Consumption-Weighted Life Cycle Assessment of a Consumer

Article pubs.acs.org/est

Consumption-Weighted Life Cycle Assessment of a Consumer Electronic Product Community Erinn G. Ryen,† Callie W. Babbitt,*,† and Eric Williams† †

Golisano Institute for Sustainability, Rochester Institute of Technology, Rochester, New York 14623, United States S Supporting Information *

ABSTRACT: A new approach for quantifying the net environmental impact of a “community” of interrelated products is demonstrated for consumer electronics owned by an average U.S. household over a 15-year period (1992−2007). This consumption-weighted life cycle assessment (LCA) methodology accounts for both product consumption (number of products per household) and impact (cumulative energy demand (MJ) and greenhouse gas emissions (MT CO2 eq) per product), analyzed using a hybrid LCA framework. Despite efficiency improvements in individual devices from 1992 to 2007, the net impact of the entire product community increased, due primarily to increasing ownership and usage. The net energy impact for the product community is significant, nearly 30% of the average gasoline use in a U.S. passenger vehicle in 2007. The analysis points to a large contribution by legacy products (cathode ray tube televisions and desktop computers), due to historically high consumption rates, although impacts are beginning to shift to smaller mobile devices. This method is also applied to evaluate prospective intervention strategies, indicating that environmental impact can be reduced by strategies such as lifespan extension or energy efficiency, but only when applied to all products owned, or by transforming consumption trends toward fewer, highly multifunctional products.

■

INTRODUCTION Greening the environmental performance of consumer electronics has been a major initiative for researchers and decision makers. Manufacturing innovations and voluntary product labeling have helped reduce energy impacts for individual products.1,2 However, U.S. households have been amassing a large and increasingly complex bundle of devices to fulfill information, communication, and entertainment functions,3 potentially offsetting environmental savings from efficiency gains. For example, while average standby power for televisions (TVs) and computers has declined since the 1990s with the introduction of Energy Star standards,4 the overall volume of new products with standby modes has increased.5 The rebound effect has also been noted at the electronic component level, for computer microprocessors.6 For efficiency improvements to result in reduced environmental impacts, technological innovations must outweigh overall consumption of the goods.7 Because of the complex relationship between consumption and technological progress, sustainability methods such as LCA struggle with characterizing dynamic changes in environmental impacts. Of the wide body of literature quantifying energy impacts of consumer electronics, all but a few8,9 compute life cycle impacts without considering consumption behavior and ownership patterns. For example, many LCAs focus on use phase at the household,10−13 state,11,14 or national scale,4,15−18 or for a single product19−22 (see Supporting Information (SI) Table SI-1). Thus, a need remains to link environmental © 2015 American Chemical Society

analyses of manufacturing and use (impact per product) with evolving trends in consumption (products owned at a given time). Because electronics are usually purchased in groups to fulfill information, communication, and entertainment needs, LCA methods must consider the number and type of devices owned within this group, or “product community”. To this end, inspiration is drawn from the field of biological community ecology, which studies groups of living organisms that persist and interact in a defined space and time.23 The organisms provide services or functions such as nutrient cycling24 and facilitating response to external stressors (e.g., changes in resources, precipitation, or temperature).24,25 Fluctuations in the structure (number and distribution of organisms) and functions in the community dictate resultant flows of inputs (e.g., energy from the sun or nutrients) and outputs (e.g., biomass) from the ecosystem.25 Similarly, household purchase and use of different numbers and types of electronic products also drive attendant inputs, such as energy (e.g., electricity and fuel) and materials (e.g., plastics, glass, and metals) and resultant outputs (e.g., used components and electronic waste). Consequently, community ecology offers a promising systemReceived: Revised: Accepted: Published: 2549

October 20, 2014 January 14, 2015 January 15, 2015 January 15, 2015 DOI: 10.1021/es505121p Environ. Sci. Technol. 2015, 49, 2549−2559

Article

Environmental Science & Technology

Figure 1. Consumption-weighted LCA methodology and scope: (a) inputs used and outputs determined in the “per product” and “per product community” analysis, and (b) devices comprising each cohort and product community per modeled year. The product community in each modeled year was divided into groups or cohorts (see color coding) based on the period during which devices were introduced.

■

MATERIALS AND METHODS Objective and Scope. To quantify the product community’s net environmental impact, the consumption-weighted LCA approach was developed for and applied to a group of interrelated electronics that provide information, communication, and entertainment services. The functional unit was an average U.S. household for one year. The metrics used to quantify environmental impact were annualized cumulative energy demand and greenhouse gas (GHG) emissions per household. Although both impact categories were calculated based on the same methods, equations, and most sources, the detailed description of methods, below, is specific to the energy analysis, with subsequent details provided for calculation of GHG emissions (see also the SI). Whereas many environmental impacts result from production and consumption of consumer electronics, cumulative energy demand, which includes both direct (electricity consumed while using the device) and indirect (i.e., upstream fossil fuel) inputs, is a well established predictor of environmental impacts including, but not limited to, the depletion of resources, acid rain, and climate change.38,39 The consumption-weighted impact (Figure 1) was calculated as the product of community structure (number (n) of products (i) owned per average U.S. household) and annualized energy demand (Ei,t) (in MJ) per product (i) per household for the modeled years (t) (eq 1).

atic approach to assessing a product community’s net environmental impact.3 Addressing household electronics as a product community3 or “portfolio”,26 builds on Levine’s product-centered approach27,28 and a small set of studies that focus on an “ensemble” of energy generating systems29,30 or a “fleet” of transportation systems.31−35 These studies show that considering products as an interconnected group has led to more comprehensive pollution reduction strategies and policies (e.g., vehicle mileage standards31). Although no single electronic device has an impact close to an automobile, the cumulative impact may be significant. For example, in many countries, mobility services are supplied by a community dominated by a few similar “species” (cars, buses, trains, etc.), in contrast to the complex, interacting, and rapidly evolving community of consumer electronics. Applying LCA at the household scale has been noted as particularly appropriate for products undergoing technology transitions34 since impacts for emerging technologies are closely linked to consumption behavior.36 Additionally, as found in environmental behavior research, residents generally have more control over the household’s purchasing decisions, as opposed to a larger scale (firm or nation), where only a few individuals have decision responsibility.37 Systematic understanding of impacts due to product interactions within households can broaden the application and scope of LCA methodology36 and lead to policies that encourage behavioral changes and reduce environmental impacts.37 Therefore, our goal is to develop and apply a new assessment approach that systematically characterizes dynamic changes in net environmental impacts for an evolving electronic product community.

E household, t =

∑ (Ei ,t × ni ,t ) i

(1)

Annualized environmental impacts were determined by a hybrid LCA approach following Hertwich and Roux9 and included production (material extraction, manufacturing, and 2550

DOI: 10.1021/es505121p Environ. Sci. Technol. 2015, 49, 2549−2559

Article


for each device (i) as an input to the EIO-LCA model. The net annualized production energy (Ep) (in MJ) for each device (i) was a product of the IO sector energy (e) ($/MJ) from the 1992, 1997, and 2002 producer price EIO models (and extrapolated for 2007) and average producer price (pp) ($) for year (t), divided by the average service life of the product (l) (eq 3).

transportation) and product use. The scope excluded end of life (EOL), because many studies have found that EOL energy impacts only contributed a small fraction to net life cycle results.9,20,21,40 Production energy and GHG emissions were estimated via the online Economic Input−Output Life Cycle Assessment (EIO-LCA) tool by Carnegie Mellon University (CMU)’s Green Design Institute.41 As a result, the dynamic analysis focused on years for which EIO-LCA data were available (1992, 1997, 2002) and reasonably extrapolated (2007, see next section). Analyzing the 2007 model year was essential for capturing the effect of newer devices (plasma TVs, tablets, and e-readers) on the overall impact. The scope included products owned per household (ni,t) as determined in a previous study,3 which first categorized consumer electronic products based on industry classifications and then estimated the number of each product per household between 1990 and 2010 using a material flow approach (see Figure 1 and SI Table SI-2 for the 19 devices included in the scope). Although products were introduced continuously to U.S. households between 1990 and 2010, this study grouped them into cohorts corresponding to the years for which EIOLCA data were available. For example, the “1992 Cohort” consisted only of devices introduced by and before 1992 (e.g., CRT TV and desktop computer), while the “1997 Cohort” comprised devices introduced between 1992 and 1997 (digital camera and camcorder). Hybrid LCA Methodology. As discussed above, the hybrid LCA methodology calculated the annualized energy demand per device (Ei,t) as the sum of production energy (Ep,i,t) (including all upstream supply chain processes, estimated via EIO-LCA) and use phase energy (Eu,i,t) (estimated via productlevel process data) for each device (i) and modeled year (t) (eq 2). Ei , t =

∑ (Ep,i ,t + Eu ,i ,t ) i

Ep , i , t =

pp , i , t × ei , t li

(3)

Impact per IO sector for 2007 was projected using linear extrapolation of existing IO sector level impacts per nominal input dollar from the 1992, 1997, and 2002 IO sector data points, an approach enabled by the relatively small year-to-year variability in the stable U.S. manufacturing sector energy and decreasing GHG emissions (SI Figure SI-1 and Tables SI-4 to SI-7). Product price was a key input to the EIO-LCA model, here determined in the following two steps: (1) collecting average consumer prices for each product in every modeled year, and (2) converting consumer prices to producer prices, the required input to the producer price model. Average consumer prices were collected from a consistent set of publicly available trade publications and commercial sources, such as the Consumer Electronics Manufacturing Association,49 review articles (e.g., Ballou50 and Cheng51), or Consumer Reports publications52−74 (see SI Table SI-8 for a complete list). Although electronic devices are available with variable customizations and sizes (e.g., screen sizes for televisions and monitors), a single model size was generally used for all years analyzed, and average prices reflected typical product models and configurations. In a few cases where consumer prices were not available for the modeled years, average consumer prices adjacent to the modeled year were adjusted using the U.S. Bureau of Labor Statistics (BLS) producer price index (PPI)75 for the specified IO sectors, or for a few cases (gaming consoles and printers), the consumer price index.76 Producer prices were converted from consumer prices using the U.S. BEA IO Bridge Tables to Personal Consumption Expenditures77−80 for each modeled year (see SO Tables SI-9 to SI-12 for details related to production input values and Table SI-13 for a summary of average producer prices). Product lifespan was also a required input, whereby total production energy could be equally divided by the average service life to determine annualized impact for each modeled year. The issues surrounding product lifespan definition and resultant contribution to uncertainty and variability to life cycle energy impacts have been widely discussed.81−83 Here, lifespan was defined as the time in use during the device’s average first life (li). In some cases where data had no delineation between use, storage, and reuse lifespans (printer, TVs, camera, camcorder, and VCR, DVD, blu-ray, and MP3 players), the total available lifespan was applied. The selection of each product’s lifespan from available sources was first based on primary data on consumer behavior, such as from consumer surveys (e.g., Williams84 and NIES85). In cases where this information was not available, lifespans were based on product studies, technical reports, or assumptions in peer-reviewed publications (SI Figure SI-2). In general, the baseline LCA analysis was based on the median of all lifespan values compiled (see SI Table SI-14). In a few cases where lifespan data were limited, but products had closely related functions or forms (e.g., basic and smart mobile phones, or tablets and e-readers),

(2)

Potential errors associated with the EIO-LCA methodology include aggregation of data to the sector level and the assumption that products were produced in the U.S.42 However, the benefits of reduced cutoff error and its quick and inexpensive nature promote EIO-LCA as an environmental policy tool.42,43 Although conducting individual process-based LCA on all 19 devices in the product community would be ideal, such an effort would have enormous financial and time constraints and data limitations. Thus, the approach used here was to demonstrate the benefit of the consumption-weighted LCA approach using a hybrid method, which could be extended in the future as product-specific data become available. In terms of geographic scope, U.S. IO sector data were initially closely aligned with the production of consumer electronics.44−46 According to the Consumer Electronics Industry,47 many consumer electronics were still produced in the U.S. as late as 1994, including half of all TVs sold domestically. However, the transition to overseas production necessitated consideration of global supply chains, modeled here with China-based IO energy data from Chang et al.48 as described in SI Table SI-3. An uncertainty analysis comparing U.S.- and China-based production impacts was based on available years in the Chinese data set (2002 and 2007).48 Calculation of Production Impacts. Production phase energy was estimated by first classifying each electronic product into appropriate U.S. Bureau of Economic Administration (U.S. BEA) IO sectors and then determining average producer prices 2551


Article


Figure 2. Net annualized energy and GHG emissions of electronics owned by U.S. households in 1992, 1997, 2002, and 2007. Data are aggregated by cohorts corresponding to the period in which groups of devices were introduced into the product community (indicated by color coding), and compared on a “per product” (ownership of one device per year) (a) and “per product community” basis (weighted by average household consumption trends) (b).

household), they would automatically be considered “primary” televisions, likely purchased and used for the main household TV viewing. The remaining households (1.0 − 0.11 − 0.30) would have used 0.59 CRTs per household as primary viewing devices and any remaining CRT TVs (2.94 − 0.59 = 2.35 per household) would be considered as secondary devices. A sample calculation of the use phase energy division for primary and secondary TVs and desktop computers is provided in SI Section 2.6.3, with resulting model inputs in Table SI-37. Future Production and Consumption Scenarios. To further demonstrate the utility of the consumption-weighted LCA approach, the method was used to analyze the extent to which common intervention strategies (e.g., green production and use behaviors) and/or shifts in the community structure could reduce the net impact. Two common production-oriented strategies were considered: increase energy efficiency during use by 10% and/or extend product lifespan by 10%. A 10% energy efficiency improvement per device is consistent with conservative estimates by U.S. EPA Energy Star program, which suggested that building occupants could achieve at least 10% in energy savings through education and behavior changes,92,93 such as unplugging devices not in use, using smart power strips to reduce standby energy,92,94,95 or implementing common denominator strategies (efficiency standards for chargers).11 Extending lifespan has been recommended as another key strategy to manage life cycle impacts of electronics.4,21,96 and is included in the rating system developed by the Electronic Product Environmental Assessment Tool (EPEAT).97 Product lifetime extension could be achieved with more durable materials, enhanced maintenance services, or product labeling.96,98 Additional descriptions of scenarios and devices included in each are noted in SI Table SI-39. To assess potential changes due to shifts in consumption, multiple scenarios were developed to reflect evolving

the same lifespan was assumed for both. To capture the range of uncertainty associated with varying lifespans, uncertainty analysis using low, median, and high data was conducted. Calculation of Use Phase Impacts. The use phase energy was derived from each product’s average power consumption per mode and time spent in each mode for typical models as reported in trade industry reports4,18 and governmental reports.12,15 For modeled years where these data were not available, energy consumption and usage per mode or unit energy consumption (UEC) data were extrapolated from adjacent years. A full description of data, extrapolation, and sources is available in SI Tables SI-15 to SI-35. The average UEC (kWh per year) per product was converted to cumulative energy demand using variable conversion factors of 10.5 to 10.8 MJ cumulative energy demand per kWh electricity. These factors, which were calculated specific to each modeled year, accounted for upstream energy inputs, changes in the U.S. grid fuel mix,86 inefficiencies,87,88 and transmission losses89 (see SI Section 2.6.3 and Table SI-36). A summary of model inputs is in SI Table SI-37. Following the same approach, conversion factors ranging from 5.9 × 10−4 to 7.0 × 10−4 MT CO2 eq/ kWh, which were obtained or estimated from the U.S. EPA Emissions & Generation Resource Integrated Database (eGRID),90 were used to calculate use phase GHG emissions (SI Section 2.6.5 and Table SI-38). When a household owned multiple devices of the same type (observed for TVs and desktop computers), the products’ usage was assumed to vary depending on whether they were the primary device in use or secondary devices used less frequently. Distinctions in use phase energy for primary and secondary products followed reported usage patterns in technical and trade publications.4,16,18,91 For example, in 2007, an average U.S. household owned 3.35 TVs, which included plasma, LCD, and CRT models. It was assumed that if a household had a plasma TV (0.11 per household) or LCD TV (0.30 per 2552


Article


Figure 3. Analysis of the 1992 cohort. Increasing average consumption of devices in the 1992 cohort (secondary y axis) offsets a decreasing trend in weighted average prices (y axis), as shown by the increasing and then decreasing trend in consumption-weighted household spending (price per product multiplied by number owned in a given year) (y axis) (a). On a “per product” basis, average standby power mode for the 1992 cohort decreases, but active power mode and viewing time increases over time (b). Increasing total number of devices in the 1992 cohort (secondary x axis) correlates with an increasing contribution of “per product community” use phase energy (c).

■

consumption trends, ongoing emergence of small, mobile devices, and the potential for design and purchase of fewer, functionally convergent devices.3 Functionally convergent or hybrid devices that provide multiple functions have been gaining momentum in the market, as seen by blurring lines between phone and tablet,95,99,100 high resolution camera and smartphone,101 and TV, gaming console, and computer.102 Future consumption scenarios consisted of three hypothetical cases that represented a potential shift in consumption away from many single- or few-function products toward a few highly multifunctional products, specifically for devices used in (1) voice communication (e.g., phone calls), (2) data manipulation (e.g., word processing, surfing the Internet), and (3) audio visual playback or recording (e.g., recording or watching movies or music). An extreme case, the “digital streamlined” scenario, was based on maximum deployment of six functionally convergent devices (see SI Table SI-39 for additional information).

RESULTS AND DISCUSSION

Product and Household Level Results. The net impact for electronics purchased and used by an average U.S. household is presented for all products, assessed independently, or “per product” (Figure 2a) and on a consumption-weighted basis for the entire household, or “per product community” (Figure 2b) (see also SI Tables SI-40 to SI-42). If products are accounted for independently (i.e., the impact of producing one of each product is summed for all products in the household for a modeled year), the net impact appears to increase over time (Figure 2a), corresponding to the introduction of new products into the household. When net impact is disaggregated into cohorts, flat (1992 cohort) or declining (1997 and 2002 groupings) trends are generally observed, corresponding to stabilizing or decreasing trends in production energy or GHG intensity (MJ or MT CO2 eq/$ in the EIO model), respectively, as well as reduced product prices. Net annualized GHG impacts reflect trends almost identical to energy when 2553


Article


Figure 4. Relative contribution to the net annualized energy impact of electronics devices owned by U.S. households in 1992, 1997, 2002, and 2007, on a “per product” (ownership of one device per year) (a) and “per product community” basis (weighted by household consumption trends) (b). Shadings indicate cohorts or groups of devices based on the first year introduced into the product community. Numerical results at right of each figure are for 2007. Each product’s net impact (in MJ) is represented by line thickness in pixels (PX).

disaggregated according to contribution by each cohort over time (SI Figure SI-3). When accounting for actual consumption of each product, the community’s net annualized energy and GHG impacts also increase over time (Figure 2b), but not due to the purchase of newly introduced products (e.g., plasma and LCD TVs), as these devices have very low ownership rates in the time period analyzed. Instead, the increase is attributed to increasing accumulation of earlier products that have become essential components of a household’s social, communication, and entertainment activities (Figure 2b). For example, households in 2007 owned an average of 3 CRT TVs. The resulting net energy impact of the product community is significant; equivalent to nearly 30% of the average annual fuel consumed by an average passenger vehicle in 2007103 (see SI Table SI-43). Energy services such as transportation and climate-control garner far more policy attention than consumer electronics, perhaps because they are supplied by a small number of high impact products (e.g., automobile and gas furnace). Viewed through the community lens, however, products associated with information, communication, and entertainment services have

comparable macro-level impacts and warrant greater policy consideration. Close examination of the 1992 cohort (Figure 3) illustrates how changes in prices, consumption, and innovation influence energy impacts. Increasing consumption of devices offsets a decreasing trend in average product prices, the main driver of production impacts (Figure 3a and SI Table SI-44). Increased viewing time and energy use during the active power mode further compounds the increasing consumption of these products (Figure 3b). For example, CRT TV and desktop computer ownership expand in this time period (nearly 40% and 300%, respectively), active usage (hours per year) increases (20% for the CRT TV and over 100% for the desktop computer), overshadowing any power mode energy efficiencies occurring at the same time (SI Tables SI-15 to SI-16 and SI-23 to SI-24). The decreasing trend in average standby power mode for the 1992 cohort (Figure 3b and SI Table SI-45) reflects the growth of environmental initiatives such as the U.S Energy Star program. Clearly, energy conservation policies focusing solely on standby power are not sufficient to handle the “per product community” impacts. Moreover, increasing number of devices in the 1992 cohort and power mode/time trends aligns with 2554


Article


Figure 5. Changes in net annualized energy for potential environmental intervention strategies relative to 2007 baseline community of U.S. household electronics. Strategies 1 and 2 represent conventional environmental improvement approaches; strategies 3−6 represent future consumption scenarios. For each strategy (a), shadings correspond to product cohort. Decreases (negative percentages) and increases (positive percentages) in the impact as a result of each strategy are denoted on a “per product community” basis (b).

increasing the contribution of use phase contribution of “per product community” energy (Figure 3c and SI Table SI-46). The upsurge in average active usage impact is not surprising, as product functionality and integration to daily life has grown significantly during the same time period.3 When the net energy impact is partitioned, the products responsible for the greatest impact vary significantly depending on whether a “per product” (Figure 4a) or “per product community” (Figure 4b) approach is used. For example, the plasma TV, as one of the highest contributors to the aggregate impact (Figure 4a), appears to be a prime candidate for environmental improvement. Whereas such energy gains would certainly not be a detriment, they may make little to no difference for the community as a whole, because ownership of plasma TV devices is low during this time period. Instead, the CRT TV and desktop computer are the main contributors to the entire household impact (Figure 4b). Going beyond 2007, these products are certainly being replaced with newer technology (e.g., LCD TVs, laptops, or tablets), but the analysis demonstrates that prioritization for environmental improvement must account for actual consumption, which may lag the introduction of emerging products. Uncertainty Analysis. We recognize that there are many sources of uncertainty associated with the model and its inputs (i.e., product ownership and impact per product). Uncertainties associated with model inputs include ownership, prices, and lifespan. For example, this approach provides a snapshot of four years of product ownership per average U.S. household, which

has changed since 2007 to include the consumption of smaller mobile devices.3 Because prices are a critical input, uncertainties may arise if consumers select products with prices significantly above or below average annual U.S. or global values. Although lifespan is expected to be a key variable in electronic product assessment, uncertainty analysis showed relatively low sensitivity to varying lifespan parameters. In 2007, high lifespans result in 3% lower impacts compared to baseline results, while low lifespans increase net product community impacts by 6% (SI Table SI-47 and Figure SI-4). The model itself introduces uncertainty related to aggregation within IO sectors, linearity of impact (MJ/$),42 and geographical coverage. A major source of uncertainty for the impact per product is attributed to regional differences in manufacturing, given the shifts from domestic to overseas production. The “per product community” impact calculated using China-based IO data remains relatively constant from 2002 to 2007, but is nearly double that of the net impact assuming U.S.-based IO data in 2007 (see SI Tables SI-48 to SI50 and Figure SI-5). There are several region-specific differences that might explain this finding, including the highly aggregated China EIO model and fuel consumption mix per country. Contributing U.S. sectors have already gone through periods of growth, innovation, and now, stability, so relatively little further improvements are observed in the sector-specific energy intensity (MJ/$) over the time period in study. On the other hand, China’s electronic device manufacturing sectors may still be experiencing production efficiency gains that 2555


Article


few as six types of products, each albeit owned at higher concentrations, could theoretically provide all required information, communication, and entertainment services used in a household,3 which would result in a significant reduction in the net annualized energy impact for the entire community (Figure 5b, strategy 6). A majority of these savings is due to eliminating highly concentrated legacy products such as the CRT TV, desktop computer, CRT monitor, and VCR, which contribute over 70% of the 2007 baseline community-level impact. In the new consumption scenario, the highest impact products would then be the tablet and LCD TV. Then, appropriate product-level intervention strategy can be applied to maximize further improvements. For example, mobile devices with short lifespans (tablet) would likely benefit from a lifespan extension strategy and high-energy use devices (LCD TV) would benefit with an operational efficiency strategy (see SI Tables SI-53 to SI-56). Encouraging the design and ownership of functionally convergent devices as an energy reduction strategy for the product community is consistent with current industry trends.95 For example, as digital content on the cloud increases, consumers are expected to “favor lighter, faster, and fewer devices” (p 6),104 resulting in multifunctional devices prevailing over single function products (e.g., e-readers) as a content delivery system. Because consumers identify price and feature variety as purchasing decisions over energy efficiency,95 designing a fewer number of functionally convergent devices may be a “big pivot” (p 60)106 strategy that disrupts the “unsustainable” product community, significantly reduces overall impact, and moves households onto a sustainable path that integrates both consumption and production improvement strategies. A trade-off would occur since many state electronics recycling policies are designed around mass-based targets, and a shift in the product community structure would have significant repercussions on recycling infrastructures and business models implemented to meet those targets. The consumption-weighted LCA methodology can therefore assist governmental and industry decision and makers as they propose, evaluate, and implement future policies, standards, and legislation to manage life cycle impacts for groups of emerging computing technologies. As these results suggest, we can no longer ignore product communities when designing, producing, and consuming green devices. The consumption-weighted LCA methodology presented here is able to capture dynamic changes in the net environmental impact (annualized energy demand) for both production and consumption of an interrelated group or “community” of consumer electronics in an average U.S. household. This approach is important since consumer electronics are experiencing rapid changes in consumption patterns and functional preferences.107 Considering products as a community answers a call for LCA to broaden its scale, address rebound, behavior, and price effects, while balancing the need for a simplified assessment tool. 36,105 The consumption-weighted results are also more relevant for design, production, and policy changes, which must target products that are high impact in their own right (“per product”) as well as those whose net contribution becomes significant due to high ownership rate (“per community”). Can households reduce their environmental footprint while preserving the features consumers demand from beloved electronic devices? Designing and encouraging the ownership of fewer multifunctional devices may simultaneously enable

outpace increasing consumption trends as shown by the decreasing contributions of the production phase energy from the 1992 and 1997 cohorts (SI Table SI-50). In the 2002 cohort, consumption changes still dominate for some products, particularly the LCD TVs and smartphones introduced in 2002, which increase by 37- and 12-fold in net energy impact for the household from 2002 to 2007 (SI Table SI-50). In general though, if consumption trends continue as is, the future net energy impact for an average U.S. household consuming products produced in Asia will likely show similar trends once these sectors stabilize production energy intensity. Evaluation of Intervention Strategies. This research thus far has demonstrated the utility of the consumptionweighted LCA methodology to illustrate the changing energy impact for a community of consumer electronics owned by an average U.S. household. It stands to reason that this methodology can also be applied to determine the effectiveness of common intervention strategies, as well as more radical changes to the community’s overall structure. Policies aimed at reducing environmental impacts often focus on regulating the purchase of products with “green” attributes, such as energy conservation (e.g., Energy Star), or designing products for longevity and for enhanced recycling and end of life management (e.g., EPEAT). Although common strategies such as energy efficiency and lifespan extension do show promise on a “per product” basis, they actually yield incremental energy reductions for the product community after accounting for consumption. For example, a 10% reduction in use phase energy can lead to as much as a 9% decrease in energy impact per product for the CRT TVs, VCR, desktop computer, CRT monitor, and plasma TV, all of which have high use phase contributions to their total life cycle impact. Similarly, a 10% increase in lifespan creates 6−8% decrease in energy impact per product for devices (camcorder, camera, smart phone, MP3 player, e-reader, and tablet) which have high production phase impacts (see SI Tables SI-51 and SI-52). However, these benefits are diminished when consumption is taken into consideration, as many of the products with high individual improvements are actually owned at low rates within the product community. When considering the community as a whole, improving operational efficiency as a strategy (Figure 5a, strategy 1) results in community-level savings of 8.4% compared to the 2007 baseline (Figure 5a and b and SI Table SI-52). However, to achieve this savings would require every product in the community to reach efficiency improvements of at least 10%, which could be difficult to achieve due to rapid changes in consumer preferences and shortened innovation cycles. In contrast, the conventional strategy of extending product lifespan (Figure 5a, strategy 2) yields incremental improvements (1.5%) for the entire product community (Figure 5b). In addition to conventional strategies, the consumptionweighted approach can quantify how potential future changes in product ownership associated with device convergence may ultimately influence overall net energy impact. In most cases, the model is very sensitive to fundamental changes in the product community structure, such as if tablets largely replace desktop computers, monitors, e-readers, and MP3 players for providing mobile data processing and browsing functionality to consumers (Figure 5, strategy 4). Certain multifunctional products (e.g., tablet), like a natural invasive species, could hypothetically disrupt the product community by changing consumption patterns. In the “digital streamlined” scenario, as 2556


Article


ACEEE Summer Study on Energy Efficiency in Buildings: Pacific Grove, CA, August 13−18, 2006. (11) Porter, S.; Moorefield, L.; May-Ostendorp, P. Final Field Research Report; Ecos Consulting: Durango, CO, October 31, 2006. (12) Bensch, I., Pigg, S.; Koski, K.; Belshe, R. Electricity Savings Opportunities for Home Electronics and Other Plug-in Devices in Minnesota Homes: A Technical and Behavioral Field Assessment; ECW Report257-1; Energy Center of Wisconsin: Madison, WI, May 2010. (13) Peters, J.; Frank, M. Van Clock, D. J.; Armstrong, A. Electronics and Energy Efficiency: A Plug Load Characterization Study; SCE0284; report prepared for Southern Californian Edison; Research Into Action, Inc.: Portland, OR, January 29, 2010. (14) McAllister, J.; Farrell, A. Electricity consumption by batterypowered consumer electronics: a household-level survey. Energy 2007, 32, 1177−1184. (15) Zogg, R.; Alberino, D. Electricity Consumption by Small End Uses in Residential Buildings Final Report; Contract DE-AC01-96CE23798; report prepared for the U.S. Department of Energy Office of Building Equipment; Arthur D. Little: Cambridge, MA, August 20, 1998. (16) Rosen, K.; Meier, A.; Zandelin, S. National energy use of consumer electronics in 1999. Presented at 2000 ACEEE Summer Study on Energy Efficiency in Buildings; American Council for an Energy-Efficient Economy: Washington, DC, 2000; http://www. aceee.org/files/proceedings/2000/data/index.htm. (17) Kawamoto, K.; Koomey, J.; Nordman, B.; Brown, R.; Piette, M.; Ting, M.; Meier, A. Electricity used by office equipment and network equipment in the U.S. Energy 2002, 27, 255−269. (18) Urban, B.; Tiefenbeck, T.; Roth, K. Energy Consumption of Consumer Electronics in U.S. Homes in 2010; final report to the Consumer Electronics Association; Fraunhofer Center for Sustainable Energy Systems: Cambridge, MA, 2011. (19) Socolof, M.; Overly, J.; Kincaid, L.; Geibig, J. Desktop Computer Displays: A Lifecycle Assessment, vol. 1; EPA-744-R-01-004; U.S. EPA: Washington, DC, 2001. (20) Williams, E. Energy intensity of computer manufacturing: Hybrid assessment combing process and economic input-output methods. Environ. Sci. Technol. 2004, 38 (22), 6166−6174. (21) Deng, L.; Babbitt, C. W.; Williams, E. Economic-balance hybrid LCA extended with uncertainty analysis: Case study of laptop computer. J. Cleaner Prod. 2011, 19, 1198−1206. (22) Teehan, P.; Kandlikar, M. Comparing embodied greenhouse gas emissions of modern computing and electronics products. Environ. Sci. Technol. 2013, 47 (9), 3997−4003. (23) Whittaker, R. H. Communities and Ecosystems, 1st ed.; MacMillan: New York, 1970. (24) Diaz, S.; Cabido, M. Vive la difference: Plant functional diversity matters to ecosystem processes. Trends Ecol. Evol. 2001, 16, 464−655. (25) Ricklefs, R.; Miller, G. L. Ecology, 4th ed.; W.H. Freeman and Company: New York, 2000. (26) Williams, E. Portfolio LCA. Presented at the 2010 American Center for Life Cycle Assessment LCA X conference; Portland, OR, November 2−4, 2010. (27) Levine, S. H. Products and ecological models: A population ecology perspective. J. Ind. Ecol. 1999, 3, 47−62. (28) Levine, S. H. Comparing products and production in ecological and industrial systems. J. Ind. Ecol. 2003, 7 (2), 33−42. (29) Gutowski, T. Energy payback for energy systems ensembles during growth. Presented at the 2010 Institute of Electrical and Electronics Engineers International Symposium on Sustainable Systems and Technology Conference; Washington, DC, May 17−19, 2010. (30) Kotaro, K.; Gershwin, S.; Buonassisi, T.; Gutowski, T. Impact of PV growth on CO2 emission in the world. Presented at the 2012 Institute of Electrical and Electronics Engineers International Symposium on Sustainable Systems and Technology; Boston, MA, May 16−18, 2012. (31) Winebrake, J.; Corbett, J.; Wang, C.; Alexander, E.; Farrell, A. E.; Woods, P. Optimal fleetwide emissions reductions for passenger ferries: An application of a mixed-integer nonlinear programming model for the New York−New Jersey harbor. J. Air Waste Manage. 2005, 55 (4), 458−466.

increased household functionality and environmental improvements. Then, applying the most suitable conventional strategy for each product can further maximize additional improvements.

■

ASSOCIATED CONTENT

* Supporting Information S

Tables SI-1 to SI-56 and Figures SI-1 to SI-5. This material is available free of charge via the Internet at http://pubs.acs.org/.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: 585-475-6277; fax: 585-475-5455 e-mail: cwbgis@rit. edu. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We gratefully acknowledge and thank Mona Komeijani, Barbara Kasulaitis, and Matthew Koskinen for assistance in preparing graphics and data collection, and Christy Tyler, Greg Babbitt, and Gabrielle Gaustad for their valuable comments. This research was supported by the STAR Fellowship Assistance Agreement FP-91736401-1 awarded by the U.S. Environmental Protection Agency (EPA). The U.S. EPA has not formally reviewed the research. This research was also supported by the Golisano Institute for Sustainability at Rochester Institute of Technology (RIT) and by the National Science Foundation (CBET Grant 1236447).

■

REFERENCES

(1) Brown, R.; Webber, C.; Koomey, J. Status and future directions of the energy star program. Energy 2002, 27, 505−520. (2) Sanchez, M.; Brown, R.; C. Webber, C.; Homan, G. K. Savings estimates for the United States Environmental Protection Agency’s Energy Star voluntary product labeling program. Energy Policy 2008, 36, 2098−2108. (3) Ryen, E.; Babbitt, C.; Tyler, A.; Babbitt, G. Community ecology perspectives on the structural and functional evolution of consumer electronics. J. Ind. Ecol. 2014, 18, 708−721. (4) Roth, K.; McKenney, K. Energy Consumption by Consumer Electronics in U.S. Residences; final report prepared for the Consumer Electronics Association; TIAX LLC: Cambridge, MA, 2007. (5) Meier, A.; Nordman, B.; Busch, J.; Payne, C.; Brown, R.; Homan, G.; Sanchez, M.; Webber, C. Low-Power Mode Energy Consumption in California Homes: Final Pier Project; prepared for Public Interest Energy Research (PIER) California Energy Commission; report CEC500-2008-035; Lawrence Berkeley National Laboratory: Berkeley, CA, September 2008. (6) Deng, L.; Williams, E. Functionality versus ‘typical product’ measures of technological progress: a case study of semiconductor manufacturing. J. Ind. Ecol. 2011, 15 (10), 108−120. (7) Dahmus, J. Can efficiency improvements reduce resource consumption? A Historical Analysis of Ten Activities. J. Ind. Ecol. 2014, 18, 883−897. (8) Malmodin, J.; Moberg, Å.; Lundén, D.; Finnveden, G.; Lövehagen, N. Greenhouse gas emissions and operational electricity use in the ICT and entertainment media sectors. J. Ind. Ecol. 2010, 14 (5), 770−790. (9) Hertwich, E.; Roux, C. Greenhouse gas emissions from the consumer electric and electronic equipment by Norwegian households. Environ. Sci. Technol. 2011, 45, 8190−8196. (10) Hendron, R.; Eastment, M. Development of an energy-savings calculation methodology for residential miscellaneous electric loads; preprint of conference paper; NREL/CP-550-39551; presented at 2557


Article


(55) Consumer Reports. Consumer Reports Buying Guide 2002; Consumers Union of United States: Yonkers, NY, 2001. (56) Consumer Reports. Consumer Reports Digital Buying Guide 2004; Consumers Union of United States: Yonkers, NY, 2003. (57) Consumer Reports. 1993 Buying Guide: Special Year-End Issue. Consumer Reports 1992, 57 (13). (58) Consumer Reports. Big computers come in small packages. Consumer Reports 1993, 58 (11), 716−721. (59) Consumer Reports. Consumer Reports 1997, 62 (5). (60) Consumer Reports. Consumer Reports 2002, 67 (3). (61) Consumer Reports. Consumer Reports 2002, 67 (9). (62) Consumer Reports. Consumer Reports 2002, 67 (12). (63) Consumer Reports. Consumer Reports 2005, 70 (3). (64) Consumer Reports. Consumer Reports 2007, 72 (1). (65) Consumer Reports. Consumer Reports 2007, 72 (6). (66) Consumer Reports. Consumer Reports 2007, 72 (11). (67) Consumer Reports. Consumer Reports 2007, 72 (12). (68) Consumer Reports. Consumer Reports 2008, 73 (2). (69) Consumer Reports. Consumer Reports 2008, 73 (10). (70) Consumer Reports. Consumer Reports 2009, 74 (12). (71) Consumer Reports. Consumer Reports 2010, 75 (7). (72) Consumer Reports. Consumer Reports 2010, 75 (12). (73) Consumer Reports. Consumer Reports Buying Guide 2007; Consumers Union of United States: Yonkers, NY, 2006. (74) Consumer Reports. Consumer Reports Buying Guide 2008; Consumers Union of United States: Yonkers, NY, 2007. (75) U.S. Department of Labor, Bureau of Labor Statistics (U.S. BLS). Producer price index (PPI). http://bls.gov (accessed March 2013). (76) U.S. BLS. Consumer price index (CPI). http://bls.gov (accessed March 2013 and July 2014). (77) U.S. Department of Commerce, Bureau of Economic Analysis (U.S. BEA). 1992 Benchmark I-O bridge tables to PCE and PDE; 1992; http://www.bea.gov (accessed February 2014). (78) U.S. BEA. 1997 Bridge tables to personal consumption expenditures and private fixed investment in equipment and software; 1997; http:// www.bea.gov (accessed February 2014). (79) U.S. BEA. Benchmark input-output data. 2002 Bridge tables to personal consumption expenditures (PCE) and private fixed investment in equipment and software; 2002; http://www.bea.gov (accessed February 2014). (80) U.S. BEA. 2007 Bridge Tables to Personal Consumption Expenditures; 2007; http://www.bea.gov (accessed February 2014). (81) Babbitt, C. W.; Kahhat, R.; Williams, E.; Babbitt, G. Evolution of product lifespan and implications for environmental assessment and management: A case study of personal computers in higher education. Environ. Sci. Technol. 2009, 43, 5106−5112. (82) Teehan, P.; Kandlikar, M. Sources of variation in life cycle assessments of desktop computers. J. Ind. Ecol. 2012, 16, S182−S194. (83) Arushanyan, Y.; Ekener-Petersen, E.; Göra, F. Lessons learned − Review of LCAs for ICT products and services. Comput. Ind. 2013, 65 (2), 211−234. (84) Williams, E. Arizona State University. Multiple functionality and personal digital infrastructure: substitution versus complementarity; Webbased survey conducted under AT&T Industrial Ecology Fellowship grant. Data available at: https://www.researchgate.net/publication/ 271767873 (accessed February 2015). (85) National Institute for Environmental Studies (NIES). Lifespan database for vehicles, equipment, and structures: LiVES; http://www.nies. go.jp/lifespan/index-e.html (accessed December 2013). (86) U.S. Energy Information Administration (EIA). Annual Energy Outlook 2014; http://www.eia.gov/forecasts/aeo/MT_electric.cfm (accessed December 2014). (87) U.S. EIA. Table 8.1. Average Operating Heat Rate for Selected Energy Sources; http://www.eia.gov/electricity/annual/html/epa_08_ 01.html (accessed December 2014). (88) U.S. EIA. Annual Energy Review; http://www.eia.gov/ totalenergy/data/annual/pdf/sec12_6.pdf (accessed December 2014).

(32) van Putten, I.; Kulmala, S.; Thébaud, O.; Dowling, N.; Hamon, K.; Hutton, T.; Pascoe, S. Theories and behavioural drivers underlying fleet dynamics models. Fish Fish. 2012, 13 (2), 216−235. (33) Field, F.; Kirchain, R.; Clark, J. Life-cycle assessment and temporal distributions of emissions: Developing a fleet-based analysis. J. Ind. Ecol. 2000, 4 (2), 71−91. (34) Levine, S. H.; Gloria, T.; Romanoff, E. A dynamic model for determining the temporal distribution of environmental burden. J. Ind. Ecol. 2007, 11 (4), 39−49. (35) Stasinopoulos, P.; Compston, P.; Newell, B.; Jones, H. M. A system dynamics approach in LCA to account for temporal effectsA consequential energy LCI of car body-in-whites. Int. J. Life Cycle Assess. 2012, 17 (2), 199−207. (36) Guinee, J. B.; R. Heijungs, R.; Huppes, G.; Zamagni, A.; Masoni, P.; Buonamici, R.; Ekvall, T.; Rydberg, T. Life cycle assessment: Past, present, and future. Environ. Sci. Technol. 2010, 45 (1), 90−96. (37) Reid, L.; Sutton, P.; C. Hunter, C. Theorizing the meso level: The household as a crucible of pro-environmental behavior. Prog. Hum. Geogr. 2010, 34 (3), 309−327. (38) Kok, R.; R. Benders, R.; Moll, H. C. Measuring the environmental load of household consumption using some methods based on input−output energy analysis: A comparison of methods and a discussion of results. Energy Policy 2006, 34, 2744−2761. (39) Huijbregts, M.; Rombouts, L.; Hellweg, S.; Frischknecht, R.; Hendriks, A. J.; van de Meent, D.; Ragas, M. J. Is cumulative fossil energy demand a useful indicator for the environmental performance of products? Environ. Sci. Technol. 2006, 40 (3), 641−648. (40) Williams, E.; Ayres, R.; Heller, M. The 1.7 kg microchip: Energy and chemical use in the production of semiconductors. Environ. Sci. Technol. 2002, 36, 5504−5510. (41) Carnegie Mellon University Green Design Institute (CMU). Economic Input-Output Life Cycle Assessment; 2008; http://www.eiolca. net (accessed January 2013). (42) Hendrickson, C. T.; Lave, L. B.; Matthew, H. S. Environmental Life Cycle Assessment of Goods and Services: An Input-Output Approach; Resources for the Future Press: Washington, DC, 2006. (43) Finnveden, G.; Hauschild, M.; Ekvall, T.; Guinee, J.; Heijungs, R.; Hellweg, S.; Koehler, A.; Pennington, D.; Suh, S. Recent developments in life cycle assessment. J. Environ. Manage. 2009, 91 (1), 1−21. (44) Duan, H.; Eugster, M.; Hischier, R.; Streicher-Porte, M.; Li, J. Lifecycle assessment study of a Chinese desktop personal computer. Sci. Total Environ. 2009, 407, 1755−1764. (45) McCormack, R. The plight of American manufacturing. American Prospect. December 21, 2009; http://prospect.org/article/ plight-american-manufacturing (accessed June 2013). (46) Electronics Industries Association (EIA). The U.S Consumer Electronics Industry in Review; Electronic Industries Association, Consumer Electronics Group: Washington, DC, 1991. (47) Consumer Electronics Industry (CEI). The U.S Consumer Electronics Industry in Review; Electronic Industries Association, Consumer Electronics Group: Washington, DC, 1995. (48) Chang, Y.; Ries, R.; Wang, Y. The quantification of the embodied impacts of construction projects on energy, environment, and society based on I−O LCA. Energy Policy 2011, 39 (10), 6321− 6330. (49) Consumer Electronics Manufacturers Association (CEMA). U.S. Consumer Electronics Industry Today; CEMA: Arlington, VA, 1998. (50) Ballou, M. DEC flexes micro muscle with new PCs. Computerworld August 31, 1992. (51) Cheng, C. Motion computing LE1700 tablet PC. PCMAG.com. March 26, 2007; http://www.pcmag.com/article2/0,2817,2107220,00. asp (accessed July 2013). (52) Consumer Reports Books. Audio/Video Buying Guide 1992− 1993 ed.; Consumers Union of United States: Yonkers, NY, 1992. (53) Consumer Reports Books. Audio/Video Buying Guide 4th ed.; Consumers Union of United States: Yonkers, NY, 1995. (54) Consumer Reports. Consumer Reports Buying Guide 2001; Consumers Union of United States: Yonkers, NY, 2000. 2558


Article

Environmental Science & Technology (89) U.S. EIA. Frequently Asked Questions: How much electricity is lost in transmission and distribution in the United States; 2014. http://www. eia.gov/tools/faqs/faq.cfm?id=105&t=3 (accessed December 2014). (90) U.S. EPA. eGRID; 2014.http://www.epa.gov/cleanenergy/ energy-resources/egrid/ (accessed December 2014). (91) Ostendorp, P.; Foster, S.; Calwell, C. Televisions active mode energy use and opportunities for energy savings; Issue paper prepared for the Natural Resources Defense Council, 2005. (92) U.S. EPA. Energy Star products: 20 years of helping America save energy; EPA 430-K-12-001; U.S. EPA: Washington, DC, March 2012; http://www.energystar.gov/ia/products/downloads/ES_Anniv_ Book_030712_508compliant_v2.pdf (accessed January 2014). (93) U.S. EPA. Engage occupants; http://www.gov/buildings/facilityowners-and-managers/existing-buildings/save-energy/engageoccupants (accessed December 2014). (94) U.S. Department of Energy (DOE). Are energy vampires sucking you dry? October 28, 2013; http://energy.gov/articles/are-energyvampires-sucking-you-dry (accessed February 2014). (95) Northeast Energy Efficiency Partnerships (NEEP). Business & Consumer Electronics: A Strategy for the Northeast; NEEP: Lexington, MA, August 2013; http://www.neep.org/efficient-products/businessconsumer-electronics/BCE-Strategy/index. (96) Cooper, T. Slower consumption reflections on product life spans and the ‘throwaway society’. J. Ind. Ecol. 2005, 9 (1−2), 51−67. (97) Electronics Product Assessment Tool (EPEAT). http://www. epeat.net (accessed December 2015). (98) Cox, J.; Griffith, S.; King, G. Consumer understanding of product lifetimes. Resour. Conserv. Recycl. 2013, 79, 21−29. (99) DesMarais, C. Phablets craze goes bigger and adds dual SIM support. PC World.com. January 26, 2013; http://www.pcworld.com/ article/2026484/phablets-craze-goes-bigger-and-adds-dual-simsupport.html#tk.nl_today (accessed February 2013). (100) Venture Beat. Phablets will be a multibillion dollar market by 2018, and Apple has no horse in the race. venturebeat.com;. January 21, 2014; http://venturebeat.com/2014/01/21/phone-Tablet-hybridswill-be-a-multi-billion-dollar-market-by-2018/ (accessed January 2014). (101) CNET. Nokia Lumia 1020 review: photographers, meet your camera phone. CNET.com. July 23, 2013; http://www.reviews.cnet. com/nokia-lumia-1020/ (accessed December 2013). (102) Hachman, M. The Future of the Desktop is a Tabletop. PCWorld.com. February 10, 2014; http://www.pcworld.com/article/ 2095487/the-future-of-the-desktop-is-a-tabletop.html (accessed February 2014). (103) U.S. Department of Transportation, Bureau of Transportation Statistics (U.S. BTS). Table 4-11: light duty vehicle, short wheel base and motorcycle fuel consumption and travel. http://www.rita.dot.gov/bts/ sites/rita.dot.gov.bts/files/publications/national_transportation_ statistics/html/table_04_11.html (accessed January 2014). (104) A. T. Kearny. The E-Reader (R)evolution Here to Stay or a Path to Digital Process?; A.T. Kearny Inc.: Chicago, IL, April 2010; http:// www.atkearney.com/paper/-/asset_publisher/dVxv4Hz2h8bS/ content/the-e-reader-revolution/10192 (accessed October 2013). (105) Winston, A. Resilience in a hotter world: Extreme weather and rising demand for resources call for a fundamentally new strategy. Harvard Bus. Rev. 2014, 4, 56−64. (106) Barns, M. Expect a seismic shift in video consumption. Recode.net. February 26, 2014. http://recode.net/2014/02/26/expecta-seismic-shift-in-video-consumption/ (accessed April 2014). (107) Hertwich, E. Consumption and the rebound effect: An industrial ecology perspective. J. Ind. Ecol. 2005, 9 (1−2), 85−98.

2559


Consumption-Weighted Life Cycle Assessment of a Consumer

Recommend Documents