Evaluating the Sustainability of Electronic Media: Strategies for …publicationslist.org/data/lorenz.hilty/ref-212/2014_Hischier_Ahmadi... · vices at the LCI level – and the consequences

This Accepted Author Manuscript is copyrighted and published by Elsevier. The final pub-lication is available via http://www.sciencedirect.com/science/article/pii/S1364815214000103. Suggested citation: Hischier, R., Ahmadi Achachlouei, M., Hilty, L.M.: Evaluating the Sustainability of Electronic Media: Strategies for Life Cycle Inventory Data Collection and their Implications for LCA Re-sults. Environmental Modelling & Software, Vol. 56, pp. 27-36 (2014)

Evaluating the Sustainabil ity of Electronic Media: Strategies for Life Cycle Inventory Data Collection and their Implications for LCA Results

Roland Hischier1*, Mohammad Ahmadi Achachlouei1,2,3, Lorenz M. Hilty1,3,4

1 Empa Swiss Federal Laboratories for Materials Science and Technology, Technology and Society Lab, St. Gallen (Switzerland) 2 KTH Royal Institute of Technology, Division of Environmental Strategies Research (fms), Stockholm (Sweden) 3 KTH Royal Institute of Technology, Centre for Sustainable Communications (CESC), Stockholm (Sweden) 4 University of Zürich, Department of Informatics, Zürich (Switzerland)

* Corresponding author: [email protected] / Empa, Technology & Society Lab, Lerchenfeldstrasse 5, 9014 St. Gallen, Switzerland / phone +41 58 765 7847

Abstract

This paper compares two Life Cycle Assessment (LCA) studies independently carried out to assess the environmental impacts of electronic versus print media. Although the two studies lead to the same overall conclusion for the case of a news magazine – namely that the tablet version of the magazine has environmental advantages over the print version – there are significant differences in the details of the LCA results. We show how these differences can be explained by differences in the methodologi-cal approaches used for Life Cycle Inventory (LCI) modelling, in particular the use of rough average data versus the attempt to use the most specific and detailed data as

R. Hischier, M. Ahmadi Achachlouei, and L.M. Hilty

2

possible. We conclude that there are several issues in LCA practice (at least when applied in the domain of media) that can significantly influence the results already at the LCI level: The data collection strategy used (e.g. relying on desk-based research or dismantling a given device) and the decisions made at inventory level with regard to parameters with significant geographic variability, such as the electricity mix or recycling quotas.

Keywords

Life Cycle Assessment, Life Cycle Inventory Modelling, Environmental Impact, Elec-tronic Media, Electronics Industry, Print Media, Recycling, Sensitivity Analysis

1. Introduction

Electronic media are increasingly replacing print media. Mobile end-user devices such as laptop and tablet computers, smartphones and e-book readers, usually combined with mobile internet access, have become serious competitors of print media (news-papers, printed books) as well as traditional TV in their role of providing daily news and entertainment. The trend towards high-definition displays and the increasing bandwidth of internet access networks is unleashing the market potential of on-demand content provision and leading us into a new era of highly personalized and “smarter” media use.

Replacing paper by information and communication services and introducing “pay per use” models instead of ownership have been part of early visions of a sustainable in-formation society (Dompke et al., 2004). However, this development has both posi-tive and negative environmental impacts. On the positive side, there is a potential to reduce paper consumption. Development in the area of display technology lead to a replacement of Cathode-Ray Tube (CRT) displays, which used to be the most en-ergy- and material-intensive parts of TV sets and desktop computers. On the nega-tive side, there are a large number of new end-user devices that are produced, used and then disposed of after a relatively short service life, combined with an increase of data traffic. Studies on the overall impact of Information and Communication Technologies (ICT) on the environment, especially on greenhouse gas emissions, usually treat the trends towards electronic media as a part of the dematerialization


3

effect that results from a shift from material goods to services (Hilty et al., 2004; Hilty et al., 2006; Erdmann and Hilty, 2010).

Life Cycle Assessment (LCA) is the method of choice to assess and compare the en-vironmental effects of traditional and electronic media. LCA is a methodology to as-sess the potential environmental impacts and resources used throughout a product’s life-cycle, i.e., from raw material extraction, production and use to waste treatment (ISO, 2006) – and is considered the most established and best-developed methodol-ogy in the area of sustainability assessment (Finnveden et al., 2009). Already two decades ago, this method has been recognized as an opportunity also for the elec-tronics industry to identify areas for improvement (Rhodes, 1993). The topic gained importance when the electronics industry became one of the fastest growing eco-nomic sectors and the problem of managing waste electrical and electronic equip-ment started to damage the image of electronics as a “clean” technology (Cui and Forssberg, 2003; Widmer et al., 2005; Hilty, 2005). In recent years, the long-term availability of some scarce metals used in electronics production became an issue as well (see e.g. Wäger et al., 2010; Manhart, 2011; Dodson et al., 2012), fostering the interest in LCA studies of electronic products. Electronics LCA plays a crucial role in the environmental assessment of other products containing electronics, such as the compact fluorescent lamp replacing banned light bulbs (Welz et al., 2011).

In the last two decades, LCA has been repeatedly used to assess specific end-user media devices or services, comparing them often with paper-based media (e.g., Reichart and Hischier, 2001; Reichart and Hischier, 2003; Duan et al., 2009; Hischier and Baudin, 2010; Weber et al., 2010; Moberg et al., 2010; Moberg et al., 2011; Achachlouei and Moberg, 2013; Achachlouei et al., 2013a; Malmodin et al., 2013). However, despite these LCA studies (a more complete overview can be found in An-drae and Andersen, 2010), there is still a lack of life cycle inventory (LCI) back-ground data in this domain.

This paper summarizes and compares two different studies of modern ICT media de-vices at the LCI level – and the consequences of the choices made at this level to the resulting data in an application context. After a short overview of the so far ex-isting LCA/LCI studies in the area of modern ICT media devices (section 2.1), the two studies used for comparing the two approaches are described (sections 2.2 and 2.3). In section 3, we present the actual comparison between these two studies – first at the level of the device, a tablet computer, and in the second part in an appli-cation case comparing a print and an electronic edition of a magazine. Finally, our


4

conclusions from this comparison, together with an outlook identifying open issues, are presented in section 4.

2. Life-Cycle Inventory Data of Electronic Media Devices

2.1 Overview

The LCI step in the LCA method involves creating an inventory of flows from and to nature caused by a technical system, called the product system. The technical sys-tem is modelled based on general knowledge and specific data on the relevant pro-cesses. The development of such inventories in the area of electronic media devices began in the late 1990s with first studies of desktop personal computers (PCs) – followed later by studies of various other devices (such as television set, laptop or tablet computers). The most frequently cited LCA study of a desktop PC is the 1998 study by Atlantic Consulting (Atlantic Consulting and IPU, 1998), although the very first such study was published a year earlier (Tekawa et al., 1997). In 2003, von Geibler and co-workers published a first study that addressed “mobile comput-ing” (von Geibler et al., 2003); this study served as a main source for the laptop computer dataset in ecoinvent (Hischier et al., 2007). In the following years, various LCA studies about desktop and mobile computing devices have been published (Hik-wama, 2005; Kemna et al., 2005; Lu et al., 2005; Duan et al., 2009; Song et al., 2013), which were all using traditional process LCA methodology. In 2004, Williams established a study on a desktop PC combining process LCA and Input-Output-Modelling (Williams, 2004), and Deng and co-workers presented a similar study for a laptop (Deng et al., 2011). St-Laurent and co-workers recently presented an ap-proach to eco-labelling of laptops (St-Laurent et al., 2012). The most recent publi-cations do not focus any more on the impact of end-user devices alone, but put them into a broader context (e.g. Andrae, 2013; Maga et al., 2013) or cover social issues as well (e.g. Ciroth and Franze, 2011; Ekener-Petersen and Finnveden, 2013).

Looking at television devices, a first LCA study has been published in 1998 (Thomas et al., 1998). In 2003, Aoe presented a model for the calculation of the eco-efficiency of electronic devices, using various television devices as examples (Aoe, 2003) and the EC-JRC’s Institute for Prospective Technological Studies (IPTS) pub-lished a report on the eco-design of such devices (IPTS, 2003). The plasma technol-ogy has been investigated by a diploma thesis at Empa (Baudin, 2006; Hischier and


5

Baudin, 2010). More recently, Feng and Ma published a study investigating color TV sets in China (Feng and Ma, 2009).

Only a few studies can be found about more modern devices such as tablets, smart-phones or e-book readers. Among the first studies is the work by Faist Emmenegger and colleagues, adressing the 3rd generation UMTS mobile network as a whole (Faist Emmenegger et al., 2006). Further studies in this area are e.g. a footprint study of mobile phones (Frey et al., 2006), a study of mobile communication as whole (Fehske et al., 2011), as well as two studies focussing mainly on the end-of-life phase in this area (Scharnhorst et al., 2006; Takahashi et al., 2009). Concerning tablet computers, first studies, based on some simplifications, have been published about four years ago (Deetman and Odegard, 2009; Moberg et al., 2010).

However, by far not all of the studies mentioned present the inventory data used for modelling the end-user devices in a transparent manner; a prerequisite for the LCA community to review and reuse such data. Publicly accessible LCI databases that have been established since the end of the 1990s, such as the ecoinvent database (ecoinvent Centre, 2003) or the ELCD database of the EC Joint Research Centre (EC-JRC), Institute for Environment and Sustainability (JIES), hardly contained any data of the electronics industry in their initial versions. Version 2 of the ecoinvent database (published as version “v2.01” in 2007) has been the first publically acces-sible and transparent LCI database covering processes for electronics production, use and disposal on the level of components and modules1. Data about end-user de-vices, however, are even in ecoinvent very limited, covering one desktop computer with CRT or LCD screen and one laptop computer only (Hischier et al., 2007). The focus of the electronics data in ecoinvent is rather on the electronic components than on whole devices, in order to enable a broader applicability of the data in this fast-changing sector.

Across all the cited studies, two basic strategies for getting LCI data for electronic devices can be distinguished:

• The “desk-based” strategy, using only existing data sources such as product declarations and roughly adapting the data (e.g., by extrapolation or interpola-tion) to a specific case when necessary (examples for this strategy are e.g. Lu et al., 2005; Duan et al., 2009; or Ciroth and Franze, 2011).

1 Two other LCI databases containing extensive information on electronics products are GaBi and EIME – but due to the high price of the data, they aren‘t considered public databases in this article.


6

• The “lab-based” strategy that involves dismantling one or more specific de-vices to generate detailed data for a specific case (examples for this strategy are e.g. Tekawa et al., 1997; Hikwama, 2005; or Baudin 2006).

Both strategies usually link their data about a device with background data on the manufacturing of components (i.e. average data about the production of standard electronic components, e.g., from ecoinvent).

In 2012, two projects at Empa (Switzerland) and at KTH Royal Institute of Technol-ogy (Sweden) on electronic and printed media were started independently, using the two different strategies to achieve a similar goal. This gives us the opportunity to investigate ex post how the two data collection strategies influenced the LCA re-sults. The following two sections (i.e. section 2.2 and 2.3) describe the two projects and the data used in more detail.

2.2 Example 1: Using a Desk-based Strategy

In a project in 2011-2012 on behalf of the “Denkfabrik visuelle Kommunikation” of viscom (the Swiss Association for Visual Communication), Empa’s Technology and Society Lab, to-gether with the Department of Informatics (IFI) of the University of Zürich, developed the web-based tool “mat” (media analytics tool), designed to ev-aluate the environmental effects of a broad variety of printed and electronic media for use patterns interactively defined by the user (Hischier et al., 2013). In a first part of this study, “classic” cradle-to-gate LCA datasets have been established for various electronic devices used by consumers in 2012. Based on published compo-sition data from various producers, composition tables of the various electronic me-dia devices considered were compiled, ranging from a smartphone up to a television device. More details about the examined devices and their data sources can be found in Hischier et al., 2013. Table 1 shows – for the example of the LCD tablet com-puter, as this device is the common denominator with the study from KTH (de-scribed in section 2.3) – how the information extracted from the source (i.e. Apple, 2011) has been linked with the data from ecoinvent 2.2 (ecoinvent Centre, 2010) as background database. Then all input data (e.g., data of electronic components and modules, basic materials, material processing, etc.) for the various devices were taken from ecoinvent 2.2, which provided all background data for the study.


7

Table 1. Example of linking project-specific data with background data from

the database ecoinvent v2.2 – shown for a 10-Inch LCD tablet computer.

Component Amount Linked with the following ecoinvent datasets to build the LCI model

Aluminium 135 g aluminium production mix & sheet rolling, aluminium Plastics 17 g 50% ABS, 50% PUR, rigid – plus 100% injection

moulding Battery 130 g LiIon-Battery Circuit Boards 38 g Printed wiring board, mounted, laptop computer Display 140 g LCD module Glass 105 g flat glass, coated

Other metals 25 g 50% copper (+ wire drawing, copper) 50% steel, unalloyed (+ profile extrusion, steel)

Charger 50 g 0.11 of dataset ‘power adapter, for laptop’ 0.312 kWh Electricity, medium voltage (China) Production

efforts 303 kg Tap water & sewage water treatment

As further described in Hischier et al., 2013, for the end-of-life treatment an up-to-date WEEE2 processing system is assumed. For all devices examined in this study a manual depollution (according to the rules shown in Tab.4.6, part V of Hischier et al., 2007), followed by a mechanical treatment in a shredder (according to the proced-ure described in chapter 4.3.5, part V of Hischier et al., 2007) is modelled. For the metals (aluminium, copper, steel) recovered in the subsequent recycling procedure, an avoided burden approach (giving credits equal to the impacts of the respective primary production) is applied.

Figure 1 shows the environmental impacts measured with the ReCiPe method (Goed-koop et al., 2012) on the midpoint and endpoint levels3 across the complete life cycle of the tablet. For this calculation the default use pattern of a tablet, as defined in consent with with the advisory board of the mat project, is used here: a total life span of 2 years, 2 hours/day active use in Switzerland over the whole life span, and the remaining 22 hours/day in the sleep mode (see also Table 5 in Hischier et al., 2013). Additionally, a data download rate of 100 MB/day is assumed (representing

2 Waste Electrical and Electronic Equipment 3 Midpoint: Hierarchist perspective (H); Endpoint: Average European Hierarchist perspective (H/A)


8

about 10% of a typical daily download rate of a private household based on expert information, cited in Achachlouei, 2013).

Figure 1. Upper part: Environmental impact of a Tablet computer, used during 2 years (2h/d). Lower part: Environmental impact of the Production only. Shown are the ReCiPe midpoint indicators Fossil Depletion Potential (FDP), Metal Depletion Potential (MDP), Global Warming Potential (GWP), Terrestrial Acidification Potential (TAP), Freshwater Eutrophication Potential (FEP), Photochemical Oxidant Formation Potential (POFP), Ozone Depletion Potential (ODP), Human Toxicity Po-tential (HTP), Freshwater Ecotoxicity Potential (FETP), Terrestrial Ecotoxicity Potential (TETP) and the ReCiPe endpoint indicators damage on Human Health (HH), damage on Ecosystem Quality (EQ), and Resource Consumption (Res).

As shown in Figure 1, the production of the device is clearly dominating the life-cycle in all environmental indicators examined, representing in all almost cases 60 % and more of the total impact. Within the production, the electronics components – i.e. the printed wiring board and its components (ICs, resistors, capacitors, etc.) is clearly responsible for the main impact (always above 75%, often above 90%). For the end-of-life step, the avoided burden approach used in this study results in most impact categories in an overall environmental benefit (i.e., a negative burden).

2.3 Example 2: Using a Lab-based Strategy

A Licentiate thesis, carried out in 2011-2012 at the Centre for Sustainable Com-munications (CESC) of the KTH Royal Institute of Technology assessed the envi-


9

ronmental impacts of a magazine in print and tablet versions, the latter of which was read from a tablet with an LCD screen (Achachlouei, 2013). The tablet device mod-elled in the study was an assumed generic tablet similar to an iPad 2. Size and weight of the tablet are as follows: height 241.2 mm; width 185.7 mm; depth 8.8 mm; weight 612 g. To model the inventory of the production of the tablet, an iPad 2 was obtained and disassembled into the pieces that could be identified with the sup-port of Apple’s technical specification4. The weight and size of the disassembled parts were measured. The manufacturing of the identified components was modelled using the ecoinvent database (Hischier et al., 2007), although the data included in ecoinvent did not represent the exact components and production processes used for the tablet under study. However, as contacting Apple for more detailed data on the environmental impacts of the manufacturing was not successful, these data from ecoinvent were considered the best available alternative. The various datasets used, as well as missing data, are further specified in Achachlouei, 2013.

Information about possible end-of-life treatment of the tablet was gathered from an electronics recycling company, examining the disassembled iPad. It was assumed that 20% of the discarded tablets are not processed in a WEEE recycling system, but directly go to municipal waste incineration plants. To model these 20%, an incin-eration dataset for equipment were chosen from the ORWARE study (described in Dalemo et al., 1997) as implemented in Arushanyan et al., 2013, for Sweden. Two types of incineration plants were used: heat only boiler (HOB) and combined heat and power (CHP) plants. It was assumed that 65% of the waste equipment was in-cinerated in CHP and 35% in HOB (Svensk Fjärrvärme, 2010). It was estimated that approximately 51 weight-% of the tablet can be directly recycled, and 49 weight-% goes to a mechanical process in which the remaining aluminium (about 15 weight-% of the panel module) can be taken out for material recycling and the residuals go to an energy recovery process and to landfill5. Table 2 presents the parts and materials used in the tablet, and the disposal treatment chosen for each one.

4 Personal communication with Patrik Daijavad, PDJ Development, 2011. 5 Personal communication with Sverker Sjölin, Stena Technoworld, 2011.


10

Table 2. Weight of components/modules of the examined tablet computer.

Material / component Weight Unit Weight-% of total

End-of-Life treatment

Display module (LCD mod-ule) 145 g 23.7 To mechanical treatment

Aluminium Back Panel 140 g 22.9 To recycling Battery 135 g 22.0 To recycling

Glass Panel Sheet 109 g 17.8 To mechanical treat-ment/MSWI

Circuit Boards (and connec-tors) 39 g 6.44 To recycling

Other Metals 26 g 4.24 Plastics 18 g 2.88 To recycling Total 612 g

Figure 2 shows the environmental impacts of the complete life cycle of the modelled tablet computer – as in Figure 1 measured with the ReCiPe method (Goedkoop et al., 2012) on the midpoint and endpoint level. This calculation is based on the default use pattern defined from Achachlouei and co-workers in the framework of their work – i.e. a total life span of 3 years for the tablet and a reading time of 14 hours/week (see Achachlouei et al., 2013b). Similar as for the results shown in Figure 1, a data download rate of 100 MB/day is assumed.


11

Figure 2. Upper part: Environmental impact of a Tablet computer, used during 3 years (2h/d). Lower part: Environmental impact of the Production only. The same impact factors are shown as in Figure 1.

Similar to the results shown in Figure 1, the production of the device is clearly domi-nating the life-cycle in (almost) all environmental indicators, representing 60% and more of the total impact. The only exception is TETP, the terrestrial ecotoxicity po-tential. This impact factor is dominated here by the disposal of ashes after combus-tion of natural wood chips at cogeneration units to produce electricity, which has a much higher share in the Swedish than in the Swiss electricity mix, which was used for the calculations in Figure 1. Within the production, which is modelled in much more detail here, the printed wiring boards are responsible for the main part of the impact; followed by the LCD module and the batteries; a result that is consistent with the results shown in Figure 1.

3. Comparison of the two Approaches

3.1 Reference for Comparison

In order to examine the influence of the data collection strategies used in the two studies, we align all the other aspects in which the studies differ to the highest de-gree possible by selecting the “reference scenario, mature” reported in Achachlouei


12

et al., 2013b, as the reference for both studies. Table 3 summarizes these aspects and the common values used in order to compare the two studies.

Table 3. Key values for the comparison of the studies (using the print and the electronic version of a

magazine as case study). Data taken from Achachlouei et al., 2013b, if not otherwise mentioned.

Aspect Unit Value Remarks (i) Tablet edition

• Total lifetime of tablet Years 3

• Daily active use of tablet h/d 2 further 22 h/d: “stand-by”

• Reading time for magazine min 41

• Data size of magazine down-load MB 163

• Electricity for Download - National Swiss for the Empa study / Swed-ish for the KTH study

(i i) Print edition

• Paper size - A4 i.e. 297 x 230 mm

• Paper weight g/m2 79.4 Calculated back from total weight of one copy of magazine

• Paper type - LWC LWC = Lightweight coated

• Printing technology - reel-fed offset

• Number of reader per copy pers. 4.4

• Average Transport distance km 298 Calculated back from information given in Achachlouei, et al., 2013b

Based on this information we reconstructed the reference scenario using Empa’s mat tool. These new results are compared in the following two sections with the original results from the KTH study (Achachlouei et al., 2013b) on two levels – the level of the electronic device only (section 3.2), and the level of the comparison between the print and the electronic edition of the modelled magazine (section 3.3).

3.2 Comparison at Device Level

The results for the the comparison of the two studies at device level are shown in Figure 3 for reading the electronic edition (i.e. for a reading time of 41 minutes) for a selection of 6 different ReCiPe midpoint indicators. The selected midpoint indica-tors were chosen to represent a reasonable mix between issues relevant in the con-


13

text here (such as resource depletion) and basic issues (such as global warming or acidification).

Figure 3. The environmental impact of reading one issue of an online magazine on a tablet computer (based on the figures in Achachlouei, 2013), calculated based on two different LCI models that were constructed with a desk-based and with a lab-based strategy, respectively. Six selected ReCiPe indicators are shown.

The comparison in Figure 3 shows for 4 out of the 6 impact factors (i.e. Global Warming, Terrestrial Acidification, Freshwater Eutrophication, and Photochemical Oxidants Formation) almost similar overall results. However, at the detailed level, the results are in agreement only for one impact factor (Global Warming). In 3 out of the 4 cases with overall similarities, the desk-based Empa model calculates a higher im-pact of the data transfer, while in the lab-based KTH study the production of the device is much more dominant.

These are two differences that only by accident roughly compensate for each other at the level of the whole life cycle. The first difference can be explained by different electricity mixes assumed behind the data transfer. In the Empa study, a part of the data transfer is assumed to use average European electricity, whereas the Swedish study assumed the cleaner Swedish electricity mix. This difference has no connec-


14

tion to the issue of data collection strategies, but may in itself represent a relevant observation. It shows the high sensitivity of the assumed electricity mix in such con-texts, also for impact factors other than Global Warming.

The second difference, substantial deviations in the impact of production, can how-ever be explained by the different data collection strategies used for the LCI model-ling. The lab-based strategy, starting with the dismantling of a real device, leads to a complete coverage of the components. In the desk-based strategy, high uncertainty is introduced due to the limited number of different types of mounted printed wiring boards for modern devices that can be found in the accessible databases – which is in the case of ecoinvent 2.2 one dataset for the mainboard of a laptop computer. Then, using this dataset may introduce a bias because the density could be much higher in a small mobile device such as a tablet than in a laptop computer, due to the much more limited space in the device. And as the IC production represents a relevant part of the overall impact of an ICT device, this may lead to significant devi-ations.

For the two other midpoint factors, Metal Depletion and Fossil Depletion, Figure 3 shows rather big differences between the two modelling approaches with a much higher value for Metal Depletion and a much lower value for the Fossil Depletion cre-ated with the lab-based strategy. These differences can be traced back to the same underlying reasons as above. Metal Depletion is dominated by the tablet’s produc-tion, which is modelled in more detail in the lab-based KTH study, which leads to a higher impact despite the fact that both studies were using the same background data about the production of components. Regarding the data transfer, the two models use slightly different approaches to allocate home networking electricity use to the reading of the electronic version, which results in higher electricity consump-tion for data transfer in the KTH study. Then, electricity consumption is connected to metal depletion via the use of copper for the power grid. The higher Fossil Deple-tion calculated with the mat tool is connected to the data transfer step in the life cycle, which is modelled with the UCTE electricity mix, including a high share of fossil energy, in the Empa study and with the Swedish electricity mix in the KTH study.

Limitations on data for electronic devices are also relevant to the components data used to model the tablet manufacturing, i.e. data being rather old and not exactly describing the components in the study. As in both studies no better data could be found, they used in most of the cases ecoinvent as best estimate, which means that there could be systematic errors shared by the two studies. It is interesting that the


15

value provided by Apple for the iPad2 production greenhouse gas emissions is 63 kg CO2-Eq per device (including retail/shipping packaging), while the tablet modelled in the KTH study, based on the dismantling of an iPad2, yields 36.2 kg CO2-Eq. This lower carbon footprint may be due to specific components which might have a more energy-demanding manufacturing process than the generic components available in the ecoinvent database. Also there were some data missing in modelling the compo-nents of the tablet, including the capacitive touch film and flex film (multi-touch screen type). For the assembly process both studies used the data on electricity use from the laptop assembly in ecoinvent. This again indicates that completeness in covering the components of the device – as opposed to focus on components the modeller are aware of and for which they find data – may be a sensitive issue in LCA of complex electronic devices.

3.3 Comparison at the Application Level

In the second step, the consequences of these two data collection strategies to the results of a comparison between the paper and the tablet edition of a magazine are shown. The original purpose of both studies has been the comparison of envi-ronmental impacts of print media with those of electronic media, given various scen-arios of content delivery and consumption. Figure 4 shows the impact of reading a printed copy of a magazine in comparison to the impact of reading it on a tablet computer, modelled with the two approaches, based on the key figures shown in Table 3.


16

Figure 4. The environmental impact of reading a magazine (print version vs. tablet version). Impact per reader of a copy, calculated with the two approaches described in this paper. The same indicators are shown as in Figure 3.

This comparison of print and tablet edition leads for 4 indicators (i.e. Global Warm-ing, Fossil Depletion, Terrestrial Acidification and Photochemical Oxidant Formation) to similar conclusions with both strategies (i.e., the impact of print version is higher than that of the electronic version, if read on this tablet). However, the results are – again – contradictory for the two remaining indicators (Metal Depletion, Freshwater Eutrophication). For Metal Depletion, one reason for the difference is the topic “pa-per production”. In the Empa study, print shows a higher impact (than the electronic version), where about 25% of the impact come from the sulphate pulp production, 20% from the infrastructure and further 20% from the consumed electricity (aver-age European electricity mix) and the rest from other processes. The KTH study pre-sents a much lower result for paper production. At the same time, the impact of the tablet production is assessed as being much higher for the reasons explained before, which together makes the print version roughly a factor of 6 more efficient in terms of Metal Depletion in the KTH study. The Empa study suggests a small advantage for the tablet here. In the case of Freshwater Eutrophication, it is noticeable that the


17

print version gets a negative impact value in the KTH study. This can be explained by the way how the benefits are calculated for the recycled paper amount. The model assumes that the collected magazine paper goes into a recycling paper mill in Sweden producing newsprint paper, and that this recycling paper – sold on the Euro-pean market – replaces average (European) newsprint paper that is produced from wood. The avoided burden – i.e. the high benefit shown as the negative part of the bar – is dominated by the long-term impact of lignite mining for the UCTE electricity consumed in the production of average European newsprint paper.

Furthermore, there are some systematic differences across all 6 impact indicators. First, it can be seen that the results for the print version of the magazine show clearly lower impacts for the paper production in the KTH study. The reason for this difference is the fact that the KTH model is based also in this area on specific data, i.e. data from one paper mill for the production of one specific type of magazine pa-per representing the situation around 2010, while the Empa approach is using aver-age European LWC (lightweight coated) paper, according to Hischier, 2007, repre-senting the situation around the year 2000. As in the case of the tablet computer, the strategy used in the KTH study is to investigate the specific case, while in the Empa study the strategy is to use average data. The latter is problematic here be-cause the average data available is much older and no attempt was made to ex-trapolate the efficiency improvements in the paper industry since the year 2000, which led to lower auxiliary consumption levels and lower emission levels in 2010. On the other hand, there is high variability within the paper industry and using the data from one specific mill may also create a bias. In the case of the KTH study, it is a logical consequence of the aim of the study, which was to investigate a specific magazine printed on paper produced in a specific mill. Generalizing the results re-quires great caution because the specific case may not be representative for the universal set the reader has in mind. By contrast, the aim of the Empa study was to create a tool providing an estimate for the average case where specific data are not available.

Across all indicators, we also observe a difference in the benefits resulting from the assumed end-of-life treatment of the magazine (i.e. the paper recycling). Although both approaches are based on a similar ‘avoided burden’ approach for the paper (as-suming that the paper replaced by recycling paper is newsprint paper), the propor-tion going into the recycling loop is 67% in the data used by Empa (the remaining 33% being incinerated), whereas the KTH study is based on the Swedish paper re-


18

cycling quota of roughly 95%. This difference of almost 30% more paper to recycl-ing has significant consequences for each of the midpoint indicators (via the benefit in the lower part of each bar diagram).

4. Discussion and Conclusions

We compared two studies that were independently conducted for the purpose of as-sessing the environmental impacts of a magazine, in particular to make a statement about the comparative environmental advantages and disadvantages of a print ver-sion vs. an electronic version of the magazine or similar media. The study done at Empa was based on rough average data collected by pure desk research. In contrast, the approach of the study done at KTH was to model the specific characteristics of the case investigated, in particular by dismantling a specific tablet device and by using the data about paper production in a specific paper mill.

Although the two studies lead to the same overall conclusion, namely that the tablet version of the magazine has advantages over the print version (in the assumed re-ference scenario, which has been the same for our comparison of the studies), there are surprising and significant differences in the details of the LCA results. These dif-ferences exist despite the fact that the two studies were using the same back-ground data from the ecoinvent database.

We did not quantify the differences between the studies, because this would pre-tend a level of accuracy that is not warranted, and did a qualitative comparison in-stead. We did neither do any uncertainty calculation in the framework of this com-parison work here. The main reason is the fact that each of the two studies has the most relevant uncertainty on a different level. While in the KTH study, the main point is the way how the identified (and counted) electronic components are actually modelled with respective (background) LCI data; in the Empa study, the crucial point is how adequate the composition of the available PWB datasets is for the specific case – here a LCD tablet. These two types of uncertainty can’t be quantified with the currently existing models in LCA (e.g. the pedigree approach of ecoinvent, re-ported in Frischknecht et al., 2007). In the same time, both of these uncertainties can vary considerably and achieve probably similar order of magnitudes … and in that sense, could be omitted in the overall evaluation (as not being more relevant in either of the two studies).


19

From the explanations for the differences we found, we can derive the following conclusions for life cycle inventory modelling in the media domain:

1. Using a “lab-based” strategy to model an ICT device, dismantling the device and then searching data for each component, can make a significant difference in comparison to a “desk-based” strategy trying to create the composition table by extrapolating data from existing sources only. The lab-based approach reduces the uncertainty regarding the actual IC density. In our case, it lead to a signifi-cantly higher production impact due to the more complete coverage of the com-ponents.

2. Focusing on a specific production plant or using average data instead can lead to substantial differences in domains where there is high variability, such as fresh fibre paper production. It depends of the purpose of the study (i.e., the intended generalizability of the conclusions) which strategy is more appropriate. Modellers also should be aware of a potential error when using older data in domains where there is considerable technological development.

3. The electricity mix assumed may affect the results substantially. Even in our case, where the use of the print magazine does not consume any electric energy and delivering and reading the tablet magazine only a very small amount, the electricity mix has turned out to be a sensitive parameter: Electricity dominates the impact of the data transfer (an issue with very inhomogeneous results pro-vided in literature, see Coroama et al., 2013), and it may have a substantial ef-fect on the impact of paper production and the avoided burden calculated for re-cycling. The latter is important to point out because the addressee of the results of an LCA study may not be aware of the sensitivity of this type of assumptions behind the data, which have nothing to do with the electronic device or paper that deliver the actual content. The fact that different parts of the life cycle and – more difficult to see – the processes for which an avoided burden is calculated, are located in different geographic regions in reality, where different electricity mixes are given, makes the life cycle inventory highly sensitive to explicit or im-plicit geographic assumptions.

4. Recycling quotas should not be treated as rough estimates, because they may have a significant effect on the overall result. The difference between 67% and 95% of a magazine’s paper going into recycling can, for specific impact catego-


20

ries and under particular conditions, create substantial variations for the print version of the magazine.

Many of the issues found in the comparison of the two studies could have been de-tected by including more systematic sensitivity analyses. How to systematically test the sensitivity of assumptions on the geographic location of processes in LCA (e.g., with implications for electricity mix, recycling quotas, avoided burden) – a methodol-ogy of “spatial sensitivity analysis” – could become a topic for future research.

Our results also show that the adequacy of strategies for LCI data collection and modelling depends on the aims and scope of a study. For studies with the focus on a specific ICT device, a lab-based strategy seems most adequate. If the aim is to make general statements, e.g., about the environmental impact of print vs. electronic me-dia, a desk-based approach using average data that is not older then the technology in use may be sufficient.

Acknowledgements

The authors would like to thank the ‘Denkfabrik visuelle Kommunikation’ and its managing director, Rudolf Lisibach, Otelfingen/Switzerland, for supporting the devel-opment of the mat tool, and Michael Keller for implementing the tool as a part of his studies at the Department of Informatics, University of Zürich. With regard to the study conducted by the second author at KTH, we would like to thank Patrik Dai-javad for contributing to data gathering and component inventorying of the tablet device, and greatly acknowledge the support of the Centre for Sustainable Com-munications (CESC), a Vinnova Centre of Excellence at the KTH Royal Institute of Technology, Stockholm/Sweden. Finally, we want to thank the three anonymous re-viewers for their constructive feedback on the first version of this article.

References

Achachlouei, M.A., 2013. Environmental Impacts of Electronic Media: A Comparison of a Magazine's Tablet and Print Editions, Licentiate Thesis, School of Architecture and the Built Environment, Department of Urban Planning and Environment, Division


21

of Environmental Strategies Research (fms). KTH Royal Institute of Technology: Stockholm (Sweden).

Achachlouei, M.A., Moberg, A., 2013. Life cycle assessment of a magazine - part 2: A comparison of print and tablet editions. Journal of Industrial Ecology submitted.

Achachlouei, M.A., Moberg, A., Hochschorner, E., 2013a. Climate Change Impact of Electronic Media Solutions: Case Study of the Tablet Edition of a Magazine, In: Hilty, L.M., Aebischer, B., Andersson, G., Lohmann, W. (Eds.), ICT4S 2013: Proceedings of the First International Conference on Information and Communication Technologies for Sustainability: ETH Zürich (Switzerland), pp. 231-236.

Achachlouei, M.A., Moberg, A., Hochschorner, E., 2013b. Life cycle assessment of a magazine - part 1: tablet edition in emerging and mature states. Journal of Industrial Ecology submitted.

Andrae, A.S.G., 2013. Comparative micro life cycle assessment of physical and vir-tual desktops in a cloud computing network with consequential, efficiency, and re-bound considerations. Journal of Green Engineering 3 193-218.

Andrae, A.S.G., Andersen, O., 2010. Life cycle assessments of consumer electronics — are they consistent? Int J LCA 15(8) 827-836.

Aoe, T., 2003. Case Study for Calculation of Factor X (Eco-Efficiency) - Comparing CRT TV, PDP TV and LCD TV, EcoDesign Conference: Tokyo (Japan).

Apple, 2011. iPad 2 Environmental Report. Apple Inc. : Cupertino, CA (USA).

Arushanyan, Y., Björklund, Y., Eriksson, O., Finnveden, G., Ljunggren Söderman, M., Sundqvist, J.-O., Stenmarck, A., 2013. Environmental Assessmen of Waste Policy In-struments in Sweden [in progress].

Atlantic Consulting, IPU, 1998. LCA Study of the Product Group Personal Computers in the EU Ecolabel Scheme.

Baudin, I., 2006. Life Cycle Analysis of a Plasma Display [Analyse de cycle de vie d'un écran plasma] comem+. Haut Ecole d'Ingénierie et de Gestion du Canton de Vaud: Yverdon (Switzerland).

Ciroth, A., Franze, J., 2011. LCA of an Ecolabeled Notebook. Consideration of Social and Environmental Impacts Along the Entire Life Cycle. GreendeltaTC GmbH, Berlin (Germany).

Coroama, V., Hilty, L.M., Heiri, E., Horn, F., 2013. The Direct Energy Demand of In-ternet Data Flows. Journal of Industrial Ecology 17 (5) 2013, 680–688.


22

Cui, J., Forssberg, E., 2003. Mechanical recycling of waste electric and electronic equipment: a review. Journal of Hazardous Materials B99 243-263.

Dalemo, M., Sonesson, U., Björklund, A., Mingarini, K., Frostell, B., Jönsson, H., Ny-brant, T., Sundqvist, J.-O., Thyselius, L., 1997. ORWARE – A simulation model for or-ganic waste handling systems. Part 1: Model description. Resources, Conservation & Recycling 21(1) 17-37.

Deetman, S., Odegard, I., 2009. Scanning Life Cycle Assessment of Printed and E-paper Documents based on the iRex Digital Reader. CML - Insitute for Environmental Sciences, Leiden University / i Rex Technologies: Leiden (the Netherlands).

Deng, L., Babbitt, C.W., Williams, E.D., 2011. Economic-balance hybrid LCA extended with uncertainty analysis: Case study of a laptop computer. Journal of Cleaner Pro-duction 19(11) 1198-1206.

Dodson, J.R., Hunt, A.J., Parker, H.L., Yang, Y., Clark, J.H., 2012. Elemental sustaina-bility: Towards the total recovery of scarce metals. Chemical Engineering and Pro-cessing: Process Intensification 51 69-78.

Dompke, M., Von Geibler, J., Göhring, W., Herget, M., Hilty, L.M., Isenmann, R., Kuhndt, M., Naumann, S., Quack, D., Seifert, E., 2004. Memorandum Nachhaltige In-formationsgesellschaft. Fraunhofer IRB Verlag, Stuttgart (Germany).

Duan, H., Eugster, M., Hischier, R., Streicher-Porte, M., Li, J., 2009. Life Cycle As-sessment Study of a Chinese Desktop Personal Computer. Science of the Total Envi-ronment 407(5) 1755-1764.

ecoinvent Centre, 2003. ecoinvent data v1.01. Swiss Centre for Life Cycle Inventor-ies: Dübendorf, CH.

ecoinvent Centre, 2010. ecoinvent data v2.2. Swiss Centre for Life Cycle Inventor-ies: Dübendorf, CH.

Ekener-Petersen, E., Finnveden, G., 2013. Potential hotspots identified by social LCA - Part 1: A case study of a laptop computer. Int J LCA 18(1) 127-143.

Erdmann, L., Hilty, L.M., 2010. Scenario Analysis: Exploring the Macroeconomic Im-pacts of Information and Communication Technologies on Greenhouse Gas Emissions. Journal of Industrial Ecology 14(5) 824-841.

Faist Emmenegger, M., Frischknecht, R., Stutz, M., Guggisberg, M., Witschi, R., Otto, T., 2006. Life cycle assessment of the mobile communication system UMTS towards eco-efficient systems. Int J LCA 11(4) 265-276.


23

Fehske, A., Fettweis, G., Malmodin, J., Biczok, G., 2011. The global footprint of mo-bile communications: The ecological and economic perspective. IEEE Communications Magazine 49(8) 55-62.

Feng, C., Ma, X.Q., 2009. The energy consumption and environmental impacts of a color TV set in China. Journal of Cleaner Production 17(1) 13-25.

Finnveden, G., Hauschild, M.Z., Ekvall, T., Guinée, J., Heijungs, R., Hellweg, S., Koehler, A., Pennington, D., Suh, S., 2009. Recent Developments in Life Cycle Assessment. Journal of Environmental Management 91 1-21.

Frey, S.D., Harrison, D.J., Billett, E.H., 2006. Ecological Footprint Analysis Applied to Mobile Phones. Journal of Industrial Ecology 10(1-2) 199-216.

Frischknecht, R., Jungbluth, N., Althaus, H.-J., Doka, G., Dones, R., Hischier, R., Hell-weg, S., Nemecek, T., Rebitzer, G., Spielmann, M., 2007. Overview and Methodology. Swiss Centre for Life Cycle Inventories: Dübendorf, CH.

Goedkoop, M., Heijungs, R., Huijbregts, M.A.J., de Schreyver, A., Struijs, J., Van Zelm, R., 2012. ReCiPe 2008 - A life cycle impact assessment method which comprises harmonised category indicators at the midpoint and the endpoint level. First edition (revised) / Report I: Characterisation. VROM - Ministery of Housing Spatial Planning and Environment: Den Haag (the Netherlands).

Hikwama, 2005. Life Cycle Assessment of a Personal Computer, p. 148.

Hilty, L. M., Wäger, P., Lehmann, M., Hischier, R., Ruddy, T., Binswanger, M., 2004. The future impact of ICT on environmental sustainability. Fourth Interim Report – Re-finement and quantification. Institute for Prospective Technological Studies (IPTS), Sevilla. Hilty, L.M., Arnfalk, P., Erdmann, L., Goodman, J., Lehmann, M., Wäger, P., 2006. The Relevance of Information and Communication Technologies for Environmental Sus-tainability – A Prospective Simulation Study. Environmental Modelling & Software 11(21) 1618-1629.

Hischier, R., 2007. Life Cycle Inventories of Packaging and Graphical Paper. Swiss Centre for Life Cycle Inventories, Empa - TSL: Dübendorf, CH.

Hischier, R., Baudin, I., 2010. LCA study of a plasma television device. Int J LCA 15 428-438.


24

Hischier, R., Classen, M., Lehmann, M., Scharnhorst, W., 2007. Life Cycle Inventories of Electric and Electronic Equipment: Production, Use and Disposal. Swiss Centre for Life Cycle Inventories, Empa - TSL: Dübendorf, CH.

Hischier, R., Keller, M., Lisibach, R., Hilty, L.M., 2013. mat - an ICT application to sup-port a more sustainable use of print products and ICT devices, In: Hilty, L.M., Aebischer, B., Andersson, G., Lohmann, W. (Eds.), ICT4S 2013: Proceedings of the First International Conference on Information and Communication Technologies for Sustainability: ETH Zürich (Switzerland), pp. 223-230.

IPTS, 2003. Environmental, Technical and Market Analysis concerning the Eco-Design of Television Devices. European Commission, Joint Research Centre (JRC) - Institute for Prospective Technological Studies (IPTS): Sevilla (Spain).

ISO, 2006. Environmental Management - Life Cycle Assessment - Principles and Framework. International Standardization Organization (ISO), European Standard EN ISO 14'040: Geneva (Switzerland).

Kemna, R., van Elburg, M., Li, W., van Holsteijn, R., 2005. Methodology Study Eco-Design of Energy-using Products (MEEUP). Product Cases Report. VHK (Van Holsteijn en Kemna BV) on behalf of European Commission, DG Entreprise (Brussels): Delft (the Netherlands).

Lu, L.-T., Wernick, I.K., Hsiao, T.-Y., Yu, Y.-H., Yang, Y.-M., Ma, H.-W., 2005. Balancing the life cycle impacts of notebook computers: Taiwan's experience. Resources, Con-servation & Recycling 48(1) 13-25.

Maga, D., Hiebel, M., Knermann, C., 2013. Comparison of two ICT solutions: desktop PC versus thin client computing. Int J LCA 18(4) 861-871.

Malmodin, J., Bergmark, P., Lunden, D., 2013. The future carbon footprint of the ICT and E&M sectors, In: Hilty, L.M., Aebischer, B., Andersson, G., Lohmann, W. (Eds.), ICT4S 2013: Proceedings of the First International Conference on Information and Communication Technologies for Sustainability: ETH Zürich (Switzerland), pp. 12-20.

Manhart, A., 2011. International Cooperation for Metal Recycling From Waste Electri-cal and Electronic Equipment. Journal of Industrial Ecology 15(1) 13-20.

Moberg, A., Borggren, C., Finnveden, G., 2011. Books from an environmental per-spective – Part 2: e-books as an alternative to paper books. Int J LCA 16 238-246.

Moberg, A., Johansson, M., Finnveden, G., Jonsson, A., 2010. Printed and tablet e-paper newspaper from an environmental perspective - A screening life cycle assess-ment. Environmental Impact Assessment Review 30(3) 177-191.


25

Reichart, I., Hischier, R., 2001. Environmental impact of electronic and print media: Television, internet newspaper and printed daily newspaper, In: Hilty, L.M., Gilgen, P.W. (Eds.), Sustainability in the Information Society - 15th International Symposium Informatics for Environmental Protection. Metropolis Verlag: Zürich (Switzerland), pp. 91-98.

Reichart, I., Hischier, R., 2003. The Environmental Impact of Getting the News - A Comparison of On-Line, Television, and Newspaper Information Delivery. Journal of Industrial Ecology 6(3-4) 185-200.

Rhodes, S.P., 1993. Applications of Life Cycle Assessment in the Electronics Industry for Product Design and Marketing Claims, International Symposium on Electronics and the Environment. IEEE: Arlington (USA), pp. 101-105.

Scharnhorst, W., Hilty, L.M., Jolliet, O., 2006. Life cycle assessment of second gen-eration (2G) and third generation (3G) mobile phone networks. Environmental Inter-national 32(5) 656-676.

Song, Q., Wang, Z., Li, J., Yuan, W., 2013. Life cycle assessment of desktop PCs in Macau. Int J LCA 18(3) 553-566.

St-Laurent, J., Hedin, D., Honee, C., M., F., 2012. Green electronics? - An LCA based study of eco-labeling of laptop computers, Electronics Goes Green 2012+, ECG 2012 - Joint International Conference and Exhibition, Proceedings: Berlin (Germany).

Svensk Fjärrvärme, 2010. Statistics 2010. Download from http://www.svenskfjarrvarme.se/Global/Statistik/Excel-filer/Statistik%202010.xls. Svensk Fjärrvärme (Swedish District Heating Association).

Takahashi, K.I., Tsuda, M., Nakamura, J., Otabe, K., Tsuruoka, M., Matsuno, Y., Ada-chi, Y., 2009. Elementary analysis of mobile phones for optimizing end-of-life scen-arios, 2009 IEEE International Symposium on Sustainable Systems and Technology, ISSST '09 in Cooperation with 2009 IEEE International Symposium on Technology and Society, ISTAS. IEEE: Tempe, AZ (USA).

Tekawa, M., Miyamoto, S., Inaba, A., 1997. Life cycle assessment; an approach to environmentally friendly PCs, IEEE International Symposium on Electronics & the Envi-ronment. IEEE: San Francisco, CA (USA), pp. 125-130.

Thomas, V., Caudill, R., Badwe, D., 1998. Marginal emissions and variation across models: Life-cycle assessment of a television, IEEE International Symposium on Elec-tronics & the Environment: Oak Brook , IL (USA), pp. 48-53.


26

von Geibler, J., Ritthoff, M., Kuhndt, M., 2003. The environmental impacts of mobile computing. A case study with HP, Digital Europe: e-business and sustainable devel-opment. Wuppertal Institute: Wuppertal, p. 43.

Wäger, P., Lang, D., Bleischwitz, R., Hagelüken, C., Meissner, S., Reller, A., 2010. Sel-tene Metalle - Rohstoffe für Zukunftstechnologien, SATW Schrift Nr. 41, p. 32.

Weber, C.L., Koomey, J.G., Matthews, H.S., 2010. The Energy and Climate Change Implications of Different Music Delivery Methods. Journal of Industrial Ecology 14(5) 754-776.

Welz, T., Hischier, R., Hilty L. M., 2011. Environmental impacts of lighting technolo-gies — Life cycle assessment and sensitivity analysis. Environmental Impact Assess-ment Review 31 (3) 2011, 334-343 Williams, E., 2004. Energy Intensity of Computer Manufacturing: Hybrid Assessment Combining Process and Economic Input-Output Methods. Environmental Science & Technology 38(22) 6166-6174.