- Email: [email protected]
End-of-life resource recovery from emerging electronic products – A case study of robotic vacuum cleaners
Journal of Cleaner Production 137 (2016) 652e666
Contents lists available at ScienceDirect
Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro
End-of-life resource recovery from emerging electronic products e A case study of robotic vacuum cleaners Keshav Parajuly*, Komal Habib, Ciprian Cimpan, Gang Liu, Henrik Wenzel SDU Life Cycle Engineering, Department of Chemical Engineering, Biotechnology and Environmental Technology, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark
a r t i c l e i n f o
a b s t r a c t
Article history: Received 15 March 2016 Received in revised form 22 July 2016 Accepted 22 July 2016 Available online 27 July 2016
Integrating product design with appropriate end-of-life (EoL) processing is widely recognized to have huge potentials in improving resource recovery from electronic products. In this study, we investigate both the product characteristics and EoL processing of robotic vacuum cleaner (RVC), as a case of emerging electronic product, in order to understand the recovery fate of different materials and its linkage to product design. Ten different brands of RVC were dismantled and their material composition and design profiles were studied. Another 125 RVCs (349 kg) were used for an experimental trial at a conventional ‘shred-and-separate’ type preprocessing plant in Denmark. A detailed material flow analysis was performed throughout the recycling chain. The results show a mismatch between product design and EoL processing, and the lack of practical implementation of ‘Design for EoL’ thinking. In the best-case scenario, only 47% of the total materials in RVCs are ultimately recycled. While this low material recovery is mainly due to the lower plastic recycling rate, other market realities and the complex material flows in the recycling chain also contribute to it. The study provides a robust methodological approach for assessing the EoL performance based on the knowledge of a product and its complex recycling chain. The lessons learned can be used to support both the design and EoL processing of products with similar features, which carry a high potential for resource recovery, especially at the initial stage of the recycling chain. © 2016 Elsevier Ltd. All rights reserved.
Keywords: Robotic vacuum cleaner Material flow analysis Material recovery Design for end-of-life Design for recycling WEEE
1. Introduction The lack of efficient collection and appropriate recycling infrastructure is one of the major challenges in achieving a ‘closed’ loop of materials from consumer goods, including electrical and electronic products (Graedel et al., 2011). Further, the complex design of modern products adds to the challenges in the end-of-life (EoL) treatment processes, as the recycling technologies are being outpaced by emerging composite and elementally diverse products, making the resource recovery process more and more difficult. Product design and EoL management have a significant impact on the resource recovery as well as the overall environmental impacts of a product (Li et al., 2015). Efforts have been made towards linking these two stages e the product design/inception and its EoL processing e by incorporating ‘design for EoL’ approach for better EoL performance of products (Lee et al., 2014). Nevertheless, there are still considerable opportunities, both technological and legislative,
* Corresponding author. E-mail address: [email protected] (K. Parajuly). http://dx.doi.org/10.1016/j.jclepro.2016.07.142 0959-6526/© 2016 Elsevier Ltd. All rights reserved.
for linking the producers and EoL managers to intensify this approach (Li et al., 2015; Mayers et al., 2011). A typical resource recovery chain consists of three main steps: collection, preprocessing, and end processing. Being the first treatment step, preprocessing has a large impact on the subsequent pathways and fate of materials and on the final resource recovery in the whole-system perspective. Manual or mechanical, preprocessing serves as the guide for material flows in the following treatment steps. It defines the effectiveness of material liberation, sorting, and diversion to the correct downstream processing path, which ultimately influences the overall recycling rate (Chancerel et al., 2009). The challenges in the recycling process e especially the preprocessing step e have been illustrated for several products, including case studies on recovery of valuable metals from desktop computers (Meskers et al., 2009; Wang et al., 2012), computer hard disk drives (HDDs) (Habib et al., 2015), and electrical and electronic components in vehicles (Widmer et al., 2015). A common conclusion of these case studies is that the existing preprocessing infrastructure is not sufficient, as it causes significant losses of valuable resources in the process. Manual dismantling and positive sorting of the valuable components like high-grade printed circuit boards (PCBs) and HDDs is
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
common for products such as computers. This step is usually more expensive, especially in developed countries, given the high labor cost and varying product types with unfamiliar product properties (Basdere and Seliger, 2003). As a rule of thumb, the EoL products with enough valuable materials to cover the cost of dismantling are processed manually for component recovery, while the rest is sent to generalized processing. Generalized mechanical shredding and separation is the most preferred way of handling low-grade waste electrical and electronic equipment (WEEE) such as household appliances category (Chancerel et al., 2011). As the technology advances, the ubiquitous use of electronics in our daily life has only been increasing. It has led to the rising use of products such as photovoltaic panels (Cucchiella et al., 2015), elec€ hler, 2013), wearable electronics such as smart tronic textiles (Ko watches, and unmanned aerial vehicles (drones) used for amateur photography. Examples of household appliances include smart weighing scales with communication capabilities and robotic vacuum cleaners (RVCs) that can operate autonomously. As products evolve, more household appliances are incorporating complex electronic components, such as PCBs, sensors, and display panels. With this changing product features and material composition comes the challenge of addressing them under the existing EoL management setups, which are evolving slower than the products. The treatment options relying on conventional physical processing technique cannot handle every product with the same efficiency of material recovery. The recovery of valuable metals from the high-grade electronic waste has been the major research focus, whereas household products get less attention. Moreover, the possibilities of improving the design of these emerging products to allow an efficient resource recovery remain largely untested. The EoL-targeted design improvements need to be based on the knowledge of the EoL fate of the product. In order to do so, it becomes crucial to understand the performance of such products in the existing recycling chain, and the role of product design in improving the overall resource recovery efficiency. In this study, we give a more nuanced and differentiated view on these aspects, including an elaboration on which materials and components in the product are influenced the most by lack of efficient preprocessing separation. We use RVC as a case to illustrate the fate of an emerging product in the existing recycling chain and identify the potential for the improvement in product design. We aim to answer three main questions: a) Where are the main material losses in the existing recycling chain? b) For which materials and components is the problem of inefficient separation in the preprocessing step most significant? c) What is the connection of these losses to the design features of the product and how it can be addressed? RVC illustrates the trend of programmable household appliance that requires minimal human involvement. One study reported almost a quadruple growth of RVC sales in Denmark between 2010 and 2014, with the total volume reaching from 12,500 units to 48,100 units (Euromonitor, 2015). Another study forecasts 3 million units of RVCs will be marketed globally in 2016, (Euromonitor, 2012), which would add 10.5 kilo tons into the WEEE stream at the end of their lives. Although this volume of EoL RVC is not of a huge significance on its own compared to total WEEE flow, it exemplifies the trend of emerging electronic appliances and serves as a representative case product. 2. Materials and methods The study consists of two main parts e the first part (product characterization) studies the product characteristics, while the
653
second part (EoL assessment) focuses on the performance of the case product in the existing recycling chain. The methodological approach is illustrated in Fig. 1 and is described in the following subsections. 2.1. Product characterization The purpose of this step was to understand the design features and the material composition of the case product. For that, ten RVCs of different brands were disassembled with the help of basic handheld tools such as screwdriver, pliers, and hammer. The different connection types and sequence of disassembly of different components were noted, which can be related to the ease of manual dismantling. An example of step-wise disassembly is illustrated in Appendix A. After the disassembly, the materials of each component were identified, and if required, the components were further dismantled to reach the material-level identification. Materials were identified using techniques including visual recognition based on the physical properties (e.g. color, density, texture) of metals and polymers, and magnetic detection for ferrous metals. The material composition was divided into three material categories: Ferrous (iron and steel), Non-ferrous (copper and aluminum) and Polymers (plastic and rubber). Components made of more than one materials (e.g. electromotors, wires, and connectors) were dismantled further to reach to the composition at material level. However, other components with much more complex material composition (e.g. PCBs and battery) were not dismantled further. 2.2. EoL assessment The fate of EoL RVCs was measured by following the material flows in the WEEE recycling chain by combining an experimental run in a preprocessing plant and a comprehensive material flow analysis. A total of 125 (349 kg) EoL RVCs was treated at a conventional preprocessing plant in Denmark. This sample size was chosen to provide enough feed for a minimum of 15-minute run in the preprocessing plant with a capacity of one metric ton per hour. The batteries (63 kg) were removed manually from the RVCs before feeding them into the plant for the trial. Therefore, the batteries are not part of the EoL assessment. Fig. 3 shows the process flow of the plant together with the material flows. In the plant, WEEE first goes through a manual dismantling and pre-sorting stage, where selected components (containing hazardous substances and/or valuable materials) are removed. Then a conveyor belt carries the WEEE to the chain shredder, which is equipped with a multi-cyclone system for the cleaning of the exhaust air, resulting three residue fractions (F1, F2 and F3). The material outflow from the shredder travels through an overbelt magnetic separator, where the ferrous fraction (F4) is partially picked by the magnet. The flow then enters a size-sorting unit with cut-off size 10 cm 10 cm, splitting the input into two streams with different particle sizes. The stream with smaller particle size passes through another overbelt magnet, followed by a drum magnet, where the remaining ferrous metals are separated more effectively, resulting in fractions F5 and F6 respectively. The remaining fraction is then sent to an eddy-current separator, where the non-metal (F7) and nonferrous metal (F8) fractions are separated. The stream with larger particle size also follows a similar route with an overbelt magnet, resulting in another ferrous fraction (F9) and the second eddy-current separator at the end of the process resulting in fractions F10 and F11. All 11 output fractions from the plant were collected and characterized. Each of the output fractions from the experimental run was sorted manually in the lab and their composition was determined. In case of non-liberated components, it was estimated based
654
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Fig. 1. Methodological approach of the study: product characterization and EoL assessment.
on the material composition of similar components and their degree of liberation. This material composition of each flow was then combined with the information provided by the recyclers in order to estimate the material recovery in the downstream recovery chain. The final material recovery was calculated using the recycling efficiencies reported by the recyclers and available literature. As reported by (Habib et al., 2015), a recovery rate of 95% is assumed for the ferrous metals. Similarly, the material recovery efficiency of the copper and aluminum smelters are considered 99% and 95% respectively. However, any other materials in this fraction are assumed lost during the smelting process. The material recovery efficiency at the plastic plant is assumed 47% for plastic and 99% for €ger and Hischier, 2013). At the metals based on the report by (Wa hammer mills, 95% of the metals are assumed to be recovered from the motor-trafo fraction. All the key assumptions are summarized in Tables D.1. 3. Results and discussion 3.1. Product characteristics A typical RVC has an average weight of 3.5 kg including its accessories, e.g. charging dock and remote control, while the robot itself weighs about 2.5 kg when the battery is removed. The shape, size, and weight of some selected models of RVCs can be found in Appendix B. All RVCs contain plastic casings, dustbin with filter, wheels, brushes, sensors, and printed circuit boards. Electromotors are used to drive the vacuum unit, wheels, and brushes. Some use small cameras for positioning along with proximity sensors. Optical and acoustic sensors are used for dirt detection and concentrated cleaning (iRobot, 2015). Most of them contain display units and user interface for setting up pre-programmed schedules. The components of RVCs can be grouped into the following five categories: Structural components (casing, cover, and body frame), Functional components (brush, wheels, and suction unit), Battery, PCBs, wires, and Other (small components such as sensors and connectors). Fig. 2 shows the weight share of each component group as well as their detailed material composition. Polymers (plastic and rubber) were the main building material for RVCs, with a share of 58% of the total product weight. Most of the
structural components and more than half of the functional components were made of plastic. Based on the labels provided on the product components, it mainly contained acrylonitrile-butadienestyrene (ABS) and polycarbonate (PC) type plastic. Ferrous metals were the other key material group in the product that summed up to 14% of the total product weight. They were mainly found in electromotors (powering the wheel and brush shafts), body frame, and screws used to connect the components. Non-ferrous metals and magnet (from electromotor) each weighed 3% of the total weight, while battery and PCBs weighed 18% and 5% respectively. Ferrous screws were found to be the most common connection type to bind the different components of RVC together. On average, each RVC used 75 screws to hold the mostly-plastic structural components together. Only four out of the 10 RVCs used snap connection to the lid covering battery, while for the remaining six, the number of screws to be removed to reach the battery varied from two to five. Similarly, at least 14 screws needed to be removed to recover the PCBs from the product. Other joint types included tiewraps to hold wires together and glue to attach small components. The WEEE Directive has categorized small household appliances (SHA) under Category 2, which includes products such as traditional vacuum cleaners, toasters, iron, and body care appliances (European Parliament, 2012). The WEEE categories provide a baseline for EoL collection and treatment. Based on the service provided by RVCs, they also belong to the Category 2 and therefore, are collected and treated together with other SHA products. However, RVCs differ from average SHA product in terms of material composition. For example, our empirical characterization study of household products showed the average weight of PCBs was found to be less than 1% for SHA (unpublished work), compared to 5% in RVCs. PCBs have higher value per unit weight compared to scrap steel or plastic, which makes it desirable to remove PCBs during the preprocessing step (Wang, 2014). While high-grade PCBs (e.g. from personal computers) are manually recovered before mechanical preprocessing, this practice is not common for household appliances. 3.2. EoL material recovery Fig. 3 shows the complete material flows starting from the shredder to final recovery. These flows are presented as two parts:
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
655
Fig. 2. Material compositions of different components and component groups in RVC, grouped into: Structural components (casing, cover, and body frame), moving components (brush, wheels, and suction unit), battery, PCBs, wires, and other (small electronic parts).
‘Preprocessing’ and ‘Downstream processing’. The first part provides the detailed picture of the processes and flows through the preprocessing plant, while the downstream part follows the material streams through multiple recycling facilities to estimate their recovery fate. 3.2.1. Preprocessing The mechanical processing resulted in 11 output fraction from the preprocessing plant, which are described below. Please refer Appendix C for the pictures of the output fractions. Fractions F1, F2 and F3 (residues) mainly consisted of dust particles and some plastic parts from the shredded RVCs. Some foreign materials were found in the dust particles that were on the filter or the dust bins. Most of it contained dust and fine particles that are most likely not a part of the RVCs. Ferrous Fraction (F4) consisted of large share of ferrous metal and non-liberated parts such as brush with plastic broom and metal shaft, plastic pieces with screws on it, electromotor in different condition (e.g. whole electromotor with plastic parts, broken steel parts, with copper and brass parts) etc. Fractions F5 and F6 mostly contained electromotors and transformers. It also contained clean Fe-metal, some of the broken electromotor parts (mainly the coils with copper and brass), and small plastic parts with ferrous screws on it. Fraction F9 also had similar composition. The non-metal output from the eddy-current separator (F7) was composed of mainly plastic. However, there were other nonliberated components including PCBs etc. The non-ferrous output (F8) mainly contained non-ferrous metals, and large amount of PCBs and capacitors detached from them during shredding. The outputs from the second eddy-current separator F10 and F11 were similar in material composition to F7 and F8, respectively. In the preprocessing plant, the primary intention of mechanical shredding is to liberate different materials and components in order to ease their separation in the subsequent processes. However, the
random breakdown of components such as electromotors, plastic casings, and PCBs resulted in their unwanted diversion. Fig. 4 illustrates the change in the state of materials and components in the process, along with the mass balance of input and output. For example, many non-liberated plastic and other non-ferrous components were picked by the magnetic separators together with the ferrous parts. Similarly, most of the electromotors in the motortrafo fraction were found broken with the magnet and copper coil missing inside. It was the result of random mechanical impact during the shredding of RVCs, which in turn, led to the mixing of different materials in the output streams. The non-ferrous fraction, however, was found to be relatively purer, but the total amount recovered was 72% lower compared to the input. Ideally, all non-ferrous metals and PCBs should also end up in the non-ferrous outputs. Therefore, the decreased weight percentage of non-ferrous output suggests losses of PCBs and non-ferrous metals combined. The non-metal fractions had the largest mass share and contained mainly shredded plastic, which is a result of the high plastic content in RVCs. Nevertheless, they also contained metal components, especially those that are not liberated properly during the shredding process. After the mechanical processing, some of the output fractions go through manual sorting, followed by a mixing step, to create homogeneous outgoing streams based the demand from the downstream recycling chain. For example, metals including copper, aluminum, and PCBs in the non-metal fraction (F7 and F10) are diverted to their respective outgoing streams. Similarly, fractions F5, F6 and F9 are combined and sent as ‘motor-trafo’ fraction to the hammer mill. This sorting and mixing also allows the preprocessing plant to prepare the volume required for shipment to the downstream recycling. 3.2.2. Downstream flows of the materials in the recycling chain The materials leave the preprocessing plant as the following six fractions (Fig. 3): Residues go to incineration or disposal, without any material recovery. Therefore, any material entering this flow is considered
656
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Fig. 3. Material flows in the WEEE recycling chain. The width of each flow represents their weight and the bars across the flows show their material compositions. [PCB: printed circuit board, Cu: copper, Al: Aluminum, NFe: other non-ferrous metals, Fe: ferrous metal].
Fig. 4. Diversion of material fractions during mechanical preprocessing. The input material composition is based on the dismantling of 10 sample RVCs (excluding batteries). The solid lines are used to show the desired material flow and dash lines for the main cross-flow of materials observed in the output streams.
lost. Ferrous stream (Fe-rich) is sent to steel smelter, where it gets recycled with high efficiency. Motor-trafo stream goes for secondary processing in Denmark, which involves shredding in hammer mills and separation of copper and small amounts of aluminum. Other materials including plastic and magnet fragments in the electromotors are not recovered in this process. The recovered steel and copper streams are
eventually sent to smelter outside the country. Plastic-rich stream goes to plastic recycler, where remaining fractions of metals are also recovered and sent to the respective smelters. Copper (Cu-rich) stream goes to copper smelter, where the copper from PCB is also recovered. RVCs contained mainly low-to medium-grade PCBs, for which the composition is assumed to be 30% metal (primarily copper) and 70% glass-reinforced polymer (WRAP, 2014). The non-metal part of PCBs is not recovered as material. As the focus of this study is not to quantify the detailed flows of trace elements, the possible presence of other metals (e.g. gold, silver, and palladium) is not accounted in the PCB material composition. It can be justified by the fact that when the copper in PCBs is recovered at the copper smelter, it also reflects the recovery of other metals. Therefore, copper stands as the reference metal to show the recovery of materials in PCBs. Aluminum (Al-rich) stream goes to aluminum smelter for material recovery. Overall, as much as 53% of the total material (151 kg out of 286 kg) coming out from the preprocessing is lost in the downstream recycling chain, with individual materials having different recovery rates. It includes the processing losses in the downstream recycling chain. The recovery rate of ferrous metals is 92%, while it is 83% and 92% for copper and aluminum, respectively (Fig. 5). Only 45% of total plastic was recovered as material. The detailed material compositions of the output fractions and final recovery are provided in Tables D.2 and D.3. These calculations are based on the best-case scenario, assuming close-to-perfect fine sorting with the given material
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
liberation achieved in the plant. In the real operational setup, more cross-flow of materials can be expected that can lead to less efficient material recovery. As it can be seen, most of the metals are recovered even if they are not sorted efficiently at the preprocessing plant. Plastics, on the other hand, have lower probability of being recycled when they are shredded and mixed with different polymer types. The shredding process causes the random size reduction and liberation of the high-grade plastic components that leads to varying size of the output recyclates, mixed with other materials.
3.3. Implications for the recycling industry and policymakers Though plastic contributes to the most significant share of the total material value in RVC, it was not found to be ‘the most important’ target material in the preprocessing step. Interestingly, the price of the plastic recyclates is arbitrarily decided based on the content of metals in it, which appears to be contrary to the aim of achieving pure material streams from the treatment process. Similarly, the recyclate is priced and accepted by the copper smelter based on the percentage of copper present in the mixture of plastic and other metals. Such market realities support the existing treatment processes, which are focused on producing ‘sellable’ outputs but not necessarily ‘clean’ material outputs. It has been documented that the recovery of the high-grade plastic components from plastic-rich products is the best option compared to other alternative disposal routes (Wager and Hischier, 2015), which can be achieved using selective dismantling of the plastic-rich products (Rios et al., 2003). While recovery of clean plastic components and the copper-rich electromotors could fetch more value, it will require labor-intensive manual dismantling resulting in higher processing cost. Likewise, advanced sorting based on sensors is an option for recovering clean plastics but is not always deemed viable for the given economies of scale. An ideal preprocessing should separate materials with a preferred consistency so that the outgoing streams can be used as feed for end processing, which will in turn help to achieve higher material recovery. The conventional size reduction that uses mechanical shredders cannot guarantee the separation of units like PCBs, electromotors, and other composite parts (metal-metal or metal-plastic combination) and have been often associated with loss of valuable materials (Chancerel et al., 2011; Wang et al., 2012). The impurities due to cross flow of materials during the process also means economic loss for the plant. Nevertheless, the final recovery fate of the materials also considerably depends on the downstream recycling pathways. Our findings show that materials with higher value and established recycling infrastructure (e.g.
Fig. 5. Overall material recovery.
657
copper) will be eventually recovered from the recyclates streams. Moreover, the downstream processing of low-value output streams (e.g. plastic-rich fraction) is driven by the content of valuable metals. Thus, the upfront separation efficiency and the purity of the streams are mainly an issue for value creation, not as much for final material recovery. It is important here to mention that the objective of this study is to understand the performance of RVCs in the existing preprocessing setup, and not to evaluate the plant itself. The reference facility used in this study represents a contemporary mechanical WEEE pre-treatment practice for categories including SHA. We measure the performance of the whole product against the recycling system in order to identify hotspots of material losses. This approach provides the full picture of the EoL processing and the loss and recovery of the key materials reported in this study thus reflect the fate of other resources as well. The results above clearly show that new products like RVCs have higher potential of resource recovery than it is being exploited by the existing recycling chain. The WEEE Directive (European Parliament, 2012) requires the member states in European Union to report the recycling rates of each WEEE category. The overall recycling rates are, therefore, calculated and collectively reported at category level by the member states. Such calculations are based on the amount of WEEE collected and the amount sent for treatment, and the information provided by the downstream recyclers on material recovery. However, our finding suggests these recycling rates reported for categories do not reflect the true material recovery of individual products in those categories. In Denmark, for example, the recycling rate for SHA was reported to be 87% for 2012 (DPA-System, 2013) compared to our estimated rate of 47% for RVCs. Higher recycling rates may be achievable for metal-rich WEEE with less complex material composition, but it is evident that not all products under one category can be processed with equal efficiency in the existing recycling chain. This is even more pertinent when there is a significant difference in material composition and design of products. 3.4. Product design and EoL processing Fig. 6 illustrates an example of issues related to the product design in the process of material liberation and separation. On the left, it shows the output fraction from the first overbelt magnet, which should ideally be pure ferrous metals. On the right is an under-shredded RVC that came out almost intact from the preprocessing plant. The inefficiencies of both the shredding process and product design have resulted in such mixed and inconsistent output streams. The plastic brooms in the brush were either weaved or glued to the metal shaft, which made it impossible to separate them by shredding. Moreover, large pieces of plastics, for example, were diverted to this fraction due to a small steel screw. It is because the steel screw did not liberate in the shredding process and was eventually picked by the magnet. Arguably, the incomplete liberation can be related to inadequate shredding of the input, but more intensive shredding is not necessarily helpful. It may reduce the component size, but without the guarantee of separating different materials. For example, the number of pure steel screws in the ferrous fractions was less than the screws found with some plastic component in it (Figure E.1). Similarly, the copper coils and steel casing of an electromotor cannot be separated by using more intense shredding as it results in the loss of magnet in the electromotor, which converts into powder due to the impact. As this powder is diluted in all the outgoing fractions, the probability of recovering the materials in the magnet becomes very low (Habib et al., 2015). Clearly, the product was not designed to perform well in the
658
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Fig. 6. Examples of incomplete size reduction and material liberation. Left: Mixture of different materials and components in the ferrous fraction (F4); Right: An under-shredded RVC piece found in the non-metal fraction (F10).
existing shredding-based EoL processing, or to say the least, there is a mismatch between EoL product type and EoL processing technique. The key design issues affecting the material recovery in the preprocessing steps are summarized below: Immature product design: Classical design errors such as ‘birdnest’ wiring were found in RVCs (Figure E.2). It makes not only a potential manual disassembly process, but also the material separation after shredding, more difficult. Compartmentalization of wires and connector will allow easier recovery of the non-ferrous metals used in them and reduce contamination to other output fractions. Use of connectors: Improper use of connectors (e.g. too many screws e resulting in longer dismantling time and incomplete separation) and incompatible materials (e.g. ferrous steel screws in plastic components e resulting in inefficient material separation) show the lack of ‘design for EoL’ thinking (Fig. 6). A slight design alternative such as using plastic-based snap fit or plastic screw will improve the material separation. Placement of key components: Some components need to be removed from the EoL products before it can be shredded. Removal of battery and PCBs (>10 cm2), for example, is required by the WEEE Directive. Making such components easily accessible by using less complicated connectors and with proper placement of the components will allow quicker removal. Moreover, material recovery from components such as PCBs and electric motors can be improved if they are treated separately. Materials compatibility: Aside from connectors, the issue of material compatibility has also been observed in other components such as broom and wheel. They usually are made of composite of plastic and steel, put together either by weaving or by using glue. This resulted in poor material liberation. Such issues can be avoided, for example, by using compatible materials or using only one material for the component (e.g. use of plastic shaft for the plastic broom/brush). Both the WEEE (European Parliament, 2012) and EcoDesign (European Parliament, 2009) Directives suggest a ‘comprehensive approach’ towards improving the environmental performance of products by means of product design, considering the whole lifecycle that includes the product EoL. However, our case study of an emerging electronic product showed little evidence of EoL thinking in product design. This can be related to, among other reasons, the lack of intensive for the producers and the designers who are not certain of the EoL fate of their product. 4. Conclusions Though the current EoL processing technologies are evolved
from years of learning and improvement by the recycling industry, they are constantly being challenged by the diversity of electrical and electronic products. In this context, achieving higher recovery of materials from EoL products will require a detailed understanding of their performance in the recycling chain. This study comprehensively illustrates the EoL fate of resources used in the case product by taking into account the existing recycling infrastructure. Our findings suggest that, in the best case, the existing recycling chain may recover up to 92% of the base metals from EoL RVCs. However, only 45% of the plastics were estimated to be recycled as material. Further, the conventional ‘shred-and-separate’ approach in the preprocessing of the EoL products resulted in dispersion and eventual loss of components such as PCBs and magnets. Tailored processing based on the understanding of the material composition and design features of the EoL product will improve the material recovery. This case study also exemplifies the lack of ‘EoL thinking’ in the design of a relatively new product. Classical design issues such as bird-nest wiring and incompatible materials and connector types in the product components added to the challenges in the material recovery process. Better recycling rate, for plastics in particular, is possible with improved material compatibility and ease of disassembly. On top of the lack of appropriate recycling technology and product design, the material losses were also caused by the market realities that are driven more by economic than environmental interests. Despite the constant discussion and increasing legislative push, practical implementation of ‘design for EoL’ thinking still remains a challenge. This study offers a robust methodological approach for assessing the EoL performance of a product in the complex material recovery chain and provides the understanding of how the design of a product is linked with the processes in the chain. The methodology and findings from this study can be used to assess the EoL performance of other products and to support the integration of the product design with the EoL resource recovery. Acknowledgement We thank Christina Kaloudaki Pangidou and Tom Ellegaard for their help during the experiment and information collection. The research was a part of the INNOSORT project (innosort.teknologisk. dk) that is funded by the Danish Agency for Science, Technology and Innovation. We thank the Agency and the project partners for their support. The views contained in the paper are those of the authors and do not represent the official views or policies of any stakeholders.
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Appendix A. Product disassembly
Figure A.1. Main Unit (Top & Bottom View).
Figure A.2. Removal of Battery.
Figure A.3. Side Brushes.
659
660
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Figure A.4. Dust Bin.
Figure A.5. Filter.
Figure A.6. Casing.
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Figure A.7. Top Cover.
Figure A.8. Printed Circuit Board.
Figure A.9. Detailed inside View.
661
662
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Figure A.10. Wheels.
Figure A.11. Power Brush Casing & Motor.
Figure A.12. Electromotors.
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
663
Figure A.13. Dismantled electromotors.
Figure A.14. Screws.
Appendix B. Robotic vacuum cleaner
Table B.1 Size of some selected RVC types. RVC Model
Weight (kg)
Dimension (mm) Diameter
Height
Samsung Navibot LG Hom-Bot iRobot Roomba
3.5 3.2 3.8
355 370 355
90 90 80
664
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Appendix C. Output fractions
Figure C.4. Fraction 5. Figure C.1. .Fraction 1.
Figure C.2. Fraction 2. Figure C.5. Fraction 6.
Figure C.3. Fraction 4.
Figure C.6. Fraction 7.
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
665
Figure C.10. Fraction 11.
Figure C.7. Fraction 8.
Appendix D. Key assumptions and calculations
Table D.1 Key assumptions. 1. Recovery Rates
Hammer mill: 95% Plastic recycler: Metal 99%, Plastic 47% Steel Smelter: 95% Copper Smelter: 99% Al Smelter: 95% 2. PCB composition 30% Cu, 70% glass-reinforced polymers 3. Material Market Price Steel/Iron: 0.12, Copper: 5.2 (V/kg) Aluminium: 1.5, Plastics: 1.2
Source: Local recyclers (W€ ager and Hischier, 2013) (Habib et al., 2015) (WRAP, 2014) (Cucchiella et al., 2015)
Figure C.8. Fraction 9.
Table D.2 Material Composition of output Fractions from preprocessing plant.
Figure C.9. Fraction 10.
Fraction
Material composition (kg) Fe
Cu
Al
PCB
Plastic Rest
%
Residues (1 þ 2þ3) Ferrous (4) Motor-trafo (5 þ 6þ9) Plastic Fraction (7 þ 10) Non-Ferrous (8 þ 11) Total (kg)
0.00 18.11 14.32 4.00
0.05 1.42 2.29 4.24
0.00 0.08 0.29 0.71
1.70 0.46 0.71 9.19
0.06 9.02 5.17 178.57
2.8% 10.9% 9.7% 73.8%
6.23 1.91 5.01 13.89
0.12 0.53 1.44 3.74 1.72 0.39 2.8% 36.55 8.53 2.52 15.80 194.54 27.41 285.35 kg (100%)
666
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Table D.3 Final Material Recovery and Value share. Materials /
Fe
Cu
Al
Plastic
Rest
Total Weight (kg) Material Recovered (kg) Material Lost (kg) % Recovered (of total of each material)
36.55 33.73 2.82 92%
13.27 11.46 1.81 83%
2.52 2.33 0.19 92%
194.54 87.13 107.41 45%
38.47 e 38.47 e
285.35 134.65 150.70
Materials market price (V/kg) Recovered Value (V) Lost Value (V) Total Value (V)
0.12 4.05 0.34 4.39
5.2 59.61 9.39 69.01
1.5 3.49 0.29 3.78
1.2 104.56 128.89 233.45
e e e e
Total 171.71 138.91 310.62
Recovered Materials (weight %) Recovered Value (%)
25.0% 2.4%
8.5% 34.7%
1.7% 2.0%
64.7% 60.9%
e e
100% 100%
% 47% 53%
Appendix E. Design features References
Figure E.1. The liberated screws (left) and non-liberated screws with plastic and other metal components (right).
Figure E.2. The bird-nest wiring found in a dismantled RVC.
Basdere, B., Seliger, G., 2003. Disassembly factories for electrical and electronic products to recover resources in product and material cycles. Environ. Sci. Technol. 37, 5354e5362. Chancerel, P., Bolland, T., Rotter, V.S., 2011. Status of pre-processing of waste electrical and electronic equipment in Germany and its influence on the recovery of gold. Waste Manag. Res. 29, 309e317. Chancerel, P., Meskers, C.E.M., Hagelüken, C., Rotter, V.S., 2009. Assessment of precious metal flows during preprocessing of waste electrical and electronic equipment. J. Industrial Ecol. 13, 791e810. Cucchiella, F., D'Adamo, I., Lenny Koh, S.C., Rosa, P., 2015. Recycling of WEEEs: an economic assessment of present and future e-waste streams. Renew. Sustain. Energy Rev. 51, 263e272. DPA-System, 2013. WEEE, BAT, and ELV Statistics Denmark, seventh ed. Euromonitor, 2012. Robotic Vacuum Cleaners Welcomes its Golden Age. Euromonitor International. Euromonitor, 2015. Vacuum Cleaners in Denmark. Euromonitor International. European Parliament, 2009. Directive 2009/125/EC of the European Parliament and of the Council of 21 October 2009 establishing a framework for the setting of ecodesign requirements for energy-related products (recast). Official J. Eur. Union. European Parliament, 2012. Directive 2012/19/EU of the European Parliament and of the Council of 4 July 2012 on waste electrical and electronic equipment (WEEE) (recast). Official J. Eur. Union. Graedel, T.E., Allwood, J., Birat, J.-P., Buchert, M., Hagelüken, C., Reck, B.K., Sibley, S.F., Sonnemann, G., 2011. What do we know about metal recycling rates? J. Industrial Ecol. 15, 355e366. Habib, K., Parajuly, K., Wenzel, H., 2015. Tracking the flow of resources in electronic waste - the case of end-of-life computer hard disk drives. Environ. Sci. Technol. 49 (20), 12441e12449. iRobot, 2015. iRobot ROOMBA Specifications. €hler, A.R., 2013. Challenges for eco-design of emerging technologies: the case of Ko electronic textiles. Mater. Des. 51, 51e60. Lee, H.M., Lu, W.F., Song, B., 1 March 2014. A framework for assessing product EndOf-Life performance: reviewing the state of the art and proposing an innovative approach using an End-of-Life Index. J. Clean. Prod. 66, 355e371. Li, J., Zeng, X., Chen, M., Ogunseitan, O.A., Stevels, A., 2015. “Control-alt-delete”: rebooting solutions for the E-waste problem. Environ. Sci. Technol. 49, 7095e7108. Mayers, K., Peagam, R., France, C., Basson, L., Clift, R., 2011. Redesigning the camel. J. Industrial Ecol. 15, 4e8. Meskers, C.E.M., Hagelüken, C., Salhofer, S., Spitzbart, M., 2009. Impact of preprocessing routes on precious metal recovery from PCs. In: GMBD (Ed.), European Metallurgical Conference, Innsbruck, Austria. Rios, P., Stuart, J.A., Grant, E., 2003. Plastics disassembly versus bulk Recycling: engineering design for end-of-life electronics resource recovery. Environ. Sci. Technol. 37, 5463e5470. Wager, P.A., Hischier, R., 2015. Life cycle assessment of post-consumer plastics production from waste electrical and electronic equipment (WEEE) treatment residues in a Central European plastics recycling plant. Sci. Total Environ. 529,158e167. Wang, F., 2014. E-waste: Collect More, Treat Better: Tracking Take-back System Performance for Eco-efficient Electronics Recycling. Delft University of Technology. Wang, F., Huisman, J., Meskers, C.E., Schluep, M., Stevels, A., Hageluken, C., 2012. The Best-of-2-Worlds philosophy: developing local dismantling and global infrastructure network for sustainable e-waste treatment in emerging economies. Waste Manag. 32, 2134e2146. Widmer, R., Du, X., Haag, O., Restrepo, E., Wager, P.A., 2015. Scarce metals in conventional passenger vehicles and end-of-life vehicle shredder output. Environ. Sci. Technol. 49, 4591e4599. WRAP, 2014. Techniques for Recovering Printed Circuit Boards (PCBs). Waste & Resources Action Programme. €ger, P., Hischier, R., 2013. Life Cycle Analysis of MBA Polymers' Post-consumer Wa Recycled Plastics. Swiss Federal Laboratories for Materials Science and Technology (Empa).
Contents lists available at ScienceDirect
Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro
End-of-life resource recovery from emerging electronic products e A case study of robotic vacuum cleaners Keshav Parajuly*, Komal Habib, Ciprian Cimpan, Gang Liu, Henrik Wenzel SDU Life Cycle Engineering, Department of Chemical Engineering, Biotechnology and Environmental Technology, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark
a r t i c l e i n f o
a b s t r a c t
Article history: Received 15 March 2016 Received in revised form 22 July 2016 Accepted 22 July 2016 Available online 27 July 2016
Integrating product design with appropriate end-of-life (EoL) processing is widely recognized to have huge potentials in improving resource recovery from electronic products. In this study, we investigate both the product characteristics and EoL processing of robotic vacuum cleaner (RVC), as a case of emerging electronic product, in order to understand the recovery fate of different materials and its linkage to product design. Ten different brands of RVC were dismantled and their material composition and design profiles were studied. Another 125 RVCs (349 kg) were used for an experimental trial at a conventional ‘shred-and-separate’ type preprocessing plant in Denmark. A detailed material flow analysis was performed throughout the recycling chain. The results show a mismatch between product design and EoL processing, and the lack of practical implementation of ‘Design for EoL’ thinking. In the best-case scenario, only 47% of the total materials in RVCs are ultimately recycled. While this low material recovery is mainly due to the lower plastic recycling rate, other market realities and the complex material flows in the recycling chain also contribute to it. The study provides a robust methodological approach for assessing the EoL performance based on the knowledge of a product and its complex recycling chain. The lessons learned can be used to support both the design and EoL processing of products with similar features, which carry a high potential for resource recovery, especially at the initial stage of the recycling chain. © 2016 Elsevier Ltd. All rights reserved.
Keywords: Robotic vacuum cleaner Material flow analysis Material recovery Design for end-of-life Design for recycling WEEE
1. Introduction The lack of efficient collection and appropriate recycling infrastructure is one of the major challenges in achieving a ‘closed’ loop of materials from consumer goods, including electrical and electronic products (Graedel et al., 2011). Further, the complex design of modern products adds to the challenges in the end-of-life (EoL) treatment processes, as the recycling technologies are being outpaced by emerging composite and elementally diverse products, making the resource recovery process more and more difficult. Product design and EoL management have a significant impact on the resource recovery as well as the overall environmental impacts of a product (Li et al., 2015). Efforts have been made towards linking these two stages e the product design/inception and its EoL processing e by incorporating ‘design for EoL’ approach for better EoL performance of products (Lee et al., 2014). Nevertheless, there are still considerable opportunities, both technological and legislative,
* Corresponding author. E-mail address: [email protected] (K. Parajuly). http://dx.doi.org/10.1016/j.jclepro.2016.07.142 0959-6526/© 2016 Elsevier Ltd. All rights reserved.
for linking the producers and EoL managers to intensify this approach (Li et al., 2015; Mayers et al., 2011). A typical resource recovery chain consists of three main steps: collection, preprocessing, and end processing. Being the first treatment step, preprocessing has a large impact on the subsequent pathways and fate of materials and on the final resource recovery in the whole-system perspective. Manual or mechanical, preprocessing serves as the guide for material flows in the following treatment steps. It defines the effectiveness of material liberation, sorting, and diversion to the correct downstream processing path, which ultimately influences the overall recycling rate (Chancerel et al., 2009). The challenges in the recycling process e especially the preprocessing step e have been illustrated for several products, including case studies on recovery of valuable metals from desktop computers (Meskers et al., 2009; Wang et al., 2012), computer hard disk drives (HDDs) (Habib et al., 2015), and electrical and electronic components in vehicles (Widmer et al., 2015). A common conclusion of these case studies is that the existing preprocessing infrastructure is not sufficient, as it causes significant losses of valuable resources in the process. Manual dismantling and positive sorting of the valuable components like high-grade printed circuit boards (PCBs) and HDDs is
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
common for products such as computers. This step is usually more expensive, especially in developed countries, given the high labor cost and varying product types with unfamiliar product properties (Basdere and Seliger, 2003). As a rule of thumb, the EoL products with enough valuable materials to cover the cost of dismantling are processed manually for component recovery, while the rest is sent to generalized processing. Generalized mechanical shredding and separation is the most preferred way of handling low-grade waste electrical and electronic equipment (WEEE) such as household appliances category (Chancerel et al., 2011). As the technology advances, the ubiquitous use of electronics in our daily life has only been increasing. It has led to the rising use of products such as photovoltaic panels (Cucchiella et al., 2015), elec€ hler, 2013), wearable electronics such as smart tronic textiles (Ko watches, and unmanned aerial vehicles (drones) used for amateur photography. Examples of household appliances include smart weighing scales with communication capabilities and robotic vacuum cleaners (RVCs) that can operate autonomously. As products evolve, more household appliances are incorporating complex electronic components, such as PCBs, sensors, and display panels. With this changing product features and material composition comes the challenge of addressing them under the existing EoL management setups, which are evolving slower than the products. The treatment options relying on conventional physical processing technique cannot handle every product with the same efficiency of material recovery. The recovery of valuable metals from the high-grade electronic waste has been the major research focus, whereas household products get less attention. Moreover, the possibilities of improving the design of these emerging products to allow an efficient resource recovery remain largely untested. The EoL-targeted design improvements need to be based on the knowledge of the EoL fate of the product. In order to do so, it becomes crucial to understand the performance of such products in the existing recycling chain, and the role of product design in improving the overall resource recovery efficiency. In this study, we give a more nuanced and differentiated view on these aspects, including an elaboration on which materials and components in the product are influenced the most by lack of efficient preprocessing separation. We use RVC as a case to illustrate the fate of an emerging product in the existing recycling chain and identify the potential for the improvement in product design. We aim to answer three main questions: a) Where are the main material losses in the existing recycling chain? b) For which materials and components is the problem of inefficient separation in the preprocessing step most significant? c) What is the connection of these losses to the design features of the product and how it can be addressed? RVC illustrates the trend of programmable household appliance that requires minimal human involvement. One study reported almost a quadruple growth of RVC sales in Denmark between 2010 and 2014, with the total volume reaching from 12,500 units to 48,100 units (Euromonitor, 2015). Another study forecasts 3 million units of RVCs will be marketed globally in 2016, (Euromonitor, 2012), which would add 10.5 kilo tons into the WEEE stream at the end of their lives. Although this volume of EoL RVC is not of a huge significance on its own compared to total WEEE flow, it exemplifies the trend of emerging electronic appliances and serves as a representative case product. 2. Materials and methods The study consists of two main parts e the first part (product characterization) studies the product characteristics, while the
653
second part (EoL assessment) focuses on the performance of the case product in the existing recycling chain. The methodological approach is illustrated in Fig. 1 and is described in the following subsections. 2.1. Product characterization The purpose of this step was to understand the design features and the material composition of the case product. For that, ten RVCs of different brands were disassembled with the help of basic handheld tools such as screwdriver, pliers, and hammer. The different connection types and sequence of disassembly of different components were noted, which can be related to the ease of manual dismantling. An example of step-wise disassembly is illustrated in Appendix A. After the disassembly, the materials of each component were identified, and if required, the components were further dismantled to reach the material-level identification. Materials were identified using techniques including visual recognition based on the physical properties (e.g. color, density, texture) of metals and polymers, and magnetic detection for ferrous metals. The material composition was divided into three material categories: Ferrous (iron and steel), Non-ferrous (copper and aluminum) and Polymers (plastic and rubber). Components made of more than one materials (e.g. electromotors, wires, and connectors) were dismantled further to reach to the composition at material level. However, other components with much more complex material composition (e.g. PCBs and battery) were not dismantled further. 2.2. EoL assessment The fate of EoL RVCs was measured by following the material flows in the WEEE recycling chain by combining an experimental run in a preprocessing plant and a comprehensive material flow analysis. A total of 125 (349 kg) EoL RVCs was treated at a conventional preprocessing plant in Denmark. This sample size was chosen to provide enough feed for a minimum of 15-minute run in the preprocessing plant with a capacity of one metric ton per hour. The batteries (63 kg) were removed manually from the RVCs before feeding them into the plant for the trial. Therefore, the batteries are not part of the EoL assessment. Fig. 3 shows the process flow of the plant together with the material flows. In the plant, WEEE first goes through a manual dismantling and pre-sorting stage, where selected components (containing hazardous substances and/or valuable materials) are removed. Then a conveyor belt carries the WEEE to the chain shredder, which is equipped with a multi-cyclone system for the cleaning of the exhaust air, resulting three residue fractions (F1, F2 and F3). The material outflow from the shredder travels through an overbelt magnetic separator, where the ferrous fraction (F4) is partially picked by the magnet. The flow then enters a size-sorting unit with cut-off size 10 cm 10 cm, splitting the input into two streams with different particle sizes. The stream with smaller particle size passes through another overbelt magnet, followed by a drum magnet, where the remaining ferrous metals are separated more effectively, resulting in fractions F5 and F6 respectively. The remaining fraction is then sent to an eddy-current separator, where the non-metal (F7) and nonferrous metal (F8) fractions are separated. The stream with larger particle size also follows a similar route with an overbelt magnet, resulting in another ferrous fraction (F9) and the second eddy-current separator at the end of the process resulting in fractions F10 and F11. All 11 output fractions from the plant were collected and characterized. Each of the output fractions from the experimental run was sorted manually in the lab and their composition was determined. In case of non-liberated components, it was estimated based
654
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Fig. 1. Methodological approach of the study: product characterization and EoL assessment.
on the material composition of similar components and their degree of liberation. This material composition of each flow was then combined with the information provided by the recyclers in order to estimate the material recovery in the downstream recovery chain. The final material recovery was calculated using the recycling efficiencies reported by the recyclers and available literature. As reported by (Habib et al., 2015), a recovery rate of 95% is assumed for the ferrous metals. Similarly, the material recovery efficiency of the copper and aluminum smelters are considered 99% and 95% respectively. However, any other materials in this fraction are assumed lost during the smelting process. The material recovery efficiency at the plastic plant is assumed 47% for plastic and 99% for €ger and Hischier, 2013). At the metals based on the report by (Wa hammer mills, 95% of the metals are assumed to be recovered from the motor-trafo fraction. All the key assumptions are summarized in Tables D.1. 3. Results and discussion 3.1. Product characteristics A typical RVC has an average weight of 3.5 kg including its accessories, e.g. charging dock and remote control, while the robot itself weighs about 2.5 kg when the battery is removed. The shape, size, and weight of some selected models of RVCs can be found in Appendix B. All RVCs contain plastic casings, dustbin with filter, wheels, brushes, sensors, and printed circuit boards. Electromotors are used to drive the vacuum unit, wheels, and brushes. Some use small cameras for positioning along with proximity sensors. Optical and acoustic sensors are used for dirt detection and concentrated cleaning (iRobot, 2015). Most of them contain display units and user interface for setting up pre-programmed schedules. The components of RVCs can be grouped into the following five categories: Structural components (casing, cover, and body frame), Functional components (brush, wheels, and suction unit), Battery, PCBs, wires, and Other (small components such as sensors and connectors). Fig. 2 shows the weight share of each component group as well as their detailed material composition. Polymers (plastic and rubber) were the main building material for RVCs, with a share of 58% of the total product weight. Most of the
structural components and more than half of the functional components were made of plastic. Based on the labels provided on the product components, it mainly contained acrylonitrile-butadienestyrene (ABS) and polycarbonate (PC) type plastic. Ferrous metals were the other key material group in the product that summed up to 14% of the total product weight. They were mainly found in electromotors (powering the wheel and brush shafts), body frame, and screws used to connect the components. Non-ferrous metals and magnet (from electromotor) each weighed 3% of the total weight, while battery and PCBs weighed 18% and 5% respectively. Ferrous screws were found to be the most common connection type to bind the different components of RVC together. On average, each RVC used 75 screws to hold the mostly-plastic structural components together. Only four out of the 10 RVCs used snap connection to the lid covering battery, while for the remaining six, the number of screws to be removed to reach the battery varied from two to five. Similarly, at least 14 screws needed to be removed to recover the PCBs from the product. Other joint types included tiewraps to hold wires together and glue to attach small components. The WEEE Directive has categorized small household appliances (SHA) under Category 2, which includes products such as traditional vacuum cleaners, toasters, iron, and body care appliances (European Parliament, 2012). The WEEE categories provide a baseline for EoL collection and treatment. Based on the service provided by RVCs, they also belong to the Category 2 and therefore, are collected and treated together with other SHA products. However, RVCs differ from average SHA product in terms of material composition. For example, our empirical characterization study of household products showed the average weight of PCBs was found to be less than 1% for SHA (unpublished work), compared to 5% in RVCs. PCBs have higher value per unit weight compared to scrap steel or plastic, which makes it desirable to remove PCBs during the preprocessing step (Wang, 2014). While high-grade PCBs (e.g. from personal computers) are manually recovered before mechanical preprocessing, this practice is not common for household appliances. 3.2. EoL material recovery Fig. 3 shows the complete material flows starting from the shredder to final recovery. These flows are presented as two parts:
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
655
Fig. 2. Material compositions of different components and component groups in RVC, grouped into: Structural components (casing, cover, and body frame), moving components (brush, wheels, and suction unit), battery, PCBs, wires, and other (small electronic parts).
‘Preprocessing’ and ‘Downstream processing’. The first part provides the detailed picture of the processes and flows through the preprocessing plant, while the downstream part follows the material streams through multiple recycling facilities to estimate their recovery fate. 3.2.1. Preprocessing The mechanical processing resulted in 11 output fraction from the preprocessing plant, which are described below. Please refer Appendix C for the pictures of the output fractions. Fractions F1, F2 and F3 (residues) mainly consisted of dust particles and some plastic parts from the shredded RVCs. Some foreign materials were found in the dust particles that were on the filter or the dust bins. Most of it contained dust and fine particles that are most likely not a part of the RVCs. Ferrous Fraction (F4) consisted of large share of ferrous metal and non-liberated parts such as brush with plastic broom and metal shaft, plastic pieces with screws on it, electromotor in different condition (e.g. whole electromotor with plastic parts, broken steel parts, with copper and brass parts) etc. Fractions F5 and F6 mostly contained electromotors and transformers. It also contained clean Fe-metal, some of the broken electromotor parts (mainly the coils with copper and brass), and small plastic parts with ferrous screws on it. Fraction F9 also had similar composition. The non-metal output from the eddy-current separator (F7) was composed of mainly plastic. However, there were other nonliberated components including PCBs etc. The non-ferrous output (F8) mainly contained non-ferrous metals, and large amount of PCBs and capacitors detached from them during shredding. The outputs from the second eddy-current separator F10 and F11 were similar in material composition to F7 and F8, respectively. In the preprocessing plant, the primary intention of mechanical shredding is to liberate different materials and components in order to ease their separation in the subsequent processes. However, the
random breakdown of components such as electromotors, plastic casings, and PCBs resulted in their unwanted diversion. Fig. 4 illustrates the change in the state of materials and components in the process, along with the mass balance of input and output. For example, many non-liberated plastic and other non-ferrous components were picked by the magnetic separators together with the ferrous parts. Similarly, most of the electromotors in the motortrafo fraction were found broken with the magnet and copper coil missing inside. It was the result of random mechanical impact during the shredding of RVCs, which in turn, led to the mixing of different materials in the output streams. The non-ferrous fraction, however, was found to be relatively purer, but the total amount recovered was 72% lower compared to the input. Ideally, all non-ferrous metals and PCBs should also end up in the non-ferrous outputs. Therefore, the decreased weight percentage of non-ferrous output suggests losses of PCBs and non-ferrous metals combined. The non-metal fractions had the largest mass share and contained mainly shredded plastic, which is a result of the high plastic content in RVCs. Nevertheless, they also contained metal components, especially those that are not liberated properly during the shredding process. After the mechanical processing, some of the output fractions go through manual sorting, followed by a mixing step, to create homogeneous outgoing streams based the demand from the downstream recycling chain. For example, metals including copper, aluminum, and PCBs in the non-metal fraction (F7 and F10) are diverted to their respective outgoing streams. Similarly, fractions F5, F6 and F9 are combined and sent as ‘motor-trafo’ fraction to the hammer mill. This sorting and mixing also allows the preprocessing plant to prepare the volume required for shipment to the downstream recycling. 3.2.2. Downstream flows of the materials in the recycling chain The materials leave the preprocessing plant as the following six fractions (Fig. 3): Residues go to incineration or disposal, without any material recovery. Therefore, any material entering this flow is considered
656
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Fig. 3. Material flows in the WEEE recycling chain. The width of each flow represents their weight and the bars across the flows show their material compositions. [PCB: printed circuit board, Cu: copper, Al: Aluminum, NFe: other non-ferrous metals, Fe: ferrous metal].
Fig. 4. Diversion of material fractions during mechanical preprocessing. The input material composition is based on the dismantling of 10 sample RVCs (excluding batteries). The solid lines are used to show the desired material flow and dash lines for the main cross-flow of materials observed in the output streams.
lost. Ferrous stream (Fe-rich) is sent to steel smelter, where it gets recycled with high efficiency. Motor-trafo stream goes for secondary processing in Denmark, which involves shredding in hammer mills and separation of copper and small amounts of aluminum. Other materials including plastic and magnet fragments in the electromotors are not recovered in this process. The recovered steel and copper streams are
eventually sent to smelter outside the country. Plastic-rich stream goes to plastic recycler, where remaining fractions of metals are also recovered and sent to the respective smelters. Copper (Cu-rich) stream goes to copper smelter, where the copper from PCB is also recovered. RVCs contained mainly low-to medium-grade PCBs, for which the composition is assumed to be 30% metal (primarily copper) and 70% glass-reinforced polymer (WRAP, 2014). The non-metal part of PCBs is not recovered as material. As the focus of this study is not to quantify the detailed flows of trace elements, the possible presence of other metals (e.g. gold, silver, and palladium) is not accounted in the PCB material composition. It can be justified by the fact that when the copper in PCBs is recovered at the copper smelter, it also reflects the recovery of other metals. Therefore, copper stands as the reference metal to show the recovery of materials in PCBs. Aluminum (Al-rich) stream goes to aluminum smelter for material recovery. Overall, as much as 53% of the total material (151 kg out of 286 kg) coming out from the preprocessing is lost in the downstream recycling chain, with individual materials having different recovery rates. It includes the processing losses in the downstream recycling chain. The recovery rate of ferrous metals is 92%, while it is 83% and 92% for copper and aluminum, respectively (Fig. 5). Only 45% of total plastic was recovered as material. The detailed material compositions of the output fractions and final recovery are provided in Tables D.2 and D.3. These calculations are based on the best-case scenario, assuming close-to-perfect fine sorting with the given material
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
liberation achieved in the plant. In the real operational setup, more cross-flow of materials can be expected that can lead to less efficient material recovery. As it can be seen, most of the metals are recovered even if they are not sorted efficiently at the preprocessing plant. Plastics, on the other hand, have lower probability of being recycled when they are shredded and mixed with different polymer types. The shredding process causes the random size reduction and liberation of the high-grade plastic components that leads to varying size of the output recyclates, mixed with other materials.
3.3. Implications for the recycling industry and policymakers Though plastic contributes to the most significant share of the total material value in RVC, it was not found to be ‘the most important’ target material in the preprocessing step. Interestingly, the price of the plastic recyclates is arbitrarily decided based on the content of metals in it, which appears to be contrary to the aim of achieving pure material streams from the treatment process. Similarly, the recyclate is priced and accepted by the copper smelter based on the percentage of copper present in the mixture of plastic and other metals. Such market realities support the existing treatment processes, which are focused on producing ‘sellable’ outputs but not necessarily ‘clean’ material outputs. It has been documented that the recovery of the high-grade plastic components from plastic-rich products is the best option compared to other alternative disposal routes (Wager and Hischier, 2015), which can be achieved using selective dismantling of the plastic-rich products (Rios et al., 2003). While recovery of clean plastic components and the copper-rich electromotors could fetch more value, it will require labor-intensive manual dismantling resulting in higher processing cost. Likewise, advanced sorting based on sensors is an option for recovering clean plastics but is not always deemed viable for the given economies of scale. An ideal preprocessing should separate materials with a preferred consistency so that the outgoing streams can be used as feed for end processing, which will in turn help to achieve higher material recovery. The conventional size reduction that uses mechanical shredders cannot guarantee the separation of units like PCBs, electromotors, and other composite parts (metal-metal or metal-plastic combination) and have been often associated with loss of valuable materials (Chancerel et al., 2011; Wang et al., 2012). The impurities due to cross flow of materials during the process also means economic loss for the plant. Nevertheless, the final recovery fate of the materials also considerably depends on the downstream recycling pathways. Our findings show that materials with higher value and established recycling infrastructure (e.g.
Fig. 5. Overall material recovery.
657
copper) will be eventually recovered from the recyclates streams. Moreover, the downstream processing of low-value output streams (e.g. plastic-rich fraction) is driven by the content of valuable metals. Thus, the upfront separation efficiency and the purity of the streams are mainly an issue for value creation, not as much for final material recovery. It is important here to mention that the objective of this study is to understand the performance of RVCs in the existing preprocessing setup, and not to evaluate the plant itself. The reference facility used in this study represents a contemporary mechanical WEEE pre-treatment practice for categories including SHA. We measure the performance of the whole product against the recycling system in order to identify hotspots of material losses. This approach provides the full picture of the EoL processing and the loss and recovery of the key materials reported in this study thus reflect the fate of other resources as well. The results above clearly show that new products like RVCs have higher potential of resource recovery than it is being exploited by the existing recycling chain. The WEEE Directive (European Parliament, 2012) requires the member states in European Union to report the recycling rates of each WEEE category. The overall recycling rates are, therefore, calculated and collectively reported at category level by the member states. Such calculations are based on the amount of WEEE collected and the amount sent for treatment, and the information provided by the downstream recyclers on material recovery. However, our finding suggests these recycling rates reported for categories do not reflect the true material recovery of individual products in those categories. In Denmark, for example, the recycling rate for SHA was reported to be 87% for 2012 (DPA-System, 2013) compared to our estimated rate of 47% for RVCs. Higher recycling rates may be achievable for metal-rich WEEE with less complex material composition, but it is evident that not all products under one category can be processed with equal efficiency in the existing recycling chain. This is even more pertinent when there is a significant difference in material composition and design of products. 3.4. Product design and EoL processing Fig. 6 illustrates an example of issues related to the product design in the process of material liberation and separation. On the left, it shows the output fraction from the first overbelt magnet, which should ideally be pure ferrous metals. On the right is an under-shredded RVC that came out almost intact from the preprocessing plant. The inefficiencies of both the shredding process and product design have resulted in such mixed and inconsistent output streams. The plastic brooms in the brush were either weaved or glued to the metal shaft, which made it impossible to separate them by shredding. Moreover, large pieces of plastics, for example, were diverted to this fraction due to a small steel screw. It is because the steel screw did not liberate in the shredding process and was eventually picked by the magnet. Arguably, the incomplete liberation can be related to inadequate shredding of the input, but more intensive shredding is not necessarily helpful. It may reduce the component size, but without the guarantee of separating different materials. For example, the number of pure steel screws in the ferrous fractions was less than the screws found with some plastic component in it (Figure E.1). Similarly, the copper coils and steel casing of an electromotor cannot be separated by using more intense shredding as it results in the loss of magnet in the electromotor, which converts into powder due to the impact. As this powder is diluted in all the outgoing fractions, the probability of recovering the materials in the magnet becomes very low (Habib et al., 2015). Clearly, the product was not designed to perform well in the
658
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Fig. 6. Examples of incomplete size reduction and material liberation. Left: Mixture of different materials and components in the ferrous fraction (F4); Right: An under-shredded RVC piece found in the non-metal fraction (F10).
existing shredding-based EoL processing, or to say the least, there is a mismatch between EoL product type and EoL processing technique. The key design issues affecting the material recovery in the preprocessing steps are summarized below: Immature product design: Classical design errors such as ‘birdnest’ wiring were found in RVCs (Figure E.2). It makes not only a potential manual disassembly process, but also the material separation after shredding, more difficult. Compartmentalization of wires and connector will allow easier recovery of the non-ferrous metals used in them and reduce contamination to other output fractions. Use of connectors: Improper use of connectors (e.g. too many screws e resulting in longer dismantling time and incomplete separation) and incompatible materials (e.g. ferrous steel screws in plastic components e resulting in inefficient material separation) show the lack of ‘design for EoL’ thinking (Fig. 6). A slight design alternative such as using plastic-based snap fit or plastic screw will improve the material separation. Placement of key components: Some components need to be removed from the EoL products before it can be shredded. Removal of battery and PCBs (>10 cm2), for example, is required by the WEEE Directive. Making such components easily accessible by using less complicated connectors and with proper placement of the components will allow quicker removal. Moreover, material recovery from components such as PCBs and electric motors can be improved if they are treated separately. Materials compatibility: Aside from connectors, the issue of material compatibility has also been observed in other components such as broom and wheel. They usually are made of composite of plastic and steel, put together either by weaving or by using glue. This resulted in poor material liberation. Such issues can be avoided, for example, by using compatible materials or using only one material for the component (e.g. use of plastic shaft for the plastic broom/brush). Both the WEEE (European Parliament, 2012) and EcoDesign (European Parliament, 2009) Directives suggest a ‘comprehensive approach’ towards improving the environmental performance of products by means of product design, considering the whole lifecycle that includes the product EoL. However, our case study of an emerging electronic product showed little evidence of EoL thinking in product design. This can be related to, among other reasons, the lack of intensive for the producers and the designers who are not certain of the EoL fate of their product. 4. Conclusions Though the current EoL processing technologies are evolved
from years of learning and improvement by the recycling industry, they are constantly being challenged by the diversity of electrical and electronic products. In this context, achieving higher recovery of materials from EoL products will require a detailed understanding of their performance in the recycling chain. This study comprehensively illustrates the EoL fate of resources used in the case product by taking into account the existing recycling infrastructure. Our findings suggest that, in the best case, the existing recycling chain may recover up to 92% of the base metals from EoL RVCs. However, only 45% of the plastics were estimated to be recycled as material. Further, the conventional ‘shred-and-separate’ approach in the preprocessing of the EoL products resulted in dispersion and eventual loss of components such as PCBs and magnets. Tailored processing based on the understanding of the material composition and design features of the EoL product will improve the material recovery. This case study also exemplifies the lack of ‘EoL thinking’ in the design of a relatively new product. Classical design issues such as bird-nest wiring and incompatible materials and connector types in the product components added to the challenges in the material recovery process. Better recycling rate, for plastics in particular, is possible with improved material compatibility and ease of disassembly. On top of the lack of appropriate recycling technology and product design, the material losses were also caused by the market realities that are driven more by economic than environmental interests. Despite the constant discussion and increasing legislative push, practical implementation of ‘design for EoL’ thinking still remains a challenge. This study offers a robust methodological approach for assessing the EoL performance of a product in the complex material recovery chain and provides the understanding of how the design of a product is linked with the processes in the chain. The methodology and findings from this study can be used to assess the EoL performance of other products and to support the integration of the product design with the EoL resource recovery. Acknowledgement We thank Christina Kaloudaki Pangidou and Tom Ellegaard for their help during the experiment and information collection. The research was a part of the INNOSORT project (innosort.teknologisk. dk) that is funded by the Danish Agency for Science, Technology and Innovation. We thank the Agency and the project partners for their support. The views contained in the paper are those of the authors and do not represent the official views or policies of any stakeholders.
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Appendix A. Product disassembly
Figure A.1. Main Unit (Top & Bottom View).
Figure A.2. Removal of Battery.
Figure A.3. Side Brushes.
659
660
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Figure A.4. Dust Bin.
Figure A.5. Filter.
Figure A.6. Casing.
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Figure A.7. Top Cover.
Figure A.8. Printed Circuit Board.
Figure A.9. Detailed inside View.
661
662
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Figure A.10. Wheels.
Figure A.11. Power Brush Casing & Motor.
Figure A.12. Electromotors.
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
663
Figure A.13. Dismantled electromotors.
Figure A.14. Screws.
Appendix B. Robotic vacuum cleaner
Table B.1 Size of some selected RVC types. RVC Model
Weight (kg)
Dimension (mm) Diameter
Height
Samsung Navibot LG Hom-Bot iRobot Roomba
3.5 3.2 3.8
355 370 355
90 90 80
664
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Appendix C. Output fractions
Figure C.4. Fraction 5. Figure C.1. .Fraction 1.
Figure C.2. Fraction 2. Figure C.5. Fraction 6.
Figure C.3. Fraction 4.
Figure C.6. Fraction 7.
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
665
Figure C.10. Fraction 11.
Figure C.7. Fraction 8.
Appendix D. Key assumptions and calculations
Table D.1 Key assumptions. 1. Recovery Rates
Hammer mill: 95% Plastic recycler: Metal 99%, Plastic 47% Steel Smelter: 95% Copper Smelter: 99% Al Smelter: 95% 2. PCB composition 30% Cu, 70% glass-reinforced polymers 3. Material Market Price Steel/Iron: 0.12, Copper: 5.2 (V/kg) Aluminium: 1.5, Plastics: 1.2
Source: Local recyclers (W€ ager and Hischier, 2013) (Habib et al., 2015) (WRAP, 2014) (Cucchiella et al., 2015)
Figure C.8. Fraction 9.
Table D.2 Material Composition of output Fractions from preprocessing plant.
Figure C.9. Fraction 10.
Fraction
Material composition (kg) Fe
Cu
Al
PCB
Plastic Rest
%
Residues (1 þ 2þ3) Ferrous (4) Motor-trafo (5 þ 6þ9) Plastic Fraction (7 þ 10) Non-Ferrous (8 þ 11) Total (kg)
0.00 18.11 14.32 4.00
0.05 1.42 2.29 4.24
0.00 0.08 0.29 0.71
1.70 0.46 0.71 9.19
0.06 9.02 5.17 178.57
2.8% 10.9% 9.7% 73.8%
6.23 1.91 5.01 13.89
0.12 0.53 1.44 3.74 1.72 0.39 2.8% 36.55 8.53 2.52 15.80 194.54 27.41 285.35 kg (100%)
666
K. Parajuly et al. / Journal of Cleaner Production 137 (2016) 652e666
Table D.3 Final Material Recovery and Value share. Materials /
Fe
Cu
Al
Plastic
Rest
Total Weight (kg) Material Recovered (kg) Material Lost (kg) % Recovered (of total of each material)
36.55 33.73 2.82 92%
13.27 11.46 1.81 83%
2.52 2.33 0.19 92%
194.54 87.13 107.41 45%
38.47 e 38.47 e
285.35 134.65 150.70
Materials market price (V/kg) Recovered Value (V) Lost Value (V) Total Value (V)
0.12 4.05 0.34 4.39
5.2 59.61 9.39 69.01
1.5 3.49 0.29 3.78
1.2 104.56 128.89 233.45
e e e e
Total 171.71 138.91 310.62
Recovered Materials (weight %) Recovered Value (%)
25.0% 2.4%
8.5% 34.7%
1.7% 2.0%
64.7% 60.9%
e e
100% 100%
% 47% 53%
Appendix E. Design features References
Figure E.1. The liberated screws (left) and non-liberated screws with plastic and other metal components (right).
Figure E.2. The bird-nest wiring found in a dismantled RVC.
Basdere, B., Seliger, G., 2003. Disassembly factories for electrical and electronic products to recover resources in product and material cycles. Environ. Sci. Technol. 37, 5354e5362. Chancerel, P., Bolland, T., Rotter, V.S., 2011. Status of pre-processing of waste electrical and electronic equipment in Germany and its influence on the recovery of gold. Waste Manag. Res. 29, 309e317. Chancerel, P., Meskers, C.E.M., Hagelüken, C., Rotter, V.S., 2009. Assessment of precious metal flows during preprocessing of waste electrical and electronic equipment. J. Industrial Ecol. 13, 791e810. Cucchiella, F., D'Adamo, I., Lenny Koh, S.C., Rosa, P., 2015. Recycling of WEEEs: an economic assessment of present and future e-waste streams. Renew. Sustain. Energy Rev. 51, 263e272. DPA-System, 2013. WEEE, BAT, and ELV Statistics Denmark, seventh ed. Euromonitor, 2012. Robotic Vacuum Cleaners Welcomes its Golden Age. Euromonitor International. Euromonitor, 2015. Vacuum Cleaners in Denmark. Euromonitor International. European Parliament, 2009. Directive 2009/125/EC of the European Parliament and of the Council of 21 October 2009 establishing a framework for the setting of ecodesign requirements for energy-related products (recast). Official J. Eur. Union. European Parliament, 2012. Directive 2012/19/EU of the European Parliament and of the Council of 4 July 2012 on waste electrical and electronic equipment (WEEE) (recast). Official J. Eur. Union. Graedel, T.E., Allwood, J., Birat, J.-P., Buchert, M., Hagelüken, C., Reck, B.K., Sibley, S.F., Sonnemann, G., 2011. What do we know about metal recycling rates? J. Industrial Ecol. 15, 355e366. Habib, K., Parajuly, K., Wenzel, H., 2015. Tracking the flow of resources in electronic waste - the case of end-of-life computer hard disk drives. Environ. Sci. Technol. 49 (20), 12441e12449. iRobot, 2015. iRobot ROOMBA Specifications. €hler, A.R., 2013. Challenges for eco-design of emerging technologies: the case of Ko electronic textiles. Mater. Des. 51, 51e60. Lee, H.M., Lu, W.F., Song, B., 1 March 2014. A framework for assessing product EndOf-Life performance: reviewing the state of the art and proposing an innovative approach using an End-of-Life Index. J. Clean. Prod. 66, 355e371. Li, J., Zeng, X., Chen, M., Ogunseitan, O.A., Stevels, A., 2015. “Control-alt-delete”: rebooting solutions for the E-waste problem. Environ. Sci. Technol. 49, 7095e7108. Mayers, K., Peagam, R., France, C., Basson, L., Clift, R., 2011. Redesigning the camel. J. Industrial Ecol. 15, 4e8. Meskers, C.E.M., Hagelüken, C., Salhofer, S., Spitzbart, M., 2009. Impact of preprocessing routes on precious metal recovery from PCs. In: GMBD (Ed.), European Metallurgical Conference, Innsbruck, Austria. Rios, P., Stuart, J.A., Grant, E., 2003. Plastics disassembly versus bulk Recycling: engineering design for end-of-life electronics resource recovery. Environ. Sci. Technol. 37, 5463e5470. Wager, P.A., Hischier, R., 2015. Life cycle assessment of post-consumer plastics production from waste electrical and electronic equipment (WEEE) treatment residues in a Central European plastics recycling plant. Sci. Total Environ. 529,158e167. Wang, F., 2014. E-waste: Collect More, Treat Better: Tracking Take-back System Performance for Eco-efficient Electronics Recycling. Delft University of Technology. Wang, F., Huisman, J., Meskers, C.E., Schluep, M., Stevels, A., Hageluken, C., 2012. The Best-of-2-Worlds philosophy: developing local dismantling and global infrastructure network for sustainable e-waste treatment in emerging economies. Waste Manag. 32, 2134e2146. Widmer, R., Du, X., Haag, O., Restrepo, E., Wager, P.A., 2015. Scarce metals in conventional passenger vehicles and end-of-life vehicle shredder output. Environ. Sci. Technol. 49, 4591e4599. WRAP, 2014. Techniques for Recovering Printed Circuit Boards (PCBs). Waste & Resources Action Programme. €ger, P., Hischier, R., 2013. Life Cycle Analysis of MBA Polymers' Post-consumer Wa Recycled Plastics. Swiss Federal Laboratories for Materials Science and Technology (Empa).