A comparison of hourly with annual air pollutant emissions: Implications for estimating acute exposure and public health risk

ABSTRACT

Health risks from air pollutants are evaluated by comparing chronic (i.e., an average over 1 yr or greater) or acute (typically 1-hr) exposure estimates with chemical- and duration-specific reference values or standards. When estimating long-term pollutant concentrations via exposure modeling, facility-level annual average emission rates are readily available as model inputs for most air pollutants. In contrast, there are far fewer facility-level hour-by-hour emission rates available for many of these same pollutants. In this report, we first analyze hour-by-hour emission rates for total reduced sulfur (TRS) compounds from eight kraft pulp mill operations. This data set is used to demonstrate discrepancies between estimating exposure based on a single TRS emission rate that has been calculated as the mean of all operating hours of the year, as opposed to reported hourly emission rates. A similar analysis is then performed using reported hourly emission rates for sulfur dioxide (SO₂) and oxides of nitrogen (NO_x) from three power generating units from a U.S. power plant. Results demonstrate greater variability at kraft pulp mill operations, with ratios of reported hourly to average hourly TRS emissions ranging from less than 1 to greater than 160 during routine facility operations. Thus, if fluctuations in hourly emission rates are not accounted for, over- or underestimates of hourly exposure, and thus acute health risk, may occur. In addition to this analysis, we also demonstrate an additional challenge when assessing health risk based on hourly exposures: the lack of human health reference values based on 1-hr exposures.

Implications: Largely due to the lack of reported hourly emission rate data for many air pollutants, an hourly average emission rate (calculated from an annual emission rate) is often used when modeling the potential for acute health risk. We calculated ratios between reported hourly and hourly average emission rates from pulp and paper mills and a U.S. power plant to demonstrate that if not considered, hourly fluctuations in emissions could result in an over- or underestimation of exposure and risk. We also demonstrate the lack of 1-hr human health reference values meant to be protective of the general population, including children.

Introduction

The Clean Air Act (CAA) directs the U.S. Environmental Protection Agency to set health-based standards for hazardous air pollutants (HAPs), criteria air pollutants (CAPs), and other designated pollutants released into the air from anthropogenic sources (CAA 1990). For example, the pulp and paper industrial sector consists of facilities that convert wood into pulp (i.e., pulp mills) and facilities that convert pulp into paper (i.e., paper mills). There are currently 107 chemical pulp mills in operation in the United States (U.S. Environmental Protection Agency [EPA] 2017a). Of these, 97 utilize the kraft pulping process. In this process, wood chips are treated under pressure and high temperature with a mixture of sodium hydroxide and sodium sulfide. Pulp is washed prior to paper making, and spent cooking chemicals are recovered for reuse in the process and to generate steam. These processes result in emissions of HAPs, CAPs, and designated pollutants that can result in human health effects. Studies of pulp mill workers have shown that chemicals from operations at these facilities can increase the risk for dermatitis, airway inflammation, cardiovascular disease, and/or cancer following occupational exposures (Andersson et al. 2007; Jungbauer et al. 2005; Rylander, Thorn, and Attefors 1999; Toren, Hagberg, and Westberg 1996; Toren, Persson, and Wingren 1996). An epidemiologic study reported a possible association between proximity to pulp mills and wheezing symptoms in adolescents who reported using cigarettes or experienced secondhand smoke exposure (Mirabelli and Wing 2006).

Figure 1. Histogram of ratios of reported to average hourly TRS emission rates for 11 recovery boilers located at eight kraft pulp and paper mills.

Source emission rates for toxic air pollutants are typically estimated by EPA from source category survey information or compiled in data sets such as the National Emissions Inventory (NEI). In most cases, these emission data are representative of an annual emission total (e.g., tons per year [TPY]). For most chemicals emitted, finer-time-scale emission rate information is unavailable. Therefore, in chronic risk assessments such as those performed by EPA under CAA section 112(f)(2) (National Research Council [NRC] 1983; EPA 1999; Smith et al. 2018), HAPs are generally assumed to be emitted at a constant rate for every hour of the year (e.g., annual emissions divided by 8760 hr in a year). Dispersion models such as AERMOD (American Meteorological Society/U.S. Environmental Protection Agency Regulatory Model; EPA 2016) then estimate hour-by-hour ambient chemical concentrations for every hour of the modeled period, taking into account the local hourly meteorological data for that period (e.g., wind direction and speed). With respect to estimating chronic noncancer health risk, all estimated hourly chemical concentrations at a location of interest (i.e., at a particular model receptor) are averaged to calculate an annual average concentration that can be compared with a chronic noncancer health reference value that is based on a long-term (at least a year) exposure level (supplemental material, Figure S1). Although there is some uncertainty in estimating hourly ambient concentrations from an annual emission rate that is assumed to be constant, the effect is expected to be mitigated given that an entire year’s worth of average emission data is being combined with an entire year’s worth of site-specific meteorological data. Thus, at a receptor of interest, there are likely to be both over- and underestimations of actual hourly concentrations that are minimized when averaged over the course of the year.

However, chemical emission rates reported as long-term averages can be problematic for estimating 1-hr exposure and public health risk using dispersion modeling. This is largely because fluctuations in short-term emission rates that result in increased pollutant concentrations are unknown, and it is these short-term increases in chemical concentrations that are of particular concern when estimating acute risk based on hourly exposures. That is, if all other modeling parameters (e.g., release parameters, meteorology, 1-hr human health reference value) are held constant, differences in exposure concentrations and health risk are largely the result of the emission rates used as dispersion model inputs (EPA 2017b). Importantly, fluctuations in hourly emission rates may be substantial for some industries. For example, facilities utilizing batch operations would be expected to have greater hourly fluctuations in emissions than facilities that operate on a continuous basis.

In situations where 1-hr pollutant exposure concentrations can be estimated with confidence—perhaps due to the availability of hour-by-hour emission rates—there is still an additional challenge for generating meaningful human health risk estimates. There is a lack of 1-hr human health reference values (HHRVs) intended for use in the general population, including for most HAP compounds. When acute reference values designed to be health protective in sensitive populations such as children and older adults are not available for a specific HAP, acute reference values designed for use in emergency response situations are often used as alternatives. However, these values are designed for a once-in-a-lifetime exposure and may not protect these sensitive groups from repeated exposures. A more complete discussion of the human health reference values for acute and chronic durations for several chemicals is available in a series of EPA reports (EPA 2012a, 2012b, 2012c, 2012d, 2012e), and a comparison of values within and between averaging times is available separately (Woodall 2005). Current short- and long-term air quality standards for CAPs can be found at https://www.epa.gov/criteria-air-pollutants/naaqs-table.

In this report, we analyze total reduced sulfur (TRS) emission rate data for recovery boilers collected under the New Source Performance Standards for Kraft Pulp Mills (40 CFR Part 60; Subparts BB and BBa). Hour-by-hour emission rate data were available for 11 recovery boilers at eight facilities. The data demonstrate the potential differences in modeled exposure estimates that would be solely attributable to using TRS emission rates calculated as the mean of all operating hours of the year, as opposed to a facility’s reported hourly emission rate. As a comparison between industries, we perform a similar analysis using reported hourly emission rates for sulfur dioxide (SO₂) and oxides of nitrogen (NO_x) from a large U.S. power plant (hourly emission rate monitoring of SO₂ and NO_x is a requirement for power plants under the EPA’s Acid Rain Program). Finally, we use information from the EPA Risk and Technology Review (RTR) program (EPA 1999, 2009) as an example of the lack of 1-hr HHRVs meant to be health protective in the general population.

Methods

Total reduced sulfur (TRS) measurements from pulp and paper mills

Source emission rates for toxic air pollutants are typically estimated by EPA from source category survey information or compiled in data sets such as the NEI. In most cases, these emission data are representative of an annual emission total (e.g., tons per year [TPY]). The NEI emission estimates are developed using data reported by state air agencies, which range from measured data, estimated data, and data supplemented with EPA estimates using emission factors and corresponding emission unit throughput (see https://www.epa.gov/air-emissions-inventories/national-emissions-inventory-nei).

Continuous emission monitoring (CEMS) of TRS is a requirement under the New Source Performance Standards for Kraft Pulp Mills (40 CFR Part 60; Subparts BB and BBa; EPA 1976). The raw TRS monitoring data used in this report were obtained from https://www.regulations.gov/document?D=EPA-HQ-OAR-2012-0640-0089 and modified as detailed below.

The continuous emission monitoring data used in this analysis included measurements at those facilities and operating units for which hour-by-hour TRS measurements were available. In total, there were 98 kraft pulp mills in the data set, and of these, eight facilities with 11 emission units met this requirement (Table 1). In contrast, 85 facilities provided 12-hr average emission estimates for TRS from 125 emission points (Table 1). One operating unit in the spreadsheet containing the raw data was labeled as only containing 12-hr data, but upon further inspection it was found to also contain hourly data and thus was included in the analysis (NEI9201). Because this analysis is meant to describe fluctuations in emissions that may occur during routine periods of operation, we removed hourly emission rates associated with start-up/shut-down, malfunction events, CEMS errors, or logging errors. Moreover, if there was no data entry associated with a particular hour of operation, that hour was also not used in this analysis. In total, operating hours associated with the emission units used in this analysis ranged from 4221 to 8493 hr, with all but one emission unit having greater than 7300 hr of normal operations (Table 2). Hourly TRS data corrected for percent oxygen were used in all cases. Notably, for facility NEI40686, corrected TRS data were not provided in the raw data but were calculated using the following formula: C_corr = C_meas × (21 − X/21 − Y), where C_corr is the concentration corrected for oxygen, C_meas is the measured concentration of TRS uncorrected for oxygen, X is the volumetric oxygen concentration in percentage to be corrected to (8% for recovery furnaces), and Y is the 12-hr average of the measured volumetric oxygen concentration. All data are from the year 2009 except for one facility, where the operating unit emission data were from 2011 (NEI41565).

Table 1. Temporal resolution of total reduced sulfur (TRS) emission rates from 98 kraft pulp and paper mill operations with 151 operating units.

Number of facilities	Number of units	TRS data temporal resolution
8	11	1-hr
85*	125*	12-hr
4	5	24-hr
18	21	No data provided

Notes.*Includes facilities with 1-hr data.

Table 2. Distribution of ratios of reported to average hourly TRS emission rates for 11 recovery boilers located at eight kraft pulp and paper mills.

Facility identification number^a	No. of operating hours	25th percentile	50th percentile	75th percentile	99th percentile	Max. ratio	Mean ± SD
NEI26471	8430	0.37	0.83	1.33	3.67	38.84	0.98 ± 0.96
NEI26504(i)	8087	0.64	0.91	1.23	3.11	6.71	1.0 ± 0.59
NEI26504(ii)	8001	0.64	0.91	1.23	2.92	10.59	0.99 ± 0.57
NEI26504(iii)	8186	0.53	0.85	1.22	4.14	14.95	1.0 ± 0.87
NEI41565	4221	0.58	0.58	0.86	7.20	163.98	1.0 ± 3.71
NEI45206	7509	0.45	0.63	1.14	5.71	11.92	0.99 ± 1.13
NEI46931(i)	7324	0.30	0.48	0.92	10.13	28.73	1.0 ± 1.86
NEI46931(ii)	8287	0.20	0.40	0.65	15.51	97.05	0.99 ± 4.27
NEI47074	8493	0.78	0.98	1.19	2.32	12.01	0.99 ± 0.46
NEI40686	8524	0.64	0.93	1.31	2.58	7.14	1.0 ± 0.53
NEI9201	8250	0.69	1.03	1.03	2.59	22.4	0.96 ± 0.75

Notes.^aNEI26504 and NEI46931 provided hourly emission data for multiple recovery boilers.

The average hourly emission rate for each emission unit was calculated by averaging all 1-hr reported emission rates included in the final data set. Hourly-to-average hourly emission ratios were then calculated by dividing measured hourly emission rate by the average hourly emission rate just obtained. Statistical distributions of these hourly-to-average hourly emission ratios were then calculated in order to describe the variations in emission ratios among different emission units within and across facilities (Table 2).

SO₂ and NO_x measurements from a large U.S. power plant

Hourly SO₂ and NO_x emission data were obtained for the Homer City power plant in Homer City, Pennsylvania, from EPA’s Clean Air Markets Data (https://ampd.epa.gov/ampd/). Monitoring of hourly SO₂ and NO_x emission data is a requirement under the EPA’s Acid Rain Program. The Homer City power plant was chosen because for the year 2014, it was the number one emitter of SO₂ and the number three emitter of NO_x in the United States—an indication of the plant’s size and output (https://ampd.epa.gov/ampd/). Hourly emissions for both SO₂ and NO_x were obtained for the same three operating units at the power plant. If there was no data entry associated with a particular hour of operation, that hour was not used in this analysis. Total hours examined in this analysis are presented in Results (Table 3), but all operating units for both SO₂ and NO_x had over 7000 valid operating hours. A similar analysis was performed for the New Madrid power plant (number five emitter of NO_x emissions in the United States in 2014) and is presented in the supplemental material. All data are from the year 2014.

Table 3. Comparison of maximum reported 1-hr SO₂ and NO_x emission rates with average hourly emission rates for three power generating units located at a large U.S. power plant.

		Max. hourly emissions		Mean hourly SO2 emissions (lbs/hr)		Max. ratio
Unit ID	Hours of data available	SO2	NOx	SO2	NOx	SO2	NOx
1	7914	25,253.0	3901.2	16,101.5	1760.0	1.6	2.2
2	7033	22,422.2	4282.5	15,564.7	1835.3	1.4	2.3
3	7538	7043.7	3538.6	1435.3	2307.9	4.9	1.5

For each operating unit, hourly-to-average hourly emission ratios for SO₂ and NO_x were calculated for each hour data were available. The resulting ratios were further analyzed to develop statistical distributions and to identify variations among emission units.

Review of reference values used in RTR assessments

Human health reference values (HHRVs) provide an essential tool to enable the transition from estimates of exposure to HAP chemicals and HAP-associated compounds in air to calculations of potential risks for chronic and acute noncancer health effects. In this study, we focused only on the acute values currently being used in the RTR program. These acute values are listed on an EPA Web site (https://www.epa.gov/fera/dose-response-assessment-assessing-health-risks-associated-exposure-hazardous-air-pollutants) along with context and criteria for inclusion of only the values deemed appropriate for RTR purposes. Note that hydrogen sulfide is listed on this Web site but is not a HAP and therefore not included in this analysis. In addition, emission levels for individual HAPs from the final 2014 National Emissions Inventory (NEI) were collated with the HAPs having available appropriate HHRVs to show the percentage of total HAP emissions covered by the available HHRVs.

Results and discussion

Reported hourly-to-average hourly TRS emission ratios from kraft pulp and paper mills

In order to demonstrate the difference between reported hourly and emission rates calculated as the mean of all operating hours of the year (i.e., average hourly emission rate), and its potential effect on exposure and risk assessment, 1-hr reported TRS emission measurements from 11 recovery furnaces located at kraft pulp mills were analyzed. Having hour-by-hour emissions information from industrial sources is not typical for most pollutants, but different aggregations of hourly TRS data were reported by the kraft paper industry in 2011 in response to a CAA section 114 information collection request (EPA 2011). Considering the 1-hr TRS emission data from these 11 recovery boilers over the course of a year, we calculated the majority of reported hourly-to-average hourly ratios to be less than 10 (Table 2 and Figure 1). However, relatively high TRS ratios did occur. The maximum reported hourly emission rate ranged from approximately 6.7 to 164 times the average hourly emission rate (Table 2). Ratios less than 1 were also found (Table 2 and Figure 1). Thus, the use of average hourly rather than reported hourly emission rates could result in an over- or underestimation of exposure concentrations and therefore health risk.

For the vast majority of toxic air pollutants, detailed hourly emission rate data like those presented above are unavailable. Part of a multistep strategy used by the EPA’s RTR program to account for the paucity of reported hourly air toxics emission rates when modeling ambient concentrations is to apply a multiplier value to average hourly emission rates (EPA 2009). The program uses a default emission rate multiplier of 10 when there is not facility-specific information to better inform the multiplier. The default factor of 10 is based on an analysis of 99th percentile hazardous air pollutant emission rates reported during short-term nonroutine events (e.g., malfunction) from facilities in the Houston-Galveston area (EPA 2009). Our analysis demonstrated that 99th percentile ratios for TRS from 10 out of 11 kraft pulp mill operating units were also approximately 10 or less (Table 2). That is, 99th percentile reported hourly-to-average hourly emission TRS ratios ranged from approximately 2.3 to 15.5 for this group of recovery furnaces (Table 2). However, it should be noted that whereas the Houston-Galveston analysis examined emission rates of a number of HAPs during short-term nonroutine events such as periods of malfunction, our analysis was specifically designed to examine emission rates of TRS (which is not a HAP and thus not analyzed under the RTR program) during routine operations. Thus, comparisons between the two data sets should be done with caution.

In addition to using a default multiplier of 10, risk assessments for the RTR program also employ a number of additional health protective assumptions when estimating hourly exposures (e.g., assuming a person is located at the site of maximum concentration for at least an hour). Thus, even if hourly-to-average hourly emission ratios are higher than the default factor of 10—as was found for a single facility in our analysis (NEI46931(ii))—the results of the RTR acute-risk assessment are still likely to be upper-bound estimates of exposure and health risk.

Reported hourly-to-average hourly SO₂ and NO_x emission ratios from a large U.S. power plant

Fluctuations in hourly emission rates are expected to be dependent on the pollutant being emitted as well as on the industry and the facilities in that industry being analyzed. To illustrate this point, we examined 1-hr emission rate data for SO₂ and NO_x from three large power generating units located at a large coal-fired power plant in the United States (hourly SO₂ and NO_x reporting is specifically required for many electric generating units to support the Acid Rain Program). Given the continuous nature of power generating operations at this facility (and at large power plants in general), it was expected that reported hourly emission rates would be relatively stable throughout the year. Indeed, we calculated most reported hourly-to-average hourly emission ratios for all units to be less than 2 for both SO₂ (Table 3 and Figure 2a) and NO_x (Table 3 and Figure 2b). Moreover, although we found that the absolute largest ratio is about 5 for SO₂ based on an isolated data point from unit 3, all other units for both SO₂ and NO_x had ratios of approximately 2.5 or less (Table 3). Conducting a similar analysis on a second large power plant yielded similar results for both SO₂ and NO_x (Figure S2 and Table S2).

Figure 2. Distribution of ratios of reported to average hourly (a) SO₂ emission rates and (b) NO_x emission rates for three power generating units located at a large U.S. power plant.

The power plant data presented above suggest that industries could provide information on whether their production processes are batch or continuous in nature to better inform the relationship between actual hourly and average hourly emission rates. That said, even more informative would be if an industry could provide limited CEM data so that an hourly emission profile could be generated. The amount of CEM data needed to establish a meaningful temporal profile would have to be the subject of future research, as it would likely be industry, chemical, and/or chemical family specific. However, this research would be a benefit to both risk assessors and regulated industrial facilities. Risk assessors would be able to provide more accurate estimations of hourly health risk. Industrial facilities could more easily avoid being evaluated using default emission multipliers, which in many cases could substantially overestimate hourly emissions and thus health risk. For example, if the RTR default factor of 10 was used in an hourly risk assessment of SO₂ and NO_x (which are not HAPs) at the power discussed above, it would result in an overestimation of acute health risk. In addition, limited monitoring for some chemicals would also be preferable to the technologically challenging and cost-prohibitive solution of providing hour-by-hour emission rates for every toxic chemical a facility emits. Moreover, even if only some facilities in a given category of sources (e.g., secondary lead smelters) provided this type of emission profile, the risk assessor or regulatory body could apply these profiles directly to sources that did not generate their own provided that they had similar production processes and control technologies. The risk assessor could also use additional information from sources without profiles (e.g., installed control technologies) in order to modify and apply an existing temporal profile. Either way, this data-driven methodology is likely to result in a more accurate estimate of hourly emission rates, modeled dispersion concentrations, and resulting health risk when compared with applying a simple default emission multiplier to every hour of the year.

In lieu of modeling, in some situations ambient monitoring could also be useful in estimating fluctuations in short-term concentrations of air toxicants. A recent study demonstrated increased concentrations of a number of toxic air pollutants at times when the community of interest was located downwind of a local pulp and paper mill (Hoffman et al. 2017). Although this study used a 24-hr average toxicant sampling time, shorter sampling times could potentially be used to determine hourly concentrations. However, ambient monitoring is time and resource intensive and may be impractical for simultaneously evaluating a large number of diverse air pollutants. Moreover, if the facility of interest is located in an area with other industrial facilities, source apportionment to the facility of interest can be difficult.

Availability of 1-hr HHRVs

Even in situations where hourly pollutant exposure concentrations can be modeled or monitored with confidence, there is still an important hurdle for generating meaningful hourly estimates of human health risk. This has to do with the comparison between the estimated exposure concentration generated via modeling or monitoring and the HHRV that it is being used for comparison. Particularly for HAP compounds, there is a lack of HHRVs intended to be used in risk assessments designed to estimate the potential for adverse health effects in the general population that includes sensitive groups such as children and older adults.

To illustrate the lack of 1-hr HHRVs available for use in a regulatory risk assessment, we reviewed the HHRVs used in the EPA RTR program (Table 4 and Figure 3). The RTR program estimates acute risk associated with hourly exposures in populations living around industrial sources and therefore typically prefers 1-hr HHRVs that are derived to be health protective in populations that include sensitive groups (EPA 2009). Moreover, in order to support a regulatory decision, HHRVs used in the RTR program are limited to those that have undergone the rigor of peer review and public comment.

Table 4. Acute inhalation human health reference values available for 303 HAP chemical and HAP-related compounds in the RTR program.

1-Hour human health reference values used in the RTR program	Number of HAP chemical or HAP-related compounds with reference value
California reference exposure level	42
Acute Exposure Guideline Level-1	50
Emergency Response Planning Guidelines-1	39
Acute Exposure Guideline Level-2	79
Emergency Response Planning Guidelines-2	59

Notes. https://www.epa.gov/fera/dose-response-assessment-assessing-health-risks-associated-exposure-hazardous-air-pollutants; some HAPs have more than one HHRV.

Figure 3. Coverage of human health reference values used in the EPA RTR program by (a) chemical and (b) percent emissions reported to NEI.

California reference exposure level (REL) values meet RTR program criteria. REL values represent 1-hr exposure concentrations below which no adverse health effect is expected, including in sensitive groups (California Environmental Protection Agency Office of Environmental Health Hazard Assessment [OEHHA] 2008, 2014) (Table S1). However, of the 303 HAP chemical and HAP-related compounds listed on the EPA Office of Air Quality Planning and Standard’s (OAQPS) air toxics dose-response value list (OAQPS 2017), only 42 (14%) of these chemicals/compounds have an REL value (Table 4). In the absence of an REL value, the RTR program may use a severity level 1 (mild, reversible effects) Acute Exposure Guideline Level (AEGL-1) (Department of Energy [DOE] 2016; NRC 2001) or Emergency Response Planning Guideline Level (ERPG-1) (American Industrial Hygiene Association [AIHA] 2002; DOE 2016) to estimate acute risk (Table S1). These other values are used even though they are meant to be used in emergency response situations, are derived for a once in lifetime exposure, and represent concentrations at which health effects are likely to occur. Moreover, it is important to note that exposure concentrations below the AEGL-1 can still result in mild adverse effects (EPA 2018) and that the elderly, sick, and very young are not covered by ERPG values (National Oceanic and Atmospheric Administration [NOAA] 2018). If these values are included, approximately 25% of HAP chemicals/compounds have either an REL, AEGL-1, or ERPG-1 (Figure 3a). If HHRVs representing severity level 2 effects (irreversible effects or impairment of the ability to escape exposure) are also included, then about 33% of these chemicals/compounds have an hourly reference value (Figure 3a).

Although a relatively small percentage of these 303 HAP chemicals and HAP-related compounds have 1-hr REL, AEGL, or ERPG values, those that do have values make up the vast majority of the estimated emissions for all HAP in the 2014 NEI. That is, chemicals/compounds that represent about 94% of HAP emissions have a 1-hr HHRV of severity 1 and 96% are represented by a 1-hr HHRV of severity 2 or less (Figure 3b). That being said, the toxicity associated with 1-hr exposures for the approximately 200 HAP chemicals or HAP-related compounds that make up the rest of these emissions are largely unknown. Thus, it is at least possible that a 1-hr exposure to a relatively small concentration to some of these HAP compounds could result in adverse health effects.

In lieu of 1-hr HHRVs for emergency response, risk assessors conducting assessments for the general population could potentially use occupational HHRVs. There are a number of HHRVs meant to protect healthy workers from occupational exposures of less than 1 hr, including the 30-min immediately dangerous to life and health (IDLH) values, the 15-min short-term limit values (STEL), as well other values that have been summarized more extensively elsewhere (EPA 2009; Woodall 2005; Woodall et al. 2017). However, similar to the use of emergency response HHRVs, occupational values such as these have obvious drawbacks for use in risk assessments applicable to populations that include sensitive groups. Occupational safety values are derived under the assumption that the person being exposed is a healthy worker with some level of choice to be in that setting, as opposed to member of an at-risk population such as a child or older adult breathing the air outside their home or school with few to no options to avoid exposure. In addition, occupational HHRVs are likely of little utility in risk assessments outside the occupational setting given that the HHRV is likely based on a concentration that is so high that a similar exposure concentration would not be experienced outside the workplace. Thus, if these values are used in a risk assessment for the general public, it could result in an underestimation of health risk that would be difficult to quantify.

Although we show the need for additional HHRVs in relation to 1-hr regulatory risk assessments, the need for these types of short-term health benchmarks is likely to increase in the future given the advancement of low-cost portable air sensors that are able to measure air pollutants in real time (Woodall et al., 2017). That is, it is very likely that citizens and advocacy groups will use these sensors to measure short-term air pollutant concentrations in their communities and then ask health and environmental agencies what the results could indicate from a health perspective. In instances such as these, short-term HHRVs will be valuable tools for federal, state, and local health and environmental agencies for evaluating a potential health risk. Thus, it would be beneficial to risk assessors and to the public if these entities placed additional resources toward the development of additional 1-hr HHRVs for toxic air pollutants.

Conclusion

We conclude that modeling pollutant exposures based on average hourly emission rates that have been calculated as the mean of all operating hours of the year, as opposed to the facility’s reported hourly emission rates, can result in a substantial over- or underestimation of potential health risk. We also conclude that there is a paucity of 1-hr HHRVs intended for use in the general population that includes sensitive groups such as children. These results suggest that further research is needed not only to more fully elucidate the relationship between reported hourly and average hourly emission rates for different categories of industrial sources, but also for developing additional 1-hr HHRVs intended for use in the general population. This research would lead to more accurate risk assessments based on 1-hr exposures and would therefore benefit both the risk assessor and the facilities being assessed.

Disclaimer

The views expressed in this paper are those of the authors and do not represent the official views or policies of the U.S. Environmental Protection Agency (EPA).

Additional information

Notes on contributors

Michael J. Stewart

Michael J. Stewart is a biologist and a risk assessor at the EPA.

James Hirtz

James Hirtz is a chemical engineer and risk assessor at the EPA.

George M. Woodall

George M. Woodall is a toxicologist at the EPA.

Chelsea A. Weitekamp

Chelsea A. Weitekamp is a postdoctoral fellow for Oak Ridge Institute for Science and Education, conducting research at the EPA.

Kelley Spence

Kelley Spence is a chemical engineer at the EPA.