Ongondo and Williams (2009), the exact number of cell phone take-back schemes operating in the world is unknown, and an estimated 102 different schemes are operating in the UK alone. One important policy developed by the Organization for Economic Cooperation and Development (OECD) is Extended Producer Responsibility (EPR). EPR extends the manufacturer's responsibilities to the post-consumer stage of the product's lifecycle, including take-back, recovery, and disposal (Nnorom and Osibanjo, 2008). The EPR policy is implemented through (i) take-back programs (like regulatory approaches, mandated recycling, minimum recycled content standards, energy efficiency standards, and disposal bans and restrictions), (ii) voluntary industry practices and environmental labeling, and (iii) economic instruments (like the Advance Recycling Fee (ARF), advance disposal fee (ADF), virgin material taxes/subsidies, and deposit/refund schemes) (Milanez and Bhrs, 2009). However, literature review shows that almost all these approaches failed to improve the return rate above 27% (achieved in California only in 2009), and even that percentage also showed a drastic decrease down to 13% within just two years. Electronic product prices in Japan include an EOL/EOU management 362

International Journal of Production Economics 219 (2020) 360–374

M. Ullah and B. Sarkar

Fig. 1. The proposed dual-channel hybrid manufacturing-remanufacturing production system.

implied to decrease inventory inaccuracies in a stochastic environment. Cui et al. (2017) investigated the effectiveness of RFID investments under coordinated and uncoordinated scenarios. The only work that studied RFID in waste management is done by O'Connell et al. (2013). However, they focused only on the separation of EOL/EOU products from waste streams, while this work considers separation from the waste stream along with tracking in storages. Ullah and Sarkar (2018) studied the use of RFID to improve electronic waste management, however, they did not consider dual channel return system. Besides, no one has studied the use of RFID to improve the return rate of EOL/EOU products in supply chain and manufacturing context.

2.3. Use of RFID in the supply chain IoT has drawn increasing attention over the past several years in every field. IoT corresponds to the interconnection of uniquely identifiable embedded computing devices within the existing Internet infrastructure. With IoT technology, a product can be equipped with a uniquely identifiable code, which can then be monitored and tracked by using sensors and wireless sensor networks (Fang et al., 2016). Cisco's Internet Business Solutions Group (IBSG) predicts that there will be 50 billion IoT devices connected to the Internet by 2020 (Yang et al., 2015). IoT uses the RFID technology to enable tracking and identification of every individual product. Like other fields, the notion of IoT and RFID in supply chain management has also shown increasing applications to enable traceability, improve visibility, and enhance inventory management in the supply chain (Yang et al., 2015). RFID finds a more extensive application in reverse logistics and closed-loop supply chain due to the possible recovery of the expensive RFID tags. Kang and Gershwin (2005) showed that RFID systems can result in lower inventory levels and higher service levels by identifying stock losses. Diekmann et al. (2007) studied a two-echelon supply chain model that used an RFID system to track the movement of products between different nodes. They studied the effects of an RFID system on the members of the supply chain and the cost distributed among the members. Kiritsis et al. (2008) studied RFID in closed-loop supply chains and proposed a way to close the information gaps with RFID technology. Xu et al. (2009) studied a closed-loop supply chain and illustrated the use of wireless-technology for product lifecycle monitoring with a case study from the PROMISE project. Luttropp and Johansson (2010) suggested that these devices can be used to store disassembling and recycling information due to the large memory offered by ultra-high frequency RFID tags, thereby making recycling easier. According to Curran and Williams (2012), RFID technology will also be investigated, designed and tested in the ZeroWIN (Toward Zero Waste in Industrial Networks) project. This shows the potential of RFIDs to reduce industrial waste. Kim and Glock (2014) studied the impacts of RFID systems on the management of returnable transport items. They showed that RFID can improve the performance of the system. Ondemir and Gupta (2014) studied a remanufacturingto-order and disassembly-to-order model that utilized RFID to reach optimum decisions related to remanufacturing, disassembly, recycling, disposal, and storage. Fan et al. (2015) discussed how RFID can be

3. Problem description and notation This section defines the proposed problem and provides notation used to develop a mathematical model of the proposed problem. 3.1. Problem definition The studied literature suggests the importance of recycling and remanufacturing in the electronics industry. However, remanufacturing in electronic products, and especially in the cell phone industry, faces the problem of low return rate. Therefore, it becomes crucial for researchers to sort out this problem and develop new approaches in order to improve the return rate. The return rate of cell phones must reach the 100% mark due to financial losses; and, more importantly, due to the potential threats to human health and the environment. This paper is the first step in this direction and uses IoT to design a second recovery channel for remanufacturer to improve the return rate. The proposed system improves the collection and separation of cell phones and other small electronic products (for example digital cameras) from municipal waste, and also track them in storages inside houses. A hybrid manufacturing-remanufacturing model is considered that produces finished products by manufacturing from raw material and remanufacturing from EOL/EOU products. Fig. 1 provides the proposed dual-channel manufacturing-remanufacturing system. In practice, remanufactured items are perceived, by some customers, to be of lower quality compared to newly produced items. In such a case, the consumer does not pay the same price for both types of products. Hence, the manufacturer sells remanufactured products with a lower price compare to manufactured products. Furthermore, the 363

International Journal of Production Economics 219 (2020) 360–374

M. Ullah and B. Sarkar

3.2. Notation

manufacturer sells remanufactured products either in primary or in a secondary market; secondary markets are developing markets where the demand for remanufactured products is higher because of its low price. Usually, remanufactured products are introduced into a market where the manufacturer believes that manufactured products cannot be sold because of their higher prices. This strategy can help organizations to increase their profit and market shares. As discussed earlier, one of the main problems in the cell phone industry is their fast-developing technological innovation which results in smaller cell phone life cycles. The demand for any product decreases very quickly once a new model is introduced into the market. Therefore, we assumed that demand for the product decreases with time such that for the t th cycle, demand is given by the function,

x t = (x 0 −

Following notations are used to develop a mathematical model of the proposed production system. Decision variables

τ1 collection rate from RFID based system (percent of demand) τ2 collection rate from the traditional market-driven system d maximum separation between two readers Parameters

x t total realized demand at manufacturer during time period t (units/period) x r,t realized demand for remanufactured products during time period t (units/period) x m,t realized demand for new items during time period t (units/ period) qfr average quality of the finished product produced through remanufacturing qrw < qfr < 100 qfm average quality of the finished product produced through manufacturing 0 < qfm < 100 Crw cost of retrieving a single product through the proposed RFID based system ($/unit) Crm cost of collecting a single product through the market-driven system ($/unit) Ccs shipment cost per EOL/EOU product collected through RFID and market-driven systems ($/unit) Cis inspection cost of EOL/EOU products collected through the market-driven system ($/unit) qrw average quality of only remanufacturable EOL/EOU products, collected through the proposed RFID based system λ ≤ qrw < 100 (percent) λ average quality of all EOL/EOU products, collected through proposed RFID based system, 0 < λ < qrw (percent) qrm average quality of only remanufacturable EOL/EOU products collected through the market-driven system μ ≤ qrm < 100 (percent) μ average quality of all EOL/EOU products collected through the market-driven system, 0 < μ ≤ qrm (percent) maq minimum acceptable quality for remanufacturing 0 < maq < 100 (percent) Cq quality upgradation cost ($/unit) Cds disposal cost of product having quality below the minimum acceptable quality level ($/unit) Z= f(d) cost of the RFID system ($) which is a function of sensing distance (d) c1 price of type-1 reader ($/item) l length of the search area (m) w width of the search area (m) c2 price of type-2 reader ($/item) rs the threshold below which type-2 reader can sense the tag with probability 1 (m) Ss sensing radius of the type-1 reader (m) St transmission radius of the type-1 reader (m) α decay parameter for sensing with distance Ω reader threshold parameter, where (0 < Ω ≤ 1) CL cost of labor unit ($/labor unit) nL number of labor units required per kilometer (labor units/km)

β *t )+

where x t is the manufacturer's total realized demand, which is the sum of demand for manufactured and remanufactured products. Furthermore, x 0 is the initial realized demand of the manufacturer and β is decrement per period. We assumed that x r,t is the demand for remanufactured products, which as gt percent of the manufacturer's realized demand x t , such that x r,t = x t gt . Then x m,t = x t − x t gt is the demand for new products. If λ is the average quality of collected products and τ1 is the percentage of collected products from RFID based channel, maq e−λλk

then (∑k = 0 k ! )(τ1*x r,t ) is the quantity of used products, collected from RFID based channel, having a quality level less than maq ; this quantity is non remanufacturable and must be produced by manufacturing. Therefore, the maximum quantity that can be satisfied from the remanufacturing xt − (∑maq k=0

process

is

xt − (∑maq k=0

e−λλk )(τ 1 * xr,t ) k!

xt

, hence, 0 < gt

e−λλk )(τ 1 * xr,t ) k!

< . However, in reality, the value of gt is much lower xt than this upper bound. For gt = 0 , the demand for remanufactured products is zero. For a higher value of gt , the demand for remanufactured products increases. Total demand for remanufactured products x t gt is satisfied by remanufacturing EOL/EOU products recovered from two different recovery channels. Whereas, the demand for manufactured products ( x m,t ) is satisfied by producing new products from raw material. The manufacturer recovers EOL/EOU products from the market through the traditional market-driven system as well as through the RFID based recovery channel. The traditional market-driven recovery system relies on financial incentives paid to the consumer. Although the quality of EOL/EOU products, obtained from this channel, is higher compared to the second channel, however, the return rate of this channel is very low. As incentives depend upon product quality, hence, consumers tend to return products with higher quality to get higher incentives. The manufacturer defines a minimum acceptable quality and purchases EOL/EOU products having quality above this level. On the other hand, the quality of used products obtained from waste streams is the lowest one. The manufacturer does not have any power over this quality and, therefore, passively accepts all products, inspects them and categorizes them according to the minimum acceptable quality level. Products above this level are remanufactured while products below this level are disposed of with some disposal cost. The manufacturer also retrieves used products from hibernation and storages. Similarly, like the waste stream channel, the manufacturer passively accepts all products and categorize them based on the minimum acceptable quality. However, quality from this source tends to be higher than the quality obtained from waste streams. To address the explained problem, this paper focuses on reverse logistics and proposes a novel system that uses RFID to track, collect and separate cell phones from storages and household waste. We first discuss the requirements of the system for effective tracking of EOL/ EOU products, followed by an explanation of the proposed system and the benefits attained by the system. Finally, we develop a mathematical model and study barriers associated with the application of the system.

4. The proposed recovery system EOL/EOU product management is mandatory in many sectors, and manufacturers are often forced by legislations to take-back their EOL/ EOU products. For instance, legislation in Europe regarding tires requires every manufacturer to dispose of one used tire for each sold tire. But there are no such laws in cell phone manufacturing, and therefore, a 364

International Journal of Production Economics 219 (2020) 360–374

M. Ullah and B. Sarkar

huge number of phones are not recycled properly. Hence, there is a huge problem regarding EOL/EOU management of cell phones. However, RFID has the potential to solve the problem and alleviate the risks by improving the return rate and separating the discarded phones from municipal waste. The main function of RFID technology is to detect objects that are tagged with the RFID tag. This is the required feature that can be used in the cell phone industry to mitigate risks and financial losses, and to improve the return rate. RFID enables the unique identification, monitoring, and tracking of different objects in a real-time environment by using radio waves to transfer the data. In simple RFID systems, a tag is attached to each item that is to be tracked. The tag is made from a tiny integrated circuit (IC), called a tag-chip, which is connected to a small antenna and can be built in a variety of shapes. The tag-chip contains read-only or read-write memory, which stores the required information. Each tag transmits a distinctive electromagnetic (EM) signature, which is captured by a device called reader that enables a host computer to recognize the object related to the tag (Mason et al., 2012). The data linked to the tag is either recalled from the memory of the computer or any other associated action is triggered. In particular, RFID systems can answer the following four questions; who/what/where/and how are you? The answer to “where are you?” is required to solve the problem of tracking the cell phone. “How are you?” would answer whether the phone is working or obsolete. “Who/what are you?” provides the dismantlement information required for the recyclers. Table 1 summarizes different RFID technologies (class) and provides an overview of the features of these classes (Types of RFID Systems, 2018). We proposed a method to embed RFID tags in cell phones during their manufacturing. Once the phone is obsolete and the user throws it in the storage, the phone battery dies after some time and it becomes impossible to locate the phone. The proposed RFID tag system can, therefore, be used to trace the position of the phone. Because RFID tags do not need power from the cell phone battery, therefore, can respond to the reader even when the phone does not have a battery. Once the reader receives the signal, the presence of the phone is noted in the read range of the reader. The location of the phone is then traced by using a location tracking system. An accelerometer is used to determine whether the phone is hibernated or in use. The accelerometer senses the motion and if the phone has not been moved in some specific time, the phone is declared hibernated and retrieved from the storage place. Tracking a cell phone in household waste is much easier than locating a hibernated cell phone. This can be done by passing the waste containers through RFID gates. When a container having a discarded cell phone is passed through the gate, the tag receives the signal from the reader, actuating a system that produces a sound. Thus, the cell phone can be traced. Alternatively, the truck that collects waste from the waste bins is equipped with the RFID system. When it approaches a waste bin that contains a discarded cell phone, the system produces a sound that informs the driver about the presence of a cell phone in the waste bin. Before collection, the cell phone is retrieved from the waste bin. Fig. 2 shows the proposed application of RFID in the EOL/EOU management of cell phones.

Fig. 2. The proposed RFID based system.

4.1. Structure of the proposed tracking system The system consists of the RFID hardware and software that provides a user interface and connection to google maps, or any other map, to locate the cell phone. The RFID hardware consists of tags and readers. Tags are embedded into cell phones, whereas readers are portable and can be carried anywhere. During operation, the reader continuously interrogates the presence of tags in its read range. It should be noted that the reader interrogation frequency is directly related to the power consumption of the tag. In the current context, this is not a problem because tags are not supposed to be interrogated frequently. Sensors are also embedded in the tag to monitor cell phone performance. An accelerometer is installed that senses the movement and stores the data in the tag. Once the tag is connected, the data is transmitted, which determines whether the phone is in use or obsolete. The storage period (sensor data) is a user-defined parameter and depends on the storage capacity of the tag. Fig. 3 shows the indoor tracking of cell phones. The tag can communicate directly with the reader or indirectly through another tag. In this way an ad-hoc network can be created on demand, thus reducing the number of required readers. The retrieval phase, however, is costly and cannot be done frequently or for fewer cell phones. Therefore, to minimize the cost, this paper divides the complete process into two phases, 1) initial tracking, 2) final tracking and retrieval. The first phase is basically a lengthy data collection phase, in which the collection agency gathers location data for obsolete cell phones. The data is accumulated in a database and after enough data is collected, the second phase is launched in which obsolete cell phones are tracked accurately and finally retrieved. 4.1.1. Initial tracking Active tags, capable of tag to tag communication, are used in the proposed system. This means that each cell phone tag can communicate with another tag. When a reader-tag (tag in a cell phone that is in-use) senses another tag in a hibernated cell phone, the reader-tag activates the onboard Global Positioning System (GPS), and determines the coordinates of the reader phone subsequently sending the information (coordinates) to a central data managing server with the received signal

Table 1 Classification of RFID systems on the basis of features. Class

Tag

Communication type

Description

Class 1 Class 2

Tag to Reader Tag to Reader

Read-only, Tag can store a unique object identifier only, communicate with a reader or active tags of class 5. Extend class 1 with read/write memory, communicate with a reader or active tags of class 5.

Class 3 Class 4

Identity Tags, Pure passive Higher Functionality Tags, Pure passive Semi-Passive Tags Active Ad-hoc Tags

Tags with internal integrated battery for operating, communicate in the absence of passive tag reader. Internal power source, can communicate with other active tags (other Class 4 and Class 5 Tags) and readers,

Class 5

Reader Tags

Tag Tag Tag Tag Tag

to to to to to

Reader Tag Reader Tag Reader

Most sophisticated system, can communicate with the reader, active (Class 4 and Class 5), and passive tags (Class 1, Class 2, and Class 3).

Source: Mason et al. (2012); Engels and Sarma (2005). 365

International Journal of Production Economics 219 (2020) 360–374

M. Ullah and B. Sarkar

Table 2 Calculation of power loss coefficients N, floor penetration loss factors Lf with n being the number of floors penetrated, for indoor transmission loss. Frequency Distance power loss coefficients, N

Floor penetration factor Lf (dB) with n being the number of floors penetrated Series (2012)

900 MHz 1.2–1.3 GHz 1.8–2.0 GHz 4 GHz 60 GHz 900 MHz

1.8–2.0 GHz

Office

Commercial

33 32 30 28 22 9 (1 floor), 19 (2 floors), 24 (3 floors) 15 + 4(n − 1)

20 22 22 22 17

6 + 3(n − 1)

of the obsolete phones are retrieved and sent to the collection centers for inspection. Based upon quality, the retrieved obsolete cell phones are then sent to either a reuse market, or for refurbishing, remanufacturing, recycling, or incineration. 4.1.2. Final tracking and retrieval The tracking agency (manufacturer, government or recycler) schedules the final tracking and retrieval of cell phones from those areas indicating the presence of a hibernated cell phone. The retrieval frequency is user-defined, decided based upon the number of hibernated cell phones initially detected in phase one. A search is launched when enough phones are detected. The location of the phone is determined by the LANDMARC tracking system developed by Ni et al. (2004). In this approach, the search area is divided into subareas, by placing reader and reference tags in a specific manner, as shown in Fig. 5. This reduces the search area, thus making it easy to locate a cell phone. For example, in Fig. 5, a cell phone in the white colored home is sensed by reader 1 and reader 9, while a cell phone in the black colored home is sensed by reader 1 and 3.

Fig. 3. Indoor tracking of cell phones and tag-reader communication.

4.1.3. LANDMARC system The objective of the LANDMARC system is to increase the accuracy of the tracked location with fewer RFID readers. To accomplish this objective, LANDMARC uses the concept of reference tags placed at fixed and known locations in order to promote tracking of the unknown tags.

Fig. 4. Flowchart of the proposed initial tracking system.

strength (RSS). This provides information about a stationary tag being discovered inside the read range of the transmitted coordinates, thus, providing a rough idea where a hibernated cell phone is placed. The tracking algorithm for initial tracking is shown in Fig. 4. The message consists of the information of the stationary tag (perspective tracking tag), the location of the reader-tag, and the RSS. An estimated distance between transmitted coordinates is calculated from the RSS using the following formula (Series, 2012),

RSS (dB ) = 20 log10 f + N log10 dr + Lf (n) − 28 dr =

(1)

28 + RSS − n Lf − 20 f log10 N log10

(2)

where,

dr the distance between reader tag and tracking tag (m) Lf floor penetration factor of the signal strength loss (dB) f frequency (MHz) n number of floors N distance power loss coefficient Table 2 provides details related to the power loss coefficient (N) and floor penetration factor (Lf ) for office and commercial buildings. When enough data about obsolete phones are collected, the second phase is launched, in which an RFID tracking system is deployed in specific areas. The exact location of the tag is now determined, and all

Fig. 5. Dividing the search area into subareas by the placement of different readers and readers tag. 366

International Journal of Production Economics 219 (2020) 360–374

M. Ullah and B. Sarkar

maq e−λλk

This system focuses on determining the nearest reference tag to a tracking tag by comparing the signal strength of both the reference tag and the tracking tag. A weighted average factor wj , which is based on the distance between the reference tag from the tracking tag, is assigned to every reference tag. The object location is determined by computing the reference tag's real position (which is already known) and the preassigned weighting average. For a system consisting of n readers, m reference tags and u tracking tags, the signal strength vector of a → tracking tag is defined as S = (S1, S2, ..., Sn ) , where S i symbolizes the signal strength of the tracking tag observed on reader i . Another vector → θ = (θ1, θ2, ..., θn ) is defined, where θi represents the signal strength of the tracking tag. For each tracking tag p ε (1, u), a Euclidean distance vector r j is defined that shows the signal strength between reference tag and the tracking tag, and is given as;

Ej =

have remanufacturable quality can be written as 1-∑k = 0 k ! . Based on these assumptions, the average quality of the remanufacturable pro( ∑100 k = maq + 1 k

lected products can be written as qrm =

For tracking tag i , Ej is set of the location relationship between the reference tag and the tracking tag. For m reference tags ⎯→ ⎯ Ej = (E1, E2, ..., Em ) the system selects k reference tags, and the tracking location is obtained using the following formula, k

(4)

where wi is the weighting factor, assigned to the i determined by the formula,

wj =

th

reference tag and is

1 2 Ei

∑ik= 1 1 E 2

(5)

i

This means the reference tag with the smallest E value has the largest weight. 5. Mathematical model

100 − maq

.

maq e−λλk )(τ1*x r,t )+ k! Ccs + Cis) + Cq (qfr2

+ Cds (∑k = 0 (τ2*x r,t )(Crm +

2 2 − qfm )(τ2*x r,t ) + W *τ22 + (x m,t ) Cq ( qfm )

The first term shows the investment to establish RFID system; the second term shows collection and transportation cost of EOL/EOU cell phones through RFID system, and the third term is quality upgradation cost of converting the retrieved quality level qrw into required quality level qfr . We assumed that the quality upgradation cost has a decreasing return value in terms of quality, and therefore, used a quadratic function to express the declining return of quality. This type of function is common to show the declining return rate and reported by many including Sarkar et al. (2017) and Maiti and Giri (2015). The fourth term gives the disposal cost of non-remanufacturable products having a quality level less than the minimum acceptable quality level. Where maq e−λλk

∑k = 0 k ! gives the cumulative Poisson probability for products having quality less than maq . The fifth term gives collection, inspection, and shipment cost of EOL/EOU products using the traditional market-driven system. The sixth term provides quality upgradation cost for products collected through the traditional market-driven system. The second last term shows the investment required to establish inspection and sorting facility at collection centers for the market-driven system, and the last term shows the cost of manufacturing x m, t products from new parts. After simplification, the total cost of the system can be written as,

5.1. Quality of collected products The quality of any EOL/EOU product ranges from 0 to 100. The minimum quality a product can have is 0, which means no part of the product is reusable and, therefore, all parts need replacement in order to get as good as new product. Conversely, if a returned product has a quality level 100 then all parts of the product are working perfectly fine and no extra procedures are required. The product can be sold in the market with a little bit of cosmetic treatment. If the quality follows a Poisson distribution with mean λ , than the probability that any product

maq e−λλk ) gt x t τ1 k!

TC = Zτ12 + Cds (∑k = 0 Cq (qfr2

e−λλk

can have a quality level K can be written as P (k , λ ) = k ! . Now the cumulative probability of all products having quality less than the minimum acceptable quality level can be written as maq e−λλk , k!

e−μμk ) k!

2 TC = Z τ12 + (τ1*x r,t )(Crw + Ccs) + Cq (qfr2 − qrw )(τ1*x r,t )

This paper considers a firm that produces cell phones by two methods: manufacturing cell phones by purchasing new parts and remanufacturing cell phones by refurbishing EOL/EOU cell phones collected from consumer and waste streams. EOL/EOU cell phones can be collected from consumers through two different channels: 1) retrieval by the proposed RFID system, and 2) market-driven system. Generally speaking, the quality of used cell phones retrieved through the above two channels differs from each other. The market-driven system basically relies on high financial incentives; therefore, consumers tend to return the product with high quality. On the other hand, hibernated cell phones or those ended up in the waste stream have low quality, however, the manufacturer does not need to pay much for these cell phones except the initial investment to establish the collection system. Now the problem is to define the optimal return strategy and the optimal percentage of demand fulfilled from remanufacturing and manufacturing cell phones from new parts.

P (k ≤ maq, λ ) = ∑k = 0

(∑100 k = maq k

The manufacturer produces finished products with two different methods, manufacturing and remanufacturing, based on demand x m, t = x t − x t gt and x r,t = x t gt respectively. For remanufacturing, used cell phones, obtained from two different channels, are the raw material. τ1 percent of demand x r,t is satisfied from RFID based channel and the remaining (τ2 x r,t ) is satisfied from the market-driven system. For the RFID based channel, the manufacturer bears an initial investment (Z ). The demand for new products is satisfied by manufacturing new cell phones from virgin raw material. Raw material obtained from all the above sources have different quality, and therefore, incurs different costs to the manufacturer. Retrieval from waste streams is the most inexpensive option, but with a very low average quality of the retrieved cell phones. A portion of these collected cell phones are not remanufacturable because of their low quality. In this case, these cell phones are divided into two categories. The manager decides a minimum acceptable quality (maq) level for remanufacturing. All products above this quality level are remanufactured while the remaining products are disposed of properly. EOL/EOU cell phones collected from the market-driven system, usually, have good quality and require a low cost to convert it into finished products. However, their procurement cost is higher because of the return incentives paid to the consumer and intensive inspection to assess the quality. Based on the above assumption total cost (TC) of the manufacturer can be written as,

(3)

(x , y ) = ∑i = 1 wi (x i , yi ).

)

5.2. 1. Model

n

∑i = 1 (θi − Si )2 , ∀ j ε (1, m)

e−λλk

k! . For duct, for channel-1, can be written as qrw = 100 − maq channel-2, the mean quality is μ > λ , and the average quality of col-

−

+ gt x t (Ccs + Crw ) τ1 + gt x t

2 qrw ) τ1+

2 gt x t (Ccs + Cis + Crm ) τ2 + gt x t Cq (qfr2 − qrm ) τ2 + Wτ22 + Cq 2 qfm (x t − gt x t (τ1 + τ2))

and the probability that a product can 367

(6)

International Journal of Production Economics 219 (2020) 360–374

M. Ullah and B. Sarkar

To develop an RFID based collection system, it is critical to determine the number of RFID readers that are deployed in the search area in order to achieve full coverage. On the one hand, if we increase the number of readers, the cost might increase significantly. On the other hand, if we decrease the number of readers, we might not get full coverage over the entire search space. Therefore, we develop a mathematical model to determine the minimum number of RFID readers required to achieve full coverage. Furthermore, the model also considers the cost of the workforce involved in the operations of the proposed system. We considered two types of readers, 1) type-1 readers with higher sensing power and, 2) type-2 readers with low sensing power. The entire search area is first divided into sub-areas by deploying type-1 readers. To achieve complete coverage, we considered the Disk Sensing Model for type-1 reader deployment. The area covered by each type-1 reader is further divided by placing type-2 readers inside the coverage area. We used the Probabilistic (exponential) Coverage Protocol to minimize the number of type-2 readers. The probabilistic model assumes that the sensing range of a sensor reduces exponentially with distance, while the Disk Sensing Model considers the sensing region as a disk with a certain radius. The probabilistic model is more cost efficient than the Desk Sensing Model. However, we used Desk Sensing model for type-1 reader deployment to reduce the chances of incorrect detection and to increase the sensing power of the system. Readers are advised to consult Hefeeda and Ahmadi (2007) and Zhang and Hou (2005) to understand these two models in detail. The number of type-1 readers can be determined by achieving complete coverage over the entire search area. According to the disk sensing model, coverage implies connectivity when 2 Ss≤ St . We used Fig. 6 to achieve complete coverage. It is clear from the figure that a complete coverage is achieved if Ss = 2 St . 2

Fig. 7. Different possible division of search area into smaller areas, (a) possible division of search area into smaller areas with type-1 readers (b) a single type-1 reader coverage area division through type-2 readers.

is the maximum sensing threshold below which the reader can sense tag with probability 1. Each sub-search area is again divided into sub-sub4 S2

areas by placing ( 2 s ) type-2 readers, and rs is the maximum sensing d threshold below which the reader can sense tag with probability 1, as shown in Fig. 6. The last term in (7) is the cost of the workforce involved in the tracking and retrieval process, which is based on the total number of workers required for the total search area. The first constraint shows that the maximum distance between any two consecutive Log [1 − 3 1 − Ω ]

), this is the small readers must be smaller than 3 (rs − α maximum separation distance between two sensor based on Probabilistic Coverage Protocol designed by Hefeeda and Ahmadi (2007). The second constraint is based on Disk Sensing model, developed by (Zhang and Hou, 2005), and shows that for type-1 readers, the sensing radius must be at least half of the transmission radius, which ensures the connectivity of the system. Fig. 7 shows different possible divisions of the search area with type-1 and type-2 readers, ensuring connectivity and complete coverage. The total cost of the production system with the investment cost of the proposed RFID based system can be determined by putting (7) in (6) and is given as,

2

As a2 + b2 = c 2 ⇒ a2 + b2 = 4 St2 ⇒ a + b = St ⇒ a = b = 2 St . 4 The total area is divided into n squares based on the transmission radius (St ) of type-1 readers as shown in Fig. 6. One reader is deployed for each square.

Z = f (d ) = c1 (

l w )( 2 S ) 2 St t

+ c2 (

4 Ss2 d2

)(

l w )( 2 S ) 2 St t

l

w

+ CL ( nL* 1000 * 1000 ) (7)

Subject to

3 (rs −

Log [1 − 3 1 − Ω ] )≥d α

TC = (c1 (

St ≥ 2 Ss

l w )( 2 S ) 2 St t l

w

4 Ss2 d2

)(

l w )( 2 S )) 2 St t maq e−λλk ) gt x t τ1 k!

τ12 + CL ( nL* 1000 * 1000 ) + Cds (∑k = 0

The formulation in (7) assumes that the total search area is divided l w into sub-search areas by placing n = ( 2 S )( 2 S ) sensing reader, and Ss t

+ c2 (

2 + gt x t (Ccs + Crw ) τ1 + gt x t Cq (qfr2 − qrw ) τ1 + gt x t (Ccs + Cis + Crm ) τ2 + gt x t

t

2 Cq (qfr2 − qrm ) τ2 + Wτ22 2 + Cq qfm (x t − gt x t (τ1 + τ2))

(8) Subject to

d−

3 (rs −

Log [1 − 3 1 − Ω ] )≤0 α

(τ1 + τ2 − 1) ≤ 0

(9) (10)

5.3. Solution methodology The optimization problem is to minimize total cost (8) subject to constraint (9) and (10). As the model itself and constraint (10) are both nonlinear, therefore, this paper uses the Karush-Kuhn-Tucker (KKT) method to solve the problem. The Lagrange function for the proposed problem can be written as,

Fig. 6. Reader deployment: (a) type-1 reader deployment based on Desk Sensing Model, (b) determination of transmission radius of type-1 readers, (c) type-2 reader deployment based on exponential sensing model. 368

International Journal of Production Economics 219 (2020) 360–374

M. Ullah and B. Sarkar

l w )( 2 S ) 2 St t

L = (c1 (

+ c2 (

4 Ss2 d2

)(

l w )( 2 S )) τ12 2 St t

+ CL (nL*

l*w ) 10002

Step 3. Utilize d* and τ2* to determine η2* from Eq. (22).

+ Cds

Step 4. Find τ1* from Eq. (18) utilizing d* and η2*.

maq e−λλk ) gt x t τ1 k!

(∑k = 0

Step 5. Finally, determine η1* from Eq. (21) utilizing τ1*.

2 + gt x t (Ccs + Crw ) τ1 + gt x t Cq (qfr2 − qrw ) τ1 + gt x t (Ccs + Cis + Crm) τ2 + gt x t 2 2 Cq (qfr2 − qrm ) τ2 + Wτ22 + Cq qfm

(x t − gt x t (τ1 + τ2)) + η1 (d −

3 (rs −

Log [1 − 3 1 − Ω ] ) α

5.4.1. Numerical example This section provides a numerical example with three different cases to illustrate model applications and obtain results based on different scenarios. Case 1 is the proposed model, developed and solved in the previous section, while Case 2 is a variant of Case 1 with one additional assumption that the proposed RFID based system is outsourced without the required investment. Case 3 only studies the cost of RFID based system with objective function f (d ) given in Eq. (7). All the three cases are solved with one date set. The values of input parameters are:

)+

η2 (τ1 + τ2 − 1) (11) Optimal values of the variables can be determined by applying KKT conditions. Note that TC and constraints are convex, and TC is twice differentiable with respect to τ1, τ2 , and d . The KKT conditions from (11) can be found as follows: ∂L (τ 1, τ2, d, η) ∂τ 1

Crw = 3, Ccs = 1.9, Cis = 2.8, Cq = 135, qfr = 0.95, qfm = 0.98, Crm = 33, qrm = 0.5, c1 = 300, l = 300, w = 300, c2 = 200, rs = 12, Ss = 50, α = 0.05, Ω = 0.999, St = 2Ss, Cds = 8, λ = 35, μ = 45, W = 1000, gt = 0.5, nL = 6, CL = 50, x 0 = 10000, maq = 30,

= RQx t Cds − q2x t Cq + x t (Ccs + Cis + Crw ) + x t 2 Cq (q2 − qrw ) + η2 8(

+ ∂L (τ 1, τ2, d, η) ∂τ2

l w )( ) c 2 Ss2 τ 1 2 St 2 St d2

=0

(12)

which are obtained from Geyer and Blass (2010) and Guide and Van Wassenhove (2001).

2 = −q2x t Cq + x t (Ccs + Crm ) + x t Cq (q2 − qrm ) + η2 + 2W

Case 1. For the values given above, the model is solved using Wolfram Mathematica 11.2 optimization software package on a desktop computer with Intel(R) Core(TM)i5-7500 [email protected] and 16.00 GB of RAM. Computational time was 0.045 s and the optimal results obtained are d = 24.43, τ1 = 0.85 and τ2 = 0.15. Total cost of the system is $1169794. In this case, the total realized demand is the sum of the demand for remanufactured products and new products. Total demand for remanufacturing is x t gt , which is further divided into two parts. The results suggest that a mixed recovery policy is the optimal one, in which 85% demand (for remanufactured products) is satisfied from channel-1 and 15% demand is satisfied from the market-driven remanufacturing channel. Furthermore, x m, t quantity is satisfied from manufacturing new parts.

τ2 = 0 (13) l w 8( )( ) c 2 Ss2 τ12 2 St 2 St d3

∂L (τ 1, τ2, d, η) ∂d

= η1 −

∂L (τ 1, τ2, d, η) ∂η1

=d−

∂L (τ 1, τ2, d, η) ∂η2

= −1 + τ1 + τ2

η1 (d −

3 (rs −

3 (−

=0

Log[1 − (1 − Ω)1/3] α

(Log[1 − 3 1 − Ω ]) )) α

+ rs ) = 0

(14) (15) (16)

=0

(17)

η2 (τ1 + τ2 − 1) 5.4.2. Impact of demand rate Impacts of demand for remanufactured products on channel selection is shown in Fig. 8. The relation shows that higher demand for remanufactured products favors the proposed RFID based channel, while low demand prefers traditional market-driven system. Therefore, supply chain managers of products with higher demand for remanufacturing are suggested to implement this system. This means, initially when the demand for a product is higher, a hybrid recovery strategy is an optimal policy. While, during last periods of the planning horizon, pure recovery policy, from the market-driven system, is the optimal policy. At the end of the planning horizon, when the demand for a particular model is too low, remanufacturing must be avoided and all collections must be recycled instead.

and η ≥ 0 Solving the above equations simultaneously, we get the optimal KKT points which are given below,

τ1* =

d2 (gt xt Ccs + gt (∑kmaq =0

e−λλk 2 + g x C q2 − g x C q2 + η ) ) xt Cds + gt xt Crw − gt xt Cq qfm 2 t t q fr t t q rw k! l w 2 8 c 2 Ss 2 St 2 St

(18)

τ2* = 1 +

d* =

e−λλk 2 2 d2 (−2W + gt xt ( (∑kmaq = 0 k ! ) Cds + Cis − Crm + Crw + Cq (qrm − qrw ))) l w 2 2 2d W + 8 c 2 Ss 2 St 2 St

3 (−Log[1 − (1 − Ω)1/3] + αrs ) α

(20)

w c 2 Ss2 τ12 2 St − 3 3 (Log[1 − (1 − Ω)1/3] − αrs )3

maq e−λλk ) Cds k!

η2* = −gt x t (Ccs + (∑k = 0 +

l 8 2 St

Case 2. In this case, we analyze the effects of the investment required for the proposed RFID system. We assume that the manufacturer does not need to invest in the proposed RFID based system. The objective function is modified for this scenario. The model is solved without RFID based investment and the optimal results show that d decreases to 12.21 and τ1 increases to 1. This means that the proposed system cost is one of the most important parameters. However, as shown in Fig. 8, the impact of cost is reduced with the increasing demand for remanufacturing products. This will also help organizations to explore new markets, thus expanding its market shares and improve profitability. Furthermore, if an organization can manage to reduce the cost of RFID based system, the proposed RFID based channel is beneficial even with a low demand for remanufactured products. The cost of RFID technology is rapidly decreasing and RFID based return channel is the future of reverse logistics. This assumption is also valid in

l * 2 St

8α3

η1* =

(19)

(21) 2 2 + Cis + Crw − Cq (qfm − qfr2 + qrw ))

w c 2 Ss2 (−1 + τ2) 2 St d2

(22) 5.4. Algorithm Step 1. Substitute parametric values in Eq. (20) and find d*. Step 2. Utilize d* and determine τ2* from Eq. (19). 369

International Journal of Production Economics 219 (2020) 360–374

Remanufacturing rate

M. Ullah and B. Sarkar

1 0.8 0.6 0.4 0.2 0 0

200

500

1000

2000

4000

6000

8000

Demand for remanufactured products (xr) (units per period)

Fig. 8. Impacts of demand rate per time period on return rate percentage of channel-1 and channel-2.

which remains unaffected to further decrease in maq . Fig. 9 shows the impact of maq on channel selection. The effect of both λ and μ is obvious, as increasing any one of these increases the return rate from the same channel, and a 25% decrease brings the return rate of the channel to zero. Another important parameter is the quality upgradation cost. Interestingly, decreasing quality cost Cq favors the return rate from channel-1. A 25% decrease in Cq increases the return rate of channel-1 to 100% while the return rate from channel-2 is reduced to zero percent. On the other hand, increasing quality upgradation cost by 50% increase the return rate of channel-2 to 88%. Here we get an interesting insight for managers, that is, the high quality upgradation cost (which in actual is production cost) favors the traditional market-driven system because of its high quality. Therefore, economically speaking, for expensive products, channel-2 is a favorable strategy, while for cheaper products, channel-1 is an economic option. However, previously we found that demand inversely affects the return rate of traditional market-driven channels, hence, supply chain managers must keep both factors in consideration during decisions making.

a situation where the RFID based system is outsourced. In this case, the outsourcer provides the tracking system to more than one manufacturer, and thus, the cost is shared among different organizations. However, total cost of the system in this case is $1137693, which is just 0.027% less than the original model, and compared to environmental benefits, this increase in total cost is nothing. The distance between two readers d reduces to half of the previous case. This validates our modeling approach; that without the investment cost the model opted for a maximum detection capability. Case 3. In this case, we study the impacts of search area dimension on the number of required readers, the maximum distance between two readers, and the required investment (Z = f (d )). The objective function is modified again and only costs related to RFID based system are considered as given by EQ. (7). Results are compiled in Table 3 which show that the number of type-1 readers highly relies on the search area, and it linearly increases with expansion in the search area. However, the number of type-2 readers/type-1 reader remains the same. The distance between two type-2 readers is not influenced by the total search area. However, it does change with changes in the transmission radius of type-1 readers. This means that the transmission radius of type-1 readers should be an integral multiple of the sensing radius of type-2 readers.

5.6. Impacts of important parameters on the RFID system In this section, we study the effect of changes in input parameters on the proposed RFID system. From the results, the effect of different parameters on the required investment and decision variables is studied. We observed that Ss and rs are the most important parameters followed by c1 and c2 , while α showed a negligible effect on decision variables and cost. The cost of type-1 reader (c1) and type-2 reader (c2 ) have no effect on decision variable, although required investment (Z) has a linear relation with them. This is because the decision variables depend upon sensing power. Because the sensing power is constant, therefore, no matter how much you increase the price of one reader, it will not affect the number of readers; however, the value of Z increases. This provides an important managerial insight and suggests that the designer of this type of system should focus on the sensing radius and the power of the readers. The cost of the reader is not the most important factor in this model. The sensing power rs of type-2 readers is the most important parameter to decide the distance between two nodes in the network of type2 readers. A decrease in the sensing range rs decreases the distance d * and increases the number of readers, therefore, the investment cost is increased. But this increase is not linear; rather it shows an exponential increase. Hence, an increase in sensing power can decrease the number of deployed sensors, and the investment is reduced. As the relation is exponential, sensors of higher power decrease the number of deployed sensors, and the investment cost does not increase linearly. The model behavior observed with changes in the sensing power of type-1 readers Ss is interesting. Sensing power first decreased the total cost of the system, but then the total cost increased as shown in Fig. 10.

5.5. Impacts of important parameters on the return rate channel selection We study the effect of average quality of returned and finished products on channel selection. The results show that finished product quality has no impact on the values of optimal return rate from each channel, although, the total cost of the system reduces drastically. Reducing finished product quality by 50% drops the total cost to $225287 , while τ1 and τ2 remains the same. Unlike finished product quality, the minimum acceptable quality (maq) of return directly influences the channel selection decisions. Reducing (maq) by 25% increase return rate of channel-1 from 85% to 98%. A further decrease in (maq) does not affect the optimal values. Similarly, increasing (maq) by 50% increase return rate of channel-2 to 59%. At 100% increase, the return rate of channel 1 drops to 39%, Table 3 Optimal number of readers and required investment for RFID based recovery channel. l (m)

300 600 900

w (m)

300 600 900

d * (m)

24.43 24.43 24.43

Number of type-1 readers

Number of type-2 readers

y1*

y2*

Total investment required for RFID based system Z = f (d) ($)

9 25 49

153 425 833

32848 91246 178843

370

International Journal of Production Economics 219 (2020) 360–374

M. Ullah and B. Sarkar

1.2

Remanufacturing rate

0.98

0.98

0.92

1 0.8 0.6

0.58

0.59

0.61

0.61

0.42

0.41

0.39

0.39

0.37

0.45

0.6

0.8

0.4 0.2

0.02

0.02

0.1

0.2

0.08

0 0.27

Manimum acceptable quality for channel 1

Fig. 9. Impacts of minimum acceptable quality on the return channel selection.

return rate and increase recycling of cell phones (Ongondo and Williams, 2009; DTSC, 2006; Ministry of Environment, 2016). All this money can be directed to implement RFID systems in this industry. Third, this system requires a one-time investment and tags retrieved with the recovered phones can be reused. Furthermore, the investment that is recouped from recycling can also be used to fund the use of RFIDs in this sector. Additionally, for a firm with an annual supply chain cost of $1.1 million, the increase in total cost is just 0.02%, which is negligible compared to the potential benefits of the system. For instance, apart from recycling, the revenue and market share of organizations can also be increased by selling the recovered phones in the second-hand market. Aside from the obvious environmental and economic benefits, the proposed system can also provide a competitive edge to the manufacturer by providing real-time information about product functionality. Delving deeper into the functionality issue, which is primarily related to the quality and performance of the product, but can also be solved using the proposed RFID based system. At the moment, there is no such technique available; and manufacturers remain unaware of the real-time status of cell phones after they are sold out. For example, RFID may help the manufacturers to handle false error reports regarding malfunctions in the product, as reported in Samsung Galaxy Note 7.

5.7. Cost justifications, associated benefits This section presents cost justifications and benefits, other than those stated above, associated with the proposed system. Moreover, it also gives significant barriers to the implementation of the system. After a comprehensive study of the literature, we concluded that cost should not be a hurdle in implementing this system for three main reasons. First, no amount of money is bigger than the fatal environmental and health risks associated with the improper disposal of cell phones. Second, all stakeholders including governments, private NGOs, customers, and manufacturers are supporting projects to improve the

Total investment

Toatl cost of RFID system

60000

Sensing redius

52513

54797

50000

44614 90

38495

40000 30000

80

32848

30095 24095

20000

21995

50

70 60 20495

35502

100

120 100 80 60

27462 40

40 30

10000 10

15

20

20

0

0

Fig. 10. Investment cost (Z) relation with type-1 reader sensing radius. 371

Sensing redius of type 1 reader

This provides an important managerial insight for this model. The designer should be aware of the relationship between type-1 reader sensing radius and the total investment required for the system. This is because the internode distance of type-2 readers remains the same and increasing the radius of type-1 readers would increase the number of deployed type-2 readers, thus increasing the total cost of the system. This extends the sensing region beyond the search area. Therefore, the supply chain managers must decide both Ss and rs in connection with each other, as they are interdependent and cannot be decided individually.

International Journal of Production Economics 219 (2020) 360–374

M. Ullah and B. Sarkar

6. Conclusions

After its initial release, Samsung Galaxy Note 7 was reported to have a problem that caused its batteries to explode during charging. Many consumers, during this time, however, falsely reported that their phones caught fire. These reports were devastating for the manufacturer; and Samsung, a big name in the smartphone industry, lost its credibility. Consequently, many airlines throughout the world banned Note 7 onboard. The proposed RFID based system can also help the manufacturer to collect real-time data about product malfunction and avoid such false error reporting from unreliable consumers. Therefore, there is an immediate demand for a system that can solve all such problems with a high level of automation. The proposed system, apart from solving the problems, is likely to provide other value-added benefits in the cell phone supply chain as well. These include theft and loss prevention, streamlined inventories, reduced turnaround time, and avoidance of unnecessary handling. Furthermore, a sensor embedded device using RFID can actively measure the estimated remaining lifetime of a product, thus enabling the manufacturer to have real-time information about product failure and the reasons behind the failure, which results in quality improvements (Ilgin and Gupta, 2010).

The literature review showed a very low recycling rate in the cell phone supply chain due to two main reasons, 1) discarding phones into municipal waste and 2) hibernation period. This paper concentrated on possible solutions to the above mentioned two problems and introduced a novel smart recovery system that minimized the hibernation period and facilitated the separation of discarded cell phones from municipal waste. A dual-channel hybrid manufacturing-remanufacturing production model was developed with different quality of used products. Production coordination and channel selection were studied in the proposed model. Channel-1 was the proposed RFID based recovery channel, while channel-2 denoted the traditional recovery channel that relied upon consumers' willingness to return the product. The manufacturers, in the traditional recovery channel, use financial and other incentives to encourage consumers to return the EOL/EOU products. The proposed recovery channel (channel-1), on the other hand, utilizes an RFID system to retrieve EOL/EOU products from waste streams and hibernation. Previously, hundreds of different collection systems operated around the world to recover obsolete phones. But they all failed to produce results. The main reason for this failure was the exogenous nature of return rate in the existing systems. In all these systems, the return rate is out of the direct control of the collector and solely rely upon consumers’ will. The proposed system utilized IoT (viz. RFID tags, readers, and sensors) to bring the return rate of EOL/EOU cell phones under collector control. Return rate in the cell phone supply chain can, therefore, reach up to 100% by using this system, and each phone that is out of use can be tracked, recovered and recycled. A tracking system was designed to increase the return rate of the hibernated cell phones from storage, and the system focused on determining the locations of hibernated cell phones and retrieval from storage. The designed system jointly exploits the traditional indoor localization techniques and the tools offered by advanced cell phones to assist in the tracking of hibernated cell phones. A separation system was also designed to separate obsolete cell phones from municipal waste. Additionally, this research simultaneously considered the cost of implementation along with the design of the system. Therefore, a mathematical model was developed to minimize the number of required readers for the tracking system, thus reducing the total cost of the designed system. The mathematical model was developed based on the Disk Sensing Model and the Exponential Sensing model to ensure proper coverage, detection, and location. An indoor localization system LANDMARC was utilized to find the coordinates of the tracking cell phone. The model was solved in three different cases and important results were obtained. The results from Case 1 showed that a hybrid collection strategy with 85% share of channel-1 and 15% of channel-2 is the optimal one. The results further showed that high demand for remanufactured products promoted the proposed RFID based recovery channel. The investment cost of the system was found one of the most important factors in channel selection, however, the impact of investment on total cost was very low. For low demand, outsourcing of RFID was found better than self-investment. However, with high demand, the impacts of investment are negligible. Channel selection is largely governed by the average quality of the products, and low minimum acceptable quality promoted higher return rate from channel-1. The results also suggested that, while designing an RFID based system, reader sensing power is more important than the cost of readers. This research can be extended in a number of ways, the first and immediate extension is considering random demand. However, in this case, the complexity of the system would increase manifold. Considering green image of the manufacturer is another extension that can produce interesting results, as a firm can get a competitive advantage over its competitors by producing green products that are 100% recyclable. Incorporation of carbon dioxide emissions from collections od used is also an important extensions of this papers, an excellent example of carbon emissions in logistics model is given by

5.8. Technical challenges and barriers to implementation of the system There are three main technical challenges that cell phone remanufacturing sector is facing. First one is the high rate of innovation in this industry, which decreases the lifespan of technologies. Obsolescence results in a rapid decline in the price of current technology, reducing the profitability of cell phone remanufacturing. However, the adverse impacts of this problem can be minimized with fast acquisition, remanufacturing, and selling of used cell phones, which was not possible with the previous collection methods. The proposed recovery system can, however, increase the speed of the recovery process, making it possible to minimize the adverse impacts. Moreover, selling the remanufactured products in developing markets, where the technology is still acceptable, can also reduce the severity of this problem. Once the technology is completely outdated, the collected cell phones must then be recycled to extract raw material. The second problem is the requirement of complex remanufacturing techniques and the sound knowledge of how to dismantle these devices? Presently, this is the most crucial factors for recyclers, because there are various models, and every model has different parts and structures. For instance, the European Telecommunication Standards Institute had registered more than 1800 different models of cell phones until 2003 (Manivannan, 2016). Which, we believe, have been increased many times by now. Thus, the provision of appropriate dismantlement information is crucial for every model. This can be attained with the use of RFID tags because tag memory can store enough information about phone dismantlement procedures and remanufacturing techniques. Furthermore, this also eliminates the need for dismantling manuals, which consume a large quantity of paper. Finally, the absence of single global-standard technology limits the number of potential secondary markets. For example, some regions use Global System for Mobile Communications (GSM) standards while other use Code Division Multiple Access (CDMA). Although, most of the recent phones come with more than one technologies, yet the remanufacturer must carefully decide the potential secondary market. An important problem with the implementation of the system is the working environment. Tag performance is affected by metallic objects and wet surfaces because electromagnetic waves cannot penetrate metal and wet surfaces well. This issue is important in tracking the hibernated cell phones and separating obsolete phones from municipal waste, as both metallic objects and wet surfaces can reduce the efficiency of the system. This makes the use of some RFID systems, particularly Class 1 and 2, unreliable in such environments.

372

International Journal of Production Economics 219 (2020) 360–374

M. Ullah and B. Sarkar

Sarkar et al. (2019).

review. Resour. Conserv. Recycl. 55 (1), 11–19. Kang, Y., Gershwin, S.B., 2005. Information inaccuracy in inventory systems: stock loss and stock out. IIE Trans. 37 (9), 843–859. Khetriwal, D.S., Kraeuchi, P., Widmer, R., 2009. Producer responsibility for e-waste management: key issues for consideration–learning from the Swiss experience. J. Environ. Manag. 90 (1), 153–165. Kim, T., Glock, C.H., 2014. On the use of RFID in the management of reusable containers in closed-loop supply chains under stochastic container return quantities. Transp. Res. E Logist. Transp. Rev. 64, 12–27. Kiritsis, D., Jun, H.B., Xirouchakis, P., 2008. In: Miles, S.B., Sarma, S.E., Williams, J.R. (Eds.), Closing Product Information Loops with Product-Embedded Information Devices: RFID Technology and Applications, Models and Metrics. RFID technology and applications, Cambridge University Press, New York, pp. 169–182 2008. Korugan, A., Dingeç, K.D., Onen, T., Ateş, N.Y., 2013. On the quality variation impact of returns in remanufacturing. Comput. Ind. Eng. 64 (4), 929–936. Li, B., Yang, J., Lu, B., Song, X., 2015a. Estimation of retired mobile phones generation in China: a comparative study on methodology. Waste Manag. 35, 247–254. Li, X., Li, Y., Cai, X., 2015b. Remanufacturing and pricing decisions with random yield and random demand. Comput. Oper. Res. 54, 195–203. Luttropp, C., Johansson, J., 2010. Improved recycling with life cycle information tagged to the product. J. Clean. Prod. 18 (4), 346–354. Maiti, T., Giri, B.C., 2015. A closed loop supply chain under retail price and product quality dependent demand. J. Manuf. Syst. 37, 624–637. Maiti, T., Giri, B.C., 2017. Two-way product recovery in a closed-loop supply chain with variable markup under price and quality dependent demand. Int. J. Prod. Econ. 183, 259–272. Manivannan, S.V., 2016. Environmental and health aspects of mobile phone production and use: suggestions for innovation and policy. Environmental Innovation and Societal Transitions 21, 69–79. Mason, A., Shaw, A., Al-Shamma’a, A., 2012. Peer-to-peer inventory management of returnable transport items: a design science approach. Comput. Ind. 63 (3), 265–274. Milanez, B., Bhrs, T., 2009. Extended producer responsibility in Brazil: the case of tyre waste. J. Clean. Prod. 17 (6), 608–615. Ministry of Environment. Waste Recycling, Extended Producer Responsibility (EPR). 2016, Available at: http://eng.me.go.kr/eng/web/index.do?menuId=372 Accessed on: 01/10/2016. Mobile cellular subscriptions, Population, total. 2018, Available at: http://data. worldbank.org/Accessed on 01/03/2018. Moshtagh, M.S., Taleizadeh, A.A., 2017. Stochastic integrated manufacturing and remanufacturing model with shortage, rework and quality based return rate in a closed loop supply chain. J. Clean. Prod. 141, 1548–1573. Nahmiasj, S., Rivera, H., 1979. A deterministic model for a repairable item inventory system with a finite repair rate. Int. J. Prod. Res. 17 (3), 215–221. Ni, L.M., Liu, Y., Lau, Y.C., Patil, A.P., 2004. LANDMARC: indoor location sensing using active RFID. Wirel. Netw. 10 (6), 701–710. Nnorom, I.C., Osibanjo, O., 2008. Overview of electronic waste (e-waste) management practices and legislations, and their poor applications in the developing countries. Resour. Conserv. Recycl. 52 (6), 843–858. O'Connell, M., Hickey, S., Besiou, M., Fitzpatrick, C., Van Wassenhove, L.N., 2013. Feasibility of using radio frequency identification to facilitate individual producer responsibility for waste electrical and electronic equipment. J. Ind. Ecol. 17 (2), 213–223. Ondemir, O., Gupta, S.M., 2014. Quality management in product recovery using the Internet of Things: an optimization approach. Comput. Ind. 65 (3), 491–504. Ongondo, F., Williams, I.D., 2009. Mobile telephone collection, reuse and recycling in the UK. In: 3rd BOKU Waste Conference 2009, pp. 175–184 Austria. 15 - 17 Apr 2009. Panda, S., Modak, N.M., Cárdenas-Barrón, L.E., 2017. Coordinating a socially responsible closed-loop supply chain with product recycling. Int. J. Prod. Econ. 188, 11–21. Richter, K., 1996a. The eoq repair and waste disposal model with variable setup numbers. Eur. J. Oper. Res. 95 (2), 313–324. Richter, K., 1996b. The extended eoq repair and waste disposal model. Int. J. Prod. Econ. 45 (1–3), 443–447. Richter, K., 1997. Pure and mixed strategies for the eoq repair and waste disposal problem. Operations-Research-Spektrum 19 (2), 123–129. Richter, K., Dobos, I., 1999. Analysis of the EOQ repair and waste disposal problem with integer setup numbers. Int. J. Prod. Econ. 59 (1–3), 463–467. Sarkar, B., 2019. Mathematical and analytical approach for the management of defective items in a multi-stage production system. J. Clean. Prod. 218, 896–919. Sarkar, B., Ullah, M., Kim, N., 2017. Environmental and economic assessment of closedloop supply chain with remanufacturing and returnable transport items. Comput. Ind. Eng. 111, 148–163. Sarkar, B., Majumder, A., Sarkar, M., Kim, N., Ullah, M., 2018. Effects of variable production rate on quality of products in a single-vendor multi-buyer supply chain management. Int. J. Adv. Manuf. Technol. 99 (1–4), 567–581. Sarkar, B., Guchhait, R., Sarkar, M., Cárdenas-Barrón, L.E., 2019. How does an industry manage the optimum cash flow within a smart production system with the carbon footprint and carbon emission under logistics framework? Int. J. Prod. Econ. 213, 243–257. Schrady, D.A., 1967. A deterministic inventory model for reparable items. Nav. Res. Logist. Q. 14 (3), 391–398. Series, P., 2012. Propagation Data and Prediction Methods for the Planning of Indoor Radio Communication Systems and Radio Local Area Networks in the Frequency Range 900 MHz to 100 GHz. Recommendation ITU-R 1238-7. Silveira, G.T., Chang, S.Y., 2010. Cell phone recycling experiences in the United States and potential recycling options in Brazil. Waste Manag. 30 (11), 2278–2291. Sullivan, D.E., 2006. US Geological Survey Fact Sheet 2006-3097. US Geol. Surv., Denver,

Acknowledgements This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (Project Number: 2017R1D1A1B03033846). References Ahmed, W., Sarkar, B., Ullah, M., 2018. Impact of reparation for imperfect quality items having shortages in the system under multi-trade-credit-period. DJ Journal of Engineering and Applied Mathematics 5 (1), 1–6. Aras, N., Verter, V., Boyaci, T., 2006. Coordination and priority decisions in hybrid manufacturing/remanufacturing systems. Prod. Oper. Manag. 15 (4), 528–543. California Department of Toxics Substances Control (DTSC), 2006. New law for recycling cellular phones fact sheet. Available at: https://www.dtsc.ca.gov/HazardousWaste/ EWaste/upload/HWMP_FS_AB2901-2.pdf, Accessed date: 13 August 2018. Chan, F.T.S., Kai Chan, H., 2008. A survey on reverse logistics system of mobile phone industry in Hong Kong. Manag. Decis. 46 (5), 702–708. Cui, L., Deng, J., Liu, F., Zhang, Y., Xu, M., 2017. Investigation of RFID investment in a single retailer two-supplier supply chain with random demand to decrease inventory inaccuracy. J. Clean. Prod. 142, 2028–2044. Curran, T., Williams, I.D., 2012. A zero waste vision for industrial networks in Europe. J. Hazard Mater. 207, 3–7. Dey, B.K., Sarkar, B., Pareek, S., 2019. A two-echelon supply chain management with setup time and cost reduction, quality improvement and variable production rate. Mathematics 7 (4), 328. Diallo, C., Venkatadri, U., Khatab, A., Bhakthavatchalam, S., 2017. State of the art review of quality, reliability and maintenance issues in closed-loop supply chains with remanufacturing. Int. J. Prod. Res. 55 (5), 1277–1296. Diekmann, T., Melski, A., Schumann, M., 2007. Analysing Impacts of RFID in Supply Chains Using Joint Economic Lot Size Models. In ECIS. pp. 1839–1850. Dobos, I., Richter, K., 2004. An extended production/recycling model with stationary demand and return rates. Int. J. Prod. Econ. 90 (3), 311–323. Dobos, I., Richter, K., 2006. A production/recycling model with quality consideration. Int. J. Prod. Econ. 104 (2), 571–579. El Saadany, A.M., Jaber, M.Y., 2010. A production/remanufacturing inventory model with price and quality dependant return rate. Comput. Ind. Eng. 58 (3), 352–362. Engels, D.W., Sarma, S.E., 2005. Standardization Requirements within the RFID Class Structure Framework. Auto-ID Labs, Massachusetts Institute of Technology, Cambridge, MA USA (Tech. Rep). Fan, T., Tao, F., Deng, S., Li, S., 2015. Impact of RFID technology on supply chain decisions with inventory inaccuracies. Int. J. Prod. Econ. 159, 117–125. Fang, C., Liu, X., Pardalos, P.M., Pei, J., 2016. Optimization for a three-stage production system in the Internet of Things: procurement, production and product recovery, and acquisition. Int. J. Adv. Manuf. Technol. 83 (5–8), 689–710. Fu, J., Zhou, Q., Liu, J., Liu, W., Wang, T., Zhang, Q., Jiang, G., 2008. High levels of heavy metals in rice (Oryza sativa L.) from a typical E-waste recycling area in southeast China and its potential risk to human health. Chemosphere 71 (7), 1269–1275. Geyer, R., Blass, V.D., 2010. The economics of cell phone reuse and recycling. Int. J. Adv. Manuf. Technol. 47 (5–8), 515–525. Giri, B., Sharma, S., 2016. Optimal production policy for a closed-loop hybrid system with uncertain demand and return under supply disruption. J. Clean. Prod. 112, 2015–2028. Govindan, K., Soleimani, H., 2017. A review of reverse logistics and closed-loop supply chains: a journal of cleaner production focus. J. Clean. Prod. 142, 371–384. Guide Jr., V.D.R., Van Wassenhove, L.N., 2001. Managing product returns for remanufacturing. Prod. Oper. Manag. 10 (2), 142–155. Guo, J., Ya, G., 2015. Optimal strategies for manufacturing/remanufacturing system with the consideration of recycled products. Comput. Ind. Eng. 89, 226–234. Habib, M.S., Sarkar, B., Tayyab, M., Saleem, M.W., Hussain, A., Ullah, M., et al., 2019. Large-scale disaster waste management under uncertain environment. J. Clean. Prod. 212, 200–222. Hefeeda, M., Ahmadi, H., 2007. A probabilistic coverage protocol for wireless sensor networks. In: 2007 IEEE International Conference on Network Protocols. IEEE, pp. 41–50. Heydari, J., Govindan, K., Sadeghi, R., 2018. Reverse supply chain coordination under stochastic remanufacturing capacity. Int. J. Prod. Econ. 202, 1–11. Huisman, J., 2003. The QWERTY/EE Concept, Quantifying Recyclability and EcoEfficiency for End-Of-Life Treatment of Consumer Electronic Products. TU Delft, Delft University of Technology. Hwang, H., Ko, Y.D., Yune, S.H., Ko, C.S., 2009. A closed-loop recycling system with a minimum allowed quality level on returned products. Int. J. Serv. Oper. Manag. 5 (6), 758–773. Ilgin, M.A., Gupta, S.M., 2010. Comparison of economic benefits of sensor embedded products and conventional products in a multi-product disassembly line. Comput. Ind. Eng. 59 (4), 748–763. Iqbal, M.W., Sarkar, B., 2019. Recycling of lifetime dependent deteriorated products through different supply chains. RAIRO Oper. Res. 53 (1), 129–156. Jang, Y.C., Kim, M., 2010. Management of used & end-of-life mobile phones in Korea: a

373

International Journal of Production Economics 219 (2020) 360–374

M. Ullah and B. Sarkar

potential of metallic elements in discarded electronics: a case study of mobile phones in China. J. Environ. Sci. 20 (11), 1403–1408. Xu, D.F., Li, Q., Jun, H.B., Browne, J., Chen, Y.L., Kiritsis, D., 2009. Modelling for product information tracking and feedback via wireless technology in closed-loop supply chains. Int. J. Comput. Integr. Manuf. 22 (7), 648–670. Xu, C., Zhang, W., He, W., Li, G., Huang, J., 2016. The situation of waste mobile phone management in developed countries and development status in China. Waste Manag. 58, 341–347. Yang, K., Forte, D., Tehranipoor, M., 2015. ReSC: RFID-enabled supply chain management and traceability for network devices. In: International Workshop on Radio Frequency Identification: Security and Privacy Issues. Springer, Cham, pp. 32–49. Zeng, A.Z., Hou, J., 2018. Procurement and coordination under imperfect quality and uncertain demand in reverse mobile phone supply chain. Int. J. Prod. Econ. 209, 346–359. Zhang, H., Hou, J.C., 2005. Maintaining sensing coverage and connectivity in large sensor networks. Ad Hoc Sens. Wirel. Netw. 1 (1–2), 89–124.

USA. Tian, X., Zhang, Z.-H., 2018. Capacitated disassembly scheduling and pricing of returned products with price dependent yield. Omega 84, 160–174. TYPES OF RFID SYSTEMS, 2018. Available at: https://www.impinj.com/about-rfid/ types-of-rfid-systems/, Accessed date: 3 January 2018. Ullah, M., Sarkar, B., 2018. Smart and sustainable supply chain management: a proposal to use rfid to improve electronic waste management. In: Proceedings of International Conference on Computers and Industrial Engineering. CIE 2018-December. Ullah, M., Sarkar, B., Asghar, I., 2019. Effects of preservation technology investment on waste generation in a two-echelon supply chain model. Mathematics 7 (2), 189. Welfens, M.J., Nordmann, J., Seibt, A., 2016. Drivers and barriers to return and recycling of mobile phones. Case studies of communication and collection campaigns. J. Clean. Prod. 132 (20), 108–121. Wilhelm, M., Hutchins, M., Mars, C., Benoit-Norris, C., 2015. An overview of social impacts and their corresponding improvement implications: a mobile phone case study. J. Clean. Prod. 102, 302–315. Wu, B.Y., Chan, Y.C., Middendorf, A., Gu, X., Zhong, H.W., 2008. Assessment of toxicity

374