Comparative Human Toxicity Impact of Electricity Produced from Shale

Oct 10, 2017 - The human toxicity impact (HTI) of electricity produced from shale gas is lower than the HTI of electricity produced from coal, with 90...
1 downloads 11 Views 793KB Size
Subscriber access provided by LAURENTIAN UNIV

Article

Comparative Human Toxicity Impact of Electricity Produced from Shale Gas and Coal Lu Chen, Shelie A. Miller, and Brian R. Ellis Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.7b03546 • Publication Date (Web): 10 Oct 2017 Downloaded from http://pubs.acs.org on October 12, 2017

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.

Page 1 of 25

Environmental Science & Technology

Shale Gas

Coal

ACS Paragon Plus Environment

Environmental Science & Technology

Page 2 of 25

Comparative Human Toxicity Impact of Electricity Produced from Shale Gas and Coal

1 2 3

Authors: Lu Chen1,2, Shelie A. Miller1,2*, Brian R. Ellis2

4 5 6 7 8 9 10 11 12 13 14 15 16

Affiliations: 1

Center for Sustainable Systems, School for Environment and Sustainability, University of Michigan, 440 Church Street, Ann Arbor MI 48109 2

Department of Civil and Environmental Engineering, University of Michigan, 2350 Hayward Street, Ann Arbor MI 48109 *Corresponding author: Shelie A. Miller, 440 Church Street, Ann Arbor MI 48109; 734-7638645; [email protected] Keywords: hydraulic fracturing, risk assessment, life cycle assessment

17

The human toxicity impact (HTI) of electricity produced from shale gas is lower than the HTI of

18

electricity produced from coal, with 90% confidence using a Monte Carlo Analysis.

19

different impact assessment methods estimate the HTI of shale gas electricity to be one-to-two

20

orders of magnitude less than the HTI of coal electricity (0.016-0.024 DALY/GWh versus 0.69-

21

1.7 DALY/GWh). Further, an implausible shale gas scenario where all fracturing fluid and

22

untreated produced water is discharged directly to surface water throughout the lifetime of a well

23

also has a lower HTI than coal electricity. Particulate matter dominates the HTI for both

24

systems, representing a much larger contribution to the overall toxicity burden than VOCs or any

25

aquatic emission. Aquatic emissions can become larger contributors to the HTI when waste

26

products are inadequately disposed or there are significant infrastructure or equipment failures.

27

Large uncertainty and lack of exposure data prevent a full risk assessment; however, the results

28

of this analysis provide a comparison of relative toxicity, which can be used to identify target

29

areas for improvement and assess potential tradeoffs with other environmental impacts.

30 31

ACS Paragon Plus Environment

Two

Page 3 of 25

Environmental Science & Technology

32 33 34

Introduction

35

operations. Recent studies quantify water quality issues associated with shale gas,1-3 although the

36

magnitude and frequency of contamination events are still not well documented. While concerns

37

of water contamination and VOC emissions associated with shale gas production may be

38

warranted, it is also important to contextualize the potential human toxicity impacts of shale gas

39

relative to other sources of electricity.

Potential contamination of drinking water is a major concern surrounding hydraulic fracturing

40

A growing number of studies investigate the potential environmental impacts related to

41

hydraulic fracturing. A recent analysis suggests that spills of fracturing fluid could impact soil

42

health due to the ionic strength of flowback water, although the study found that human health

43

concerns were minimal.4 A variety of studies have highlighted concerns regarding major water

44

withdrawals and contamination events due to stray gas leakage, accidental spills, and wastewater

45

disposal.3,

46

fracturing, although some localized analyses are being conducted.8 VOC signatures from winter

47

haze events have been linked to the Bakken,9 although it is difficult to differentiate between

48

activities associated with hydraulic fracturing and other oil and gas extraction activities. Other

49

analyses suggest the contribution of VOCs to changes in air quality is minimal,8, 10 indicating the

50

need for further investigations into VOC emissions associated with hydraulic fracturing.

51

Additional studies have begun to assess the toxicity of components of fracturing fluids and

52

flowback water.11

53

on a variety of environmental concerns, including greenhouse gas emissions, water use, and

54

criteria air emissions;1,

55

analysis comparing electricity produced from shale gas and coal.

5-7

Data are scarce regarding changes to air quality associated with hydraulic

Life cycle assessments of electricity produced from shale gas have focused 12-16

however, there does not appear to be a comprehensive toxicity

56

The relative human toxicity impact (HTI) of electricity produced from shale gas and coal

57

are compared to better understand how increased penetration of shale gas will affect overall toxic

58

releases associated with the power sector. HTI is commonly used within life cycle assessment to

59

quantify the inherent toxicity burden associated with emissions from a product or process.17 The

60

HTI is measured as disability-adjusted life years per unit of electricity generation (DALY/GWh)

61

and is calculated by means of generic fate and exposure assumptions applied consistently across

62

all systems. The HTI serves as a useful screening metric and an initial step toward a full risk

2

ACS Paragon Plus Environment

Environmental Science & Technology

63

assessment. Characterization factors (CFs), sometimes referred to as human toxicity potentials

64

(HTP), estimate human health damage for a wide range of chemicals and are expressed as

65

DALY/kg emission. The results of life cycle impact assessment are subject to a great deal of

66

variability and uncertainty. Therefore, this analysis uses a variety of approaches to test the

67

robustness of the results. Two scenarios, representing a baseline case and an accidental release

68

scenario, are used to examine different sets of potential assumptions. Both the USEtox 2.0 and

69

ReCiPe 2016 impact assessment methods are used to demonstrate how different sets of CFs may

70

influence the results.18, 19 Local sensitivity analysis in the form of one-at-a-time perturbation is

71

used to assess the degree of influence associated with individual parameters. Finally, Monte

72

Carlo Analysis is used as a global sensitivity method to simulate how both variability and

73

uncertainty across parameter ranges affect the outcome of the analysis.

74

This study conducts a comparative analysis of the toxicological human health effects of

75

electricity produced from shale gas and coal. Human toxicity is a single life cycle impact

76

assessment indicator and is not a comprehensive analysis of environmental factors. A complete

77

life cycle assessment (LCA) should include a broader suite of impacts, including climate change,

78

ecosystem quality, resource depletion, land use, and water use. The evolving nature of research

79

on the environmental impacts of hydraulic fracturing is characterized by data limitations and lack

80

of consensus on individual indicators such as climate change.20 Focused, in-depth analyses on

81

discrete indicators are still needed to be able to compile the necessary information to complete a

82

comprehensive LCA. Therefore, the results of this analysis on human toxicity must be used

83

alongside compatible studies focusing on other indicators in order to understand the overall

84

environmental context.

85 86 87

Methods The scope of this analysis is a human toxicity impact assessment of direct chemical

88

emissions associated with shale gas and coal electricity.

The inventory includes chemical

89

emissions that occur during resource extraction and electricity generation, the stages with the

90

greatest associated toxicity. Baseline case and accidental release scenarios are constructed for

91

both energy systems. The baseline scenario approximates normal operating conditions, while the

92

accidental release scenario simulates major unintended releases of emissions for each system to 3

ACS Paragon Plus Environment

Page 4 of 25

Page 5 of 25

Environmental Science & Technology

93

calculate an upper toxicity limit. For the shale gas system, the study includes aquatic emissions

94

of fracturing fluid chemicals and produced water associated with shale gas extraction, air

95

emissions of NOx, PM, and VOCs during shale gas extraction, and emissions of PM, VOCs, NOx

96

and SOx during electricity generation. To compensate for data scarcity associated with shale gas

97

production and improve confidence in the robustness of the results, the toxicity of the shale gas

98

system is overestimated whenever there is insufficient or missing data. For the coal system, the

99

study estimates the toxicity associated with effluent loadings from coal mine outfalls, air

100

emissions of PM, Hg, VOCs, NOx, and SOx during electricity generation, and aqueous emissions

101

from coal ash impoundments. For the coal accidental release scenario, emissions of metals

102

associated with a coal ash spill are also included.

103

transportation, infrastructure, and cooling water are not included in the analysis, due to minor

104

contributions to overall toxicity.1, 21 Figure 1 depicts the shale gas and coal systems.

Chemical emissions associated with

105

.

106 107 108

a)

b)

109 110 111 112

Figure 1. a) Sketch of shale gas electricity system, including hydraulic fracturing operations and electricity generation; b) Sketch of coal electricity system, including coal mining and electricity generation.

113 114 115

Pennsylvania is used as the point of origin for both shale gas and coal, given the

116

abundance of both sources of energy within the region. When Pennsylvania specific data are not

117

available, alternate datasets are used that can be reasonably assumed to be consistent with coal 4

ACS Paragon Plus Environment

Environmental Science & Technology

118

and shale gas resources originating from the region. National averages are used for efficiency

119

and air emissions associated with natural gas and coal electricity generation, as the resources

120

produced within the region are distributed nationally.22

121 122

Human Toxicity Calculations

123

Chemical emissions that have a direct toxicological effect on human health are included in the

124

analysis. Factors that are non-toxicity related (e.g., noise, light, stress), have an indirect effect on

125

human health (e.g., ozone depletion, climate change), or cause environmental damage not related

126

to human health (e.g., salinity, pH, turbidity) are outside the scope of the analysis. Toxic releases

127

are measured using disability-adjusted life years (DALY) per GWh of electricity produced. A

128

DALY is a common metric to measure impact on human health and corresponds to the number

129

of years lost due to poor health, disability, or premature death. 18, 19

130 131

To calculate HTI of electricity produced from coal and shale gas, the USEtox 2.0 (released in

132

2015) and ReCiPe 2016 characterization factors are used.18, 19 USEtox and ReCiPe are impact

133

assessment methods that estimate the toxicity of a given quantity of emissions by providing CFs

134

that estimate human health damage for a wide range of chemicals. Human toxicity CFs are

135

averages for North America. Both USEtox and ReCiPe are long-term exposure assessment

136

methods and consider the impact on the general population. The time horizon is 100 years for

137

USEtox and infinite for ReCiPe’s Egalitarian assumptions. Human health impacts due to acute

138

workplace exposures are not taken into account. USEtox 2.0 is a scientific consensus model

139

recommended as the preferred database for calculating HTI by the United Nations Environment

140

Programme and the Society of Environmental Toxicology and Chemistry’s Life Cycle Initiative.

141

ReCiPe was developed by a consortium of life cycle assessment practitioners. The two methods

142

were developed independently and both provide endpoint level CFs for carcinogenic and non-

143

carcinogenic toxicity. Life cycle impact assessment is an accepted method to quantify the

144

relative hazard and importance of pollutants when data for a full-scale risk assessment are not

145

available.23

146 147

CFs from both models are reported due to the large degree of uncertainty in developing

148

appropriate CFs and differences in CFs for key constituents of this analysis. The sensitivity 5

ACS Paragon Plus Environment

Page 6 of 25

Page 7 of 25

Environmental Science & Technology

149

analysis addresses concerns regarding CF uncertainty in greater detail. Eqn. 1 is the generic form

150

for calculating HTI for an individual emission.

151 152

 =  × 

Eqn. 1.

153 154

Where i represents each unique chemical,  is the human toxicity impact for each emission

155

(DALY/GWh),  is the characterization factor for each chemical (DALY/kg emission),  is

156

the emissions of each chemical associated with a given amount of electricity generation

157

(kg/GWh). Total HTI ( ) for life cycle electricity generation is the sum of the individual

158

chemical impacts (Eqn. 2).

159 160

 = ∑ 

Eqn 2.

161 162

The HTI of PM reported in this analysis is the aggregate human health damages resulting

163

from both PM2.5 and PM10 formed from both primary (directly emitted to the atmosphere) and

164

secondary (generated by chemical reactions associated with emissions such as NOx and SOx)

165

sources.

166

PM has a different mechanism for impacting human health than other chemicals in the

167

inventory and the CF for PM is derived via a different approach. The ReCiPe model contains a

168

CF for PM2.5, as well as for NH3, NOx, and SO2, which contribute to secondary PM aerosol

169

formation. USEtox does not include a CF for PM, so the USEtox method is supplemented with a

170

method described by Gronlund et al.

171

and thirteen coal-fired power plants within Pennsylvania are used to calculate the HTI of PM,

172

using production-weighted averages of electricity generation for each plant.25 The Supporting

173

Information contains relevant assumptions and calculations for the HTI of PM.24, 25

24

Emission data collected from twenty-three natural gas

174 175

Shale Gas

176 177

In an effort to be conservative due to uncertainty and lack of data, assumptions are made to

178

overestimate the potential toxic emissions for the shale gas baseline whenever limited data

179

inhibit the analysis. A number of pathways exist for chemicals within fracturing fluid and 6

ACS Paragon Plus Environment

Environmental Science & Technology

Page 8 of 25

180

produced water to enter the environment. Potential emission pathways include storage failures of

181

fracturing fluids and produced water at the surface, produced water migration through the

182

subsurface into groundwater, contamination from faulty casings, and as effluent from wastewater

183

treatment plants.26

184

To estimate potential releases of fracturing fluid chemicals, baseline and accidental

185

release scenarios are constructed. Data from 2990 wells from FracFocus within Pennsylvania are

186

used, containing a total of 368 chemicals. FracFocus 3.0 is a machine-readable database that

187

includes self-reported chemicals used in fracturing fluid formulations from over 62,000 wells

188

throughout the United States.27 The ID number, total fracturing fluids volume, purpose of each

189

chemical (corrosion inhibitor, proppant, etc.), and the concentration of each component were

190

recorded for each well. To account for differences in chemical compositions due to geographic

191

and operational variations, average well information for each Pennsylvania County represented

192

in FracFocus as of October 2015 was used. Not all wells use the same chemicals, so the average

193

concentration was used for the 100 most frequently used chemicals from 2010 to 2015 (Table

194

S1). Most of the fracturing formulations surveyed within FracFocus include less than 40

195

chemicals, so using the average concentration of the 100 most frequently used chemicals

196

overestimates the HTI for shale gas.

197 198

Table 1. Assumptions used for baseline and accidental release shale gas scenarios and parameter ranges found

199

within the literature for comparison.

200

Information.

Calculations and further justifications are found within the Supporting

201 Parameter

Unit

Average fracturing fluid volume

m3/well

Flowback water percentage share of injected fluid Produced water percentage share of injected fluid Shale gas electricity generation efficiency Shale gas well lifetime span Shale gas production volume Barium removal efficiency Fracturing fluid release rate

Baseline case 13000

Accidental release 13000

Literature Range 3500-26000

Reference 27

%

15

15

10-15

%

5

5

1-7

Mcf / kWh year MMcf %

0.0101

0.0101

28-30

15 4300 90

15 4300 0

0.009640.0134 5-30 1600-9000 0 – 99

% released

1

100

See details in text

3

7

ACS Paragon Plus Environment

1, 26

1

31 32, 33 34

Page 9 of 25

Environmental Science & Technology

202 203 204

Currently, no database exists that systematically records spills and other releases.35 The baseline

205

case is conservative to compensate for data gaps in toxicity and chemical composition. The

206

baseline case assumes that 1% of fracturing fluids are emitted to surface water without any

207

treatment, although the discharge could occur via any of the potential emissions release

208

mechanisms. A recent study estimated the total known spill volume of fracturing fluid from 6000

209

wells to be 174 m3 from 2008 to 2013,35 averaging 0.03 m3 per well during that period. The

210

assessment report from the EPA lists the median spill volume as 1.6 m3 per well that reports a

211

spill (range 0.19-72 m3), with 0.4-12.2 spills per 100 wells, which is similar to data reported

212

elsewhere.36, 37, although it does not necessarily account for spills that go unreported. Of the

213

spills that are reported, the EPA estimates that only 9% of spills reach surface water. Assuming

214

13000 m3 of fracturing fluid for each well (Table 1),27 a more reasonable spill rate is 0.001% of

215

injected fluids, as opposed to the 1% assumed in the baseline scenario.

216

In addition to fracturing fluids injected into the wells, produced water has the potential to

217

contaminate drinking water sources. The chemical composition of flowback and produced water

218

are different, with larger concentrations of naturally dissolved materials in produced water than

219

flowback water. Some of the chemicals in produced water, especially organic compounds,

220

originate from the injected fracturing fluids. It is difficult to differentiate naturally-occurring

221

organic compounds from those that derive from fracturing fluid formulations. Because of the

222

difficulty distinguishing between the sources of chemicals, all chemicals within produced water

223

are treated as a separate source of emissions, effectively double-counting the fracturing fluid

224

chemicals contained in produced water. The volume of flowback and produced water is assumed

225

to be 20% of the injected water volume and have the average chemical composition of a

226

compilation of 35,000 data entries for the Marcellus shale gas region reported by Abualfaraj et

227

al.38 (Table S2).

228

Produced water from hydraulic fracturing operations is assumed to be sent to a dedicated

229

water treatment plant, which is the most common method of disposal in Pennsylvania.39

230

Alternative methods of disposal include injection into Class II disposal wells and recycling for

231

additional shale gas extraction operations.

232

Chemical constituents of produced water can enter surface water via the same

233

mechanisms as fracturing fluids, and also as effluent from wastewater treatment due to 8

ACS Paragon Plus Environment

Environmental Science & Technology

234

incomplete removal. For the baseline shale gas case, 10% of the chemicals within produced

235

water are assumed to be discharged to surface water. For context, a recent study estimated the

236

total known spill volume of produced water from 6000 wells to be 473 m3 from 2008 to 2013,35

237

averaging 0.079m3 per well. The assessment report from the EPA lists the median spill volume

238

of produced water as 37.5 m3 per well that reports a spill, with 0.4-12.2 spills per 100 wells.

239

Assuming 2700 m3 of produced water per well (Table 1), then the spill rate of produced water is

240

likely less than 1%. In addition, some fraction of the constituents of produced water are

241

discharged as effluent from the wastewater treatment facility, so the overall estimate of 10%

242

discharge is assumed to be reasonable and likely an overestimate for the baseline case.

243

The accidental release scenario for shale gas assumes 100% of injected fracturing fluid

244

and 100% of produced water is directly discharged to surface water over the entire lifetime of the

245

well. While this scenario is highly improbable, it provides a maximum threshold for possible

246

toxicity loads associated with shale gas.

Page 10 of 25

247

For air emissions from shale gas fields, values are obtained from a comprehensive study

248

by Roy et. al, which quantifies total load of NOx, PM2.5, and VOC emissions per well associated

249

with hydraulic fracturing operations.40 Their study estimates total load, but does not provide

250

compositional analysis of VOCs. In the absence of compositional data from this study, the CF

251

for benzene was used for all VOCs in this analysis. VOC composition from oil and gas sites tend

252

to have significant proportions of alkanes and alkenes that have lower toxicity than aromatics.

253

Therefore, the use of the benzene CF is considered an overestimation of VOC contribution to

254

HTI.

255 256 257

Coal

258

An EPA study on coal mine discharge41 was used in conjunction with annual

259

Pennsylvania coal production estimates42 to estimate toxicity associated with coal mining.41 The

260

study calculated loading of aluminum, iron, manganese, and total suspended solids in

261

Pennsylvania streams impacted by acid mine drainage.

262

To estimate human toxicity of mercury air emissions from coal-fired power plants, data

263

was collected from the 100 largest power producers in the United Sates in 2013.43 Combustion

264

residuals, also known as coal ash or fly ash, are assumed to be stored in surface impoundments. 9

ACS Paragon Plus Environment

Page 11 of 25

Environmental Science & Technology

265

Some of the chemicals within the coal ash leach out of the disposal site and are released to the

266

environment (Table S3).44 The coal ash effluent values are assumed to be representative of coal

267

originating in Pennsylvania, given the locations of the power plants where data were obtained.

268

An EPA study on power plant effluent, which sampled coal-fired power plants in the

269

Appalachian region45, reported similar values. It is assumed that, in addition to ash, the coal ash

270

impoundment also receives discharges of flue gas desulfurization waste water and transport

271

water for fly and bottom ash. These waste streams appear to be included in reported outflow

272

values and are therefore not calculated as separate emissions.

273

Electricity generation was calculated using nameplate capacities of each power plant,

274

assuming continual operation of all plants, 365 days/year, which overestimates the generation

275

capacity of each plant, underestimating the HTI associated with electricity production. Wet

276

disposal of coal ash is not the only possible method of disposal. Coal ash can also be reused in

277

engineering applications or landfilled in a dry state. This choice is further addressed in the

278

sensitivity analysis and discussion.

279

The accidental release scenario for coal uses the same assumptions for all parameters of

280

the baseline case, and also includes unintentional release of coal ash into surface water, similar to

281

the impoundment failures that occurred in Kingston, TN in 2008 and Eden, NC in 2014.

282

Although major releases of coal ash spills are infrequent, they have historical precedent and their

283

inclusion provides an analogous system to compare with the accidental release scenario for shale

284

gas.

285

For the coal accidental release scenario, post-remediation data from the Kingston coal ash

286

spill were used.46 Remediation efforts took place in 2009 and 2010 leaving 0.18 million m3 of

287

ash in the Emory River. The density of ash is assumed to be 1500 kg/m3 and the HTI is

288

calculated based on the volume of the ash left in the Emory River post-remediation. Babbit et al.

289

calculated the average coal ash generation from four power plants in Florida as 0.0568 kg/kWh.21

290

The National Renewable Energy Laboratory estimates the average coal ash generation rate as 0.1

291

kg/kWh

292

more conservative HTI estimate.

47

. The smaller coal ash generation value is used in this analysis in order to provide a

293 294

Data Limitations

10

ACS Paragon Plus Environment

Environmental Science & Technology

295

The scope of this study is intended to address the full life cycles of shale gas and coal

296

electricity; however, data limitations necessitated a number of choices to be made regarding the

297

life cycle inventory of both shale gas and coal. Because of these limitations, systematic choices

298

were made to overestimate the coal inventory while underestimating the shale gas inventory.

299

Most studies that report VOCs or non-methane hydrocarbons report total load but do not

300

provide compositional analysis. Benzene is used as the CF for all VOCs in this analysis, which

301

likely overestimates VOC contributions to total HTI for all scenarios.

302

Page 12 of 25

Data relevant to calculating the HTI of coal mining is rarely reported in a format that is

303

conducive to translation into inventory data.

Mining emissions tend to be reported as

304

concentrations, without sufficient data to relate concentrations to mass emission/mass coal

305

extracted needed for the inventory (e.g. stream flow, time since coal extraction, overall amount

306

of coal extracted from site). Although other studies have found that coal mining may have

307

significant human health impacts on nearby communities,48-50 lack of appropriate data inhibit

308

inclusion of several aspects of the coal system into this analysis. Potential toxicity impacts are

309

associated with cadmium and selenium in mine drainage; however, limited data prevented

310

quantification of loading data from either element.41 Similarly, calculation of HTI associated

311

with coal mining does not include particulate matter or other air emissions associated with

312

mining that are analogous for the data obtained for the shale gas baseline case. Fugitive coal

313

dust emissions along transportation routes has been linked to chronic community-level exposures

314

of particulate matter and metals,51 but the format of data available inhibits translation into

315

inventory metrics useful for this analysis.

316

In addition to limitations associated with life cycle inventory collection, ReCiPe and

317

USEtox do not contain a uniform list of chemicals, so a given chemical may be associated with a

318

CF in one database and not in another. Characterization factors are also highly uncertain and a

319

standard operating assumption is that a CF may vary by three orders of magnitude in either

320

direction. 18, 19

321

Different oxidation states of metals have different toxicities; however, oxidation states

322

are not available from the datasets used in this analysis. To be consistent with the systematic

323

overestimation of the shale gas HTI, the more toxic oxidation state is assigned for shale gas

324

emissions whenever available. Similarly, mercury in the coal system is assumed to be present in

325

its inorganic form, even though it has the potential to form organic mercury, such as 11

ACS Paragon Plus Environment

Page 13 of 25

Environmental Science & Technology

326

methylmercury and dimethylmercury, which are more toxic than inorganic mercury.52 CFs for

327

radionuclides are included in ReCiPe but not in USEtox. The effects on results of these issues

328

are further discussed in the Results and Discussion.

329 330

Uncertainty and Sensitivity Analysis

331

A Monte Carlo Analysis (MCA) is used to determine the range of possible results and the

332

extent of overlap between the shale gas and coal cases. Triangular distributions are assigned to

333

each of the parameters within the model, using the most likely value and upper and lower range

334

boundaries (see Tables 1, S10-S12). For the CFs, which span six orders of magnitude, triangular

335

distributions are assigned to the exponent of each parameter to reduce positive bias in sampling.

336

The results of 10,000 trials are reported.

337

A one-at-a-time perturbation method is used to assess the sensitivity of the results to each

338

input parameter. The effect of each parameter on the HTI is determined by changing its value to

339

the extreme ends of its range while keeping all other parameters at their initial model value (see

340

Tables 1, S10-S12). CFs from USEtox are used as the default for both the sensitivity analysis and

341

the MCA, unless a CF was available only in ReCiPe.

342

In addition to the sensitivity analysis, two alternate operating scenarios are explored for

343

coal. The scenarios are disposal of coal ash via a dry storage method and implementation of the

344

Mercury Air Toxics Standards (MATS) regulation.53

345

12

ACS Paragon Plus Environment

Environmental Science & Technology

346 347 348 349

Page 14 of 25

Results and Discussion

PM

VOC

As

Others

Ba

DMF

Hg (air)

Hg (coal ash)

100

HTI (DALY/GWH)

10

1

0.1

0.01 USEtox

ReCiPe

USEtox

Shale Gas

ReCiPe

Coal

USEtox

ReCiPe

Shale Gas

Baseline Scenario

USEtox

ReCiPe

Coal

Accidental Release Scenario

350 351 352 353 354 355 356

Figure 2. Baseline and accidental release scenarios for electricity produced from shale gas and coal using the USEtox and ReCiPe2016 methods to quantify HTI. Data is depicted on a log scale. Major components of HTI include particulate matter (PM), volatile organic compounds (VOC), arsenic (As), barium (Ba), N, N’-dimethylformamide (DMF), and mercury (Hg). Constituents that appear in both systems are depicted in gray scale, whereas those only found in coal system are in orange and those only found in shale gas system are in blue.

357

Figure 2 shows that the baseline scenario for coal-fired electricity has a greater HTI than

358

both the baseline and accidental release scenarios for shale gas, by at least an order of magnitude.

359

The HTI baseline values are 0.016 DALY/GWh (USEtox) and 0.024 DALY/GWh (ReCiPe) for

360

shale gas and 0.69 DALY/GWh (USEtox) and 1.7 DALY/GWh (ReCiPe) for coal. A prior study

361

on coal toxicity using CML2001, another popular impact assessment method, estimated an HTI

362

of coal electricity to be between 0.24 DALY/GWh to 2.2 DALY/GWh,54 which aligns with the

363

results of this analysis.

364

Particulate matter is the dominant toxicity contributor for both shale gas (86% USEtox,

365

93% ReCiPe) and coal (92% USEtox, 98% ReCiPe), and includes calculation of primary

366

emissions of particulate as well as secondary aerosol formation (see Tables S6-S9). Other

367

contributors to the HTI of the coal baseline are air emissions of mercury (5.6% USEtox,