Comment submitted by Kimberly A. Warner, Ph.D., Marine Pollution Scientist and Eric A. Bilsk, Assistant General Counsel, Oceana

August 11, 2008 Via Federal eRulemaking Portal: http://www.regulations .gov National Emission Standards for Hazardous Air Pollutants for Mercury Cell Chlor-alkali Plants Docket Environmental Protection Agency EPA Docket Center (EPA/DC) Air and Radiation Docket Mail Code 2822T, 1200 Pennsylvania Ave., NW. Washington, DC 20460. Re: National Emission Standards for Hazardous Air Pollutants: Mercury Emissions from Mercury Cell Chlor-Alkali Plants, Proposed Rule, 73 Fed. Reg. 33258 (June 11, 2008); EPA-HQ-OAR-2002-0017; RIN 2060-AN99 Dear Sirs: Thank you for the opportunity to comment on the proposed amendments to the National Emissions Standards for Hazardous Air Pollutants from mercury cell chlor-alkali plants (“2008 MACT rule”). These proposed amendments are a result of the EPA’s reconsideration of the 2003 MACT rule. As you know, mercury pollution is a serious threat to public health and the environment. As you also know, the use of mercury in chlor-alkali plants is simply unnecessary. Over 95% of the chlorine produced in the United States is already produced using mercury-free technology. The few remaining plants that continue to use mercury pollute at an inordinately high rate. Switching these plants to mercury-free technology would eliminate over a ton of mercury releases, plus the potentially vast quantity of unreported releases. And the switchover would be economical as well. Each plant that converted could save millions of dollars annually in energy costs alone. Yet EPA’s proposed MACT rule fails to recognize or take advantage of this win-win situation, instead allowing a small number of plants to eke out their useful lives while emitting the maximum amount of neurotoxin into the environment. The notice of proposed rule discusses EPA studies conducted since the 2003 MACT rule. These studies make clear what has long been suspected--that cell room fugitive emissions are by far the largest source of mercury emissions from mercury-cell chlor-alkali plants (over twice the measurable stack emissions). Yet the proposed 2008 MACT rule does not set reportable limits on these emissions. That failure is plainly inconsistent with the Clean Air Act. If the EPA persists, to the economic and environmental detriment of the country and the chlor-alkali industry, in failing to determine that the Clean Air Act requires a mercury-free production process, at the minimum the EPA must set reportable limits on the significant quantity of fugitive emissions from mercury-cell rooms. BACKGROUND Notwithstanding the claims of the proposed rule, the problem of “missing” mercury and the extent to which the missing mercury is lost through fugitive emissions has not been resolved. The proposed rule was developed in response to a petition from the Natural Resources Defense Council to reconsider the 2003 MACT rule. In the intervening years since the petition for reconsideration was granted (April 8, 2004), the EPA conducted and sponsored several studies of cell-room or fugitive mercury emissions in order to determine if these emissions were responsible for the 65 tons of “missing” mercury reported by the industry in 2000, the fate of which was called an “enigma” in the 2003 MACT rule. Although the 2008 MACT rule asserts that NRDC claimed that all the missing mercury left the plants as fugitive emissions, NRDC made no such claim: We do not claim to know for certain what is happening to all of the mercury lost from the chlor-alkali process. Certainly some is lost to the ambient air, and certainly some accumulates onsite - both NRDC and the Chlorine Institute can agree on that much. The agency’s task is to solve the “enigma” of how much mercury is being lost as fugitive emissions. The preamble of the proposed rule further faults NRDC for ignoring emission studies, (e.g. those studies conducted in Maine in 1998 and in Sweden in 2001) and for ignoring the “primary conclusion” of the published 2000 EPA emission study in Georgia. In fact, the fault lies with the review of prior literature that EPA itself sponsored – that review failed to consider the Maine and Sweden studies. Further, the primary conclusion of the Southworth, et al., paper, which reports results from the EPA 2000 study in Georgia, was that fugitive emissions under the worst case estimate would equal all the mercury that was lost. From the intervening studies summarized in the 2008 MACT rule, it appears that the EPA has inadequately addressed the issue of missing mercury. The EPA, with no independent verification, has simply taken the industry at its word that all the missing mercury can be accounted for. We note that in using mass balance methods, one facility (ASHTA) has accomplished not just “eliminating” the missing mercury but actually “creating” substantial amounts of it, as well! Without presenting any verifiable data, the EPA proclaims that all the mercury that was missing in the past was likely due to bad accounting rather than actual fugitive losses to air, water, or wastes. The EPA acknowledged in the 2003 MACT rule that the industry could not “fully substantiate” its theory that the missing mercury was contained in process equipment, although the EPA seemed more than ready to believe this view. Yet in the 2008 MACT rule, the EPA refers to vague “accumulation factors” that the industry now uses to make this unaccounted for mercury vanish in the balance sheets. Further, EPA adopts this unsubstantiated use of “accumulation factors” to partly account for all the mercury in the present and past that went missing. These assumptions do not address the valid point raised by NRDC in its petition that if every plant lost the amount of mercury in its pipes yearly that industry claimed was lost (7 tons), over the lifetime of the plant, the resulting implausible mass of mercury in the pipes would roughly equal the total amount in the cells (273 tons). The EPA attributes the rest of the many tons of missing mercury to error rates in the radioactive tracer method now apparently widely used by the industry to more accurately gauge how much mercury is contained in their cells. The EPA claims that the error rate is 1% and extrapolates to demonstrate that this “error rate” factor alone could account for nearly all the 65 tons of missing mercury from the 2000 estimate. But the actual error rate for this method, as reported by the Chlorine Institute, the industry trade group, is actually closer to 0.1 %, which would hardly account for much of the mercury that was unaccounted for. The EPA should provide references if it has information to the contrary to allow for public review. DETAILED COMMENTS I. EPA SHOULD REQUIRE THE CONVERSION OF CHLOR-ALKALI PLANTS TO NON-MERCURY TECHNOLOGY The proposed rule erroneously addressed the issue of creating a sub-category of mercury cell-chlorine production by establishing a plausible criterion for determining whether to create a sub-category but then misapplying the criterion to the facts. EPA proposes that a subcategory should be created when “where differences in design and operation between types of sources in a category clearly do affect the achievability of standards” (emphasis added). The proposed rule then argues, not that non-mercury production standards are not achievable, but that mercury-cell production is not the same as non-mercury cell production, a point irrelevant to the decision. The proposed rule also erroneously argued that conversion to non-mercury cell technology is not appropriate as a “beyond the floor control” “because of the high cost impact this forced conversion would impose on the facilities in the industry.” In making this determination, the EPA considered only the cost associated with the conversion, without considering the net cost or economic benefit. In fact, as an Oceana study indicates, it is likely that any plant that converts will experience substantial benefits including an increase in energy efficiency between 25 and 37 percent per ton of chlorine produced. This increased energy efficiency could amount to savings of $98.6 million or more over five years. Furthermore, membrane cells are smaller than mercury cells, allowing plants to increase their capacity as much as 25 percent, leading to increased sales and additional energy savings on the additional capacity. The Oceana report shows, in individualized analyses for each of the remaining chlor-alkali plants that has not committed to convert, that in fact, conversion would pay for the majority of its cost in five years. Thus the proposed rule is incorrect to suppose a “high cost impact.” Accordingly, EPA should heed the record evidence that conversion is not only economically feasible, but beneficial, and EPA should mandate conversion to non-mercury technology as the beyond-the-floor control requirement. II. IN THE ALTERNATIVE, EPA SHOULD REQUIRE CELL ROOM (FUGITIVE) NUMERIC EMISSION LIMITS Even if EPA ignores the record evidence and does not require chlor-alkali plants to convert to non-mercury technology, it should still require cell room numeric emissions limits. The Clean Air Act defines hazardous air pollutant (“HAP”) emissions limits which are “not feasible to prescribe or enforce” as either (A) HAPs which cannot be emitted through a conveyance designed or constructed to emit or capture HAP or (B) when measurement methodology is not practicable due to technological or economical limitations. The proposed rule fallaciously argued that mercury cell rooms do not meet the “conveyance” requirement of the Clean Air Act and that it is neither technologically nor economically feasible to monitor emissions from cell rooms. The language of the statue and the evidence before the agency contradicts both these claims. A. The cell room should be treated as a “conveyance” for Clean Air Act purposes The plain language of the Clean Air Act and the evidence before the agency dictate that the cell room should be treated as a “conveyance.” The proposed rule argues that even though the cell room emits mercury, it is not a conveyance designed to do so. But the proposed rule fails to address the plain language of the definition which does not state that the test is whether the current design was intended to be a “conveyance” but whether HAPs “cannot” be emitted through a conveyance. Thus, the EPA’s interpretation of this Clean Air Act requirement improperly departs from the plain language of the statute. In fact, the evidence shows that cell rooms can be redesigned to serve as conveyances, as shown by the successful modification of cell rooms at existing chlor-alkali plants. This point is explained in detail in Section B. below. The proposed rule misleadingly contends that configuring cell rooms to monitor fugitive emission is economically infeasible, by presenting cost estimates ($10-20 million) only for completely replacing existing cell room buildings at plants to mimic the allegedly “perfect” configuration of the Olin GA plant. The EPA failed to consider the costs of any other cell room modifications that would facilitate fugitive emission quantification, such as reconfiguration of fans and vents to channel air towards roof vents where emissions could be monitored. These costs would be considerably lower than those for complete building replacement and should not be prohibitive. The EPA’s discussion of reconfiguration is patently irrational. B. Monitoring cell room emissions is technologically and economically feasible The proposed rule erroneously contends that even though cell room fugitive emission monitoring was demonstrated to be feasible through the EPA’s own published studies on the Olin GA plant, this plant had the best cell room configuration and other plant cell rooms lack this optimal configuration, thus rendering them unsuitable for fugitive emissions monitoring. Yet the discussion in the proposed rule demonstrates nothing of the sort. The proposed rule describes cell room configurations of two plants which have committed to convert to mercury-free technology (ERCO, WI and Pioneer, LA) and of one plant, Olin TN, which has not. Thus, of the four plants which remain committed to mercury technology, two of those (Olin GA and TN) have already demonstrated, through EPA sponsored studies, that their fugitive emissions can be quantified. Furthermore, the record for the proposed rule demonstrates that emissions monitoring was actually accomplished at two other plants – the Occidental Chemical facilities in Delaware and Alabama. Thus, while the proposed rule fails to consider the other two plants which continue to use mercury technology (ASHTA, OH, and PPG, WV) it appears that emission monitoring should be technologically feasible at them, as well, since the record shows that emissions can be measured at 4 different plants (i.e. Olin GA, Olin TN, Oxy DE, and Oxy AL), each with different cell room configurations. The proposed rule also fallaciously argues that monitoring is not technologically feasible because of (mischaracterized) error rates in monitoring data. To try to establish this point, the proposed rule summarizes the results of EPA verification studies of industry-sponsored emission monitoring at two plants. EPA reports that it found differences of up to 29% on air flow measurements and 20% on mercury concentration measurements in these emission validation studies. By only highlighting the maximum error rates, the EPA concluded that emission monitoring was not suitable for demonstrating compliance with a numeric standard. On the contrary, the average error rates were very suitable for demonstrating compliance (10% for the mercury concentration and 13% for the air flow) and could be accommodated in setting a numeric limit. Furthermore, the mercury concentration measurement was validated by comparisons to calibration standards. The error rate on this measurement could be easily reduced even further by requiring frequent calibration of the monitoring equipment -- which should be standard operating procedure. As for the air flow error rate, the air flow models upon which the industry relied in its emission monitoring studies could be improved with further validation studies with anemometers. If the models consistently underestimated air flow by a certain percentage, then this underestimation percentage could be added into the reported emission. Alternatively, the cell room vents and fans could be configured to direct air flow to the roof vent where airflow could be directly measured. Taken together, the alleged problems EPA discusses vanish upon a careful reading of the data. Accordingly, if conversion to mercury-free technology is not required, cell room emission limits should be mandated based on the Clean Air Acts standard of “the average emission limitation achieved by the best performing 5 sources.” The proposed rule states that the average mercury fugitive emissions estimated for the industry, based on data from three plants besides ASHTA, is 450 g/d. ASHTA itself is said to emit 0g/d according to the 2006 Toxics Release Inventory. Therefore the emissions limitation should be 337.5 g/d, if one assumes that the other three plants committed to mercury technology each emit 450g/d. III. THE PROPOSED RULE’S RELAXATION OF STANDARDS UNDER THE COMBINED WORK PRACTICE STANDARDS AND MERCURY MONITORING PROGRAM IS FATALLY FLAWED While the proposed rule’s preferred alternative is unlawful and far inferior to either converting to mercury-free technology or setting numeric limitations for cell room emissions, even combined work practices and monitoring can be prescribed well or poorly. It is clear from EPA’s studies spanning several years at a cost of over 1 million dollars that the EPA still has no idea what causes fugitive emissions. The proposed rule does a very poor job at prescribing work practices or monitoring to gain a better understanding. What the work practice standards and mercury monitoring will not do, based on the discussion presented in the proposed rule, is either adequately control emissions or protect worker health. Rather, the modest goal of these procedures is simply to detect whether there is an unusual level of mercury in the cell room which indicates some kind of operating problem, malfunction, spill, or leak which should be corrected. The 2008 MACT rule has • Traded informative record keeping requirements for check lists; • Relaxed triggers for corrective action in response to mercury spikes in the cell room monitoring plan; and • Failed to consider EPA sponsored data on the timeliness needed to respond to cell room mercury spikes. The 2003 MACT rule required that detailed records be kept of daily work practices for those plants which chose the work practice standards. For those plants which chose the mercury monitoring plan, the 2003 MACT required one setting of the cell room “action level” based on the 75th percentile of 30 days of cell room mercury concentration monitoring data after the compliance date. The 2008 MACT rule now requires record keeping of work practices for only a minimum of 28 days out of the year (during the setting of the semiannual action level), rather than full time, and replaces the daily record keeping for the rest of the year with weekly checklists. The 2008 MACT also raises the cell room action level trigger to the 90th percentile of baseline monitoring from the more conservative 75th percentile and allows re-setting of the action level whenever the plant chooses. It is unclear how relaxing the standards of each of the programs described in the proposed rule will help the EPA or the industry better understand the causes of, and methods of controlling, fugitive emissions since these changes will only serve to reduce the amount of information that can be collected and considered. Since control of HAPs should be the goal of such rule making rather than relieving industry of the “burden” of record keeping, the original standards of the 2003 MACT rule would be preferable. Under an adaptive management plan, the industry should adopt the strict record keeping for the work practices standards and mercury monitoring described in the 2003 MACT rule for a full year after the compliance date. Careful review of these records, corrective actions taken, and gathered emission data will help identify the causes of, and control strategies for, fugitive emissions. Only after such analysis should any relaxation of these standards be considered. Analyses of mercury monitoring data from an industry-sponsored study of two plants revealed that cell room mercury concentrations and emissions rose when the ambient temperature increased, apparently due to volatilization of mercury previously absorbed onto cell room surfaces. Thus, if the action level, based on the 75th percentile of cell room monitoring, was set when the ambient temperature was low, one would expect more frequent exceedances of this action level when the temperature warmed up. Apparently at least one of these plants complained that setting the action level at the 75th percentile resulted in “chasing alarms” rather than fixing problems. So the EPA overcompensated on at least three fronts to make the requirements for responding to mercury cell room spikes less stringent than originally proposed. First, the EPA changed the required baseline monitoring period from 30 days to 14-30 days. This change was likely instituted to decrease the “burden” of work practice standards record keeping even further (from all year to 30 days during the setting of the action level to now only 14 days at a minimum). Next, the EPA allowed the industry to reset the action level at least twice per year to account for the ambient temperature effect. This requirement alone should have corrected any ‘false alarms” due to the setting the action level when ambient temperatures were low. Yet, the EPA now proposes to allow the industry to reset this action level whenever it chooses, essentially permitting higher and higher levels of mercury escapes before any corrective action is taken. Furthermore, apparently even these measures were not enough to satisfy the industry, so EPA also allowed the action level trigger to be raised to the 90th percentile of the 14-30 days of baseline monitoring. This further reduces the frequency of any action or response and further increases the amount of mercury releases tolerated. Instead of making these changes, EPA should keep the original action level trigger based on the 75th percentile and EPA and industry should use the first full year of cell room monitoring data and record keeping for each plant to determine mathematical relationships between temperature and baseline mercury levels in order to prescribe an appropriate temperature-based action level. In addition, the EPA failed to consider the cell room monitoring data it gathered since the 2003 rule was promulgated. Considering the data, the “What work practice standards must I meet?” section should have been revised to accommodate the recent findings. The current proposed rule states that plants need not take any measures to identify and control fugitive emissions until the average concentration of mercury exceeds the action level for one full hour. From the studies conducted since the 2003 rule, it appears that, by only requiring action after the average mercury concentration exceeds the action level for a full hour, many instances of leaks will go undetected. In one study of alarm events corresponding to action level exceedences and subsequent corrective actions, 80% of the leaks identified from alarm events lasted less than one hour. Action was apparently taken when peak, rather than average, mercury concentrations were observed. Waiting for one full hour to average the mercury concentrations to determine whether or not the arbitrary action level is exceeded will result in more fugitive losses. The EPA should revise this language in the rule based on these data, and require action based on exceedences of the action level lasting a minimum of 15 minutes. The current proposed rule calls for mercury monitoring only 75% of the time (9 months out of the year). This requirement should be changed to a full year of monitoring for at least the first full year after the compliance date or until the EPA and industry can identify the causes of fugitive emissions. IV. THE EPA SHOULD RECONSIDER THE LAX MONITORING AND REPORTING REQUIREMENTS FOR. STACK EMISSIONS The 2003 MACT rule allowed industry to opt out of continuous mercury monitoring of the stack emissions and instead only monitor them for 6 hours per week. It is completely inappropriate to regulate the one remaining source of mercury emissions that the EPA deems quantifiable and then only require industry to monitor these emissions 4% of the time (assuming 24 hr plant operation) and to have no requirement for reporting these emissions. It is plainly irrational for EPA to deem the approximate 10-20% uncertainty in fugitive emissions unacceptable yet simultaneously allow over 90% uncertainty in the magnitude of the so-called measurable stack emissions. CONCLUSION The proposed rule fails to set forth a rational reconsideration of the 2003 MACT rule. In particular, the EPA’s stated basis for declining to require chlor-alkali plants to convert to mercury-free technology is simply contradicted by the evidence. Furthermore, the EPA’s stated basis for refusing to impose cell room numeric emissions limits is also contradicted by the evidence. EPA must go back to the drawing board, consider the evidence before it in good faith, and require the elimination of unnecessary mercury pollution from the few remaining antiquated chlor-alkali facilities that uneconomically persist in using nineteenth century mercury cell technology. Very truly yours, Kimberly A. Warner, Ph.D. Marine Pollution Scientist Eric A. Bilsky Assistant General Counsel