at SciVerse ScienceDirect

Journal of Cleaner Production 44 (2013) 18e29

Contents lists available

Journal of Cleaner Production

journal homepage: www.elsevier .com/locate/ jc lepro

Manufacturing-focused emissions reductions in footwear production

Lynette Cheah a, Natalia Duque Ciceri b,1, Elsa Olivetti c,*, Seiko Matsumura d, Dai Forterre e, Richard Roth f,Randolph Kirchain g

aMaterials Systems Laboratory, Massachusetts Institute of Technology, 77 Massachusetts Avenue, E38-435, Cambridge, MA 02139, USAbMaterials Systems Laboratory, Massachusetts Institute of Technology, 77 Massachusetts Avenue, E38-436, Cambridge, MA 02139, USAcMaterials Systems Laboratory, Massachusetts Institute of Technology, 77 Massachusetts Avenue, E38-434, Cambridge, MA 02139, USAdASICS Corporation, Head Office, 7-1-1 Minatojima-Nakamachi, Chuo-ku, KOBE 650-8555, JapaneASICS Corporation, ASICS Europe BV, Legal Department, Taurusavenue 125, 2132 LS Hoofddorp, The NetherlandsfMaterials Systems Laboratory, Massachusetts Institute of Technology, 77 Massachusetts Avenue, E38-428, Cambridge, MA 02139, USAgMaterials Systems Laboratory, Massachusetts Institute of Technology, 77 Massachusetts Avenue, E38-432, Cambridge, MA 02139, USA

a r t i c l e i n f o

Article history:Received 20 February 2012Received in revised form14 October 2012Accepted 26 November 2012Available online 12 December 2012

Keywords:Manufacturing vs. materialsUncertaintyFootwearCarbon footprint

* Corresponding author. Tel.: þ1 617 253 0877; faxE-mail addresses: lynette@alum.mit.edu (L.

(N.D. Ciceri), elsao@mit.edu (E. Olivetti), inoue-se@d.forterre@eu.asics.com (D. Forterre), rroth@mit.edu(R. Kirchain).

1 Natalia’s family name is Duque Ciceri.

0959-6526/$ e see front matter � 2012 Elsevier Ltd.http://dx.doi.org/10.1016/j.jclepro.2012.11.037

a b s t r a c t

What is the burden upon your feet? With sales of running and jogging shoes in the world averaginga nontrivial 25 billion shoes per year, or 34 million per day, the impact of the footwear industryrepresents a significant portion of the apparel sector’s environmental burden. A single shoe can contain65 discrete parts that require 360 processing steps for assembly. While brand name companies dictateproduct design and material specifications, the actual manufacturing of footwear is typically contractedto manufacturers based in emerging economies. Using life cycle assessment methodology in accordancewith the ISO 14040/14044 standards, this effort quantifies the life cycle greenhouse gas emissions, oftenreferred to as a carbon footprint, of a pair of running shoes. Furthermore, mitigation strategies areproposed focusing on high leverage aspects of the life cycle.

Using this approach, it is estimated that the carbon footprint of a typical pair of running shoes made ofsynthetic materials is 14 � 2.7 kg CO2-equivalent. The vast majority of this impact is incurred during thematerials processing and manufacturing stages, which make up around 29% and 68% of the total impact,respectively. Other similar studies in the apparel industry have reported carbon footprints of runningshoes ranging between 18 and 41 kg CO2-equivalent/pair (PUMA, 2008; Timberland, 2009).

For consumer products not requiring electricity during use, the intensity of emissions in themanufacturing phase is atypical; most commonly, materials make up the biggest percentage of impact.This distinction highlights the importance of identifying mitigation strategies within the manufacturingprocess, and the need to evaluate the emissions reduction efficacy of each potential strategy. By sug-gesting a few of the causes of manufacturing dominance in the global warming potential assessment ofthis product, this study hypothesizes the characteristics of a product that could lead to highmanufacturing impact. Some of these characteristics include the source of energy in manufacturing andthe form of manufacturing, in other words the complexity of processes used and the area over whichthese process are performed (particularly when a product involves numerous parts and light materials).Thereby, the work provides an example when relying solely on the bill of materials information forproduct greenhouse gas emissions assessment may underestimate life cycle burden and ignore poten-tially high impact mitigation strategies.

� 2012 Elsevier Ltd. All rights reserved.

: þ1 617 258 7471.Cheah), nduquec@mit.eduasics.co.jp (S. Matsumura),(R. Roth), kirchain@mit.edu

All rights reserved.

1. Introduction

In 2010 the world produced and purchased almost 25 billionshoes, nearly all of which (more than 90%) were manufactured indeveloping and transitional economies (IBISWorld, 2010; SportBusiness Research Network, 2011). Not surprisingly, an industryof that scale and geographic footprint has come under great

2 The embodied energy of plastics, excluding PVC is between 75 and 120 MJ/kg.3 Specific energy consumption values for hydraulic, hybrid and all-electric

injection molding machines are 19.0, 13.2 and 12.6 MJ/kg, respectively.4 Energy difference between virgin and recycled aluminum.

L. Cheah et al. / Journal of Cleaner Production 44 (2013) 18e29 19

pressure regarding its social and environmental impact (SportBusiness Research Network, 2011).

In response to these pressures within the footwear and thebroader apparel industry, many companies are active in publishingreports on their products’ environmental performance and corpo-rate social responsibility. Collaborative efforts within the sectorhave also begun to emerge. An industry consortium, driven by theOutdoor Industry Association and the Sustainable Apparel Coali-tion, is developing an Eco Index assessment tool that defines sharedguidelines to help companies measure and evaluate the environ-mental impact of their products (Choinard et al., 2011; Zeller, 2011).One reason that companies are partnering on this issue is theinherent complexity in measuring and improving social and envi-ronmental performance. This complexity stems, in part, froma geographically distributed supply chain and low marketconcentration, as the top players in the sector account for less than10% of total industry revenue (IBISWorld, 2010). Furthermore, dueto the diluted nature of brand owner influence on sizeable,consolidated upstream manufacturers within the apparel sector,the ability to impact and direct the actions of suppliers remainschallenging (Locke and Romis, 2010; Plambeck et al., 2012; Zhuet al., 2011). Additionally, due to the increasing trend of suppliersreceiving inquiries for the same information from differentcustomers, a supplier’s collaboration can become increasinglyreluctant. Partnering enables method harmonization and stan-dardization of the multiple, overlapping survey inquiries on thesupply chain.

One element of environmental performance reporting includesmeasurement of the greenhouse gas (GHG) emissions over a prod-uct’s life cycle, which, even taken on its own, can be a complexprocess. This study documents the process of assessing the so-called carbon footprint of a common footwear product e one pairof running shoes. The first objectives is to carry out a benchmarkingexercise in order to understand the life cycle GHG emissions of anexisting shoe design, including the uncertainty associated withthese calculations, and thereby contribute to carbon footprintingmethodology. This contribution helps shoe designers understandthe impact of the products they are designing, as well as to identifyand evaluate potential improvements in future designs. As will beshown below, this case is of particular interest because, unlike mostreported in the literature, the burden from manufacturing is foundto be the dominant life cycle phase. This paper postulates some ofthe characteristics of a product that lead to manufacturing domi-nance in the resulting footprint, as well as discusses the potentialstrategies for reducing the impact of products without use phaseenergy consumption.

Often in product life cycle assessment (LCA) one or two of thelife cycle phases drive the total impact within a single impactcategory (Ashby, 2009). These main drivers of impact typicallyvary depending on the impact category. Because full assessment istime consuming and complex, such phase dominance has beenused to qualitatively streamline LCA efforts (Kaebernick et al.,2003). Particularly in the conceptual or early phase of productdesign, intuiting a particular product’s impact before executinga complete analysis (or before all information about a product isavailable) enables quick evaluation of product environmentalperformance. Beyond this dominance analysis, previous authorshave recommended focusing instead on decision maker analysisto help support specific actors in prioritizing action, particularlyin the case of manufacturing (Löfgren et al., 2011). Because thephase that dominates impact may vary depending on the cate-gory, any streamlining approach (full LCA, as opposed to onlycarbon footprinting considered here) should consider whethersimplification masks potential trade-offs often occurring amongimpact categories. For the purposes of this study, understanding

the drivers of dominance in the life cycle points to areas whereefforts should focus when developing mitigation strategies forGHG emissions.

In order to systematize this phase dominance streamliningapproach, efforts have been made to group products by character-istics and environmental performance to identify common envi-ronmental behavior of product groups. Through this categorization,several authors have identified the raw materials production phase(hereafter referred to more simply as materials phase) and/or theuse phase as most typically dominant within the life cycle (Sousaand Wallace, 2006; Kaebernick and Soriano, 2000; Hanssen,1999). Sousa performed an empirical study building on the workof others to classify products as material and use phase dominantproducts by product characteristics (Sousa, 2002). This classifica-tion can be summarized in three types:

a) Products with (internal) power consumption during use,including home appliances and vehicles (family cars, aircraft,etc.), among others.

b) Products with (external) power consumption, including prod-ucts that require a particular power consumption activityduring use such as washing, refrigerating, and heating.

c) Products with no (or negligible) power consumption duringuse such as furniture, paper/plastic bags, coffee filters, etc.

In general, it is straightforward to recognize when a product’sburden will be focused in the use phase, based on whether theproduct consumes power during use. The degree of use phasedominance is dictated by its lifetime and energy intensity duringuse. The relative importance of materials (as opposed tomanufacturing) is generally less obvious, but derives directly frombroad trends in energy requirements. Examining the embodiedenergy of materials, or energy to produce a workable unit mass ofraw material, in contrast to the typical manufacturing processesreveals why the materials phase generally dominates the life cyclefor non-power using products. The embodied energy of materialsillustrated in (Ashby, 2009) clusters by materials type (i.e., metals,polymers, ceramics, and hybrids) and ranges from 10e103 MJ/kg(excluding precious metals that are in the order of 103e105 MJ/kg; below 10 MJ/kg are cement and concrete, which do not typi-cally undergo additional processing). Ashby shows that thesematerials are uniformly distributed within this range for metals,ceramics and hybrids; polymers exhibit a more narrowed distri-bution clustered in the median (102 MJ/kg). Compare this distri-bution to the expected range for commonmanufacturing processes,which vary from 1 to 50 MJ/kg (Gutowski et al., 2009) normallydistributed with a mean of approximately 20 MJ/kg. Therefore, theupper bound ofmanufacturing energy requirement is typically at orbelow the lower bound of embodied energy for materials. Forexample, in the case of plastics, the embodied energy of differenttypes of polymers (including feedstock energy)2 (Smil, 2008) isoften one order of magnitude higher than the energy requirementsfor injection molding of polymers3 (Thiriez, 2006). In the case ofaluminum and steel, the embodied energy is 80 to3004 times largerthan the overall energy requirements for machining aluminum and30 times higher for steel. Given their even higher embodied energy,precious metals would be even more subject to this argument (notrelevant for the case of shoes). Nevertheless, as the case presentedhere demonstrates, there are cases where these trends can be

L. Cheah et al. / Journal of Cleaner Production 44 (2013) 18e2920

reversed. As described further in Section 5, reliably identifyingthese cases is valuable to minimize analytical burden and avoidoverlooked opportunities for improvement.

The first four sections of this paper provide an empirical carbonfootprint analysis for the apparel and footwear industry, providingdetails around the overall methodology and documenting theapproach to relevant uncertainty calculations. Beyond this casestudy, Section 5 explores the drivers of dominance among life cyclephases of a product that, in turn, point to potentially high leveragemitigation strategies for GHG emissions. This may help practi-tioners develop a sense of the potential “hotspots” of particularproducts before performing a complete carbon footprinting anal-ysis. Identifying which phases may be significant can informwhereeffort should be directed in resource-intensive, detailed datacollection for LCA more generally. For example, a practitioner whois relying primarily on bill of materials information for a productLCA (rather than quantifying the manufacturing burden) mayunderestimate life cycle burden and preclude identifying truly highimpact mitigation strategies.

2. Case description: materials and methods

There have been several efforts to standardize methodologiesfor quantitative evaluation of environmental impact of products orprocesses. Proposed standards or specifications include those fromthe British Standards Institute (offering the PAS 2050 specification)(British Standards Institute, 2008), the World Resources Institute/World Business Council on Sustainable Development GreenhouseGas Protocol (WRI and WBCSD, 2004) and the InternationalStandards Organization (Choinard et al., 2011) among others(International Organization for Standardization, 2006). The ISO hasdeveloped a standard methodology for LCA as part of its ISO 14000environmental management series. The ISO 14040 standard seriesoutlines four main steps in an LCA: goal and scope definition,inventory analysis, impact assessment, and interpretation ofresults. There exists also ISO series related specifically to GHGaccounting including a draft on carbon footprinting of products(i.e., ISO 14067).

A few companies and reports in the academic literature havedescribed environmental footprint results for apparel. Productassessments have emphasized the high impact of the materialsprocessing and manufacturing phase, accounting for upwards of90% of the burden in synthetic running shoes (Nike, 2010; PUMA,2010). In the apparel and cotton industry, other studies focusedon water footprint have shown the use phase as the highercontributor of impact due to the washing of the fabric (Joule, 2011).Milà et al. (1998) applied LCA to identify the high-impact life cyclephases of women’s leather shoes. These shoes are quite differentfrom synthetic athletic shoes, due to the particularly high impact ofcattle raising and leather processing. Companies have also lookedinto the footprint of leather shoes reporting greater than 90% of theburden arising from materials production, particularly from cattleor pig processing (Barling, 2008). Finally, Woolridge et al. (2006)and Staikos and Rahimifard (2007) explored alternative end-of-life treatments for textiles and footwear. Woolridge et al. quanti-fied the energy footprint of recycling clothing and concludes thatthe reuse and recycling of donated clothing results in a net energybenefit compared to clothing made from virgin materials. Staikosand Rahimifard presented a decision-making model for end-of-lifeshoe waste management (i.e., identifying the most appropriatereuse, recovery and recycling option for postconsumer shoes). Thismodel combined different techniques to calculate qualitative andquantitative factors, including Analytic Hierarchy Process, Cost-Benefit Analysis, and LCA. In addition, as mentioned above, theapparel industry has formed a coalition to develop environmental

assessment tools for their products (Sustainable Apparel Coalition,2012).

2.1. Goal and scope definition

This first part of this study evaluated the life cycle GHG emis-sions associated with a specific model of running shoes, andinvestigated the important drivers behind this GHG impact. Thestudy examined a pair of size 9men’s ASICS GEL -KAYANO 17 shoes,which were manufactured in year 2010 by a major contractmanufacturer in China.

First, the impact of raw material extraction and processing wasincorporated for all materials in the shoe, production scrap, as wellas shoe packaging materials (though not the raw materials’ pack-aging). Secondly, manufacturing and assembly of the product wasconsidered, which encompassed the impact associatedwith the useof factory equipment in China, but did not involve themanufacturing and transport of the equipment to the factory (asthis is assumed to be a small part of the burden for this industry(Frischknecht et al., 2007a)). Transportation to keymarket locationswas included. The burden associated with retailing the product wasnot explicitly included in the footprint analysis, but a smallassessment was done to demonstrate theminor contribution of thisphase to the overall footprint, shown in Section 4.5. Furthermore,the use of the product was included, namely the impact associatedwithwashing the shoes. The impact of producing detergent towashthe shoes, however, was not included based on previous workindicating the small contribution to overall burden (Cullen andAllwood, 2009). Finally, end-of-life disposition and transportationbetween the life cycle phases were integrated in the analysis.

2.2. Inventory data collection

Several sets of data were required in order to estimate the GHGimpact associated with the shoes, including the materials,manufacturing and transportation. For the materials, informationwas gathered on the 26 distinct materials used within the threemain shoe components: the upper, the sole (together weighing674 g), and the packaging that the shoes are sold in (178 g). Theupper is the part of shoe that covers the sides and top of the user’sfoot. The sole can be further decomposed to several parts e theoutsole, midsole, trusstic, gel, and sock lining. A breakdown of theshoes’ mass by part and by material is illustrated in Fig. 1. Pulp inthe packaging, polyurethane and polyester use in the upper, olefincopolymer in the midsole, and diene rubber in the outsole make upa majority (74%) of the shoes’ mass. In addition, material used inproduction but not in the final product was accounted for, such asmaterial lost to scrap.

Information was gathered on the location of each material orpart supplier. Detail was assembled around the processes andgeneral processing parameters used to manufacture and assemblethe shoes.

There are 360 process steps involved in the manufacturing andassembly of running shoes (see Fig. 2). Manufacturing shoes islabor-intensive and most of the processing steps involve cuttingand stitching together the 53 parts in the shoes’ upper. The mid-and out-soles are pressed, while the trusstic is injection molded.The pressed parts have to be buffed and the shoes are assembled(glued and pressed). Most of these steps are performed either byhand or by workers operating individual machines.

For the manufacturing phase, GHG emissions are known to ariseprimarily from fuel combustion and use of electricity to powerequipment and machines, therefore information on fuel and elec-tricity use was collected for each factory building involved inproducing the specific shoe model. Fugitive emissions from the

Fig. 1. Mass of a pair of running shoes, by percent, including packaging (total around 850 g) including a) breakdown by part and b) breakdown by material.

L. Cheah et al. / Journal of Cleaner Production 44 (2013) 18e29 21

manufacturing facility, other than from fuel use, were not includedin the analysis based on lack of data.

In this case, the manufacturing of the shoes was contracted toa large footwear manufacturer based in China. Data on the amountof resources consumed by and emissions from each factory buildinginvolved in the shoe production were collected from the manu-facturer over the production period of August 2010eJanuary 2011,during which time 3.6 million pairs of shoes were produced. Thisincludes data on electricity use, fuel use, water use, and solid wasteemissions. Data on air and water emissions were not available. Dataon the number of parts/shoes produced during this 6-month periodwere also collected (see Appendix). Electricity use was the greatestin the Assembly and Midsole buildings, with a total of 23 GWh overall buildings. Coal was reported to be used in the Outsole andMidsole buildings for heating purposes as part of themanufacturing process. The access to these data was possiblethrough the industryeacademia partnership. The strong involve-ment of the brand owner through the engagement with theirsuppliers played a key role in the collection of these extensive data.

Transportation data were gathered, including transportationmode, shipment volumes and distance for transportingmaterials orgoods in between the shoes’ life cycle phases. For transportation ofraw materials or parts to the manufacturing facility, supplier loca-tions (by country) were obtained. For transportation of finished

Fig. 2. Processing stages in p

shoes, the shipment volumes of shoes to various globalmarkets andtheir destination ports, as well as locations of first distributioncenters, were collected over the period of August 2009 to August2010. This was used to estimate the shipment volume-weightedaverage impact of transporting finished shoes. It was assumedthat no air transportation was used.

For the use phase, the recommended washing instructions forshoes by the company is to use a shoe cleaner or to hand wash withcold water and mild detergent, followed by an air dry. It wasassumed that most shoe users do not wash their shoes more thanthree times over the lifetime of the product.

Assumptions weremade to consider the impact of the end of lifephase. Various regions of the world were investigated in terms ofrates of municipal solid waste disposal practices. In Europe, theEU27 reported in 2008 that 40% of the municipal waste was land-filled, 20% incinerated, 23% recycled and 17% composted (ChineseNational Development and Reform Commission, 2009). In otherless developed countries (e.g., non-OECD country members), ratesof municipal waste to landfill average higher than 90% with the restbeing incinerated and a very small percentage recycled (ChineseNational Development and Reform Commission, 2009). For thecurrent study, based on statistics on U.S. municipal solid wastedisposal (United States Environmental Protection Agency, 2008a), itwas assumed that, at the end of their useful lives, 80% of the shoes

roducing running shoes.

L. Cheah et al. / Journal of Cleaner Production 44 (2013) 18e2922

sold were eventually landfilled, and the remaining shoes wereincinerated. None of the shoes were recycled, so there was nomaterial recovery.

We obtained background inventory data from the life cycleinventory database ecoinvent 2.2. The detailed inventories used forkey materials are provided in the appendices. This European focusof the ecoinvent life cycle inventory database provided one sourceof uncertainty in the study (Frischknecht et al., 2007b; Frischknechtand Rebitzer, 2005).

2.3. Impact assessment

The Global Warming Potential (GWP) impact was determinedaccording to gaseous emissions’ potential to contribute to globalwarming over a 100-year period, based on values published in 2007by the Intergovernmental Panel on Climate Change (IPCC)(Intergovernmental Panel on Climate Change, 2007). The impactsfor all gaseous emissions were evaluated relative to carbon dioxide.Impact values for GWP were reported in terms of an equivalentmass of carbon dioxide (kg CO2-equivalent). Life cycle analyseswere conducted using a combination of models in SimaPro LCAsoftware and Microsoft Excel with a Crystal Ball extension.

The authors acknowledge the limits of a single categoryassessment as potentially neglecting other significant environ-mental impacts. In addition, this work does not investigate socialand economic impacts, both of significant importance to thisindustry. A more comprehensive assessment of the product impactwould include those factors.

3. Uncertainty calculation

While the calculation methods for GWP evaluation aredescribed elsewhere, such as in PAS 2050 (British StandardsInstitute, 2008), this section provides more detail on the methodemployed to estimate the uncertainty in these calculations. Athorough investigation of uncertainty is not the focus of this work;however, providing a description of themodeling approach adds anexample of quantitative uncertainty evaluation to the LCAliterature.

Within LCA literature, uncertainty is divided into two broadclasses: measurement and complex system modeling (Williamset al., 2009). Measurement uncertainty refers to the precision orindeterminate error that is generated by a spread of measurementvalues of a quantity. This is also called parameter uncertainty, asthese observed values (input data) in a model are inherently vari-able and random (Huijbregts et al., 2001). Complex systemmodeling refers to the systematic error generated in the modelingframeworks developed to estimate the LCI impact results and theactual emissions type. The difference in scenarios (normativechoices) and models (mathematical relationships) used for theconstruction of these reference models that estimate impacts arethe main components of source of variability for impact type (Lloydand Ries, 2007). Therefore, data collected on the quantity of mate-rial and energy used in the production of the shoe, as well as theGWP impact associated with the type of material and energy sourcecontain uncertainty. Data uncertainty was incorporated in quantityand type by establishing appropriate ranges and distributionalinformation for the point values provided by the product designersand manufacturers. To calculate the total impact and its associateduncertainty, Monte Carlo statistical simulations were performed toaggregate the uncertainty associated with quantity and type. Thisapproach towards incorporating uncertainty has been used byother LCA practitioners (Sonnemann et al., 2003; Maurice et al.,2000; Lo et al., 2005).

In terms of material quantity, a uniformly distributed 20% errorwithin the scrap loss was assumed to account for the ordinaryvariation of the amounts of raw material used in productioncompared with the amount incorporated in the shoe (e.g., mass ofpelletized polymers into mixing machines to process the materialsinto the shoe soles or the mass of injected material into the molds).For instance, if the reported mass of polyurethane entering theproduction process for a particular part was 100 g and its reportedmass in the shoe was 60 g, the resulting scrap loss is 40%. Thus, thepolyurethane scrap loss used in the analyses was 30%e50% (i.e., 20%error in the scrap loss).

Uncertainty in the material typewas incorporated by estimatinga surrogate impact from the ecoinvent database based on a cate-gorization of the key parts by material family and deriving theuncertainty from the range of materials within the relevant cate-gory. For example, the GWP impact of nylon in the upper part of theshoe was modeled by sampling with equal probability from thetwelve individual database inventories of nylon-related processeswithin ecoinvent. In addition, a middle range of uncertainty wasestimated from the pedigree matrix for the individual entries(Frischknecht et al., 2007b). This approach to materials type spec-ification will be described in greater detail in a forthcomingpublication by several of the authors of this work.

The variation of monthly manufacturing production volumesand utility use determined the uncertainty around the productionquantity parameters, which includes both the electricity and coalused in manufacturing. Lognormal distributions were assumed forboth the electricity and coal quantity with the average and standarddeviation was derived from six months of data. The variation inelectricity GWP impact was a function of grid fuel mix; the emis-sions factor was assumed to be a lognormal distribution witha mean of 0.88 kg CO2-eq/kWh and a standard deviation of0.21 kg CO2-eq/kWh based on the location and temporal variationin the Chinese electric power generation grid mix (China-Electricity-Council, 2010). Coal was used for heating equipment inmanufacturing. The average emissions associated with the use ofcoal was 2.0 kg CO2/kg of coal combusted (see Section 4.3 forfurther explanation). The uncertainty in coal combustion impactwas a function of coal type. China’s coal resources includemost coaltypes, but mainly lignite, sub-bituminous, and bituminous varie-ties, accounting for about 13%, 43%, and 28% respectively of accu-mulated proven coal resources (Sun, 2010). These ratios and theenergy content of each coal type enabled the approximation of0.34 kg CO2-eq/kg of coal combusted for the standard deviation onthe lognormally distributed emissions factor.

4. Case results

4.1. Total life cycle impact

Based on the data furnished by the shoe manufacturer, the totalGHG emissions over the shoes’ life cycle are estimated at14 � 2.7 kg CO2-eq (coefficient of variation ¼ 20%). This includesemissions associated with the scrap material lost during themanufacturing phase. Fig. 3 shows a breakdown of the total GWPimpact by life cycle phase. The material processing andmanufacturing phases of the product’s life cycle dominate its lifecycle GHG emissions, accounting for 97% of total emissions. Themanufacturing burden is over twice that of the materials burden.The impact of each phase is now examined separately in detail.

4.2. Material extraction and processing

Based on the materials described in the inventory section andlife cycle inventory information from ecoinvent 2.2, the material

Fig. 3. Breakdown of total GWP impact of running shoes by life cycle phase.

L. Cheah et al. / Journal of Cleaner Production 44 (2013) 18e29 23

processing phase of the shoes is determined to emit4.0 � 0.36 kg CO2-eq (�9%) of greenhouse gases. Fig. 4 shows thebreakdown of the shoes’ GWP impact, again by part as well as bymaterial, for the mass of materials that are embodied in the shoesonly (excluding scrap). Several insights can be gained from thisfigure. Firstly, the shoes’ upper makes up only 23% of the shoes’mass, but is responsible for 41% of the shoes’ emissions. Secondly,57% of the material emissions arise from the use of polyester andpolyurethane alone. This indicates that shoe designers should focuson upper and sock lining materials, specifically polyester andpolyurethane, in order to reduce the materials impact.

The emissions attributed to scrap material are about a third ofthe total materials processing emissions (see Fig. 5). Material scraploss is the difference between the amount of material used duringthe production of the shoes and the amount of material that endsup within the shoes. Material scrap loss during production is thus

Fig. 4. GWP (kg CO2-eq) impact of materials by percent within a pair of runni

significant and indicates potential for both cost and emissionssavings. For example, scrap loss in the total use of polyurethane toproduce the shoes is 50%. This loss is associated with injectionmolding of more complex parts.

4.3. Manufacturing

GHG emissions from manufacturing are expected to arisefrom three sources: electricity use, fuel combustion, and wastedisposal. Based on this facility-level data, the emissions frommanufacturing activity is estimated at 9.5 � 2.7 (coefficient ofvariation ¼ 28%) kg CO2-eq per pair of shoes produced, repre-senting 68% of the total burden per pair of shoes. A breakdown ofthis impact between the three sources is shown in Fig. 6. The effectof waste disposal is minimal, while the burden is roughly equallyattributed to electricity use and the combustion of coal.

It was assumed that the emissions factor associated with theChinese electricity grid is 0.88 kg CO2/kWh (Chinese NationalDevelopment and Reform Commission, 2009). Uncertainty isincorporated as the variation in electricity GWP impact as a func-tion of grid fuel mix (see Section 3 for further explanation). Theemissions factor for coal combustion in an industrial setting is94 kg CO2/mmBTU (United States Environmental ProtectionAgency, 2008b), and the average heating value of Chinese coal is23 MJ/kg (Sun, 2010), so the emissions associated with the use ofcoal is 2.0 kg CO2/kg of coal combusted. The uncertainty is incor-porated in coal combustion impact as a function of coal type. Forthe disposal of mixed plastics in a sanitary landfill, ecoinvent 2.2estimates the impact at 0.09 kg CO2-eq/kg waste disposed. Forwaste that is sent to amunicipal incinerator, the impact is greater at2.4 kg CO2-eq/kg. However, in China, it is estimated that only 4% ofsolid waste that are disposed of ends up being incinerated, whilethe rest are landfilled (Huang et al., 2006).

4.4. Transportation/logistics

Considering the volume of shipped materials and the locationsof suppliers, as well as emissions factors for transport via freightship (0.011 kg CO2-eq/ton-km) and average truck (0.21 kg CO2-eq/ton-km) obtained from the ecoinvent 2.2 database, the GWP impactof transporting raw materials is estimated at 0.034 kg CO2-eq/pairof shoes.

Given that a pair of shoes in their packaging weighs 852 g, theimpact of this final product’s transportation to market turns out to

ng shoes including a) breakdown by part and b) breakdown by material.

0.0

1.0

2.0

3.0

4.0

5.0

Production (incl. scrap) In shoe only

Ma

te

ria

ls

G

WP, k

gC

O2

-e

q, p

er s

ho

e

Other

PPG

Carbon black (filler)

Pulp

Thermo plastic elastomer

EVA

Diene rubber

Nylon

Olefin copolymer

Polyester

Polyurethane

Polyurethane, -50%

Polyester, -32%Polyester, -32

Fig. 5. GWP (kg CO2-eq) impact of materials due to scrap loss compared with materials in shoe only, per pair of running shoes.

L. Cheah et al. / Journal of Cleaner Production 44 (2013) 18e2924

be 0.24 kg CO2-eq/pair of shoes. Note that this figure varies widelyfrom 0.092 to 1.0 kg CO2-eq/pair of shoes, depending on theproximity of the market to the manufacturing facility. The lowestfigure is for a pair of shoes shipped from China to Hong Kong. Thehighest figure is for a pair of shoes shipped from the samemanufacturing facility to the Quebec province in Canada.

The total emissions associated with transportation phase for anaverage pair of running shoes is 0.30 kg CO2-eq, which makes uponly 2% of the shoe’s total impact. So the impact of transportation isfound to be trivial over the shoes’ life cycle. If the highest figure forthe transport of finished shoes is used (that is, considering a pair ofshoes shipped to Quebec, Canada rather than the average pair ofshoes shipped), this fraction increases to 7%. This is not insignifi-cant, but still small compared to the material processing andmanufacturing phases. More details on the transportation phaseare available in the Appendix.

A small separate analysis on retail impact has been madeto understand the order of magnitude impact of that phase.This analysis includes the energy impact from warehousingand retail. Based on the dimensions of a typical shoe box(35 cm � 15 cm � 22 cm) and energy consumption figures fromtypical warehouse and retail buildings from the Energy InformationAgency (Chinese National Development and Reform Commission,2009), a total impact of 0.25 kg CO2-eq/pair is estimated.

Fig. 6. Manufacturing GWP (kg CO2-eq) impact, per pair of running shoes.

4.5. Use

It is assumed that the GWP impact over the shoes’ use phasewillarise only from the washing of the shoes, as there is no impact fromwearing the shoes. It is assumed that consumers will follow thecompany’s recommendation to hand wash their shoes with coldwater and a mild detergent, followed by an air dry. So emissionswill only arise from the treatment of wastewater generated fromwashing shoes. Not including the life cycle emissions of thedetergent, and assuming 90 L of water is used over the lifespan ofthe shoe to wash the shoes, the magnitude of treating this volumeof wastewater is only 0.03 kg CO2-eq.

4.6. End-of-life treatment

Emissions associated with the end-of-life treatment of therunning shoes are also small. As mentioned, it has been assumedthat at the end of their useful lives, 80% of the shoes are landfilled,while the remaining shoes are incinerated. This is based on thestate of non-recyclable municipal solid waste treatment in the U.S.The end-of-life emissions are 0.37 kg CO2-eq, or only 3% of the totallife cycle impact. In terms of the potential for recycling runningshoes, efforts to date seem limited to recycling of outsole, midsoleand upper materials into filler for indoor and outdoor flooringmaterial. Closed loop recycling of shoe materials into shoe mate-rials is rare in the industry (Staikos and Rahimifard, 2010).

5. Discussion

Because rawmaterials can be a primary driver of carbon impact,practitioners rely heavily on the bill of materials to approximateproduction impact to lower the data collection burden of LCA.Previous work on product attribute-based categorization and theorder of magnitude analysis described in the introduction abovesupport this approach to streamlining LCA. However, where themanufacturing contributes to a significant portion of the overallburden this practicewould underestimate the total carbon footprintand overlook high leverage mitigation strategies. Furthermore,scrap generated in production may be very significant and wouldmisjudge the real impact if omitted, because of focus on just mate-rials in the final product (Löfgren et al., 2011). It is therefore useful tohypothesize characteristics of a system when manufacturing is ex-pected to be a major contributor and what drives this contribution.

One challenge in identifying such situations is the assignment ofdistinct boundaries between materials and manufacturing in cases

L. Cheah et al. / Journal of Cleaner Production 44 (2013) 18e29 25

when the structure of the supply chain blurs the boundariesbetween the two phases or when the production of semi-finishedand finished goods occur at the same facility. For example, in thecase of electronic-containing products, it is complex to define theseboundaries, as the production of components often combines bothphases (e.g., metal deposition and substrate etching to produce anintegrated circuit). Other industries, such as paper production(Lopes et al., 2003), food production, or cement manufacture, arechallenging to divide by phase because raw material trans-formation and “product” manufacturing happen at the samefacility. In paper production, for instance, the pulping and paper-making often occur using the same machine to produce a semi-finished good (i.e., a one-ton paper roll) which is then transferredto a paper converting step to produce the finished products (e.g.,paper towels, facial tissues, etc.). In general, where these bound-aries are not distinct, the materials andmanufacturing contributioncan be accounted for jointly and the limitations of depending on thebill of materials are not as significant.

For cases where such a distinction does exist, this sectionexplores two potential reasons for the importance ofmanufacturingin the overall impact of this product: the source of energy employedat the facility and the form of manufacturing (coupled with thecharacteristics of the material being manufactured).

5.1. Source of energy

An important driver for manufacturing as a major contributor isthe source of energy used in manufacturing or the fuel mix forelectricity available at the facility. In this particular study, resultsshow that not only is the electricity in China highly derived fromcoal (which leads to a high grid emissions factor), but also, half ofthe emissions in manufacturing derive from the direct use of coalfor heat required in production.

In order to examine the breakdown of impact independently ofthe GWP-intensity of the energy source, the burden in terms ofcumulative energy demand (CED) per pair of shoes is calculated(Fig. 7). The fraction of the energy used in manufacturing is equallyattributed to the use of electricity and the combustion of coal, and isstill on the order of the materials impact. While recalculating theproduct burden based on energy explains some of why the GWP-intensity for manufacturing is high as shown in Fig. 3, the burdenis still on the order of material; therefore materials non-dominancestill requires some exploration.

5.2. Form of manufacturing

Another way that manufacturing can provide an appreciableportion of the life cycle burden is through the type of

Fig. 7. Cumulative energy demand (CED) of running shoes by life cycle phase.

manufacturing required in production. For example, productscontaining electronics can be manufacturing energy intense (e.g.,personal computers (Williams, 2004; Deng et al., 2011) and mobilephones (Yu et al., 2010)). A study of energy requirements on 20different manufacturing processes showed how energy intensity issignificantly increased with micro/nano processes ranging from104e109 MJ/kg of processed material. The trend observed on theseprocesses showed thework being performed at smaller dimensionsas well as at lower rates result in lowering the throughput in termsof mass and increasing the electrical work requirements (Gutowskiet al., 2009). This suggests that the drivers may be the complexnature of the manufacturing processes, the low mass throughput,and the magnitude of the area over which the processes areperformed.

While running shoes do not require anywhere near the processcomplexity seen in electronics, the required manufacturingprocesses do occur on the small, light components. As an example,in the case of running shoes, one of the main process contributors isthe injection molding of several pieces of the sole. In order toinvestigate this hypothesis, the power consumption, cycle time, andamount of material processed for the specific injection moldingmachines used by the upstream footwear manufacturer werecollected, resulting in 19e29 MJ/kg of polymer processed(excluding any auxiliary equipment or the efficiency of the electricgrid). An empirical environmental study of injection moldingconcluded that the average values for hydraulic, hybrid, and all-electric machines are 1.4e3.4 MJ/kg of polymer processed(Thiriez, 2006).

A direct comparison of these two outcomes suggests that theenergy magnitude of the processes in shoe manufacturing mayindividually contribute to a greater percentage of the total burden.Similar relationships are observed for foaming machines, as well assome of the heating processes. It is noted that a much morecomplete quantitative analysis should be conducted for severaladditional steps within the process in order to draw generalizedconclusions. The above observations would suggest thatmanufacturing may be significant when occurring in areas withcarbon intensive energy sources and when the product involvesmany small light components e even if produced using conven-tional processes.

5.3. Emissions reduction mitigation strategies

By understanding the drivers of GWP burdenwithin a product, itis possible to identify high leverage approaches to mitigate theseimpacts. The field of industrial ecology proposes implementationstrategies toward improving environmental impact, includingsubstitution, dematerialization, and waste mining. Several guide-lines for ecodesign have also been proposed including 1) choice ofmaterials and components towards recyclability, durability, reus-ability and low toxicity, 2) renewable energy use, 3) reduction inenergy or materials intensity and 4) wasteminimization (Borchardtet al., 2011; Niinimaki and Hassi, 2011). If product LCA or carbonfootprinting results emphasize primarily the raw materials phase,reduction efforts might focus on substitution of the product’smaterials, where possible, for recycled or even bio-derived mate-rials. Use of more recycled materials enables some GWP mitigationin the case of the product explored here, but finding ways to couplebeneficial material use with streamlined manufacturing processeshas the potential to create significant emission reductions. Thissection outlines potential mitigation strategies that focus onmanufacturing in terms of two major categories: parts consolida-tion and process efficiency.

With over 65 parts in the shoe, many of which require labor-intensive activities like stitching or cutting to assemble,

Fig. 8. Rank ordered list of mitigation strategies by percent potential reduction from the baseline footprint.

L. Cheah et al. / Journal of Cleaner Production 44 (2013) 18e2926

opportunities are explored where components of the same mate-rial could be consolidated in order to eliminate steps in themanufacturing process. Two adjacent parts of the upper shoe havebeen identified that can be combined to achieve two potentialimprovements. Firstly, the amount of material lost to scrap isreduced, thereby enabling the purchase of less material. Secondly,the combination of parts eliminates three cutting and weldingprocess steps, reducing the electricity burden to operate themachines.

In terms of design, alternative strategies for including estheticfeatures not related to performance of the shoe are assessed. Forexample, there are opportunities where design features could beprinted onto the base fabric rather than affixing the features to theshoe through cut and weld processes. This change has the dualbenefit of eliminatingmaterial and avoiding thewelding and cuttingprocess steps. The most significant savings arise from the avoidedelectricity burden of the cutting and welding machines, but thereduction in materials use also makes a contribution. Furthermore,opportunities are identifiedwhere adjacent componentsof differentcolors could be combined to be the same color. This small changeeliminates cutting process steps and reduces the operation time ofthe pressing machine without impacting product performance.

Some processes have been identified where machinery could beused more efficiently. In each case, there is downtime betweenproduction cycles where the machinery is left to idle (consumingenergy) while the next batch of components is being prepared. Bymodifying the procedure such that while a batch is in the machine,staff could be staging the next set of components, the machinerycould remain in continuous use. In all cases, a costebenefit analysisis required to fully understand the economic benefits of potentialstrategies.

The paragraphs above described qualitatively some generalstrategies for mitigation that are explored by this research after thebaseline footprint is determined. Fig. 8 below shows a rank orderedlist of specific strategies the study quantifies. These strategiesexplore design, materials, and process changes by the potentialmagnitude of reduction from the baseline footprint described in thecase study results and are done in close cooperation with productdesigners at ASICS. Specifically three strategies for consolidationare investigated, one for the upper and one for the lower and onethat eliminate parts by printing color directly onto base fabric;three strategies involving materials substitution for lighter weight,recycled and bioderived. The final strategy explored looks at

process modification. The percentages shown in Fig. 8 are deter-mined based on the assumed embodied energy change, changes inmaterial mass, and assumptions around energy savings inproduction (by looking at individual machine impact). The lattertwo quantities are provided through discussions with the brandowner and manufacturing partner. While material substitution forlighter weight, recycled, or bio-derived materials does provideimpact reduction, the greatest GHG emissions savings can be foundin the strategies that involve parts consolidation. As discussedabove, this is because of the dematerialization as well as processefficiency improvements, which are reflected also in process timeimprovements, highlighting the importance of targetingmanufacturing aspects. It is noted that the rank order of theseresults depends on the particular mass of components underconsideration.

As a result of this analysis, along with detailed discussions withdesigners, product managers and corporate strategy individualswithin the company, several mitigation strategies have beenimplemented in the production of the next model of the KAYANO17. Enacting the materials focused strategies alone results in a 24%reduction of the materials portion of the total impact. Whilemanufacturing focused strategies are also being targeted, thequantification of the manufacturing impact reduction needs to becoupled with further energy auditing of the facility and will thus beincluded in future work by the company and its upstream partner.

6. Conclusions

One purpose of this study is to determine the carbon footprint,or life cycle GWP, of a pair of running shoes and to suggest strate-gies to reduce the product’s impact. The analysis is also used toinvestigate the characteristics of products that show a highmanufacturing phase burden. The carbon footprint results indicatethat the impact is 14 � 2.7 kg CO2-equivalent over the shoes’ life-span, from cradle to grave. There are negligible emissions expectedover the shoes’ use phase, and the transportation and end-of-lifephases only contributed nominally to the shoes’ overall impact.By carrying out this study, GWP hotspots, or materials/processes ofparticularly high impact, are identified. It has been determined thatmost of the emissions are released during shoes’ material pro-cessing (29%) and manufacturing phase (68%).

Using the hotspots as a guide, a number of mitigation strategieswithin the materials and manufacturing phases are considered that

Table 2Process materials inventory used for each component.

Category Component Materials SimaPro process

Upper Polyester Polyester resin, unsaturated,at plant/RER U

Polyurethane Polyurethane, flexible foam,at plant/RER U

Nylon Nylon 66, at plant/RER UCotton Textile, woven cotton, at

plant/GLO UPPG Polyols, at plant/RER UTDI Toluene diisocyanate, at

plant/RER UPolypropylene Polypropylene, granulate,

at plant/RER UInsole Olefin copolymer n-Olefins, at plant/RER U

EVA Ethylene vinyl acetatecopolymer, at plant/RER U

Sole Midsole Olefin copolymer n-Olefins, at plant/RER UEVA Ethylene vinyl acetate

copolymer, at plant/RER UDi-a-cumylperoxide(curing agent)

Hydrogen peroxide, 50% inH2O, at plant/RER U

Magnesiumcarbonate(filler)

Magnesium sulfate, atplant/RER U

Azodicarbonamide(blowing agent)

N,N-Dimethylformamide,at plant/RER U

Outsole Diene rubber Synthetic rubber, at plant/RER U

Natural rubber Natural rubber based sealing,at plant/DE U

Silicon dioxide(filler)

Silica sand, at plant/DE U

Paraffin oil Paraffin, at plant/RER UN,N0-Dinitrosopentamethylene(blowing agent)

Methylamine, at plant/RER U

Rubber colorconcentrates(pigment)

Pigments, paper production,unspecified, at plant/RER U

Carbon black(filler)

Carbon black, at plant/GLO U

Other Polyurethane Polyurethane, flexible foam,at plant/RER U

Socklining Polyester Polyester resin, unsaturated,at plant/RER U

Thermoplasticelastomer/styrenebutadienecopolymer

Acrylonitrile-butadiene-styrenecopolymer, ABS, at plant/RER U

Other Shoelace Polyester Polyester resin, unsaturated, atplant/RER U

Packaging Pulp Kraft paper, bleached, at plant/RER UThermo-mechanical pulp, atplant/RER UPaper, newsprint, at regionalstorage/RER UCore board, at plant/RER U

L. Cheah et al. / Journal of Cleaner Production 44 (2013) 18e29 27

do not affect the product’s performance. In the material processingphase, use of polyester and polyurethane in the shoes’ upper isfound to contribute to almost 60% of emissions; substituting a lesscarbon-intensive material, such as a recycled material, couldgreatly reduce the impact. Furthermore, consolidating similaradjacent parts could eliminate production steps and minimizescrap loss. Also, shifting to printing design elements onto the shoerather than affixing additional material can save energy andmaterial. In terms of the manufacturing phase, finding cleaneralternatives for heating, pursuing energy-efficiency improvementsrelated to the sole production and assembly processes, andreducing machinery idle time would help to lower the GWP of theproduct.

Gaining an understanding of product types that may trendtowards a higher impact in the manufacturing phase could help tocue practitioners when to look beyond the bill of materials whenquantifying life cycle impacts and developing mitigation strategies.This analysis points to two factors that appear to drive an increasedenergy burden in the manufacturing phase of a product’s lifecycle. The form of manufacturing can lead to a higher carbonfootprint, particularly when a product involves many small, lightcomponents e even if produced using conventional processes. Thesource of energy for the manufacturing site can contribute signifi-cantly to the impact, especially when production occurs in areaswith carbon intensive energy sources.

Acknowledgments

The authors would like to thank other ASICS team members forbeing forthcoming with the data and providing logistical support,specifically the contribution of Norio Takaoka and Toshirou Ikezaki.Finally, the authors would like to acknowledge the researchconsultation of Edgar Blanco and the help of Suzanne Greene.

Appendix

There are several materials used in the shoes that are notavailable in the ecoinvent 2.2 life cycle inventory (LCI) database. Inthese cases, the closest match is used, and these substitutes arelisted in Table 1. The impact of these materials on the total GWPimpact is noted to be small. Table 2 shows the materials inven-tory for each component and the respective process data used fromthe SimaPro database. Table 3 shows the manufacturing datacollected from various factory buildings involved in shoe produc-tion, over the period of August 2010 to January 2011. Otherassumptions are made on the transportation distances as well asmode for the other transportation legs, and all transportation legsaccounted for are summarized in Table 4 below. For example, it isassumed that waste generated from the manufacturing facility inChina is transported over a distance of 100 km by lorry to thelandfill or incinerator.

Table 1Closest matches for materials in shoe that are not available in the ecoinvent database.

Material in shoe Substitute material referenced in ecoinventfor LCI data

Resulting GWP impact,kg CO2-eq

% of total GWPimpact

Aluminum coated glass beads (in logo) Glass fiber 0.0012 0.024%Di-a-cumyl peroxide (curing agent) Hydrogen peroxide 0.0019 0.038%Magnesium carbonate (filler) Magnesium sulfate 0.0039 0.078%Azodicarbonamide (blowing agent) N,N-dimethylformamide 0.027 0.54%N,N0-dinitroso pentamethylene

(blowing agent)Methylamine 0.015 0.30%

Chemicals (in outsole) Acrylonitrile butadiene styrene (ABS) 0.022 0.45%Total: 0.071 1.4%

Table 3Manufacturing data for production period August 2010 to January 2011.

Factory building Production Resource use & emissions

Total Units Electricity demand, kWh Coal use, tons Water use, tons Solid waste, tons

Assembly 3,640,833 Pairs of shoes 7,393,934 0 506,324 276Outsole 12,959,528 Fores, heels 2,026,545 2074 27,747 223Midsole 6,062,499 Soles 11,636,850 10,158 1,014,218 484Injection molding 14,856,145 Parts 1,391,566 0 55,819 20Cardboard box 9,176,255 Boxes 113,702 0 0 427

Total: 22,562,597 12,232 1,604,108 1431

Table 4Assumptions made (*) and resulting emissions for the transportation phase of shoes.

Product Origin Destination Distance Mode Emissions, kg CO2-eq

Raw materials Various China Various Various 0.03Production waste China Landfill/incinerator 100 km* Freight lorry 0.01Finished shoes China 1st distribution center Various Various 0.15 (average)Finished shoes Distribution center Retailer 500 km* Freight lorry 0.09Retired shoes Customer Landfill/incinerator 100 km* Freight lorry 0.01

Total: 0.30

L. Cheah et al. / Journal of Cleaner Production 44 (2013) 18e2928

References

Ashby, M.F., 2009. Materials and the Environment: Eco-informed Material Choice.Butterworth-Heinemann/Elsevier, Amsterdam; Boston.

Barling, R.L., 2008. Carbon Footprint of a Shoe Produced by an InternationalEnterprise with an Implemented Corporate Social Responsibility Strategy,Taking PUMA AG as Example. Fachhochschule für Technik und Wirtschaft,Berlin, Germany.

Borchardt, M., Wendt, M.H., Pereira, G.M., Sellitto, M.A., 2011. Redesign ofa component based on ecodesign practices: environmental impact and costreduction achievements. Journal of Cleaner Production 19 (1), 49e57.

British Standards Institute, 2008. PAS 2050:2008 Specification for the Assessmentof the Life Cycle Greenhouse Gas Emissions of Goods and Services. BSI, London.

China-Electricity-Council, 2010. China Electric Power Industry Statistics Analysis.Chinese National Development and Reform Commission, 2009. Baseline Emission

Factor for Chinese Regional Grids.Choinard, Y., Ellison, J., Rideway, R., 2011. The sustainable economy. Harvard Busi-

ness Review.Cullen, J.M., Allwood, J.M., 2009. The role of washing machines in life cycle

assessment studies. Journal of Industrial Ecology 13 (1), 27e37.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 CleanerProduction 19 (11), 1198e1206.

Frischknecht, R., Rebitzer, G., 2005. The ecoinvent database system: a comprehen-sive web-based LCA database. Journal of Cleaner Production 13 (13e14), 1337e1343.

Frischknecht, R., Althaus, H.J., Bauer, C., Doka, G., Heck, T., Jungbluth, N., et al., 2007.The environmental relevance of capital goods in life cycle assessments ofproducts and services. The International Journal of Life Cycle Assessment 12, 7e17.

Frischknecht, R., Jungbluth, N., Althaus, H.-J., Doka, G., Heck, T., Hellweg, S., et al.,2007. The Ecoinvent Database System: Overview and Methodology. SwissCentre for Life Cycle Inventories, Dubendorf, Switzerland.

Gutowski, T.G., Branham, M.S., Dahmus, J.B., Jones, A.J., Thiriez, A., 2009. Thermo-dynamic analysis of resources used in manufacturing processes. EnvironmentalScience & Technology 43 (5), 1584e1590.

Hanssen, O.J.,1999. Sustainable product systemsdexperiences based on case projectsin sustainable product development. Journal of Cleaner Production 7, 27e41.

Huang, Q., Wang, Q., Dong, L., Xi, B., Zhou, B., 2006. The current situation of solidwaste management in China. Journal of Material Cycles andWaste Management8 (1), 63e69.

Huijbregts, M., Norris, G., Bretz, R., Ciroth, A., Maurice, B., von Bahr, B., et al., 2001.Framework for modelling data uncertainty in life cycle inventories. The Inter-national Journal of Life Cycle Assessment 6 (3), 127e132.

IBISWorld, 2010. Industry Report: Global Footwear Manufacturing.Intergovernmental Panel on Climate Change, 2007. Contribution of Working Group I

to the Fourth Assessment Report of the Intergovernmental Panel on ClimateChange. In: Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B.,Tignor, M., Miller, H.L. (Eds.), Climate Change 2007: the Physical Science Basis.Cambridge, United Kingdom.

International Organization for Standardization, 2006. ISO 14044 e EnvironmentalManagement e Life Cycle Assessment e Principles and Framework. ISO,Geneva, Switzerland.

Joule, E., 2011. Fashion-forward thinking: sustainability as a business model at LeviStrauss. Global Business and Organizational Excellence 30 (2), 16e22.

Kaebernick, H, Soriano, V. 2000. Approach to simplified environmental assessmentby classification of products. In: 7th CIRP International Seminar on Life CycleEngineering, Tokyo, pp. 163e169.

Kaebernick, H., Sun, M., Kara, S., 2003. Simplified lifecycle assessment for the earlydesign stages of industrial products. CIRP Annals e Manufacturing Technology52 (1), 25e28.

Lloyd, S.M., Ries, R., 2007. Characterizing, propagating, and analyzing uncertainty inlife-cycle assessment e a survey of quantitative approaches. Journal of Indus-trial Ecology 11 (1), 161e179.

Lo, S.-C., Ma, H.-W., Lo, S.-L., 2005. Quantifying and reducing uncertainty in lifecycle assessment using the Bayesian Monte Carlo method. Science of the TotalEnvironment 340 (1e3), 23e33.

Locke, R., Romis, M., 2010. The promise and perils of private voluntary regulation:labor standards and work organization in two Mexican garment factories.Review of International Political Economy 17 (1), 45e74.

Löfgren, B., Tillman, A.-M., Rinde, B., 2011. Manufacturing actor’s LCA. Journal ofCleaner Production 19 (17e18), 2025e2033.

Lopes, E., Dias, A., Arroja, L., Capela, I., Pereira, F., 2003. Application of life cycleassessment to the Portuguese pulp and paper industry. Journal of CleanerProduction 11 (1), 51e59.

Maurice, B., Frischknecht, R., Coelho-Schwirtz, V., Hungerbühler, K., 2000. Uncer-tainty analysis in life cycle inventory. Application to the production of electricitywith French coal power plants. Journal of Cleaner Production 8 (2), 95e108.

Milà, L., Domènech, X., Rieradevall, J., Fullana, P., Puig, R., 1998. Application of lifecycle assessment to footwear. The International Journal of Life Cycle Assess-ment 3 (4), 203e208.

Niinimaki, K., Hassi, L., 2011. Emerging design strategies in sustainable productionand consumption of textiles and clothing. Journal of Cleaner Production 19 (16),1876e1883.

Nike. Corporate Responsibility Report FY07-09. 2010.Plambeck, E., Lee, H., Yatsko, P., 2012. Improving environmental performance in

your Chinese supply chain. MIT Sloan Management Review.PUMA, 2008. PUMAVision Sustainability Report 2007/2008. PUMA, Germany.PUMA. Annual Report 2010. 2010.Smil, V., 2008. Energy in Nature and Society: General Energetics of Complex

Systems. MIT Press, Cambridge, Mass.Sonnemann, G.W., Schuhmacher, M., Castells, F., 2003. Uncertainty assessment by

a Monte Carlo simulation in a life cycle inventory of electricity produced bya waste incinerator. Journal of Cleaner Production 11 (3), 279e292.

Sousa, I., 2002. Approximate Life-cycle Assessment of Product Concepts UsingLearning Systems. Massachusetts Institute of Technology, Cambridge,Massachusetts.

Sousa, I., Wallace, D., 2006. Product classification to support approximate life-cycleassessment of design concepts. Technological Forecasting & Social Change 73,228e249.

Sport Business Research Network, 2011. Consumer Expenditures on Running/Jogging Shoes.

Staikos, T., Rahimifard, S., 2007. A decision-making model for waste management inthe footwear industry. International Journal of Production Research 45 (18e19),4403e4422.

Staikos, T., Rahimifard, S., 2010. Post-consumer waste management issues in thefootwear industry. Proceedings of the Institution of Mechanical Engineers, PartB: Journal of Engineering Manufacture, 363e368.

Sun, G., 2010. Coal in China: Resources, Uses, and Advanced Coal Technologies. In:White Paper, Pew Coal Initiative.

Sustainable Apparel Coalition. 2012.

L. Cheah et al. / Journal of Cleaner Production 44 (2013) 18e29 29

Thiriez, A., 2006. An Environmental Analysis of Injection Molding. MassachusettsInstitute of Technology.

Timberland. Timberland Green Index Report. 2009.United States Environmental Protection Agency, 2008a. Municipal Solid Waste

Generation,RecyclingandDisposal in theUnitedStates: Facts andFigures for2008.United States Environmental Protection Agency, 2008b. Direct Emissions from

Stationary Combustion Sources, Climate Leaders Greenhouse Gas InventoryProtocol CoreModule Guidance. United States Environmental Protection Agency.

Williams, E., 2004. Energy intensity of computer manufacturing: hybrid assessmentcombining process and economic input-output methods. EnvironmentalScience & Technology 38 (22), 6166e6174.

Williams, E.D., Weber, C.L., Hawkins, T.R., 2009. Hybrid framework for managinguncertainty in life cycle inventories. Journal of Industrial Ecology13 (6), 928e944.

Woolridge, A.C., Ward, G.D., Phillips, P.S., Collins, M., Gandy, S., 2006. Life cycleassessment for reuse/recycling of donated waste textiles compared to use ofvirgin material: an UK energy saving perspective. Resources, Conservation andRecycling 46 (1), 94e103.

WRI and WBCSD, 2004. The Greenhouse Gas Protocol: a Corporate Accounting andReporting Standard.

Yu, J., Williams, E., Ju, M., 2010. Analysis of material and energy consumption ofmobile phones in China. Energy Policy 38 (8), 4135e4141.

Zeller, T., 2011. Clothes makers join to set ‘green score’. New York Times. p. B1.Zhu, Q.H., Geng, Y., Sarkis, J., Lai, K.H., 2011. Evaluating green supply chain

management among Chinese manufacturers from the ecological modernizationperspective. Transportation Research Part E e Logistics and TransportationReview 47 (6), 808e821.