OFFICE OF INSPECTOR GENERAL
Catalyst for Improving the Environment
Evaluation Report
Improvements in Air Toxics
Emissions Data Needed to Conduct
Residual Risk Assessments
Report No. 08-P-0020
October 31, 2007
-------
Report Contributors:
Jim Hatfield
Michael Young
Kevin Good
Erica Hauck
Geoff Pierce
Abbreviations
AFS Air Facility Subsystem
CAA Clean Air Act
CEMS Continuous Emissions Monitoring System
CFR Code of Federal Regulations
ECMC Emission Calculation Method Code
EPA U.S. Environmental Protection Agency
GACT Generally Available Control Technology
GAO Government Accountability Office
HAP Hazardous Air Pollutant
IDEA Integrated Data for Enforcement Analysis
LDAR Leak Detection and Repair
MACT Maximum Achievable Control Technology
NEI National Emissions Inventory
NESHAP National Emission Standards for Hazardous Air Pollutants
OAQPS Office of Air Quality Planning and Standards
OECA Office of Enforcement and Compliance Assurance
OIG Office of Inspector General
PAH Polycyclic Aromatic Hydrocarbon
POM Polycyclic Organic Matter
TRI Toxics Release Inventory
VOC Volatile Organic Compound
Cover photos: Industries that represent the three MACT sources categories on which our
review concentrated are shown. These include, clockwise from top left:
an aluminum plant, a petroleum refinery, and a paper mill. (All photos
courtesy U.S. Department of Energy)
-------
U.S. Environmental Protection Agency
Office of Inspector General
At a Glance
08-P-0020
October 31, 2007
Why We Did This Review
The 1990 Clean Air Act
Amendments required the
U.S. Environmental Protection
Agency (EPA) to develop
maximum achievable control
technology (MACT) standards
to reduce air toxics emissions
from stationary sources. In
2004, EPA completed the last
of its MACT standards. We
conducted this evaluation to
assess the effectiveness of
those standards in reducing air
toxics emissions.
Background
EPA has issued 96 MACT
standards covering
174 different categories of
industrial sources of air toxics.
Now that the MACT standards
have been issued, EPA must
assess the public health risk
remaining after each MACT
standard is implemented. If
the risk from a MACT
category is "unacceptable,"
EPA must promulgate
additional regulations to
reduce air toxics emissions
from that category. Excess
exposure to air toxics can
increase one's risk of
developing cancer and other
serious ailments.
For further information,
contact our Office of
Congressional and Public
Liaison at (202) 566-2391.
To view the full report,
click on the following link:
www.epa.qov/oiq/reports/2008/
20071031-08-P-0020.pdf
Catalyst for Improving the Environment
Improvements in Air Toxics Emissions Data
Needed to Conduct Residual Risk Assessments
What We Found
EPA's National Emissions Inventory (NEI) data indicate an overall decline in air
toxic emissions concurrent with implementation of the MACT standards.
Although NEI data reliability is uncertain, it is reasonable to conclude that air
toxics emissions have decreased. Our work suggests that the MACT program has
played a role in these reductions.
EPA plans to use NEI data to assess the public health risk remaining from MACT
sources' air toxics emissions, but the reliability of NEI data for site-specific
emissions varies considerably. EPA has not established objectives to define an
acceptable level of quality for NEI data used in the residual risk process. EPA
guidance recommends that program offices develop data quality objectives for
using data in such decision-making processes. Given the uncertainties associated
with NEI data, EPA could over- or under-estimate public health risk from MACT
sources' emissions. Overstating risk could result in EPA placing on industries
regulations that are not cost beneficial. Conversely, understating risk could result
in EPA not requiring regulations where needed to protect public health.
In our March 2004 report on EPA's air toxics performance measures, we
recommended that EPA require State reporting of air toxics emissions data. EPA
has not implemented this recommendation, citing unclear statutory authority and
the belief that voluntary reporting can achieve this goal. However, such a
requirement could help EPA obtain more reliable and complete NEI data.
In December 2006, EPA presented its plan for conducting residual risk
assessments to EPA's Science Advisory Board. The Board's June 2007 report
recommended several actions to improve this process. These recommendations
included developing a framework for improving the NEI data and conducting an
analysis to determine the impact of data uncertainty on the risk assessments. In
March 2007, EPA solicited public comment on the NEI and other data it plans to
use for conducting residual risk assessments.
What We Recommend
We recommend that EPA develop data quality objectives for using NEI data in
conducting residual risk assessments, and establish requirements for State
reporting of air toxics emissions data and compliance monitoring information.
EPA disagreed with our recommendations, but stated that it had activities
underway to improve the NEI data. However, EPA's planned actions do not
sufficiently address the problems identified, and we consider the issues unresolved.
-------
4-
MEMORANDUM
SUBJECT:
FROM:
TO:
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
OFFICE OF
INSPECTOR GENERAL
October 31,2007
Improvements in Air Toxics Emissions Data Needed to Conduct
Residual Risk Assessments
Report No. 08-P-0020
WadeT. Najjum
Assistant Inspector General for Program Evaluation
Robert J. Meyers
Principal Deputy Assistant Administrator for Air and Radiation
This is our report on the subject evaluation conducted by the Office of Inspector General (OIG)
of the U.S. Environmental Protection Agency (EPA). This report contains findings that describe
the problems the OIG has identified and the corrective actions the OIG recommends. This report
represents the opinion of the OIG and does not necessarily represent the final EPA position.
Final determinations on matters in this report will be made by EPA managers in accordance with
established audit resolution procedures.
The estimated cost of this report - calculated by multiplying the project's staff days by the
applicable daily full cost billing rates in effect at the time - is $1,288,566.
Action Required
In accordance with EPA Manual 2750, you are required to provide a written response to this
report within 90 calendar days. You should include a corrective action plan for agreed upon
action, including milestones. This report will be available at http://www.epa.gov/oig.
If you or your staff have any questions regarding this report, please contact me at (202) 566-0827
or najjum.wade@epa.gov: or Rick Beusse, Product Line Director, at (919) 541-5747 or
beusse.rick@epa.gov.
-------
Improvements in Air Toxics Emissions Data Needed to Conduct
Residual Risk Assessments
Table of Contents
Chapters
1 Introduction 1
Purpose 1
Background 1
Noteworthy Achievements 6
Scope and Methodology 6
2 EPA Inventory Data Show General Decrease in Air Toxics Emissions 7
Available Data Has Limitations 7
NEI Data Suggest Air Toxics Emissions Have Decreased
Since MACT Program Implemented 7
Air Toxics Ambient Monitoring Data Indicate
General Reduction in Emissions 13
External Factors and Changes in Emissions Estimation
Methods Impact Air Toxics Emissions Reported 13
EPA and Some State Compliance Monitoring Activities
Should Result in Emissions Reductions 14
Key Compliance Monitoring Information Lacking
to Evaluate MACT Effectiveness 14
Conclusions 16
Agency Comments and OIG Evaluation 16
3 Air Toxics Inventory Data Need to Be Improved
Before Being Used in Residual Risk Assessments 17
Reliability of NEI Data is Uncertain 17
EPA and Other Studies Have Noted Uncertainties
with Air Toxics Data 19
Data Quality Objectives Have Not Been Established 20
No Reporting Requirement for Air Toxics Emissions Data 21
Use of NEI Inventory Data Could Result in
Unreliable Residual Risk Assessments 22
Conclusions 23
Recommendations 23
Agency Comments and OIG Evaluation 23
Status of Recommendations and Potential Monetary Benefits 26
- continued -
-------
Improvements in Air Toxics Emissions Data Needed to Conduct
Residual Risk Assessments
Appendices
A Details on Scope and Methodology 27
B Amount and Percent of Change in NEI Air Toxics Emissions
Between 1990 and 2002 32
C NEI Emissions Reductions Compared to
Federal Register Anticipated Reductions 35
D Detailed Results - Petroleum Refinery MACT 37
E Detailed Results -Primary Aluminum MACT 45
F Detailed Results - Pulp and Paper MACT 52
G MACT Compliance Monitoring, Enforcement, and
Compliance Assistance Activities 57
H Compliance Assistance Tools Developed by EPA
for Three Sample MACTs 61
I Agency Response to the Draft Report 63
J Distribution 66
-------
Chapter 1
Introduction
Purpose
The 1990 Amendments to the Clean Air Act (CAA) required the U.S.
Environmental Protection Agency (EPA) to develop and issue maximum
achievable control technology (MACT) standards to reduce emissions of air
toxics from certain categories of stationary sources. EPA issued the last of its
MACT standards in 2004. EPA anticipates that when fully implemented in 2007,
MACT standards will reduce air toxics emissions from stationary sources by
1.7 million tons per year. We conducted this evaluation to assess the progress of
MACT standards in reducing air toxics emissions. Specifically, we sought to
determine:
1. To what extent MACT standards have reduced emissions of air toxics and
achieved the emission reductions anticipated when the rules were developed?
2. What factors are associated with the effectiveness of MACT standards in
achieving emission reductions, including:
• What factors external to MACT standards have impacted changes in
emissions?
• Is there a relationship between the quantity and quality of compliance
monitoring and enforcement and the effectiveness of MACT standards in
achieving the desired emissions reductions?
• Is there a relationship between the adequacy of compliance assistance
tools and the effectiveness of MACT standards in achieving the desired
emissions reductions?
Background
Toxic air pollutants are those pollutants known or suspected to cause cancer or
other serious health effects. The 1990 CAA Amendments established a list of 189
air toxics that EPA is required to control. Since 1990, EPA has revised the list
slightly and currently regulates 187 air toxics. These pollutants are also referred
to as hazardous air pollutants (HAPs). Congress selected these toxic air pollutants
on the basis of potential health and/or environmental hazard.
EPA's latest National Air Toxics Assessment estimated that in 1999 more than
270 million people lived in U.S. census tracts where the combined upper bound
lifetime cancer risk from air toxics exceeded 10 in 1,000,000. People who live
-------
near certain major industrial plants may face even higher cancer risks from air
toxics. The following table shows the sources of air toxics emissions:
Table 1.1: Sources of Air Toxics Emissions
Source
Large
Stationary
or Major
Area
Mobile
Description
Any building, structure, facility, or installation that emits or may emit any air pollutant.
Large or major stationary sources are those that emit 1 0 tons or more per year of any
of the listed toxic air pollutants, or 25 tons or more per year of a combination of air
toxics.
Stationary sources that emit less than 1 0 tons per year of a single air toxic, or less
than 25 tons per year of a combination of air toxics. While emissions from individual
area sources are often relatively small, collectively their emissions can be of concern,
particularly where large numbers of sources are located in heavily populated areas.
Includes cars and light trucks, heavy trucks and buses, nonroad recreational vehicles
(such as dirt bikes and snowmobiles), farm and construction machines, lawn and
garden equipment, marine engines, aircraft, and locomotives.
Source: Developed by Office of Inspector General (OIG) staff based on information from EPA
According to the most recent National Air Toxics Assessment, major stationary
sources, which are the principal sources regulated by the MACT program,
account for approximately 11 percent of the average national cancer risk from
exposure to air toxics. Examples of toxic air pollutants include benzene, which is
emitted from petroleum refineries and other sources; hydrogen fluoride, which is
emitted from primary aluminum facilities and other sources; and formaldehyde,1
which is emitted from pulp and paper mills as well as petroleum refineries and
other sources.
MACT Is First of Two Phases to Reduce Air Toxics Risks
To control air toxics emissions from major stationary sources, the CAA requires a
two-phased approach. The first phase requires EPA to develop emission
standards, referred to as MACT standards, for sources that emit the air toxics
identified in the CAA. To identify those sources requiring emission standards, the
CAA requires EPA to list the categories of major and area stationary sources of
air toxics emissions. In 1992, EPA published the initial list of air toxics source
categories requiring emissions standards. This list included both major and area
source categories, but the majority were major source categories.2 Based on that
initial list and subsequent revisions, EPA has promulgated 96 MACT standards
covering 174 different source categories. The second phase is commonly referred
to as the residual risk and technology review program. This phase requires EPA
to assess the public health risk remaining after implementation of the MACT
1 EPA staff informed us that in the last 3-4 years the toxicity risk for formaldehyde has been lowered, and is no
longer considered a key HAP. However, controversy remains over this action.
2 EPA has identified additional categories of area sources that present health risks in urban areas, as required by the
CAA, that were not included in this initial list. On November 22, 2002, EPA issued the final list of 70 source
categories for regulation under the area source program as required under CAA sections 112(c)(3) and
112(k)(3)(B)(ii). EPA is under court order to complete regulations for these area source categories.
-------
standards for both major and area sources.3 This second phase also requires that
EPA review the technology-basis of the major and area source standards for
advancement in technology and develop additional standards if needed. Table 1.2
describes the two phases and their implementation status as of August 2007.
Table 1.2: Two Phases for Reducing Air Toxics Emissions and Risks from Stationary Sources
Description
Phase 1
Phase 2
EPA was required to set technology-based MACT
standards for major sources of air toxics. These
standards were to reflect, at a minimum, the level of
emissions that the best performing 12 percent of
sources in the category were achieving in practice.
EPA was also required to set technology-based
standards for area sources. These could be either
MACT (setting standards as least as stringent as the
best performing 12 percent) or could be based on
generally available control technology (GACT).
EPA must promulgate stricter standards if the original
MACT standard is not sufficiently protective of human
health. These standards are referred to as residual
risk standards, since they address the remaining or
residual risk after implementation of the original
MACT. Section 112 requires these stricter standards
if the MACT standard has not reduced excess lifetime
cancer risk to the most exposed individual to less
than one in 1,000,000. The CAA requires that EPA
establish these standards within 8 years after
promulgation of the MACT.
In addition, section 112 requires EPA to review, and if
necessary, revise both MACT and GACT standards
for both major and area sources considering
advancements in control technologies. The CAA
requires that EPA conduct these technology reviews
every 8 years after promulgation of the MACT/GACT
standards.
Status
EPA promulgated the last of
its 96 MACT standards in
September 2004.
EPA promulgated 27 GACT
standards as of July 2007
Ongoing. As of June 2007,
EPA had completed eight
residual risk and technology
review standards and had
proposed risk and
technology review standards
for petroleum refining in
September 2007. EPA has
begun a new streamlined
process for developing
standards, called Risk and
Technology Review. EPA
expects to propose six
additional standards by the
end of the calendar year
Source: Developed by OIG staff based on information from EPA
MACT standards generally establish emission limits for the air toxics of concern
emitted from the covered sources. These limits are based on the emission
reductions achieved by the highest performing sources in that source category.
However, not all MACT standards include emission limits. The MACT standards
for different source categories can vary greatly in the number and complexity of
requirements. Regardless of the standard's specific requirements, each MACT
has an effective date of compliance by which all covered facilities must comply
These residual risk assessments are not required where EPA set a generally available control technology standard
instead of a MACT standard for area sources, as provided by CAA section 112(d)(5).
-------
with the MACT's requirements. By the end of 2007, all 96 MACT standards
would have reached their effective date of compliance.4
Eight years after each MACT standard is promulgated, EPA is required to assess
the public health risks remaining from that source category's air toxics emissions.
The CAA stipulates that EPA must consider the excess lifetime cancer risks to the
individual most exposed to HAP emissions from a facility in a specific source
category. If, after implementation, EPA finds that a MACT standard is not
sufficiently protective of health, the Agency must issue additional standards that
provide an "ample margin of safety" to protect public health.
EPA's approach to determine whether a MACT standard is sufficiently protective
of health considers risk and other factors. A lifetime cancer risk of greater than
1 in 1,000,000 is considered a trigger point requiring more detailed analysis to
determine whether additional reductions are required to provide an "ample margin
of safety." EPA considers cost, technical feasibility, and other factors in deciding
whether a risk greater than 1 in 1,000,000 provides an ample margin of safety.
However, a risk greater than 1 in 10,000 is generally considered as not providing
an ample margin of safety and would require EPA actions to reduce that risk.
Table 1.3 shows EPA's decision approach to conducting residual risk
assessments:
Table 1.3: Decision Making Process for Residual Risk
Maximum Individual Risk =
1 in 1,000,000 or Less
Maximum Individual Risk =
Between 1 in 1,000,000 and
1 in 10,000
Maximum Individual Risk:
1 in 10,000 or Greater
"Ample margin of safety" is
met. No additional action is
needed.
Costs, technical feasibility, and
other factors are considered in
determining whether additional
actions are required.
Standard is generally not
considered sufficiently
protective of public health,
and additional actions are
needed to reduce risk.
Source: Developed by OIG staff based on information from EPA
EPA is behind in conducting residual risk reviews. In addition, EPA is behind in
conducting technology reviews, required every 8 years after promulgation of
MACT and GACT standards for both major and area sources. The Government
Accountability Office (GAO) recently found5 that because EPA issued most of
the MACT standards behind schedule, the residual risk reviews "have been
pushed back commensurately, thereby delaying any additional public health
protection that these reviews may provide." As of June 2007, EPA had completed
eight residual risk standards. These standards cover 14 of the 174 MACT source
categories for which EPA is required to make a residual risk determination.
4 Three EPA MACT standards were vacated by court decisions. As a result of these court decisions, it is now
incumbent upon the States to develop emission standards for these three categories and incorporate these standards
and compliance dates into State-issued operating permits for the applicable facilities.
5 EPA Should Improve the Management of its Air Toxics Program, Report No. GAO-06-669, June 26, 2006.
-------
Given resource constraints and pending court orders to complete residual risk
reviews, EPA has begun a streamlined approach to analyze risks from large groups
of source categories at once. In addition, EPA is combining the residual risk reviews
with technology reviews for each MACT standard. This combined, streamlined
approach is referred to as Risk and Technology Review. This approach is discussed
in more detail in Chapter 3.
State and Local Agency Roles in MACT Implementation
Although EPA developed the MACT standards, EPA has generally delegated
implementation of the standards to the States. Implementation activities include:
• Incorporating the standards into facility operating permits.
• Providing compliance assistance to regulated facilities.
• Maintaining required records.
• Monitoring compliance.
• Taking enforcement actions against facilities that violate the standard.
Once a standard has been delegated, the State becomes the primary agency for
implementation. However, this delegation does not prohibit EPA from enforcing
any applicable emission standard or requirement of the MACT.
Three types of activities considered crucial to successful implementation of the
MACT standards are compliance assistance, compliance monitoring, and
enforcement actions.
• Compliance assistance is help provided by regulatory agencies to enhance
a facility's understanding of the regulation and increase compliance.
Examples include training sessions and workshops, plain language guides,
compliance checklists, and applicability flowcharts.
• Compliance monitoring consists of activities conducted by regulatory
agencies to determine whether a facility is complying with environmental
regulations. These activities include full or partial compliance evaluations
conducted by qualified inspectors, and review of information submitted by
the facility, such as monitoring reports. A full compliance evaluation
includes:
o A review of all required reports and the underlying records.
o An assessment of air pollution control devices and operating
conditions.
o Observing visible emissions, as appropriate.
o A review of facility records and operating logs.
o An assessment of process parameters, such as feed rates, raw
material compositions, and process rates.
o A stack test if there is no other way to determine compliance with
the emission limits.
-------
Enforcement Actions consist of actions taken to compel compliance with
an environmental statute (such as the CAA), Federal rules and regulations,
or federally-enforceable State rules and regulations. Enforcement actions
may consist of civil or criminal judicial enforcement proceedings taken as
a result of an identified violation.
Noteworthy Achievements
To improve its use in measuring air toxics trends and progress, as well as improve
its use for air quality modeling, EPA conducted an extensive revision of the 1990
National Toxics Inventory baseline. The original baseline was comprised of
county-level data, but EPA revised the baseline when it created the 1990 National
Emissions Inventory (NEI) for air toxics, including providing individual facility
point source data. EPA has made other changes to improve the 1990 NEI
baseline, including adding missing MACT source categories and updating
selected non-point emission estimates and emissions allocation procedures.
Scope and Methodology
We conducted our field work from August 2005 to June 2007 at the EPA Office
of Air and Radiation's Office of Air Quality Planning and Standards (OAQPS) in
Research Triangle Park, North Carolina; and the EPA Office of Enforcement and
Compliance Assurance (OECA) in Washington, DC. We also interviewed
managers and collected data from five EPA regions, seven State agencies, and
nine MACT-regulated facilities.
We selected three MACT source categories for in-depth analysis from the
universe of MACTs with compliance dates prior to 2002. These three were the
Petroleum Refinery MACT, the Primary Aluminum MACT, and the Pulp and
Paper MACT I & III. For these three MACT categories, we compared 1990 and
2002 NEI emissions of the specific air toxics targeted by the MACT standard. To
address the factors associated with the effectiveness of the MACT standards in
achieving emission reductions, we selected three facilities from each of the three
MACT categories above for additional analyses. We selected these nine facilities
based on emissions of certain pollutants of concern contained in the 2002 NEI
Version 1.
We conducted this program evaluation in accordance with generally accepted
government auditing standards. Those standards require that we plan and perform
the evaluation to obtain sufficient, appropriate evidence to provide a reasonable
basis for our findings and conclusions based on our evaluation objectives. We
believe that the evidence obtained provides a reasonable basis for our findings and
conclusions based on our evaluation objectives.
Appendix A describes our scope and methodology in more detail, including
information on prior reports, data limitations, and management control review.
-------
Chapter 2
EPA Inventory Data Show General Decrease
in Air Toxics Emissions
EPA data indicate an overall decline in air toxic emissions concurrent with the
implementation of the MACT standards. Several factors may have contributed to
this decline. These include the requirements of the MACT standards as well as
external factors, such as industry trends or other CAA regulations. Some of the
decline could also be due to changes in emission calculation methodologies rather
than actual changes in emissions. Based on the data available, we were unable to
determine the extent to which MACT standards caused emission reductions.
Available Data Has Limitations
While we believe EPA's data can be used to conclude that air toxics emissions are
generally declining, we do not attest to the accuracy of the specific emission totals
presented in our analyses. EPA's air toxics emission inventory is based on
emission estimates developed using methodologies that have varying degrees of
reliability. Emissions data obtained from continuous emissions monitoring or
source-specific emission tests are generally the preferred method for estimating a
source's emissions. Other estimation methodologies, such as emissions factors,
are less reliable than source monitoring.
EPA has not quantified the uncertainty associated with the various estimation
methodologies used to compile the air toxics emissions inventory. Further, the
methodology used to develop facility-specific emission estimates can change over
time, resulting in a reported emission reduction being the result of changed
methodology rather than an actual decrease in emissions. Chapter 3 discusses the
uncertainties and limitations of EPA's air toxics inventory data, and EPA's plans
to use this data to conduct residual risk assessments.
NEI Data Suggest Air Toxics Emissions Have Decreased
Since MACT Program Implemented
We conducted a number of analyses comparing 1990 and 2002 NEI emissions.
Based on our analysis of NEI data pertaining only to MACT standards, we
determined that from 1990 to 2002:
• Total air toxics emissions from stationary sources declined by
approximately 1.7 million tons.
• Emissions of the six air toxics designated by EPA as presenting the
highest cancer risk nationwide have declined.
-------
• Overall air toxics emissions declined for the 27 MACTs with compliance
dates prior to 2002. Further, the percentage reductions in emissions for
16 of these 27 MACTs met or exceeded the percentage reductions
anticipated by EPA when it developed the standards.
• Emissions of the primary air toxics targeted by the petroleum refinery,
primary aluminum, and pulp and paper MACTs declined by 93 percent,
37 percent, and 90 percent, respectively.
These downward trends suggest that the MACT standards have helped to reduce
air toxics emissions. Based on our interviews with EPA staff, State
environmental staff and managers, and industry representatives, other factors also
played a role in reducing air toxics emissions. These other factors, which include
decreases in production and air rules other than the MACT standards, are
discussed later in this chapter. Details on the emissions trends we analyzed
follow.
NEI Overall Air Toxics Emissions Declined Between 1990 and 2002
According to the NEI, total air toxics emissions from stationary sources (both
major and area) assigned to a MACT category declined by approximately
1.7 million tons, or 70 percent. In computing the total emissions for any given
MACT category, we only included emissions of specific air toxics or air toxic
categories reported in both 1990 and 2002 for that MACT code. Figure 2.1
illustrates the change.
Figure 2.1: Air Toxics Emissions Assigned to a MACT Code, 1990 - 2002
y enn nnn -,
o nnn nnn
"tn
c
o
** 1 inn nnn
in
c
O -I nnn nnn
in
I
cnn nnn
1990
2002
Source: Developed by OIG staff from NEI data
Appendix B presents the results of our analysis for each MACT category.
-------
NEI Emissions of Key Air Toxics Declined Between 1990 and 2002
EPA's OAQPS had identified six key air toxics, listed in Table 2.1, that present
the highest average cancer and non-cancer risks to the public nationwide. The
key air toxics are a subset of the 33 Urban Air Toxics and reflect air toxics posing
high public health risks.
We compared emissions of these six key air toxics for the 27 MACT standards
with compliance dates prior to 2002. This comparison indicated that emissions of
the six key air toxics decreased by significant percentages. Aggregate emissions
of the key air toxics from these source categories declined by 83 percent over this
period. Table 2.1 shows the emissions changes for the six key pollutants.
Table 2.1: Emissions of Key Air Toxics from 1990 to 2002 for MACTs
with Compliance Dates Prior to 2002
Key Pollutant
Acrolein
Arsenic3
Benzene
1,3-Butadiene
Chromium Vlb
Formaldehyde0
Totals d
1990 NEI
Emissions
(Tons per year)
822.8
40.9
35,301.5
14,891.7
251.6
17,268.4
68,577.0
2002 NEI
Emissions
(Tons per year)
100.1
5.2
10,083.4
606.1
32.4
1,102.5
11,929.8
Change
(2002-1990)
(Tons per year)
-722.7
-35.7
-25,218.2
-14,285.6
-219.1
-16,165.9
-56,647.2
Change
-88%
-87%
-71%
-96%
-87%
-94%
-83%
= Total includes all arsenic compounds.
= Total includes all chromium compounds. Chromium VI is a subset of this total.
c = There is currently a great degree of uncertainty as to the cancer risk posed from inhalation of formaldehyde.
OAQPS is currently supporting a cancer potency that would indicate the cancer risks from formaldehyde are
not expected to be one of the "key air toxics" as noted above. EPA continues to study the effects of this
pollutant and is expecting a resolution on this uncertainty in the near future.
d = Totals may not equal due to rounding.
Source: Developed by OIG staff from the NEI databases
NEI Air Toxics Emissions Declined for MACTs with
Compliance Dates Prior to 2002
For each of the 27 MACTs with compliance dates prior to 2002, we compared
emissions reductions as indicated by the NEI to the emissions reductions
anticipated when the MACT standard was developed. The 2002 inventory should
represent emissions after the MACT requirements went into effect for these
MACTs. For these 27 MACTs, the NEI data show a decline in air toxics
emissions of over 1.5 million tons from 1990 to 2002, or 85 percent, for those
pollutants recorded in both years' inventories. Further, 16 of these 27 MACTs
achieved the emissions reductions that EPA anticipated when the standard was
developed.
When developing the MACT standards, EPA normally estimated the emission
reductions expected to occur as a result of the MACT. The anticipated
reductions, along with the baseline emissions for that MACT category, were
-------
generally published in the Federal Register notice or supporting documents for the
rule. Table 2.2 presents the results of our review, while Appendix C provides a
detailed comparison for each of these 27 MACT standards.
Table 2.2: Status of Anticipated Emissions Reduction Achievement for MACTs with
Compliance Dates Prior to 2002
Status of Anticipated
Reduction Achievement
Anticipated Reductions Achieved
Anticipated Reductions Not Achieved
Unable to Determine if Anticipated Reductions Achieved
Total
No. of
MACT
Standards
16
6
5
27
Percent of
MACT
Standards
59%
22%
19%
100.0%
Source: Developed by OIG staff from NEI databases
NEI Emissions of Targeted Air Toxics Declined for Three Sample
MACTs between 1990 and 2002
We selected the following three MACT standards for a more in-depth analysis of
emissions changes:
• Petroleum Refinery MACT (Subpart CC),
• Primary Aluminum MACT (Subpart LL), and
• Pulp and Paper MACT I & III (Subpart S).
NEI data indicated that emissions from two of these three MACT categories have
declined since implementation of the MACT standard. However, EPA has not yet
conducted technology reviews or residual risk assessments for these three
MACTs. The residual risk assessments will determine whether the MACT
standards are sufficiently protective of public health. For each of the three
MACTs, we reviewed changes in emissions of those air toxics specifically
targeted by the MACT standard. The results of our analyses follow.
NEI-Reported Emissions for Petroleum Refineries Declined Significantly.
NEI data show a significant decline in emissions from the petroleum refinery
sector. Industry representatives and EPA staff attributed this decline to the
numerous CAA regulations, in addition to the MACT, applicable to this
industry. OECA staff also told us that the new source review enforcement
initiative resulted in decreased emissions for this industry sector.
The NEI data indicated that total air toxics emissions decreased by over
90,000 tons, or 91 percent, from 1990 to 2002. This exceeded the 59 percent
reduction EPA anticipated when the rule was developed. Further, the NEI
data indicated that emissions of the primary pollutants targeted by the MACT
also declined. Figure 2.2 depicts the emissions of targeted pollutants between
1990 and 2002. Appendix D provides more information on the Petroleum
Refinery MACT.
10
-------
Figure 2.2: Emissions of Targeted Air Toxics for Petroleum Refinery MACT, 1990-2002
25,000
<£ 20,000
| 15,000
o
«> 10,000
5,000
Source: Developed by OIG staff from NEI databases
NEI Emissions of Pollutants of Most Concern from Primary Aluminum
Production Have Declined. The NEI data indicated that total air toxics
emissions increased for the primary aluminum sector because of increases in
emissions of carbonyl sulfide. However, emissions of poly cyclic aromatic
hydrocarbons (specifically PAH-7 and PAH-15) and hydrogen fluoride, which
EPA considers the pollutants of most concern from primary aluminum
facilities, all declined. According to the NEI, total emissions of these three air
toxics declined by 37.4 percent from 1990 to 2002, which did not meet the
50-percent reduction anticipated when the MACT was developed. OAQPS
told us that based on industry comments received in response to its Advance
Notice of Proposed Rulemaking, it plans to revise the 2002 NEI emission
estimates for this sector. Figure 2.3 depicts the emissions of targeted
pollutants between 1990 and 2002. Appendix E provides more information on
the Primary Aluminum MACT.
11
-------
Figure 2.3: Emissions of Targeted Air Toxics for the Primary Aluminum MACT
2,500
^ 2,000
i/)
£ 1,500
o
w 1,000
500
0
nt
Hydrogen Fluoride
PAH 15
PAH 7
Source: Developed by OIG staff from NEI databases.
NEI Emissions from Pulp and Paper Have Declined. The NEI data
indicated that total air toxics emissions decreased by 192,197 tons, or
90 percent. This overall reduction of 90 percent exceeded the 58 percent
reduction EPA anticipated when the MACT standard was developed. Further,
the data indicate that emissions of the primary pollutants targeted by the
MACT also declined. For example, emissions of formaldehyde, which is a
pollutant of concern from pulp and paper mills, declined from 4,124 tons in
1990 to 196 tons in 2002. The following figure depicts the emissions of
targeted pollutants between 1990 and 2002. Appendix F provides more
information on the Pulp and Paper MACT.
Figure 2.4: Emissions of Targeted Air Toxics for the Pulp and Paper MACT, 1990 - 2002
160,000n
140,000
£ 120,000
:§. 100,000
c 80,000
o
w 60,000
E 40,000
20,000
•
II rLniffi^^i^^ — _
• 1990 NEI
• 2002 NEI
N/V
-------
Air Toxics Ambient Monitoring Data Indicate General Reduction in
Emissions
Trends in ambient air concentrations can be indicative of trends in air toxics
emissions. ERG6 conducted a study, Evaluating HAP Trends: A Look at
Emissions, Concentrations, and Regulation Analyses for Selected Metropolitan
Statistical Areas, to identify trends in air toxics emissions and ambient
concentrations in selected metropolitan areas. This study's results suggest a
decline in air toxics7 concurrent with the implementation of the MACT standards.
The study's objectives included determining whether air toxics-specific Federal
regulations have been effective at reducing ambient concentrations of air toxics,
and whether air toxics emissions show a decline due to air toxics-specific Federal
regulations. The study focused on trends for 9 specific air toxics in
10 metropolitan areas. ERG concluded that ambient concentrations generally
declined between 1992 and 2003; during this same period, EPA issued 64 air
toxics-specific regulations. The study also noted that air toxics emissions
decreased substantially between 1990 and 2002; during that period, EPA
implemented over 40 air toxics-specific regulations. The study's authors noted
that their analysis was based on the unrevised 1990 NEI and draft 2002 NEI. Our
analysis, which was based on the revised 1990 NEI and a later version of the 2002
NEI, showed similar trends.
External Factors and Changes in Emissions Estimation Methods
Impact Air Toxics Emissions Reported
Factors external to the MACT standards can cause reductions in air toxics
emissions. For example, a decline in production for a particular industry can
result in reduced emissions, regardless of the MACT requirements. While it
varies by industry and facility, implementation of some National Ambient Air
Quality Standards regulations often produces a co-benefit of reducing some air
toxics. Similarly, State or local air toxics regulations that are more stringent than
Federal MACTs may be responsible for some air toxics emission reductions.
Further, changes in emissions factors or other estimation methods have impacted
reported emissions for some source categories. Representatives from six of the
eight facilities we contacted said that improved emissions factors and/or increased
emissions monitoring8 have led to more accurate estimates of emissions than in
the past. Because reported emissions are often based on emissions factors,
6 ERG is a multidisciplinary consulting firm that offers a broad range of services, including environmental services.
While EPA is one of ERG' s clients, this study was self-initiated and not requested or funded by EPA.
7
Study results presented October 2006 at the Air and Waste Management Association's Environmental Data
Analysis Workshop in Chicago, Illinois.
8 Of the three MACTs we reviewed in-depth, only the Primary Aluminum MACT required emissions monitoring of
air toxics. The other two MACTs required monitoring of processes or surrogate pollutants to determine compliance
with the MACT standard.
13
-------
changes in these factors over time may make it appear that actual emissions have
increased or decreased, when in fact it is the estimation methods that have
changed. OECA managers and staff told us that a general lack of monitoring
requirements for the MACT program impedes their ability to analyze emissions
trends and assess the likelihood of potential MACT violations.
Further details on external factors changes in emission estimation techniques for
the three MACT standards on which we focused are in Appendices D, E, and F.
EPA and Some State Compliance Monitoring Activities Should Result
in Emissions Reductions
While data were not available to assess the overall effectiveness of compliance
monitoring activities, the results of specific EPA compliance monitoring
initiatives and one State initiative have disclosed significant noncompliance with
some MACT standards. Corrective actions to bring the sources into compliance
with the MACT should result in reduced air toxics emissions. For example, EPA
estimates that its MACT-related enforcement actions for Fiscal Years 2005 and
2006 will result in a projected 332,000 and 356,000 pounds of air toxics
reductions, respectively, once the sources come into compliance with the MACT.
We were not able to identify any specific impact on emission reductions from
compliance assistance activities (e.g., guidance documents, training workshops,
etc.). However, the State and industry representatives we talked with told us that
EPA's compliance assistance activities and tools were generally useful. See
Appendices G and H for more information on compliance monitoring and
assistance activities conducted by EPA.
Key Compliance Monitoring Information Lacking to Evaluate
MACT Effectiveness
Sufficient data on all MACT-related compliance monitoring and assistance
activities were not available to quantitatively determine the impact of these
activities on air toxics emissions. EPA's database for recording MACT-related
compliance monitoring activities, generally known as the Air Facility Subsystem
(AFS), lacked essential program information needed to conduct such analyses.
We noted the following limitations:
• AFS did not identify the specific MACT subpart applicable to a facility.
• AFS did not indicate the pollutants analyzed during stack tests to measure
the amount of a specific pollutant or pollutants being emitted through the
stacks at a facility.
• AFS did not indicate the specific CAA regulations assessed during partial
compliance evaluations.
• Some States did not submit required information to AFS.
14
-------
In some instances, these data elements were not included because EPA policy did
not require States to submit this data. In other instances, incompatibility between
State data systems and AFS hampered reporting. The absence of key compliance
monitoring information hinders EPA's ability to assess MACT program
performance and properly oversee State compliance monitoring efforts.
AFS is the primary source of compliance data for facilities regulated by the CAA,
including MACT sources. AFS is comprised of the minimum data requirements
that State and local agencies are required to submit for CAA-regulated facilities,
such as compliance monitoring activities and enforcement actions. AFS supplies
all of the air data to the EPA IDEA (Integrated Data for Enforcement Analysis)
system, which is an integrated, multi-media data system that contains information
on all EPA-regulated facilities.
The minimum data requirements are approved by the Office of Management and
Budget through the Information Collection Request process. The most recent set
of minimum data requirements became effective in Fiscal Year 2006. In that
year, EPA began requiring States to submit MACT subpart information for each
MACT facility. However, this requirement is being phased in and EPA does not
expect AFS to contain this information for the entire MACT universe until Fiscal
Year 2011. As part of future requests to the Office of Management and Budget,
OECA plans to request the addition of partial compliance evaluation and stack
test data to the list of minimum data requirements, pending AFS modernization.
Additionally, some States were not reporting all of the minimum data
requirements to AFS. For example, staff from Region 5 told us that Illinois had
not submitted a large amount of the required minimum data to AFS due to
resource constraints and the lack of an interface between the State's database and
EPA's AFS database. Texas has also had a long-running problem submitting data
to AFS because the State's system was not compatible with AFS. The State
recently uploaded new data into AFS, but this update had not yet been completed
at the time we conducted our analysis. EPA regional air program managers told
us of at least two other States with similar problems, but said these States have
since taken action to improve their data submission to AFS.
OECA staff and managers told us that more detailed and complete information on
State compliance monitoring activities would enable them to more effectively
manage the MACT compliance monitoring program. For example, OECA could
compare a facility's emissions trends to the type of compliance monitoring
activity conducted at a facility. This information would help OECA evaluate the
adequacy of the State's compliance monitoring program and identify whether a
State needed additional oversight or guidance. In addition, this information
would enable OECA to better target its own compliance monitoring activities.
15
-------
Conclusions
NEI data and ambient monitoring data indicate that air toxics emissions declined
after implementation of the MACT program. Based on our analysis, we believe it
is reasonable to conclude that the MACT standards have reduced air toxics
emissions. However, because of data limitations, it is uncertain the extent to
which emissions reductions have occurred or can be attributed to the MACT
standards. In addition to emissions data of questionable reliability, a significant
limitation to evaluating the effectiveness of the MACT program is the lack of key
information in EPA databases for tracking MACT-related compliance monitoring
activities.
Agency Comments and OIG Evaluation
The Agency did not have any comments on this chapter.
16
-------
Chapter 3
Air Toxics Inventory Data Need to Be Improved
Before Being Used in Residual Risk Assessments
EPA needs to improve its NEI data before the Agency uses it to conduct air toxics
residual risk assessments. The quality of EPA's air toxics inventory data has
taken on increased importance with EPA's decision to rely on NEI data to
conduct residual risk assessments. Residual risk assessments require accurate
facility-specific emission data in order for EPA to determine the public health risk
resulting from exposure to air toxics. However, EPA has not established data
quality objectives specifying the quality of data needed for this decision making
process as recommended by EPA guidance. Further, EPA continues to rely on
voluntary efforts to collect air toxics emissions data from States, which hampers
its ability to obtain complete, consistent, and timely data. If emissions and the
subsequent risk derived from those emissions for specific source categories are
understated, important regulations to protect public health may go undeveloped.
Conversely, if emissions are overstated, requirements that are not cost-beneficial
could be placed on industries.
Reliability of NEI Data is Uncertain
Air toxics emissions data come from several sources, often with associated data
limitations. Data sources for the NEI include State-reported data, data gathered
by EPA during industry-specific rulemakings, industry-reported data from the
Toxics Release Inventory, and data derived from national estimates. Further, the
methodology used to develop emission estimates reported in the NEI can differ
between and within the sources of the data. For example, different States may use
different methodologies to develop the emission estimates sent to EPA. Also,
most air toxics emissions data are provided to EPA on a voluntary basis.
Review of the point source codes for EPA's 2002 NEI data showed improvement
from the 1990 NEI data. The source codes identify the sources of data used in the
development of NEI data, and are useful in evaluating the data's quality. We
found data source codes for point sources generally to be well populated in EPA's
2002 NEI.
Within EPA's NEI database, Emission Calculation Method Codes (ECMCs) are
used to indicate how emissions measurements are derived. ECMCs provide
important information on the quality of the emissions data, since the reliability of
emissions estimates varies greatly based on the calculation method used. We
reviewed ECMC records from an updated dataset of 2002 NEI point sources
provided by EPA's Emissions Inventory and Analysis Group in June 2007. We
found that the ECMS field was blank for 57 percent of the data points. The
proportion of the unpopulated records represented about 39 percent of total point
17
-------
source emissions. Further, emissions factors accounted for 37 percent of the data
points and nearly 40 percent of the reported emissions. Emissions factors are
representative values that relate the amount of pollutant emitted to the atmosphere
to an activity associated with the source (e.g., kilograms of nitrogen oxide emitted
per unit of fuel burned). As noted in prior OIG reports, emissions estimates
derived from emissions factors are generally not considered to be of high quality,
particularly for site-specific emissions estimates, and EPA recommends against
such use. Table 3-1 shows the number and types of ECMCs recorded for the
2002 NEI Data.
Table 3-1: Type and Number of Emission Calculation Methods Used for NEI Data
(Emission Emission Emissions
Calculation Calculation Method Number of Percentage of (Tons per
Method Code Description Records Records year)
363,240
nission Emission
culation Calculation Method Number of
lod Code Description Records
NA NO EMISSION 1 022 797
CALCULATION
METHOD CODE
01 CONTINUOUS 5 707
EMISSIONS
MONITORING SYSTEM
04 STACK TEST 14,786
03 MATERIAL BALANCE 48,953
08 EPA EMISSION 315330
FACTOR
09 STATE/LOCAL 329 384
EMISSION FACTOR
10 SITE-SPECIFIC 8920
EMISSION FACTOR
11 VENDER EMISSION 4680
FACTOR
12 TRADE GROUP 11 634
EMISSION FACTOR
05 EPASPECIATION 12151
PROFILE
06 STATE/LOCAL 106
SPECIATION PROFILE
07 MANUFACTURER 735
SPECIFICATION
02 ENGINEERING 31 895
JUDGMENT
56.60%
0.32%
13,004
Proportion
of
Emissions
38.96%
1.39%
Totals3
' = Totals may not equal due to rounding.
1,807,079
Source: Data extracted from the NEI National Output format database by OAQPS.
0.82%
2.71%
17.45%
18.23%
0.49%
0.26%
0.64%
0.67%
0.01%
0.04%
1 .77%
100.00%
27,023
86,587
252,474
53,483
36,200
19,075
7,832
6,480
109
1,213
65,518
932,237
2.90%
9.29%
27.08%
5.74%
3.88%
2.05%
0.84%
0.70%
0.01%
0.13%
7.03%
100.00%
The heavy use of emissions factors in the NEI makes the reliability of the data
highly uncertain. Emissions factors can result in emissions data of questionable
reliability, particularly at the facility level. For example, managers at two primary
aluminum plants we contacted stated that they found the emissions factors used to
estimate hydrogen fluoride emissions were underreporting emissions prior to
implementation of the MACT requirements. A prior OIG report also noted
instances where the use of unreliable emissions factors resulted in underreporting
of emissions. Additionally, the Agency has not completed the corrective actions
for the emissions factors program it agreed to address based on a March 2006
18
-------
OIG report.9 Our prior report noted that the percentage of emissions factors rated
below average or poor increased from 56 percent in 1996 to 62 percent in 2004.
The most reliable methods of measuring emissions as indicated in EPA's 2002
NEI documentation are continuous emissions monitoring systems (CEMS) and
stack tests. However, actual measurements such as stack tests and CEMS account
for less than 2 percent of the measurements and less than 5 percent of total
emissions. Further, this percentage is unlikely to improve since, according to
EPA staff, none of the MACT standards requires air toxics to be measured by
CEMS. The lack of emissions calculation methodology data, coupled with the
heavy reliance on emissions factors to develop the NEI, casts doubt on the quality
and reliability of NEI data for making residual risk decisions.
The NEI data are an important starting point for the residual risk process. The
NEI data are input into air quality models used to estimate ambient air toxic
concentrations. In turn, these modeled ambient concentrations are input into other
models that estimate public health risk from exposure to air toxics. The public
health risk estimate is a key factor in EPA deciding whether or not to issue
additional standards to reduce air toxics emissions.
EPA and Other Studies Have Noted Uncertainties with Air Toxics Data
EPA's OAQPS acknowledged that there are uncertainties in the NEI data when it
submitted its residual risk plan to the Science Advisory Board for review.
OAQPS noted that:
... the accuracy of emissions values will vary depending on the
source of the data present in NEI, incomplete or missing data,
errors in estimating emissions values, and other factors.
Additionally, some emission values in the ANPRM [Advance
Notice of Proposed Rulemaking] data set are estimates developed
from emission factors, mass-balance calculations, or other
methods and thus lack the precision and verifiable nature of actual
stack measurements. Even stack measurements have associated
uncertainty due to factor[s] such as measurement error and the
long term representativeness of monitoring data.
OAQPS also acknowledged that some of the NEI data were incomplete, as
follows:
The NEI contains incomplete emissions data for some source
categories. With regard to these source categories, we derived
emissions data from the 1999 NEI, which, absent more specific or
recent data, we assumed to be equivalent to 2002 emissions. Where
' EPA Can Improve Emissions Factors Development and Management, Report No. 2006-P-00017, March 22, 2006.
19
-------
data were not available in the 1999 NEI, we derived data from the
Toxics Release Inventory (TRI). TRI emissions values are not
subject to the same QA procedures that NEI estimates are, and the
TRI database lacks some data fields and levels of specificity that
are required for accurate dispersion modeling.
EPA's Science Advisory Board reviewed EPA's Risk and Technology Review
plan and made several recommendations for improving the residual risk process
in a June 2007 memorandum to the EPA Administrator. The following two
recommendations in particular addressed the role of the NEI accuracy and
uncertainty of the emissions data:
• The plan should incorporate a framework for improving the NEI as new/more
accurate data become available. There is a concern that the use of the
voluntary NEI data base for a regulatory purpose could induce changes in
reporting that modify the data base.
• A sensitivity analysis should be conducted to determine: (1) which inputs are
the main drivers of the residual risk estimates, and (2) if differences in the
levels of uncertainty for those inputs (for example, uncertainties in emissions
from some sources compared to others) may potentially result in
misclassification.
EPA plans to do a sensitivity analysis for the more complex residual risk
assessments with higher estimated risks that may trigger the need for additional
standards. EPA did not conduct sensitivity analyses for the first eight assessments
conducted under its current Risk and Technology Review approach, since EPA
determined that these categories posed low risks. EPA plans to rely on its current
approach for improving the NEI data and the premise that using the data to
conduct residual risk assessments will induce improvements in NEI reporting.
An August 2005 NARSTO report10 discussed the uncertainty associated with air
toxics inventories. NARSTO noted that:
The data that do exist for these pollutants are generally derived not from
direct measurements, but from models that frequently rely on limited out-
of-date data, and which are rarely subject to analyses of uncertainty.
Data Quality Objectives Have Not Been Established
EPA guidance recommends the use of the Data Quality Objectives Process when
using environmental data. However, EPA has not established data quality
objectives for using NEI data in residual risk assessments. According to EPA's
10 Improving Emission Inventories for Effective Air Quality Management Across North America, NARSTO 05-001,
August 2005. NARSTO is a public/private partnership including the United States, Canada, and Mexico that was
formed to improve management of air quality in North America.
20
-------
February 2006 Guidance on Systematic Planning Using the Data Quality
Objectives Process:
EPA developed the Data Quality Objectives (DQO) Process as the
Agency's recommended planning process when environmental data
are used to select between two alternatives or derive an estimate of
contamination. The DQO Process is used to develop performance and
acceptance criteria (or data quality objectives) that clarify study
objectives, define the appropriate type of data, and specify tolerable
levels of potential decision errors that will be used as the basis for
establishing the quality and quantity of data needed to support
decisions.
The Data Quality Objectives Process also calls for the user to establish "acceptable
quantitative criteria on the quality and quantity of the data to be collected, relative to
the ultimate use of the data." These acceptance criteria typically involve measures of
precision, bias, representativeness, completeness, comparability, and sensitivity.
EPA's plan for conducting residual risk assessments does not include
pre-established performance criteria for accepting the NEI air toxics emissions
data. Instead, EPA issued an Advance Notice of Proposed Rulemaking asking for
public comments on the NEI data EPA plans to use in its residual risk
assessments. At the time we issued our report, EPA was reviewing the public
comments and revising the inventory where comments supported changes.
According to OAQPS staff, comments had resulted in a 4-percent decrease in NEI
total emissions as of the end of July 2007. EPA has also begun to qualitatively
rank the data from A to E. OAQPS plans to consider these qualitative rankings
when conducting residual risk assessments to the extent they are available.
Although OAQPS plans to describe the limitation and uncertainties associated
with the NEI data, it does not plan to reject data based on its quality.
No Reporting Requirement for Air Toxics Emissions Data
Since EPA has not established an air toxics emissions reporting requirement, EPA
continues to employ voluntary efforts to collect air toxics emissions data. While
the percentage of State-reported data has increased from 54 percent in 1990 to
67 percent in 2002, EPA's reliance on voluntary efforts to provide air toxics NEI
data has not provided assurance that the data are accurate and reliable. A
reporting requirement would help ensure that all States submit data, the States
submit data in a consistent format, and the data include important information
such as the methodology used to develop the data. Although our 2004 report11 on
air toxics recommended that EPA develop such a requirement, EPA did not
implement our recommendation because EPA questioned whether it had the
statutory authority to require the inventories. In contrast, EPA does require States
1: EPA 's Method for Calculating Air Toxics Emissions for Reporting Results Needs Improvement, Report No.
2004-P-00012, March 31, 2004.
21
-------
to submit emissions data for criteria pollutants in accordance with the
requirements of the 2002 Consolidated Emissions Reporting Rule.
According to OAQPS staff, EPA considered including an air toxics emissions
reporting requirement in a January 2006 proposed Air Emissions Reporting
Requirements rule. Although this proposed rule did not include such a
requirement, EPA received a comment from the National Association of Clean
Air Agencies12 strongly encouraging EPA to require Title V facilities13 to report
air toxics emissions data. As of April 2007, the final rule had not been published.
OAQPS management and staff re-iterated that the lack of clear statutory authority
to establish an air toxic reporting requirement was a major obstacle to
implementing this rule.
Use of NEI Inventory Data Could Result in Unreliable Residual
Risk Assessments
EPA's residual risk assessments and the resulting decisions of whether to develop
residual risk standards represent a significant CAA requirement. These decisions
can have important ramifications for public health and control costs for affected
industries. Uncertainties or errors in the air toxics emissions inventories used to
develop these risk assessments could impact EPA's final decisions. EPA's Data
Quality Objectives guidance describes the potential impact of data errors on
decisions as follows:
. . . these data [that is, data used in decision-making] are subject
to various types of errors due to such factors as how samples were
collected, how measurements were made, etc. As a result,
estimates or conclusions that you make from the collected data
may deviate from what is actually true within the population.
Therefore, there is a chance that you will make erroneous
conclusions based on your collected data or that the uncertainty in
your estimates will exceed what is acceptable to you.
With respect to residual risk decisions, if emissions and the subsequent risk
derived from those emissions are understated, EPA may not develop regulations
that are needed to protect public health. Conversely, if emissions and the
subsequent residual risk are overstated, regulatory requirements that are not cost
effective could be placed on industries.
12 This association represents air pollution control agencies in the 50 U.S. States, 4 U.S. Territories, and over
165 metropolitan areas. The association was formerly known as STAPPA/ALAPCO (State and Territorial Air
Pollution Program Administrators/Association of Local Air Pollution Control Officials).
13 All major sources (i.e., any source that emits or has the potential to emit 100 tons or more a year of any criteria
pollutant), and certain areas sources are required by the 1990 CAA Amendments to obtain Title V permits.
A Title V permit grants a source permission to operate, and includes all air pollution requirements that apply to the
source. It also requires that the source report its compliance status with respect to permit conditions.
22
-------
Conclusions
EPA has not established a process to ensure that the air toxics emissions data used
in residual risk assessments meet data quality objectives as recommended by
Agency guidance. Ideally, data that meet pre-determined performance criteria
should be used for residual risk decisions because of the significant impact these
decisions could have on the public's health and the regulated community. While
OAQPS plans to characterize the limitations and uncertainties of their risk
assessments, these characterizations as planned would not quantitatively describe
the range of the potential errors associated with these assessments.
Recommendations
We recommend the Principal Deputy Assistant Administrator for Air and
Radiation:
3-1 Establish data quality objectives for data used in conducting residual risk
assessments.
3-2 Revise the Consolidated Emissions Reporting Rule to require standardized
State reporting requirements and methods for calculating air toxics
emissions data.
Agency Comments and OIG Evaluation
The Agency did not express disagreement with the report's findings, but
disagreed with several conclusions as well as with the OIG's recommendations.
The Agency stated that it plans to improve the quality and consistency of the NEI
data, but does not commit to establishing data quality objectives. The Agency
agreed that establishing air toxics emissions reporting requirements could
improve the quality of NEI data. However, the Agency said that such
requirements were not appropriate at this time. Thus, the Agency plans to
continue to rely on voluntary efforts to obtain air toxics NEI data.
We do not believe the Agency's response sufficiently addresses the problems
disclosed in this report, and we consider the issues unresolved. A summary of the
Agency's response to each recommendation and our evaluation of the Agency's
comments follows. A copy of the Agency's complete response is included in
Appendix I.
Recommendation 3-1
The Agency disagreed with our recommendation that it establish data quality
objectives for data used in conducting residual risk assessments. The Office of
Air and Radiation commented that quantifiable data quality objectives or
23
-------
acceptance criteria for the use of data within the NEI were not practicable or
appropriate within the current assembly and use of the NEI. The Office of Air
and Radiation noted that for the NEI, it uses a Quality System based on an
acceptance process to determine if the inventory is suitable for the purpose(s) for
which it is to be used. According to the Office of Air and Radiation, this process
includes numerous quality reviews by both EPA and its State/local partners and
generally incorporates multiple elements, such as: (1) data/information flow
monitoring and checks; (2) clear decision points in the data flow; (3) decisions by
professional experts in assessing data elements through mass balance calculations,
statistical determinations, and other tools; and (4) data acceptance determinations
regarding the usability of the data/information for the intended purpose.
The OIG continues to believe that data quality objectives should be established
for emissions data used in conducting residual risk assessments. As the Office of
Air and Radiation noted in its response, the intended use of data determines the
quality of data needed. In the case of the Risk and Technology Review process,
the emissions data serve as the foundation for EPA's residual risk assessments
and the establishment of residual risk standards under section 112(f) of the CAA.
In response to EPA's Advance Notice of Proposed Rulemaking, several State
agencies and industry groups cited NEI data quality problems and questioned its
suitability for residual risk assessments. Of particular note:
• The National Association of Clean Air Agencies expressed serious
concerns about the quality of NEI data and its use in conducting residual
risk assessments, and suggested that EPA use its CAA section 114
authority to collect the emissions data needed for residual risk
assessments.
• The American Shipbuilding Association stated that it was deeply
concerned about EPA making decisions affecting their industry "based
upon an unrealistic selection of data that is not only incomplete and
inaccurate, but one that almost seems to disregard the importance of
relevance as a determinative factor." They also stated that "Some of the
data is by far too inaccurate to be used for modeling and residual risk
assessment for individual facilities."
• The American Petroleum Institute expressed similar concerns in stating
that "It is inappropriate for the Agency to use low quality data on the
basis that the regulated community did not provide better data. Yet this is
what the Agency proposes to do for the RTR [Risk and Technology
Review] 2 rulemaking." In order to obtain better data, the American
Petroleum Institute recommended that EPA work directly with sources to
obtain the information voluntarily, use its section 114 authority, or
promulgate a rule requiring the submission of necessary data.
As noted in comments to the Advance Notice of Proposed Rulemaking, the
existing NEI database is not the only option for obtaining emissions data for risk
assessments. The Agency has explicit authority under CAA section 114 to obtain
24
-------
emissions data from stationary sources in order to develop or assist in the
development of any emissions standard under section 112. The Agency could use
such authority to collect emissions data for residual risk assessments if NEI data
are not of sufficient quality.
Recommendation 3-2
The Office of Air and Radiation agreed that establishing requirements for
reporting air toxics emissions data to the NEI could improve the quality of such
data. However, the Agency commented that such requirements were not
appropriate at this time. The Agency pointed to its evolving efforts to improve
the air toxics emission inventory by working with its State and local partners and
industry.
EPA outlined several reasons why an air toxics emissions reporting requirement
was needed when it proposed the 2002 Consolidated Emissions Reporting Rule.
In the rule's proposal EPA stated that air toxics emissions inventories were
needed to: conduct better oversight of the Title V program; develop and assess
progress of the CAA-required control strategy for reducing public health risk
from area sources; and assess progress in meeting Government Performance and
Results Act goals for reducing the public's cumulative health risk from exposure
to air toxics emissions. Although it did not include an air toxics reporting
requirement in the final rule, EPA stated that it planned "to develop HAP
reporting measures at a future date." EPA continues to rely on voluntary
reporting efforts which, in our view, have not provided reasonable assurance that
the NEI data are accurate and reliable. Given the Agency's decision to also use
NEI data for residual risk assessments, we believe it is inappropriate for the
Agency to delay implementation of an air toxics reporting requirement to improve
data quality.
As mentioned above, EPA already has authority under CAA section 114 to collect
air toxics emissions data from stationary sources. Thus, the Agency could use
section 114 authority to obtain data for residual risk assessments, even if routine
reporting of air toxics emissions was not required at this time.
25
-------
Status of Recommendations and
Potential Monetary Benefits
POTENTIAL MONETARY
RECOMMENDATIONS BENEFITS (in $OOOs)2
Planned
Rec. Page Completion Claimed Agreed To
No. No. Subject Status1 Action Official Date Amount Amount
3-1 23 Establish data quality objectives for data used in 0 Principal Deputy
conducting residual risk assessments. Assistant Administrator
for Air and Radiation
3-2 23 Revise the Consolidated Emissions Reporting Rule 0 Principal Deputy
to require standardized State reporting Assistant Administrator
requirements and methods for calculating air toxics for Air and Radiation
emissions data.
1 O = recommendation is open with agreed-to corrective actions pending;
C = recommendation is closed with all agreed-to actions completed;
U = recommendation is undecided with resolution efforts in progress.
2
Identification of potential monetary benefits was not an objective of this evaluation.
26
-------
Appendix A
Details on Scope and Methodology
To address the extent to which MACT standards have reduced air toxics and achieved the
emission reductions anticipated, we assessed the changes in air toxics emissions between 1990
and 2002 using EPA's NEI. Since all of the MACTs were promulgated after passage of the 1990
CAA Amendments, the 1990 NEI serves as a baseline of pre-MACT air toxics emissions. The
2002 NEI was the most current inventory available. After we had completed the majority of our
field work and analyses, EPA released an updated version of the 2002 inventory in March 2007.
Accordingly, we updated our analyses using the March 2007 revised updated NEI data. This
latest update is generally known as NEI 2002 Version 3. Our overall analyses of MACT
emissions included the following:
• First, we compared 1990 and 2002 emissions for only those emissions assigned to
specific MACT codes. Since differences exist between the 1990 and 2002 inventories,
we used only NEI records that were comparable between the 2 years' inventories when
comparing emissions for specific MACT categories. That is, we only included emissions
for air toxics that were included in both inventories for that MACT category. After
presenting our preliminary results to OAQPS, they commented that our emissions totals
did not accurately account for certain categories of air toxics. This was typically the case
for speciated pollutants such as Polycyclic Organic Matter (POM), Chromium, Xylene,
and Cresol that needed to be compared on a category basis. The HAP category is an EPA
convention that includes all the species of a particular HAP. We adjusted our MACT
category emission totals to properly account for speciated toxics as recommended by
OAQPS. These adjustments did not materially impact the total MACT emissions.
• Second, we compared 1990 and 2002 emissions for only those MACT standards that had
compliance dates prior to 2002. Since these MACT source categories should have been
in compliance before 2002, the 2002 NEI should serve as a post-MACT measure of air
toxics emissions reductions.
• We also compared emissions of the 6 key HAPs from these 27 MACT categories in 1990
and 2002. The key HAPs are those air toxics identified by OAQPS as posing the highest
cancer and non-cancer risks on average nationwide.
We selected three MACT source categories for more in-depth analysis. We selected these
MACTs from the universe of MACTs with compliance dates prior to 2002. We used the
following criteria in selecting the three MACTs for in-depth analyses:
• Air toxics emissions from the source category pose a relatively large public health risk.
• The NEI data for the source category are among the most reliable/accurate.
• Anticipated emission reductions from the MACT are large.
• Changes in air toxics emissions are not due to obvious external factors.
27
-------
The following table illustrates the three MACTs selected using the above criteria:
Table A.1: MACTS Selected for Review
MACT
Petroleum Refinery MACT (Subpart CC)
Primary Aluminum MACT (Subpart LL)
Pulp and Paper MACT I & III (Subpart S)
Date Issued
August 18, 1995
October/, 1997
April 15, 1998
No. Air Toxics Targeted
by the MACT Standard
10
3
13
Source: Developed by OIG staff based on information from EPA
We also interviewed EPA staff and industry groups, and conducted literature searches, to
identify external factors that may have impacted emissions for these three MACT categories.
To address what factors are associated with the effectiveness of the MACT standards in
achieving emission reductions, we selected nine facilities from the three MACTs above for even
greater in-depth analyses. We selected these nine facilities based on emissions of certain
pollutants of concern contained in the 2002 NEI Version 1. Specifically:
• For the Petroleum Refinery MACT, we selected the three facilities with the highest
reported emissions of benzene. We used benzene because EPA had identified benzene as
one of six key HAPs, and petroleum refineries are large emitters of benzene.
• For the Primary Aluminum MACT, we selected the three facilities with the highest
reported emissions of POM. We used POM because it is highly toxic and is a common
pollutant from primary aluminum smelters.14
• For the Pulp and Paper MACT I & III, we selected the three facilities with the highest
reported emissions of formaldehyde. We used formaldehyde because EPA had identified
it as one of six key HAPs, and pulp and paper mills are relatively large emitters of
formaldehyde.15
For each of these nine facilities, we reviewed and compared emissions data from the TRI and
NEI, as well as compliance and enforcement data from EPA's IDEA system. We conducted
semi-structured interviews with representatives from the facilities, and also with agency staff
from the States in which the facilities are located. We also conducted interviews with staff from
the EPA regions where these facilities and States are located. However, because of limitations in
the compliance monitoring and compliance assistance data, we were unable to fully address our
objective of determining whether there was an association between compliance monitoring and
compliance assistance activities and emission trends. Instead, we relied upon interviews of
knowledgeable individuals and various studies to qualitatively assess the impact of compliance
monitoring and compliance assistance on air toxics emissions reductions from MACT sources.
Since the three MACT categories and nine facilities we selected for in-depth review were not
14 One of the top emitters of POM in the primary aluminum industry was not in operation at the time of our review,
so we selected the next highest emitter to be part of our sample.
15 According to EPA staff, in the last 3-4 years, the toxicity risk for formaldehyde has been lowered, and is no longer
considered a key HAP, but controversy remains over this action.
28
-------
randomly selected, our findings and recommendations may not be representative of the MACT
universe as a whole.
While conducting field work, we identified several significant problems with the NEI data, and
learned that EPA is planning to use the 2002 NEI to conduct residual risk analyses for a large
number of source categories under the Risk and Technology Review process.16 Given our
concerns with the quality of the air toxics emissions inventory data and EPA's intent to use this
same data to make regulatory decisions, we conducted additional steps not included in our
original project plan and evaluation objectives. For example, we reviewed documents pertaining
to the Risk and Technology Review process, reviewed the updated Risk and Technology Review
data files, and interviewed EPA staff involved in the residual risk process.
Locations Reviewed
Our evaluation focused on two EPA headquarters offices: (1) OAQPS within the Office of Air
and Radiation, and (2) OECA. OAQPS, located in Research Triangle Park, North Carolina,
developed the MACT standards and many of the compliance assistance tools for affected
industries to use in reducing air toxics emissions. OECA, located in Washington, DC, is
responsible for ensuring compliance with the MACT standards. Since the regions and States do
most of the actual implementation of the MACT standards, we also contacted the following five
regions and seven States:
Table A.2: Regions and States Contacted
EPA Region
Region 3 - Philadelphia
Region 4 -Atlanta
Region 5 - Chicago
Region 6 - Dallas
Region 7 - Kansas City
States
West Virginia
Mississippi, South Carolina
Oklahoma, Texas
Kansas, Missouri
Source: Developed by OIG staff
Additionally, to determine how facilities calculated their reported emissions and the extent to
which external factors may have influenced reported reductions, we interviewed representatives
of eight of the nine facilities selected for greater in-depth analyses. Representatives from one
pulp and paper facility, despite repeated attempts to schedule an interview, did not respond to our
requests. We also conducted interviews with outside stakeholders.
Review of Management (Internal) Controls
Government Auditing Standards require that auditors obtain an understanding of internal controls
significant to the audit objectives and consider whether specific internal control procedures have
been properly designed and placed in operation. Our overall objective was to determine the
extent to which MACT standards have reduced air toxics and achieved the emission reductions
anticipated. We reviewed management controls related to the objective of this evaluation. This
included EPA's policy and procedures for administering and managing the MACT Program for
16 As explained in Chapter 3, EPA has asked stakeholders to review the NEI data for 22 MACT source categories
before using the data to complete the mandated residual risk analyses.
29
-------
the three sample MACT categories issued by EPA's OAQPS. As part of this review, we
examined analytical, documentary, and testimonial evidence from five EPA regions, seven
States, and nine facilities (three each in the Petroleum Refinery industry, Primary Aluminum
industry, and the Pulp and Paper industry).
We found the internal controls did not ensure reliable data from the AFS. This system is the
primary source of compliance data for facilities regulated by the C AA, including MACT sources.
Our concerns with this system are discussed in Chapter 2. We also found the internal controls
did not ensure that data generated by or for OAQPS are of known and acceptable quality. Our
concerns with data quality are discussed in Chapter 3.
Data Limitations
EPA has taken steps to improve the NEI since our 2004 report (see "Prior Reports" below)
discussed concerns with the accuracy of air toxics emissions data. These improvements include
revisions made to the 2002 NEI after we started our field work. However, as discussed in
Chapter 3, significant data quality concerns still exist with this inventory.
The reliability of trends analyses is limited by improvements to the 2002 data, which impacts the
comparability of this data to the 1990 data. For example, changes to the 2002 NEI have resulted
in:
• increases in the number of facilities and emissions sources,
• increases in the number of air toxics with reported emissions, and
• changes in the MACT code assigned to some sources.
Prior Reports
The OIG conducted one prior evaluation that indirectly related to the effectiveness of the MACT
standards. This report, EPA 's Methods for Calculating Air Toxics for Reporting Needs
Improvement (2004-P-00012), issued March 31, 2004, noted that EPA was measuring the overall
air toxics program's progress based on total air toxics emissions reductions without considering
the toxicity of the pollutants. The report recommended that EPA include a risk-based factor in
its program measurement. In response to our recommendation, EPA has started using toxicity-
weighted measures to evaluate the air toxics program.
The OIG and GAO conducted several prior audits/evaluations that addressed EPA's
development of the MACT standards in accordance with the timelines established by the CAA
and efforts to incorporate MACT requirements into individual facility operating permits. These
prior reports did not specifically review the effectiveness of the MACT standards in reducing air
toxics emissions. These prior reports are:
30
-------
Prior OIG Reports:
• EPA and State Progress in Issuing Title VPermits, Report No. 2002-P-00008, March 29,
2002
• Substantial Changes Needed in Implementation and Oversight of Title V Permits If
Program Goals Are To Be Fully Realized, Report No. 2005-P-00010, March 9, 2005
Prior GAO Reports:
• Air Pollution: Status of Implementation and Issues of the Clean Air Act Amendments of
1990, GAO/RCED-00-72, April 2000
• EPA Has Completed Most of the Actions Required by the 1990 Amendments, but Many
Were Completed Late, GAO-05-613, May 2005
• EPA Should Improve the Management of Its Air Toxics Program, GAO-06-669, June
2006
31
-------
Appendix B
Amount and Percent of Change in NEI
Air Toxics Emissions Between 1990 and 2002
MACT Source Category
Aerospace Industries
Agricultural Chemicals and Pesticides
Manufacturing
Auto & Light Duty Truck (Surface
Coating)
Boat Manufacturing
Brick and Structural Clay Products
Manufacturing
Carbon Black Production
Cellulose Products Manufacturing
Chromium Electroplating
Clay Ceramics Manufacturing
Coke Ovens: Charging, Top Side, and
Door Leaks
Coke Ovens: Pushing, Quenching, &
Battery Stacks
Commercial Sterilization Facilities
Dry Cleaning Facilities:
Perchloroethylene
Engine Test Cells/Stands
Ethylene Processes
Ferroalloys Production
Flexible Polyurethane Foam Fabrication
Operations
Flexible Polyurethane Foam Production
Friction Materials Manufacturing
Gasoline Distribution (Stage I)
Halogenated Solvent Cleaners
Hazardous Waste Incineration:
Commercial
Hydrochloric Acid Production
Hydrogen Fluoride Production
Industrial/Commercial/ Institutional
Boilers & Process Heaters
Integrated Iron & Steel Manufacturing
Iron and Steel Foundries
Large Appliance (Surface Coating)
Leather Tanning & Finishing Operations
Lime Manufacturing
Magnetic Tapes (Surface Coating)
Manufacture of Nutritional Yeast
Marine Vessel Loading Operations
Mercury Cell Chlori-Alkali Plants
1990 NEI
(tons per year)
209,666
1,147
35,565
8,868
5,562
6,970
4,696
184
982
2,214
904
308
77,698
745
439
714
12
16,335
1,948
33,241
137,982
10,605
61,048
4
30,323
1 1 ,444
7,658
12,027
2,508
1
4,478
254
7,449
88
2002 NEI
Version 3
(tons per year)
2,305
1,433
6,762
7,325
6,635
1,550
2,182
5
771
64
1,327
46
22,117
716
500
293
20
2,589
225
80,028
39,451
9,616
399
5
221,990
706
1,628
2,498
216
999
434
93
216
18
Change
(2002-1990)
(tons per year)
-207,361
286
-28,803
-1,543
1,073
-5,420
-2,514
-179
-211
-2,150
423
-262
-55,581
-29
61
-421
8.00
-13,746.00
-1,723.00
46,787
-98,531
-989
-60,649
1
191,667
-10,738
-6,030
-9,529
-2,292
998
-4,044
-161
-7,233
-70
Percent
Change
(2002-1990)
-99%
25%
-81%
-17%
19%
-78%
-54%
-97%
-21%
-97%
47%
-85%
-72%
-4%
14%
-59%
67%
-84%
-88%
141%
-71%
-9%
-99%
25%
632%
-94%
-79%
-79%
-91%
99800%
-90%
-63%
-97%
-80%
32
-------
MACT Source Category
Metal Can (Surface Coating)
Metal Coil (Surface Coating)
Metal Furniture (Surface Coating)
Mineral Wool Production
Miscellaneous Coating Manufacturing
Miscellaneous Metal Parts & Products
(Surface Coating)
Miscellaneous Organic Chemical
Manufacturing
Municipal Landfills
Natural Gas Transmission & Storage
Off-Site Waste and Recovery
Operations
Oil & Natural Gas Production
Organic Liquids Distribution
(Non-Gasoline)
Paper & Other Webs (Surface Coating)
Petroleum Refineries - Catalytic
Cracking, Catalytic Reforming, & Sulfur
Plant Units
Petroleum Refineries - Other Sources
Not Distinctly Listed
Pharmaceutical Production
Phosphate Fertilizers Production
Phosphoric Acid Manufacturing
Plastic Parts & Products (Surface
Coating)
Plywood and Composite Wood
Products
Polyether Polyols Production
Polymers and Resins 1 &II
Polymers and Resins III
Polymers and Resins IV
Polyvinyl Chloride & Copolymers
Production
Portland Cement Manufacturing
Primary Aluminum Production
Primary Copper Smelting
Primary Lead Smelting
Primary Magnesium Refining
Printing, Coating & Dyeing Of Fabrics
Printing/Publishing (Surface Coating)
Publicly Owned Treatment Works
Pulp and Paper (Combustion) MACT II
Pulp & Paper (non-combustion), MACT
land III
Reciprocating Internal Combustion
Engines (RICE)
Refractory Products Manufacturing
Reinforced Plastic Composites
Production
1990 NEI
(tons per year)
40,574
4,013
15,664
3,253
8,853
61,134
96,625
309
469
435
42,589
147
30,706
3,793
101,305
40,958
558
128
58,696
18,702
3,699
26,248
1,082
4,008
1,127
9,691
4,069
157
297
37,535
16,065
35,860
8,119
56,922
247,777
28,484
238
14,814
2002 NEI
Version 3
(tons per year)
3,797
2,285
759
482
3,988
8,643
8,507
28,945
2,153
17
21,541
1,146
3,059
1,030
8,717
2,458
117
62
8,015
10,124
170
4,763
2,042
730
59
4,143
4,829
37
60
7,385
4,562
11,089
12,031
35,637
22,971
8,294
78
3,730
Change
(2002-1990)
(tons per year)
-36,777
-1,728
-14,905
-2,771
-4,865
-52,491
-88,118
28,636
1,684
-418
-21,048
999
-27,647
-2,763
-92,588
-38,500
-441
-66
-50,681
-8,578
-3,529
-21 ,485
960
-3,278
-1,068
-5,548
760
-120
-237
-30,150
-11,503
-24,771
3,912
-21 ,285
-224,806
-20,190
-160
-11,084
Percent
Change
(2002-1990)
-91%
-43%
-95%
-85%
-55%
-86%
-91%
9267%
359%
-96%
-49%
680%
-90%
-73%
-91%
-94%
-79%
-52%
-86%
-46%
-95%
-82%
89%
-82%
-95%
-57%
19%
-76%
-80%
-80%
-72%
-69%
48%
-37%
-91%
-71%
-67%
-75%
33
-------
MACT Source Category
Rubber Tire Production
Secondary Aluminum Production
Secondary Lead Smelting
Semiconductor Manufacturing
Shipbuilding & Ship Repair (Surface
Coating)
Site Remediation
Solvent Extraction for Vegetable Oil
Production
Stationary Combustion Turbines
Steel Pickling - HCL Process
Synthetic Organic Chemical
Manufacturing (Hazardous Organic
NESHAP)
Taconite Iron Ore Processing
Viscose Process Manufacturing
Wet-Formed Fiberglass Mat Production
Wood Building Products (Surface
Coating)
Wood Furniture (Surface Coating)
Wool Fiberglass Manufacturing
Totals
1990 NEI
(tons per year)
2,766
21,860
2,116
1,915
3,462
28
25,549
550
3,316
546,063
0
3,216
322
12,203
24,403
6,665
2,373,604
2002 NEI
Version 3
(tons per year)
1,247
2,688
951
389
1,455
186
16,270
1,294
1,038
12,264
133
2,477
51
2,790
10,128
2,302
709,280
Change
(2002-1990)
(tons per year)
-1,519
-19,172
-1,165
-1,526
-2,007
158
-9,279
744
-2,278
-533,799
133
-739
-271
-9,413
-14,275
-4,363
-1,664,324
Percent
Change
(2002-1990)
-55%
-88%
-55%
-80%
-58%
564%
-36%
135%
-69%
-98%
443233%
-23%
-84%
-77%
-58%
-65%
70%
Source: Data extracted from NEI databases
34
-------
Appendix C
NEI Emissions Reductions Compared to
Federal Register Anticipated Reductions
MACT Category
Aerospace Industries
Chromium Electroplating
Coke Ovens: Charging, Top
Side, and Door Leaks
Commercial Sterilization
Facilities
Dry Cleaning Facilities:
Perchloroethylene
Ferroalloys Production
Flexible Polyurethane Foam
Production
Gasoline Distribution (Stage 1)
Halogenated Solvent Cleaners
Industrial Cooling Towers
Magnetic Tape (Surface
Coating)
Marine Vessel Loading
Operations
Off-Site Waste and Recovery
Operations
Petroleum Refineries
Pharmaceutical Production
Polymers & Resins 1 "
Polymers & Resins II
Polymers & Resins IV
Primary Aluminum Production c
Primary Lead Smelting
Federal
Register
Baseline
(tons per year)
209,661
175
1,830
1,188
46,500
NA
19,924
50,706
135,397
25
4,470
NA
57,320
89,618
36,923
14,618
135
19,974
11,000
NA
Federal
Register
Anticipated
Reduction
(tons per year)
123,700
174
1,532
1,140
25,800
0
13,800
2,866
853,000
25
2,300
4,565
47,000
52,911
24,000
7,046
105
3,880
5,500
0
Percent
Federal
Register
Anticipated
Reduction
59%
99%
84%
96%
55%
0%
69%
6%
63%
99%
51%
NA
82%
59%
65%
48%
78%
81%
50%
0%
1990 NEI
(tons per year)
209,666
184
2,214
308
77,698
714
16,335
33,241
137,982
NA
4,478
7,449
435
101,305
40,958
26,248
NA
4,008
3,178
297
2002 NEI
Version 3
(tons per year)
2,305
5
64
46
22,117
293
2,589
80,028
39,451
2.5
434
216
17
8,717
2,458
4,763
NA
730
1,832
60
NEI
Change
(2002-1990)
(tons per year)
207,361
179
2,150
-262
-55,581
-421
-13,746
46,786
-98,531
NA
-4,044
-7,233
-418
-92,588
-38,500
-21,485
NA
-3,278
1,346
-237
Percent NEI
Change
(2002-1990)
-99%
-97%
-97%
-85%
-72%
-59%
-84%
141%
-71%
NA
-90%
-97%
-96%
-91%
-94%
-82%
NA
-82%
42%
-80%
Anticipated
Emissions
Reduction
Achieved3
Yes
No
Yes
No
Yes
Yes
Yes
No
Yes
NA
Yes
NA
Yes
Yes
Yes
NA
NA
Yes
No
Yes
35
-------
MACT Category
Printing/Publishing (Surface
Coating)
Pulp & Paper Production
(Subpart S)
Secondary Lead Smelting
Shipbuilding & Ship Repair
(Surface Coating)
Steel Pickling - HCL Process
Synthetic Organic Chemical
Manufacturing (Hazardous
Organic NESHAP)
Wood Furniture (Surface
Coating)
Federal
Register
Baseline
(tons per year)
23,871
264,555
2,106
1,497
3,289
573,202
NA
Federal
Register
Anticipated
Reduction
(tons per year)
7,400
153,221
1,411
350
2,500
510,000
32,795
Percent
Federal
Register
Anticipated
Reduction
31%
58%
67%
23%
76%
89%
NA
1990 NEI
(tons per year)
35,848
247,777
2,116
3,462
3,316
546,063
24,400
2002 NEI
Version 3
(tons per year)
11,088
22,971
862
1,353
1,038
12,260
10,116
NEI
Change
(2002-1990)
(tons per year)
-24,760
-224,806
-1,253
-2,109
-2,278
-533,803
-14,285
Percent NEI
Change
(2002-1990)
-69%
-90%
-59%
-61%
-69%
-98%
-59%
Anticipated
Emissions
Reduction
Achieved3
Yes
Yes
No
Yes
No
Yes
NA
3 Achievement of the anticipated emissions reduction is based on comparison of NEI percentage reduction to Federal Register percentage reduction.
b The 1900 and 2002 NEI totals for this MACT include both Polymers and Resins I and II. This is because the emissions for these 2 MACTS cannot be
separated in the 1990 NEI.
c The Federal Register baseline and anticipated reduction figures for the Primary Aluminum MACT include only emissions of POM and Hydrogen Fluoride. In
order to compare the anticipated emissions reductions for this MACT to the NEI emissions changes, we included only POM and Hydrogen Fluoride emissions
in the NEI figures in this table. POM emissions from the NEI were determined by summing emissions of POM as PAH-7 and POM as PAH-15.
NA Data not available to make a determination.
Source: Developed by OIG staff from Federal Register notices, EPA rule development documents, and 1990 and 2002 NEI databases
36
-------
Appendix D
Detailed Results - Petroleum Refinery MACT
Petroleum refining is a complex process involving the physical, thermal, and chemical separation
of crude oil into its major components, which are further processed into a variety of finished
petroleum products. Petroleum products represent the single largest source of energy for the
United States, and include gasoline, jet fuel, home heating oil, and kerosene. According to the
Energy Information Administration, as of November 2006, 142 refineries were operating in the
United States. Refineries are located throughout the United States. The State with the largest
number of refineries is Texas, followed by California and Louisiana. Each refinery can have
thousands of possible emission points.
The 1995 Petroleum Refinery MACT (Subpart C) controls air toxics emissions from several
different emission points, including process vents, storage vessels, wastewater streams, gasoline
loading racks, marine tank vessel loading, and equipment leaks. The MACT incorporates other
standards, including the 1989 Benzene National Emission Standards for Hazardous Air
Pollutants (NESHAP)17 and the 1994 regulation for fugitive emissions at certain MACT
sources,18 often referred to as Leak Detection and Repair (LDAR). The MACT standard targets
10 primary air toxics emitted by refineries, including 2,2,4-trimethylpentane, benzene, cresols,
ethylbenzene, hexane, methyl tert butyl ether, naphthalene, phenol, toluene, and xylenes.
Facilities were to be in full compliance with the Petroleum Refinery MACT by August 18, 1998.
Aggregate and Targeted Air Toxics for the Petroleum Refining Industry
Decreased Dramatically
According to the NEI, air toxics emissions from petroleum refineries decreased dramatically
between 1990 and 2002. Total emissions of air toxics decreased from 101,305 tons in 1990 to
8,717 tons in 2002, a reduction of 91 percent. This exceeds the total air toxics emissions
reductions that EPA anticipated when the MACT was developed in 1995. EPA anticipated the
standard would reduce total air toxics emissions from the industry by 59 percent.
Emissions of all the individual air toxics specifically targeted by the MACT standard also
decreased significantly between 1990 and 2002. For example, emissions of benzene decreased
from 6,762 tons in 1990 to 861 tons in 2002, a reduction of 87.3 percent. Table D-l shows the
1990 and 2002 NEI emissions figures for each of the targeted air toxics for the Petroleum
Refinery MACT.
17 40 Code of Federal Regulations (CFR) 61 Subpart FF
18 40 CFR 63 Subpart H
37
-------
Table D-1: Emissions of Targeted Air Toxics for the Petroleum Refinery MACT
Targeted Air Toxics a
2,2,4-Trimethylpentane
Benzene
Cresol
Ethylbenzene
Hexane
Methyl Tert Butyl Ether
Naphthalene
Phenol
Toluene
Xylenes (Mixed Isomers)
Total
1990 NEIb
23,589.0
6,762.0
513.5
3,482.4
24,804.4
3,605.3
796.1
216.5
15,394.8
13,911.0
93,075.0
2002 NEI
Version 3 b
138.9
861.3
13.1
280.7
1,174.1
427.5
69.1
31.2
1,848.7
1,615.5
6,460.1
Change °
(2002-1990)
-23,450.1
-5,900.7
-500.4
-3,201.7
-23,630.3
-3,177.8
-727.0
-185.3
-13,546.1
-12,295.5
-86,614.9
Percent
Change
-99.4%
-87.3%
-97.5%
-91 .9%
-95.3%
-88.1%
-91 .3%
-85.6%
-88.0%
-88.4%
-93.1%
Primary air toxics targeted by the Petroleum Refinery MACT per the Federal Register.
bTons per year.
Source: Developed by OIG staff from NEI data
The decrease in emissions of total air toxics and targeted air toxics in the petroleum refining
industry coincides with the implementation of the MACT standard. The compliance date of the
MACT was August 18, 1998. Thus, based on the NEI data, it appears that the MACT has been
effective in reducing emissions in the industry. However, it is not possible to determine from
this analysis if the decreases in emissions were due only to the MACT standard. Other external
factors, such as industry trends or other regulations, may have also impacted emissions at
petroleum refineries.
Petroleum Refinery Emissions Decreased Despite Increase in Industry Production
and Capacity
Air toxics emissions from petroleum refineries decreased significantly between 1990 and 2002,
despite an increase in industry production over the same time period. According to the Energy
Information Administration, net U.S. production of crude oil and petroleum products increased
by 13.1 percent over this period, from approximately 5.6 billion barrels in 1990 to about 6.3
billion barrels in 2002.
38
-------
Figure D-1: Comparison of Industry Production to Total Air Toxics Emissions
7,000,000
6,000,000
•2 5,000,000
ro
i/>
O 4,000,000
I Production
-Total HAP Emissions
=- 3,000,000
jj 2,000,000
00
1990
2002
Source: Developed by OIG staff based on data from NEI and the Energy Information Administration
Similarly, total capacity in the refining industry increased between 1990 and 2002. According to
the Energy Information Administration, crude oil distillation capacity at operating U.S. refineries
increased by 7.3 percent, from about 15.1 millions barrels per day in 1990 to about 16.2 million
barrels per day in 2002. At the same time, the number of operating refineries in the Unites States
decreased, from 194 in 1990 to 144 in 2002. Thus, while the total number of U.S. refineries
decreased, the average capacity at individual refineries has increased. According to the Energy
Information Administration, capacity per operating refinery increased by 28 percent from 1990
to 1998.
Emissions are generally expected to increase as production and capacity increase. However, the
opposite has occurred for the petroleum refining industry - emissions of aggregate and targeted
air toxics have decreased dramatically, while total production and capacity levels in the industry
increased. Thus, the decreases in emissions from petroleum refineries cannot be explained by
industry trends in production or capacity.
Other Clean Air Act Regulations May Have Impacted Air Toxics Emissions from
Refineries
The refining industry is subject to a number of CAA requirements other than the Petroleum
Refinery MACT, and these requirements may have impacted air toxics emissions at refineries.
These regulations include New Source Performance Standards, Prevention of Significant
Deterioration/New Source Review, reformulated gasoline rules, and other MACT standards.
An industry association regulatory analyst told us that, depending upon its operation and
configuration, a refinery could be subject to more than 15 different MACT standards, including
the Hazardous Organic NESHAP, the Miscellaneous Organic NESHAP, and the Cooling Towers
MACT. Representatives of two of the refineries we contacted said they are subject to the
39
-------
Hazardous Organic NESHAP as well as the Petroleum Refinery MACT, and that the Hazardous
Organic NESHAP has had an impact on emissions at the facility.
The National Petroleum Refinery Initiative Is Expected to Reduce Benzene and
Volatile Organic Compound (VOC) Emissions
EPA's National Petroleum Refinery Initiative is an integrated compliance and enforcement
strategy to address air emissions from the petroleum refining industry. The Refinery Initiative is
focused on the four most significant CAA compliance issues for the industry: New Source
Review/Prevention of Significant Deterioration; Flaring/New Source Performance Standards;
LDAR; and the Benzene NESHAP. The Refinery Initiative does not focus directly on the
Petroleum Refinery MACT standard. However, LDAR and the Benzene NESHAP are
incorporated into the MACT standard; in order to comply with the Petroleum Refinery MACT,
facilities must also comply with these standards.
Since March 2000, EPA has entered into 17 consent decree settlements with U.S. companies that
refine nearly 77 percent of the nation's petroleum. These settlements cover 85 refineries in
25 States and, when fully implemented, should result in annual emissions reductions of
approximately 80,000 tons of nitrogen oxides and approximately 235,000 tons of sulfur dioxide.
The consent decrees are expected to result in reductions of fugitive emissions of VOCs,
including benzene. It is uncertain if any reductions in benzene emissions resulting from the
Initiative are reflected in the 2002 NEI figures. Over half of the affected refining capacity was
not yet covered under the settlements in 2002.
Two of three refineries we contacted have entered into consent decrees with EPA under the
Refinery Initiative. The third refinery has a consent decree with EPA covering similar issues,
but a refinery official told us that it is separate from the National Refinery Initiative. The
consent decrees for all three of these facilities address LDAR, and officials from two of the
facilities told us that enhanced LDAR activities have reduced emissions from their facilities. For
example, one facility representative told us that the consent decree reduced the leak level for
which repairs are required from 10,000 parts per million to 500 parts per million. This has led to
a reduction in fugitive emissions, which the representative told us is reflected in the facility's
TRI figures for 2005.
Results for Sample Facilities
We selected three petroleum refineries for more in-depth review of emission changes, actions
taken to comply with the MACT, external factors impacting emissions, compliance and
enforcement activities, and use of compliance assistance tools. We selected these facilities based
on the highest reported emissions of benzene from the 2002 NEI Version 1. We selected
benzene because EPA considers it to be the air toxic posing the largest potential risk from the
refining industry. Since these facilities were not randomly selected, the experiences of these
facilities, as described below, may not be representative of the industry as a whole.
EPA revised the 2002 NEI after we selected our sample facilities and after we had conducted the
majority of our field work. We updated our analyses to reflect the revised NEI emissions (i.e.,
Version 3) as described in the following sections.
40
-------
Facility A
Facility A is one of the largest refineries in the United States, ranking in the top 10 percent in
terms of capacity. As of January 2006, Facility A had an operating capacity of almost 290,000
barrels per day. The refinery is located on the Gulf of Mexico, and its primary products are low
sulfur gasoline, diesel fuel, and jet fuel. An environmental engineer from Facility A told us that
the refinery is located in an area that was designated as non-attainment in the 1980s under the
National Ambient Air Quality Standards for ozone. As a result, the facility already had
significant controls in place when the MACT was implemented, and did not have to make a large
number of changes to some of the refinery's processes. The engineer told us that in order to
comply with the MACT, the facility added controls to its truck racks, which were undergoing an
expansion, and upgraded controls on its storage tanks. It also expanded and upgraded its LDAR
controls and made adjustments to its startup, shutdown, and malfunction plan.
This facility's NEI Version 3 emissions show a decrease in emissions from 1990 to 2002 for 9 of
the 10 air toxics targeted by the Petroleum Refinery MACT. Reported decreases in the 9
pollutants ranged from about 15 percent to 98 percent,19 with emissions of 8 of the 10 targeted
air toxics reduced by more than 50 percent. We were unable to determine if emissions of
2,2,4-Trimethylpentane decreased due to missing data in the 2002 NEI. This facility's updated
2002 NEI Version 3 emissions are relatively consistent with the facility's TRI-reported
emissions. Benzene emissions do not show a marked decrease from 1990 levels. According to a
facility engineer, the refinery made expansions in the mid-1990s that involved benzene sources.
He also told us that after 2001, the refinery began identifying and reporting more leaks under
LDAR, which includes benzene emissions.
The following table shows 1990 NEI, 2002 NEI (Versions 1 and 3), and 2002 TRI-reported
emissions for all targeted pollutants.
19
1990 NEI compared to 2002 NEI Version 3.
41
-------
Table D-2: Emissions of Targeted Air Toxics for Facility A
20
Targeted Air Toxic a
2,2,4-Trimethylpentane
Benzene
Cresol c
Ethylbenzene
Hexane a
Methyl Tert Butyl Ether
Naphthalene
Phenol
Toluene
Xylenes6
1990 NEIb
204.3
54.4
4.6
27.7
214.8
28.6
6.8
1.7
127.0
195.6
2002 NEI
Version 1 b
NA
55.8
0.1
11.5
11.7
20.6
1.0
0.2
35.0
178.3
2002 NEI
Version 3 b
NA
46.4
0.1
10.0
8.0
13.9
0.8
0.1
29.2
130.3
Percent
Reduction'
14.7%
97.8%
63.9%
96.3%
51 .4%
88.2%
94.1%
77.0%
33.4%
2002 TRI b
NA
44.0
0.1
10.7
6.8
10.6
0.8
0.2
30.8
44.5
Primary air toxics targeted by the Petroleum Refinery MACT per the Federal Register.
bTons per year.
CTRI amount represents Cresols (Mixed Isomers).
d TRI amount represents n-Hexane.
eTRI amount represents Xylenes (Mixed Isomers).
f 1990 NEI compared to 2002 NEI Version 3.
NA= Reported emissions not available for this air toxic.
Source: Developed by OIG staff from NEI and TRI databases
Facility B
Facility B is a mid-sized refinery with a capacity of about 112,000 barrels per day. It is located
in the Midwest, and its primary products include gasoline and diesel fuels. The facility was
purchased out of bankruptcy by the current owners in 2004. Officials from the facility told us
they were not employed by the refinery at the time the Petroleum Refinery MACT standard was
implemented. Based on facility records and discussions with other employees, they believe the
previous owners took several actions to comply with the MACT. These actions included:
• Rebuilding the loading rack and installing vapor controls;
• Adding a flare to the coker, and modifying a number of atmospheric vents so that
emissions were treated and sent to the coker flare;
• Increasing LDAR monitoring; and
• Stopping the discharge of slop oil into the sewer, which resulted in a significant drop in
air toxics emissions.
The facility's NEI emissions data show that 2 of the 10 targeted air toxics increased from 1990 to
2002, while 5 of the 10 decreased. Data for 3 of the 10 targeted air toxics are missing and thus
we were unable to determine if emissions decreased for these pollutants. Benzene emissions
increased slightly during this period, and hexane emissions increased by over 47 percent.
Refinery officials attributed the increase in these pollutants to an increase in production at the
facility in 1998. For the 5 targeted air toxics showing emissions decreases, reductions ranged
from about 83 percent to 99 percent. The refinery official attributed this overall decrease to a
20 The name of Facility A changed between the 1990 and 2002 NEI. We included emissions only from those records
that matched according to the State Facility Identification or the Federal Registry Identification in the different
inventories.
42
-------
combination of factors, including permitting, offsets to avoid triggering Prevention of Significant
Deterioration requirements, and the MACT standard.
The following table shows 1990 NEI, 2002 NEI (Versions 1 and 3), and 2002 TRI-reported
emissions for all targeted pollutants.
Table D-3: Emissions of Targeted Air Toxics for Facility B
Targeted Air Toxic a
2,2,4-Trimethylpentane
Benzene
Cresol
Ethylbenzene
Hexane c
Methyl Tert Butyl Ether
Naphthalene
Phenol
Toluene
Xylenes0
1990 NEI b
97.4
25.9
NA
13.2
102.4
13.7
3.2
0.8
60.5
51.4
2002 NEI
Version 1 b
0.8
26.1
NA
0.1
150.9
NA
0.1
NA
6.3
16.7
2002 NEI
Version 3 b
0.8
26.1
NA
0.1
150.9
NA
0.1
NA
6.3
8.7
Percent
Reduction e
99.2%
(+ .8%)
99.2%
(+ 47.4%)
96.9%
89.6%
83.1%
2002 TRI
b
NA
26.2
NA
0.1
157.5
NA
NA
NA
6.3
8.7
Primary air toxics targeted by the Petroleum Refinery MACT per the Federal Register.
bTons per year.
CTRI amount represents n-Hexane.
dTRI amount represents Xylenes (Mixed Isomers).
e1990 NEI compared to 2002 NEI Version 3.
NA = Reported emissions not available for this air toxic.
Source: Developed by OIG staff from NEI and TRI databases
Facility C
Facility C is one of the largest refineries in the United States, ranking in the top 10 percent in
terms of capacity. As of January 2006, Facility C had an operating capacity of 330,000 barrels
per day. The refinery is located on the Gulf of Mexico, and its primary products are gasoline, jet
fuel, and diesel fuel.
Officials from Facility C told us they took a number of actions in order to comply with the
Petroleum Refinery MACT standard. These actions included:
• Installing seals on their floating roof racks;
• Adding a water scrubber to a process vent;
• Installing closed loop sampling systems;
• Implementing LDAR requirements at a number of emission units; and
• Installing a lean absorption system to comply with the Marine Vessel Loading MACT
(Subpart Y), which is incorporated into the Petroleum Refinery MACT.
According to the NEI, emissions of 9 of the 10 air toxics targeted by the MACT decreased
between 1990 and 2002. We were unable to determine if emissions of cresol decreased because
of missing data. Benzene emissions from Facility C decreased significantly between 1990 and
2002, from 128.4 tons to 37.1 tons (about 71 percent). Reductions for the other air toxics
43
-------
targeted by the MACT ranged from about 1 percent to nearly 100 percent. Emissions of eight of
the targeted air toxics decreased by more than 50 percent.
However, it is possible that emissions decreased even more than is indicated by the NEI since the
NEI emissions for 2002 are based on the facility's potential to emit (i.e., the maximum level of
emissions the facility could emit) for that year, rather than actual emissions. State agency staff
who provided the NEI emissions data to EPA told us that they did not have the facility's actual
emissions data for 2002. Instead, they reported the facility's potential to emit. The facility's
2002 TRI-reported emissions were considerably lower than the 2002 NEI emissions for the
targeted air toxics we compared.
The following table shows 1990 NEI, 2002 NEI (Versions 1 and 3), and 2002 TRI-reported
emissions for all targeted pollutants.
Table D-4: Emissions of Targeted Air Toxics for Facility C
Targeted Air Toxic a
2,2,4-Trimethylpentane
Benzene
Cresol
Ethyl benzene
Hexanec
Methyl Tert Butyl Ether
Naphthalene
Phenol
Toluene
Xylenes0
1990 NEI b
482.2
128.4
NA
65.3
507.0
67.6
16.0
4.0
299.7
492.1
2002 NEI
Version 1 b
2.3
37.1
0.0
27.5
90.5
4.8
2.1
0.0
122.2
488.9
2002 NEI
Version 3 b
2.3
37.1
0.0
27.5
90.5
4.8
2.1
0.0
122.2
488.9
Percent
Reduction e
99.5%
71.1%
57.9%
37.8%
82.1%
92.9%
86.9%
100.0%
59.2%
0.7%
2002 TRI b
NA
27.0
0.0
8.8
52.5
NA
1.7
0.0
43.5
103.4
Primary air toxics targeted by the Petroleum Refinery MACT per the Federal Register.
bTons per year.
CTRI amount represents n-Hexane.
d TRI amount represents sum of m-Xylene, o-Xylene, and p-Xylene.
e 1990 NEI compared to 2002 NEI Version 3.
NA = Reported emissions not available for this air toxic.
Source: Developed by OIG staff from NEI and TRI databases
44
-------
Appendix E
Detailed Results - Primary Aluminum MACT
The United States aluminum supply is comprised of three basic sources:
• Primary aluminum (domestic production from ore material);
• Imported aluminum (of primary and secondary ingot and mill products); and
• Recycled aluminum (metal recovered from scrap, also known as secondary recovery).
Companies in the primary aluminum industry are primarily engaged in making aluminum from
alumina. Alumina is refined from bauxite ore. The companies in this industry may also roll,
draw, extrude, or cast the aluminum they make into primary forms. These forms can be bars,
billets, ingots, plates, rods, sheets, or strip. Companies in this industry may make primary
aluminum or aluminum-based alloys from alumina.
The intent of the 1997 Primary Aluminum MACT was to reduce emissions of the two targeted
air toxics, hydrogen fluoride and POM. Facilities were expected to be in compliance by October
7, 1999. The MACT added extensive monitoring and reporting requirements. Monitoring
requirements included:
• Monthly measurements of total fluorides secondary emissions (unless one obtains a
monitoring alternative) with annual measurements of primary emissions;
• Quarterly measurements of POM secondary emissions from each Soderberg potline with
annual measurements of primary emissions;
• Continuous parameter monitoring system (e.g., scrubber air flow rate, dry alumna flow
rate, water flow rate, voltage, etc.) for each emission control device; and
• Monitoring device(s) to determine the daily weight of aluminum produced.
Reporting requirements included:
• Start-up, shutdowns, and malfunctions reports;
• Annual reports on performance tests done after the initial performance test; and
• Semi-annual excess emissions reports.
Aggregate and Targeted Air Toxics for the Primary Aluminum Industry Decreased
Between 1990 and 2002
NEI data indicates that hydrogen fluoride and two subcategories of POM (PAH 7 and PAH 15)
emissions declined for this MACT sector between 1990 and 2002. Hydrogen fluoride emissions
decreased by 27.6 percent. PAH 15 emissions decreased by 91.4 percent, and PAH 7 emissions
declined by 94.5 percent. For hydrogen fluoride, this is less than the reduction EPA estimated
for total fluorides when it issued the Primary Aluminum MACT standards. EPA had estimated
45
-------
that total fluoride emissions would decrease by 46 percent. EPA estimated an approximately
91 percent decrease in POMs, which includes PAH 15 and PAH 7.
21
Table E-1: NEI Emissions of Targeted Air Toxics for the Primary Aluminum MACT
Targeted Air Toxic a
Hydrogen Fluoride
PAHC 15
PAHC 7
Totals
1990 NEI b
2,476.00
427.5
274.2
2,903.5
2002 NEI
Version 3 b
1,792.30
36.7
15.3
1,829.0
Change
(2002-1990)"
-683.7
-390.7
-258.9
1,074.50
Percent
Change b
-28%
-91%
-94%
-37%
Primary air toxics targeted by the MACT per the Federal Register.
bTons per year.
0 Polycyclic aromatic hydrocarbons.
d Total represents the sum of Hydrogen Fluoride and PAH 15 since PAH 7 is a subset of PAH 15.
Source: Developed by OIG staff from NEI databases
Decreases in Production and Changes in Emission Reporting Impact Emission
Trends
Two unrelated factors should be considered in analyzing emission trends for the primary
aluminum industry. First, there was a decline in U.S. capacity and production, which would be
expected to cause industry-wide emissions to decrease. Second, industry representatives told us
that the methodology used to estimate emissions for TRI reporting was revised after the MACT
implementation date in 1999 because company officials realized they could provide more
accurate emissions data. Company representatives told us that prior to the implementation of the
Primary Aluminum MACT, their company relied on emissions factors to estimate emissions.
Following the implementation of the MACT, more reliable monitoring became available. The
representatives told us that they discovered that the use of the emissions factors resulted in the
underreporting of emissions. A change in methodology to estimate emissions complicates
analyzing emissions trends. We do not know the extent to which reliable methodologies were
used to develop a facility's emissions total in the NEI. EPA representatives told us that the NEI
does not always identify the methodology used to calculate the amount of emissions. If a less
reliable methodology, such as an emission factor, is still being used in the current NEI, it could
result in underreporting of emissions.
Downward Trend in Capacity and Production in the Industry
The primary aluminum industry is subject to world markets, and United States production has
decreased between 1990 and 2002. According to the Vice President for Environmental Health
and Safety of the Aluminum Association, there has been a "dramatic downward trend" in U.S.
capacity and production in the industry. Primary aluminum production decreased 33.1 percent
from 1990 and 2002. Between 1999, the compliance date of the MACT, and 2002, industry wide
U.S. production decreased by 27.1 percent.
21 The Federal Register Notice for the Primary Aluminum MACT identified a 50-percent reduction in HAPs from
11,000 tons per year to 5,500 tons per year. EPA's estimated HAPs and reductions are significantly more than the
emissions in the NEI for 1990 and 2002.
46
-------
These decreases in production are influenced by rising energy costs, particularly in the
Northwest, where some facilities have shut down. Since the compliance date of the MACT in
1999, several facilities have closed temporarily or permanently. According to the U.S.
Geological Survey, in 1999, 23 primary aluminum facilities were in operation. In contrast, in
2002, 16 primary aluminum plants were in operation, and 7 were temporarily or permanently
idled. At the end of 2002, about 1.5 million metric tons per year of domestic primary smelting
capacity, equivalent to 35 percent of total capacity, was closed.
Figure E-1: Primary Aluminum Production in the United States, 1990 - 2002
4 500
t/> '
£
0)
E^ nnn
•5
•c 2 500
ro
§9 nnn
£
c
'•=r 1 ^nn
o
IB
° 1 nnn
E
0- ^nn
n
1990
1991
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
Source: Developed by OIG staff from U.S. Geological Survey data
Improved Emission Estimates
Improvements in the methodology for estimating emissions have impacted reported emissions
trends since the MACT compliance date. Prior to implementation of the MACT, industry
facilities used emissions factors to estimate the amount of air toxics emitted by their facilities
and reported in the TRI. However, the MACT required facilities to conduct self-monitoring and
reporting. This additional monitoring resulted in companies obtaining more accurate information
about their emissions. Representatives from two of the three facilities we contacted told us that
they revised their methodology for reporting emissions after the MACT was implemented. They
told us that the revised methodology resulted in higher reported emissions than what they had
reported under the prior methodology.
Results for Sample Facilities
We selected the three facilities with the largest reported emissions of POM in the 2002 NEI
(Version 1) for review. Since these facilities were not randomly selected their experiences may
not be representative of the industry as a whole. In particular, production at two of these
facilities increased while production for the industry overall has decreased. Further, NEI and
47
-------
TRI data show an increase in hydrogen fluoride emissions for each of these three facilities. This
is contrary to overall industry trends, which indicate an overall decrease in hydrogen fluoride
emissions.
EPA revised the NEI after we selected our sample facilities, and had conducted the majority of
our field work. We updated our analyses to reflect the revised NEI emissions (i.e., Version 3).
However, the changes from Version 1 to Version 3 were minor and were not noticeable when the
data was converted from pounds to tons.
Facility D
Facility D is an aluminum reduction plant built in 1979. Company officials told us they have
been operating at or near capacity since 1999. Further, this facility was one of the best
performing facilities in terms of emission rates when the MACT was developed. Thus, this
facility's emission rates were used to establish the MACT floor for the entire industry. In
response to the MACT, the facility took steps to meet industrial ventilation criteria, installed
parametric monitors to conduct continuous monitoring, added computers for data collection and
reporting systems, and updated inspection practices.
According to the NEI Version 3 data, hydrogen fluoride emissions increased from 20.4 to
63.3 tons per year (210.3 percent) between 1990 and 2002 for this facility. PAH-15 emissions
decreased from 5.4 to 3.5 tons per year (35.2 percent) during the same time period. We also
compared the facility's 2002 NEI emissions to the facility's 2002 TRI self-reported emissions.
The TRI-reported emissions for hydrogen fluoride were close to the number reported in NEI.
Company officials could not explain the increase in EPA's NEI emissions for hydrogen fluoride
since they did not know how the NEI numbers were developed. We were unable to determine
how the NEI emissions were developed. An Agency official noted problems with the accuracy
of the 1990 NEI. The company officials told us that their TRI-reported emissions have remained
constant from 1997 through 2004. They told us that even though facility production has
increased, reported emissions have remained constant. Thus, the facility's emission rate for
hydrogen fluoride (amount of emissions per unit of production) has decreased. Company
officials told us that, in their opinion, the MACT has resulted in emissions reductions for the
industry, particularly at the older facilities.
The following table shows 1990 NEI, 2002 NEI (Versions 1 and 3), and 2002 TRI-reported
emissions for all targeted pollutants.
48
-------
Table E-2: Emissions of Targeted Air Toxics for Facility D
Targeted Air Toxics a
PAH c 15
PAHC7
Hydrogen fluoride
Totals e
1990 NEIb
5.4
2.1
20.4
25.8
2002 NEI
Version 1 b
3.5
3.5
63.3
66.8
2002 NEI
Version 3 b
3.5
3.5
63.3
66.8
Percent
Reduction d
35.2%
(+66.7%)
(+210.3%)
(+158.9%)
2002 TRI b
NA
NA
62.5
Not
Applicable
Primary air toxics targeted by the MACT per the Federal Register.
bTons per year.
0 Polycyclic aromatic hydrocarbons.
d 1990 NEI compared to 2002 NEI Version 3.
e Total represents the sum of Hydrogen Fluoride and PAH 15 since PAH 7 is a subset of PAH 15.
NA = Emissions for this subcategory of polycyclic organic matter is not available in the TRI.
Source: Developed by OIG staff from NEI databases
Facility E
Facility E is an aluminum reduction plant built in 1957. According to facility representatives,
they made several changes in response to the 1997 Primary Aluminum MACT. For example,
they installed a pitch fume scrubber, started daily testing, and maintained daily logs. Facility
representatives told us they switched from using pencil pitch to liquid pitch around 1999-2000
and this resulted in a substantial reduction in polycyclic aromatic compounds.22 The change in
pitch type reduced fugitive emissions.
This facility's hydrogen fluoride emissions increased from 74.7 to 119.2 tons per year
(59.6 percent) between 1990 and 2002, according to the NEI, and the facility's PAH-15
emissions decreased from 20 to 8.4 tons per year (58 percent) between 1990 and 2002. We also
compared the facility's 2002 NEI emissions to the facility's 2002 TRI self-reported emissions.
The reported emissions for hydrogen fluoride were very similar in both databases.
Facility E representatives told us that the company changed the way TRI-reported emissions
were calculated, resulting in higher emission estimates despite installing additional controls.
Earlier, Facility E used AP-42 emissions factors to calculate its TRI emissions, but in 2001
changed to collecting actual emissions data from monitoring. They found that estimated
emissions based on emissions factors were "a little bit lower" than emissions based on
monitoring data. The company also reported confusion about reporting polycyclic aromatic
compounds in the TRI. In their opinion, the MACT reduced emissions because it forced them to
pay more attention to daily compliance and fugitive emissions.
The following table shows 1990 NEI, 2002 NEI (Versions 1 and 3), and 2002 TRI-reported
emissions for all targeted pollutants.
! Polycyclic aromatic compounds include PAH 7 and PAH 15 and are reported in the TRI.
49
-------
Table E-3: Emissions of Targeted Air Toxics for Facility E
Targeted Air Toxic a
PAH c 15
PAHC7
Hydrogen fluoride
Totals e
1990 NEI
20
7.8
74.7
94.7
2002 NEI
Version 1
8.4
8.4
119.2
127.6
2002 NEI
Version 3
8.4
8.4
119.2
127.6
Percent
Reductiond
58.0%
(+7.7%)
(+59.6%)
(+34.7%)
2002 TRI
NA
NA
119.3
Not
Applicable
Primary air toxics targeted by the MACT per the Federal Register.
bTons per year.
c Polycyclic aromatic hydrocarbons.
d 1990 NEI compared to 2002 NEI Version 3.
e Total represents the sum of Hydrogen Fluoride and PAH 15 since PAH 7 is a subset of PAH 15.
NA = Emissions for this subcategory of polycyclic organic matter is not available in the TRI.
Source: Developed by OIG staff from NEI and TRI databases
Facility F
Facility F is an aluminum reduction plant that started producing aluminum in 1971. According
to facility officials, production at their plant has increased 14 percent since 1999. They told us
that in response to the MACT, they (1) installed a scrubber on their carbon based oven,
(2) modified bag houses, (3) and implemented monitoring methods.
Based on the NEI data, hydrogen fluoride emissions increased from 88.4 to 143.1 tons per year
(61.9 percent) between 1990 and 2002. The facility's PAH-15 emissions decreased during this
period from 23.6 to 6.8 tons per year (71.2 percent), according to NEI data. We also compared
the facility's 2002 NEI emissions to the facility's TRI self-reported emissions. The TRI-reported
emissions were slightly less than the amount in the NEI.
Facility F representatives told us that their method for reporting hydrogen fluoride emissions in
TRI has improved since 1990. They are now using actual emissions testing data rather than
emissions factors. These officials told us that emissions of hydrogen fluoride were
underreported until they started using testing data to report emissions. They also noted that there
was some confusion in the past regarding the reporting of polycyclic aromatic compound
numbers. In their opinion, the MACT reduced emissions for this industry since the majority
facilities within the industry had to make changes to come into compliance.
The following table shows 1990 NEI, 2002 NEI (Versions 1 and 3), and 2002 TRI-reported
emissions for all targeted pollutants.
50
-------
Table E-4: Emissions of Targeted Air Toxics for Facility F
Targeted Air Toxic a
PAHC15
PAHC7
Hydrogen fluoride
Totals e
1990 NEIb
23.6
9.2
88.4
112.0
2002 NEI b
6.8
6.8
143.1
149.9
2002 NEI
Version 3 b
6.8
6.8
143.1
149.9
Percent
Reduction d
71 .2%
26.1%
(+61 .9%)
71 .2%
2002 TRI b
NA
NA
136.2
Not
Applicable
Primary air toxics targeted by the MACT per the Federal Register.
bTons per year.
cPolycyclic aromatic hydrocarbons.
d 1990 NEI compared to 2002 NEI Version 3.
6 Total represents the sum of Hydrogen Fluoride and PAH 15 since PAH 7 is a subset of PAH 15.
NA = Emissions for this subcategory of polycyclic organic matter is not available in the TRI.
Source: Developed by OIG staff from NEI and TRI databases
51
-------
Appendix F
Detailed Results - Pulp and Paper MACT
The Pulp and Paper MACT category includes two types of mills: (1) mills that process raw wood
fiber or recycled fiber to make pulp and/or paper; and (2) mills that convert pulp and paper into
more specialized products such as paperboard boxes, writing paper, and sanitary paper. Pulp and
paper production involves both combustion processes and non-combustion processes. The Pulp
and Paper MACT (MACTs I and III)23 apply to non-combustion sources of pulp and paper
production. MACT I establishes controls for air toxics emissions from mills using the kraft,
sulfite, semi-chemical, and soda pulping processes. MACT III establishes controls for air toxics
emissions from mills using mechanical, secondary fiber, and non-wood pulping, and
papermaking systems at all mills.
The 1998 Pulp and Paper MACT established separate emission limits for pulping system vents,
pulping process condensate streams, and bleaching systems. MACT I and MACT III provide
several alternative emission limits for the pulping vent standard, pulping condensates standard,
and bleaching system standard. For each regulated emission point, a mill can choose the
emission limit it will comply with and can use any emission control technology to achieve
compliance. In general, mills must conduct an initial performance test and then continuously
monitor a set of control devices or process operating parameters. Exceeding a monitoring
parameter constitutes a violation of the standard. Mills must continuously monitor operating
parameters and report all instances where the values deviate from the maximum or minimum
values established during the initial performance test.
The Pulp and Paper MACT targets 13 air toxics that are included in EPA's list of the 33 highest
priority air toxics. These include acetaldehyde, acrolein, carbon tetrachloride, chloroform,
formaldehyde, and methylene chloride, among others. Formaldehyde is estimated to present the
most risk to public health from the pulp and paper production industry.
NEI Data Shows Significant Decrease in Air Toxics Emissions
According to the NEI, from 1990 to 2002, air toxics emissions from pulp and paper facilities
decreased from 213,652 tons to 21,455 tons (90 percent). When EPA developed the Pulp and
Paper MACT it estimated that these standards would reduce emissions by 64 percent. Emissions
for the 13 air toxics specifically targeted by the Pulp and Paper MACT have also decreased. The
following table shows the emission changes for targeted pollutants from 1990 to 2002, based on
the NEI.
23 MACT II (subpart MM) covers chemical production for the pulp and paper industry and had a compliance date in
2004. Our review did not include this MACT since we limited our review to those MACTs with compliance dates
prior to 2002
52
-------
Table F-1: NEI Emissions of Target Air Toxics for the Pulp and Paper MACT
Targeted Air Toxic a
1 ,2,4-Trichlorobenzene
Acetaldehyde
Acrolein
Carbon Tetrachloride
Chloroform
Cresol/Cresylic Acid
(Mixed Isomers)
Cumene
Formaldehyde
Methanol
Methylene Chloride
(Dichloro methane)
Phenol
Propionaldehyde
Xylenes (Mixed Isomers)
Totals
1990 NEI b
5,710.0
10,921.1
283.0
4,270.0
18,500.1
NA
8,270.0
4,124.4
152,292.9
1,740.0
3,060.0
2,040.0
2,440.0
213,651.5
2002 NEI
Version 3 b
74.6
1,432.8
55.5
54.9
463.5
NA
16.0
196.4
18,959.3
79.6
68.0
11.1
43.3
21,455.0
Change b
(2002-1990)
5,635. .4
9,488.3
227.5
4,215.1
18,036.6
8,254.0
3,928.0
133,333.6
1,660.4
2,992.0
2,028.9
2,396.7
192,196.5
Percent
Change
-98.7%
-86.9%
-80.4%
-98.7%
-97.5%
-99.8%
-95.2%
-87.6%
-95.4%
-97.8%
-99.5%
-98.2%
-90.0%
Primary air toxics targeted by the MACT per the Federal Register.
bTons per year.
Source: Developed by OIG staff from NEI database
We did not determine the extent to which these emissions were attributed to the MACT rules.
As explained below, decreases in production had an impact on overall emissions.
Decreases in Production May Have Impacted Overall Emissions
A decrease in U.S. production of paper and paperboard products was one possible reason for an
overall decline in air toxics emissions from this industry. Between 1997 and 2000, U.S. exports
of paper and paperboard products declined 5.5 percent, while imports increased 20 percent.
Capital investments in recent years were well below historic levels, and no new mills have been
built in the United States since 1970. For the first time, U.S. industry capacity declined in 2001.
Review of Sample Facilities
We selected the three facilities with the largest reported emissions of formaldehyde in the 2002
NEI (Version 1) for review. These three pulp and paper mills represent three industry
components: pulp mills, paper mills, and paperboard mills. Since these facilities were not
randomly selected, the experiences of these facilities may not be representative of the industry as
a whole.
Pulp and Paper emissions were not affected by the update from NEI Version 1 to Version 3.
Facility G
Facility G is a paperboard mill. Facility G managers told us that changes in emission calculation
methods and increases in production were two non-MACT related factors that impacted air
toxics emissions. Facility G was granted a 1-year extension to comply with Phase 1 of the
53
-------
MACT standard, resulting in a compliance date of April 16, 2002. Managers told us that Facility
G made several changes to the mill to comply with the MACT. In 2001, Facility G installed a
stream stripper, made improvements to its thermal oxidizer, and made minor improvements to
the non-condensable gas collection system. In 2003, changes were made to the washer system to
comply with Phase 2 of the MACT. Facility G used the Clean Condensate Alternative, shower
water improvements and changes to the tanks and piping were made, and it began using cleaner
water in the shower water system, which helped reduce the amount of methanol emissions in the
wastewater.
According to the NEI, emissions of targeted air toxics decreased significantly for this facility
from 1990 to 2002. We also compared the facility's 2002 NEI emissions to the facility's self-
reported emissions in TRI. The TRI-reported emissions were greater for two pollutants:
acetaldehyde and formaldehyde. However, NEI 2002 methanol emissions were greater than
TRI-reported emission by more than 500 tons per year.
Facility G managers told us that they had made significant changes in their method for
calculating TRI-reported emissions since 1999. Specifically, emissions factors for methanol and
other air toxics have changed, and industry guidelines for calculating emissions have improved
over the years making them more accurate. In their estimation, the majority of the emissions
changes have resulted in decreased reported emissions, since the old emissions factors over-
estimated emissions. However, any decreases in emissions at Facility G resulting from changes
in emissions factors may have been offset by increases in production. The following table shows
1990 NEI, 2002 NEI (Versions 1 and 3), and 2002 TRI-reported emissions for all targeted
pollutants.
Table F-2: Emissions of Targeted Air Toxics for Facility G
Targeted Air Toxic a
1,2,4-
Trichlorobenzene
Acetaldehyde
Acrolein
Carbon Tetrachloride
Chloroform
Cumene
Formaldehyde
Methanol
Methylene Chloride
O-Cresol
Phenol
Propionaldehyde
1990 NEI b
176.2
333.1
8.6
130.2
563.9
255.3
125.8
4,691.7
53.0
NA
94.5
63.0
2002 NEI
Version 1 b
8.2
38.9
4.0
12.1
10.4
NA
17.1
1642.8
4.1
NA
1.1
NA
2002 NEI
Version 3 b
8.2
38.9
4.0
12.1
10.4
NA
17.1
1642.8
4.1
NA
1.1
NA
Percent
Reduction c
95.3%
88.3%
53.5%
90.7%
98.2%
86.4%
65.0%
92.3%
NA
1.1
NA
2002 TRI b
NA
49.9
NA
NA
NA
NA
37.8
1,068.2
NA
NA
1.3
NA
Primary air toxics targeted by the MACT per the Federal Register.
bTons per year.
°1990 NEI compared to 2002 NEI Version 3.
NA = Reported emissions not available for this air toxic.
Source: Developed by OIG staff from NEI and TRI databases
54
-------
Facility H
Facility H is a paper and paperboard mill. Facility H managers told us they had to make process
and control changes to comply with the MACT. Initially the stripper-off gas systems from the
steam stripper were incinerated in the lime kiln. However, variability in the stripper-off gases
thermal value caused swings in the lime kiln, which resulted in damage to the kiln. Facility H
installed a methanol liquefaction system to condense the stripper-off gases and burn the
methanol in the recovery boilers.
According to the NEI, emissions of targeted air toxics decreased significantly for this facility
from 1990 to 2002. We also compared the facility's 2002 NEI emissions to the facility's TRI
self-reported emissions. The 2002 emissions data from the NEI and TRI were very consistent for
this facility. The following table shows 1990 NEI, 2002 NEI (Versions 1 and 3), and 2002 TRI-
reported emissions for all targeted pollutants.
Table F-3: Emissions of Targeted Air Toxics for Facility H
Targeted Air Toxic a
1,2,4-
Trichlorobenzene
Acetaldehyde
Acrolein
Carbon Tetrachloride
Chloroform
Cumene
Formaldehyde
Methanol
Methylene Chloride
O-Cresol
Phenol
Propionaldehyde
1990 NEI bc
94.5
211.4
4.6
69.8
302.4
136.9
67.4
2,731.4
28.4
NA
50.6
33.8
2002 NEI
Version 1 bc
3.3
77.0
3.5
0.1
NA
0.5
14.8
892.9
NA
NA
3.6
0.2
2002 NEI
Version 3 bc
3.3
77.0
3.5
0.1
NA
0.5
14.8
892.9
NA
3.6
0.2
Percent
Reduction
96.5%
63.6%
23.9%
99.9%
99.6%
78.0%
67.3%
92.9%
99.4%
2002 TRI b
NA
77.0
NA
NA
NA
NA
14.8
892.9
NA
NA
3.6
NA
Primary air toxics targeted by the MACT per the Federal Register.
bTons per year.
cThe NEI emissions include totals from two different Pulp and Paper MACT codes: 1626-1 (subpart S)
and 1626-2 (subpart M). In the 1990 NEI and the 2002 NEI Version 1, the majority of emissions for this
facility were included in the 1626-1 code. However, in the 2002 NEI Version 3, the majority of
emissions were included in the 1626-2 code. For this facility, we combined emissions from both codes
in the NEI emissions figures.
d 1990 NEI compared to 2002 NEI Version 3.
NA= Reported emissions not available for this air toxic.
Source: Developed by OIG staff from NEI and TRI databases
Facility I
Facility lisa pulp mill that produces cellulose fibers. According to the NEI, emissions of
targeted air toxics decreased significantly for this facility from 1990 to 2002. However, the data
from the 2002 NEI appeared questionable since the reported emissions for three pollutants
(acetaldehyde, formaldehyde, and methanol) were the same. Further, the NEI emissions were
significantly lower than TRI reported emissions for two of the pollutants.
55
-------
Despite repeated attempts, we were unable to schedule a meeting with representatives from this
facility to discuss the MACT's impact on their facility's emissions. The following table shows
1990 NEI, 2002 NEI (Versions 1 and 3), and 2002 TRI-reported emissions for all targeted
pollutants.
Table F-4: Emissions of Targeted Air Toxics for Facility I
Targeted Air Toxic a
1,2,4-
Trichlorobenzene
Acetaldehyde
Acrolein
Carbon Tetrachloride
Chloroform
Cumene
Formaldehyde
Methanol
Methylene Chloride
O-Cresolc
Phenol
Propionaldehyde
1990 NEI b
26.3
49.7
1.3
19.4
84.1
38.1
18.8
700.0
7.9
NA
14.1
9.4
2002 NEI
Version 1 b
NA
22.2
NA
NA
NA
NA
22.2
22.2
22.2
NA
NA
NA
2002 NEI
Version 3 b
NA
22.2
NA
NA
NA
NA
22.2
22.2
22.2
NA
NA
NA
Percent
Reduction d
55.3%
(+18.1%)
96.8%
(+181.0%)
2002 TRI b
NA
53.2
NA
NA
NA
NA
7.6
188.2
NA
19.7
1.3
NA
a Primary air toxics targeted by the MACT per the Federal Register.
bTons per year.
CTRI amount represents Cresol (Mixed Isomers).
d1990 NEI compared to 2002 NEI Version 3.
NA= Reported emissions not available for this air toxic.
Source: Developed by OIG staff from NEI and TRI databases
56
-------
Appendix G
MACT Compliance Monitoring, Enforcement,
and Compliance Assistance Activities
Our evaluation objectives included the following questions:
• Is there a relationship between the quantity and quality of compliance monitoring and
enforcement and the effectiveness of MACT standards in achieving the desired emissions
reductions?
• Is there a relationship between the adequacy of compliance assistance tools and the
effectiveness of MACT standards in achieving the desired emissions reductions?
Limitations to the amount and quality of data regarding compliance monitoring and compliance
assistance activities prevented us from quantitatively assessing the impact of these activities on
MACT effectiveness. Instead we reviewed results from previous studies and initiatives, and
discussed the impact of these activities on MACT effectiveness with representatives from
selected facilities and States. The sections below discuss the results of our review.
Results of Statistically Valid Compliance Rate Studies
We identified two statistically valid compliance rate studies that have been conducted for MACT
industries, one by EPA and the other by the California Air Resources Board. Both of these
studies found high noncompliance rates for their respective industries.
In Fiscal Year 2002, EPA conducted a statistically valid study of compliance with the Ethylene
Oxide Sterilizer MACT. This is the only statistically valid study EPA has conducted for the
MACT standards. Generally, sources that operate ethylene oxide sterilizers are smaller
stationary sources that sterilize or fumigate medical supplies, pharmaceuticals, and/or spices.
EPA randomly selected 70 facilities to inspect out of a universe of 222 facilities. The study
found that 50.8 percent of the selected facilities were out of compliance with the MACT
standard.
In a statistically valid study conducted in 1999 and 2000, the California Air Resources Board
found significant noncompliance with California's Chrome Plating Airborne Toxic Control
Measure, the State's equivalent to the Federal Chrome Electroplating MACT. Inspectors in five
air districts conducted on-site inspections of 188 chrome platers and found 73 percent of them
had some type of violation. The majority of the violations pertained to monitoring,
recordkeeping, and reporting: 68 percent of the facilities had a non-emission-related violation;
39 percent had an emission-related violation (a violation that could potentially result in an
emission exceedance); and 8 percent had a direct excess emission violation (an emission
exceedance witnessed by the inspector).
In a followup to the Chrome Plating compliance rate study, the California Air Resources Board
and the individual air districts conducted a number of targeted compliance-related activities for
57
-------
the chrome platers in the study. After conducting these permitting, training, enforcement, and
outreach activities, the facilities were inspected again in 2003. The California Air Resources
Board found that the noncompliance rate at these facilities had decreased from 78 percent in
1999-2000 to 14 percent in 2003.
EPA's Targeted MACTInitiatives Have Found Noncompliance with MACT
Standards
OECA has included air toxics as a National Enforcement Priority since Fiscal Year 2000. For
the planning cycle Fiscal Years 2005-2007, OECA implemented an air toxics initiative in which
each region selects MACT source categories for targeted compliance investigations and
enforcement. EPA's original goal under this initiative was to achieve air toxics emissions
reductions of 12,000 pounds each year from these targeted MACTs, for a total reduction of
36,000 pounds by the end of Fiscal Year 2007. However, reported emissions reductions for
Fiscal Years 2005 and 2006 under the initiative have greatly exceeded this goal. In Fiscal Year
2005, total reported air toxics emission reductions from regionally targeted cases were 332,000
pounds; for Fiscal Year 2006, they were 356,000 pounds. Most of the violations and resulting
emissions reductions were from facilities subject to the Secondary Aluminum MACT, the
Hazardous Organic NESHAP, and the Pharmaceutical MACT.
Based on its enforcement activity in Fiscal Years 2004-2005 related to the national air toxics
strategy, EPA found significant noncompliance with the LDAR requirements of the MACT
standards. Twenty-two MACTs, including the Petroleum Refining MACT and the Hazardous
Organic NESHAP, have LDAR requirements. In response to this noncompliance, OECA is
conducting a national targeted initiative in Fiscal Year 2006-2007 to focus on MACT equipment
leak requirements.
Impact of Self-Monitoring and Reporting Requirements on Compliance and
Emissions Reductions
Representatives from five of the eight sampled facilities we interviewed told us that self-
monitoring requirements were important to achieving compliance with MACT requirements and
gaining emissions reductions. In general, the facility representatives we spoke with told us that
internal compliance monitoring, rather than State-conducted on-site inspections, is what drives
compliance and emissions reductions. They told us that the self-monitoring required by the
MACT standard can help a facility identify a problem before it becomes a violation, and can help
ensure that the facility fixes problems more quickly.
The only MACT violations at the nine sample facilities we reviewed were discovered through
internal facility monitoring and self-reported by the facilities to State agencies. All three of the
primary aluminum facilities have reported violations to the State, two for emissions exceedances
and one for failure to record startup and shutdown events. One of the petroleum refineries also
self-reported violations pertaining to startup and shutdown maintenance.
Industry representatives told us that self-reporting requirements enable regulatory agencies to
become aware of MACT violations that they otherwise may not detect. For example, one
58
-------
refinery representative told us that finding violations would require review of the facility's
extensive records, noting that there are about 250,000 components in the LDAR database alone.
He said the State does not have the time or expertise to inspect at that degree of specificity.
Compliance and enforcement staff from Kansas told us that self-reporting is very important
because the State does not have the number of staff nor expertise to find every violation at
petroleum refineries. They said the refineries have more knowledge and expertise about their
operations than Kansas' inspectors, and this makes it difficult for them to determine if the facility
is in compliance. According to one Kansas staff member, the State has to have a lot of trust in
the facility. If a facility reports that it is in compliance, the State has to accept that unless their
inspectors find otherwise.
Sfafe and Industry Representatives Generally Find Compliance Assistance Useful
EPA developed several compliance assistance tools for each of the three MACTs we selected for
in-depth analysis. These tools24 include such documents as plain language guides, applicability
flowcharts, and Questions and Answers documents. Six of the seven States and six of the eight
facilities we spoke with had used one or more of EPA's compliance assistance tools in
implementing the three MACTs in our sample. Representatives from the six States that used the
tools, and from five of the six facilities that used the tools, said that they were useful for
implementing the MACT. The petroleum refinery representatives we spoke with said they found
EPA's Questions and Answers document particularly helpful.
The results of EPA-conducted online surveys on the usefulness of its compliance assistance tools
generally showed good results. In 2004 and 2005, EPA surveyed industry and regulators on the
usefulness and effectiveness of EPA's compliance assistance tools for four MACT standards: the
Aerospace MACT, Plywood MACT, Reciprocating Internal Combustion Engines MACT, and
Industrial Boiler MACT. In general, the surveys found that the tools were obtained and used by
a large portion of the respondents. Most of the tools were rated as "good" or "very good" in
terms of usefulness by a majority of respondents.
State agency staff and industry managers told us that compliance assistance tools could be
improved in some areas. For example, EPA did not issue some of the tools until after the
compliance date of the MACT. Representatives from three States told us that some of EPA's
inspection checklists are of limited value because they cover too much information or are too
complicated to use. In addition, an official from one of the refineries told us that many of the
tools are labeled "Draft" or included a disclaimer. Thus the guidance may not be very helpful
when it comes to inspections or enforcement. For example, the Petroleum Refining MACT
Standard Guidance contains a disclaimer that the document is only guidance and does not create
any enforceable rights.
24
A complete list of the tools developed by EPA for these three MACTs is located in Appendix G.
59
-------
Sfafe Representatives Indicated that Compliance Assistance Is More Important
for Smaller Facilities
In general, State agency staff told us compliance assistance is most important and useful for
smaller facilities. They explained that smaller sources may not be accustomed to being
regulated, and that compliance assistance can be critical for effective implementation of the
standards at these facilities. In contrast, larger facilities often have environmental staff to
implement the regulations and are often members of industry associations that provide
environmental compliance training.
60
-------
Appendix H
Compliance Assistance Tools Developed by EPA
for Three Sample MACTs
Petroleum Refinery MACT (Subpart CC)
Petroleum MACT Standard Guidance - Revised to Include Rule Amendments (November
2000)
• Petroleum MACT Standard Guidance - Revised to Include Rule Amendments
(November 2000). This document includes:
o An overview of the standard
o A summary table of control requirements
o Applicability flowcharts
o A summary of testing, monitoring, and reporting requirements
o A summary table of the overlap of the MACT with other regulations
o Additional resources and contacts
o A glossary of terms and definitions used in the standard
o A summary of amendments to the standard
• Summary of Testing, Monitoring, Recordkeeping, and Reporting Requirements of
40 CFR 63 Subpart CC Petroleum Refineries NESHAP (November 2000)
• Benzene NESHAP FAQ Handbook for Subparts BB and FF (September 1997)
• Questions and Answers for the Refinery MACT I Rule (November 1998)
• List of Petroleum Refineries Affected by the Petroleum Refineries Standards (September
1997)
• Petroleum Refinery MACT - Compliance Assistance/Inspection Checklist (November
2000)
• Petroleum Refinery MACT - Determining Applicability (November 2000)
Primary Aluminum MACT (Subpart LL)
• Plain Language Guide to the Primary Aluminum NESHAP (October 1999). This
document includes:
o An overview of the standard
o An applicability flowchart
o A compliance timeline
o Summary tables of monitoring, recordkeeping, and reporting requirements
o A summary of test methods and calculations
o A schedule of notification and reporting due dates
o Example reports and notifications
o Contacts for additional help and information
o An index for the standard, cross-referenced to the Federal Register notice
o A description of the primary aluminum process and emission points
o A site-specific test plan
61
-------
• Inspection Checklist - Primary Aluminum NESHAP (September 2001)
• 1996 Primary Aluminum Source List
Pulp and Paper MACT (Subpart S)
• Pulping and Bleaching System NESHAP for the Pulp and Paper Industry: A Plain
English Description (September 2001). This document includes:
o An overview of the standard
o A list of Federal Register notices pertaining to the MACT
o A description of the pulping and bleaching processes
o Summary tables of monitoring, reporting, and recordkeeping requirements
o A comparison of MACT and New Source Performance Standards requirements
for pulp and paper sources
o A summary of other air regulations affecting the pulp and paper industry
• Summary Flow Diagrams of the Pulp and Paper MACT Standard (December 1997)
• List of Pulp and Paper Mills Subject to Combined Air and Water Rules (EPA; September
1997)
• Questions and Answers (Q& As) for Pulp and Paper NESHAP (September 1999)
• Questions and Answers (Q&A's) for the Pulp and Paper NESHAP - Second Volume
(March 2000)
• Memo from OAQPS - Clean Condensate Alternative for the Pulp and Paper NESHAP
(April 2004)
• Letter from Region 4 - Pulp and Paper MACT Condensate Averaging Times (November
1999)
• Letter from Region 4 - Clarification of the Clean Condensate Alternative (March 2003)
• EPA Office of Compliance Sector Notebook Project: Profile of the Pulp and Paper
Industry - 2nd Edition (November 2002)
62
-------
Appendix I
Agency Response to the Draft Report
OCT 23 2007
MEMORANDUM
SUBJECT: Comments of OIG Evaluation Report: "Improvements in Air Toxics
Emissions Data Needed to Conduct Residual Risk Assessments" (OIG-
Assignment No. 2005-001117)
FROM: Robert J. Meyers
Principal Deputy Assistant Administrator
for Air and Radiation (6101 A)
TO: Wade T. Najjum
Assistant Inspector General for Program Evaluation (2460T)
We appreciate the opportunity to review and comment on the final Office of
Inspector General (OIG) report entitled "Improvements in Air Toxics Emissions Data
Needed to Conduct Residual Risk Assessments" (OIG-Assignment No. 2005-001117).
The Office of Air and Radiation (OAR) disagrees with several conclusions reached by
OIG as well as with OIG's recommendations contained in the final report. Below, we
explain our disagreements along with the activities we have underway to implement the
recommendations.
(1) 3-1) We recommend that the Environmental Protection Agency (EPA)
develop data quality objectives for using NEI data in conducting residual
risk assessments.
Response: It is OAR's view that quantifiable Data Quality Objectives (DQOs) or
acceptance criteria for use of data within the National Emissions Inventory are not
practicable or appropriate within the current assembly and use of the National
Emissions Inventory (NEI). To fulfill Agency mission goals and objectives using
data of known quality, EPA programs develop and implement supporting quality
systems. The Office of Air Quality Planning and Standards (OAQPS) uses an
applied Quality System, implemented through a formal Quality Management Plan
(QMP), to manage the quality of its environmental data collection, generation, and
use. To expedite these concepts, OAQPS' QMP requires that projects meeting
certain criteria are required to create Quality Assurance Project Plans (QAPP).
63
-------
For the NEI, EPA uses a Quality System that is based on an acceptance process to
determine if the inventory is suitable for the purpose(s) for which it is to be used.
This process includes numerous quality reviews by both EPA and our State/local
partners. These reviews are both automated (e.g., State IT system checks) and
professionally evaluated by experts (e.g., EPA and State staff). The review
process generally incorporates multiple elements such as 1) data/information flow
monitoring and checks; 2) clear decision points in the data flow; 3) decisions by
professional experts in assessing data elements through mass balance calculations,
statistical determinations, and other tools; and 4) data acceptance determinations regarding
the usability of the data/information for the intended purpose.
These reviews are documented in the plans for inventory development which are
updated with each Inventory Cycle (e.g., 1999, 2002, and 2005). For the NEI,
project specific plans, similar to a QAPP, have been prepared at the beginning of
each 3-year project cycle. Project specific documentation has served as an
updated plan by specifying differences from the original plan for each finalized
version of a project. The 2005 NET data plan development is located at
http ://www. epa. gov/ttn/chief/net/2005dataplan.html.
Also, on the EPA CHIEF website
(http://www.epa.gov/ttn/chief/net/2002inventory.html) are the other elements of the
DQOs and QAPPs such as the Input formats, the Consolidated Emissions Reporting
Rule, and Documentation of the 2002 NEI:
In the future, as resources permit, EPA can work with our partners, EPA
researchers, and the broader scientific community to identify and incorporate new,
and possibly quantifiable, quality measures and acceptance criteria. At this time,
quantifiable acceptance criteria for NEI data national data elements are not
meaningful or useful. EPA continues to improve the NEI and is currently
developing a new data base to improve the timeliness and quality of data and
reports from the NEI. As we work with our partners on improving the quality and
consistency of data provided, we will continue to consider new Data Quality
Objectives appropriate for the use of NEI data and information.
(2) 3-2) We also recommend that EPA establish requirements for State reporting
of air toxics emissions data and compliance monitoring information.
Response: While we agree that establishing requirements for reporting air toxics
emissions data to the EPA could improve the quality of such data, we do not
believe such requirements are appropriate at this time. We arc currently revising
the requirements for State reporting of criteria pollutant emissions to EPA in the
Air Emissions Reporting Requirements rule. These changes were proposed last
year and we are currently considering public comments, including a comment on
this issue. As this is a policy issue under consideration, it is premature to convey our
program's position on this issue.
64
-------
We note that our current, evolving HAP emission inventory efforts should
continue to improve progress in improving HAP emissions inventory data quality.
Working with our State and local partners and industry, we believe these efforts
can achieve the desired data quality results. The EPA has developed national
level inventories of HAP emissions on a 3 year cycle since 1990. The success of
EPA's efforts rests in large measure on the cooperation of states in submitting
HAP data to EPA to support the development of national HAP emission
inventories. Each emission inventory cycle has benefited from increased state
agency participation in submitting HAP data. We believe it will he possible to
continue developing and improving national level HAP inventories using the
cooperative approach employed to date but intend to closely monitor the
participation of State agencies and industry in this effort. We also expect EPA's
new Emissions Inventory System (EIS), when completed, to improve HAP
inventory quality. The EIS, by further automating data handling and analysis, and
providing inventory transparency, will facilitate identification and resolution of
both HAP and criteria pollutant inventory data issues. We understand and
appreciate the objectives inherit in your recommendation regarding the need for a
HAP data reporting rule. We may revisit this recommendation in the future
should our voluntary efforts to improve this data not meet our program needs.
In closing, I would like to thank the Office of the Inspector General for working
with us in developing the final report and offering its recommendations. If you have
additional questions after reviewing this response and the documents cited therein, please
do not hesitate to request a briefing. Thank you for the opportunity to formally respond
to the recommendations contained in the Office of Inspector General (OIG) report
entitled "Improvements in Air Toxics Emissions Data Needed to conduct Residual
Risk Assessments.
Attachment
cc: Pete Cosier, OAR Audit Follow-up Coordinator
Michael Boucher, OAQPS Audit Follow-up Coordinator
Rick Beusse, Director for Program Evaluation, Air Issues, OIG
65
-------
Appendix J
Distribution
Office of the Administrator
Principal Deputy Assistant Administrator for Air and Radiation
Agency Followup Official (the CFO)
Agency Followup Coordinator
Office of General Counsel
Associate Administrator for Congressional and Intergovernmental Relations
Associate Administrator for Public Affairs
Audit Followup Coordinator, Office of Air and Radiation
Acting Inspector General
66
------- |