Scholarly article on topic 'Indoor Air Pollutant Exposure for Life Cycle Assessment: Regional Health Impact Factors for Households'

Indoor Air Pollutant Exposure for Life Cycle Assessment: Regional Health Impact Factors for Households Academic research paper on "Environmental engineering"

0
0
Share paper
Academic journal
Environmental Science & Technology
OECD Field of science
Keywords
{""}

Academic research paper on topic "Indoor Air Pollutant Exposure for Life Cycle Assessment: Regional Health Impact Factors for Households"

Subscriber access provided by NEW YORK MED COLL

Article

Indoor air pollutant exposure for life cycle assessment: regional health impact factors for households

Ralph K Rosenbaum, Arjen Meijer, Evangelia Demou, Stefanie Hellweg, Olivier Jolliet, Nicholas L. Lam, Manuele Margni, and Thomas E. McKone

Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.5b00890 • Publication Date (Web): 07 Oct 2015

Downloaded from http://pubs.acs.org on October 7, 2015

Just Accepted

"Just Accepted" manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides "Just Accepted" as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. "Just Accepted" manuscripts appear in full in PDF format accompanied by an HTML abstract. "Just Accepted" manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). "Just Accepted" is an optional service offered to authors. Therefore, the "Just Accepted" Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the "Just Accepted" Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these "Just Accepted" manuscripts.

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036

Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

i Indoor air pollutant exposure for life cycle assessment:

2 regional health impact factors for households

* 12 * 3 45 6 7

3 Ralph K. Rosenbaum ' ' , Arjen Meijer ' , Evangelia Demou ' , Stefanie Hellweg , Olivier Jolliet,

4 Nicholas L. Lam , Manuele Margni , Thomas E. McKone

5 1. Irstea, UMR ITAP, ELSA Research group & ELSA-PACT—Industrial Chair for Environmental

6 and Social Sustainability Assessment, 361 rue J.F. Breton, 5095, 34196 Montpellier, France

7 2. Department of Management Engineering, Technical University of Denmark, Lyngby, Denmark

8 3. OTB Research for the Built Environment, Faculty of Architecture and the Built Environment,

9 Delft University of Technology, Delft, The Netherlands

10 4. Healthy Working Lives Group, Institute of Health and Wellbeing, College of Medical,

11 Veterinary and Life Sciences, University of Glasgow, Glasgow, UK

12 5. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow, Glasgow, UK

13 6. Institute of Environmental Engineering, ETH Zurich, Zurich, Switzerland

* Corresponding authors: R.K. Rosenbaum, Irstea, Industrial Chair ELSA-PACT, 361 Rue Jean-François Breton, BP 5095, 34196 Montpellier, France, Ph. +33 499612048, ralph.rosenbaum@irstea.fr; A. Meijer, TU Delft / Faculty of Architecture and The Built Environment, OTB - Research for the Built Environment, PO Box 5043, 2600 GA Delft, The Netherlands, Ph. +31 152785658, a.meijer@tudelft.nl

7. Environmental Health Sciences, School of Public Health, University of Michigan, Ann Arbor,

U.S.A.

8. School of Public Health, University of California Berkeley, Berkeley, U.S.A.

9. Department of Mathematical and Industrial Engineering, CIRAIG - Polytechnique Montreal,

Montreal, Canada

Abstract

Human exposure to indoor pollutant concentrations is receiving increasing interest in Life Cycle Assessment (LCA). We address this issue by incorporating an indoor compartment into the USEtox model, as well as by providing recommended parameter values for households in four different regions of the world differing geographically, economically, and socially. With these parameter values, intake fractions and comparative toxicity potentials for indoor emissions of dwellings for different air tightness levels were calculated. The resulting intake fractions for indoor exposure vary by two orders of magnitude, due to the variability of ventilation rate, building occupation and volume. To compare health impacts as a result of indoor exposure with those from outdoor exposure, the indoor exposure characterization factors determined with the modified USEtox model were applied in a case study on cooking in non-OECD countries. This study demonstrates the appropriateness and significance of integrating indoor environments into LCA, which ensures a more holistic account of all exposure environments and allows for a better accountability of health impacts. The model, intake fractions, and characterization factors are made available for use in standard LCA studies via www.usetox.org and in standard LCA software.

Keywords

35 Life Cycle Assessment, indoor exposure, USEtox, characterization factor, comparative toxicity

36 potential, cooking, solid fuels, household air pollution

37 TOC/Abstract Art

39 Introduction

40 Life cycle Assessment (LCA) has broad applications in supply chain management and policy analysis,

41 helps to identify effective improvement strategies for the environmental performance of products or

42 services and to avoid burden shifting between different environmental issues.1 Current LCA

43 methodology covers more than a dozen impact categories such as climate change, acidification,

44 eutrophication, land-use, or water-use, as well as toxicity, distinguishing ecotoxicity and human toxicity.

45 The latter currently only considers outdoor exposure to ubiquitous chemical concentrations in the

46 environment (or food) from emissions of a product's or service's life cycle, while indoor exposure with

47 proximity to sources emitting in confined (dilution) volumes have not yet been integrated. It is important

48 to note that LCA employs an "emitter perspective" aiming to assess potential impacts of chemical

49 exposure related to a given emission, i.e. marginal exposure or impact attributable to a specific emission

50 source. This is different from Environmental Risk Assessment, which is based on a "receptor

51 perspective" aiming to measure the level of cumulative exposure from single or multiple sources of

52 chemical emission, no matter where these occur.

Human exposure to indoor concentrations of chemicals is receiving increasing interest in LCA. Due to the often high concentrations of harmful substances in indoor environments and the long periods people spend indoors, the indoor intake per unit of (indoor) emission of these substances can be equal or higher than outdoor intake, by up to several orders of magnitude.4'5 Inclusion of indoor exposure in LCA has been acknowledged as an area of need by the UNEP/SETAC Life Cycle Initiative (http://www.lifecycleinitiative.org), which is taking up recommendations and conclusions toward the enhancement of the current LCA framework study. Within this initiative, an international expert group on the integration of indoor and outdoor exposure in LCA has formulated a framework for integration of indoor exposure in LCA.6 They found that a single-compartment box model is most compatible with LCA and therefore recommended it for use as a default in LCA. Indoor intake fractions were found to be several orders of magnitude higher in many cases than outdoor intake fractions, which highlights the relevance of considering indoor exposure. While an initial set of model parameter values was provided and the integration of the model into the USEtox model was suggested in the previous study, a full set of representative parameter values for various indoor settings is still missing to make this approach operational.6 The model parameters given in the framework have been presented as ranges of values.6 The actual values of the parameters depend on the geographical region of the assessed site, the type and characteristics of the dwelling, and the characteristics and behavior of the occupants. In LCA, when no data are available about the actual dwelling or the occupants, average parameter values are generally used.

USEtox is a tool for calculation of comparative toxicity potentials (characterization factors) for human health and freshwater ecosystems, developed under the auspices of the UNEP/SETAC Life Cycle Initiative. It models a cause-effect chain that links emissions to impacts through three steps: environmental fate, exposure, and effects. It was developed as a methodology simple enough to be used on a worldwide basis and for a large number of substances while incorporating broad scientific

consensus. , It is the recommended LCA (midpoint) toxicity characterization model of the European

Union14, endorsed by the UNEP/SETAC Life Cycle Initiative, and adopted by the US-EPA's life cycle

Page 5 of 23 Environmental Science & Technology

79 impact assessment tool TRACI.15 Therefore, it is regarded as the relevant basis to integrate indoor and

80 outdoor exposure characterization into one consistent method for use in LCA, as also discussed by

81 Hellweg et al.6

82 The aims of this paper are 1) extending the USEtox model to include the indoor environment as a

83 compartment; 2) providing an overview of recommended parameter values to be used as default

84 household model parameters for different geographical settings; 3) comparing intake fractions calculated

85 with these recommended default parameters with intake fractions for outdoor exposure; and, 4) applying

86 the new characterization factors for indoor exposure to a comprehensive case study on cooking

87 worldwide. The scope of this paper is restricted to the LCA emitter perspective, i.e. the calculation of

88 potential health effects from indoor emissions modeled as the cumulative impacts from indoor exposure

89 and outdoor exposure due to indoor emissions only. The focus was on indoor emissions of volatile and

90 semi-volatile organic compounds, because pollutants such as particles, ozone or NOx require specific

91 model processes for transport and transformation and are currently not addressed by LCA toxicity

92 models and not included in USEtox. In LCA their impacts on human health are assessed respectively in

93 the separate impact categories "particulate matter formation" and "photochemical ozone formation".

94 Materials and Method

95 The one-box model recommended by Hellweg et al. for estimation of indoor air intake fraction is

96 given as (Equation 1b in 6):

97 iF = ——--N (1)

V ■ m ■ kex

98 where iF is the population intake fraction of a chemical (-), IR is the daily inhalation rate of air of an

99 individual (m /day), N is the number of people exposed (-),V is the volume of the exposure area (m ),

100 kex is the air exchange rate of the volume in the exposure area (-) and m is the mixing factor (-). The

101 following sections describe how this has been implemented into the matrix-algebra framework of the

102 USEtox model.13

Environmental Science & Technology Page 6 of 23

Overall framework: In the USEtox framework based on Rosenbaum et al.16, the characterization factor matrix that represents the impact per kg substance emitted is obtained by multiplying an intake fraction matrix (iF) by an effect factor matrix (EF). The intake fraction is the product of a fate matrix (FF) and an exposure matrix (XF):16

CF = EFiF = EFXFFF (2)

The unit of the elements in FF is [d], in XF [1/d], in iF [kgintake/kgemitted], in EF [disease cases/kg of chemical intake], and in CF [disease cases/kg emitted] or CTUh, which is the name given by the USEtox developers to the results (characterization factors) of their model for human health (as opposed to CTUe

- Comparative Toxic Unit for ecosystems). For the concept and interpretation of these matrices, their elements and their units we refer to Rosenbaum et al.16 The matrix-algebra based calculation framework of USEtox allows for the straightforward integration of additional compartments and exposure pathways by simply adding the corresponding columns or rows to the respective fate and exposure matrices.16 All parameters describing the indoor compartment and the resulting exposure are provided as recommended value sets for household settings in different regions, but can also be modified freely by the user in the model to represent more site-specific conditions. Fate: The fate matrix FF [d] is calculated as the inverse of the exchange-rate matrix K [1/d]: FF = (- K )-1 (3)

The exchange-rate matrix K represents the exchange rate between compartments in the non-diagonal terms and the overall removal rate in the diagonal term (with a negative sign). The indoor environment is modeled as a separate air compartment contributing to the overall inhalation exposure of humans.

This compartment is added to the existing 11 USEtox compartments. Three removal mechanisms are

considered according to Wenger et al. :

1) The advective ventilation flow, parameterized as the air exchange rate kex [h-1] (as in Equation 1b in

6). The air exchange rate does not depend on the substance, but on the building characteristics, such as

type and size of windows and doors, type of walls, and the number of cracks in the façades. Average

values for several regions are given in Table S1 in the Supporting Information (SI). kex is not a loss, but

ACS Paragon Plus Environment 6

129 an inter-media transport mechanism connecting indoor with outdoor compartments. Based on the

130 average distribution of the global population between urban and rural areas of about 50% respectively ,

131 half of the ventilation flow is directed to each of the urban and continental rural environments of

132 USEtox (Figure 1). This is taken into account in the model by a non-diagonal term from indoor to

133 compartment i given as: kindoori = f<xik<x, with fexi = 0.5 for transfers to both urban and continental

134 rural air compartments (i).

135 2) The gas-phase (g) air-degradation rate kg,deg [h-1] is mainly related to reactions with ozone, hydroxyl

136 radicals, and nitrate radicals (gas-phase degradation). The overall degradation rate in the indoor air is

137 calculated as the average radical concentration ([OH], [O3], [NO3]) multiplied by the corresponding

138 second order degradation rate constant: kg,deg = kOH • [OH] + kO3 • [O3] + kNO3 • [NO3] . Long-term

139 averaged indoor concentrations of ozone ([O3] = 8 ppb), hydroxyl radical ([OH] = 3*10"6 ppb) and

140 nitrate radical ([NO3] = 10° ppb) were taken from Wenger at al. and second order degradation rate

141 constants from the EPI Suite v4.1 software19, which provides OH rate constants for most substances, but

142 only few for O3 and NO3.

143 3) An equivalent removal rate by adsorption to indoor surfaces, ks [h-1] can be calculated as a net

144 removal rate from the air, assuming steady-state conditions between the air and room surface without

145 adding a separate compartment. This approach is similar to the net removal rate calculated in USEtox

146 from the freshwater outdoor environment to the sediments, which are not considered as separate

147 compartments to limit the model complexity. Since degradation on surfaces is not well characterized,

148 this removal rate to surfaces is subject to high uncertainty. Surface removal in the current model is

149 applied primarily to Semi-Volatile Organic Compounds (SVOCs), for which additional gaseous dermal

150 exposure may also be relevant and may compensate this removal. If the model is eventually used for

151 particulate matter (PM) and ozone, then surface removal could become more important and requires

152 further assessment of the literature on indoor ozone and PM deposition including the work of

153 Weschler and Nazaroff . We therefore do not include the sorption removal pathway in the default

model, but only consider it for the sensitivity study together with the dermal gaseous exposure pathway. A more detailed description of the calculation of the equivalent removal rate to the surface ks is given in SI (section S3).

The air degradation rate and the equivalent removal rate to the surface directly add up to the air exchange rate for the diagonal term of K.

I odour air

■ ? Ï

HI air I I

iK'umd Jtmn in

i Boundary laver

Boundary layer-surf wee mu» partíliotirritf

agricultural soil

Surface

lK-urudtflM»n un iurfiRï ^^

natural soil

coastal

freshwater marine _ water

agricultura soil

natural soil

freshwater

Figure 1: Schematic representation of the USEtox model with indoor compartment embedded; adapted from Rosenbaum et al. 3 and Wenger et al.17

Exposure: The exposure pathway considered in this paper is inhalation. The relevant parameters for inhalation exposure in households are the following: individual daily inhalation (breathing) rate (IR)

[m3/d], average number of people in the building N [dimensionless], building volume V [m ], and daily time fraction spent indoors ft [dimensionless]. The latter is the quotient of the time spent indoors and the total time of a day (24h). Recommendations, assumptions, and choices for these parameter values are further discussed below. The exposure factor XF [1/d] for the indoor exposure setting is then calculated based on Equation 1b in Hellweg et al.6 (with mixing factor m = 1, assuming that complete mixing within the indoor volume is an inherent hypothesis of the indoor iF model):

171 XF = — • f • N (4)

172 The calculated XF values are placed in the corresponding element of the exposure matrix XF in

173 USEtox. For SVOCs the dermal absorption of gas-phase chemicals may become important and means

174 that the validity of equation (4) is restricted to VOCs. In this paper the potential influence of the

175 dermal gaseous uptake pathway is considered as a sensitivity study together with the influence of

176 adsorption removal on indoor surfaces which competes with this exposure pathway. Existing

177 approaches17,26 were adapted to determine the convective transfer at body surface as a function of heat

178 transfer coefficients , which might be added to USEtox in a later stage once data will be broadly

179 available and the models further evaluated, in conjunction with the introduction of a dermal pathway

180 within USEtox.

181 Effect and characterization factor: The human health effect factor EF is the same as for outdoor

182 exposure in USEtox and thus also independent of the exposure setting or region. Therefore, EF was

183 taken directly from the USEtox database. According to Rosenbaum et al.16 the characterization factor

184 matrix CF (named HDF in 16) is then obtained by multiplying the matrices FF, XF, and EF (Equation

185 2).

186 Model Parameterization

187 In order to calculate characterization factors (and intake fractions) for indoor exposure, the parameters

188 discussed above are needed in the USEtox model. In LCA, the exact situation where the indoor exposure

189 takes place is seldom known. In order to calculate characterization factors for generic situations, regions

190 can be defined, for each of which a characterization factor can be calculated using region-specific

191 parameters. Regions can be defined as 1) countries or continents, 2) based on the level of economic

192 development or urbanization, or 3) as a combination of 1) and 2).

193 For several parameters, the data availability is limited for most regions, especially for non-OECD

194 countries. Especially for houses with low air-exchange in non-OECD countries, few data about the

195 parameters needed for the calculations are available, specifically for building volumes (V), occupation

(N), and air exchange rate (kex). You et al. found air exchange rates in 41 elderly homes in China

ranging from 0.29 hr-1 to 3.46 hr-1 in fall (median: 1.15 hr-1), and from 0.12 hr-1 to 1.39 hr-1 in winter -1 28

(median: 0.54 hr- ). Massey et al. found air exchange rates in 10 houses in northern India ranging from 2.5 hr-1 to 3.1 hr-1 in winter and 4.6 hr-1 to 5.1 hr-1 in summer.29 These data suggest that air exchange rates in houses with low air-exchange in non-OECD countries may be higher than in houses with low air-exchange in OECD countries. However, it is not clear how representative the dwellings described by You et al. and Massey et al. are for all houses with low air-exchange in the respective countries.

Therefore, four regions have been defined in this study: Europe (EU-27), North America (USA), OECD countries, and non-OECD countries. We assume that a population-weighted average from EU-27 countries is representative for Europe, that an average from the USA is representative for North America, that a population-weighted average from EU-27 countries and the USA is representative for OECD countries, and that a population-weighted average from China, India, Uganda, Brazil, and Guatemala is representative for non-OECD countries. The region-specific parameters considered are the building volume (V) and the number of people in the building (N). For the air exchange rate (kex) data availability is even less robust than for N and V. Therefore, a distinction has been made between houses with a low air exchange rate (kex < 8 h-1) named "L-AER" and houses with higher air exchange rates (kex > 8 h-1, especially for houses with no windows and/or doors) named "H-AER". All houses in OECD countries were assumed as having a relatively low air-exchange, while in non-OECD countries, houses with both low and high air exchange (e.g. houses with no glass in the windows) exist. In the absence of data for houses with low air-exchange in non-OECD countries, we assume the same value for kex as for OECD countries. In Table 1, the recommended values of the region-specific parameter sets are summarized. In SI (Table S1), the parameter values are given for the different countries within the regions.

Table 1: Recommended parameter values and standard deviations (SD) for the indoor exposure model per region, calculated as averages from the individual countries and weighted over the population of those countries

Region V [m3l N [-l kex [h-1l IR [m3/dl ft [-l

Average SD Average SD Average SD 13 0.58

Non-OECD countries (H-AER building) 119 25.6 4.0 0.87 15.6 0.85

Non-OECD countries (L-AER building) 0.64 0.08

OECD countries 236 37.9 2.5 0.22

Europe (EU-27) 209 22.9 2.4 0.26

North America (USA) 277 a 2.6 a

a single data point (US average) as we are using country averages and hence no variability assessed on sub-country level See Table S1 in SI for data per country and literature references

223 We assume the daily individual inhalation rate for humans for indoor exposure to be 13 m /d, the same

224 as USEtox assumes for outdoor exposure. The average time spent indoors needs to be differentiated

225 between time spent at work and time spent at home (which could even be further distinguished between

226 private and public buildings such as shops, restaurants, etc.), where exposure conditions can be very

227 different. As we are focusing here on household exposure, we assume a daily average of 14 hours spent

228 at home. These can be complemented by 7-8 hours at work, leaving 2-3 hours outdoors. The time

229 fraction spent indoors (at home) is then calculated as ft = 14h/24h = 0.58.

230 Although, these parameters have a strong regional dependency based on cultural and climatic

231 variability , it was not possible to consider this due to very limited data availability and a strong bias

232 towards OECD country-data where data are available. The European Expolis study for example,

233 measured between 18 to 23 hours spent indoors (total) and a range of 0.06 to 5 hours spent outdoors

234 (total) for the adult population (25-55 y) in the seven participating urban areas. The Expolis time-use

235 dataset is the largest multinational European time-use data set, which has been gathered specifically for

236 exposure assessment purposes. Time activity data were gathered from 808 persons in seven European

237 cities: Athens, Basel, Grenoble, Helsinki, Milan, Oxford, and Prague. For North America, the U.S.

238 National Human Activity Pattern Survey (NHAPS) showed that the mean percentage of time spent

239 indoors was 21 hours, with 14 hours of this time spent in a residence and 4 hours of the time spent in

240 other indoor locations. Similar time-patterns were also observed in the Canadian Human Activity

241 Pattern Survey (CHAPS), with some seasonal variations from the U.S. pattern.34 Smith reports that even

242 in developing countries, people spend 70% or more of the day indoors.

243 Sensitivity and variability analysis

For those chemicals with an indoor iF dominated by removal via ventilation rather than by degradation or adsorption, a parameter sensitivity and variability analysis was performed, in order to determine their contribution to variance. Since the ranges of these parameters (Table S1, SI) represent variability (between countries, building types, or individual persons) rather than uncertainty, the analysis only quantifies some of the overall variance, essentially being a variability analysis. The following parameters used to calculate indoor iF were included in the variability analysis using Monte Carlo simulation with 50,000 iterations and Latin Hypercube Sampling (Crystal Ball 11.1.2): 1) building volume V; 2) number of people in the building N; 3) air exchange rate kex; 4) individual daily inhalation rate (at home) IR, 5) daily time at home thome (used to calculate the daily time fraction spent indoors ft). For the values of V and N the sampling method has been adapted to reflect the dependency between these parameters: for each Monte Carlo run, a corresponding set of values for N and V for one country was selected out of their discrete distribution over all countries, with a probability-weighting based on its population. The average individual inhalation rate at rest for households was sampled from the reported interval of 0.44-1.04 m3/h 36 assuming a beta distribution between these limits. The air exchange rate (kex) was sampled from a discrete distribution representing L-AER and H-AER buildings respectively from various countries using a probability-weighting based on their respective population. The daily time at home was assumed to be normally distributed with an assumed standard deviation of 2, resulting in a 95% confidence interval ranging from 10 to 18 hours per day at home. For further details and values the reader is referred to SI. Case Study

To illustrate the application of the method developed, an LCA of cooking in non-OECD countries was performed. This case study was chosen for its relevancy: Air pollution originating from households account for approximately 4% of global health burden and was the leading environmental health risk

factor. The functional unit was defined as the delivery of 1 MJ of useful heat, delivered with stoves based on different fuels: wood, charcoal, liquefied petroleum gas (LPG), and coal. These fuels are the principal fuels being used in non-OECD countries; for example, in India 78% of the population lives in

270 houses where wood or LPG is used as main cooking fuel. Background data for the fuel supply chain of

271 coal, charcoal and LPG were taken from the inventory database ecoinvent. Wood was assumed to be

272 manually collected (no emissions from transport and harvesting), and only land use and the emissions

273 during combustion were accounted for. For the integrated toxicity assessment of indoor and outdoor

274 emissions, the USEtox outdoor model and effect factors (with integrated indoor model) were used

275 according to equation (2), extended to endpoint results expressed as Disability Adjusted Life Years

276 (DALY) using the following disability weights: 11.5 DALY/CTUh for cancerous effects and 2.7

277 DALY/CTUh for non-cancerous effects (CTUh - Comparative Toxic Unit for humans corresponding to

278 cases of cancer or of non-cancer).40 Respiratory inorganics impacts of PM25, NOx, SOx and NH3 were

279 estimated using the effect and characterization factors from Gronlund et al41 The direct emissions are

280 displayed in Table S2 of SI together with further details on the background processes given in section S2

281 of SI.

282 Results

283 Intake fractions and characterization factors

284 With the methodology described and the list of parameters given, intake fractions and characterization

285 factors for indoor exposure in residential settings (i.e. households) can be calculated for the defined

286 regions. For volatile substances, ventilation is the only sink in the indoor environment. Since ventilation

287 is chemical independent, no substance-related parameters are used in these calculations. Therefore, the

288 intake fractions for indoor exposure to volatile substances are the same for all substances and are given

289 in Table 2 for the defined regions. Due to the substance-dependency of the toxicity-effect factor, the

290 characterization factors for these substances vary among chemicals (Equation 2). The substance-specific

291 characterization factors for the USEtox chemical database are given in Excel format as part of SI for 946

292 substances. The characterization factors, in literature sometimes also referred to as comparative toxicity

293 potentials, vary over 12 orders of magnitude from least to most toxic and are up to five orders of

294 magnitude higher for household indoor emissions relative to continental rural emissions for the same

substance (see Figure 2). However, with future updates to the database, the characterization factors will likely change. Therefore, future updates to the latest (indoor and outdoor) characterization factors will be available on the USEtox website (www.usetox.org) and should always be taken from there.

Figure 2: Comparison of characterization factors (CFs) for indoor emissions in non-OECD countries and L-AER buildings (x-axis) relative to CFs for continental urban and rural outdoor emissions (y-axis); the difference between indoor and outdoor iFs is smaller for the other regions

The average house size in non-OECD countries is lower than that in OECD countries, and the average household size is larger (see Table 1). Therefore, intake fractions in L-AER houses in non-OECD countries are about three times higher than those in OECD countries. Intake fractions in H-AER houses in non-OECD countries are a factor of 10 lower because of the higher ventilation rates (Table 1). The results of the variability analysis of household indoor intake fractions are given as standard deviations in Table 2. The variability within the regions is influenced by the amount of data available, which is much

309 lower for non-OECD compared to OECD countries, making those results somewhat less representative

310 for variability between countries.

311 Table 2: Intake fractions (iF) for household indoor exposure with standard deviations (SD) and results of

312 the importance analysis of the parameters used to calculate iF for the defined regions (negative

313 contributions represent an inverse correlation between parameter and result)

Region iF [-] SD IR/h thome N/V kex

Non-OECD countries (H-AER building) 6.8-10"4 8.8-10-4 48% 34% -16% -2%

Non-OECD countries (L-AER building) 1.710-2 1.6-10-2 45% 31% -15% -9%

OECD countries 5.2-10"3 1.7 10-3 41% 29% -21% -9%

Europe (EU-27) 5.7-10"3 3.4-10-3 12% 8% -7% -73%

North America (USA) 4.6-10-3 a a a a a

314 asingle data point (US average) as we are using country averages and hence no variability assessed on sub-country level

316 The results of the importance analysis are given in Table 2. For each region the contribution to total

317 variance per parameter is given, providing an importance ranking of these parameters. Despite some

318 variation in the percentage of contribution the ranking is the same for the OECD and Non-OECD

319 scenarios. Due to the large variability in air-tightness of buildings within Europe, the air exchange rate

320 varies the most and hence contributes the most to total variance of iF in this region with the remaining

321 parameters ranking the same way as for the other regions.

322 For substances with significant indoor degradation (e.g. ozone-sensitive substances) or adsorption to

323 surfaces (e.g. semi-volatile substances), the intake fraction is substance-specific. The intake fractions

324 and characterization factors for these substances can be calculated using the USEtox model version 2.0.

325 The sensitivity study carried out to determine the influence of degradation and surface adsorption

326 delivers the following conclusions: Degradation plays a relatively minor role for the removal of

327 substances emitted into indoor air, by increasing the removal rate by a maximum 20% (Figure S1, SI).

328 The effect of adsorption on room surfaces may be more substantial, since it reduces inhalation intake

329 fraction at high vapor pressure by up to a factor of 60 for substances like benzo[a]pyrene with vapor

330 pressure below 1 Pa (Figure 3, first 4 columns, Figure S2, SI), even for degradation rates on surfaces as

331 low as 1 per thousand of the air degradation (low surface degradation). On the contrary, dermal gaseous

332 exposure uptake increases with the octanol-air partition coefficient Koa and tends to compensate the

333 reduction due to surface adsorption (Figure 3, 4 central columns) for substances with high Koa, leading

ACS Paragon Plus Environment 15

to a total intake with adsorption that is close to the default inhalation intake without adsorption. However, additional information is needed to better characterize surface adsorption and degradation and the way it may compensate the increase in dermal gaseous uptake, hence the choice to only consider indoor air advective removal, degradation, and inhalation pathways in the default model at this stage. More details on the sensitivity study can be found in section S3 of SI.

Figure 3: Variations in indoor intake fractions for the 3073 organic substances in the USEtox 1.01 database considering the inhalation, dermal gaseous and sum of these two exposure pathways with four assumptions: No, low, medium, and high surface degradation rates following sorption, respectively corresponding to surface degradation rates of 0, 0.001, 0.01 and 0.1 of the indoor air OH degradation rate

Case study: world cooking

Figure 4 shows that the health impacts from indoor emissions are dominating the overall health effects. Assuming equal weighting between cancer, non-cancer, and respiratory effects, the respiratory effects from PM emissions represent clearly the most relevant effect for all cooking alternatives analyzed. Total health impacts are more than one order of magnitude lower for cooking with gas compared to charcoal and two orders of magnitude smaller compared to wood and coal.

355 The framework to calculate intake fractions and characterization factors for indoor exposure to

356 substances in households in the USEtox model is described. With this framework and the recommended

357 parameter values given, the iF and characterization factors for household indoor exposure to substances

358 can be calculated for different regions. However, given the uncertainties behind these estimates, the iF

359 for OECD countries, Europe, and North America are essentially equal (Table 2) and we recommend

360 using the OECD value for Europe and North America as well. It should be noted that the distinction

361 between L-AER and H-AER buildings is a strongly simplified, binary classification due to lack of more

362 detailed data. These two classes essentially distinguish between 1) basic constructions ranging from

363 buildings with simple or no sealing and cracked walls to huts or tents without windows and/or doors (H-

364 AER) as opposed to 2) fairly modern buildings eventually with ventilation systems, sealing and

365 insulation (L-AER), which is how they should be used in LCA practice.

366 The observed differences in iF of almost two orders of magnitude between the regions (Table 2) are

367 caused by differences in ventilation rate, building occupation and volume. The dermal absorption of gas-

368 phase chemicals may become important in particular for SVOCs and the calculated intake fractions must

369 be used with care for this class of compounds, as these will require further attention, both for their

370 adsorption and potential degradation rates on surfaces and for dermal uptake.

E-04 E-05 E-06 E-07 E-08 E-09 E-10 E-11 E-12 E-13 E-14

• •

o n o

n U □

■ ■

n ■ ■

▲ ▲

A A i

Open fire (wood)

Charcoal

Coal (China)

• Respiratory inorganics indoor O Respiratory inorganics outdoor ■ carcinogenic effects indoor □ carcinogenic effects outdoor ▲ non carcin. effects indoor A non carcin. effects outdoor

Figure 4: Human health impacts in DALY from indoor and outdoor exposure Discussion

Environmental Science & Technology Page 18 of 23

The USEtox intake fractions for inhalation exposure to outdoor emissions range from 3*10-6 (continental urban air emission) and 7*10-9 (continental rural air emission) respectively for dioxathion (CAS 78-34-2), and up to 3 *10-4 (for continental urban and rural air emission) for 1,1,1,2-tetrafluoroethane (CAS 811-97-2). The intake fractions for indoor air emissions as given in Table 2 are thus at least two and up to seven orders of magnitude higher than the intake fractions for outdoor air emissions.

With the indoor exposure model implemented in USEtox and the resulting characterization factors, it is now operational to integrate household indoor exposure to substances into life cycle assessment studies. Both, iF and characterization factors calculated in this study are based on the still sparse data sources available, which highly influenced the number of regions that could be defined. When more data become available the definitions of regions should be revised in order to better represent global variability, and the iF and characterization factors should be updated. Meanwhile, the parameters in Table 1 for the OECD and non-OECD scenarios are recommended for LCA application of Hellweg et al.'s one-box indoor exposure model. Since the present intake fractions are based on average occupancy and continuous emission, further efforts are needed in the future to better assess emissions with non-continuous sources related to the nexus of occupant and source activity patterns (e.g. cooking), in particular emission patterns that involve near-person releases. Another refinement would be to account for substance removal by filters in centrally air-conditioned buildings, a region-specific removal rate that may be substantial in hot climate. Moreover, whereas degradation was not an important removal process we underline that impacts from the products of homogenous reactions in air or other degradation processes may have significant impacts42-44 but are not taken into account in the CFs calculated by this research work. According to current practice, LCA practitioners can take them into account by adding the amount of reaction products generated from a parent compound to the life cycle emission inventory and characterize them with their corresponding characterization factors.

The case study on cooking in non-OECD countries demonstrates the appropriateness and significance of integrating indoor environments into LCA. Approximately 2.4 billion people, concentrated largely

Page 19 of 23 Environmental Science & Technology

397 within low- and middle-income countries,45 continue to rely on solid fuels as main sources of household

398 energy without access to clean energy or appropriate technologies to prevent exposure to harmful levels

399 of indoor air pollutants from inefficient burning of biomass fuels.46 The results of the case study confirm

400 that health impacts from indoor exposure are relevant. Neglecting these impacts would have provided an

401 incomplete and misleading picture: While cooking with wood would have performed best if only the

402 outdoor emissions were considered (as usually done in LCA), it was the worst alternative after coal if

403 health impacts from indoor exposure were considered. Given the current limits in data availability to

404 parameterize the indoor exposure model for the most affected regions, more robust datasets will likely

405 increase the discrimination of baseline and proposed alternatives. Thus, incorporating the indoor

406 environment in LCA ensures a more holistic consideration of all exposure environments and allows for

407 a better accountability of health impacts. Furthermore, while developing countries transition towards

408 more processed fuels (e.g. petroleum, or electricity from coal), the holistic approach of LCA remains

409 relevant and necessary for assessing both health and environmental implications.

410 Databases providing emission data for different materials, products, and surfaces are an essential

411 element needed towards operationalization of indoor exposure assessment within LCA. Currently,

412 indoor emission data are not widely available or not in a suitable format for LCA (e.g. given as

413 concentrations whereas emitted mass or emission rates would be required to link with our model

414 results).

415 Adapting current tools, such as the USEtox toxicity characterization model, by investigating their

416 applicability under various situations and providing regional specific parameters, allows for identifying

417 "hot-spots" of disease burdens as well as pointers for solutions using a consistent and transparent

418 method. This study, using an illustrative case of cooking, quantified indoor intake fractions for

419 households in various regions of the world that differ geographically, economically, and socially, and

420 provided information on the impact that human behavior, energy use, and technology can have on

421 human health. The modification to the USEtox model, with the integration of the indoor environment, is

422 part of the official update to USEtox version 2.0 and can contribute in providing a clearer assessment of

the source of burden of disease and provide a more informed basis for decision making for all stakeholders.

Supporting information

Parameter values for the individual countries, inventory data for case study on cooking, sensitivity of iF to degradation rates and adsorption on surfaces (PDF), intake fractions, effects factors and characterization factors for household indoor air emissions for three regions calculated with USEtox (Excel). This material is available free of charge via the Internet at http://pubs.acs.org.

Acknowledgements

Most of the work for this project was carried out on a voluntary basis and financed by in-kind contributions from the authors' home institutions which are therefore gratefully acknowledged. The work was performed under the auspices of the UNEP-SETAC Life Cycle Initiative which also provided logistic and financial support. Ralph K. Rosenbaum acknowledges the support from the EU-funded TOX-TRAIN project (project no. 285286, FP7-PEOPLE-IAPP Marie Curie Actions) and the Industrial Chair ELSA-PACT (a research unit of the ELSA research group) with its partners SUEZ, BRL, SCP, UCCOAR-Val d'Orbieu, EVEA, ANR, Irstea, Montpellier SupAgro, Ecole des Mines d'Ales, CIRAD, ONEMA, ADEME, and the Region Languedoc-Roussillon. Evangelia Demou acknowledges financial support by the Medical Research Council (partnership grant MC/PC/13027). The authors are grateful for the participation of Mariano della Chiesa for the calculations of the case study in SimaPro.

References

(1) Hellweg, S.; Mila i Canals, L. Emerging approaches, challenges and opportunities in life cycle assessment. Sci. 2014, 344 , 1109-1113.

(2) Keller, D.; Wahnschaffe, U.; Rosner, G.; Mangelsdorf, I. Considering human toxicity as an impact category in Life Cycle Assessment. Int. J. Life Cycle Assess. 1998, 3, 80-85.

446 (3) Jonsson, A. Is it feasible to address indoor climate issues in LCA? Environ. Impact Assess. Rev.

447 2000, 20, 241-259.

448 (4) Meijer, A.; Huijbregts, M. A. J.; Reijnders, L. Human health damages due to indoor sources of

449 organic compounds and radioactivity in life cycle impact assessment of dwellings. Part 2:

450 Damage scores. Int. J. Life Cycle Assess. 2005, 10, 383-392.

451 (5) Meijer, A.; Huijbregts, M. A. J.; Reijnders, L. Human health damages due to indoor sources of

452 organic compounds and radioactivity in life cycle impact assessment of dwellings - Part 1:

453 Characterisation factors. Int. J. Life Cycle Assess. 2005, 10, 309-316.

454 (6) Hellweg, S.; Demou, E.; Bruzzi, R.; Meijer, A.; Rosenbaum, R. K.; Huijbregts, M. A. J.;

455 McKone, T. E. Integrating Indoor Air Pollutant Exposure within Life Cycle Impact Assessment.

456 Environ. Sci. Technol. 2009, 43, 1670-1679.

457 (7) Kikuchi, Y.; Hirao, M. Local risks and global impacts considering plant-specific functions and

458 constraints: A case study of metal parts cleaning. Int. J. Life Cycle Assess. 2010, 15, 17-31.

459 (8) Skaar, C.; J0rgensen, R. B. Integrating human health impact from indoor emissions into an LCA:

460 A case study evaluating the significance of the use stage. Int. J. Life Cycle Assess. 2013, 18, 636461 646.

462 (9) Collinge, W.; Landis, A. E.; Jones, A. K.; Schaefer, L. A.; Bilec, M. M. Indoor environmental

463 quality in a dynamic life cycle assessment framework for whole buildings: Focus on human

464 health chemical impacts. Build. Environ. 2013, 62, 182-190.

465 (10) Demou, E.; Hellweg, S.; Wilson, M. P.; Hammond, S. K.; Mckone, T. E. Evaluating indoor

466 exposure modeling alternatives for LCA: A case study in the vehicle repair industry. Environ. Sci.

467 Technol. 2009, 43, 5804-5810.

468 (11) Chaudhary, A.; Hellweg, S. Including Indoor Offgassed Emissions in the Life Cycle Inventories

469 of Wood Products. Environ. Sci. Technol. 2014, 48, 14607-14614.

470 (12) Hauschild, M. Z.; Huijbregts, M. A. J.; Jolliet, O.; MacLeod, M.; Margni, M.; Van de Meent, D.;

471 Rosenbaum, R. K.; McKone, T. E. Building a model based on scientific consensus for Life Cycle

472 Impact Assessment of Chemicals: the Search for Harmony and Parsimony. Environ. Sci. Technol.

473 2008, 42, 7032-7037.

474 (13) Rosenbaum, R. K.; Bachmann, T. M. K.; Gold, L. S.; Huijbregts, M. A. J.; Jolliet, O.; Juraske,

475 R.; Koehler, A.; Larsen, H. F.; MacLeod, M.; Margni, M.; et al. USEtox - The UNEP/SETAC-

476 consensus model: recommended characterisation factors for human toxicity and freshwater

477 ecotoxicity in Life Cycle Impact Assessment. Int. J. Life Cycle Assess. 2008, 13, 532-546.

478 (14) EC-JRC. International Reference Life Cycle Data System (ILCD) Handbook - Recommendations

479 for Life Cycle Impact Assessment in the European context; First edit.; European Commission,

480 Joint Research Centre, Institute for Environment and Sustainability: Ispra, Italy, 2011.

481 (15) Bare, J. TRACI 2.0: the tool for the reduction and assessment of chemical and other

482 environmental impacts 2.0. Clean Technol. Environ. Policy 2011, 13, 687-696.

483 (16) Rosenbaum, R. K.; Margni, M.; Jolliet, O. A flexible matrix algebra framework for the

484 multimedia multipathway modeling of emission to impacts. Environ. Int. 2007, 33, 624-634.

485 (17) Wenger, Y.; Li, D. S.; Jolliet, O. Indoor intake fraction considering surface sorption of air organic

486 compounds for life cycle assessment. Int. J. Life Cycle Assess. 2012, 17, 919-931.

UN. World Urbanization Prospects: The 2011 Revision; New York, USA, 2011. US EPA. Estimation Programs Interface EPI Suite Version 4.11, 2012.

Henderson, A.; Hauschild, M. Z.; Van de Meent, D.; Huijbregts, M. A. J.; Larsen, H. F.; Margni, M.; McKone, T. E.; Payet, J.; Rosenbaum, R. K.; Jolliet, O. USEtox fate and ecotoxicity factors for comparative assessment of toxic emissions in life cycle analysis: sensitivity to key chemical properties. Int. J. Life Cycle Assess. 2011, 16, 701-709.

Weschler, C. J. Ozone in Indoor Environments: Concentration and Chemistry. Indoor Air 2000, 10, 269-288.

Nazaroff, W. W. Indoor particle dynamics. Indoor Air 2004, 14, 175-183.

Weschler, C. J.; Nazaroff, W. W. Dermal Uptake of Organic Vapors Commonly Found in Indoor Air. Environ. Sci. Technol. 2013, 48, 1230-1237.

Gong, M.; Zhang, Y.; Weschler, C. J. Predicting dermal absorption of gas-phase chemicals: transient model development, evaluation, and application. Indoor Air 2014, 24, 292-306.

Weschler, C. J.; Nazaroff, W. W. SVOC exposure indoors: fresh look at dermal pathways. Indoor Air 2012, 22, 356-377.

Tibaldi, R.; ten Berge, W.; Drolet, D. Dermal Absorption of Chemicals: Estimation by IH SkinPerm. J. Occup. Environ. Hyg. 2013, 11, 19-31.

Csiszar, S. A.; Ernstoff, A. S.; Fantke, P.; Jolliet, O. Stochastic modeling of near-field exposure to parabens in personal care products. Submitt. Rev.

You, Y.; Niu, C.; Zhou, J.; Liu, Y.; Bai, Z.; Zhang, J.; He, F.; Zhang, N. Measurement of air exchange rates in different indoor environments using continuous CO2 sensors. J. Environ. Sci. 2012, 24, 657-664.

Massey, D.; Kulshrestha, A.; Masih, J.; Taneja, A. Seasonal trends of PM10, PM5.0, PM2.5 &amp; PM1.0 in indoor and outdoor environments of residential homes located in North-Central India. Build. Environ. 2012, 47, 223-231.

Rotko, T.; Oglesby, L.; Kunzli, N.; Jantunen, M. J. Population sampling in European air pollution exposure study, EXPOLIS: Comparisons between the cities and representativeness of the samples. J. Expo. Anal. Environ. Epidemiol. 2000, 10, 355-364.

Hanninen, O. O.; Alm, S.; Katsouyanni, K.; Kunzli, N.; Maroni, M.; Nieuwenhuijsen, M. J.; Saarela, K.; Sram, R. J.; Zmirou, D.; Jantunen, M. J. The EXPOLIS study: Implications for exposure research and environmental policy in Europe. J. Expo. Anal. Environ. Epidemiol. 2004, 14, 440-456.

Schweizer, C.; Edwards, R.; Bayer-Oglesby, L.; Gauderman, W.; Ilacqua, V.; Jantunen, M.; Lai, H.; Nieuwenhuijsen, M.; Kunzli, N. Indoor time-microenvironment-activity patterns in seven regions of Europe. J. Expo. Sci. Environ. Epidemiol. 2007, 17, 170-181.

Klepeis, N. E.; Nelson, W. C.; Ott, W. R.; Robinson, J. P.; Tsang, A. M.; Switzer, P.; Behar, J. V; Hern, S. C.; Engelmann, W. H. The National Human Activity Pattern Survey (NHAPS): a resource for assessing exposure to environmental pollutants. J. Expo. Sci. Environ. Epidemiol. 2001, 11, 231-252.

526 (34) Leech, J. A.; Nelson, W. C.; Burnett, R. T.; Aaron, S.; Raizenne, M. E. It's about time: A

527 comparison of Canadian and American time-activity patterns[dagger]. J. Expo. Sci. Environ.

528 Epidemiol. 2002, 12, 427-432.

529 (35) Smith, K. R. Looking for pollution where the people are. In AsiaPacific issues no. 10; East-West

530 Center: Honolulu, Hawaii, USA, 1994.

531 (36) Nazaroff, W. W. Inhalation intake fraction of pollutants from episodic indoor emissions. Build.

532 Environ. 2008, 43, 269-277.

533 (37) Lim, S. S.; Vos, T.; Flaxman, A. D.; Danaei, G.; Shibuya, K.; Adair-Rohani, H.; Amann, M.;

534 Anderson, H. R.; Andrews, K. G.; Aryee, M.; et al. A comparative risk assessment of burden of

535 disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990-2010:

536 a systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012, 380, 2224537 2260.

538 (38) Census of India. CensusInfo India 2011

539 http://www.devinfolive.info/censusinfodashboard/website/index.php/pages/kitchen_fuelused/Tota

540 l/insidehouse/IND.

541 (39) ecoinvent Centre. ecoinvent data v2.1, 2007.

542 (40) Huijbregts, M. A. J.; Rombouts, L. J. A.; Ragas, A. M. J.; Van de Meent, D. Human-

543 Toxicological Effect and Damage Factors of Carcinogenic and Noncarcinogenic Chemicals for

544 Life Cycle Impact Assessment. Integr. Environ. Assess. Manag. 2005, 1, 181-192.

545 (41) Gronlund, C.; Humbert, S.; Shaked, S.; O'Neill, M.; Jolliet, O. Characterizing the burden of

546 disease of particulate matter for life cycle impact assessment. Air Qual. Atmos. Heal. 2015, 8, 29547 46.

548 (42) Terry, A. C.; Carslaw, N.; Ashmore, M.; Dimitroulopoulou, S.; Carslaw, D. C. Occupant

549 exposure to indoor air pollutants in modern European offices: An integrated modelling approach.

550 A tmos. Environ. 2014, 82, 9-16.

551 (43) Kim, S.; Hong, S.-H.; Bong, C.-K.; Cho, M.-H. Characterization of air freshener emission: the

552 potential health effects. J. Toxicol. Sci. 2015, 40, 535-550.

553 (44) Rohr, A. C. The health significance of gas- and particle-phase terpene oxidation products: a

554 review. Environ. Int. 2013, 60, 145-162.

555 (45) Banerjee, S. G.; Bhatia, M.; Azuela, G. E.; Jaques, I.; Sarkar, A.; Portale, E.; Bushueva, I.;

556 Angelou, N.; Inon, J. G. Global tracking framework: Sustainable energy for all; Washington,

557 DC, USA, 2013.

558 (46) Wilkinson, P.; Smith, K. R.; Joffe, M.; Haines, A. A global perspective on energy: health effects

559 and injustices. Lancet 2007, 370, 965-978.