An integrated hazard screening and indexing system for hydraulic fracturing chemical assessment

Process Safety and Environmental Protection 130 (2019) 126–139

Contents lists available at ScienceDirect

Process Safety and Environmental Protection journal homepage: www.elsevier.com/locate/psep

An integrated hazard screening and indexing system for hydraulic fracturing chemical assessment Guangji Hu, Haroon R. Mian, Kasun Hewage, Rehan Sadiq ∗ School of Engineering, The University of British Columbia, Okanagan Campus, Kelowna, BC, V1V 1V7 Canada

a r t i c l e

i n f o

Article history: Received 2 January 2019 Received in revised form 12 June 2019 Accepted 2 August 2019 Available online 11 August 2019 Keywords: Hydraulic fracturing Environmental and human health Oil and gas chemical Hazard screening Hazard indexing Chemical hazard assessment

a b s t r a c t Various chemicals used in hydraulic fracturing have raised environmental and human health (EHH) concerns regarding water resources contamination, leading to the transition towards the use of chemicals with minimum EHH hazards. Chemical hazard screening and indexing approaches have been used to measure the chemical hazard of hydraulic fracturing, and each approach is associated with inherent advantages and limitations. In this study, the two chemical hazard assessment approaches were discussed, and an integrated chemical hazard screening and indexing system was developed to combine the strengths of the two approaches. The integrated system was applied to assess the EHH hazards of representative hydraulic fracturing chemicals used in British Columbia, Canada. The hazard screening results showed that more than half of the ingredients and additives were classified into high hazard groups. Moreover, the integrated system generated more critical hazard assessment results than two hazard indexing systems, revealing that using the individual hazard indexing approach could result in underestimated EHH hazards for chemicals. The integrated system can significantly improve the data confidence levels of hazard assessment results compared to a previously developed indexing system. The integrated system can also help formulate fracturing fluids with low EHH hazards by identifying ingredients of high hazard concerns. © 2019 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

1. Introduction Over the past decades, the unconventional natural gas production has been rapidly increasing in Canada (NEB, 2017). The rapid growth of the unconventional gas industry is mainly attributed to the combined use of horizontal drilling and hydraulic fracturing, which allows for the economic extraction of natural gas from low-permeability formations such as gas-bearing shales and sandstones (Gallegos and Varela, 2014). Despite the promising resource potential and economic benefits, the rapid expansion of the unconventional gas industry has triggered considerable public debate on possible environmental and human health (EHH) impacts posed by hydraulic fracturing (Boudet et al., 2014; Soeder et al., 2014; Vengosh et al., 2014). The chemicals used in hydraulic fracturing are of particular concern due to the potential contamination of ground and surface water supplies and the associated health risks

∗ Corresponding author at: School of Engineering, The University of British Columbia, Okanagan 3333 University Way, Kelowna, British Columbia, V1V 1V7, Canada. E-mail address: [email protected] (R. Sadiq).

to aquatic ecosystems and water resource users (Akob et al., 2016; Kahrilas et al., 2014; Orem et al., 2017; Renock et al., 2016). In hydraulic fracturing, various chemical additives are used to improve fracturing performance and gas recovery. An additive typically consists of several ingredients at different concentrations. According to the downhole functions, additives can be divided into different categories such as gelling agents, friction reducers, and crosslinkers (Hu et al., 2018a; Kahrilas et al., 2016; Stringfellow et al., 2014). Different additives are mixed with water and proppants (commonly quartz sands) to formulate a fracturing fluid, which can be pumped into underground under high pressures to initiate fractures in the low-permeability formations (FracFocus, 2014). A fracturing fluid may contain three to twelve additives, depending on geological characteristics of the target formations and requirements of the operators (Soeder et al., 2014). Additives may only account for a small fraction (e.g., < 2%) of fracturing fluid; however, the use of millions of gallons of fracturing fluid for a single hydraulic fracturing operation still involves a substantial amount of chemicals (All Consulting, 2012; Engle et al., 2014; Soeder et al., 2014). More critically, some of the ingredients have been identified as carcinogens, mutagens, and substances with acute and chronic toxic effects on human health and aquatic ecosystems (Cozzarelli et al., 2017; Hu et al., 2018a,2018b; Stringfellow et al., 2014),

https://doi.org/10.1016/j.psep.2019.08.002 0957-5820/© 2019 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

G. Hu et al. / Process Safety and Environmental Protection 130 (2019) 126–139

List of acronyms acute human oral toxicity AhT AHP analytical hierarchy process AT aquatic toxicity (acute/chronic) B bioaccumulation potential carcinogenicity C CASRN chemical abstracts service registry number ChT chronic human oral toxicity chemical toxicological data CTD DCIE/H/A/F data confidence index for ingredientenvironmental health/ingredient-human health/additive/fracturing fluid DCS data confidence score endocrine disruptor E EHH environmental and human health GHS globally harmonized system of classification and labelling of chemicals HGI/A/F hazard group for ingredient/additive/fracturing fluid hazard index for ingredient/additive/fracturing HII/A/F fluid HS hazard score HyFFGAS hydraulic fracturing fluid greenness assessment system HyFI/A/F hazard assessment results for ingredient/additive/fracturing fluid from HyFFGAS ICHSIS integrated chemical hazard screening and indexing system IES ingredient environmental health hazard score HIS ingredient human health hazard score hazard value for ingrediIHVI/A/F integrated ent/additive/fracturing fluid mutagenicity M P environmental persistence reproductive toxicity R

increasing public concern over the chemical hazard of hydraulic fracturing. Although there are various government regulations, industry codes-of-practices, and company standard operating procedures in place to minimize the likelihood of unintended release of fracturing fluids, the health risk posed by hydraulic fracturing chemicals to surrounding ecosystems and resource users cannot be neglected. The downhole performance and cost are two major criteria determining the selection of hydraulic fracturing chemicals; When the two criteria are met, the use of chemicals with minimum EHH effects should be encouraged by both the regulatory organizations and industries for reasons of responsible production and public confidence (Brannon et al., 2012; CAPP, 2012; Thomas et al., 2019). The transition towards the use of more environmentally responsible chemicals has posed several challenges such as developing frameworks and methodologies which can provide meaningful and reliable chemical hazard assessment results. As a response, various chemical hazard assessment systems have been developed to systematically evaluate EHH hazards and generate outcomes for informed decision making in hydraulic fracturing chemical management. The representative systems include the Quantitative Ranking Measure of Oil Field Chemical Environmental Impacts (Jordan et al., 2010), Chemical Hazard Rating System (Hepburn, 2012), Chemical Scoring Index (Verslycke et al., 2014), Intrinsik Screening-level Assessment System (Intrinsik, 2013), Hydraulic Fracturing Fluid Greenness Assessment System (HyFFGAS) (Hurley et al., 2016), and GreenScreen System (CPA,

127

2016). These chemical hazard assessment systems can generally be divided into two approaches: hazard screening approach and hazard indexing approach, and the two approaches are inherently linked with different advantages and limitations. It is highly desirable to develop an integrated chemical hazard assessment system that combines the strengths of the two approaches for more effective chemical hazard evaluation. In this study, the advantages and limitations of hazard screening and indexing systems were discussed. Based on the discussion, an integrated chemical hazard screening and indexing system (ICHSIS) was developed. The objective defining, parameters selection, data uncertainty analysis, and chemical hazard weighing, aggregation, and classification of ICHSIS were elucidated. The integrated system was used to assess the representative hydraulic fracturing chemicals used in British Columbia, Canada, and the assessment results were compared with those from the previously developed HyFFGAS.

2. Hazard screening and indexing systems Both hydraulic fracturing chemical hazard screening and indexing systems are qualitative methodologies, either use descriptive terms or numerical rating scales to describe chemical hazards (Ferrari et al., 2016). These systems share a common feature that the assessment processes all begin with ingredients as they are the essential components of additives/fracturing fluids. However, the hazard screening and indexing systems use different methods to present ingredients’ hazards and aggregate the hazards to additive/fluid levels. The assessment results from the two approaches are also associated with different hazard implications and data uncertainties.

2.1. Hazard screening systems Chemical hazard screening aims to select an appropriate hazard designation for a given chemical. The hazard designations are assigned based on qualitative hazard description and potency consideration, rather than numerical scales. The Intrinsik Screening-level Assessment System and the GreenScreen System are two representative chemical hazard screening systems (CPA, 2016; Intrinsik, 2013). The two systems operate at screening levels with a focus on the defined series of hazard endpoints (e.g., carcinogenicity, aquatic toxicity) relevant to the EHH hazard profile of a chemical. The chemical toxicological data (CTD) of an ingredient is screened against the selected hazard endpoints to determine whether the concerned hazard exists or not, and the severity of hazard if it exists. The resultant hazard profile is presented in qualitative hazard designations, such as the three hazard categories used in the Intrinsik Screening-level Assessment System and the four hazard benchmarks used in the GreenScreen System (CPA, 2016; Intrinsik, 2013). Different hazard designations represent different severity levels of EHH hazards. The concentrations of ingredients within an additive/fluid are also screened against the cut-off concentrations of hazard endpoints to determine whether the hazard is in-effect or not. Chemical hazard screening systems can generate descriptive hazard designations reflective of chemical hazards, without involving any numerical conversion and aggregation algorithm. Thus, the results are relatively objective and easy for hazard communication. Nonetheless, hazard screening systems are less applicable when comparing two chemicals that have the same hazard designation. Also, the ingredient concentration evaluation mechanism (i.e., the “cut-off” concentration) is binary in hazard screening systems, neglecting the fact that the higher concentrations of hazardous

128

G. Hu et al. / Process Safety and Environmental Protection 130 (2019) 126–139

ingredients contained in an additive/fluid, the higher EHH risks might be posed by the ingredients. 2.2. Hazard indexing systems Various chemical hazard indexing systems have been developed to translate the hazard information of hydraulic fracturing chemicals to a single measure (i.e., index, score) that reflects an aggregated chemical hazard, allowing for comparing the EHH hazards of different chemicals (Hepburn, 2012; Hurley et al., 2016; Jordan et al., 2010; Verslycke et al., 2014). Hazard indexing systems are composed of various hazard endpoints, scoring rules, and score aggregation algorithms. The CTD of ingredients on the selected hazard endpoints and their concentrations in an additive/fluid are transformed to a numerical scale using specific sub-index functions, scoring rule sets, or implicit rating curves (Hurley et al., 2016). The resultant sub-indices or scores are weighted and aggregated to produce a final index. Qualitative hazard descriptions are established based on the scales of the final indices to facilitate hazard interpretation and decision making on chemical selection. When using indexing systems, a few issues can occur as a result of abstracting information and data. Indexing systems are not entirely successful in providing the true picture of the assessed subject due to diverse types of input data and partly because they are insufficient to aggregate diverse data properly (Sadiq et al., 2010). The improper aggregation could generate eclipsed, exaggerated, and ambiguous results (Sadiq et al., 2010; Swamee and Tyagi, 2000). For instance, eclipsing occurs when a chemical being assessed is associated with critical EHH hazards, yet the derived hazard index comes out at a moderate level, failing to show any critical hazard due to improper aggregation. Moreover, weighing sub-indices/scores is a subjective process depending on assessor and system developers’ opinions. Thus, it is possible that different hazard indexing systems generate different assessment results for the same chemical (Hurley et al., 2016). There is a growing need to develop an integrated system that can not only objectively reflect the EHH hazard of a chemical for hazard monitoring but also compare the EHH hazards of different chemicals for informed chemical selection. 3. Integrated screening and indexing system Informed by the advantages of chemical hazard screening and indexing systems, ICHSIS was developed to characterize the EHH hazards of hydraulic fracturing chemicals at ingredient, additive, and fluid levels. In ICHSIS, EHH hazard is defined as the properties and characteristics of a chemical that can cause adverse effects on EHH. The EHH hazards of hydraulic fracturing chemicals are primarily evaluated through water exposures since there is substantial concern regarding the potential for hydraulic fracturing operations to contaminate water sources. 3.1. Framework The framework of ICHSIS is outlined in Fig. 1. The assessment process begins with chemical data acquisition, including identifying the chemical abstracts service registry numbers (CASRN) and concentrations of ingredients contained in an additive/fracturing fluid and searching the CTD for the ingredients. The CTD and concentration of each ingredient are processed through hazard screening and indexing approaches to generate two hazard assessment outcomes coupled with data confidence indicators. An integrated hazard assessment outcome is generated for each ingredient by amalgamating the hazard screening and indexing outcomes. The ingredients’ hazard assessment outcomes are then

aggregated to generate a hazard assessment outcome for the additive/fracturing fluid. Based on the hazard assessment outcomes, informed decisions on chemical use can be made for EHH hazard mitigation. 3.2. Hazard endpoints and criteria The hazard endpoints and criteria inclusive in ICHSIS are shown in Table 1. The hazard endpoint denotes the type of adverse effects on EHH, such as carcinogenicity, mutagenicity, and aquatic toxicity (Exon, 2006). The hazard criteria indicate the severity of a specific type of adverse effects, such as the Category 1–4 acute/chronic aquatic toxicity for evaluating the severity of aquatic toxicity. The definitions of the selected hazard endpoints and criteria can be referred to the Globally Harmonized System of Classification and Labelling of Chemicals (GHS) (UN, 2013). In addition to the conventional hazard endpoints defined in the GHS, an emerging hazard endpoint-endocrine disruptor (E), was included in ICHSIS because there is a growing concern about the extent of risk posed by endocrine disruptors to human and wildlife health (European Commission, 2018; Kassotis et al., 2017). The EHH hazard is assessed in the context of unintended releases of hydraulic fracturing chemicals caused by spill accidents, equipment failure, or loss of well-bore integrity. Thus, physical hazards (e.g., flammability, explosiveness, corrosiveness) and health hazards due to dermal and inhalation exposure are not assessed as those hazards are more relevant to workplace safety. However, the hazard endpoints and exposure routes considered in ICHSIS can be modified to suit different assessors’ need. As shown in Table 1, each hazard endpoint was assigned a cut-off concentration for determining whether the concentration of an ingredient within a mixture is high enough to trigger the concerned hazard endpoint. The cut-off concentrations are consistent with those used in other chemical hazard classification systems, including the Health Canada’s Workplace Hazardous Materials Information System and the GHS (Health Canada, 2015; UN, 2013). Also, hazard scores (HS) ranging from 0 to 10 were assigned to the hazard criteria under each hazard endpoint to differentiate and scale the hazard criteria. A higher HS indicates a higher level of hazard for the respective endpoint. 3.3. Chemical toxicological data Hazard screening and indexing rely on the CTD of ingredients. The quality and availability of CTD greatly affect the data confidence level of hazard assessment results. A variety of CTD sources, including peer-reviewed chemical toxicity databases, material safety data sheets, suitable ingredient analogs, and chemical toxicity model simulations, are used in ICHSIS to maximize the availability of CTD (Appendix I). As Fig. 2 shows, different data sources were divided into four tiers based on the data confidence implication (CPA, 2016). A data confidence score (DCS) was assigned to each tier of data sources for indexing purpose. DCSs range from 0 to 10, where a higher value indicates higher data certainty of the data source. A CTD searching rule was established to ensure that assessment results can be generated with the highest possible data confidence. As Fig. 2 shows, tier 1 data sources are searched first after identifying the CASRN for an ingredient. If CTD cannot be found in tier 1 data sources, then tier 2 data sources will be searched. Similarly, tier 3 analog ingredient will be used if CTD lacks in tier 2 data sources. Analog ingredients can be found using the Analog Identification Methodology developed by the US Environmental Protection Agency (USEPA, 2018). If CTD cannot be found in tier 1–4 data sources, then a data gap will be identified for the target ingredient concerning the hazard endpoint. A moderate-level hazard will be assumed for the data-missing endpoint to generate an assessment outcome due to the conservativeness of hazard assessment.

G. Hu et al. / Process Safety and Environmental Protection 130 (2019) 126–139

129

Fig. 1. Flowchart of integrated hazard screening and indexing system for hydraulic fracturing chemical assessment.

3.4. Hazard screening The acquired CTD is screened against the hazard endpoints to generate a hazard screening result for the ingredient. As shown in Table 2, four hazard groups (HG) were established as possible results of chemical hazard screening. The HGs were developed in light of the four benchmarks used by the GreenScreen System (CPA, 2016). Each HG has several hazard classification criteria, and each criterion is the combination of different EHH hazards. For example, if an ingredient is associated with high environmental persistence (P) and high bioaccumulation potential (B), then this ingredient will be classified into HG1. Numerical values (NHG ) were assigned to the HGs for further integration with hazard indexing results. The ingredient hazard screening process is shown in Fig. 3. Inorganic ingredients are required to be assessed differently because most of them are inherently environmentally persistent. For naturally occurring inorganic substances, persistence is not necessarily considered a negative characteristic (CPA, 2016). If an inorganic ingredient is identified as having no significant EHH toxicity, then it will still be classified into HG4 (CPA, 2016). Also, endpoint chronic human oral toxicity (ChT) was selected to accommodate the pos-

sibility of long-term, repetitive exposure to hazardous ingredients present in potable water resources as a result of spill accidents. This endpoint is reserved for ingredients that are identified as highly persistent in the environment and having the potential to cause chronic hazard exposure (Intrinsik, 2013). Hence, rapidly and inherently biodegradable organic ingredients are exempted from the ChT screening. As Fig. 4 shows, determining the HG for an additive is a hierarchical screening process starting with searching HG1 ingredients. An additive will be classified into HG1 if it contains at least one HG1 ingredient with a concentration higher than the cutoff concentration of the concerned hazard endpoint (Table 1). If an ingredient’s concentration is lower than the cut-off concentration of a hazard endpoint, then the respective hazard will be considered negligible (CPA, 2016; Intrinsik, 2013). However, there is an exceptional situation: additives containing any ingredients of high P and B hazards will be classified into HG1 regardless of the cut-off concentrations screening results. This is because if the additive is spilled, the high P and B ingredients may be longlasting in the environment and consequently accumulate in the receptors’ bodies to cause adverse health effects (CPA, 2016). If

130

G. Hu et al. / Process Safety and Environmental Protection 130 (2019) 126–139

Fig. 2. Four tiers of chemical toxicological data sources and data searching rule (CASRN: chemical abstracts service registry number, DCS: data confidence score; target ingredient and analog ingredient have different DCSs).

Fig. 3. Hazard screening process at the ingredient level (ChT: chronic human oral toxicity, P: environmental persistence, HG: hazard group).

G. Hu et al. / Process Safety and Environmental Protection 130 (2019) 126–139

131

Table 1 Environmental and human health hazard endpoints, cut-off concentrations, hazard criteria, and hazard scores inclusive in integrated chemical hazard screening and indexing system. Environmental health hazard Endpoints

Human health hazard Cut-off conc. (%)

Environmental persistence (P)

Bioaccumulation potential (B) Acute/chronic aquatic toxicity (AT)

1.0

1.0

1.0

Criteria Rapidly/readily biodegradable Inherently biodegradable Not rapidly/inherently biodegradable High Low Category 1b Category 2 Category 3 Category 4

HSa

Endpoints

10

Carcinogenicity (C), mutagenicity (M), reproductive toxicity (R), and endocrine disruptor (E)

5 0

10 0 10 6 3 0

Cut-off conc. (%)

HSa

Criteria c

d

Group /Category 1

10

Group/Category 2 Not Group/Category 1 or 2 Yes No Category 1e Category 2 Category 3 Category 4

5

0.1

Chronic human oral toxicity (ChT)

1.0

Acute human oral toxicity (AhT)

1.0

0

10 0 10 6 3 0

Hazard scores. Hazard scores ∈ [0, 10], and a higher hazard score indicates a higher hazard with respect to the hazard endpoint. Based on acute or chronic LC50 , EC50 , or IC50 values for fish, algae, or daphnia (Category 1: ≤ 1 mg/L, Category 2: > 1 to ≤ 10 mg/L, Category 3: > 10 to ≤ 100 mg/L, and Category 4: >100 mg/L). c Based on the Globally Harmonized System definitions (Group 1: known or presumed carcinogens, mutagens, or reproductive toxicants; Group 2: suspected carcinogens, mutagens, or reproductive toxicants). d Based on the European Commission Endocrine Disruptor Strategy (Category 1: evidence of endocrine disrupting activity in animal tests; Category 2: in vitro evidence of biological activity related to endocrine disruption). e Based on oral toxicity values (Category 1: < 5 mg/kg bodyweight; Category 2: > 5 to ≤ 50 mg/kg bodyweight; Category 3: >50 to ≤ 300 mg/kg bodyweight; Category 4: > 300 mg/kg bodyweight). a

b

Fig. 4. Hazard screening process at additive/fracturing fluid levels (HG: hazard group, P: environmental persistent, B: bioaccumulation potential).

the additive does not contain any HG1 ingredients, then search for HG2 ingredients and screen the concentrations of ingredients using the same method. Screening ingredients from HG1 to HG4 and their concentrations to determine the suitable HG for the additive. In ICHSIS, the hazard assessment of fracturing fluids is the same as that of additives. This is because a fracturing fluid can be regarded

as an additive which contains a large number of ingredients diluted in millions of gallons of water (Kargbo et al., 2010). The hazard screening results of fracturing fluids are also represented by four HGs. It is important to note that if a fluid consists of two or more additives containing the same ingredient, then the total concentration of this ingredient in the fluid will be screened against the cut-off concentrations.

132

G. Hu et al. / Process Safety and Environmental Protection 130 (2019) 126–139

Table 2 Four hazard groups as chemical hazard screening outcomes. HGa HG 1

HG 2

HG 3

HG 4

a b c d e f g h

NHG b

Classification criteria c

d

P : Inherently biodegradable AND B : High AND ATe : Category 2 P: Not rapidly/inherently biodegradable AND B: High P: Not rapidly/inherently biodegradable AND AT: Category 1 OR C, M, R, or Ef : Group 1 OR AhTg : Category 1 OR ChTh : Yes B: High AND AT: Category 1 OR C, M, R or E: Group 2 OR AhT: Category 2 OR ChT: Yes C, M, R or E: Group 1 P: Inherently biodegradable AND AT: Category 3 OR C, M, R, or E: Group 2 OR AhT: Category 3 P: Inherently biodegradable AND B: High P: Not rapidly/inherently biodegradable AND AT: Category 3 OR C, M, R or E: Group 2 OR AhT: Category 3 B: High AND AT: Category 3 OR C, M, R or E: Group 2 OR AhT: Category 3 C, M, R or E: Group 2 AT: Category 1 OR AhT: Category 1 OR ChT: Yes P: Not rapidly/Inherently biodegradable OR Inherently biodegradable B: High AT: Category 2 OR 3 AhT: Category 2 OR 3 P: Rapidly biodegradable AND B: Low AND AT: Category 4 AND C, M, R, and E: Not Group 1 or 2 AND AhT: Category 4 AND ChT: No

10

6

3

0

Hazard group. Numerical value of hazard group. Environmental persistence. Bioaccumulation potential. Acute/chronic aquatic toxicity. Carcinogenic, mutagenic, reproductive toxic, and endocrine disruptive effects. Acute human oral toxicity. Chronic human oral toxicity.

3.5. Hazard indexing A hazard indexing approach is used in combination with the hazard screening approach. The conversion of multi-dimension non-commensurate chemical hazard data into numerical indices involves several steps, including scoring hazard criteria, weighing hazard endpoints, aggregating scores to generate an index, and scaling and interpreting the derived index. The HSs of hazard criteria (Table 1) are aggregated according to the weights of hazard endpoints to generate a final hazard index. A set of weights of m hazard can be written as W = (w1 , w2 , endpoints m . . ., wi , . . ., wm ), where w = 1. The analytical hierarchy proi i=1 cess (AHP) was used to assign weights (wi ) to hazard endpoints. AHP can generate a weight for each hazard endpoint according to pairwise comparisons of the relative importance of endpoints (Saaty, 2008). A higher weight indicates higher relative importance of the endpoint. As shown in Table 3, the pairwise relative importance comparison of two endpoints are measured according to a numerical scale from 1 to 9. In the indexing approach, an ingredient’s environmental health hazard and human health hazard are assessed separately, enabling assessors to know the most critical hazard more intuitively. As a result, two matrices (JE and JH ) were established for pairwise comparisons of environmental health and human health hazard

endpoints on their relative importance (Table 3). Each element in the lower triangle of the matrix is the reciprocal of an element in the upper triangle. The geometric mean of each row of the matrix is calculated, and then the weight of each endpoint (wi ) can be derived by normalizing the geometric means. The importance values were assigned according to experts’ opinions, and they can be modified as required if better information becomes available. Endpoint AT was assigned the highest importance among the three environmental health hazard endpoints since the acute aquatic toxic effect is lethal and immediate. Endpoint B was given the second-high importance since a high B ingredient can potentially be accumulated in receptors’ bodies and cause chronic adverse health effects. Endpoint P was assigned the lowest importance because if a highly persistent ingredient without any significant AT or B effect is released into the environment, it will not be an immediate health hazard to the environment. Among the human health hazard endpoints, endpoint C was assigned the highest importance because it has a serious health hazard implication to the public. Endpoint AhT was assigned the second-high importance since the toxic effect is immediate and lethal. Endpoints M and R were assigned moderate importance because their effects are not lethal and immediate compared to C and AhT. Endpoint ChT was assigned relatively low importance since the adverse effect can only be triggered by repetitive chronic hazard exposure. Endpoint E was also assigned low importance because very few chemical toxicity databases have the data for its evaluation. An ingredient environmental hazard score (IES) and a human hazard score (IHS) are calculated through a weighted sum aggregation, respectively: IES or IHS =

m 

wi · HSi

(1)

i=1

where wi and HSi are the weights and HS of hazard endpoint i, respectively, and m is the total number of environmental health and human health hazard endpoints, respectively. An ingredient hazard index (HII ) is determined using the maximum operator function: HII = Max (IES, IHS)

(2)

The HII reflects the highest possible hazard of an ingredient to either environmental health or human health. After hazard screening and indexing, a HG designation (HGI ) and a HII can be generated for an ingredient. HGI shows the inherent EHH hazard objectively, while HII incorporates assessors’ judgments on the relative importance of different hazard endpoints. Both HGI and HII are integrated to generate a final hazard assessment outcome for the ingredient. Because HG can reflect the EHH hazard more objectively than HI, a higher weight should be assigned to HG. Different pairwise mathematical weights (wHG , wHI ) were assigned to HG and HI under five scenarios: S1 (0.5, 0.5), S2 (0.6, 0.4), S3 (0.7, 0.3), S4 (0.8, 0.2), and S5 (0.9, 0.1), for integrating hazard screening and indexing results, where a higher wHI suggests more influence of experts’ judgment on assessment outcomes. Under each scenario, an integrated hazard value (IHV) can be calculated: IHV = wHG · NHG + wHI · HI

(3)

where NHG is the numerical value of HG (Table 2). For an additive consists of n ingredients, the additive hazard index (HIA ) is calculated: HIA =

n  j

HIIj · Cj

(4)

G. Hu et al. / Process Safety and Environmental Protection 130 (2019) 126–139

133

Table 3 Matrices of relevant importance of environmental health and human health hazard endpoints. Matrix

JE a

JH

b

Pairwise comparison of endpoints P B AT C M R AhT ChT E

P 1.00 2.00 5.00c C 1.00 0.20 0.20 0.33 0.20 0.20

B 0.50 1.00 3.00 M 5.00 1.00 0.33 3.00 0.33 0.33

AT 0.20 0.33 1.00 R 5.00 3.00 1.00 3.00 0.33 0.33

AhT 3.00 0.33 0.33 1.00 0.33 0.33

ChT 5.00 3.00 3.00 3.00 1.00 1.00

E 5.00 3.00 3.00 3.00 1.00 1.00

wi

Rank

0.12 0.23 0.65

3 2 1

0.44 0.14 0.10 0.22 0.05 0.05

1 3 4 2 5 5

a

Environmental health hazard endpoints importance matrix (consistency ratio = 0.4% < 10%, acceptable). Human health hazard endpoints importance matrix (consistency ratio = 7.3% < 10%, acceptable). Relative importance value: 1-two endpoints are equally important, 3-one endpoint is moderately more important than the other, 5-strongly more important, 7-very strongly more important, 9-extremely more important; importance values of 2, 4, 6, 8 in-betweens. The example shows that hazard endpoint AT is 5 times (strongly) more important than hazard endpoint P. b c

Table 4 A scaling system for interpreting environmental and human health hazard levels based on chemical hazard assessment outcomes. Hazard assessment outcome

Hazard level

Interpretation

(7, 10]

Very high

The chemical is a serious threat (e.g., immediate toxic and/or lethal effect) to environmental and/or human health. Its use should be avoided. The chemical is an environmental and/or human health hazard of high concern (e.g., sub-lethal effect). It is not allowed for use unless chemicals with lower hazards cannot be found. The chemical is of moderate environmental and/or human health hazards. It is allowed for use, but opportunities exist for hazard mitigation by using chemicals with lower EHH hazards. The chemical is a potential hazard to environmental and/or human health (e.g., the chronic adverse effects resulted from long-term hazard exposures), which is allowed for use with cautions. The chemical is not a significant hazard to environmental or human health, which is recommended for use.

(5, 7]

High

(3, 5]

Medium

(0, 3]

Low

0

No hazard

where HIIj and Cj are the hazard index and the normalized maximum concentration of the jth ingredient within the additive, respectively. If an additive contains any undisclosed ingredients, then the undisclosed ingredients will be considered as having no significant EHH hazards (CCOHS, 2018). Similarly, a fracturing fluid hazard index (HIF ) can be calculated using Eq. (4). Based on the determined HGA/F and HIA/F , an IHVA/F can also be calculated for an additive/fracturing fluid using Eq. (3), respectively. As shown in Table 4, a scaling system was established for hazard assessment outcome interpretation. The scaling system was developed in light of the environmental and health hazard scoring system used in HyFFGAS (Hurley et al., 2016). 3.6. Data confidence evaluation It is important to evaluate the data confidence levels of assessment results since they can greatly affect the decision making in chemical management. A data confidence index (DCI) is calculated for each ingredient, additive (DCIA ), and fracturing fluid (DCIF ) as a measurement of the data certainty/uncertainty of hazard assess-

ment results. At the ingredient level, DCI is presented separately for environmental health (DCIE ) and human health (DCIH ) hazard assessment results. The data confidence index is calculated based on the DCSs (Fig. 2) of different CTD sources for endpoint i using Eq. (5): m 

DCSi

i

DCI =

Max

m 

(5) DCSi

i

At the ingredient level, the maximum sum of DCS for environmental health hazard is 30 and for human health hazard is 60. The resultant DCI is in a range from 0 to 1, where a higher value denotes higher data confidence of assessment results. Similarly, the DCIA and DCIF can be calculated for additive/fluid hazard assessment results using Eq. (5), respectively. A DCI scaling system (Table 5) was also established to interpret the data certainty of hazard assessTable 5 A scaling system for evaluating data certainty/uncertainty levels of hazard assessment results. DCIa

Level

Interpretation

(0.75, 1.00]

High

(0.50, 0.75]

Medium-tohigh

[0.25, 0.50)

Low-tomedium

[0, 0.25)

Low

Signals that a substantial amount of credible CTDb exists for the substance across the EHH hazard endpoints. Uncertainty with respect to the assessment result is low. Signals a lower, but still appreciable amount of CTD exists for the substance across the EHH hazard endpoints. The credibility of data is also lower. The uncertainty of the assessment result remains relatively low. Signals a lower amount of CTD exists for the substance across the EHH hazard endpoints. Also, the data credibility is not high. Uncertainty surrounding the assessment result is increased. Further review is recommended. Signals both the amount of CTD and the data credibility are low for the substance across the EHH hazard endpoints. Uncertainty surrounding the assessment result is high. Further review is strongly recommended.

a DCI: data confidence index calculated for ingredients, additives, and fracturing fluids. b CTD: Chemical toxicological data.

134

G. Hu et al. / Process Safety and Environmental Protection 130 (2019) 126–139

Fig. 5. Ingredient hazard screening results.

ment results. The four data confidence levels used in the DCI scaling system were established in accordance with the data availability measurements of the Intrinsik Screening-level Assessment System (Intrinsik, 2013).

assessment. As a result, a total of 25 additives were selected. The ingredient compositions of the selected additives are included in Appendix II. 4.1. Ingredient hazard assessment

4. Application The developed ICHSIS was applied to assess the representative hydraulic fracturing additives used in British Columbia. The additive data was collected from the FracFocus database from November 2011 to August 2014. Different additives were grouped into 13 functional categories such as gelling agent, crosslinker, and biocide according to their designed functions (Hu et al., 2018a). Additives with a use percentage > 10% within the respective functional category were considered representative and selected for

The selected additives comprise 43 different ingredients. Among these ingredients, twelve were reported without CASRNs; hence, analog ingredients were used to substitute the CASRN-missing ingredients. As Fig. 5 shows, about 21% of the assessed ingredients were classified into HG1, and these ingredients should be avoided for use in hydraulic fracturing operations from a hazard mitigation perspective. No special HG1 ingredient characterized by high P and B hazards was identified. Nearly half of the assessed ingredients were HG2 substances, suggesting that the supplementary hazard

G. Hu et al. / Process Safety and Environmental Protection 130 (2019) 126–139

135

Fig. 6. Comparisons of ingredient hazard indexing results (HII ), integrated hazard values (IHVI ) under five scenarios (S1 to 5), and results from HyFFGAS (HyFI ).

indexing approach is desirable to aid in the comparison of ingredients within the same HG. Only 9% of the ingredients were identified as non-hazardous HG4 ingredients. Among the ingredients assessed, no confirmed or suspected endocrine disruptor was found. Environmental persistence and high aquatic toxicity are the main environmental health concerns for the ingredients. This finding is generally consistent with the results from the previous comprehensive chemical hazard assessment, in which P (30%) and AT (27%) were confirmed as the causes of high environmental health hazard (Hu et al., 2018a). About 20 to 25% of the assessed ingredients were associated with the GHS Group 1 or 2 adverse human health effects (i.e., C, M, and R). AhT was not identified as a significant hazard because the majority of ingredients (80%) are associated with the GHS Category 4 acute oral toxicity. Nevertheless, roughly 21% of the ingredients were confirmed with ChT, which might cause long-term adverse health effects. The ingredients’ hazard indexing results (i.e., HII ) were compared with the integrated hazard results (i.e., IHVI ) and the results from HyFFGAS. The IHVs were calculated under five scenarios (S1 to S5) defined by different pairwise mathematical weights of HG and HI. HyFFGAS generates ingredient greenness scores on a scale from 0 to 10, where a higher greenness score indicates a lower level of EHH hazard, so the greenness scores were subtracted by ten to generate a hazard index (HyF) for comparison (Hurley et al., 2016). Based on the results, the hazard levels of ingredients were determined and compared according to the hazard scaling system shown in Table 4. As Fig. 6 shows, there is a significant difference between the hazard assessment results from the individual hazard indexing and the integrated approaches. The HII show that about 35% of the assessed ingredients fall within “high” and “very high” hazard levels, while the IHVI indicate that higher percentages (40 to 68%) of ingredients were categorized into “high” and “very high” hazard levels under five weighing scenarios. The difference reveals that the individual indexing approach could underestimate ingredients’ hazards. Moreover, the scenario-based assessments found that a higher wHG could lead to more critical hazard assessment results. The percentage of “high” hazard ingredients increased from 21 to 47% as wHG increases from 0.5 (S1) to 0.9 (S5), while the percentage of “medium” hazard ingredients decreased from 37 to 9%. The percentages of ingredients with “no”, “low”, and “very high” hazards were relatively insensitive to the change of wHG . The comparison between the IHVI and HyFI also shows a difference. About 9% of the ingredients were categorized into the “very high” hazard level based on the indexing results from HyFFGAS (Fig. 6). This value is significantly lower than the percentage (21%) identified by ICHSIS. A total of 20% of ingredients were categorized into the “high” hazard level by HyFFGAS, which is also lower than the percentage (30%) determined by ICHSIS under a moderate

weighting scenario (S3). The results further suggest that individual hazard indexing systems could generate underestimated hazard assessment results compared to the integrated system. The underestimated EHH hazards by HyFFGAS can be attributed to its hazard aggregation method, in which equal weights are assigned to environmental health and human health hazard assessment outcomes to generate a final EHH hazard score. For example, ingredient Benzene, C10-16-alkyl derivatives (CASRN: 68648-873) is a significant environmental health hazard concern because of its high bioaccumulation potential (ECCC, 2018) and the GHS Category 1 aquatic toxicity (ECHA, 2018); however, this substance is not linked to any significant human health hazard. This ingredient was classified into HG1 and a HII of 8.8 was calculated by using ICHSIS. Under the least strict weighting scenario (S1), an IHVI of 9.4 was determined, indicating that the ingredient is associated with “very high” EHH hazard. In comparison, HyFFGAS generated an environmental health hazard score of 9 and a human health hazard score of 2 for this ingredient. An ingredient hazard score of 5.5 was calculated by assigning equal weight to the two types of hazard. The hazard score indicates that the ingredient can be categorized into the “high” hazard level (Hurley et al., 2016). Thus, for the same ingredient, the hazard assessment result of HyFFGAS is less critical than that of ICHSIS. The IHVI derived from a moderate weighting scenario (S3) were selected for further comparison of hazard assessment and data confidence results. The distribution of values from different ingredient hazard assessment approaches are shown in Fig. 7a. Both the median and mean values of IHVI -S3 are higher than those of HII and HyFI , suggesting a more critical ingredient EHH hazard profile. It is also noteworthy that the mean and median values of IES are significantly higher (p < 0.05) than those of IHS, indicating that environmental health hazard is more critical than human health hazard at the ingredient level. The data confidence performance of different ingredient hazard assessment results is shown in Fig. 7b. Since HyFFGAS does not have a function for data confidence evaluation, the DCIs of the assessment results from HyFFGAS were calculated using the same approach (Eq. 5) used in ICHSIS. It can be seen that the data confidence performance was significantly (p < 0.05) improved by ICHSIS. The ingredient assessment results from ICHSIS are associated with “high” level of data confidence (Table 5), while the results from HyFFGAS are associated with much lower data confidence. The high data confidence can be attributed to the diverse CTD sources used in ICHSIS. In comparison, HyFFGAS uses material safety data sheets as the only CTD source, which is equivalent to the tier 2 data source used in ICHSIS (Hurley et al., 2016). Using tier 1 chemical toxicity databases not only increases the availability of CTD but also improves data credibility. Moreover, ICHSIS uses analog ingredients to substitute the CASRN-missing ingredients, which can also significantly reduce the data uncertainty.

136

G. Hu et al. / Process Safety and Environmental Protection 130 (2019) 126–139

Fig. 7. Comparisons of (a) ingredient hazard assessment results (IES: ingredient environmental health hazard score, IHS: ingredient human health hazard score, HII : ingredient hazard index, IHVI -S3: integrated hazard value of ingredients under scenario 3, HyFI : ingredient hazard index from HyFFGAS) and (b) data confidence indices (DCIE : data confidence index-ingredient environmental health hazard, DCIH : data confidence index-ingredient human health hazard).

4.2. Additive hazard assessment Among the additives assessed, about 28% were categorized into HG1. These HG1 additives contain at least one HG1 ingredient at a concentration higher than the cut-off values. The relatively large percentage of HG1 additives indicates that the need for using additives with lower EHH hazards is significant. About 44% and 16% of additives were identified as HG2 and HG3 chemicals, respectively. The high percentage of HG2 additives also indicates that it is necessary to use the hazard indexing approach to compare the EHH hazards of additives within the same HG. Only 12% of the additives were identified as having no significant EHH hazard. As Fig. 8 shows, about 20% of the assessed additives were classified into “high” and “very high” hazard levels based on HIA . After integrating the hazard screening results, the percentages of additives with “very high” and “high” hazards increased from 8 to 28%

Fig. 9. Comparisons of (a) additive hazard assessment outcomes (HIA : additive hazard index, IHVA -S3: integrated hazard values of additives under scenario 3, HyFA : additive hazard index from HyFFGAS) and (b) data confidence indices (DCIA : data confidence index-additive, DCIA -HyF: data confidence index-results from HyFFGAS).

and 12 to 24%, respectively. These percentages also increase as a result of increasing wHG from S1 to 5. Nevertheless, the percentage of additives identified without any significant hazard remains the same from S1 to 5. On the other hand, the indexing results (i.e., HyFA ) from HyFFGAS indicate that a total of 20% of the additives are associated with “very high” and “high” hazards, which is significantly lower than the result (52%) from ICHSIS under a moderate weighting scenario (S3). The comparison shows that using individual hazard indexing methods could also generate underestimated hazard assessment outcomes at the additive level. The integrated additive hazard assessment results calculated under S3 (i.e., IHVA -S3) were selected for comparing the distributions of values. As Fig. 9a shows, the mean and median values of IHVA -S3 are higher than those of HIA and HyFA , suggesting a higher overall EHH hazard at the additive level. As Fig. 9b shows, the DCIA from ICHSIS are much higher than the DCIA -HyF from HyFFGAS, and the improvement is significant (p < 0.05). The data confidence of the

Fig. 8. Comparisons of additive hazard indexing results (HIA ), integrated hazard values of additives (IHVA ) under five scenarios (S1 to 5), and results from HyFFGAS (HyFA ).

G. Hu et al. / Process Safety and Environmental Protection 130 (2019) 126–139

137

Table 6 Hazard assessment results of three hypothetical fracturing fluids. Additive

Iron control agent

Anti-sludge agent

Activator

Scale control agent

Biocide

Breaker

Clay control agent

Gelling agent

Crosslinker

a

Ingredient

2-Mercaptoethanol Cupric chloride Monoethanolamide Alkyl benzene sulphuric acid Methanol Benzene, C10-16-alkyl derivatives Sulphuric acid Methanol Alcohols, C12-14 secondary, ethoxylated Ethylene glycol Non-hazardous ingredients Glutaraldehyde Methanol Non-hazardous ingredients Ammonium persulfate Non-hazardous ingredients 1,6-Hexandiamine, dihydrochloride Non-hazardous ingredients Phenol formaldehyde resin Phosphoric Acid Monoethanolamide Boric acid Non-hazardous ingredients

CASRN

Conc. (%) in additive

Assessment resultsa

Conc. (%) in fracturing fluid

F1

F2

F3

Ingredient

Additive

60-24-2 7447-39-4 141-43-5 68584-22-5

90 10 30 85

2.40E-03 2.60E-04 8.00E-04 2.05E-03

2.40E-02 2.60E-03 8.00E-03 2.05E-02

2.40E-03 2.60E-04 8.00E-04 2.05E-02

(1, 7.7, 9.3) (1, 7.7, 9.3) (3, 0.7, 2.3) (2, 1.0, 4.5)

67-56-1 68648-87-3

10 5

2.41E-04 1.20E-04

2.41E-03 1.20E-03

2.41E-03 1.20E-03

(3, 3.9, 3.3) (1, 7.7, 9.3)

7664-93-9 67-56-1 84133-50-6

5 50 70

1.20E-04 5.00E-03 8.00E-03

1.20E-03 5.00E-02 8.00E-02

1.20E-03 5.00E-02 8.00E-02

(3, 1.2, 2.5) (2, 0.5, 4.4) (2, 6.5, 6.2)

107-21-1 –

60 60

8.29E-04 8.29E-03

8.29E-03 8.29E-02

8.29E-03 8.29E-03

(2, 0.5, 4.4) (4, 0.0, 0.0)

(2, 0.3, 4.3)

111-30-8 67-56-1 –

20 10 70

1.95E-04 1.60E-06 1.12E-05

1.95E-03 1.60E-05 1.12E-04

1.95E-03 1.60E-05 1.12E-04

(2, 6.5, 6.2) (2, 0.5, 6.0) (4, 0.0, 0.0)

(2, 1.4, 4.6)

7727-54-0

5

4.24E-04

4.24E-03

4.24E-04

(3, 3.2, 3.1)

–

95

8.05E-03

8.05E-02

8.05E-03

(4, 0.0, 0.0)

6055-52-3

40

7.47E-03

7.47E-02

7.47E-02

(4, 0.0, 0.0)

–

60

1.12E-02

1.12E-01

1.12E-01

(4, 0.0, 0.0)

9003-35-4

95

6.11E-02

6.11E-01

6.11E-02

(1, 7.7, 9.3)

7664-38-2 141-43-5 10043-35-3 –

5 60 13 27

3.21E-03 8.75E-04 1.17E-03 1.02E-03

3.21E-02 8.75E-03 1.17E-02 1.02E-02

3.21E-03 8.75E-04 1.17E-03 1.02E-03

(3, 1.2, 2.5) (3, 0.7, 2.3) (1, 3.2, 8.0) (4, 0.0, 0.0)

Fracturing fluid

(1, 6.1, 8.8)

(1, 1.6, 7.5)

(2, 4.0, 5.4)

F1: (4, 4.6, 1.4) F2: (1, 4.6, 8.4) F3: (2, 2.5, 5.0)

(3, 0.2, 2.1)

(4, 0.0, 0.0)

(1, 7.4, 9.2)

(1, 0.8, 7.3)

Assessment results are presented as (hazard group, hazard index, integrated hazard value-scenario 3) for ingredients, additives, and fracturing fluids.

results from ICHSIS is primarily at the “high” level, while the data confidence of HyFFGAS assessment results mainly lies between the “low-to-medium” and “medium-to-high” levels. The high data confidence of additive hazard assessment results from ICHSIS can be attributed to the high data confidence of ingredient hazard assessment results. 4.3. Fracturing fluid hazard assessment Three hypothetical fracturing fluids (i.e., F1, F2, and F3) were designed for hazard assessment at the fluid level. As Table 6 shows, the fracturing fluids contain the same additives/ingredients from different functional categories, but the concentrations of ingredients are different. F1 contains the lowest concentrations of ingredients among all the three fluids. The ingredients’ concentrations in F2 are ten times higher than the concentrations of the respective ingredients in F1. F3 contains several additives such as iron control agent, activator, and biocide at high concentrations, while the concentrations of the remaining additives are the same to those in F1. The total concentration of ingredients is < 1% in each fracturing fluid, which is reasonable as chemicals normally only account for less than 2% of fracturing fluid (All Consulting, 2012; Soeder et al., 2014). The IHVF of the three fracturing fluids were calculated under a moderate weighting scenario (S3). The hazard screening results (Table 6) show that F1 can be categorized into HG4, which means that no significant EHH hazard was determined in F1 despite the fact that it contains several HG1 ingre-

dients. This is because the concentrations of ingredients in F1 are lower than the cut-off concentrations of various hazard endpoints. The IHVF of F1 was calculated as 1.4, reflective of a “low” hazard level according to the hazard scaling system (Table 4). In comparison, F2 was categorized into HG1 due to the high concentrations of the HG1 ingredients. Accordingly, the IHVF of F2 (8.4) is much higher than that of F1, indicating a “very high” level EHH hazard. Hence, from an EHH hazard mitigation perspective, F2 needs to be reformulated by using alternative ingredients with lower EHH hazards or lowering the concentrations of the HG1 ingredients (i.e., < the cut-off concentrations). Nonetheless, the individual hazard indexing result suggests that F2 is associated with a “medium” level EHH hazard, which is allowed for use. F3 was categorized into HG2, suggesting that its EHH hazard lies between F1 and F2. The concentrations of all the HG1 ingredients in F3 are lower than the cut-off values, so the HG2 ingredients were screened following the fluid screening hierarchy (Fig. 4). The concentrations of HG2 ingredients, such as alkyl benzene sulphuric acid and methanol, exceed the cut-off values, resulting in a positive HG2 designation for F3. The IHVF of F3 (5.0) indicates that the fluid has a “medium” level EHH hazard, in which the HG2 ingredients are recommended to be used at lower concentrations or substituted by HG4 alternatives for hazard mitigation. In comparison, the individual hazard indexing approach generated an underestimated hazard (i.e., a “low” level hazard based on a HIF of 2.5) for F3. The results once again show that the individual hazard indexing approach could result in eclipsed EHH hazard at the fluid level.

138

G. Hu et al. / Process Safety and Environmental Protection 130 (2019) 126–139

5. Conclusions Different chemical hazard assessment methods have been developed to measure the hazard profiles of hydraulic fracturing chemicals. The existing methods can be divided into hazard screening and indexing approaches. By reviewing the advantages and limitations of the two approaches, ICHSIS was developed to assess the EHH hazard of hydraulic fracturing chemicals used at ingredient, additive, and fracturing fluid levels. The integrated system was applied to the representative chemicals used in British Columbia. The results from the individual hazard indexing, ICHSIS, and the previously developed HyFFGAS were compared. The hazard screening results show that more than half of the ingredients and additives can be grouped into high EHH hazard designations such as HG1 and 2, suggesting that the need for hazard mitigation is necessary. More critically, the comparison of results from different approaches indicates that the individual hazard indexing approach could generate underestimated EHH hazard assessment outcomes at different chemical use levels, and thus an integrated hazard assessment approach is required for more realistic chemical hazard assessments. The comparison also shows that the data confidence level of the results was significantly improved by ICHSIS. The developed ICHSIS represents an improved chemical hazard assessment framework, which can promote progress toward more sustainable unconventional gas production. Acknowledgments The authors would like to thank the Shale Water Steering and Technical Committee of Canadian Association of Petroleum Producers (CAPP), British Columbia Oil and Gas Commission (BCOGC), British Columbia Oil and Gas Research and Innovation Society (BC OGRIS), and Mitacs Accelerate Program for their financial and technical support for this study. The authors would also like to thank the editor and anonymous reviewers for their help in improving the quality of manuscript. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.psep.2019.08. 002. References Akob, D.M., Mumford, A.C., Orem, W., Engle, M.A., Klinges, J.G., Kent, D.B., Cozzarelli, I.M., 2016. Wastewater disposal from unconventional oil and gas development degrades stream quality at a West Virginia injection facility. Environ. Sci. Technol. 50, 5517–5525. All Consulting, 2012. The Modern Practices of Hydraulic Fracturing: A Focus on Canadian Resources. Petroleum Technology Canada and Science and Community Environmental Alliance Knowledge Fund, Tulsa, Oklahoma, Accessed 20 Sept 2018 www.allllc.com/publicdownloads/ModernPracticesHFCanadianResources.pdf. Boudet, H., Clarke, C., Bugden, D., Maibach, E., Roser-Renouf, C., Leiserowitz, A., 2014. “Fracking” controversy and communication: using national survey data to understand public perceptions of hydraulic fracturing. Energy Policy 65, 57–67. Brannon, H.D., Daulton, D.J., Post, M.A., Hudson, H.G., Jordan, A.K., 2012. The Quest to Exclusive Use of Environmentally Responsible Fracturing Products and Systems. SPE 152068, Presented at the SPE Hydraulic Fracturing Technology Conference. Society of Petroleum Engineers (SPE), Woodlands, USA, pp. 6–8, Feb 2012. CAPP, 2012. CAPP’s Guiding Principles and Operating Practices for Hydraulic Fracturing. Canadian Association of Petroleum Producers (CAPP), Calgary, Alberta, Accessed 20 Sept 2018 http://www.capp.ca/canadaindustry/ naturalGas/ShaleGas/Pages/default.aspx#operating. CCOHS, 2018. Health and Safety Fact Sheets, WHMIS 1988 - Material Safety Data Sheets (MSDSs): Creating. Canadian Center for Occupational Health and Safety (CCOHS), Accessed 16 Sept 2018 https://www.ccohs.ca/oshanswers/ legisl/msds prep.html. Cozzarelli, I.M., Skalak, K.J., Kent, D.B., Engle, M.A., Benthem, A., Mumford, A.C., Haase, K., Farag, A., Harper, D., Nagel, S.C., Iwanowicz, L.R., Orem, W.H., Akob, D.M., Jaeschke, J.B., Galloway, J., Kohler, M., Stoliker, D.L., Jolly, G.D., 2017. Environ-

mental signatures and effects of an oil and gas wastewater spill in the Williston Basin, North Dakota. Sci. Total Environ. 579, 1781–1793. CPA, 2016. GreenScreen® For Safer Chemicals Hazard Assessment Guidance. Version 1.3, March 2016. Clean Production Action (CPA), Somerville, Massachusetts. ECCC, 2018. Environment and Climate Change Canada (ECCC), the Canadian Environmental Protection Act, Searching Engine for the Results of Domestic Substances List (DSL) Categorization, Accessed 20 Sept 2018 https://pollutionwaste.canada.ca/substances-search/Substance?lang=en. ECHA, 2018. Inventory Substance Information Database. European Chemicals Agency (ECHA), Helsinki, Finland, Accessed 20 Sept 2018 https://echa.europa. eu/information-on-chemicals. Engle, A.M., Cozzarelli, I.M., Smith, B.D., 2014. USGS Investigations of Water Produced During Hydrocarbon Reservoir Development. U.S. Geological Survey Fact Sheet 2014–3104, United States Geological Survey, Virginia, US, http://dx.doi.org/10. 3133/fs20143104. European Commission, 2018. European Commission Environment Strategyendocrine Disruptors, Annex 1-Candidate List of 553 Substances, Accessed 19 Sept 2018 http://ec.europa.eu/environment/archives/docum/pdf/bkh annex 01.pdf. Exon, J.H., 2006. A review of the toxicology of acrylamide. J. Toxicol. Environ. Health B Crit. Rev. 9, 397–412. Ferrari, F., Giacomini, A., Thoeni, K., 2016. Qualitative rockfall hazard assessment: a comprehensive review of current practices. Rock Mech. Rock Eng. 49, 2865–2922. FracFocus, 2014. Ground Water Protection Council and Interstate Oil and Gas Compact Commission, FracFocus Chemical Disclosure Registry, Accessed 19 Sept 2018 http://fracfocus.ca. Gallegos, T.J., Varela, B.A., 2014. Trends in Hydraulic Fracturing Distributions and Treatment Fluids, Additives, Proppants, and Water Volumes Applied to Wells Drilled in the United States From 1947 Through 2010-data Analysis and Comparison to the Literature. The U.S. Geological Survey, Reston, Virginia. Health Canada, 2015. Guidance: Disclosure of Ingredient Concentrations and Concentration Ranges on Safety Data Sheets. Health Canada, Ottawa, Ontario, July 31, 2015, http://whmis.org/documents/HPR Concentration ranges guidance 2015 final en.pdf Accessed 19 Sept 2018. Hepburn, K., 2012. Development and Practical Application of a Chemical Hazard Rating System. SPE 160548, Presented at the SPE Annual Technical Conference and Exhibition. Society of Petroleum Engineers (SPE), San Antonio, USA, pp. 8–10, Oct 2012. Hu, G., Liu, T., Hager, J., Hewage, K., Sadiq, R., 2018a. Hazard assessment of hydraulic fracturing chemicals using an indexing method. Sci. Total Environ. 619–620, 281–290. Hu, G., Kaur, M., Hewage, K., Sadiq, R., 2018b. Fuzzy clustering analysis of hydraulic fracturing additives for environmental and human health risk mitigation. Clean Technol. Environ. Policy., http://dx.doi.org/10.1007/s10098-018-1614-3. Hurley, T., Chhipi-Shrestha, G., Gheisi, A., Hewage, K., Sadiq, R., 2016. Characterizing hydraulic fracturing fluid greenness: application of a hazard-based index approach. Clean Technol. Environ. Policy 3, 647–668. Intrinsik, 2013. A Screening-level Assessment System for Categorizing Hydraulic Fracturing Fluid Additives According to Potential Human Health and Environmental Risks. Calgary, Alberta. Jordan, A., Daulton, D., Cobb, J.A., Grumbles, T., 2010. Quantitative Ranking Measures Oil Field Chemicals Environmental Impact. SPE 135517SPE 84576-MS, Presented at the SPE Annual Technical Conference and Exhibition. Society of Petroleum Engineers (SPE), Florence, Italy, pp. 19–22, Sept 2010. Kahrilas, G.A., Blotevogel, J., Stewart, P.S., Borch, T., 2014. Biocides in hydraulic fracturing fluids: a critical review of their usage, mobility, degradation, and toxicity. Environ. Sci. Technol. 1, 16–32. Kahrilas, G.A., Blotevogel, J., Corrin, E.R., Borch, T., 2016. Downhole transformation of the hydraulic fracturing fluid biocide glutaraldehyde: implications for flowback and produced water quality. Environ. Sci. Technol. 50, 11414–11423, http://dx. doi.org/10.1021/acs.est.6b02881. Kargbo, D.M., Wilhelm, R.G., Campbell, D.J., 2010. Natural gas plays in the marcellus shale: challenges and potential opportunities. Environ. Sci. Technol. 44, 5679–5684, http://dx.doi.org/10.1021/es903811p. Kassotis, C.D., Tillitt, D.E., Davis, J.W., Hormann, A.M., Nagel, S.C., 2017. Estrogen and androgen receptor activities of hydraulic fracturing chemicals and surface and ground water in a drilling-dense region. Endocrinology 155, 897–907, http://dx. doi.org/10.1210/en.2013-1697. NEB, 2017. Canada’s Role in the Global LNG Market. Energy Market Assessment. July 2017. National Energy Board (NEB), Ottawa, ON, Canada. Orem, W., Varonka, M., Crosby, L., Haase, K., Loftin, K., Hladik, M., Akob, D.M., Tatu, C., Mumford, A., Jaeschke, J., Bates, A., Schell, T., Cozzarelli, I., 2017. Applied geochemistry organic geochemistry and toxicology of a stream impacted by unconventional oil and gas wastewater disposal operations. Appl. Geochem. 80, 155–167. Renock, D., Landis, J.D., Sharma, M., 2016. Reductive weathering of black shale and release of barium during hydraulic fracturing. Appl. Geochem. 65, 73–86, http:// dx.doi.org/10.1016/j.apgeochem.2015.11.001. Saaty, T.L., 2008. Decision making with the analytic hierarchy process. Int. J. Serv. Sci. 1, 83–98. Sadiq, R., Haji, S.A., Cool, G., Rodriguez, M.J., 2010. Using penalty functions to evaluate aggregation models for environmental indices. J. Environ. Manage. 3, 706–716. Soeder, D.J., Sharma, S., Pekney, N., Hopkinson, L., Dilmore, R., Kutchko, B., Stewart, B., Carter, K., Hakala, A., Capo, R., 2014. An approach for assessing engineering

G. Hu et al. / Process Safety and Environmental Protection 130 (2019) 126–139 risk from shale gas wells in the United States. Int. J. Coal Geol. 126, 4–19, http:// dx.doi.org/10.1016/j.coal.2014.01.004. Stringfellow, W.T., Domen, J.K., Camarillo, M.K., Sandelin, W.L., Borglin, S., 2014. Physical, chemical, and biological characteristics of compounds used in hydraulic fracturing. J. Hazard. Mater. 275, 37–54, http://dx.doi.org/10.1016/ j.jhazmat.2014.04.040. Swamee, P.K., Tyagi, A., 2000. Describing water quality with aggregate index. J. Environ. Eng. 5, 451–455. Thomas, L., Tang, H., Kalyon, D.M., Aktas, S., Arthur, J.D., Blotevogel, J., Carey, J.W., Filshill, A., Fu, P., Hsuan, G., Hu, T., Soeder, D., Shah, S., Vidic, R.D., Young, M.H., 2019. Toward better hydraulic fracturing fluids and their application in energy production: a review of sustainable technologies and reduction of potential environmental impacts. J. Petrol. Sci. Eng. 173, 793–803, http://dx.doi.org/10. 1016/j.petrol.2018.09.056. UN, 2013. Globally Harmonized System of Classification and Labelling of Chemicals (GHS), 5th edn. United Nations, New York http://www.unece.org/fileadmin/ DAM/trans/danger/publi/ghs/ghs rev05/English/ST-SG-AC10-30-Rev5e.pdf.

139

US EPA, 2018. Predictive Models and Tools for Assessing Chemicals Under the Toxic Substances Control Act (TSCA), Analog Identification Methodology (AIM) Tool. The United States Environmental Protection Agency (EPA), Washington, D.C, Accessed 20 Sept 2018 https://www.epa.gov/tsca-screening-tools/analogidentification-methodology-aim-tool. Vengosh, A., Jackson, R.B., Warner, N., Darrah, T.H., Kondash, A., 2014. A critical review of the risks to water resources from unconventional shale gas development and hydraulic fracturing in the United States. Environ. Sci. Technol. 15, 8334–8348. Verslycke, T., Reid, K., Bowers, T., Thakali, S., Lewis, A., Sanders, J., Tuck, D., 2014. The chemistry scoring index (CSI): a hazard-based scoring and ranking tool for chemicals and products used in the oil and gas industry. Sustainability 6, 3993–4009.