Browse: Departments   Dates   Agencies  

CFR Citation: 40 CFR Part 63

EPA ID: [EPA-HQ-OAR-2002-0017; FRL-8576-3]

RIN ID: RIN 2060-AN99

NOTICE: Part V

DOCUMENT ACTION: Proposed rule.

SUBJECT CATEGORY: National Emission Standards for Hazardous Air Pollutants: Mercury Emissions from Mercury Cell Chlor-Alkali Plants

DATES: Comments. Comments must be received on or before August 11, 2008.

Public Hearing. If anyone contacts EPA by June 23, 2008 requesting to speak at a public hearing, a hearing will be held on July 11, 2008.

DOCUMENT SUMMARY: This action proposes amendments to the national emission standards for hazardous air pollutants (NESHAP) for mercury emissions from mercury cell chloralkali plants. This NESHAP (hereafter called the ``2003 Mercury Cell MACT'') limited mercury air emissions from these plants. Following promulgation of the 2003 Mercury Cell Maximum Achievable Control Technology (MACT) NESHAP, EPA received a petition to reconsider several aspects of the rule from the Natural Resources Defense Council (NRDC). NRDC also filed a petition for judicial review of the rule in the U.S. Court of Appeals for the DC Circuit. By a letter dated April 8, 2004, EPA granted NRDC's petition for reconsideration, and on July 20, 2004, the Court placed the petition for judicial review in abeyance pending EPA's action on
reconsideration. This action is EPA's proposed response to NRDC's petition for reconsideration.

We are not proposing any amendments to the control and monitoring requirements for stack emissions of mercury established by the 2003 Mercury Cell MACT. This proposed rule would amend the requirements for cell room fugitive mercury emissions to require work practice standards for the cell rooms and to require instrumental monitoring of cell room fugitive mercury emissions. This proposed rule would also amend aspects of these work practice standards and would correct errors and inconsistencies in the 2003 Mercury Cell MACT that have been brought to our attention.

SUMMARY: Environmental Protection Agency,

SUPPLEMENTAL INFORMATION

A. Does this action apply to me?

The regulated categories and entities potentially affected by this proposed action include:
Category NAICS code \1\ Examples of regulated entities Industry.......................... 325181............... Alkalis and Chlorine Manufacturing. Federal government................ ..................... Not affected. [[Page 33259]]
State/local/tribal government..... ..................... Not affected. \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 affected by this action. To determine whether your facility would be regulated by this action, you should examine the applicability criteria in 40 CFR 63.7682 of subpart IIIII, National Emission Standards for Hazardous Air Pollutants (NESHAP): Mercury Emissions from Mercury Cell ChlorAlkali (hereafter called the ``2003 Mercury Cell MACT''). If you have any questions regarding the applicability of this action to a particular entity, consult either the air permitting authority for the entity or your EPA regional representative as listed in 40 CFR 63.13 of subpart A (General Provisions).

B. What should I consider as I prepare my comments to EPA?

Do not submit information containing confidential business information (CBI) to EPA through www.regulations.gov or email. Send or deliver information identified as CBI only to the following address: Roberto Morales, OAQPS Document Control Officer (C40402),
Environmental Protection Agency, Office of Air Quality Planning and Standards, Research Triangle Park, North Carolina 27711, Attention Docket ID EPAHQOAR20020017. Clearly mark the part or all of the information that you claim to be CBI. For CBI information in a disk or CDROM that you mail to EPA, mark the outside of the disk or CDROM as CBI and then identify electronically within the disk or CDROM the specific information that is claimed as CBI. In addition to one complete version of the comment that includes information claimed as CBI, a copy of the comment that does not contain the information claimed as CBI must be submitted for inclusion in the public docket. Information so marked will not be disclosed except in accordance with procedures set forth in 40 CFR part 2.

C. Where can I get a copy of this document?

In addition to being available in the docket, an electronic copy of this proposed action will also be available on the Worldwide Web (WWW) through the Technology Transfer Network (TTN). Following signature, a copy of this proposed action will be posted on the TTN's policy and guidance page for newly proposed or promulgated rules at the following address: http://www.epa.gov/ttn/oarpg/. The TTN provides information and technology exchange in various areas of air pollution control. D. When would a public hearing occur?

If anyone contacts EPA requesting to speak at a public hearing concerning the proposed amendments by June 23, 2008, we will hold a public hearing on July 11, 2008. If you are interested in attending the public hearing, contact Ms. Pamela Garrett at (919) 5417966 to verify that a hearing will be held. If a public hearing is held, it will be held at 10 a.m. at the EPA's Environmental Research Center Auditorium, Research Triangle Park, NC, or an alternate site nearby.

E. How is this document organized?

The supplementary information in this preamble is organized as follows:
I. General Information

A. Does this action apply to me?

B. What should I consider as I prepare my comments to EPA?

C. Where can I get a copy of this document?

D. When would a public hearing occur?

E. How is this document organized?
II. Background Information

A. Reconsideration Overview

B. Industry Description

C. Regulatory Background

D. Details of the Petition for Reconsideration
III. Summary of EPA's Reconsideration and Proposed Amendments

A. What were the issues that EPA reconsidered, and what are EPA's proposed responses?

B. What amendments are EPA proposing?

C. What are the impacts of these proposed rule amendments? IV. Statutory and Executive Order Reviews

A. Executive Order 12866: Regulatory Planning and Review B. Paperwork Reduction Act

C. Regulatory Flexibility Act

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 and Safety Risks

H. Executive Order 13211 (Energy Effects)

I. National Technology Transfer Advancement Act

J. Executive Order 12898: Federal Actions to Address Environmental Justice in Minority Populations and LowIncome Populations
II. Background Information

A. Reconsideration Overview

On December 19, 2003, EPA promulgated the National Emission Standards for Hazardous Air Pollutants for Mercury Emissions from Mercury Chloralkali Plants (40 CFR part 63, subpart IIIII, 68 FR 70904), hereafter called the ``2003 Mercury Cell MACT.'' This rule for mercury cell chloralkali plants implemented section 112(d) of the Clean Air Act (CAA), which required all categories and subcategories of major sources listed under section 112(c) to meet hazardous air pollutant emission standards reflecting the application of the maximum achievable control technology (MACT). Mercury cell chloralkali plants are a subcategory of the chlorine production source category listed under the authority of section 112(c)(1) of the CAA. In addition, mercury cell chloralkali plants were listed as an area source category under section 112(c)(3) and (k)(3)(B) of the CAA. The 2003 Mercury Cell MACT satisfied our requirement to issue 112(d) regulations under each of these listings (for mercury).

The 2003 Mercury Cell MACT contained numerical emission limitations for the point sources of mercury emissions at mercury cell chloralkali plants. It also required that the plants either install mercury monitoring systems on the point source vents or that they test each vent manually at least once per week. The compliance date for the 2003 Mercury Cell MACT was December 19, 2006.

The 2003 Mercury Cell MACT also contained a set of work practice standards to address fugitive mercury emissions from the cell rooms. We determined that these procedures represented the MACT for the industry, and were considerably more stringent than the 40 CFR part 61 subpart E NESHAP requirements for control of mercury emissions (hereafter called the ``part 61 Mercury NESHAP'') that were applicable to this industry prior to the 2003 Mercury Cell MACT. An alternative compliance option was included in the 2003 Mercury Cell MACT that required mercury [[Page 33260]]
monitoring systems to be installed in the cell rooms with mandatory problem correction when a sitespecific mercury concentration action level is exceeded. As of December 19, 2006, the compliance date for the 2003 Mercury Cell MACT, all facilities but one have chosen this alternative compliance option.

On February 17, 2004, the Natural Resources Defense Council (NRDC) submitted to EPA an administrative petition asking us to reconsider several aspects of the 2003 Mercury Cell MACT under Clean Air Act section 307(d)(7)(B). On the same day, NRDC and the Sierra Club filed a petition for judicial review of the 2003 Mercury Cell MACT in the U.S. Court of Appeals for the DC Circuit (Civ. No. 041048). The focus of many of the issues raised in the petition for reconsideration was EPA's treatment of the fugitive cell room emissions in the 2003 Mercury Cell MACT. Specifically, NRDC asked EPA to reconsider (1) the decision to develop a set of work practice requirements under Clean Air Act section 112(h) in lieu of a numeric emission limitation for cell rooms; (2) the decision to make the promulgated work practices optional for sources that choose to undertake continuous monitoring; (3) the decision to not require existing facilities to convert to a mercuryfree chlorine manufacturing process; (4) the elimination of the previously applicable part 61 rule's 2,300 grams/day plantwide emission limitation; and (5) the decision to create a subcategory of mercury cell chloralkali plants within the chlorine production category.

By a letter dated April 8, 2004, Jeffrey Holmstead, thenEPA Assistant Administrator for Air and Radiation, notified the NRDC that EPA had granted NRDC's petition for reconsideration of the 2003 Mercury Cell MACT. On July 20, 2004, the Court granted EPA's motion to hold the case in abeyance pending EPA's action on reconsideration of the 2003 Mercury Cell MACT. Today's notice is EPA's proposed response to NRDC's petition for reconsideration.

B. Industry Description

There currently are five operating mercury cell chloralkali plants in the U.S., with one of these plants planning to convert to non mercury technology by 2012. These five plants are in Augusta, Georgia; Ashtabula, Ohio; Charleston, Tennessee; New Martinsville, West Virginia; and Port Edwards, Wisconsin. The Port Edwards, Wisconsin facility is the one that is expected to convert to nonmercury technology.

Mercury cell chloralkali plants produce chlorine and caustic soda (sodium hydroxide) or caustic potash (potassium hydroxide) in an electrolytic reaction using mercury. A mercury cell plant typically has many individual cells housed in one or more cell buildings. Mercury cells are electrically connected together in series.

At a mercury cell chloralkali plant, mercury is emitted from point sources (i.e., stacks) and fugitive sources. Mercury also leaves the plant in wastewater and solid wastes. There are three primary point sources of mercury emissions at mercury cell plants: The endbox ventilation system vent, the byproduct hydrogen system vent, and the mercury thermal recovery unit vents. Every mercury cell plant has a hydrogen byproduct stream, and most have an endbox ventilation system. However, not all of the plants have thermal mercury recovery units. Of the five plants currently operating, all five facilities have endbox ventilation systems and two have thermal mercury recovery units.

In addition to the stack emissions, there are fugitive mercury emissions at these plants. The majority of fugitive mercury emissions occur from sources inside the cell room such as leaks from cells, decomposers, hydrogen piping, and other equipment. Fugitive mercury emissions also occur during maintenance activities such as cell or decomposer openings, mercury pump changeouts, and endbox seal replacements, etc. All of this equipment and activities are located in the cell room, so these fugitive mercury emissions would be emitted via the cell room ventilation system.

There are potential fugitive air emission sources outside of the cell room. These potential outside sources include leaks of mercury contaminated brine in the brine treatment area, the wastewater system, and the handling and storage of mercury contaminated wastes. C. Regulatory Background

The part 61 Mercury NESHAP, which applied to all mercury cell chloralkali chlorine production plants prior to the 2003 Mercury Cell MACT, contained a numerical emission limit for mercury of 2,300 grams per day (g/day) for the entire plant. Point sources were limited to 1,000 g/day of mercury. If plants conducted a series of detailed design, maintenance, and housekeeping procedures, they were permitted under the part 61 rule to assume that fugitive mercury emissions from the cell room were 1,300 g/day, without having to demonstrate as such. All the mercury cell plants complied with the part 61 Mercury NESHAP using these assumptions rather than testing and determining actual fugitive cell room mercury emissions. Therefore, the extent of actual plantwide and cell room emissions that occurred under the part 61 rule could not be precisely determined.

In the 2003 Mercury Cell MACT rulemaking, pursuant to Clean Air Act section 112(d)(2) and (3), the regulatory analyses for the stack control requirements were based on the practices and controls of the lowest emitting plants out of the eleven facilities operating at the time of the MACT analyses. Existing mercury cell chloralkali production facilities with endbox ventilation systems were required by the 2003 Mercury Cell MACT to limit the aggregate mercury emissions from all byproduct hydrogen streams and endbox ventilation system vents to not exceed 0.076 grams (g) mercury (Hg) per megagram (Mg) chlorine (Cl2) for any consecutive 52week period. Existing mercury cell chloralkali production facilities without endbox ventilation systems were required to limit the mercury emissions from all byproduct hydrogen streams to not exceed 0.033 g Hg/Mg Cl2 for any consecutive 52week period.

The 2003 Mercury Cell MACT contained a set of work practice standards to address and mitigate fugitive mercury releases at mercury cell chloralkali plants. The MACT analysis for the requirements to reduce fugitive mercury emissions was based on the best practices of the eleven facilities operating at the time of the July 2002 proposal for the Mercury Cell MACT (see 67 FR 44672, July 3, 2002). These work practice provisions included specific equipment standards such as the requirement that end boxes either be closed (that is, equipped with fixed covers), or that end box headspaces be routed to a ventilation system (40 CFR 63.8192, ``What work practice standards must I meet?'', and Tables 1 through 4 to subpart IIIII of part 63). Other examples include requirements that piping in liquid mercury service have smooth interiors, that cell room floors be free of cracks and spalling (i.e., fragmentation by chipping) and coated with a material that resists mercury absorption, and that containers used to store liquid mercury have tightfitting lids (Table 1 to subpart IIIII of part 63). The work practice standards also included operational requirements. Examples of these include requirements to allow electrolyzers and decomposers to cool before opening, to keep liquid mercury in end boxes and mercury [[Page 33261]]
pumps covered by an aqueous liquid at a temperature below its boiling point at all times, to maintain end box access port stoppers in good sealing condition, and to rinse all parts removed from the decomposer for maintenance prior to transport to another work area (Table 1 to subpart IIIII of part 63).

A cornerstone of the work practice standards was the inspection program for equipment problems, leaking equipment, liquid mercury accumulations and spills, and cracks or spalling in floors and pillars and beams. Specifically, the 2003 Mercury Cell MACT required that visual inspections be conducted twice each day to detect equipment problems, such as end box access port stoppers not securely in place, liquid mercury in open containers not covered by an aqueous liquid, or leaking vent hoses (Table 2 to subpart IIIII of part 63). If a problem was found during an inspection, the owner or operator was required to take immediate action to correct the problem. Monthly inspections for cracking or spalling in cell room floors were also required as well as semiannual inspections for cracks and spalling on pillars and beams. Any cracks or spalling found were required to be corrected within 1 month. Visual inspections for liquid mercury spills or accumulations were also required twice per day. If a liquid mercury spill or accumulation was identified during an inspection, the owner or operator was required to initiate cleanup of the liquid mercury within 1 hour of its detection (Table 3 to subpart IIIII of part 63). In addition to cleanup, the 2003 Mercury Cell MACT required inspection of the equipment in the area of the spill or accumulation to identify the source of the liquid mercury. If the source was found, the owner or operator was required to repair the leaking equipment as discussed below. If the source was not found, the owner or operator was required to reinspect the area every 6 hours until the source was identified or until no additional liquid mercury was found at that location. Inspections of specific equipment for liquid mercury leaks were required once per day. If leaking equipment was identified, the 2003 Mercury Cell MACT required that any dripping mercury be contained and covered by an aqueous liquid, and that a first attempt to repair leaking equipment be made within 1 hour of the time it is identified. Leaking equipment was required to be repaired within 4 hours of the time it is identified, although there are provisions for delaying repair of leaking equipment for up to 48 hours (Table 3 to subpart IIIII of part 63) under certain conditions.

Inspections for hydrogen gas leaks were required twice per day. For a hydrogen leak at any location upstream of a hydrogen header, a first attempt at repair was required within 1 hour of detection of the leaking equipment, and the leaking equipment was required to be repaired within 4 hours (with provisions for delay of repair if the leaking equipment was isolated). For a hydrogen leak downstream of the hydrogen header but upstream of the final control device, a first attempt at repair was required within 4 hours, and complete repair required within 24 hours (with delay provisions if the header is isolated) (Table 3 to subpart IIIII of part 63).

The work practice standards in the 2003 Mercury Cell MACT required that facilities institute a floor level mercury vapor measurement program (See Sec. 63.8192, ``What work practice standards must I meet?'', specifically paragraph (d)). Under this program, mercury vapor levels are periodically measured and compared to an action level of 0.05 mg/m3. The 2003 Mercury Cell MACT specified the actions to be taken when the action level is exceeded. If the action level was exceeded during any floorlevel mercury vapor measurement evaluation, facilities were required to take specific actions to identify and correct the problem (Sec. 63.8192(d)(1) through (4)).

As an alternative to the full set of work practice standards (including the floorlevel monitoring program), the 2003 Mercury Cell MACT included a compliance option to institute a cell room monitoring program (See Sec. 63.8192, ``What work practice standards must I meet?'', specifically paragraph (g)). In this program, owners and operators continuously monitor the mercury concentrations in the upper portion of each cell room and take corrective actions as soon as practicable when a sitespecific mercury vapor level is detected. The cell room monitoring program was not designed to be a continuous emissions monitoring system inasmuch as the results would be used only to determine relative changes in mercury vapor levels rather than compliance with a cell room emission or operating limit (68 FR 70922).

As part of the cell room monitoring program, the owner or operator was required to establish an action level for each cell room based on preliminary monitoring to determine normal baseline conditions (See Sec. 63.8192, ``What work practice standards must I meet?'', specifically paragraph (g)(2)). Once the action level(s) was established, continuous monitoring of the cell room was required during all periods of operation. If the action level was exceeded at anytime, actions to identify and correct the source of elevated mercury vapor were required to be initiated as soon as possible. If the elevated mercury vapor level was due to a maintenance activity, the owner or operator was required to ensure that all work practices related to that maintenance activity were followed. If a maintenance activity was not the cause, inspections and other actions were needed to identify and correct the cause of the elevated mercury vapor level. Owners and operators utilizing this cell room monitoring program option were required to develop sitespecific cell room monitoring plans describing their monitoring system and quality assurance/quality control procedures that were to be used in their monitoring program (Table 5 to subpart IIIII of part 63).

The 2003 Mercury Cell MACT established the requirement for owners and operators to routinely wash surfaces throughout the plant where liquid mercury could accumulate (See Sec. 63.8192, ``What work practice standards must I meet?'', specifically paragraph (e)). Owners and operators were required to prepare and follow a written washdown plan detailing how and how often certain areas specified in the 2003 Mercury Cell MACT were to be washed down to remove any accumulations of liquid mercury (Table 7 to subpart IIIII of part 63).

For new or reconstructed mercury cell chloralkali production facilities, the 2003 Mercury Cell MACT prohibited mercury emissions.

Several mercury cell plants have closed or converted to membrane cells since the promulgation of the 2003 Mercury Cell MACT. When these situations have occurred at plants with onsite thermal mercury recovery units, it has been common for these units to continue to operate to assist in the treatment of wastes associated with the shutdown/conversion. Under the applicability of the 2003 Mercury Cell MACT, these units are no longer an affected source after the chlorine production facility ceased operating. Although these mercury recovery units were required to continue to use controls as per their state permits, these proposed amendments would require any mercury recovery unit to continue to comply with the requirements of the Mercury Cell MACT for such units even after closure or conversion of the chlorine production facility, as long as
[[Page 33262]]
the mercury recovery unit continues to operate to recover mercury. D. Details of the Petition for Reconsideration

On February 17, 2004, under section 307(d)(7)(B) of the Clean Air Act, the NRDC submitted to EPA an administrative petition asking us to reconsider the 2003 Mercury Cell MACT. NRDC and the Sierra Club also filed a petition for judicial review of the rule in the U.S. Court of Appeals for the DC Circuit (NRDC v. Sierra Club v. EPA, Civ. No. 04 1048). Underlying many of the issues raised in the petition for reconsideration was the uncertainty associated with the fugitive emission estimates used by EPA in the rulemaking. In particular, the NRDC had concerns over the inability of mercury cell plants to account for all the mercury added to their processes to replace mercury that leaves in products or wastes or leaves via air emissions. NRDC, along with a number of other concerned parties who submitted comments on the July 2002 proposed rule, believed that the majority of this ``missing'' or unaccounted mercury must be lost through fugitive emissions. They also contended that recognition of this asserted fact would cause EPA to change many of the decisions that had been made in developing and promulgating the 2003 Mercury Cell MACT. Specifically, NRDC raised the following five issues in its petition:
(1) EPA refused to establish a numeric emission standard for the cell room, choosing instead to develop a set of work practices designed to minimize emissions. NRDC argued that under Clean Air Act section 112(h) EPA is permitted to substitute work practices for emission limits only upon a finding that ``it is not feasible * * * to prescribe or enforce an emission standard.''
(2) EPA's 2003 Mercury Cell MACT unreasonably backtracked from the work practices the Agency proposed. As part of the regulatory effort, EPA had surveyed the work practices used by facilities in the industry and concluded that the housekeeping activities that sources followed to comply with the part 61 Mercury NESHAP represented the MACT floor. The EPA then required these detailed housekeeping practices that were based upon the best levels of activity in the industry. But despite the results of its survey and findings, EPA made the work practices optional in the 2003 Mercury cell MACT, allowing facilities to choose not to do the housekeeping activities and to instead perform continuous monitoring. EPA then stated that ``a comprehensive continuous cell room monitoring program should be sufficient to reduce fugitive mercury emissions from the cell room without imposing the overlapping requirements of the detailed work practices.''
(3) EPA failed to consider nonmercury technology as a beyond thefloor MACT control measure for existing sources even though eliminating the mercury cell process would totally eradicate mercury emissions and also would be costeffective, based on NRDC's expectations of the amount of fugitive mercury emissions from subject sources.
(4) EPA eliminated a 2,300 g/day limit on plantwide mercury emissions that existed under the part 61 Mercury NESHAP. NRDC stated that doing so violated the CAA because the law generally prohibits the new emission standards under section 112 from weakening more stringent existing requirements.
(5) EPA inappropriately decided to create a subcategory of mercury cell plants within the chlorine production category.

In a letter dated April 8, 2004, EPA generally granted NRDC's petition for reconsideration, and indicated we would respond in detail in a subsequent rulemaking action. In addition, in meetings between EPA staff and NRDC representatives, EPA agreed to address the uncertainty of EPA's fugitive mercury emissions from this industry. The Court stayed the litigation while the Agency addressed the uncertainty issues, conducted additional testing, and reconsidered the rulemaking. III. Summary of EPA's Reconsideration and Proposed Amendments

In this section, we describe actions that we undertook in support of the proposed reconsideration of the rule, especially as related to the issues raised by NRDC in its petition for reconsideration. We present our proposed conclusions and decisions in response to NRDC's petition, and we summarize the rule amendments that we are proposing in today's action, along with our estimate of the impacts of these amendments.

These proposed amendments would be applicable to affected facilities when the final rule amendments are published, with proposed compliance periods of 60 days for facilities that have complied with the 2003 Mercury Cell MACT by selecting the continuous cell room monitoring option of that rule, and 2 years for facilities that have complied with the 2003 Mercury Cell MACT by selecting the work practice option. Mercury recovery units at sites where mercury cells are closed or converted after the date that the final rule amendments are published would be required to comply with the requirements of the final amendments as long as they are in operation.
A. What were the issues that EPA reconsidered, and what are EPA's proposed responses?

As discussed above in section (II)(D), NRDC's petition listed five specific issues. Our reconsideration of each of these issues is addressed below. First, however, we also present a discussion of another issue that we believe relates to much of NRDC's petition: The magnitude of the fugitive mercury emissions from mercury cell chlor alkali plants.
1. Magnitude of Fugitive Mercury Emissions from Mercury Cell Chlor alkali Plants

It has been difficult to quantify fugitive mercury emissions from mercury cell chloralkali plants. During most of the time when the 2003 Mercury Cell MACT was being developed, we were aware of fewer than five mercury emissions studies conducted over the last 30 or more years in the U.S. and Europe that measured fugitive emissions from mercury cell plants. Two of these studies were conducted by EPA in the early 1970's and formed the basis for the assumption of 1,300 g/day mercury cell room emissions of the part 61 Mercury NESHAP. During the development of the 2003 Mercury Cell MACT, EPA conducted a study at Olin Corporation's mercury cell plant in Augusta, Georgia (hereafter called ``Olin Georgia''), that provided an additional estimate of fugitive mercury emissions.

In the time period since mercury cell chloralkali plants were required to comply with the part 61 Mercury NESHAP, which was promulgated in April of 1973, we are not aware of any facility that conducted testing to demonstrate compliance with the cell room emission limitation of the part 61 Mercury NESHAP. Instead, all facilities carried out the set of approved design, maintenance, and housekeeping practices and assumed fugitive mercury emissions of 1,300 g/day, as was permitted by the part 61 NESHAP.

The sensitivity and concern over the actual levels of fugitive mercury emissions from the cell rooms was exacerbated by the inability of the industry to fully account for all the mercury that was added to the cells. In the preamble to the final 2003 Mercury Cell MACT (68 FR 70920), we stated the following: ``Even with this decrease in consumption, significant mercury remains unaccounted for by the industry. The mercury releases reported to the air, water, and solid wastes in the 2000 Toxics Release Inventory (TRI) totaled around 14 tons. This leaves approximately 65 tons of consumed mercury that is not accounted for in the year 2000.'' While industry representatives provided explanations for this discrepancy, they could not fully substantiate their theories.

[[Page 33263]]

Although we acknowledged the uncertainty in the accounting of all the mercury, we stated in the 2003 Mercury Cell MACT that no evidence has ever been provided to indicate that the unaccounted mercury is emitted to the atmosphere via fugitive emissions from the cell room or otherwise. In its petition for reconsideration and in other correspondence, NRDC cites information that it believes supports a conclusion that the unaccounted mercury is emitted from the cell room. However, NRDC did not address studies that have been conducted to measure fugitive mercury emissions from mercury cell plants that rebut that conclusion.

Historically, the highest daily emission rate reported for any cell room has been approximately 2,700 g/day for a plant operating in 1971, which was before the part 61 Mercury NESHAP was in effect. More recent studies show fugitive mercury emissions considerably lower than the 1,300 g/day assumption in the part 61 Mercury NESHAP. For example, a study in 1998 at the Holtrachem facility in Orrington, Maine, estimated a fugitive mercury emission rate between 85 and 304 g/day. A study in Sweden in 2001 estimated a daily fugitive emission rate of 252 g/day. While NRDC cites various peripheral aspects of the EPA study in 2000 study at Olin's Georgia mercury cell plant, NRDC does not discuss a primary conclusion of the test: That the facility was estimated to have an average fugitive mercury emission rate of 472 g/day.

While we were confident that the fugitive emissions from cell rooms were not at the very high levels estimated by NRDC (at several tons per year (tpy) per plant), we recognized that the body of fugitive mercury emissions data could be improved. Therefore, as part of our reconsideration of the 2003 Mercury Cell MACT, we collected additional information on fugitive mercury emissions from mercury cell chlor alkali plants. The primary purpose of this effort was to address whether the fugitive emissions from a mercury cell chloralkali plant are on the order of magnitude of the historical assumption of 1,300 g/ day, corresponding to 0.5 tons per year (tpy) per plant, or on the order of magnitude of the unaccounted for mercury in 2000, which would correspond to 3 to 5 tpy per plant, or at some other level.

In planning our information gathering efforts for this test program, we recognized that all of the previous studies were relatively short term. Fugitive mercury emissions from a mercury cell plant occur for numerous reasons, with significant emission sources likely being leaking or malfunctioning equipment and maintenance activities that expose mercury normally enclosed in process equipment to the atmosphere. One noteworthy NRDC criticism of the Olin Georgia study was that no major ``invasive'' maintenance activities were performed during the testing. Therefore, in designing our new study, we collected data over a number of months during a wide range of operating conditions and during times when all major types of maintenance activities were conducted.

Consequently, as part of the reconsideration efforts for the 2003 Mercury Cell MACT, EPA sponsored a test program to address the issue of the magnitude of the fugitive mercury emissions at mercury cell chlor alkali plants. We visited five mercury cell chloralkali plants to identify and evaluate the technical, logistical, and/or safety issues associated with the measurement of fugitive emissions from the mercury cell rooms as part of a test program. The result of these efforts was that we sponsored two emissions testing programs: One at the Olin mercury cell chloralkali plant in Charleston, Tennessee (hereafter called ``Olin Tennessee''), to estimate mercury emissions from one of its three cell rooms; and the other at the Occidental Chemical mercury cell chloralkali plant in Muscle Shoals, Alabama (hereafter called ``Occidental Alabama''), to estimate their total site mercury emissions. These testing programs are discussed in detail later in this notice.

In addition to these emissions measurements, we also collected mercury emissions data from the continuous mercury monitoring system installed at three mercury cell plants: The Occidental facility in Delaware City, Delaware (hereafter called ``Occidental Delaware''); Occidental Alabama; and Olin Tennessee, which was also a site for the EPA emissions measurement tests. We also performed validation studies of the air flow measurement systems and mercury monitors at these three facilities.

In addition, we compared maintenance logs and mercury emissions data to establish the correlation, if any, between maintenance activities and mercury emissions using data from Occidental's facilities. And finally, we addressed the issue of significant sources of fugitive mercury emissions from outside the cell room from the data acquired at the EPAsponsored total site emissions tests at Occidental Alabama.

The descriptions of the emissions testing and data gathering efforts are summarized below along with our estimates of fugitive mercury emissions derived from these studies. The full emissions test reports, two memoranda that summarize the test reports, validation reports, and summaries of the mercury monitoring system emissions data analyses can be found in the docket to this proposed rule (EPAHQOAR 20040017), and were previously provided to NRDC and industry representatives.
a. Description of EPASponsored Mercury Emissions Tests at Two Facilities

OlinCharleston, Tennessee. This test was performed over a six week period from August to October 2006 using a longpath ultraviolet differential optical absorption spectrometer (UVDOAS) to continuously measure the mercury concentration in the ventilator and an optical scintillometer (anemometer) to measure the velocity. Emission estimates were reported for each 24hour period. The test report can be found in the docket, item number EPAHQOAR200200170056.3.

The Olin Tennessee facility has three cell rooms installed adjacent to one another. The E510 cellroom (startup in 1962) is a simple rectangular design with two rows of cells. The E812 cell room (startup in 1968) is also a simple rectangular design with two rows of cells. In 1974, Olin added a third cell room with additional E812 cells just south of the existing E812 cell room. A central control area was installed between the E510 and E812 cell rooms. In addition, an elevator and computer equipment area was installed between the two original plants. The area between the original E812 cells and the E812 10cell Expansion is fully open. Each of the three cell rooms has a full length, natural draft ventilator mounted on the roof. Fans have been installed at the cell floor level around the perimeter of the E510 and E812 cell rooms to enhance cool air flow in key work areas. In addition, high velocity fans were installed near the central control area to aid air movement in ``dead zones'' created by the control area walls. There are no exhaust fans in any of the cell rooms.

Logistical and cost considerations resulted in the E510 cell room being selected for the EPA test. Continuously measuring the mercury emissions from more than one ventilator simultaneously was not practical, based on the limited availability of equipment and the complexities related to the operation of a number of highly sophisticated
[[Page 33264]]
measurement devices. The small size of the E812 Expansion cell room excluded it from consideration, and the complicated flow patterns between the E812 and E812 Expansion rooms would have made it very difficult to account for all the associated uncertainties using only one monitor. The configuration of the E510 cell room, the relatively straightforward air flow pattern, and the structure of the ventilator (which allowed easy access and a clear path for the beams) made it the obvious choice for the test program to optimize our ability to obtain the most reliable data.

OccidentalMuscle Shoals, Alabama. This test was conducted over 53 days, from September 21, 2006, through November 12, 2006, to measure total site mercury emissions. For this study, the ``total site'' included emissions via the cell room ventilation system, the stacks/ point sources (thermal mercury recovery unit vent, hydrogen byproduct vent, endbox ventilation vent), and any fugitives that occurred outside of the cell room in adjacent process areas. The measurement approach used a Vertical Radial Plume Mapping (VRPM) measurement configuration employing three openpath UVDOAS instruments for elemental mercury concentration measurements, in conjunction with multipoint ground level mercury measurements with a Lumex mercury analyzer. The total site mercury emissions were estimated using these concentration measurements and meteorological data (e.g., wind speed, wind direction).

The measurement systems operated on a 24 hour, 7 day per week basis for the 53day campaign. The 3beam VRPM configuration used to estimate elemental mercury emissions from the facility was located at a fixed position and fixed orientation on site for the duration of the project. Calculations of mercury flux through the VRPM plane were conducted only when specific data quality indicators involving wind speed, wind direction, path averaged concentration ratios and instrument operation were met. During the 53day emissions test program, VRPM mercury flux values were able to be calculated for 23 days. Data were reported as daily (24 hour) emission values that were extrapolated from rolling 20 minute averages calculated every four minutes. A total of 1,170 mercury emission flux estimates were produced during the 23 days. The test report can be found in the docket, item number EPAHQOAR20020017 0056.5.

The cell room at the now closed Occidental Alabama plant was a rectangular building measuring 260 feet by 357 feet. The cell room consisted of two rows of cells broken into four sections. The cell room took up half of a larger building, with a wall separating the cell room from the other half of the building that was used for equipment storage. The peak of the roof was over the wall separating the cell room from the other side of the building. The ventilation for the cell room consisted of both induced and forced draft fans. There were 43 forceddraft fans positioned on the side wall of the building pushing air towards the center of the building. There were two rows of induced draft fans on the roof of the cell building. One row, containing 33 fans, was directly over the center of the two rows of cells. The other row, which contained 32 fans, was at the peak of the roof. The result was that the building was constantly under a slightly negative pressure.
b. EPA Validations of Mercury Monitoring Systems in Cell Rooms of Mercury ChlorAlkali Plants

During the time we were planning the testing programs to estimate fugitive mercury emissions via an EPAsponsored test program, the mercury cell chloralkali industry was undertaking its own longterm mercury emissions estimation efforts. Two Occidental mercury cell plants (Delaware and Alabama) installed mercury monitoring systems in their cell rooms in 2005, and the Olin Tennessee facility installed a mercury monitoring system in 2006. The plants used these systems to identify and correct mercury emission episodes in accordance with the alternative cell room monitoring program of the 2003 Mercury Cell MACT. Specifically, the facilities monitored physical and chemical parameters in the cell room, such as air flow and mercury concentration, that allowed the continuous estimation of the relative mass of mercury emissions leaving the cell room. Since these plants had already installed and were currently running their mercury monitoring systems, we included the collection and evaluation of data from these systems in our data gathering program. The overall goal of our validation program was to provide a qualitative assessment of the mercury monitoring systems at these three facilities.

There were three specific objectives of the EPA validation studies. The first objective was to verify that facility data processing and archiving were being performed correctly. This was accomplished through comparison of facility data with independently calculated values for elemental mercury mass emission rates. These independent calculations utilized the same equations and raw input data as the company data systems. The second objective was to establish a confidence level for the accuracy of the measured elemental mercury concentrations. To accomplish this, a systems assessment was performed using calibration standards to challenge the mercury analyzer with a known concentration of mercury and to compare the analysis results with the certified concentration of the calibration standard. The goal of this assessment was an evaluation of shortterm operation of the elemental mercury analyzer and effectiveness of routine maintenance and calibration activities that may impact longterm operation of the instrument. The third objective was to establish a confidence level associated with the flow determinations. Since each cell room has a unique ventilation system, this flow determination validation was done somewhat differently for each mercury monitoring system.

The following are descriptions of the mercury monitoring system at each faculty and the results of the corresponding validation studies. The final reports for the validation program at the two Occidental facilities can be found in the docket to this rule (see docket items EPAHQOAR200200170057 and 00170058). The validation tests
performed at Olin's Tennessee facility are included within the emissions test report described above (see docket item number EPAHQ OAR200200170056.3).

OccidentalDelaware City, Delaware. Validation tests were performed by EPA at Occidental's now closed facility in Delaware the weeks of August 22, 2005, and September 9, 2005. The cell room at the Delaware City Plant was a rectangular building measuring 352 feet by 140 feet. The cell room consisted of two independent circuits, and each circuit was broken into two sections, resulting in four quadrants. The air flow in the cell room was via natural convection; there were no fans to provide either induced or forced draft air flow. During the summer months, approximately 40 percent of the sides on the lengthwise span were removed to improve ventilation. There were two rows of roof ventilators. Each ventilator was in two discrete sections for a total of four sections (corresponding to the four quadrants of the cell room).

The mercury monitoring system at the Occidental Delaware facility was a Mercury Monitoring System Model MMS16 analyzer manufactured by Mercury Instruments GmbH Analytical Instruments in Germany. It collects samples from 16 points and analyzes
[[Page 33265]]
them for elemental mercury using a Model VM3000 ultraviolet absorption analyzer. The mercury monitoring system takes one sample per minute, meaning that a sample is taken from each point once every 16 minutes. The sampling sequence is established so that a sample is taken from each quadrant once every four minutes. The flow rate for the building is estimated using a convective air flow model. The inputs to this model are atmospheric and ridge vent temperatures (which are continuously monitored), intake and discharge areas, and stack height.

The validation of the Occidental Delaware mercury monitoring system confirmed the accuracy of the data collection, calculation, and archiving system. With regard to the data quality of the mercury analyzer, mercury calibration accuracy results for the Delaware City instrument were 20 percent and 10 percent for the mid and highrange calibration standards, respectively. Specifically, the analyzer reported a concentration of 8 micrograms per cubic meter ([mu]g/ m³) for the 10 [mu]g/m³ standard and a concentration of 45 [mu]g/m³ for the 50 [mu]g/m³ standard. These results, along with the line integrity test results, suggest that the high range calibration of this instrument was offset in a negative direction.

A qualitative assessment of the accuracy of the Delaware City facility's approach to flow estimation was made with independent, on site, flow measurements using a vane anemometer at the roof vents. These measurements, covering multiple sampling points, were averaged and compared to the average air flow determined using the convective flow model equations used to estimate the flow. This evaluation showed that the difference between the anemometer and convective flow model methods was 29 percent, with the convective flow model reporting a higher value than the anemometer tests.

OccidentalMuscle Shoals, Alabama. Validation tests were performed by EPA at Occidental Alabama the week of September 12, 2005. The mercury monitoring system at this facility was a Mercury Monitoring System Model MMS16 analyzer manufactured by Mercury Instruments GmbH Analytical Instruments in Germany. The elemental mercury concentration is measured using a Model VM3000 ultraviolet absorption analyzer. The mercury monitoring system collects samples from 65 points (at the inlet to each induced draft fan) and combines them in groups of three or four to provide a representative profile of the cell room in a 20 point sample array. The mercury monitoring system takes one sample per minute, meaning that a sample is taken from each point once every 20 minutes. We previously described the cell room at Occidental Alabama, above.

To estimate the flow rate from the cell room, Occidental tested each fan to determine the flow rate at standard conditions and to correct the actual flow rate based on continuous monitoring of temperature, pressure, and humidity. The assessment of the accuracy of the Muscle Shoals facility's flow estimation procedure was made with independent, onsite, flow measurements at each of the 65 fan outlets. The total flow through all 65 fans was measured at five points within the fan exhaust area using an anemometer. The exhaust flow from each fan was determined by averaging these five flow values. Total flow from the cell room was determined by subsequently summing the flow from each fan during the test period. The difference between the anemometer and fan flow model methods was slightly more than 7 percent, with the exhaust fan model reporting a higher value than the anemometer validation tests.

The validation of the Occidental Alabama continuous mercury monitoring system confirmed the accuracy of the data collection, calculation, and archiving system of the facility. The mercury calibration accuracy results for the Muscle Shoals facility instruments were 4.0 percent and 0.2 percent, for the mid and highrange calibration standards, respectively. These results indicate that the Muscle Shoals mercury analyzer was in good operating condition with no apparent calibration problems at the time of the validation test.

OlinCharleston, Tennessee. Validation tests were performed by EPA at the Olin Tennessee facility during the month of September 2006. We previously described the cell rooms at the Olin Tennessee plant, above. This facility has two separate mercury monitoring systems: One for the E510 cell room and one for the E812/E812 Expansion rooms. These mercury monitoring systems are Mercury Monitoring System Model MMS16 analyzers manufactured by Mercury Instruments GmbH Analytical Instruments in Germany. The mercury monitoring system collect samples from individual points and analyze them for elemental mercury using a Model VM3000 ultraviolet absorption analyzer. In each of the cell rooms, there are five sampling points evenly spaced along the ventilators. In addition to the sample points in the ventilators (five for the E510 system and ten for the E812/812 Expansion system), each mercury monitoring system has one sample point dedicated to continuously measuring mercury for point sources subject to the 2003 Mercury Cell MACT, and one point used for calibration. Each point is sampled for one minute and the concentration is held and used in calculating the overall cell room average concentration until the point is sampled in the next cycle. Hourly and daily rolling averages are then calculated and stored. The flow rates for the cell rooms are estimated separately using a convective air flow model. The inputs to this model are atmospheric and ridge vent temperatures (which are continuously monitored), intake and discharge areas, discharge height, and fans on/off operation.

The mercury calibration accuracy results for the instrument in the E510 cell room were approximately 8 percent and 19 percent for the mid and high range calibration standards, respectively. For the E812/812 Expansion System, the results were approximately 5 percent and 20 percent for the mid and high range calibration standards, respectively. Both analyzers indicated higher concentrations than the certified calibration standards provided by the manufacturer.

Manual flow measurements were made in each of the cell room roof vents using a vane anemometer. These manual flow measurements were not compared directly with flow rates estimate by Olin's convective flow model. The accuracy of the facility's model was assessed in a twostep process. The manual measurements for the E510 cell room were first compared with the air flow measurements estimated using the optical anemometer in the EPA test, and then compared with the estimates from the Olin flow model. The accuracy determination between the optical flow monitor and the manual flow measurements was slightly lower than 10 percent. The flow rate estimated using the Olin flow model was approximately 5 percent higher than the flow rate measured by the optical flow monitor over the entire testing period.
c. Analyses of Cell Room Maintenance Logs and Mercury Emissions Data

Occidental also provided detailed maintenance records for the April through November 2005 (Delaware) and August 2005 through January 2006 (Alabama) time periods in addition to their emissions data. They also provided production data and details of ``alarm events'' for this period, where an alarm event was a situation in which the monitoring system recorded a mercury concentration above established action levels. When such an alarm occurred,
[[Page 33266]]
Occidental personnel were dispatched to the area of the cell room where the elevated concentration was detected to identify the specific cause and to take corrective actions. We performed an analysis of the effect of maintenance activities, alarm events, production levels, and ambient conditions on daily fugitive mercury emission levels. While we recognize that maintenance activities and alarm events can result in shortterm spikes in emissions, our analyses of the data did not show any correlation between daily fugitive mercury emissions and these events. The only factor that showed any correlation, albeit weak, to daily emissions was the ambient temperature. The report of these analyses can be found in the docket.
d. No Significant Fugitive Sources of Mercury From Outside the Cell Room

In addition to obtaining total site emission estimates at Occidental Alabama, we attempted to ascertain whether fugitive sources outside of the cell room were contributors of measurable emissions by performing a material balance on the contributors to the total site emissions and solving for the outside fugitive component.

The ``total site'' mercury emissions for this study included emissions via the cell room ventilation system, the stacks/point sources (thermal mercury recovery unit vent, hydrogen byproduct vent, endbox ventilation vent), and any fugitives that occurred outside of the cell room in adjacent process areas. From a material balance analysis of these data, we concluded that fugitive sources outside the cell room do not contribute measurable mercury emissions when compared to fugitive emissions from the cell room (see docket items EPAHQOAR 200200170056.5 and 00170056.6).
e. New EPA Fugitive Mercury Emission Estimates for Cell Rooms

We used eight separate fugitive mercury emission data sets from three different mercury cell chloralkali plants in 2005 and 2006 to produce a new estimate of fugitive mercury emissions from cell rooms. The time periods of data collection range from 6 weeks to over 30 weeks, all of which provided an opportunity to include a complete range of maintenance activities and operating conditions. Two of the data sets were generated via EPAsponsored test programs and the others were collected from cell room mercury monitoring systems that were validated by EPA. Summaries of the data sets can be found in the docket.

The daily mercury emission rates extrapolated from these data sets ranged from around 20 to 1,300 g/day per facility. The average daily emission rates ranged from around 420 g/day to just under 500 g/day per facility, with the mean of these average values being slightly less than 450 g/day per facility.

The purpose of this effort was to address whether the fugitive emissions from a mercury cell chloralkali plant are on the order of magnitude of the historical assumption of 1,300 g/day (or 0.5 tpy per plant) or on the order of magnitude of the unaccounted for mercury in 2000 (3 to 5 tpy per plant, which equates to around 10,000 g/day). The information we obtained shows that fugitive emissions are on the order of magnitude of the historical assumption of 1,300 g/day. There was no evidence obtained during any of the studies that indicated that fugitive mercury emissions were at levels higher than 1,300 g/day. In addition, all of the studies that produced these data were of sufficient duration to encompass all types of maintenance activities, including the major ``invasive'' procedures that were not conducted during the earlier test at the Olin Georgia facility. The length of these studies was also sufficient to include emissions from a variety of process upsets, such as: Liquid mercury spills, leaking cells, and other process equipment, and other process upsets (see docket items EPAHQOAR200200170021 and 00170029).

The results of the almost one million dollar study of fugitive emissions from mercury cell chloralkali plants sponsored by EPA enables us to conclude that the levels of fugitive emissions for mercury chloralkali plants are much closer to the assumed emissions in the part 61 Mercury NESHAP, of 1,300 g/day/plant (around 0.5 tons/yr/ plant) than the levels assumed by NRDC (3 to 5 tons/yr/plant). The results of this study suggest that the emissions are routinely less than half of the 1,300 g/day level, with overall fugitive emissions from the five operating facilities estimated at less than 1 ton per year of mercury.
f. Conclusions on the Use of Mercury Monitoring Systems as a Work Practice Tool

In the data we obtained or examined, we saw discrepancies between the measured concentrations and the calibrated standards, and differences between the flow rates estimated by the cell room systems and those estimated by anemometers (manual or optical), as summarized above. The differences for the measurement of the mercury concentration were as high as 20 percent, and the differences in the measurements for the flow rates were as high as 29 percent. Such differences lead us to conclude that these systems would not be suitable to accurately demonstrate compliance with a numeric standard, because of the potential for errors in compliance determinations due to uncertainties in the measurement techniques. However, since the goal of this effort was to assess the order of magnitude of fugitive mercury emissions from the cell room, we concluded that data from these systems were appropriate for that purpose since the differences were well within an order of magnitude.

Our observations at these three plants during the validation programs resulted in recognition of the ability of the mercury monitoring system to be used as a work practice tool to reduce fugitive emissions in the cell room. When the 2003 Mercury Cell MACT was promulgated, we thought that the mercury monitoring system could help identify problems before significant emission events occurred. However, at that time no mercury cell plant in the United States had installed such technology so there was no opportunity to assess their effectiveness. Now, with data from the three plants described above, we can conclusively say that the mercury monitoring systems aid in the identification and correction of fugitive emission problems and help plants refine their standard operating procedures and work practices to further reduce emissions. Therefore, we believe that the use of such systems as a tool to determine the effectiveness of work practices has been demonstrated. We estimate that the cost of installing a system in a cell room is about $120,000, which equates to a total annual cost (including annualized capital cost and operation and maintenance costs) of slightly over $25,000 per year. We believe that in the long term these systems will result in continued decreases in fugitive mercury emissions as plants will be able to identify emissionreducing improvements in their processes and practices. Therefore, we are proposing to require all mercury cell chloralkali plants to install cell room mercury monitoring systems and to develop a cell room monitoring plan.
g. Estimate of the Efficiency of the Cell Room Monitoring Program To Reduce Fugitive Emissions

In the 2003 Mercury Cell MACT, we noted our inability at that time to quantify the emission effects of adopting the cell room work practices, a point also noted by NRDC in its petition for
[[Page 33267]]
reconsideration. However, we are now able to better estimate the emissions reductions achieved by the cell room monitoring program and work practices for these amendments using the results of the test programs and other information gathering efforts, as described above.

We estimated that baseline mercury emissions prior to the 2003 Mercury Cell MACT were 1,300 g/day per facility (68 FR 70923). This equated to nationwide preMACT baseline fugitive emissions of 4.7 tpy. The test program data suggest that on average, the fugitive mercury emissions from a single facility are approximately 450 g/day, which equates to nationwide emissions of 0.9 tpy. Therefore, we estimate that the combination of the work practices promulgated in the 2003 Mercury Cell MACT combined with cell room monitoring reduces fugitive mercury emissions from a single facility by over 65 percent from the preMACT levels. On a nationwide basis, we estimate that fugitive mercury emissions have been reduced by approximately 86 percent, including plant closures.

The point source emissions (from hydrogen vents, endbox ventilation systems, and mercury recovery units) from the five mercury cell plants expected to be in operation after these amendments are finalized are around 0.4 tons/yr total. Therefore, our estimate of the nationwide total mercury emissions from all emission sources (point and fugitive) at these plants is around 1.3 tons/yr.
2. Elimination of Uncertainty Regarding the ``Missing'' Mercury

Mercury is not consumed in the mercury cell chloralkali plant process. Therefore, in theory, the amount of mercury that is added to the process should be equal to the amount of mercury that leaves the process in either air, water, or waste pathways. In other words, the mercury going into the system should approximately equal the mercury leaving the system, where the ``system'' is the entire plant. Historically, the industry has had a difficult time closing this mercury balance, as the amount of mercury added has exceeded the amount measured in the wastes, wastewater, products, and air leaving the plant. This difference has been referred to as the ``missing'' or unaccounted mercury. The primary basis for NRDC's estimates of fugitive mercury emissions from mercury cell chloralkali plants was the 65 tons of mercury that could not be fully accounted for by the industry at that time in their plantwide inventories (in 2000).

The EPA emissions testing and data gathering efforts discussed above did not independently resolve the unaccounted mercury issue. However, since promulgation of the 2003 Mercury Cell MACT, the level of mercury that is unaccounted for by the industry has diminished drastically. The industry reported a total of 7 tons of unaccounted for mercury in 2004, and 3 tons in 2005,\a\ with the estimate for 2006 even lower.
\a\ ``NINTH ANNUAL REPORT TO EPA for the Year 2005, May 15, 2006.'' http://www.epa.gov/region5/air/mercury/9thcl2report.pdf.

This reduction in the unaccounted mercury is likely due to increased efforts by the affected industry to inventory and track mercury in their plants, rather than to large reductions in mercury being released to the air, water, or in wastes. During our visits to mercury cell plants since promulgation of the 2003 Mercury Cell MACT, we have developed a fuller understanding of the components of a plant wide mercury balance.

One of the most significant improvements in estimating this balance has been in the estimation of the amount of mercury in the cells. Most plants now utilize a radioactive tracer method to estimate the mercury inventory in the cells. Previously, some plants did not use scientific methods to conduct an inventory of the mercury in the cells. The radioactive tracer method is accurate to around 1 percent. So, for a mercury cell plant that has about 300 tons of mercury in the cells, this error could cause the mercury balance to be inaccurate by about 3 tons. For plants that did not conduct a scientific inventory, their errors could result in significantly greater variability in the mercury inventory estimates for the mercury cells. If each of 10 plants had only factors of two errors in the accuracy of their mercury cell measurements, the effect could be 60 or more tons of unaccounted mercury for the cells alone.

Another area where significant improvement in the mercury balances has occurred is in estimating the amount of liquid mercury present in pipes and other process equipment. As plants perform maintenance on process equipment, they have measured the amount of mercury recovered and have developed accumulation factors that are now incorporated into the mercury balances procedures.

The 3 tons of unaccounted mercury reported in 2005 for the eight plants then in operation is, on average, approximately 750 pounds (lb) per plant. Significantly contributing to this number are the uncertainties in the various measurement techniques used to develop the

FOR FURTHER INFORMATION CONTACT Dr. Donna Lee Jones, Sector Policies and Programs Division, Office of Air Quality Planning and Standards (D24302), Environmental Protection Agency, Research Triangle Park, North Carolina 27711, telephone number: (919) 5415251; fax number: (919) 5413207; email address: jones.donnalee@epa.gov.