National Emission Standards for Hazardous Air Pollutants From the Portland Cement Manufacturing Industry, 76518-76552 [E6-21405]

Agencies

• ENVIRONMENTAL PROTECTION AGENCY

[Federal Register Volume 71, Number 244 (Wednesday, December 20, 2006)]
[Rules and Regulations]
[Pages 76518-76552]
From the Federal Register Online via the Government Printing Office [www.gpo.gov]
[FR Doc No: E6-21405]



[[Page 76517]]

-----------------------------------------------------------------------

Part IV





Environmental Protection Agency





-----------------------------------------------------------------------



40 CFR Part 63



National Emission Standards for Hazardous Air Pollutants From the 
Portland Cement Manufacturing Industry; Final Rule and Proposed Rule

Federal Register / Vol. 71, No. 244 / Wednesday, December 20, 2006 / 
Rules and Regulations

[[Page 76518]]


-----------------------------------------------------------------------

ENVIRONMENTAL PROTECTION AGENCY

40 CFR Part 63

[EPA-HQ-OAR-2002-0051; FRL-8256-4]
RIN 2060-AJ78


National Emission Standards for Hazardous Air Pollutants From the 
Portland Cement Manufacturing Industry

AGENCY: Environmental Protection Agency (EPA).

ACTION: Final rule.

-----------------------------------------------------------------------

SUMMARY: On June 14, 1999, under the authority of section 112 of the 
Clean Air Act (CAA), EPA promulgated national emission standards for 
hazardous air pollutants (NESHAP) for new and existing sources in the 
Portland cement manufacturing industry. On December 15, 2000, the 
United States Court of Appeals for the District of Columbia Circuit 
(D.C. Circuit) remanded parts of the NESHAP for the Portland cement 
manufacturing industry to EPA to consider, among other things, setting 
standards based on the performance of the maximum achievable control 
technology (MACT) floor standards for hydrogen chloride (HCl), mercury, 
and total hydrocarbons (THC), and metal hazardous air pollutants (HAP).
    EPA published a proposed response to the court's remand on December 
2, 2005. We received over 1700 comments on the proposed response. This 
action promulgates EPA's final rule amendments in response to the 
court's remand and the comments received on the proposed amendments.

DATES: This final rule is effective on December 20, 2006.

ADDRESSES: EPA has established a docket for this action under Docket ID 
No. EPA-HQ-OAR-2002-0051. All documents in the docket are listed on the 
www.regulations.gov Web site. Although listed in the index, some 
information is not publicly available, e.g., confidential business 
information (CBI) or other information whose disclosure is restricted 
by statute. Certain other material, such as copyrighted material, is 
not placed on the Internet and will be publicly available only in hard 
copy form. Publicly available docket materials are available either 
electronically through www.regulations.gov or in hard copy at EPA 
Docket, EPA/DC, EPA West, Room 3334, 1301 Constitution Ave., NW., 
Washington, DC. The Public Reading Room is open from 8:30 a.m. to 4:30 
p.m., Monday through Friday, excluding legal holidays. The telephone 
number for the Public Reading Room is (202) 566-1744, and the telephone 
number for EPA Docket Center is (202) 566-1742.

FOR FURTHER INFORMATION CONTACT: Mr. Keith Barnett, EPA, Office of Air 
Quality Planning and Standards, Sector Policies and Programs Division, 
Metals and Minerals Group (D243-02), Research Triangle Park, NC 27711; 
telephone number (919) 541-5605; facsimile number (919) 541-3207; e-
mail address barnett.keith@epa.gov.

SUPPLEMENTARY INFORMATION:

I. General Information

    A. Does this action apply to me? Entities potentially affected by 
this action are those that manufacture Portland cement. Regulated 
categories and entities include:

                   Table 1.--Regulated Entities Table
------------------------------------------------------------------------
                                                   Examples of regulated
           Category                 NAICS \1\             entities
------------------------------------------------------------------------
Industry......................  32731............  Owners or operators
                                                    of Portland cement
                                                    manufacturing
                                                    plants.
State.........................  None.............  None.
Tribal associations...........  None.............  None.
Federal agencies..............  None.............  None.
------------------------------------------------------------------------
\1\ North American Industry Classification System.

    This table is not intended to be exhaustive, but rather provides a 
guide for readers regarding entities likely to be regulated by this 
action. This table lists the types of entities that may potentially be 
regulated by this action. To determine whether your facility is 
regulated by this action, you should carefully examine the 
applicability criteria in 40 CFR 63.1340 of the rule. If you have 
questions regarding the applicability of this action to a particular 
entity, consult the person listed in the preceding FOR FURTHER 
INFORMATION CONTACT section.
    B. Judicial Review. The NESHAP for the Portland Cement 
Manufacturing Industry were proposed in December 2, 2005 (70 FR 72330). 
This action announces EPA's final decisions on the NESHAP. Under 
section 307(b)(1) of the CAA, judicial review of the final NESHAP is 
available only by filing a petition for review in the U.S. Court of 
Appeals for the D.C. Circuit by February 20, 2007. Under section 
307(d)(7)(B) of the CAA, only an objection to a rule or procedure 
raised with reasonable specificity during the period for public comment 
can be raised during judicial review. Moreover, under section 307(b)(2) 
of the CAA, the requirements established by the final NESHAP may not be 
challenged separately in any civil or criminal proceeding brought to 
enforce these requirements.
    C. How is this Document Organized? The information presented in 
this preamble is organized as follows:

I. General Information
II. Background
III. Summary of the National Lime Association v. EPA Litigation
IV. EPA's Final Action in Response to the Remand
    A. Determination of MACT for Mercury Emissions
    B. Determination of MACT for HCl Emissions
    C. Determination of MACT for THC Emissions
    D. Evaluation of a Beyond-the-Floor Control Option for Non-
Volatile HAP Metal Emissions
V. Other Rule Changes
VI. Responses to Major Comments
VII. Summary of Environmental, Energy, and Economic Impacts
    A. What facilities are affected by the final amendments?
    B. What are the air quality impacts?
    C. What are the water quality impacts?
    D. What are the solid waste impacts?
    E. What are the energy impacts?
    F. What are the cost impacts?
    G. What are the economic impacts?
VIII. Statutory and Executive Order Reviews
    A. Executive Order 12866, Regulatory Planning and Review
    B. Paperwork Reduction Act
    C. Regulatory Flexibility Analysis
    D. Unfunded Mandates Reform Act
    E. Executive Order 13132, Federalism
    F. Executive Order 13175, Consultation and Coordination With 
Indian Tribal Governments
    G. Executive Order 13045, Protection of Children From 
Environmental Health Risks and Safety Risks
    H. Executive Order 13211, Actions That Significantly Affect 
Energy Supply, Distribution, or Use
    I. National Technology Transfer and Advancement Act
    J. Congressional Review Act

[[Page 76519]]

II. Background

    Section 112(d) of the CAA requires EPA to set emissions standards 
for major stationary sources based on performance of the MACT. The MACT 
standards for existing sources must be at least as stringent as the 
average emissions limitation achieved by the best performing 12 percent 
of existing sources in the category or subcategory or the best 
performing five sources for source categories with less than 30 sources 
(CAA section 112(d)(3)(A) and (B)). This level is called the MACT 
floor. For new sources, MACT standards must be at least as stringent as 
the control level achieved in practice by the best controlled similar 
source (CAA section 112(d)(3)). EPA also must consider more stringent 
``beyond-the-floor'' control options. When considering beyond-the-floor 
options, EPA must consider not only the maximum degree of reduction in 
emissions of HAP, but must take into account costs, energy, and non-air 
quality health environmental impacts when doing so.
    On June 14, 1999 (64 FR 31898), in accordance with these 
provisions, EPA published the final rule entitled ``National Emission 
Standards for Hazardous Air Pollutants From the Portland Cement 
Manufacturing Industry'' (40 CFR part 63, subpart LLL).\1\
    The legacy public docket for the final rule is Docket No. A-92-53. 
The final rule provides protection to the public by requiring Portland 
cement manufacturing plants to meet emission standards reflecting the 
performance of the MACT. Specifically, the 1999 final rule established 
MACT-based emission limitations for particulate matter (as a surrogate 
for non-volatile HAP metals), dioxins/furans, and for greenfield \2\ 
new sources, THC (as a surrogate for organic HAP). We considered, but 
did not establish limits for, THC for existing sources and HCl or 
mercury for new or existing sources. In response to the mandate of the 
D.C. Circuit arising from litigation summarized below in this preamble, 
on December 2, 2005, we proposed amendments addressing standards for 
these pollutants. We received over 1700 comments on the proposed 
amendments. Most of these comments were from the general public and 
addressed the lack of a mercury emission limitation in the proposed 
amendments. This final action reflects our consideration of these 
comments. We have previously amended the Portland Cement NESHAP. 
Consistent with the terms of a settlement agreement between the 
American Portland Cement Alliance and EPA, EPA adopted final amendments 
and certain interpretative clarifications to the rule on April 5, 2002 
(76 FR 16614), July 5, 2002 (67 FR 44766), and December 6, 2002 (67 FR 
72580). These amendments generally relate to the rule's applicability, 
and to the performance testing, and monitoring provisions of the rule. 
In this action, we are also amending the rule to re-insert two 
paragraphs relating to the applicability of the Portland cement new 
source performance standards that were deleted in error in a previous 
amendment.
---------------------------------------------------------------------------

    \1\ Cement kilns which burn hazardous waste are in a separate 
class of source, since their emissions differ from Portland cement 
kilns as a result of the hazardous waste inputs. Rules for hazardous 
waste-burning cement kilns are found at subpart EEE of part 63.
    \2\ A new greenfield kiln is a kiln constructed after March 24, 
1998 at a site where there are no existing kilns.
---------------------------------------------------------------------------

    It should be noted that the rule text presented in this notice 
includes parts of the rule that are not being amended. This is done 
because, in some cases, adding additional rule text reduces the 
possibility of errors in updating the Code of Federal Regulations.

III. Summary of the National Lime Association v. EPA Litigation

    Following promulgation of the NESHAP for Portland cement 
manufacturing, the National Lime Association and the Sierra Club filed 
petitions for review of the standards in the D.C. Circuit. The American 
Portland Cement Alliance, although not a party to the litigation, filed 
a brief with the court as amicus curiae. The court denied essentially 
all of the petition of the National Lime Association, but granted part 
of the Sierra Club petition.
    In National Lime Association v. EPA, 233 F. 3d 625 (D.C. Cir. 
2000), the court upheld EPA's determination of MACT floors for 
particulate matter (PM) (as a surrogate for non-volatile HAP metals) 
and for dioxin/furan. However, the court rejected EPA's determination 
that it need not determine MACT floors for the remaining HAP emitted by 
these sources, namely, mercury, other organic HAP (for which THC are a 
surrogate), and HCl (233 F. 3d at 633). The court specifically rejected 
the argument that EPA was excused from establishing floor levels 
because no ``technology-based pollution control devices'' exist to 
control the HAP in question (Id. at 634). The court noted that EPA is 
also specifically obligated to consider other pollution-reducing 
measures including process changes, substitutions of materials inputs, 
or other modifications (Id.). The court remanded the rule to EPA to set 
MACT floor emission standards for HCl, mercury, and THC. (Id. At 641.)
    The Sierra Club also challenged EPA's decision not to set beyond-
the-floor emission limits for mercury, THC, and non-volatile HAP metals 
(for which PM is a surrogate). The court only addressed the absence of 
beyond-the-floor emission limits for non-volatile HAP metals since EPA 
was already being required to reconsider MACT floor emission standards 
for mercury, THC, and HCl, and thus, by necessity, also must consider 
whether to adopt beyond-the-floor standards for these HAP. The Sierra 
Club argued, and the court agreed, that in considering beyond-the-floor 
standards for non-volatile HAP metals, EPA considered cost and energy 
requirements but did not consider non-air quality health and 
environmental impacts as required by the CAA (Id. at 634-35). The court 
also found EPA's analysis of beyond-the-floor standards deficient in 
its assertion that there were no data to support fuel switching 
(switching to natural gas) as a viable option of reducing emissions of 
non-volatile HAP metals (Id. at 635).

IV. EPA's Final Action in Response to the Remand

A. Determination of MACT for Mercury Emissions

1. Floor Determinations
    In developing the proposed amendments we systematically evaluated 
all possible means of developing a quantified floor standard for 
mercury emissions from these sources, including both back end 
technology-based pollution control devices and front end feed and fuel 
control. See National Lime, 233 F. 3d at 634 (finding that EPA had 
erred in examining only technological (i.e., back-end) controls in 
considering a level for a mercury floor). We also were unable to devise 
any type of work practice standard that would result in mercury 
emissions reductions (70 FR 72332--72335, December 2, 2005).\3\
---------------------------------------------------------------------------

    \3\ Indeed, most of the options EPA considered are really 
beyond-the-floor alternatives, because they reflect practices that 
differ from those now in use by any existing source (including the 
lowest emitters). (Coal switching, switching to natural gas, and raw 
material switching are examples.) In EPA's view, a purported floor 
standard which forces every source in a category to change its 
practices is a beyond-the-floor standard. Such a standard may not be 
adopted unless EPA takes into account costs, energy, and non-air 
health and environmental impacts. 70 FR 72335.
---------------------------------------------------------------------------

    In response to comments on the proposed standards, we have 
performed additional evaluations of potential floors for mercury 
emissions (and also performed additional evaluations of

[[Page 76520]]

beyond-the-floor options for mercury control). We obtained additional 
mercury emissions test data during and after the two comment periods on 
the proposed amendments and once again evaluated setting a floor based 
on the median of the 12 percent of the kilns demonstrating the lowest 
mercury emissions in stack tests. We discuss each of these 
possibilities in turn below.
    a. Control of Mercury in Primary \4\ Raw Materials and Fossil 
Fuels. i. Mercury Emission Levels Reflecting Raw Material and Fossil 
Fuel Contributions are Inherently Site-Specific.
---------------------------------------------------------------------------

    \4\ We discuss in section IV.A.1.c below floor determinations 
for cement kilns using secondary materials (utility fly ash) as raw 
materials, in place of primary materials.
---------------------------------------------------------------------------

    As stated at proposal, mercury emissions come from the predominant 
input to a cement kiln by volume: The limestone which is the chief raw 
material for the kiln.\5\ Small amounts of mercury also are found in 
other raw material inputs to the process.\6\ Fossil fuel, almost always 
coal, is the other source of mercury emissions. Mercury levels in 
limestone vary enormously, both within a single quarry and between 
quarries, the result being that a single source may be unable to 
replicate its own performance in different tests, and could not 
duplicate a second source's performance since a kiln lacks access to 
any other kiln's limestone. Mercury levels in coal likewise vary 
significantly, although mercury emissions due to coal are normally 
swamped by the emissions attributable to limestone (70 FR 72333-34).
---------------------------------------------------------------------------

    \5\ Limestone makes up approximately 75 percent of the mass 
input to the kiln. Typically the way a cement plant is sited is that 
a limestone quarry suitable for cement production and that is 
expected to provide many years of limestone is identified and the 
plant is built next to the quarry. There are cases where a cement 
plant may purchase small amounts of limestone to blend with the 
limestone from its quarry. However, this close proximity of the 
quarry and cement plant is an inherent part of the cement 
manufacturing process and, therefore, a cement plant does not have 
the flexibility to obtain the bulk of its limestone from any other 
source. See 70 FR 72333.
    \6\ Post-proposal review of available data on other mercury raw 
materials indicates that other feed materials also contribute some 
mercury, though, in most cases, less than limestone. Other raw 
materials include (but are not limited to): shale or clay to provide 
alumina; iron ore to provide iron; and sand to provide silica. These 
raw materials are used in lesser amounts than limestone, and a 
cement plant may have some flexibility in the sources of other raw 
materials. As noted in the preamble to the proposed amendments, 
there are cases where a facility made changes to their raw materials 
(other then limestone) to reduce mercury emissions. However, this 
type of control is site specific based on the available materials 
and the chemical composition of the limestone. The site specific 
factors preclude using this as a basis for a national rule (70 FR 
72334).
---------------------------------------------------------------------------

    In an attempt to quantify the potential variability, we looked to 
see if there were facilities with multiple stack tests for mercury. We 
do have multiple test results for one of the lowest mercury emitters in 
the data base. During the first test with the raw mill on \7\ the 
facility was one of the lower emitting facilities in the source 
category demonstrating emissions of 7.8 micrograms per dry standard 
cubic meter ([mu]g/dscm) (all test values are corrected to seven 
percent oxygen). During a second test 8 years later (reflecting raw 
materials from the same quarry) mercury emissions with the raw mill on 
were 60 [mu]g/dscm, a variability factor of roughly 8 times. We could 
identify no facility operational changes between the times of the two 
tests that would account for this large difference in mercury 
emissions.
---------------------------------------------------------------------------

    \7\ See section c. below discussing operation of the in-line raw 
mill and its implication for mercury control.
---------------------------------------------------------------------------

    We also obtained data from a facility that was retested for mercury 
in July 2005, within 3 months of an initial test. With the raw mill on, 
mercury emissions averaged 0.00138 pounds per hour in the April test 
and 0.00901 pounds per hour in the July test, a variability factor of 
7. With the raw mill off, emissions averaged 0.00823 pounds per hour in 
the April test and 0.0189 pounds per hour in the July test. We also 
noted that during the April test mercury emissions with the raw mill 
off were below mercury emissions with the raw mill on in the July test. 
Because it is known that when the raw mill is on the raw meal adsorbs 
mercury, thereby reducing measured mercury emissions in the short term, 
we can only assume that the uncontrolled variation in the mercury 
levels in the raw materials--all of which come from the same quarry--
was so great between the two tests that it negated the effect of the 
operating condition of the raw mill.
    We also assessed potential variability by examining daily 
variations in cement kilns' raw materials and fuel mercury contents. We 
obtained data from an operating facility that analyzed samples of raw 
material and fuel each day over a 30 day period. We calculated average 
daily emissions assuming all the mercury in the raw materials and fuel 
was emitted. The average daily emissions would vary from a low of 0.09 
lb to a maximum of 16.44 lb, or a factor of 183 (See Summary of Mercury 
Test data in Docket 2002-0051).
    These are enormous swings in variability.\8\ Moreover, it is 
virtually certain that the variability reflected in these results fails 
to cabin the total raw material and emissions variability experienced 
by the plants in the source category, since we have only a handful of 
results. These data confirm our tentative conclusion at proposal that 
constantly changing concentrations of mercury in kiln inputs leave no 
reliable way to quantify that variability. 70 FR 72333.
---------------------------------------------------------------------------

    \8\ Variability of emissions based on the operation of air 
pollution controls are typically lower that those shown above 
because air pollution controls are typically designed to meet 
certain percent reduction or outlet emissions levels and to account 
for variations in inlet conditions.
---------------------------------------------------------------------------

    In the proposed amendments we also evaluated requiring facilities 
to switch from coal to natural gas as a method to reduce mercury 
emissions, or requiring use of so-called clean coal (70 FR 72333-34). 
We tentatively concluded that this was not feasible on a national basis 
due to insufficient supply and lack of infrastructure, and reiterate 
that conclusion here. One commenter noted that petroleum coke was 
another fuel that is lower in mercury and is currently used as a cement 
kiln fuel. However, a mercury standard based on requiring fuel 
switching to petroleum coke suffers from the same defects as requiring 
facilities to switch to natural gas. This fuel may not be available in 
all areas of the country and there may not be sufficient availability 
of the fuel to replace a significant percentage of the coal burned in 
cement kilns. Petroleum coke is a byproduct of petroleum refining, 
therefore the supply is limited by the demand for refined petroleum 
fuels. Petroleum coke has a low volatile matter content which can lead 
to ignition problems if burned without a supplemental fuel. It also 
typically has a higher sulfur content than coal. This can adversely 
affect kiln refractory life and increase internal corrosion of the kiln 
shell. As previously noted, each individual facility has specific 
requirements for raw material additives based on the chemical 
composition of its limestone. The minerals present in the coal ash 
fulfill part of those requirements. Therefore, replacing part or all of 
the coal currently used at a facility with petroleum coke, which has 
almost no ash, may force the facility to incorporate additional raw 
material additives containing mercury to compensate for the loss of the 
coal ash.
    Thus, we adhere to the tentative conclusion reached at proposal: 
front end feed and fuel control of cement kilns is inherently site 
specific, and basing limits on kiln performance in individual 
performance tests which reflect only those inputs will result in 
limitations that kilns can neither duplicate (another kiln's 
performance) nor replicate (its own).

[[Page 76521]]

    ii. Implications of Permit Limits for Mercury. There are currently 
19 cement kilns (out of 70 cement kilns for which we reviewed permit 
requirements) that have permit limits for mercury. At first blush, it 
might be argued that these permit limits demonstrate that variability 
of mercury emissions can be controlled, since sources must comply with 
the limitations. It might further be argued that these permit limits 
are ``emission limitations achieved,'' the statutory basis for 
establishing floors for existing sources under section 112(d)(3). 
Likewise, for new sources, the lowest permit limit is arguably a 
measure of performance of the ``best controlled similar source'' (the 
permit itself being the means of control). We have determined, however, 
that for most facilities, the permit limit was established based on an 
estimate provided by the facility of the annual amounts of mercury that 
would enter the kiln with the raw materials and fuels. One facility had 
a mercury limit based on its estimated annual emission from an 
emissions test, and one facility had a limit based on a State law, 
although in neither case did the resulting permit cause a cement kiln 
source to alter or otherwise modify its existing practices to meet the 
limit. Thus, we find no cases where a facility actually has had to take 
any steps, either through the imposition of process changes or add-on 
controls, to reduce its mercury emissions as a result of any of these 
permit limits. See ``Summary of Cement Kiln Permit Data for Mercury'' 
in the docket.
    We considered the option of setting an emissions limit, either on a 
pounds per year (lb/yr) or a pound per ton of clinker basis, based on 
the median of the top 12 percent of the 17 kilns with permit 
limitations. However, we repeat that none of the facilities with permit 
limits were required to actually take action to reduce mercury 
emissions. Their limits were all based on site specific factors 
(expected maximum conceivable levels of mercury emissions), and were 
set at a level that did not require the imposition of add-on controls, 
feed or fuel substitution, or any other constraint. Any limit we set 
based on these permits would require that at least some facilities 
apply beyond-the-floor control technology to meet the limit since feed 
and fuel control via substitution is not possible. Such a standard 
would impermissibly apply beyond-the-floor emission control without 
consideration of costs and other non-air health and environmental 
impacts.
    We also considered a limit where each facility would set their own 
site specific limit based on the same procedures the facilities with 
permits used: determining in the course of the permitting process what 
its maximum conceivable mercury emissions are likely to be based on the 
facility's raw material and fuel inputs, and tacking on an additional 
variability factor. However, this would require that we set a separate 
limit for each facility, with each facility being its own subcategory 
(i.e. a different type of facility) based on its site specific raw 
materials and fuels. See 70 FR 72334, alluding to this possibility. EPA 
has great discretion in deciding whether or not to subcategorize within 
a source category. We do not believe a decision to individually 
subcategorize is warranted considering the fact that the result will be 
no discernable environmental benefit because conduct will be unaltered. 
Chemical Mfr's Ass'n v. EPA, 217 F. 3d 861, 866-67 (D.C. Cir. 2000) 
(arbitrary and capricious for EPA to impose costly regulatory 
obligations without some showing that the requirement furthers the 
CAA's environmental goals).
    Therefore, we have determined that even though these permit limits 
exist, they have not resulted in a quantifiable reduction of mercury 
emissions. Any option to develop a MACT floor for mercury with these 
limits would either result in an unnecessarily complex rule with no 
environmental benefit, or a rule which improperly imposes a de facto 
beyond-the-floor standard without the required consideration of costs, 
energy and non-air quality impacts.
    iii. Why not Average the Performance Test Data? Some commenters 
stated that EPA must simply average the results of the 12 per cent 
lowest mercury performance test data to establish the floor for 
existing sources, and establish the new source performance floor at the 
level of the lowest test result. We rejected this approach at proposal, 
and do so here, because it fails to account for the variability of 
mercury levels in raw materials and fuels and hence variability in 
performance. See 70 FR 72335; see also 70 FR 59436 (Oct. 12, 2006). We 
must, of course, account for sources' variability in establishing a 
MACT floor. Mossville Environmental Action Now v. EPA, 370 F. 3d 1232, 
1241-42 (D.C. Cir. 2004). The only way all kilns, including the kilns 
with the lowest emission levels in individual tests, could meet this 
type of standard continuously, as required, would be to install backend 
technology-based control equipment. However, this would be a de facto 
beyond-the-floor standard, adopted impermissibly because of failure to 
assess cost, energy, and non-air quality health and environmental 
impacts. See 70 FR 72335.
    We are aware that in the case of the NESHAP for Industrial, 
Commercial, and Institutional Boilers and Process Heaters (Boiler 
NESHAP), we used short term emissions data and applied a variability 
factor to determine a floor for mercury emissions (69 FR 55236, 
September 13, 2004). We do not believe that approach is applicable to 
the Portland cement source category. First, in the case of the Boiler 
NESHAP the floor was based on performance of a control technology, 
fabric filters, which means that facilities were exercising some 
control over mercury emissions and variability could be realistically 
cabined and quantified, so that an emission limit could be replicable 
and duplicable. Though the majority of cement kilns also use fabric 
filters, the collected particulate in this source category consists of 
product and, to some extent, unprocessed raw materials. As a result 
most of the collected particulate is recycled back to the process, 
largely negating any impact of the particulate control technology on 
mercury emissions.\9\ Second, the variabilities seen as a result of 
fuel inputs in the Boiler NESHAP are much lower than the variabilities 
indicated in the Portland cement industry where the mercury fuel 
variability is a distant second to the enormous variability of mercury 
in the raw materials. We do not believe the data exist to accurately 
quantify this variability.
---------------------------------------------------------------------------

    \9\ As explained in the following section of the preamble, 
however, EPA has determined that the floor for both existing and new 
sources involves the removal from the kiln system of collected 
particulate under designated circumstances. In addition, the floor 
for new sources reflects reductions in mercury based on performance 
of a wet scrubber.
---------------------------------------------------------------------------

    Another option we considered was using long term data to set a 
floor. However, since, to our knowledge, continuous emission monitors 
for mercury have not been demonstrated on cement kilns, and none 
currently exist on cement kilns, there is no long term stack 
performance data on mercury emissions from cement kilns that we could 
use to set a numerical emissions limit. The only available long term 
data of which we are aware is from several facilities which have a 
requirement to perform monthly analyses of composited daily samples of 
fuels and raw materials to calculate a 12 month mercury emissions 
total. However, all these kilns are located in one state (Florida) with 
unrepresentatively low levels of mercury in limestone (so far as we can 
determine). We do not believe these data would be representative of

[[Page 76522]]

the source category as a whole. More basically, basing a standard on 
one set of kilns' raw material inputs still suffers from the defect 
that no facility has access to another's raw materials.
    b. Floors for Facilities Using Utility Fly Ash as Raw Material. 
Some cement kilns use utility fly ash as an alternative raw material to 
replace shale or clay.\10\ These kilns replace a natural material, 
shale or clay, with a secondary material (i.e. a recycled air pollution 
control residue), fly ash. Approximately 34 cement manufacturing 
facilities are currently using utility boiler fly ash as a feedstock. 
We reviewed the available data and have come to the conclusion that 
cement kilns using fly ash are a different type of kiln, within the 
meaning of section 112 (d) (1) of CAA, and that for cement kilns 
currently using fly ash, the current use would be considered the MACT 
floor. Our reasoning is as follows.
---------------------------------------------------------------------------

    \10\ Though these are also raw materials inputs, the mass of 
clay or shale is typically less than 15 percent of the mass input to 
the kiln. Limestone makes up approximately 80 percent of the mass 
input.
---------------------------------------------------------------------------

    Use of fly ash can have an effect on mercury emissions since fly 
ash contains mercury in varying amounts. As discussed below, mercury 
emissions may be higher or lower depending on the amounts of mercury 
involved vis-a-vis the raw materials that would otherwise be used (if 
available). But as also explained more fully below, some cement kilns 
using fly ash do not have an alternative raw material source. Given 
that these kilns use a different raw material, not always replaceable, 
and that the material affects mercury emissions, we believe that these 
kilns are a separate kiln type, and hence a separate subcategory, for 
purposes of mercury emissions. For a similar conclusion see 64 FR at 
52871 (Sept. 30, 1999) (cement kilns that choose to burn hazardous 
waste in place of fossil fuels are a separate source category for MACT 
purposes).
    We attempted to determine if, in general, facilities that use fly 
ash have higher emissions of mercury than those that do not. An 
analysis of data for EPA's toxic release inventory and the National 
Emissions Inventory did not show differences significant enough that we 
could draw any definitive conclusions. We considered reviewing the 
available mercury emissions test data to determine if we could discern 
a trend. However, as previously discussed, we do not believe these data 
are representative of long term mercury emissions. We also attempted to 
obtain data on the important issue of the amounts and mercury contents 
of fly ash used relative to other raw materials. These data apparently 
do not exist, with one exception discussed in the next paragraph. We do 
know that the two highest mercury emitting facilities (in individual 
performance tests) do not use fly ash. Without data on the actual 
mercury contributions of all materials, we do not believe we can draw 
any valid general conclusions on the impact of the use of fly ash on 
mercury emissions.
    We do have detailed data from one facility that used fly ash where 
50 percent of the total mercury input to the kiln is in the fly ash. 
However, even for this facility, we cannot accurately quantify the 
impact on mercury emissions of the decision to replace the shale used 
at this facility with fly ash because we have been unable to obtain 
data on the mercury content of the shale the fly ash replaced. We also 
have no mercury analysis data from the time period when the facility 
used shale.
    There are other factors to consider when we evaluate the 
environmental effects--generally quite positive--of substituting fly 
ash for shale or clay. First, fly ash in general has a lower organic 
material content than shale or clay. At the facility just mentioned, 
replacing the shale with fly ash reduced emissions of THC from around 
80 parts per million by volume (ppmv) to 3 ppmv. Because fly ash can 
reduce kiln fuel consumption, it reduces emissions of sulfur dioxide 
(SO2), oxides of nitrogen (NOX), and carbon 
monoxide (CO2). Using fly ash as a kiln feed reduces the 
landfill requirements for disposal of utility fly ash. Use of fly ash 
reduces cement plant power consumption because it is usually fine 
enough that it can be added directly to the kiln rather then being 
ground in a mill. Use of fly ash also reduces fuel consumption because 
compared to the raw materials it typically replaces it is already 
highly calcined; it does not have the same types of large crystals as 
the raw materials it replaces (this improves burnability); some fly 
ashes have lower metal alkali content, thus avoiding hard burning to 
drive off alkali metals and reducing the need to operate the alkali 
bypass; it is drier than quarried materials, thus saving fuel used to 
dry materials. Many domestic cement plants have high pyrites in their 
quarry, especially in the shale or clay. In most cases, this pyrite is 
the main source of SO2 emissions from the kiln. Using fly 
ash can significantly reduce the SO2 emissions that result 
from pyrite in the raw materials. It also reduces the energy required 
for the quarring, milling, and transporting of the shale or clay prior 
to its use as a feedstock, as well as the associated air emissions.
    It should also be noted that there are at least two new facilities 
whose permits specifically require use of fly ash as their alumina 
source, as they have no source for shale or clay, the primary material 
alternatives for alumina. Finally, a facility that currently uses fly 
ash may not be able to return to using the natural (i.e. primary) raw 
materials it replaced. For example, if the replaced raw materials were 
shale, the shale quarry may now be closed and the facility may not have 
access to a suitable shale supply.
    Given the lack of any data to positively state the impact of fly 
ash on mercury emissions for the source category in general, as well as 
the positive environmental effects of using fly ash, there is no basis 
for a floor standard based on substituting other potential raw 
materials (such as shale or clay) for fly ash. At the same time, we do 
not see any means of identifying a floor for existing fly ash users 
based on substituting different fly ash types reflecting different 
mercury content. The recycled fly ash is not fungible. Cement kilns 
must carefully select only fly ash with needed properties within a 
relatively small tolerance. Cement kilns also usually are limited to 
fly ash available from boilers which are reasonably close to the kiln 
(typically within a few hundred miles) or shipping expense becomes 
prohibitive. The fly ash selection process is involved; it has taken 
years for kilns to identify a suitable fly ash source. Accordingly, we 
evaluate fly ash like the other raw material inputs into cement kilns, 
and do not believe that a floor that is based on substitution of either 
raw materials or other fly ash is justified because the input is 
variable and uncontrollable. We discuss in section IV.A.2 below the one 
exception to this conclusion for fly ash where the mercury content has 
been artificially increased by sorbent injection.
    c. Control of Collected Particulate (Cement Kiln Dust). There are 
two operation factors that impact measured mercury emissions at the 
kiln stack. These are the use of in-line raw mills and the recycling of 
cement kiln dust (CKD).
    Many (but not all) kiln systems have in-line raw mills. In these 
systems the kiln exhaust gas is routed through the raw mill to dry the 
raw materials. This process results in mercury contained in the flue 
gas being adsorbed by the raw meal.\11\ This results in an apparent

[[Page 76523]]

reduction if mercury emissions are being measured at the kiln stack. 
However, the captured mercury is reintroduced into the kiln which 
creates a recycle loop of mercury until the captured mercury eventually 
escapes and is emitted to the atmosphere. Also, raw mills do not run 
continuously. When the raw mill is turned off, this effect of raw meal 
adsorption of mercury is negated and mercury emissions appear to 
increase. However, the increase is actually mercury that would have 
previously been emitted but was captured by the raw meal and returned 
to the kiln. The net effect is that an in-line raw mill does not 
increase or reduce mercury emissions over the long term; it simply 
alters the time at which the mercury is released.
---------------------------------------------------------------------------

    \11\ More specifically, when the mill is on-line, the kiln gas 
containing volatilized mercury is used to sweep the mill of the 
finely ground raw feed particles. Since the mill temperature is only 
about 90 to 120 [deg]C during this operational mode, the fine PM can 
adsorb the mercury in the gas stream, and the particles containing 
condensed mercury are stored in the raw feed silos. This stored raw 
mix then is fed to the kiln. The captured mercury is again 
volatilized and returned in the gas stream to the raw mill, only to 
be captured again in the raw mill, as described above. This process 
continues as long as the raw mill is on-line, and the raw feed 
continues to adsorb additional mercury through this process.
---------------------------------------------------------------------------

    Mercury is also adsorbed on the CKD collected in the particulate 
control device, typically a fabric filter or an electrostatic 
precipitators (ESP). Because the collected CKD mainly consists of 
product, and sometimes small amounts of raw materials, the collected 
CKD is recycled back to the kiln to the extent possible. The portion 
that cannot be recycled to the kiln is either sent to a landfill, or 
used in some other manner (i.e. some type of beneficial use). Most 
facilities require that a portion of the CKD be removed from the kiln 
system rather than returned to the kiln. This is done to bleed the kiln 
system of alkali materials that build up as they circulate which would 
otherwise contaminate product and damage the kiln lining. This practice 
necessarily reduces the overall volume of mercury emitted by cement 
kilns, as noted by several commenters, since the entrained mercury in 
the CKD is no longer available for release from the kiln. The amount of 
reduction is kiln-specific, based on the level of alkali materials in 
the kiln's raw materials and required product specifications, and 
therefore not quantifiable on a national basis. Nor would kiln-by-kiln 
site-specific emission standards be warranted, for the same reasons 
that site-specific limits based on mercury levels on raw material and 
fuel inputs are not justified. EPA is instead determining that a floor 
standard for both existing and new sources is the work practice that 
cement kiln dust be removed from the kiln system at the point that 
recirculation causes adverse effect on product.
    d. Standards Based on Performance of Wet Scrubbers. There are at 
least five cement kilns that have limestone (wet) scrubbers installed 
for control of SO2. Commenters noted that based on 
experience with utility boilers, and other similar combustion devices, 
there is reason to expect that the scrubbers installed on cement kilns 
also remove oxidized mercury.
    To our knowledge, we obtained all the available data on wet 
scrubber controlled facilities after the comment period on the proposed 
amendments. This consists of data from 2004 and 2005 tests at two 
facilities measured exclusively at the scrubber outlet. These data 
range from 0.42 to 30 [mu]g/dscm. Variability of mercury emissions at 
the scrubber-equipped kilns for which we have multiple test data 
differs by orders of magnitude. These data fall within the range of 
test data from all kilns (those with wet scrubbers and those without 
wet scrubbers). We have no test data for mercury measured at the 
scrubber inlet. As a result, we cannot, on the basis of the current 
data, determine with absolute certainty (though we believe it is 
reasonably certain) if the outlet mercury emissions from the wet 
scrubber equipped kilns are a result of mercury removal by the 
scrubber, or simply reflect the amounts of mercury in the raw 
materials. We now discuss the implications of this information for 
purposes of existing and new source floors. Note that the following 
discussion assumes the scrubbers remove oxidized mercury for reasons 
discussed below.
    First, there are an insufficient number of wet-scrubber equipped 
kilns on which to base an existing source floor. The scrubber-equipped 
kilns would represent the best performing sources since data from other 
kilns simply reflect the mercury levels in kiln inputs on the day of 
the test. There are 158 operating kilns, and the information available 
to us indicates that only five of them are equipped with wet scrubbers. 
The median kiln of the top 12 percent would, therefore, not be a 
scrubber equipped kiln.\12\
---------------------------------------------------------------------------

    \12\ Choosing the median source for assessing an existing source 
floor here is a reasonable manner of determining ``the average 
emission limitation achieved by the best performing 12 percent of 
existing sources'' (section 112 (d)(3)). Not only can the statutory 
term ``average'' be reasonably interpreted to mean median, but it is 
appropriate to do so here in order not to adopt a de facto beyond 
the floor standard. If one were simply to combine the mercury 
emission levels of the kilns equipped with wet scrubbers with other 
kilns whose mercury levels reflect raw material and fuel mercury 
levels at the time of the performance test, the resulting limit 
would not be achievable over time by any source other than one with 
a wet scrubber. Ostensible best performers would consequently have 
to retrofit with back end control, since otherwise they could not 
consistently achieve the results of their own performance tests.
---------------------------------------------------------------------------

    However, for new sources mercury emissions would not be 
uncontrolled--solely dependent on raw material mercury content--but 
rather would reflect performance of ``the best controlled similar 
source'' (section 112 (d)(3)). A kiln so-equipped would thus have the 
best performance over time, since variability in mercury attributable 
to raw material and fuel inputs would be controlled in part.\13\
---------------------------------------------------------------------------

    \13\ That is, variability would no longer be purely a function 
of the happenstance of the amount of mercury in raw materials (and 
fossil fuels) used in the test condition. As explained more fully 
below, performance of wet scrubbers, however, is variable, based not 
only on operation of the device but on mercury levels in input 
materials. Wet scrubbers on utility boilers, for example, are 
documented to remove between 0 to 72 percent of incoming mercury. 
See Control of Mercury Emissions from Coal-Fired Electric Utility 
Boilers: Interim Report Including Errata available at https://
www.epa.gov/nrmrl/pubs/600r01109/600r01109.htm. We should note, 
however, that because utility boilers do not have the significant 
levels of alkaline materials that are present in cement kilns, which 
alkaline materials would impede mercury oxidation and scrubber 
efficacy, we do not view utility boilers as a ``similar source'' for 
purposes of section 112(d)(1).
---------------------------------------------------------------------------

    We believe there is a reasonable basis that wet scrubbers remove 
oxidized mercury from cement kiln emissions. First, wet scrubbers are 
known to remove oxidized mercury in most combustion applications though 
removal rates vary. We have speciated mercury test data on two kilns 
that indicate that there is a significant amount of oxidized mercury in 
at least some cement kilns. See mercury emission test data for Holcim, 
Dundee, MI and Lafarge, Alpena, MI, in docket EPA-HQ-OAR-2002-0051. 
Second, the limited data we have from cement kilns equipped with wet 
scrubbers is among the lowest end-of-stack mercury data in our data 
base (although not the lowest), which could indicate that some removal 
mechanism is involved. An important caveat, however, is that these data 
are exclusively end-of-stack, without paired inlet concentrations. 
These data thus do not with absolute certainty demonstrate that mercury 
removal is occurring or how much.
    We estimated the performance of the best performing scrubber, and 
hence the new source MACT floor, to be 41 [mu]g/dscm (corrected to 7 
percent oxygen) using the following rationale. First, we limited the 
analysis to data from wet scrubber equipped kilns because, as just

[[Page 76524]]

discussed, the wet scrubber equipped kilns represent the best 
performing sources, regardless of their actual outlet emissions levels 
in individual performance tests. Second, we ranked all the wet scrubber 
mercury emissions with the raw mill off. We believe this is appropriate 
because the condition of raw mill off represents a normal operating 
mode for a cement kiln (albeit the operating mode when mercury 
emissions would be highest, as discussed above in section a.i). We then 
took the mean raw mill off value for mercury emissions from a cement 
kiln in our (limited) data base, and added to it a variability factor 
to account for normal variation in emissions. This variability factor 
is the standard deviation of the data multiplied by 2.326 (the z 
statistic) to produce the 99th confidence interval. We looked to all of 
the data, rather than to the data from the single lowest emitting kiln, 
because there are too few data points from that kiln (or from any one 
kiln) to estimate that kiln's variability. Given that variability is 
known to occur, we believe that this is the best approximation of 
variability of the best performing kiln presently available.
    The result of this analysis is a new source floor of 41 [mu]g/dscm 
that must be met continuously (raw mill on and raw mill off) (see 
further discussion in section A.3 below). This is an emissions limit 
that we believe will not be exceeded 99 percent of the time by the best 
performing kiln whose performance is used to set the standard.
    Because of the limited performance data characterizing performance 
of the lowest-emitting scrubber-equipped kiln, the rule also contains 
an alternative new source mercury floor. The best performing kiln is 
equipped with a wet scrubber, although there could be questions about 
its performance over time. Therefore, if a new source installs a 
properly designed and operated wet scrubber, and is unable to achieve 
the 41 [mu]g/dscm standard, then whatever emission level the source 
achieves (over time, considering all normal sources of variability) 
would become the floor for that source. Based on the design of the wet 
scrubber that is the basis of the new source floor, this would be a 
packed bed or spray tower wet scrubber with a minimum liquid-to-gas 
ratio of 30 gallons per thousand cubic feet of exhaust gas.
    In sum, we conclude that floors for mercury for all existing cement 
kilns should be to remove accumulated mercury-containing cement kiln 
dust from the system at the point product quality is adversely 
affected. The floor for new sources is to utilize this same work 
practice, and in addition, to meet a standard of either 41 [mu]g/dscm 
or a site-specific limit based on performance of a properly designed 
and operated wet scrubber.
    As just explained, the mercury data on which the new source floor 
is based are not only limited, but fails to definitively answer the 
critical question of whether wet scrubbers are removing oxidized 
mercury, and, if so, to what extent. We are taking immediate steps to 
address this issue and augment the data base. In an action published 
elsewhere in this Federal Register, we are granting reconsideration of 
the new source standard adopted in this rule, both due to substantive 
issues relating to performance of wet scrubbers and because information 
about their performance in this industry has not been available for 
public comment. We also have initiated actions to obtain inlet and 
outlet test data for cement kilns equipped with wet scrubbers in order 
to determine if these controls remove mercury, and to what extent. In 
addition, we are committing to completing this reconsideration process 
within one year from December 20, 2006.
2. Beyond-the-Floor Determinations
    During development of the original NESHAP for Portland cement 
manufacturing, we conducted MACT floor and beyond-the-floor analyses 
for kiln and in-line kiln/raw mill mercury emissions (63 FR 14182, 
March 24, 1998 and 64 FR 31898, June 14, 1999). We also conducted a 
beyond-the-floor analysis for mercury, based on the performance of 
activated carbon injection with an additional PM control device. Costs 
for the system would include the cost of the carbon injection system 
and an additional fabric filter (FF) to collect the carbon separately 
from the CKD. Based on the low levels of mercury emissions from 
individual Portland cement kilns, as well as the high cost per ton of 
mercury removed by the carbon injection/FF system, we determined that 
this beyond-the-floor option was not justified (63 FR 14202, March 24, 
1998).
    At proposal, EPA again concluded tentatively that a beyond the 
floor standard based on performance of activated carbon is not 
justified (70 FR 72335). We have since reevaluated beyond-the-floor 
control options for mercury emissions. This evaluation included both 
process changes and add-on control technology.
    There are two potential feasible process changes that have the 
potential to affect mercury emissions. These are removing CKD from the 
kiln system and, for the subcategory of kilns that currently use fly 
ash as a raw material, replacing the fly ash with a lower mercury raw 
material. Substituting raw materials or fossil fuels with lower-mercury 
inputs could in theory reduce mercury emissions, but this alternative 
is infeasible for the reasons explained at 70 FR 72333-72334.
    Generally, once mercury enters a kiln system, it has five potential 
fates: it may remain unchanged and become part of the final product; it 
may react with raw materials and exit the kiln in the clinker; it may 
vaporize in the high temperature of the kiln and/or preheater; it may 
condense or react with the cement kiln dust and be removed from the 
system; or it may exit the kiln system in vapor form or be adsorbed to 
a dust particle through the stack. In general, mercury in the fuel 
becomes volatilized near the kiln's combustion zone and is carried 
toward the feed end of the system along with combustion gases. Some of 
the mercury compounds pass through the entire system and exit in vapor 
phase through a stack. However, as the flue gas cools, some mercury may 
adsorb/condense onto dust particles in the cooler regions of the kiln 
system. Much of this dust containing condensed mercury would then be 
captured by the PM control device and for most kiln systems, returned 
to the kiln.
    We evaluated, requiring a facility to further reduce the recycling 
of CKD beyond the wastage already needed to protect product quality, 
the floor for both existing and new sources. For a 600,000 tpy (tpy) 
kiln the estimated total annual cost would be $3.7 million just for 
replacement of CKD (which is actually product) and disposal of 
additional solid waste. This cost does not account for the increased 
raw materials costs and energy costs associated with reducing the 
recycling of the CKD. The mercury emissions reduction would range from 
0.012 to 0.055 tpy based on assumed CKD mercury concentrations of 0.33 
and 1.53 parts per million (ppm) respectively. The cost per ton of 
mercury reduction would range from $67 million to $308 million. See 
Costs and Impacts of Wasting Cement Kiln Dust or Replacing Fly Ash to 
Reduce Mercury Emissions in docket EPA-HQ-OAR-2002-0051. We note that 
the median value for the mercury content of recycled CKD for one study 
was only 0.053 ppm. See the report Mercury and Lead Content in Raw 
Materials in docket EPA-HQ-OAR-2002-0051. This would indicate that for 
the majority of the facilities the costs per ton would be even higher 
that those presented above. In addition, we

[[Page 76525]]

estimate that wasting 50 percent of the recycled CKD would reduce the 
energy efficiency of the process by six percent due to the need to 
process and calcine additional feed to replace the wasted CKD. It is 
possible that in some cases the wasted cement kiln dust could be mixed 
with the cement product rather than landfilled, or that some other 
beneficial use could be found. This would reduce the costs and non-air 
adverse impacts of this option. However, there are currently barriers 
to directly mixing CKD with clinker due to product quality and product 
specification issues. We do not have data available to evaluate the 
potential for beneficial use of the CKD. Based on these costs, the 
adverse energy impacts, and the increased adverse waste disposal 
impacts (see 64 FR 45632, 45635-36 (Aug. 20, 1999) for examples of 
potential hazards to human health and the environment posed by disposal 
of cement kiln dust), we do not believe this beyond-the-floor option is 
justified and therefore are not selecting it.
    As previously noted, for the subcategory of facilities that use 
utility boiler fly ash as a kiln feed we determined that the current 
use represented the MACT floor. We considered two beyond-the-floor 
options for this subcategory. One option was to ban the use of any fly 
ash if it resulted in a mercury emissions increase over a raw material 
baseline, and the second was to only ban the use of fly ash whose 
mercury content had been artificially increased through the use of a 
sorbent to capture mercury in the utility boiler flue gas.
    If we were to ban the use of utility boiler fly ash for any case 
where it has been shown to increase mercury emissions from the kiln 
over a raw material baseline, facilities would have to revert to using 
their previous raw materials, or to find alternative raw materials that 
provide the same chemical constituents as the fly ash. As previously 
noted, if a facility replaces their shale or clay with fly ash, the 
quarry for that material may now be closed and it may not be possible 
to cost-effectively obtain the previously used raw materials. And for 
at least two new facilities, the original raw materials used at startup 
will include fly ash, so there is no previously used material with 
which to compare the mercury content of the fly ash. Due to the site 
specific costs associated with raw materials, we don't have any data to 
calculate the costs of the beyond-the-floor option for the industry as 
a whole. In one example, we estimated the cost as approximately $136 
million per ton of mercury reduction. See Costs and Impacts of Wasting 
Cement Kiln Dust or Replacing Fly Ash to Reduce Mercury Emissions in 
docket EPA-HQ-OAR-2002-0051. Also, this option would mean that all the 
fly ash currently being used as a cement kiln feed would now 
potentially have to be landfilled. This would generate an additional 3 
million tpy of solid waste, with potential adverse health and 
environmental impacts associated with management of these wastes. There 
would also be adverse environmental air and non-air quality health and 
environmental impacts associated with the mining of additional raw 
materials that would have to be utilized. In addition, the overall kiln 
efficiencies (i.e. the amount of fuel required per ton of clinker 
produced) at the facilities using fly ash would be expected to decrease 
if the fly ash were replaced with shale or clay. This decrease may be 
as large as 10 percent (See Site Visit to Lafarge Cement in Alpena 
Michigan in the docket).
    Based on the cost, energy, and adverse non-air quality health and 
environmental impacts, we believe that banning the current use of 
utility boiler fly ash is not justified.
    We also separately evaluated the use of fly ash from a utility 
boiler where activated carbon, or some other type of sorbent injection, 
has been used to collect mercury. This practice does not currently 
occur. See 70 FR 72344 (voicing concern about potential for increased 
mercury emissions from cement kilns were such fly ash to be used). The 
mercury concentration in this type of fly ash will vary widely. 
However, full scale testing of fly ash from utility boilers using 
various sorbent injection processes has indicated there is a potential 
for sorbent injection to significantly increase fly ash mercury content 
(Characterization of Mercury-Enriched Coal Combustion Residues from 
Electric Utilities Using Enhanced Sorbents for Mercury Control in the 
docket EPA-HQ-OAR-2002-0051). Testing to date has shown increases by a 
factor of 2 to 10, and in one case of a very low mercury fly ash the 
increase was by a factor of 70.
    Data from 16 cement facilities currently using fly ash not 
reflecting sorbed mercury showed mercury concentrations in the fly ash 
from 0.002 ppm to 0.685 ppm with a median of 0.136 ppm. Data on the fly 
ash mercury content of currently operating utility boilers testing 
sorbent injection showed levels ranging from 0.071 ppm up to 1.529 ppm 
with a median level of 1.156 ppm, significantly higher than the fly ash 
currently in use. Therefore, we see a potential for fly ash with 
enhanced mercury content due to sorbent injection at the utility site 
to increase mercury emissions from cement kilns, and for the increase 
to be much more significant than emissions attributable to the current 
fly ash being used.
    We do not see a ban on the use of this type of fly ash as 
significantly affecting the overall current beneficial uses of fly ash. 
First, we do not anticipate the widespread use of activated carbon 
injection ACI in the utility industry until 2010 or later. Therefore, 
both the cement industry and the utility industry will have a 
significant amount of time to adjust to this requirement. Second, a 
utility boiler that decides to apply ACI for mercury control has the 
option of collecting the fly ash from sorbent injection systems 
separately from the rest of the facility's fly ash (e.g., EPRI'S 
TOXECON system). Therefore, the utility boiler could continue to supply 
non-sorbent fly ash to a cement kiln even after the application of ACI 
for mercury control. Finally, technology is being developed that would 
allow the mineral-rich portion of fly ash to be separated from the high 
carbon/high mercury portion.
    Based on these factors, we are banning the use of utility boiler 
fly ash in cement kilns where the fly ash mercury content has been 
increased through the use of activated carbon or any other sorbent 
unless the facility can demonstrate that the use of that fly ash will 
not result in an increase in mercury emissions over baseline emissions 
(i.e., emissions not using the mercury increased fly ash). The facility 
has the burden of proving there has been no emissions increase over 
baseline. This requirement, adopted as a beyond-the-floor control, 
applies to both existing and new sources.
    We also reevaluated our analysis of potential control options based 
on add-on control technology. These were control options based on the 
use of a limestone scrubber, and ACI.
    As previously noted there are at least five cement kilns that have 
limestone (wet) scrubbers. As discussed in section IV.A.1.d above, 
there is a reasonable basis for believing that the wet scrubbers remove 
the oxidized mercury. There are no data available to allow us to 
definitively estimate the percent reduction expected. We performed a 
cost analysis based on an assumed mercury removal efficiency of 42 
percent, which is transferred (solely for purposes of analysis) \14\ 
from

[[Page 76526]]

performance of wet scrubbers in the utility boiler category and