The Systematic Underreporting of Toxic Air Emissions

A joint study by the Environmental Integrity Project and the Galveston-Houston Association for Smog Prevention

The U.S. Environmental Protection Agency (EPA) releases the results of the annual Toxic Release Inventory (TRI) each summer. The TRI has proved to be a powerful tool for raising consciousness about sources of toxic pollution and encouraging companies to act voluntarily to reduce their emissions. The official TRI, however, tells only part of the story because it dramatically underestimates the amount of toxic pollution from the petrochemical industry. The Texas Commission on Environmental Quality (TCEQ) has conducted studies which demonstrate the extent to which emissions of toxic chemicals from petrochemical facilities in Texas are underreported. This report applies the TCEQ's findings nationwide and reveals that emissions of toxic chemicals, including known carcinogens such as benzene and butadiene, are four to five times higher than is reflected in the TRI.

Federal and State environmental regulators have known for more than a decade that toxic air emissions are widely underreported. The primary reason for this problem is that most air emissions are not actually monitored. Instead, industrial facilities report their toxic emissions based on calculations that are often outdated and inaccurate. Rather than addressing the problem of systematic underreporting, the EPA has weakened monitoring requirements and continues to provide the public inaccurate data regarding toxic air emissions.

In 1984, an explosion at a chemical facility in Bhopal, India sent a toxic cloud into the neighboring community, killing thousands of people. This led many Americans to question the safety of industrial facilities in our own backyards and to call for stronger reporting of chemicals released into local communities. As a result, Congress passed a law which made industrial facilities report annually on certain toxic releases. This data is made available to the public in the Toxics Release Inventory (TRI).

The TRI is obviously not serving its purpose if information it provides the public is inaccurate. Numerous studies across the country have documented the underreporting of industrial emissions. Studies by the TCEQ have actually quantified the extent to which refineries and chemical plants in Texas underreport certain toxic emissions.¹ To gain an idea of how significantly the national TRI may underreport these emissions, we applied Texas' findings to the national inventory for 2001, the most recent year for which data is publicly available. Our study reveals that in 2001, if Texas' results are applied nationwide, refineries and chemical plants failed to report at least 330 million pounds of toxic hydrocarbon emissions. See Table 1.

While the Texas study looked only at a small subset of the chemicals reported to the TRI, applying the Texas results nationwide provides a glimpse of the startling magnitude of industry underreporting. Figure 1 compares the national reported to unreported emissions for the ten chemicals studied. It shows that releases of carcinogens such as benzene and butadiene may be four to five times higher than what is reported in the national TRI.

These findings are consistent with reports by the U.S. General Accounting Office and the EPA's Office of Inspector General, among others, which show that reported air emissions are often inaccurate and underestimated. It is time for EPA and the states to deal with the problem of inaccurate reporting of toxic releases. EPA should require more industrial sources to actually monitor their emissions. It should also improve emissions estimation methods for sources for which actual monitoring is not possible. The stakes for public health are too high for the government to continue to rely on data it knows to be inaccurate.

STUDY METHODOLOGY AND FINDINGS

Methodology

Numerous studies across the country have concluded that the refining and chemical manufacturing industries release significantly greater emissions than they report.² Texas was the first state to estimate the magnitude of the problem and to develop a system for adjusting reported emissions to more accurately reflect actual emissions. Texas measured ambient quantities of select hydrocarbons in the Houston area and compared ambient quantities to reported emissions of those hydrocarbons. Texas then identified the sources of the emissions and developed "adjustment factors" to account for underreporting.³ See Appendix D.

Texas officials limited their research to certain hydrocarbons believed to play a major role in causing rapid ozone formation in the Houston area. Ten of those hydrocarbons - ethylene, toluene, n-hexane, xylene, propylene, styrene, benzene, cyclohexane, ethylbenzene and 1-3 butadiene - are chemicals that are reported to the TRI. In this report, we adjusted the 2001 TRI chemical plant and refinery emissions for those ten hydrocarbons based on the Texas methodology. See Appendix D. Emissions were adjusted for only chemical plants and refineries in four Standard Industrial Codes (SICs).

It is likely that industrial sources are also underreporting many other toxic pollutant emissions. Studies similar to Texas', however, have not been conducted for the vast majority of the hundreds of pollutants reported to the TRI. This report, therefore, provides just a glimpse of a much broader problem. See Appendix E.

Study Results

Applying the Texas methodology to TRI emissions for the ten selected hydrocarbons dramatically increases the amounts of those chemicals known to be in the air. For example, according to company data reported to the 2001 TRI, nearly 6 million pounds of benzene were released into the nation's air. Adjusting the reported benzene emissions based on the Texas methodology shows the actual amount of benzene released to be more than 20 million pounds -- a 248 percent increase. Similarly, butadiene increases by 432 percent, ethylene by 417 percent, and propylene by 440 percent. See Table 1.

Obviously, adjusting the TRI data for only ten pollutants, which are emitted primarily by petrochemical facilities, moves states such as Texas and Louisiana, which contain most of the nation's petrochemical facilities, higher in the TRI state ranking. (County rankings also shift markedly.) Texas moves from number three to number one in terms of overall quantity of air emissions. Louisiana moves from number nine to number two. The adjusted data also indicate at least five million pounds of unreported emissions in Illinois, Iowa, Kentucky, Oklahoma, Mississippi, Pennsylvania and Ohio. See Appendices A and C.

Likewise, our analysis results in individual refineries and chemical plants moving up in the TRI facility rankings. As listed in Appendix B, 14 refineries and chemical plants which do not appear on a list of the 50 plants releasing the most air pollution based on industry-reported TRI data, appear on the top 50 list based on adjusted emissions. ⁴

The changes in the state, county and facility rankings demonstrate the magnitude of the underreported toxic air pollution problem. While sources at refineries and chemical plants, such as flares, cooling towers and leaks, have been identified by several studies as sources of underreported emissions, the fundamental problem with reliance on emission calculations, rather than actual monitoring, is far more widespread and could affect many additional air pollutants reported to the TRI.⁵

HEALTH EFFECTS OF UNDERREPORTED EMISSIONS

The ten pollutants studied by Texas and adjusted in this report are volatile organic compounds (VOCs). VOCs react in the air to form ozone.⁶ A number of the VOCs are also carcinogenic and otherwise toxic to humans. See Table 3. The actual health impacts of toxic air pollution releases depend on the duration and concentration of exposure. The concentration depends on both the amount of pollution released and local conditions, such as topography and weather. In addition, some of the harmful effects from VOCs are caused by secondary pollutants, such as ozone and formaldehyde, formed after the VOCs are released.

Short Term Exposure

VOCs react with other chemicals in the air to form ozone. Ozone can cause acute health reactions such as respiratory distress and eye irritation, often almost immediately upon exposure. Ozone reduces breathing capacity, which is especially serious in persons with respiratory disease. Exposure to ozone also increases a person's susceptibility to allergens (such as pollen), respiratory infections and the effects of other air pollutants. Among asthmatics, exposure to ozone is associated with increased emergency room visits, hospital admissions and deaths.⁷ Acute health effects associated with benzene, styrene and toluene include reproductive, developmental, respiratory, central nervous system and eye problems.⁸ In addition, VOC emissions can lead to the secondary formation of formaldehyde, a human carcinogen, and similar chemicals, which themselves cause acute health reactions.⁹

Long Term Exposure

Long-term exposure to the ten pollutants covered in this report is associated with serious health effects. Both benzene and 1,3-butadiene are carcinogens associated with cancers including leukemia. Benzene is ranked by EPA as one of two chemicals posing the greatest national cancer risk. Butadiene is listed by EPA as one of the two most significant probable carcinogens contributing to regional cancer risk.¹⁰ All ten pollutants are also associated with the risk of one or more non-cancer chronic diseases, especially respiratory and developmental diseases, as described in Table 2.

The lack of accurate information regarding toxic emissions means that the public may be exposed to unhealthy levels of toxics without their knowledge. It also means that states and local agencies may be developing their state implementation plans - local air pollution clean-up plans - based on incorrect emission estimates. The result may be time and money spent on plans that will not result in compliance with health-based air quality standards.

ENVIRONMENTAL PROTECTION AGENCY INACTION

Documentation of Problem

EPA and states have known for more than a decade that emissions are systematically underreported, but have not taken adequate steps to address the problem. As has been documented in numerous studies and reports, the root problem is the lack of adequate emissions monitoring. Instead of actually monitoring many air emissions, industry uses unreliable calculation methods for estimating emissions.

International studies since the early 1990s have shown that actual hydrocarbon emissions were underestimated in emissions inventories.¹² Likewise, a 1999 report published by the Minority Staff on the U.S. House of Representatives Committee on Government Reform found refineries vastly underreport leaks from valves, adding millions of pounds of harmful pollutants to the atmosphere each year, including over 80 million pounds of VOCs and over 15 million pounds of toxic pollutants.¹³

In 2001, the U.S. General Accounting Office (GAO) called on EPA to improve its oversight of emissions reporting from large facilities. The GAO study documented that only four percent of all emissions "determinations" used direct monitoring or testing. The other 96 percent were based on estimates calculated using emissions factors.¹⁴ Emissions factors were developed by EPA as a means of estimating the long-term average emissions for all facilities in a particular source category. These factors do not reflect the variations within a source category due to different processes, controls or operating systems at individual facilities. The factors, therefore, are often not accurate for calculating a particular facility's emissions.

EPA itself developed a rating system for the accuracy of its emission factors. As of 1999, EPA had rated seventy-five percent of its emission factors. Twenty-nine percent of the emission factors were rated average or above. Forty-six percent were rated below average or poor.¹⁵ Despite this poor rating, these emission factors are still the basis for many facilities' toxic emission estimates.

In 2004, the EPA's Office of Inspector General issued its own report regarding the agency's methods for calculating air toxic emissions. That report confirmed that toxic air emissions data submitted by the states is inconsistent and that EPA's emission factors for toxic emissions are not reliable.¹⁶

A number of state and local studies have likewise documented the problem of inaccurate air emission reporting. In addition to the Texas studies, the California Bay Area Air Quality Management District (BAAQMD), the University of California at Irvine (UCI), and the Mid-Atlantic Regional Air Management Association (MARAMA) have conducted studies of their own. The BAAQMD studies documented inadequacies in the reporting of emissions from pressure relief devices and from flares at petroleum refineries.¹⁷ The UCI study found ambient levels of alkane hydrocarbons in the Southwest to be higher than reported.¹⁸ The MARAMA study concluded that VOC pollution in Philadelphia area ambient air is greater than the emissions reported by industry. MARAMA found that emissions from refineries, in particular emissions from flares, cooling towers and non-routine operations, are likely underestimated.¹⁹

EPA's Response

In spite of this evidence, EPA has failed to improve monitoring and reporting of toxic air pollution. In fact, EPA has moved in the opposite direction and has weakened some federal monitoring requirements.

In 2000, EPA proposed regulations that would have standardized the types of toxics data gathered by the states, as well as the methods used to calculate emissions. Despite the fact, however, that seventeen out of the twenty-two state and local governments commenting on the proposed regulations favored toxics reporting requirements, the toxics provisions were dropped from the regulations.²⁰

Likewise, EPA established an Emission Inventory Improvement Program (EIIP), which was designed to develop standard procedures for collecting and reporting emissions data. The EIIP workgroup officials, however, decided to eliminate toxics emissions estimation from their scope of work. The EIIP is no longer active due to lack of funding.²¹

Recently, the EPA's Emissions Factors and Policy Applications Group (EFPAG) held a workshop to discuss the use of emission factors. Its survey of various stakeholders suggested that:

The EFPAG is expected to produce a "decision on options for further development" by April 2005, but does not have any specific goals for requiring improvements to emissions monitoring or reporting.²²

Finally, in 2004 EPA adopted new rules that actually weakened air emission reporting requirements.²³ Pursuant to Title V of the Clean Air Act, EPA's old rules required that major air pollution sources conduct monitoring sufficient to reveal whether or not the source was complying with federal pollution limits. This provision was used by states and EPA to add monitoring to Title V permits whenever additional monitoring was necessary to track facilities' compliance. In 2004, EPA revised these rules to only require monitoring that occurs more than once every five years. Such infrequent monitoring is clearly inadequate for tracking compliance and means that more sources will be using emission calculations and estimations, rather than actual monitoring, to report emissions. This is obviously a step in the wrong direction.

EPA has shirked its responsibility to provide the public with accurate information regarding toxic emissions. Overwhelming evidence indicates that EPA's emission factors are inaccurate for developing emission estimates. Additional real monitoring of air emissions sources is clearly needed. Yet, instead of improving monitoring requirements, EPA appears to be moving in the opposite direction by weakening residual monitoring requirements. Unfortunately, most states have done little to pick up the slack.²⁴

RECOMMENDATIONS

The primary purpose of the Toxics Release Inventory is to provide members of the public with information regarding toxic releases in their communities. This information is intended to "empower citizens, through information, to hold companies and local governments accountable in terms of how toxic chemicals are managed."²⁵ Instead, because EPA continues to knowingly allow industrial facilities to underreport toxic emissions, the public remains in the dark about the true extent of their exposure.

In order to fulfill its mandate to protect public health and the environment and to make the TRI the useful tool it is intended to be, EPA should take the following steps:

• EPA should amend its Title V regulations to clearly require that all major sources conduct monitoring sufficient to demonstrate whether or not they are in compliance their federal emission limits.
• EPA should prioritize review of state-issued Title V permits to ensure that adequate monitoring is required.
• EPA should set a schedule to re-examine its emission factors within twoyears. Priority should be placed on emissions factors for toxic chemicals and on those that are known to be unreliable. These include flares, fugitives and cooling towers at refineries and chemical plants.
• EPA should clarify that its emission factors should not be used in the permitting process (for determining permit applicability or emission limits) or for permit fee calculations. Instead, actual emissions estimates based on plant-specific data should be used.

Likewise, states should take independent action to ensure that state-issued Title V permits require adequate monitoring, and that emission factors are not the sole basis for emissions estimates used in other circumstances such as fee calculations.

Industry and the government have known for years that the calculation methods used to report most emissions are inaccurate. It is time to significantly increase the number of air pollution sources that are actually monitored and to improve emission calculation methods for those that are not. The public deserves to know the true extent of toxic pollution in the air.

APPENDIX A
*See pdf version
APPENDIX B: Toxic Air Pollution by Facility (pounds released in 2001)
*See pdf version
APPENDIX C: Toxic Air Pollution by County (pounds released in 2001)
*See pdf version

APPENDIX D: Study Methods

Adjustments to the US Environmental Protection Agency 2001 Toxics Release Inventory were based on the initial database released by the US EPA in 2001. Revisions made by the US EPA since its initial database release are not included so that the numbers could be compared with the EPA's 2001 TRI Data Release.

The emission adjustment was made for all facilities reporting a primary or secondary SIC code listed in Table D-1. These four SIC codes were selected based on a database analysis of the accounts listed in tables 6.2-1 and 6.2-2, "Complete Attainment Demonstration SIP for the Houston/Galveston Ozone Nonattainment Area" (Texas Commission on Environmental Quality, March 18, 2003). The listed SIC classifications are those for which Texas has identified major or moderate revisions to the highly reactive VOC emissions. No similar data are available from TCEQ for other reactive VOC emissions.

Where facilities or plants are counted in this report, the term "facilities" or "plants" refers to the number of reporting entities with non-zero air emissions. When placed in the context of "adjusted," the terms refer to the number of reporting entities whose data were adjusted.

Of the hundreds of chemicals reported to the Toxics Release Inventory, only ten were adjusted. The Texas Commission on Environmental Quality has estimated only two adjustment factors. One is used for the "highly reactive" VOCs and the other is for several dozen other VOCs. However, only ten of these VOCs are reported to the Toxics Release Inventory.

APPENDIX E:

Questions & Answers Regarding Study Methodology

Is it reasonable to adjust emissions from refineries and chemical plants outside Texas?
Studies in various parts of the country have confirmed that hydrocarbon emissions are underreported in other states. While the extent of underreporting may vary, a common problem is that industry reports are based on federal emissions factors that are known to be inaccurate. Until other states conduct studies like those done in Texas to actually quantify underreporting, the Texas data is the best available.

Is it reasonable to adjust emissions for plants in such a limited number of industrial classifications?
While the lack of emissions monitoring is a problem for many other types of facilities, there are not studies available quantifying the accuracy (or inaccuracy) of emissions inventories for other major sources of air toxics. For example, findings from Europe suggest that large storage tank leaks (particularly older tanks) are a major source of unreported emissions, but similar studies have not been conducted in the US that would allow the findings to be applied to bulk storage facilities. The findings in Texas suggest that a somewhat broader group of industrial facilities may be responsible for unreported hydrocarbon emissions, but these findings have not been validated with field studies. Further studies clearly need to be conducted so that the public can know the true extent of toxic air pollution.

Is it reasonable to apply uniform adjustments to individual plants?
On-site monitoring would make this report obsolete by providing useful plant-specific data. Unfortunately, the findings from Texas remain quite general. The findings in this report indicate the potential size and nature of a systematic problem with reporting, but cannot be directly translated into accurate emissions estimates for specific facilities. Some plants may use more actual monitoring and, therefore, have fewer problems with underreporting.

ENDNOTES

¹ The Texas studies include: (1) Estes, Mark, et al, "Analysis of Automated Gas Chromatograph Data from 1996-2001 to Determine VOCs with Largest Ozone Formation Potential" (Texas Commission on Environmental Quality, November 11, 2002); (2) Estes, Mark, et al, "Preliminary Emission Adjustment Factors Using Automated Gas Chromatography Data" (Texas Commission on Environmental Quality, November 5, 2002); (3) Smith, Jim, "HGB Modeling Update" (Texas Commission on Environmental Quality, April 1, 2004) and (4) Texas Commission on Environmental Quality, "Revisions to the State Implementation Plan (SIP) for the Control of Ozone Air Pollution Houston/Galveston/Brazoria Ozone Nonattainment Area" (Draft Appendix D, Tables D.4 and D.26, and Appendix GG, Table 2, May 26, 2004).

² See notes 12-19.

³ See note 1.

⁴ The broad-scale, uniform adjustment to reported emissions is the only practical method of illustrating the size of the problem with industry's self-reported data. The errors in reporting are not however uniform across all facilities. In studies examining just four hydrocarbon species (see note 1), Texas environmental officials found that refineries and chemical plants in some areas of the Houston region appeared to be releasing 13-14 times more pollution than they were reporting, while in a few other areas, the emissions reporting error was significantly smaller, on the order of 20-90%.

⁵ Reports documenting emissions estimation problems with flares, cooling towers and leaks include: (1) Environ International, "Measurement and Assessment of Equipment Leak Fugitives and Vent Emissions in Industrial Ethylene and Other Chemical Sources" (Texas Environmental Research Consortium, June 2003) and (2)Galveston-Houston Association for Smog Prevention, "Smoke in the Water: Air Pollution Hidden in the Water Vapor from Cooling Towers - Agencies Fail to Enforce Against Polluters" (February 2004). See also reports referenced in notes 17 and 19.

⁶ The relative contribution to ozone formation among the ten VOCs varies. Ethylene, propylene and butadiene are considered by Texas to be "highly reactive." For example, ethylene has the potential to form ten times more ozone than benzene.

⁷ Brunekreef, B and Holgate, S T, "Air Pollution and Health," Lancet (Oct 19, 2002).

⁸ California Environmental Protection Agency, "The Determination of Acute Reference Exposure Levels for Airborne Toxicants," March 1999.

⁹ See note 8.

¹⁰ EPA "National Air Toxics Assessment" at http://www.epa.gov/ttn/atw/nata/risksum.html. See also, California Environmental Protection Agency, "Toxicity Criteria Database - OEHHA Cancer Potency Values" (Office of Environmental Health Hazard Assessment, December 2002).

¹¹ "OEHHA-CREL" refers to the California EPA, Office of Environmental Health Hazard Assessment. Air Toxics Hot Spots Program Risk Assessment Guidelines, Part III: Technical Support Document "Determination of Noncancer Chronic Reference Exposure Levels," adopted and draft proposed Chronic Reference Exposure Levels (CRELs), http://www.oehha.ca.gov/air/chronic_rels/index.html, September 2002. "TRI" refers to the US EPA, Office of Pollution Prevention and Toxics. TRI Risk-Screening Environmental Indicators Version 2.0, Technical Appendix A - Available Toxicity Data for TRI Chemicals of the RSEI User's Manual, http://www.epa.gov/opptintr/rsei/, February 2002. These data were used as compiled by Environmental Defense at www.scorecard.org/chemical-profiles/def/rav_edf.html. For endpoints, see Scorecard at http://www.scorecard.org/health-effects/.

¹² Several studies in Europe have also concluded that emissions are under reported by a large factor. For instance, a 1993 study using aircraft measurements around the Rijnmond (The Netherlands) area suggested that emissions of ethane, propane, n-butane, i-butane, n-pentane, i-pentane, 1,1,1- trichloroethene, tetrachloromethane, tetrachloroethene were under reported by industrial sources. The study cited earlier work in support of its findings, going back as early as 1988. Michiel Roemer, "Aircraft Measurements around the Rijnmond Area" (TNO Institute of Environmental Sciences, Delft, The Netherlands, January 1993).

¹³ U.S. House of Representatives, Minority Staff, Special Investigations Division, Committee on Government Reform, "Oil Refineries Fail to Report Millions of Pounds of Harmful Emissions" (Prepared for Rep. Henry A. Waxman, November 10, 1999).

¹⁴ United States General Accounting Office, "Air Pollution: EPA Should Improve Oversight of Emissions Reporting by Large Facilities (GAO-01-46, April 2001).

¹⁵ See note 14.

¹⁶ United States Environmental Protection Agency, Office of Inspector General, "EPA's Method for Calculating Air Toxics Emissions for Reporting Results Needs Improvement" (Report No. 2004-P-00012, March 31, 2004).

¹⁷ Bay Area Air Quality Management District, "Technical Assessment Document: Further Study Measure 8, Flares" (Draft Revision 2, December 2002), "Technical Assessment Document: Further Study Measure 8, Pressure Relief Devices" (Draft Revision 2, December 2002), and "Proposed Regulation 12, Rule 11: Flare Monitoring at Petroleum Refineries" (Draft Staff Report, March 2003).

¹⁸ Katzenstein, Doezema, Simpson, Blake and Rowland, "Extensive Regional Atmospheric Hydrocarbon Pollution in the Southwestern United States" (August 2003).

¹⁹ Mid-Atlantic Regional Air Management Association, "Evaluating Petroleum Industry VOC Emissions in Delaware, New Jersey and Southeastern Pennsylvania" (October 2003).

²⁰ See note 16.

²¹ See note 16.

²² Driscoll, Tom, "Emissions Factors Program Fact Finding Survey" (US Environmental Protection Agency, Emissions Factors and Policy Applications Group (EFPAG), June 2004).

²³ 69 Fed. Reg. 3201 (Jan. 22, 2004).

²⁴ Most states have not acted to improve monitoring or reporting of toxic air pollution. Texas has adopted regulations requiring increased monitoring of cooling towers for leaks and better monitoring of the gases sent to flares, but only for equipment related to ethylene, propylene, butadiene and butenes. In addition, for the past two years, the Texas Commission on Environmental Quality has used a "top-down" emissions inventory for certain hydrocarbons rather than the unadjusted data submitted by industry. The California Bay Area Air Quality Management District has adopted a rule requiring better monitoring of flares and is considering rules adopting control requirements for flares. Bay Area Air Quality Management District, "Proposed Regulation 12, Rule 11: Flare Monitoring at Petroleum Refineries" (Draft Staff Report, March 2003).

²⁵ US Environmental Protection Agency website (www.epa.gov/tri/whatis.htm).