# Total Mercury and Particulate Continuous Emissions Monitoring Systems; Measurement of Low Level Particulate Emissions; Implementation at Hazardous Waste Combustors; Proposed RuleNotice of Data Availability and Request for Comments > Briefs, arguments, decisions, and more. URL: https://www.frixlaw.com/law-library/documents/fr%3A97-33740 ## Record - **Collection:** Federal Register - **Document type:** Proposed Rule - **Published:** December 30, 1997 - **Citation:** 62 FR 67788 ## Text ENVIRONMENTAL PROTECTION AGENCY 40 CFR Parts 60 and 63 [FRL-5941-4] Total Mercury and Particulate Continuous Emissions Monitoring Systems; Measurement of Low Level Particulate Emissions; Implementation at Hazardous Waste Combustors; Proposed Rule--Notice of Data Availability and Request for Comments AGENCY: Environmental Protection Agency (EPA). ACTION: Notice of data availability and request for comments. ----------------------------------------------------------------------- SUMMARY: This announcement is a notice of data availability and invitation for comment on the following reports pertaining to total mercury and particulate continuous emissions monitoring systems: DRAFT: Total Mercury CEMS Demonstration, Summary Table, dated December 1997; and DRAFT: Particulate Matter CEMS Demonstration, Volume I (with appendices), dated December 1997. EPA proposed requiring these monitors for hazardous waste combustors in the hazardous waste combustor proposed rule published on April 19, 1996. In addition, this document discusses topics for implementing particulate matter continuous emissions monitoring systems at hazardous waste combustors. Readers should note that only comments about new information discussed in this document will be considered. Issues related to the April 19, 1996, proposed rule and subsequent documents that are not directly affected by the documents or data referenced in this Notice of Data Availability are not open for further comment. DATES: Written comments on these documents and this document must be submitted by January 29, 1998. ADDRESSES: Commenters must send an original and two copies of their comments referencing Docket Number F-97-CS6A-FFFFF to: RCRA Docket Information Center, Office of Solid Waste (5305G), U.S. Environmental Protection Agency Headquarters (EPA, HQ), 401 M Street, SW, Washington, D.C. 20460. Comments may also be submitted electronically through the Internet to: [email protected]. Comments in electronic format should also be identified by the docket number F-97-CS6A-FFFFF. All electronic comments must be submitted as an ASCII file avoiding the use of special characters and any form of encryption. Commenters should not submit electronically any confidential business information (CBI). An original and two copies of the CBI must be submitted under separate cover to: RCRA CBI Document Control Officer, OSW (5305W), 401 M Street, SW, Washington D.C. 20460. For other information regarding submitting comments electronically, viewing the comments received, and supporting information, please refer to the proposed rule (61 FR 17358 (April 19, 1996)). The RCRA Information Center is located at Crystal Gateway One, 1235 Jefferson Davis Highway, First Floor, Arlington, Virginia and is open for public inspection and copying of supporting information for RCRA rules from 9:00 a.m. to 4:00 p.m. Monday through Friday, except for Federal holidays. The public must make an appointment to view docket materials by calling (703) 603-9230. The public may copy a maximum of 100 pages from any regulatory document at no cost. Additional copies cost $0.15 per page. FOR FURTHER INFORMATION CONTACT: For general information, call the RCRA Hotline at 1-800-424-9346 or TDD 1-800-553-7672 (hearing impaired) including directions on how to access some of the documents and data referred to in this notice electronically. Callers within the Washington Metropolitan Area must dial 703-412-9810 or TDD 703-412-3323 (hearing impaired). The RCRA Hotline is open Monday-Friday, 9:00 a.m. to 6:00 p.m., Eastern Time. The documents referred to in this notice are available over the Internet. The documents can be accessed by typing the following universal resource locator (URL): http://www.epa.gov/epaoswer/hazwaste/combust/cems This URL provides a home page through which all electronically available documents can be downloaded. The Technology Transfer Network (TTN) also provides a link to this page. CEMS information is available on TTN at the following URL: http://ttnwww.rtpnc.epa.gov/html/emtic/cem.htm The home page contains links to the files that are available electronically. The files are in an executable, compressed format to facilitate [[Page 67789]] downloading. Once extracted, each compressed file may result in more than one decompressed file. The reports are in Adobe Acrobat, PDF format. The reader should note that figures, diagrams, and appendices may not be available electronically or may only be available in other formats. For other information regarding the information contained in Sections I, II, IV, and V of this notice, contact Mr. Scott Postma, (5302W), Office of Solid Waste, 401 M Street, SW, Washington, D.C. 20460, phone (703) 308-6120, E-MAIL: [email protected]. For information regarding Section III of this notice, contact Mr. H. Scott Rauenzahn (5302W), Office of Solid Waste, 401 M Street, SW, Washington, D.C. 20460, phone (703) 308-8477, e-mail: [email protected]. SUPPLEMENTARY INFORMATION: On April 19, 1996, EPA proposed revised standards (herein referred to as ``the proposed rule'') for hazardous waste combustors (HWCs, i.e., incinerators and cement and lightweight aggregate kilns that burn hazardous waste). See 61 FR 17358. Comments received from the public in response to the proposed rule are found in RCRA docket F-96-RCSP-FFFFF. A previous notice of data availability (NODA), published on March 21, 1997, gave the public the opportunity to review the Agency's approach to demonstrating CEMS for HWCs. This previous NODA is herein referred to as ``the first CEMS NODA'' or ``CEMS NODA 1.'' See 62 FR 13776. Comments received from the public in response to the first CEMS NODA are found in RCRA docket F-97-CS3A-FFFFF. Readers should note that a separate docket was established for this document. See the Addresses section above for more information. Table of Contents I. Introduction and Background II. The Hg CEMS Demonstration Tests III. The PM CEMS Demonstration Tests A. PM performance Specification (PS) 11 Levels 1. Revised Specification Levels for the Correlation Coefficient, Confidence Interval, and Tolerance Interval 2. Data Availability 3. Data Quality Objectives: New Procedures to Appendix F of 40 CFR Part 60 B. Manual Method Accuracy 1. Modification of the Filter Recovery Process 2. Improved Sample Collection 3. Elimination of Contamination 4. Improved Sample Analysis 5. Comparison of M5 and M5i Method Precision 6. Paired Data C. Transferability of These Demonstration Test Results to Other HWC Sources IV. PM CEMS: Implementation and Compliance A. PM CEMS Compliance Schedule B. PM CEMS Operating Parameter Limit 1. Introduction 2. Data Excluded from Calculating the PM CEMS Operating Parameter Limit 3. Determining the Normality of the Data 4. Averaging Periods for the PM CEMS Operating Parameter Limit 5. Options for Calculating the PM CEMS Operating Parameter Limit a. Using rank statistics to calculate the PM CEMS-based operating parameter limit at one, fixed averaging period b. The traditional standard setting approach 6. Consideration of a Variance Procedure to Project a Higher PM CEMS Operating Parameter Limit a. HAPs for which PM control is necessary to ensure compliance b. Projecting a higher PM CEMS operating parameter limit considering the ratio of the standard to the measured level of a HAP c. Ensuring that the higher projected PM CEMS operating parameter limit does not exceed the MACT PM standard d. Establishing revised operating parameter limits for the PM control device corresponding to the higher projected PM CEMS operating parameter limit e. Implementing the variance 7. EPA's PM CEMS Testing to Identify a CEMS-Based Emission Level Achievable by MACT-Controlled Sources C. RCA Test Frequency D. Extrapolating PM CEMS Calibration Data 1. Extrapolating Light-Scattering PM CEMS Calibration Data 2. Extrapolation of Beta-gage Calibration Data E. Need to Calibrate to Twice the Emission Standard F. Allowing PM CEMS to be Used In-lieu of Method 5 Tests G. Waivers from the PM CEMS Requirements 1. Waiver of PM and Hg CEMS Requirements for Small On-Site Incinerators 2. PM CEMS Waiver for Sources with Short Life-Spans 3. Other Sources V. Other Issues Concerning CEMS and Test Methods for HWCs A. Performance Specifications for Optional CEMS B. Stack Sampling Test Methods PART 60--STANDARDS OF PERFORMANCE FOR NEW STATIONARY SOURCES I. Introduction and Background In the proposed rule, EPA proposed that continuous emissions monitoring systems (CEMS) be used for compliance with the HWC total mercury (Hg) and particulate matter (PM) standards. See 61 FR at 17426 and 17435. To require CEMS for compliance the Agency, among other things, must determine that the CEMS are commercially available and have been demonstrated to meet certain performance specifications. To make these determinations, the Agency tested various Hg and PM CEMS being marketed in the U.S. and Europe. The first CEMS NODA described the approach EPA is using to demonstrate the feasibility of PM and Hg CEMS and requested comment on certain technical issues arising from this program. This testing is now complete. Today the Agency is providing notice of an opportunity to comment on the following documents resulting from these CEMS demonstration test program: DRAFT: Total Mercury CEMS Demonstration, Summary Table, dated December 1997; and DRAFT: Particulate Matter CEMS Demonstration, Volume I (with appendices), dated December 1997. The remainder of this notice describes important information bearing upon how the reports' results relate to EPA's approach to demonstrating Hg and PM CEMS and how PM CEMS could be used for compliance. Many of these issues were raised by commenters in response to CEMS NODA 1 and the proposed rule. The reader is referred to the referenced documents for specific information regarding the Hg and PM CEMS demonstration test program and the comments cited here. II. The Hg CEMS Demonstration Tests EPA seeks comment on the document DRAFT: Total Mercury CEMS Demonstration, Summary Table, dated December 1997, provided in the above referenced docket for this NODA. This table summarizes results from the Hg CEMS demonstration tests EPA conducted. In summary, the Agency found certain aspects of the testing program revealed substantial problems regarding the measurement of the Hg CEMS accuracy and precision. EPA found it difficult to dynamically spike known amounts of mercury (in the elemental and ionic form) and obtain manual method and Hg CEMS measurements that agree at the test source. As a result, the Agency now believes it has not sufficiently demonstrated the viability of Hg CEMS as a compliance tool at all hazardous waste combustors and should not require their use. Nonetheless, EPA still believes Hg CEMS can and will work at some sources but does not have sufficient confidence that all HWC conditions are conducive to proper operation of the Hg CEMS tested. Facilities should have the choice of using Hg CEMS if desired so long as the [[Page 67790]] permitting agency approves on a site-specific basis the Hg CEMS and its site-specific performance specifications. See a related issue in section V.A. of this NODA regarding the implementation of optional CEMS. III. The PM CEMS Demonstration Tests This section describes the report DRAFT: Particulate Matter CEMS Demonstration, Volume I (with appendices), dated December 1997, contained in the docket identified in the Addresses section, above. EPA issued previous notices asking vendors to participate in this program (see 61 FR 7232, February 27, 1996) and to allow the public to comment on the Agency's approach to demonstrating these monitors (see 62 FR 13775, March 21, 1997). Since this could represent the first time EPA requires PM CEMS for compliance at stationary sources, the technical discussion contained in this section is expected to have general applicability beyond sources that burn hazardous waste. In particular, EPA invites comment from all parties concerning the following documents attached to this notice: Method 5I for the determination of low level particulate emissions; Performance Specification 11 for PM CEMS; and Quality Assurance Requirements for PM CEMS. A. PM Performance Specification (PS) 11 Levels In CEMS NODA 1, the Agency stated it intended to loosen the proposed PM CEMS Performance Specification (PS) 11 to reflect what was achievable by the monitors during this demonstration test. The test was designed to be a reasonable worst case investigation of what performance (relative to the proposed PS11) the monitors could achieve. Many comments received in response to CEMS NODA 1 stated that the proposed performance specifications were not sufficiently stringent and opposed loosening the specification levels. Concurrent with the Agency's invitation in CEMS NODA 1 to comment on our approach to demonstrate PM CEMS, EPA determined that much of the variability in the calibration curves resulted from inaccuracies in performing the manual method, Method 5 (M5). Since the fundamental approach in PS11 involves correlating manual method results to PM CEMS outputs, the PS11 statistical results reflected this variability in the manual method. Consequently, EPA undertook a systematic effort to identify and remove this error from the manual method measurement process. Manual method improvements were developed and observed, and performance specification results for the PM CEMS improved as a result. (See a related discussion in section III.B, below, regarding these improvements to the manual method.) 1. Revised Specification Levels for the Correlation Coefficient, Confidence Interval, and Tolerance Interval As a result of comments on CEMS NODA 1, EPA decided to accept a slightly modified version of the more stringent International Standards Organization (ISO) specification 10155 for PM CEMS. Four of the five PM CEMS tested during the PM CEMS demonstration tests were able to meet all three performance specifications (i.e., those for the correlation coefficient (r), confidence interval (CI), and Tolerance Interval (TI)) at all three of the emissions levels discussed in the May 2, 1997, NODA as alternatives to the proposed emissions standards (69 mg/dscm for cement kilns, 50 mg/dscm for light-weight aggregate kilns, and 34 mg/ dscm for hazardous waste incinerators.) See 62 FR 24212 for a discussion of those alternative emissions standards. One technology, an extractive light-scattering technology, did not meet all the performance specification levels at all of the alternative standards.1 Since EPA must show that at least one commercially available PM CEMS can meet the proposed performance specification, the fact that 4 of the 5 monitors were able to meet these performance levels under reasonable worst-case test conditions adequately shows that the modified specification levels are achievable. The revised performance specification levels are presented in Table 1, below. --------------------------------------------------------------------------- \1\ This technology did meet 10 of the 14 specification comparisons. Table 1: Revised Performance Specifications for PM CEMS ------------------------------------------------------------------------ Confidence Tolerance Correlation coefficient interval interval ------------------------------------------------------------------------ 0.90 10% 25% ------------------------------------------------------------------------ As previously stated, these performance levels are nearly identical to the ISO specification for PM CEMS. The only major difference between these and the ISO specification levels is the correlation coefficient, which is 0.95 in ISO 10155 and 0.90 in the modified PS11. This is acceptable since the correlation coefficient does not directly relate to measurement error while the confidence and tolerance intervals do.2 The revised PS11 also requires that a minimum of 15 runs be used for the calibration while the ISO specification requires only 9 runs. The ISO specification is also vague regarding the PM concentration ranges required for a calibration. The revised PS11 stipulates three ranges: 0 to 30, 30 to 60, and 60 to 100% of the facility's range of PM emissions. --------------------------------------------------------------------------- \2\ The correlation coefficient is defined as the ratio Sxy divided by the square root of the quantity Sxx times Syy. Sxy is a measurement of error as x relates to y, while Sxx and Syy are a reflection of the range of the data set. As a result, the correlation coefficient is not as useful a tool to evaluate measurement error as the correlation and tolerance intervals. This is particularly true as the correlation coefficient approaches 1. --------------------------------------------------------------------------- 2. Data Availability EPA had proposed that PM CEMS be used at all times that hazardous waste is in the unit. See 61 FR at 17441. Commenters to the proposed rule did not view this favorably. They said this proposal is equivalent to a 100% data availability requirement for PM CEMS. Commenters stated that this requirement is not achievable since all mechanical devices fail at some point, often without warning. They said a data availability requirement in the 85, 90, or 95% range would be more acceptable. Commenters suggested that when the PM CEMS were not available, the PM-related operating parameter limits EPA proposed should be used in place of the PM CEMS. EPA largely agrees with this comment. The PM CEMS demonstration tests show that a 100% data availability requirement is not achievable for all PM CEMS in all instances. The Agency also agrees that when PM CEMS are not operating, it is reasonable to provide for some back-up compliance system in lieu of requiring sources to either stop burning hazardous waste or have a back-up PM CEMS available. The PM APCD operating parameters proposed in the event there is no PM CEMS requirement are a good starting point for identifying such a back-up system. See the discussion later in this section for more information on this issue. Based on these demonstration tests and the comments received, EPA concludes that a 95% data availability requirement is achievable for most PM CEMS. Therefore, EPA intends that PM CEMS be used 95% of the time for compliance. However, there are useful technologies that cannot meet this 95% data availability requirement. This data availability requirement should be relaxed in certain instances. For instance, beta- gages did not meet the 95% data availability requirement during the PM CEMS tests. EPA believes beta-gages, with their relatively superior [[Page 67791]] performance and ability to measure PM emissions at truly wet stacks (i.e., those with entrained water droplets in the stack gas), will be useful at some sources for compliance. Therefore, EPA will consider an 85% data availability requirement for beta-gage technology PM CEMS.3 The Agency anticipates other case-by-case determinations will be made in the future as more is learned about the performance, benefits, and data availability limitations of other PM CEMS technologies. --------------------------------------------------------------------------- \3\ One of the beta gage PM CEMS experienced a 74% data availability during the PM CEMS demonstration test program. Much of the additional downtime was because no U.S.-based technicians were fully trained to service this instrument during this program and parts and personnel had to be brought to the U.S. Once EPA requires a new technology, such as PM CEMS, the market for that new technology is expected to mature in the US similar to how one exists overseas. As a result, data availability will be better than what was experienced here. --------------------------------------------------------------------------- Finally, EPA believes that increasing the amount of PM data available will enable sources to improve their understanding and better define the relationship between operating parameters and emission levels. EPA is aware of two relevant examples which are described here. The first one is an ongoing cooperative effort with industry, regulatory agencies, and the local public. This effort is focused on a venturi scrubber-controlled lime kiln at a pulp and paper plant where testing is being conducted to evaluate the feasibility of a predictive emission monitoring system (PEMS). Following preliminary measurements and an experimentally designed test matrix, 595 Method 5 runs were performed over a wide variety of process and scrubber operating conditions, and PM emission levels. A correlation coefficient above 0.9 was obtained in correlating PM emission levels with 54 operating parameters. In comparison, use of PM CEMS represents a powerful tool for accumulating much data at a cost that is far less than performing hundreds of Method 5 runs. As a result, PM CEMS allow for a cost- effective way to implement a PM PEMS model than making hundreds of Method 5 measurements. The second example directly relates to the first. The Electric Power Research Institute (EPRI) has already produced a means to characterize and correlate PM emissions with operating conditions at coal-fired utility boilers. PM emissions from utility boilers are similar to HWCs in that their emissions are affected by a complexity of variations from a number of fuel and feed characteristics, combustor operations, and electrostatic precipitator (ESP) operations. As in the lime kiln case mentioned above, use of PM CEMS represents a more powerful tool for accumulating and correlating vast amounts of PM emission data with PM-related operating parameters at a cost that is far less than performing a large number of Method 5 runs. These two examples, therefore, lead EPA to believe that a PM CEMS requirement will allow HWC facilities to better define what PM APCD operating parameter limits correspond to a given PM emissions concentration. As a result, the Agency encourages HWC facilities to use PM CEMS data to better define what operating parameters correspond to compliance with a facility's PM CEMS limit. The site-specific limit is discussed further in section IV.B. of this notice. 3. Data Quality Objectives: New Procedure 2 to Appendix F of 40 CFR Part 60 EPA intends to expand the ISO specification to include certain data quality objectives. For example, PM CEMS routinely and automatically check and correct their raw outputs to compensate for phenomena such as ``fogging'' of the optics and drift of the measurement signal. A large and sudden auto correction is indicative of the need to perform maintenance on the PM CEMS. To address this concern, EPA intends to include certain data quality objectives such as: The (PM CEMS internal) calibration drift not exceed 8% during any drift check; the (PM CEMS internal) calibration drift not exceed more than 4% per day for five consecutive days; and the automated (PM CEMS internal) calibration drift adjustment not exceed 2% for five consecutive days. These data quality criteria would appear in a new Procedure 2 of Appendix F to 40 CFR part 60. Just as Procedure 1 of Appendix F deals with data quality objectives for gaseous CEMS for other CAA rules (CEMS such as NOX and SOX), Procedure 2 would address data quality objectives for PM CEMS. Procedure 2 would also include the following data quality objectives: treatment of ``flagged data;'' PM CEMS automatic zero and calibration span requirements; conduct of the Absolute Calibration Audit and the quality of the standards used for these audits; sample volume audit requirements for extractive systems; relative calibration audit (RCA) requirements; the treatment of audit failures; how manual method paired data outliers (see the CEMS NODA 1, 62 FR at 13780) are handled; definition of ``out-of-control'' situations; and how facilities are to respond to these ``out-of- control'' situations. See Procedure 2 (which appears at the end of this notice) and the draft final PM CEMS Demonstration Test Report for more information regarding these requirements and Procedure 2. B. Manual Method Accuracy One outgrowth of these PM CEMS demonstration tests is that EPA has made significant improvements in making Method 5 particulate measurements. As previously mentioned, the calibration process for PM CEMS involves correlating PM CEMS outputs to manual method results. High variability in the manual method results will negatively affect the PS11 calibration statistics. Therefore, one important way to improve PS11 statistics is to improve the way manual method measurements are made. These improvements involve the use of a new Method 5I (M5i) for low level PM emissions. M5i consists of the following improvements: improved sample collection; elimination of possible contamination; and improved sample analysis. Each will be discussed in the following paragraphs. M5i will be instrumental in correlating PM CEMS outputs to manual method results. EPA also expects this new method will be preferred in all cases where low level (i.e., below 45 mg/dscm [ cents0.02 gr/dscf]) measurements are required. In practice, this means that M5i is expected to become the standard method for most HWCs and many other MACT sources. EPA expects many of the improvements in M5i can and will be implemented whenever Method 5 (collectively, including Method 5, 5A, ..., 5H) is used to make particulate measurements. M5i is almost identical to the traditional M5. Differences are discussed below. We also present a comparison of the precision of M5i (which fully implements these improvements) and the traditional M5 (which largely does not use these improvements) and discuss the need for and handling of paired M5 data. 1. Modification of the Filter Recovery Process One way M5i differs from the traditional M5 is through the use of a light-weight and integrated filter and filter assembly that can be tared and weighed together. This improves M5 by eliminating the filter recovery step. The filter recovery step can be a significant source of measurement error at some sources. In some cases as the filter dries, the filter adheres itself to the filter assembly. Recovery of the filter then involves scraping the filter off the filter assembly leaving some of the filter (and sample) on the filter assembly or otherwise losing it to the environment. In other cases, the filter recovery [[Page 67792]] process can lead to the loss of sample to the environment as light- weight particles are lost to the air during handling. It can also lead to contamination of the sample in cases where fugitive dust from the environment lands on the filter during the recovery step. Simplifying the recovery step through the use of a light-weight, integrated filter and assembly addresses these concerns and thereby improves the reliability of making PM measurements. One consequence of this improvement, though, is that the filter used M5i is smaller than that used in the traditional M5 (47 mm compared to approximately 90 to 100 mm in the traditional M5). This smaller filter can plug at higher emissions levels. For this reason, this aspect of M5i may not be implementable at sources with emissions above 45 mg/dscm (that is, total train catches exceeding 50 mg). 2. Improved Sample Collection Another important improvement to M5 is to the sample collection process itself. These improvements include: ensuring that the nozzle is 90 deg. to the direction of flow at each traverse location; and using Pesticide Grade (i.e., low residue) acetone for probe rinse. Each is discussed in the following paragraphs. These improvements to the sample collection process may also be implemented over time into other versions of Method 5. Test crews routinely check the ``level'' of the probe only once during sampling--prior to or at the beginning of the sampling process itself. As the traverse progresses, the probe can become ``unlevel,'' i.e., it is no longer at a right (90 deg.) angle to the direction flow in the stack gas. As the angle of the probe departs from 90 deg., inconsistent amounts of sample are collected and thereby causes error in M5 measurements. This can be corrected by applying a level to the sample probe and checking the level continuously throughout the traverse. Ensuring that the probe angle is constant and level throughout the traverse eliminates this potential source of measurement error. Finally, residue contained in the acetone used for the probe rinse is another source of sampling error. Acetone is used for the probe rinse since it is a solvent that evaporates readily at room temperatures and thereby allows rapid weighing of the specimen following sampling. The standard M5 procedures require that acetone residue blank levels be determined and that reagent-grade acetone in glass bottles with no more than 0.001% residue be used for probe rinses. Acetone comes in many grades, including reagent grade, depending on how the purchaser intends to use the acetone. Some grades of acetone contain higher levels of residue. This residue remains after the acetone evaporates and contaminates the probe rinse, making the ``catch'' during the probe rinse greater than what it really is. Acetone blanks above 0.001% are not allowed by M5, so the acetone itself must have a concentration of residue no more than this requirement for the blanks. M5 also requires that the acetone be stored in glass containers because acetone from metal containers generally have a high residue blank level. Test crews routinely use reagent-grade acetone purchased in small, glass containers since large quantity purchases create a fire safety and storage issue. Though ordered with the intent of meeting specifications, acetone suppliers often store bulk, reagent grade acetone in metal containers and transfer this acetone to glass containers only to ship the small quantities sold. This means that the residue concentrations found in reagent grade acetone are often higher than what is allowed by M5. The unallowable amount of residue from high blank levels would have a negative effect on the accuracy and precision of M5 results. This can be avoided by requiring low-residue, Pesticide grade acetone. 3. Elimination of Contamination In a general sense, eliminating contamination in the filter handling processes will eliminate potential sources of error. Contamination can be avoided by: using a portable desiccator for use in transporting and holding the filters to and on the stack; using glass plugs on the filter assemblies to keep them ``pure'' prior to and after sampling; covering the desiccant with a 0.1 micron screen to eliminate potential external contamination of filter housing during transport; and handling the filter assemblies with powder free latex gloves. As previously discussed, contamination in the M5 process will make measured PM levels appear to be higher than what they truly are. Each of these steps to eliminate contamination of the sample will ensure that fugitive particulate from the environment does not contaminate, thereby inadvertently causing a positive bias to the measured PM levels. 4. Improved Sample Analysis Finally, improved sample analysis will help eliminate error in the Method 5 measurement process. Specific steps to improve M5 sample analysis include: Elimination of all sources of static charge (such as those on the operator, beakers, liners, and balance); use of light weight Teflon beaker liners for gravimetric analysis of the probe rinse; maintaining the laboratory area at a humidity level of 30% or less; and putting a covered desiccant container in the balance weighing chamber. Each is discussed in the following paragraphs. High, varying levels of static charge typically produce variations in repeated weighing of susceptible materials such as glass filter holder assemblies and Teflon beakers (probe rinse catches). The need for maintaining a relatively dry atmosphere in the analytical room further exaggerates the negative effect of static charge on the weighing process. To control and minimize the consequences of static charge, EPA found it was necessary to preclude any and all aspects of static electricity. This entailed using: (1) A static-free mat on the floor area under the desiccator and balance; (2) a small charge- neutralizer in the desiccator; (3) another small charge-neutralizer in the weighing chamber of the balance; and (4) a static dissipator aerosol spray to prevent static buildup on the Teflon beakers. EPA found it was not possible to consistently reproduce the same weight results (that is, within 0.5 mg) until all four measures were done. The particulate, filter assembly, and filter are often hydrophilic in nature, i.e., they tend to adsorb water from the air. The amount of water these materials adsorb depends on the amount of water in the air. The moisture content of air is often quantified in terms of the relative atmospheric humidity, or what percentage the actual water concentration of the air is relative to the saturation concentration. The higher the relative humidity of the air, the more water is adsorbed. The converse is also true. As a result, if the relative humidity in the analysis room is high, the amount of water adsorbed onto the particulate, filter, and assembly becomes variable and it becomes increasingly difficult to obtain a stable measurement. Ensuring that the relative humidity in the analysis room remains at a constant, low level will ensure that the amount of water adsorbed by these materials remains relatively small and constant. EPA found that maintaining the relative humidity of the room to below 30% will control this source of error.4 --------------------------------------------------------------------------- \4\ As humidity levels decrease, static charge tends to increase. The elimination of static charge, previously discussed, will aid at eliminating this problem. --------------------------------------------------------------------------- To further control and minimize the adverse affects of humidity on reproducing results in the weighing [[Page 67793]] process, a small covered desiccant container was placed in the balance's weighing chamber. This ensures that the humidity level in the weighing chamber is consistent with the humidity level in the desiccator. The desiccant in the weighing chamber dries fugitive air entering the chamber from the room, preventing the adsorption of room air humidity on the materials being weighed. 5. Comparison of M5 and M5i Method Precision M5i has been validated against Method 5.5 It is also important to quantify the improvements to M5 just discussed. This can be done by comparing the precision of both methods at each of the three proposed PM standards: 34, 50, and 69 mg/dscm for hazardous waste burning incinerators, LWAKs, and cement kilns, respectively. Precision at the standards is important since measurements at the standard deal with compliance determinations at facilities. The best estimate of the standard deviation is presented to represent this precision. --------------------------------------------------------------------------- \5\ Only the filter, extraction, and weighing steps were tested. --------------------------------------------------------------------------- For M5i, the results presented are the best estimate of the standard deviation at each of the three proposed emissions standards. These results are calculated directly from the data obtained during the PM CEMS demonstration tests. The relative standard deviation (e.g., the best estimate of the standard deviation at an emissions concentration, divided by that emissions concentration) for M5i is in all cases less than 5%. Historical data from Method 5 was derived from PM concentrations ranging from 80 to 255 mg/dscm. This data indicates that the relative standard deviation for M5 is constant at 10%. Therefore, if one were to multiple the emissions standard by 10%, one can derive the best estimate of the standard deviation at the three proposed emissions standards. Table 2, below, illustrates this comparison and shows that M5i is an improvement to M5. Table 2: Comparison of Method 5 and Method 5i Standard Deviations ------------------------------------------------------------------------ Best estimate of the standard deviation (mg/ Proposed PM emissions standard (mg/dscm) at 7% dscm) oxygen ------------------------- Method 5 Method 5i ------------------------------------------------------------------------ 34 3.4 1.67 50 5.0 2.24 69 6.9 2.85 ------------------------------------------------------------------------ 6. Paired Data Throughout most of the PM CEMS demonstration tests, EPA used two simultaneous Method 5i sampling trains. These two simultaneous trains are called ``paired trains'' and the paired train data are called ``paired data.'' The average of the paired data from the two trains was considered the method result. If the ``paired data'' differed by more than 30% from the method result, EPA eliminated the method result from the calculation of the calibration. EPA's experience is that, despite the efforts just discussed to control method variability, intangibles which are unknown or unquantifiable can cause variability in M5 (and M5i) measurements. Since it is important that highly accurate M5 measurements be obtained for calibrating PM CEMS, these intangibles must be identified in some way and data affected by these intangibles must be eliminated from the PM CEMS calibration. The fact that two simultaneously run M5 (or M5i) measurements do not agree is ample evidence that something in the sample collection and analysis process was not consistent. Comments received during the comment period for CEMS NODA 1 stated that this process would not be allowed if a facility were doing a calibration for compliance. To address this concern, EPA has incorporated this outlier procedure in the new Procedure 2 and M5i. Having this procedure in the regulations will allow facilities to exclude this type of erroneous data from their PM CEMS calibrations. Please note that this procedure applies only if paired data are obtained. Single measurements obtained at different times are not paired data. Single runs cannot be eliminated by comparing these results to other single measurements. EPA strongly encourages facilities to use paired data during their calibrations. Beyond the ability to eliminate paired data outliers from the PM CEMS calibration, using the average of two runs as the method result has a moderating affect on the calibration statistics EPA calculated and used to base the PS11 revised in today's notice. This moderating effect improves the PS11 criteria relative to what they would be if paired data were not used. Some facilities may find it difficult to obtain a suitable calibration using only single M5 measurements. However, while we encourage using paired data, we are not requiring paired data for PM CEMS calibrations. This choice can be left to the facilities to determine what makes the most economic and technical sense at their site. C. Transferability of These Demonstration Test Results to Other HWC Sources EPA believes this demonstration test program adequately shows that PM CEMS will meet PS11 at most hazardous waste incinerators, cement kilns, and light-weight aggregate kilns. These tests were conducted at a reasonable worst-case facility for performance relative to the proposed performance specifications6. Therefore, a PM CEMS should pass the performance measures described in the revised draft PS11 at most HWC sources. The paragraphs below discuss specific aspects of PM CEMS and their applicability to each HWC source category. --------------------------------------------------------------------------- \6\ For more on EPA's rationale that this is a reasonable worst- case test, see CEMS NODA 1 and the PM CEMS Demonstration Test report sited here. --------------------------------------------------------------------------- For cement kilns, in-situ light scattering PM CEMS are operationally very similar to continuous opacity monitors (COMs), a technology employed at these sources for many years. Light-scattering PM CEMS differ from COMs only in the way they obtain and interpret the light from the source. As shown in the LaFarge tests,7 though, an informed decision is required to determine what type of in- situ light-scattering PM CEMS is best suited for these sources. One PM CEMS used at LaFarge was built with a heated air purge system to blow cement kiln dust away from and out of the optics of the monitor while the other was not. The monitor with the heated air purge performed very well over the course of the tests, though improvements to the Method 5 measurements and a routine cleaning of the optics could have improved performance. The PM CEMS without the heated air purge suffered operational difficulty. In addition, PM from cement kilns is mostly process dust (i.e., raw material). As such, its physical properties are not significantly affected by changes in waste or fuel feeds. Accordingly, in-situ light-scattering PM CEMS will pass performance specifications at cement kilns if an informed decision is made to [[Page 67794]] purchase a monitor that is designed to address the dusty environment at these facilities. --------------------------------------------------------------------------- \7\ See section 2.6.7 of the PM CEMS Demonstration Test Report. --------------------------------------------------------------------------- EPA believes that LWAKs are very similar to cement kilns relative to the applicability of PM CEMS, and therefore the conclusions drawn in the preceding paragraph also apply to LWAKs. For incinerators, there are certain unique situations which must be discussed: incinerators with truly wet stacks; incinerators with waste heat boilers; and mobile incinerators. As was the case with cement kilns, HWC incinerators with truly wet stacks (i.e., those with entrained water droplets in the stack gas) need to make an informed choice regarding what PM CEMS technology they elect to use. In-situ light-scattering PM CEMS are likely to have operational difficulty since the water droplets entrained in the stack gas will be mistaken for particulate. This is a readily accepted source of error and means that in-situ light-scattering PM CEMS are not a practical choice for these sources. Beta-gage and certain other light- scattering PM CEMS, however, are designed with extractive reheat systems which heat up the extracted gas to above the water condensation temperature. Incinerator groups are currently working to test these types of systems to gain first-hand experience and data regarding the use of PM CEMS at facilities with truly wet stacks. EPA encourages these tests since they will result in valuable data which can be communicated to personnel at incinerators with truly wet stacks to assist their PM CEMS purchasing decisions. Incinerators equipped with waste heat boilers (WHBs) downstream of the combustion chamber(s) also require special consideration. Like boilers, these incinerators blow soot periodically to clean the boiler tubes. PM emissions will increase and the physical properties (pertinent to PS11) of the PM may change during periods of soot blowing. To help address the impact of soot-blowing, sources would be required to include soot-blowing episodes during a minimum of three calibration runs. This will ensure that calibration captures the higher emissions that can occur during soot-blowing, thus minimizing the need to extrapolate the calibration curve beyond measured values. In addition, including soot-blowing during calibration runs will enable the source to determine whether any change in the physical properties of the PM during soot-blowing has adversely affected the calibration (i.e., as evidenced by an inability to meet PS 11 when the soot-blowing runs are included). EPA requests comment on this approach to address the special problems that soot-blowing may cause. In particular, EPA seeks the following information: --How many incinerators are currently equipped with WHBs? Are sources likely to remove WHBs to facilitate compliance with the MACT standards (e.g., D/F)? --The normal frequency and duration of soot-blowing. Under what conditions does the frequency and duration of soot-blowing change? How often does this change(s) occur? --How do PM emissions for runs that include episodes of soot blowing compare to runs without soot blowing? --How does the effect of the APCS, waste and fuel types, and other relevant factors impact changes to the PM concentrations and physical properties when one compares PM during soot-blowing and PM at other times. The reader should note that EPA intends to promulgate MACT standards for hazardous waste boilers as part of Phase II of the HWC rulemaking. EPA intends to address the applicability of PM CEMS to boilers then. If because of unforeseen reasons EPA provides a PM CEMS waiver for incinerators with WHBs, EPA would readdress the applicability of PM CEMS to hazardous waste incinerators with waste heat boilers in Phase II. Finally, another class of hazardous waste incinerators are used at Superfund sites during the clean-up process. These mobile incinerators have small, limited waste processing capacity and are often trucked to the site as needed. EPA is concerned that the variability of the feed to mobile incinerators is beyond what was experienced at the DuPont facility. As such, a unique calibration might be required for every clean-up site, which is unnecessarily burdensome. Given the PM CEMS implementation schedule discussed in section IV.A., below, implementing PM CEMS at these incinerators may not be feasible, and EPA is considering whether to waive the PM CEMS requirement for Superfund mobile incinerators. If the PM CEMS requirement is waived for certain facilities, the other, traditional operating parameters discussed in this NODA would be used instead to document compliance. IV. PM CEMS: Implementation and Compliance 8 --------------------------------------------------------------------------- \8\ The reader should note that HWCs are currently regulated under RCRA. Sources with a different regulatory history are likely to have a different compliance regime than the one described here. One should not assume that the compliance and implementation scheme described here will necessarily be applied to sources with a different regulatory history. --------------------------------------------------------------------------- A. PM CEMS Compliance Schedule Many comments received in response to the proposed rule stated that facility personnel are not familiar with the operation and maintenance characteristics of PM CEMS, or how to control their operating conditions to ensure compliance using PM CEMS. For this and reasons explained in section IV.B., EPA plans to allow a 12-month phase-in period before PM CEMS would be used as a compliance parameter. This section describes this compliance schedule. Prior to the date PM CEMS would be used for compliance (i.e., during the 12-month phase-in period), limits on key PM-related and other key operating parameters (e.g., metals feedrate) would be used to ensure compliance with the MACT standards for PM, SVM, and LVM. This one year phase-in period has four key milestones: The Compliance Date; the performance Test Date; the Certification of Compliance (CoC) date; and the Certification of PM CEMS performance date. By the Compliance Date,9 facilities would determine, using engineering judgment, the operating parameter limits necessary to ensure compliance with the standards. These initial operating parameter limits would be specified in a Precertification of Compliance (Pre-CoC) that would be submitted to the permitting authority by the Compliance Date. By the Test Date, which is nominally no later than six months after the Compliance Date, facilities would have to conduct a performance test to document compliance with the MACT emissions standards and identify operating parameter limits based on levels achieved during the test. Results of the performance tests and these revised operating parameter limits would be submitted to the permitting authority in a Certification of Compliance (CoC) nominally no more than nine months after the Compliance Date. The operating parameter limits in the CoC would be used as surrogate compliance measures to ensure that the efficiency of [[Page 67795]] the PM control device was maintained at performance test levels until the one-year anniversary date of the Compliance Date. Beginning at that time, facilities would start using the PM CEMS and cease using the operating parameters as their primary operating parameter for PM control. --------------------------------------------------------------------------- \9\ The Clean Air Act states that the Compliance Date can be no more than three years after the effective date of the rule (i.e., date of publication in the Federal Register), unless a source obtains an (up to) one-year time extension of the Compliance Date. --------------------------------------------------------------------------- During this phase-in year, there are important PM CEMS-related activities being performed. For instance, the PM CEMS, like all other equipment necessary for compliance with the MACT standards, must be installed by the Compliance Date. Like all other tests for the rule, the PM CEMS calibrations and initial certifications (see section 8.3 of PS11 and section 4 of Procedure 2) must also be performed by the Test Date. As discussed in the PM CEMS Demonstration Test Report, the mathematical characteristics of a light-scattering PM CEMS calibration curve can be difficult to determine. For this reason, a second calibration would be required within 9 months of the Compliance Date if a light-scattering PM CEMS is used. After this second calibration is performed the source would compare the two calibrations separately to determine which mathematical model best represents the data. This information (i.e., the analysis of which mathematical model is best suited for the calibration at this source and the calibration comprised of all valid calibration data obtained) would be included in a Certification of PM CEMS Performance (CoP) submitted within 12 months after the Compliance Date. (A CoP for beta-gage CEMS would also be submitted at this time, but would certify performance based on a single calibration.) This CoP would also include (for all types of CEMS) the analysis of CEMS data to identify an achievable CEMS-based PM operating parameter limit. See section IV.B., below, for more information regarding the PM CEMS operating parameter limit. On the one-year anniversary of the Compliance Date, facilities would also cease using the PM control device operating parameter limits (such as pressure drop across a fabric filter or total power input to an ESP) and start using the PM CEMS operating parameter limit as their primary compliance parameter for the PM control device. A source using a light-scattering PM CEMS would be required to perform a third calibration of the PM CEMS within 12 months of the Compliance Date. The third calibration would verify that the mathematical model selected by comparing the first two calibrations is correct. If not, to the approach must be modified based on the new data. If the model needed to be revised, the source would be required to recalculate the PM CEMS operating parameter limit considering the 12 months of data following the Compliance Date. If the model did not need to be revised, the source could elect to recalculate the PM CEMS operating parameter limit considering the full 12 months of CEMS recordings. (We request comment on whether all sources required to perform a third calibration should be required to recalculate the PM CEMS operating limit even if the calibration curve model did not need to be revised.) The results of the third calibration,10 reassessment of the calibration model, and recalculation of the PM CEMS operating parameter limit would be submitted in a second CoP. This second CoP would be provided to the permitting authority within 15 months of the Compliance Date. Note that this second CoP would not be required if a source uses a beta-gage type PM CEMS that needs only one calibration. A source using a PM CEMS that requires only one calibration (i.e., a beta-gage) would have the option, however, of submitting a second CoP if it wants to update the PM CEMS operating parameter limit based on a full year of data. --------------------------------------------------------------------------- \10\ This would involve the verification that the mathematical model used for the calibration is correct, a recalculation of the ``master'' calibration comprised of all three calibration curves, and a revised site-specific PM CEMS operating parameter limit (this time, using the first 12 months of data). --------------------------------------------------------------------------- Table 3 summarizes how PM CEMS would be implemented for compliance. Following this implementation schedule, facilities would be required to document compliance with the MACT PM standard during periodic performance testing. As discussed in section IV.F. below, a source would have the option of using the PM CEMS or the manual method for this determination. In addition, sources would be required to recalculate their PM CEMS operating parameter limit based on the previous year of CEMS data recorded when the source operated within the operating parameter limits established during the new performance test. This recalculation of the PM CEMS operating parameter limit is necessary since the new performance test is likely to result in numerically different PM-related APCD operating parameters than resulted from the previous test.11 Facilities would submit the revised operating parameter limits, including the revised PM CEMS operating parameter limit, in a CoC describing the results of the new performance test. --------------------------------------------------------------------------- \11\ If the PM concentration and operating parameter limits resulting from the subsequent performance test are more stringent than those from the previous test, the facility would have the option of not recalculating their PM CEMS operating parameter limit and continue to operate under the older, more stringent limit. [[Page 67796]] Table 3: PM CEMS Compliance Schedule ---------------------------------------------------------------------------------------------------------------- By this date These PM CEMS activities would be performed ---------------------------------------------------------------------------------------------------------------- Compliance date............................. PM CEMS installed Precertification of Compliance (Pre-CoC) submitted that establishes PM APCD (and other) operating parameter limits to ensure compliance with the SVM and LVM (and possibly D/F) standards based on engineering judgment. CD + 6 months (the ``Test Date'')........... PM CEMS calibration tests performed during MACT performance test. Method 5i used to demonstrate compliance with the manual method- based PM standard. CD + 9 months............................... Certification of Compliance (CoC) submitted that establishes PM APCD (and other) operating parameter limits to ensure compliance with the SVM and LVM (and possibly D/F) standards based on what levels were determined to correspond to compliance with the standard during the Performance Test. (These limits supersede those identified in the Pre-CoC.) A second calibration of light-scattering PM CEMS is performed. CD + 12 months.............................. Source identifies calibration curve, calculates the PM CEMS operating parameter limit (through 9 months), recommends alternative PM control device operating parameters and their numerical limits, and submits a Certification of PM CEMS Performance (CoP). Source ceases using PM control device operating parameters as the primary mode of compliance for PM and starts using the PM CEMS. The operating parameters defined in the CoC are used for compliance only when the PM CEMS is unavailable. Source reports initial calibration (composite of all calibrations through 9 months), PM CEMS-based limit, and revised operating parameter limits (for use during CEMS malfunctions) to permitting authority. A third calibration of light-scattering PM CEMS is performed. CD + 15 months.............................. Sources using light-scattering PM CEMS revise the calibration curve if necessary based on the third calibration, recalculate the PM CEMS operating parameter limit (through 12-months), update the PM control device operating parameter limits for use during CEMS malfunctions, and submit a second CoP documenting this information. The source starts using the revised operating parameters reported in the CoP during periods when the PM CEMS is unavailable unless those alternatives have been disapproved by the permitting authority. ---------------------------------------------------------------------------------------------------------------- B. PM CEMS Operating Parameter Limit EPA proposed using site-specific limits on key operating parameters of the PM control device (e.g., pressure drop across a fabric filter) to ensure that the device maintained its collection efficiency at performance test levels. These limits, in combination with other operating parameter limits (e.g., metals feedrate controls) would ensure compliance with the semivolatile metal (SVM) and low volatile metal (LVM) MACT standards. See 61 FR 17376 and 17430 (April 19, 1996). These operating parameter limits on the PM control device would also ensure compliance with the MACT PM standard 12, and possibly the MACT dioxin and furan (D/F) and mercury (Hg) standards if the source uses activated carbon injection to control these HAPs. 13 The availability of PM CEMS allows the Agency to improve upon this approach through the use of PM CEMS as the sole PM control device operating parameter 14. PM CEMS are a more sensitive and accurate operating parameter than the conventional PM-related operating parameters now used. --------------------------------------------------------------------------- \12\ The Agency has proposed a MACT PM standard as a surrogate to control emissions of non-enumerated metals HAPs (i.e., metal HAPs other than those for which specific standards have been proposed-- Hg, SVM, and LVM). Those non-enumerated HAPs are Sb, Co, Mn, Ni, and Se. \13\ The Agency proposed that the site-specific PM limit be a compliance parameter for the D/F standard irrespective of whether activated carbon injection was used as a control device. This requirement was grounded upon EPA's initial view that a significant amount of D/F (and other heavy organic compounds) are adsorbed onto particulate. As a result, PM needed to be controlled to ensure continuous compliance with the D/F standard. The Agency is now considering comments that significant D/F may not be adsorbed onto PM. Cement and lightweight kiln PM, in particular, is generally process dust (i.e., processed raw material). This process dust has little affinity for adsorbing D/F. However, EPA's ultimate decision on whether to limit PM on a site-specific basic does not depend on whether there is a need to control PM at all HWCs for D/F control. A site-specific PM limit is still needed to ensure compliance with the SVM and LVM standards at all HWCs. Sources that use activated carbon injection, howver, would be expected to have a significant amount of D/F on the PM. \14\ Note that the MACT standard for PM would continue to be a manual methods-based standard. See subsection IV.F., below, for options facilities could have to use PM CEMS for direct compliance with this PM standard. --------------------------------------------------------------------------- This section describes how PM CEMS would be implemented as an operating parameter for the SVM, LVM, PM, and possibly D/F and Hg standards. The reader should note that the proposed MACT standard for PM is and will continue to be a manual methods-based standard. The reader is referred to section IV.F., below, for options a facility could choose to use PM CEMS as a direct indicator of compliance with the MACT PM standard. 1. Introduction The PM CEMS operating parameter limit would be determined using the PM CEMS data obtained during normal operations from the Compliance Date to the time when the calculation of the operating parameter limit is performed. Although the PM CEMS would be used as an operating parameter limit here, an approach to establishing this limit could be very similar to how EPA establishes national standards from CEMS data. The municipal waste combustor rule published on February 11, 1991, has an example of how this is done 15. --------------------------------------------------------------------------- \15\The methods used for establishing CEMS standards in the February 1991 MWC rule are described in Appendices A and B of EPA document number EPA-450/3-91-004, dated December 1990. This document can also be found in the Air Docket, located in the Mall area of EPA Headquarters, 401 M Street, SW, Washington, DC 20460. It is part of docket number A-89-08-V-B-3. --------------------------------------------------------------------------- EPA notes that even though the PM CEMS operating parameter limit and the manual methods-based PM standard are both in units of particulate concentration, it is likely that the PM CEMS operating parameter limits will have a different numerical value than the manual methods-based MACT PM standards. This is because the MACT PM standards would be based on manual methods testing with no fixed averaging period. PM CEMS operating parameter limits would have both a fixed averaging period and a calculated [[Page 67797]] numerical limit. As discussed in section 4, below, the numerical value of a limit or standard is a function of the averaging period. Since it is likely that the PM MACT standard and the PM CEMS operating parameter would have different averaging periods, one would expect the numerical value of the PM CEMS operating parameter limit that indicates compliance with the MACT standards would differ from the numerical value of the MACT PM standard. 2. Data Excluded From Calculating the PM CEMS Operating Parameter Limit Before calculating the PM CEMS operating parameter limit, the PM CEMS data set must be screened to remove PM CEMS data recorded when the PM CEMS was not available or the source was out of compliance with the operating parameter limits established during the CoC. First, the facility must remove from the data set all PM CEMS data accumulated while the PM CEMS was not available or not performing acceptably as defined by the regulations. Examples of the data not included in the calculation of the PM CEMS operating parameter limit include data obtained when the PM CEMS was ``out-of-control'' as defined in Procedure 2 and PS11, periods when the PM CEMS was not analyzing stack gas (as would happen during calibrations, maintenance, etc.), and periods when the facility was not in operation. Next, the facility would further screen the data to exclude times when the facility was not operated in accordance with the operating parameter limits resulting from the performance test and reported in the CoC 16. Note that the CoC operating parameter limits would supersede the Pre-CoC operating parameter limits for this screening purpose. Although the Pre-CoC operating parameter limits may be less stringent than the CoC limits and were valid limits prior to submitting the CoC, the CoC limits are based on performance testing and as such show what operating parameter levels reflect compliance with the standards. The facility must also remove any data collected during periods of PM APCS upset irrespective of whether the operating parameter limits were exceeded. 17 --------------------------------------------------------------------------- \16\ For simplicity, EPA proposes to exclude data from all periods in which the facility operated outside of the operating envelope defined in the CoC irrespective of whether the parameter in question affects PM control. Defining what operating parameters are or are not related to PM control would force another layer of complexity in this step. \17\ Episodes of high PM emissions caused by periodic, routine maintenance cycles (e.g., ESP rapping; soot-blowing for waste heat boiler equipped incinerators, etc.) would not be considered upset conditions. We request information on how to objectively distinguish between high PM emissions attributable to PM control device upset conditions versus normal emissions variability. --------------------------------------------------------------------------- 3. Determining the Normality of the Data To calculate the PM CEMS operating parameter limit, the CEMS recordings 18 must be averaged over an appropriate averaging period. (See the discussion in the following section.) Accordingly, sources would be required to identify the mathematical model that best fits the screened CEMS data for purposes of averaging the data. For example, a log-or exponential fit may better represent a ``normal'' fit relative to an arithmetic model. To identify which mathematical model represents the best fit, facilities would calculate the Shapiro-Wilk Normality test statistic (W) at the 95% confidence level using the data obtained from the PM CEMS 19. The mathematical model with a Shapiro-Wilk test statistic closest to one (1) would be the model used for averaging at the facility. This mathematical model would be used for all PM CEMS emissions averaging at the facility. --------------------------------------------------------------------------- \18\ The light-scattering CEMS provide instantaneous data, recorded every minute as one-minute block averages. Beta-gage CEMS have sampling periods longer than 1 minute. \19\ Note that batch CEMS, such as beta-gages, may have sampling periods longer than 1 minute. In this case, the test statistic would be performed using the batch results. --------------------------------------------------------------------------- 4. Averaging Periods for the PM CEMS Operating Parameter Limit Fundamental to any emissions control parameter is the way averaging affects an emissions standard or limit. At a fixed numerical value, a standard or limit is more stringent as the averaging period decreases and less stringent as the averaging period increases because of emissions variability. In the proposed rule, EPA said that an appropriate averaging period for PM CEMS would be the length of time it takes to make three Method 5 runs. The Agency still believes this is an appropriate point of departure for the averaging period for the PM CEMS operating parameter limit. We proposed a 2 hour averaging period for PM CEMS, reasoning that it would take 40 minutes to accumulate enough PM sample to meet Method 5 requirements for sample ``catch.'' See 61 FR at 17379. Commenters argued, however, that although it takes 40 minutes to accumulate enough sample, test crews routinely sample for one hour. In addition, comments received in response to CEMS NODA 1 said the sampling time for a Method 5 run can vary from 1 to 8 hours. Basing the averaging period for the PM CEMS operating parameter limit on the length of time it takes to perform three Method 5 runs would result in an averaging period in the range of 3 to 24 hours 20. This is still being evaluated. --------------------------------------------------------------------------- \20\ Three to 24 hours is within the range of CEMS averaging periods EPA typically promulgates. From a broader perspective, averaging periods vary from regulation to regulation depending on the analysis of issues pertaining to the technical, policy, and regulatory history of each particular situation. Therefore, other source categories may or may not have the same averaging period as the one established for PM CEMS at HWC, depending on the outcome of this analysis of issues. --------------------------------------------------------------------------- 5. Options for Calculating the PM CEMS Operating Parameter Limit As discussed above, the stringency of the standard is a function of two variables--its numerical limit and the averaging period. Equally stringent standards would have a higher numerical limit at shorter averaging period and a lower numerical limit at a longer averaging period. Thus, to calculate the PM CEMS operating parameter limit, one of these two variables must be held constant--either the numerical limit or the averaging period. This section discusses two options for calculating an achievable PM CEMS operating parameter limit by defining the averaging period. EPA investigated ways to define the numerical limit and allow facilities to calculate the averaging period associated with that numerical limit, but found these alternatives often resulted in trial-and-error type calculations and might result in all facilities having different limits and averaging periods. We believe these alternatives are too labor intensive and confusing--both for facilities and the enforcement authority--and rejected this approach. Based on comments and further analysis, the Agency will prescribe one methodology in the final rule. a. Using Rank Statistics to Calculate the PM CEMS-based Operating Parameter Limit at One, Fixed Averaging Period. Under this approach, the Agency would establish an averaging period common to all sources and each source would calculate its PM CEMS operating parameter limit using rank statistics. (As discussed above, EPA is considering selecting an averaging period from within the range of three to 24 hours.) The averaging period would be the same for all facilities but the numerical value of the PM CEMS operating parameter limit would differ from facility-to-facility based on the historical data obtained at each facility. [[Page 67798]] Using the rank statistics option to calculate the limit would involve the following steps. First, the facility would take the screened PM CEMS data (i.e., after non-compliance data has been removed) and calculate rolling averages sequentially from the Compliance Date 21 using the best-fit mathematical model and the averaging period EPA promulgates. The facility would then take the resulting rolling averages and sort them in order from lowest to highest. The facility's PM CEMS operating parameter limit would be the 95th percentile highest PM CEMS rolling average, by rank, experienced during the period the PM CEMS data was accumulated. The 95th percentile is proposed here because it is the percentile level EPA historically uses for these types of calculations. EPA could promulgate some other percentile level, the 90th or 99th for example, if another percentile level is achievable and better represents good PM control. --------------------------------------------------------------------------- \21\ For simplicity, we believe it is best for facilities to ignore periods when the CEMS recorded data which was screened out and calculate the rolling averages as if the remaining data occurred sequentially. EPA specifically requests comment on this approach. --------------------------------------------------------------------------- This rank statistics option is easier to implement, relative to the other options. It also would result in a PM CEMS operating parameter limit that is in the range of actual emissions experienced by the facility (i.e., as opposed to statistically projected emissions) and demonstrated by the facility to be achievable over time. Since the limit for all sources would be based on an averaging period that would be fixed in the rulemaking, the limit would be easier to enforce as well. b. The Traditional Standard Setting Approach. Another approach the Agency is considering to determine the PM CEMS operating parameter limit is to use the way EPA has established CEMS-based standards in the past. This approach involves calculating the average and standard deviation of the data set and projecting an emissions level associated with the data. As discussed in the MWC rule, EPA calculated ``continuous compliance levels'' for each source using the equation, below. where: y=x+K*5 y = the continuous compliance level; ``x-bar'' = the sample average k = a constant associated with the averaging period and one exceedance per year; and s = the sample standard deviation. This option has some benefits and weaknesses. As discussed, it reflects a procedure EPA has previously used to establish CEMS-based standards. It would also result in every facility having the same averaging period and thus making it easier to track and enforce. However, more complicated statistics are involved. EPA also compares emissions from more than one facility when it uses this approach to set standards and would be unable to oversee the application of this approach on a site-specific basis. As a result, this approach may be unworkable as a way to establish a PM CEMS operating parameter limit. 6. Consideration of a Variance Procedure to Project a Higher PM CEMS Operating Parameter Limit As discussed previously in today's notice, the PM CEMS operating parameter limit would be based on CEMS recordings during the nine months after the Compliance Date during those periods of time that the source was operating within the operating parameter limits established during the performance test (i.e., the operating parameter limits established in the Certification of Compliance (CoC)). Comments received in response to the proposed rule questioned the need to establish PM-related operating parameters based on the performance test if: (1) PM emissions measured using manual methods during the performance test were well below the PM MACT standard; and (2) emissions of HAPs (i.e., SVM, LVM, and possibly Hg, and D/F) for which PM would be used as an operating parameter limit were well below their MACT standards. Commenters were concerned that, although their sources may readily achieve the MACT PM standard, it may be difficult 22 or expensive to ensure that performance test PM levels are representative of the full range of levels achieved during operations. The same situation could occur with the PM CEMS operating parameter limit just discussed. Infrequent exceedances of the PM CEMS operating parameter limit might or might not be an indication that the SVM, LVM, Hg, D/F, or PM MACT emission standards have been exceeded. Accordingly, commenters recommended that the rule allow sources to project higher PM-related operating parameters based on how much performance test emissions for these HAPs were below their MACT standards. --------------------------------------------------------------------------- \22\ For example, some PM control devices are so over-designed that it is difficult to force them to operate at elevated PM levels for the duration of a performance test. --------------------------------------------------------------------------- EPA agrees in theory that establishing the PM CEMS operating parameter limit considering performance test operations (i.e., historical CEMS data when the source operated within the CoC operating parameter limits) could result in an overly conservative operating parameter for PM control at sources with low PM and low HAPs that require PM control to ensure compliance. To address the concerns expressed in the comments received on the proposed rule, the Agency is considering a variance procedure to establish an higher projected PM CEMS operating parameter limit. The variance procedure would allow facilities with very low concentrations of PM and HAPs requiring PM control for compliance to increase their PM CEMS operating parameter limit (derived from operations within the CoC operating parameter limits). The factor used to increase the PM CEMS operating parameter limit could be defined as the ratio of the MACT standards for which PM control is required to assure compliance, to the performance test levels of those HAPs. To ensure that the source is still be in compliance with the MACT PM standard, the same ratio would be calculated for the PM standard to the unadjusted PM CEMS operating parameter limit. This approach is based on the principle that, at a facility which has experienced no changes in facility operations, the ratio of emissions of HAPs which require PM control to ensure compliance to the PM concentration in the stack is either constant or decreases as PM increases. In addition, revised (i.e., less stringent) traditional operating parameter limits for the PM control devices corresponding to the higher projected PM CEMS operating parameter limit could be established based on historical operating data at levels near the higher projected PM CEMS operating parameter limit. For illustration, an example follows. Assume that a hazardous waste incinerator has low metals in the feed and uses a HEPA filter for PM control. Further assume that: This incinerator's measured metals emissions during the performance test were 10 g/dscm and 7 g/dscm for SVM and LVM, respectively; that the PM concentration measured during the performance test was 5 g/ dscm; for simplicity that PM control is not required to assure compliance with the D/F and Hg standards; the unadjusted PM CEMS operating parameter limit is 15 mg/dscm; and from the HWC NODA published on May 2, 1997, that the promulgated standards are 100 g/dscm, 55 g/dscm, and 34 mg/dscm for SVM, LVM, and PM, respectively. The ratio of the standard to the measured levels are 10 and 7.8 for SVM and LVM. For PM, [[Page 67799]] the ratio is 6.8. The unadjusted PM CEMS operating parameter limit would be increased by a factor of 6.8 since the ratio calculated for PM has the lowest numerical value. a. HAPs for which PM control is necessary to ensure compliance. PM would be used as an operating parameter limit for semivolatile metals (SVM), low volatility metals (LVM), and if activated carbon is used, dioxin and furan (D/F) and mercury (Hg). See 61 FR 17422 and 17430 (April 19, 1996). Although the Agency is reconsidering whether PM is an appropriate operating parameter to ensure compliance with the D/F standard in some cases,23 PM would be an appropriate operating parameter if activated carbon injection were used to control D/F or mercury. This is because D/F and mercury adsorb onto the activated carbon, and as PM emissions increase, emissions of activated carbon with adsorbed D/F and mercury increase. --------------------------------------------------------------------------- \23\ The Agency is considering comments that significant D/F may not be adsorbed on emitted PM in all cases. Cement and lightweight kiln PM, in particular, is generally process dust (i.e., raw material) that has little affinity for absorbing D/F. --------------------------------------------------------------------------- b. Projecting a higher PM CEMS operating parameter limit considering the ratio of the standard to the measured level of a HAP. The variance would be based on the principle that, as PM emissions increase, the ratio of emissions of each HAP for which PM is an operating parameter limit (i.e., SVM, LVM, and possibly D/F and Hg) to PM emissions either is constant or decreases. Thus, the PM CEMS operating parameter limit derived from operations within the CoC operating parameter limits could be increased without exceeding the MACT standards for those HAPs by a factor considering the ratio of the standard for each of those HAPs to the performance test level of each HAP. LVM are generally not volatilized in the combustion chamber and thus are evenly distributed over all sizes of particulates. Thus, as PM emissions increase, the ratio of LVM emissions to PM emissions will be constant. SVM are generally volatilized during combustion and condense preferentially on small particulates prior to (or in) the PM control device. Thus for many PM control devices, as PM emissions increase, the ratio of larger particulates to smaller particulates increases, and the ratio of SVM emissions to total particulate emissions decreases (i.e., because the larger particulates have a lower concentration of SVM). For emission control trains where PM particle size may remain constant with an increase in PM emissions, the ratio of SVM emissions to PM emissions would remain constant. D/F and Hg are adsorbed onto the surface of carbonaceous particulates (e.g., activated carbon). Smaller particulates have a larger surface area per mass of particulate than larger particulates, and thus D/F and Hg concentrations would be higher for smaller particulates. Thus, similar to SVM, as PM emissions increase and the ratio of larger particulates to smaller particulates increases, the ratio of D/F and Hg emissions to total particulate emissions should decrease (i.e., because the larger particulates have a lower concentration of D/F and Hg). The PM CEMS operating parameter limit derived from performance test operations (i.e., calculated from historical CEMS data when the source operated within the regulations) could be increased without exceeding (theoretically) the MACT standards for SVM, LVM, and possibly D/F and Hg by the ratio of the standard for each of those HAPs to the performance test level of each HAP. It would be reasonably conservative, however, to project the higher PM operating parameter limit by the ratio of some fraction of the standard for those HAPs to the performance test level of each HAP. This fraction of the standard would need to allow for adequate flexibility for sources with low PM and HAPs for which PM control is required while ensuring that the standards are being met continuously. A specific percentage of the standard within the range of reasonable values--50% to 100%-- could be selected and would be appropriate given the uncertainty of projecting a PM operating limit that is a primary compliance measure for several MACT emissions standards. EPA believes choosing 75% of the standard as the basis for calculating the ratio is a reasonable balance of these issues. The percentage that would be appropriate is a point of interest for the Agency. Given that the PM CEMS operating parameter limit is a compliance measure for SVM, LVM, and possibly D/F and Hg the allowable higher projected PM CEMS operating parameter limit would be the lowest of the values projected for each of these standards. For example, if the projected PM CEMS operating parameter limit based on the ratio of 75% of the SVM standard to the SVM performance test level was lower than the PM CEMS operating parameter limit projections for LVM (and possibly D/F and Hg), then the SVM-projected PM CEMS operating parameter limit would be used to ensure that the SVM standard was not exceeded at the higher projected PM operating parameter limit. The Agency is concerned, however, about increasing the PM CEMS operating parameter limit itself by the ratios discussed above. This is because the limit would be established at the upper end of the range of actual CEMS readings, or perhaps at levels that statistically exceed what would be expected. See above discussion of options for calculating the PM CEMS operating parameter limit. It may be more appropriate to project the higher limit using the following options. Under option 1, the ratio determined above would be applied to the average PM emissions over time determined by the PM CEMS instead of applying the ratio directly to the unadjusted PM CEMS limit itself. The product of the ratio and the average PM emissions would then be subtracted from the average emissions to determine the correction to the PM CEMS operating parameter limit. This correction would then be added to the PM CEMS operating parameter limit to determine the revised PM CEMS operating parameter limit. Using the example described above and assuming average PM CEMS emissions are 2 mg/dscm, the lowest ratio (6.8 for PM) would be multiplied by the average PM emissions (2x6.8 = 13.6) and the average emissions would be subtracted from this product (13.6-2). This difference (11.6) would be added to the PM CEMS operating parameter limit to obtain the revised PM CEMS operating parameter limit. Under option 2, the PM CEMS recordings during the performance test would be analyzed to calculate a PM CEMS operating parameter limit and that limit would be increased by the factor defined by the ratio discussed above (e.g., 6.8 in the first example). This approach would ensure that infrequent high PM episodes that occurred over months of CEMS operations would not be driving a PM CEMS limit that was then projected further upward using the factors discussed above (unless those high PM episodes actually occurred during the performance test). Given the truncated emissions database (i.e., the performance test) for calculating the higher projected limit under this option, however, the limit may in fact be lower than the limit normally calculated from the full CEMS emissions database (i.e., without attempted to project a higher limit). In this case, the limit which is numerically higher would be used. The Agency requests information on which approach would be more appropriate for projecting a higher PM CEMS operating parameter limit. c. Ensuring that the higher projected PM CEMS operating parameter limit [[Page 67800]] does not exceed the MACT PM standard. The PM CEMS operating parameter limit would also be used to ensure compliance with the MACT PM standard. We reiterate that the PM CEMS operating parameter limit is not a measure of the emissions standard--the emissions standard is defined in the rule as being measured by using manual methods--it is instead an operating parameter limit used to ensure compliance with the applicable standards. As discussed in section 1, above, it is likely that due to several factors the PM CEMS operating parameter would have a different numerical value that the MACT PM standard. One reason for different numerical values is the use of different techniques (i.e., one is manual methods based while the other is CEMS- based) to determine a PM emissions value. The use of a manual method test to determine a value is only a limited-time (e.g., 3 to 24 hours every five years) measure of emissions, whereas a CEMS is a continuous measure of emissions (e.g., 1 minute readings all the time). Although the manual method will likely be a measure of ``high- end'' PM emissions during performance testing, it may not account for all potential variability during normal operations. The use of a CEMS to monitor PM emissions is a way to continuously measure the variability of (both low and high) PM emissions, inherent in any engineered system. Additionally, different values may be a result of the different averaging periods stated for manual methods-based PM standard and the PM CEMS operating parameter limit. See section 4, above, for a discussion of the interrelationship between a numerical value of a limit or standard and the averaging period. Having a PM CEMS operating parameter limit with a different, possibly higher, numerical limit is permissible and does not negate the value of the PM CEMS operating parameter, provided there is reasonable correlation between the operating parameter and the MACT PM standard. This section explains how EPA would ensure that the numerically larger, revised PM CEMS operating parameter limit would not violate the national PM standard. Ensuring that the higher projected PM CEMS operating parameter limit does not exceed the MACT PM standard is complicated by the fact that the PM operating parameter limit would be CEMS-based while the MACT PM standard would be manual method-based. Nonetheless, compliance with the MACT PM standard can be ensured by limiting the increase in the PM CEMS operating parameter limit (i.e., the projected PM CEMS operating parameter limit divided by the limit prior to projection) to the ratio of the MACT PM standard to the performance test PM level on a manual method basis. Given that projections rather than measured values would be used to ensure compliance with a standard, it may be prudent to limit the increase in the projected PM CEMS operating parameter limit to the ratio of 75% of the MACT PM standard to the performance test PM level. A conservative factor of 75% is within the range of reasonable values the Agency could have selected--50% to 100%. Regarding the specific percentage EPA chooses, the reader is referred to the previous discussion regarding the percentage EPA chooses for the HAP standards that require PM control to ensure compliance. Using the example from above and making the same assumptions, the ratios would be calculated using 75 /dscm, 41/dscm, and 26 mg/dscm in the numerator for SVM, LVM, and PM. These values are 75% of the standards for incinerators EPA discussed in the May 2, 1997, HWC NODA. The resulting rations would be 7.5, 5.9, and 5.1. Since the ratio calculated for PM is the lowest, ratio used to determine the revised PM CEMS operating parameter limit would be 5.1. d. Establishing Revised Operating Parameter Limits for the PM Control Device Corresponding to the Higher Projected PM CEMS Operating Parameter Limit. Ideally, PM control device operating parameter limits (e.g, pressure drop across a fabric filter) should be established to ensure compliance with the higher projected PM CEMS operating parameter limit for compliance purposes while the CEMS is malfunctioning. Absent these revised (i.e., less stringent) operating parameter limits, the source would be required to: (1) Comply with the more stringent operating parameter limits established during the performance test that correspond to the original PM CEMS operating parameter limit; or (2) ensure that a back-up CEMS is always available. The Agency is considering an approach to establish revised operating parameter limits for the PM control device corresponding to operations at the higher projected PM CEMS operating parameter limit. Under this approach, the source would analyze the historical operating parameter values during those periods of time that PM emissions were close to the higher projected PM CEMS operating parameter limit. Issues that must be addressed, include: (1) What range of PM CEMS operating parameter limit values should be considered to develop the database for PM control device operating parameter values; and (2) how should the database be analyzed to identify appropriate limits. It may be appropriate to establish the revised PM control device operating parameter limits based on the 90th percentile of values that occur when PM levels are within 75% of the higher projected PM CEMS operating parameter limit. This would help ensure that a significant data set was available for evaluation and that the limits were not based on the most lenient values recorded. This is important because the higher projected PM CEMS operating parameter limit is likely to be well beyond the calibration curve. 24 --------------------------------------------------------------------------- \24\ A source would be allowed to operate infrequently at levels approaching a higher, projected PM CEMS operating parameter limit that is beyond the calibration curve. If, however, a source operates for prolonged periods at levels above the calibration curve, it must perform Method 5 tests at those higher concentrations and include those higher PM levels in the POM CEMS calibration. See discussion on extrapolating PM CEMS calibration data elsewhere in today's notice. --------------------------------------------------------------------------- Based on further analysis, the Agency may consider other approaches to define an appropriate data set of PM control device operating limits and identify appropriate limits (e.g., considering a different percentage of the historical data and/or basing the limit on a different percentile of data). Alternatively, the Agency may conclude that these approaches to revise the performance test-based operating parameter limits would be too complicated or difficult for regulatory officials to oversee, or that it would be difficult to confirm compliance with the standards. In this event, sources would be required to continue to comply with the PM control device operating parameter limits established during the performance test when the CEMS malfunctions even though the PM CEMS operating parameter limit has been projected upward under procedures discussed above. e. Implementing the Variance. Sources requesting the variance to project a higher PM CEMS operating parameter limit would include the request with the Certification of PM CEMS Performance (CoP) that would be submitted within 12 months after the Compliance Date. The variance request must include documentation of the analyses described above to identify the higher projected PM CEMS operating parameter limit and the revised, PM control device operating parameters associated with the higher projected PM [[Page 67801]] operating parameter limit. Sources would be allowed to comply with the higher projected PM CEMS operating parameter limit immediately upon submitting the CoP. Regulatory officials would have three months to review the variance request, however, and to notify the source of intent to disapprove the higher projected PM CEMS operating parameter limit (or the associated revised PM control device operating parameter limits for use during CEMS malfunctions). In such cases, the regulatory officials would provide the basis of their initial decision and provide the source with an opportunity to present, within 30 calendar days, additional information before final action on the variance. 7. EPA's PM CEMS Testing Program to Identify a CEMS-Based Emission Level Achievable by MACT-Controlled Sources The Agency is undertaking an additional PM CEMS testing program to identify CEMS-based emission levels that are achievable by hazardous waste combustors (i.e., hazardous waste burning incinerators, cement kilns, and lightweight aggregate kilns) using MACT control. The testing is scheduled to begin in December 1997 and results should be analyzed by December 1998. The Agency is working with representatives of the regulated community to identify one source in each of the three source categories that is using MACT control and that would be likely to define the most achievable level (i.e., considering average PM emissions and emissions variability) for MACT-controlled sources. Although these test results will not be used as part of the final rule, the data will be valuable to permitting authorities and the regulated community as a PM CEMS emission benchmark that is achievable using MACT controls. Permitting authorities could use the data to identify sources that appear to have established an anomalously high CEMS-based PM operating parameter limit, and as a framework within which to review Certifications of Performance in a cost-effective manner. Likewise, sources wanting to ensure that their facility is operating in a manner representative of MACT control could use this information to see if their CEMS-based PM operating parameter limits are below the levels that MACT sources show are achievable using MACT control. C. RCA Test Frequency In the proposed rule, EPA said that facilities would be required to perform relative calibration audits (RCAs) on their PM CEMS every 18 months. This testing interval would be relaxed to 30 months for small on-site incinerators. These time intervals coincided with the proposed Performance Test intervals. If these tests can be performed at the same time as the performance tests, cost savings can be realized by the facility relative to what the costs would be if the tests were not conducted at the same time. As a result of the analysis of comments on the performance test frequency, the Agency is considering requiring all facilities to conduct comprehensive performance tests every five (5) years. Therefore, we prefer that RCA tests be performed every five years. One of the goals of the PM CEMS Demonstration Test program was to quantitatively define what RCA frequency is appropriate for PM CEMS. Unfortunately, variability in the manual method masks any error that can be identified as being caused by drift in the PM CEMS over time. Therefore, we are unable to use the PM CEMS data from the demonstration tests to extrapolate to an appropriate re-test frequency \25\. --------------------------------------------------------------------------- \25\ In fact as method accuracy improved, the PM CEMS calibration statistics got better over time. Extrapolating this data would lead to erroneous conclusion that no retesting is ever needed, since the PM CEMS calibration keeps getting better. --------------------------------------------------------------------------- Lacking these long-term data, it is important to look at what is done in other countries to qualitatively determine this RCA test frequency. The United Kingdom (UK) requires that retesting be conducted at least every year, and in Germany testing is required every 3 to 5 years. The RCA test frequency could therefore reasonably be between one and five years. The UK, though, heavily relies on manual methods testing--so much so that they believe using gas bottles is cost prohibitive for gaseous (e.g., NOx and SOx) CEMS testing and rely on manual methods testing instead. EPA is inclined to believe that the German's longer retest frequency is more consistent with our regulatory framework. D. Extrapolating PM CEMS Calibration Data One-minute or batch PM CEMS readings during the course of operations are likely to occasionally exceed the highest M5i calibration point during the course of PM CEMS use. This is because the manual method results used to derive the calibration are (nominally) one hour block averages of emissions over the sampling period while the PM CEMS readings are averages of emissions on the order of minutes. See section 4, above, regarding the interrelationship the numerical limits of a standard or limit and its associated averaging period. In addition, emissions variability within the sampling period of M5 is not likely to represent the full range of emissions variability over all periods of PM CEMS operation. Therefore, a system is needed to allow the extrapolation of data beyond the calibration curve. The revised calibration and implementation scheme described in today's notice (i.e., multiple calibrations (for light-scattering PM CEMS) over the full range of emissions at the facility) will result in a calibration from which some reasonable and limited extrapolation is reasonable. Therefore, the Agency proposes to allow the calibration curve to be used for measurements up to 25% more than the maximum M5 \26\ measurement observed during the calibration. (This will be referred to as the ``125% point.'') Beyond this point (125% of the highest M5 measurement) EPA is concerned that extrapolating the calibration data might lead to false compliance determinations. Therefore, some environmentally conservative approach must be employed. --------------------------------------------------------------------------- \26\ In this context, M5 is meant to refer to all the methods (Method 5, Method 5A, . . ., Method 5I) used to calibrate PM CEMS. --------------------------------------------------------------------------- Note that the ability to extrapolate beyond the calibration curve in no way would mitigate the facility's requirement to calibrate over its full range of PM emissions. If a facility experiences continuous periods of PM emissions beyond the calibration curve, it would be obligated to perform tests to capture these data into the calibration curve. For example, a facility may determine that it occasionally has several continuous hours of PM CEMS readings which are greater than the 125% point. Several continuous hours are enough time to conduct a M5 test, so the facility would be obligated to conduct M5 tests at this emissions level and include these data in the calibration curve used at the facility. EPA requests comment on how long a period of sustained operations at emissions levels greater than the 125% point would be necessary to require these additional calibration data points. 1. Extrapolating Light-Scattering PM CEMS Calibration Data If it is necessary to extrapolate beyond the 125% point, an environmentally conservative approach would consist of determining the slope of the calibration curve at the 125% point and have the calibration continue with a slope equal to or greater than the slope of the curve at the 125% point. For example if the [[Page 67802]] curve is a log-normal relation, the slope of the curve at the 125% point would be positive, but less than the slope of a straight line that would also describe the correlation between method results and PM CEMS outputs. Therefore, if a log-normal relationship best describes the calibration curve, facilities should extrapolate beyond the 125% point using a straight line beyond the 125% point. The slope of the straight line would be the slope of the log-normal curve, taken from the points on the calibration curve associated with the lowest M5 measurement and the 125% point. If the calibration curve is best described by a straight line arithmetic fit, then extrapolating beyond this 125% point would depend on the slope of any quadratic fit of the data. If the quadratic curve slopes negative at higher values of PM CEMS outputs, then the straight line defined by the calibration would be used to extrapolate beyond the 125% point. If the quadratic fit slopes positive at higher values of PM CEMS outputs, then the quadratic fit would be used beyond the 125% point. Finally, if the calibration curve is a quadratic fit, then the quadratic fit can be used to extrapolate all data \27\. --------------------------------------------------------------------------- \27\ Note: If the slope of the quadratic fit is ever less than zero for values of PM CEMS output above what was measured by the manual method (that is, it ever has a negative slope), this indicates that the correlation between M5 measurements and PM CEMS output is not represented by a quadratic fit and that another mathematical model should be used. --------------------------------------------------------------------------- 2. Extrapolation of Beta-gage Calibration Data For Beta-gage PM CEMS, extrapolating beyond the 125% point would involve continuing the straight line defined by its linear calibration equation. Beta-gage PM CEMS apparently are not sensitive to particle changes in the physical characteristics of particulate, as the light- scattering PM CEMS are. Therefore, a straight-line fit best represents the calibration for beta-gage PM CEMS at all times. E. Need to Calibrate to Twice the Emissions Standard One issue raised by commenters during the comment period for the proposed rule was EPA's proposal that facilities calibrate the PM CEMS to twice the emissions limit. Commenters raised concerns that facilities might not be able to emit PM at a concentration equal to twice the standard. They also said this aspect of the proposal in essence asks facilities to violate the emission standard and could lead to an enforcement action against the facility. Commenters also had concerns that facility personnel may not be sufficiently familiar with the various process and APCD factors to acceptably calibrate the PM CEMS over the full range of operations experienced at the facility. Each of these points are discussed in the following paragraphs. EPA agrees that it would be difficult for many facilities to emit PM at any prescribed level. Many facilities have redundancies in their PM APCDs to such an extent that emitting to the emissions limit may be problematic. However to have accurate PM CEMS measurements, facilities need to calibrate the PM CEMS over the full range of emissions experienced at the facility. As a result, it would be necessary to require facilities to calibrate the PM CEMS over the full range of operations, including PM emissions. This would eliminate the prescriptive nature of how high the calibration needs to be while still addressing the issue that the site-specific calibration of PM CEMS covers the broad range of PM emissions experienced at the facility. EPA does not agree, however, that this approach could cause facilities to violate the manual method MACT PM standard. The PM standard would be defined as the average of three manual method measurements. Any single run above the standard would not be a violation by itself. Average emissions over the calibration would be below the standard for a source equipped with MACT controls. Therefore, we expect that sources would be able to calibrate PM CEMS at levels higher than the PM emissions standard and still remain in compliance with the standard. If this is not practical, however, EPA may consider a waiver of the manual method PM standard during periods of calibrating (and performing RCA tests of) the PM CEMS. The need to obtain and audit an accurate calibration at and above the PM standard may override any concerns about high short-term PM emissions. EPA would want to limit the frequency and duration of calibration runs that exceed the standard, however. We request comment regarding how such limits could be implemented. One way this could occur is to require that sources request in the performance test plan approval to exceed the standard during calibration. Approval to exceed the standard would only be required if the average of all PM CEMS calibration runs is greater than the PM standard. The revised draft PS 11 states that different PM levels should be obtained by varying process conditions or, alternatively, by adjusting the APC system. It is relatively silent in presenting a well-defined protocol with guidelines on how EPA expects calibration tests to be performed. This is because individual sources should know best how to vary their PM emissions. For instance, inserting a throttle plate in lieu of one (or several) bags in a baghouse and varying the opening of the throttle plate(s) is likely an effective way to vary PM concentration for the calibration at a facility equipped with a baghouse. Varying power to an ESP and simulating various failure modes (such as lowering the temperature in the ESP to cause condensation on the plates) is likely vary PM sufficiently for the calibration at sources equipped with an ESP. The experience gained during the PM CEMS Demonstration tests suggests that one can obtain a suitable range of emissions by varying process conditions that affect inlet PM loading to the last in a series of PM APCDs and adjusting the performance of that last APCD. Exactly how this is accomplished at a given facility will vary and depend on the waste fed to the unit, how the facility is designed and operated, and in what order the APCDs are configured. Therefore, the language in the revised PS11 is adequate. More prescriptive language may not work in most cases. Finally, EPA, will be working with industry representatives to develop approaches to better describe how calibration tests should be performed at individual HWC facilities. EPA expects to provide this information in a technical implementation guide. F. Allowing PM CEMS to be Used In-lieu of Method 5 Tests Although the PM CEMS would be required only as an operating parameter, EPA intends to allow facilities to voluntarily elect to use the PM CEMS for compliance with manual methods-based PM standards. Using the PM CEMS for compliance is expected to provide a cost savings to the facility since the facility would not have to conduct periodic Method 5 tests to document compliance with PM standards. Instead a facility could elect to use the PM CEMS measurements during these periodic tests. This would be acceptable if the facility uses the block average of the PM CEMS readings during the M29 tests for the SVM and LVM standards as the particulate ``method result.'' G. Waivers from the PM CEMS Requirements In the proposed rule, EPA requested comment on waiving the PM and Hg CEMS requirement for small, on-site incinerators. See 61 FR at 17439. Upon [[Page 67803]] further consideration, EPA has identified other classes of incinerators where a PM CEMS requirement may be impractical. If the PM CEMS requirement is waived for a given source, the facility would have to comply with operating parameter limits to assure compliance for PM. Of course, a facility could always elect to use a PM CEMS for compliance even if a waiver procedure is promulgated for that facility. 1. Waiver of PM and Hg CEMS Requirements for Small On-site Incinerators EPA is considering whether to waive the PM and Hg CEMS requirements for small, on-site incinerators (SOSI). See the proposed rule, 61 FR at 17439. If a waiver is promulgated, a SOSI would be required to use existing operating parameters in lieu of a PM CEMS to document compliance with the PM, SVM, and LVM standards. 2. PM CEMS Waiver for Sources With Short Life-Spans Given the PM CEMS compliance schedule discussed in section IV.A, above, facilities with short, fixed life-spans raise several issues. For instance, certain government-run incinerators are constructed for the purposes of destroying waste that is too hazardous to transport off-site. These incinerators often have short life spans (ranging from months to a few years) and are constructed to fulfill the requirements of a consent decree, memorandum of understanding (MOU), or other legally binding enforcement agreement. For example the Department of Defense (DoD), acting under a MOU with EPA, may construct an incinerator to destroy nerve-gas agents that are too hazardous to transport. When this activity is complete, the MOU would obligate DoD to dismantle and destroy the incinerator. It does not seem practical to mandate that these facilities use PM CEMS if they will be in service for less than, or slightly longer than, the implementation schedule just discussed. Therefore, EPA is considering a waiver of the PM CEMS requirement for HWCs operating under a legally binding agreement that ensures the source will stop burning hazardous waste within three years of the Compliance Date. EPA could likewise grant a waiver from the PM CEMS requirement for facilities with short life-spans that lack the legally binding agreement discussed above. However, EPA is concerned that without a legally binding agreement to cease operations, the Agency lacks certainty that operations will cease by a prescribed date. For this reason, EPA would consider a waiver for other facilities that plan to cease operations within the first year of compliance with the HWC regulations, that is, prior to the need to use PM CEMS as the operating parameter for PM control. Facilities that operate after the first year would need to have PM CEMS installed, calibrated, meet data availability requirements, determine the PM CEMS operating parameter limit, and use the PM CEMS as the primary operating parameter for PM control. 3. Other Sources As discussed in section III.C. of this NODA, EPA may be unable to determine whether the results of the PM CEMS demonstration test can be transferred to two classes of incinerators: Those with waste heat boilers and mobile incinerators. See section III.C. for more information. V. Other Issues Concerning CEMS and Test Methods for HWCs A. Performance Specifications for Optional CEMS In the proposed rule, EPA proposed other performance specifications for multi-metals, hydrochloric acid (HCl), and chlorine gas (Cl2) CEMS. These performance specifications were proposed as PS10, 13, and 14, respectively. Based on what EPA has learned during the course of demonstrating PM and Hg CEMS, EPA expects not to promulgate the draft performance specifications (PS) for these CEMS at the time of the HWC final rule. As discussed in section II of today's notice, EPA does not plan to promulgate a PS for total mercury (Hg) CEMS either. The Agency has not tested MM and Cl2 CEMS to determine what performance is achievable by the CEMS. Hg CEMS have not been demonstrated as a compliance tool for universal application to all HWCs. EPA has tested HCl CEMS in preparation for the medical waste incinerator rulemaking but did not require the use of HCl CEMS in that rulemaking (see discussion starting at 62 FR 48360, September 15, 1997) and does not believe requiring HCl CEMS for the HWC rulemaking is appropriate either (see 61 FR at 17433). Instead, EPA will consider enabling sources to demonstrate these CEMS on a site-specific basis and to develop performance levels for the CEMS as part of the demonstration. The Agency's only concern is that the CEMS be proven to be a better and more reliable indicator of compliance for the HAP or standard than the requirements specified in the regulations. This approach is now being used to demonstrate a multi-metals CEMS at the Von Roll incinerator in East Liverpool, Ohio. EPA intends to accumulate the CEMS demonstration results and experience and will share that information with permitting authorities and sources wishing to document compliance with CEMS. Since the HCl CEMS have been demonstrated by EPA, we believe the HCl CEMS performance specification could more easily be used as a point of departure for implementing HCl CEMS at a given facility. B. Stack Sampling Test Methods Another question is whether EPA should simplify the task of determining the appropriate manual method tests to be used for compliance. Currently, stack sampling and analysis methods for HWCs are (with a few exceptions) located in RCRA's SW-846 for compliance with the BIF and incinerator rules, and in 40 CFR part 60, Appendix A for compliance with the NSPS and other air rules. Facilities could be required to perform two identical tests, one for compliance with MACT or RCRA and one for compliance with other air rules, using identical test methods simply because one method is an ``SW-846'' method and the other an ``air method.'' Stack test methods HWCs use for compliance should be found in one place to facilitate compliance. EPA intends to reference 40 CFR part 60, Appendix A, when it requires a specific stack-sampling test method. A few SW-846 methods do not have equivalents in 40 CFR part 60, Appendix A, namely the VOST and semi-VOST methods. In these few cases, EPA would continue to refer to these SW-846 methods as well. This discussion only affects stack sampling methods and has no affect on feedstream sampling and analysis. Dated: December 19, 1997. Matt Hale, Acting Director, Office of Solid Waste. Appendix I--Method 5i Method 5I--Determination of Low Level Particulate Matter Emissions From Stationary Sources 1. Applicability and Principal 1.1 Applicability. This method applies to the determination of low level particulate matter (PM) emissions from stationary sources and facilities performing calibrations or calibration audits of particulate matter continuous emission monitors as specified in the regulations. The method is effective for total train catches of 50 mg or less. The minimum detection limit for this method can be determined by repeatedly collecting and analyzing blank samples. A blank sample is a sample of blank air collected and analyzed [[Page 67804]] in the normal manner. The limit of detection can be calculated by collecting and analyzing seven blank samples and then calculating an estimate of the sample standard deviation of these blanks. The limit of detection would be three times the estimated sample standard deviation. 1.2 Principal. The PM is withdrawn isokinetically from the source and collected on a 47 mm glass fiber filter maintained at a temperature of 120 deg. 14 deg.C (248 deg. 25 deg.F). The PM mass, which includes any material that condenses at or above the filtration temperature, is determined gravimetrically after the removal of uncombined water. 2. Apparatus 2.1 Sampling Train. The sampling train configuration is the same as shown in Method 5, Figure 5-1. The sampling train consists of the following components: Pitot Tube, Probe liner Differential Pressure Gauge, Filter Heating System, Condenser, Metering System, Barometer, and Gas Density Determination Equipment. Same as Method 5, Sections 2.1.2 to 2.1.4, 2.1.6 and 2.1.7 to 2.1.10, respectively. 2.1.1 Probe Nozzle. Same as Method 5, Sections 2.1.1 with the exception that it is constructed of Borosilicate or quartz glass tubing with sharp, tapered leading edge. 2.1.2 Filter Holder. The filter holder for this sampling train is constructed of Borosilicate or quartz glass front cover designed to hold a 47 mm glass fiber filter, with a stainless steel filter support, a silicone rubber or Viton O-ring and Teflon tape seal. The holder design will provide a positive seal against leakage from the outside or around the filter. The filter holder assembly fits into a stainless steel filter holder and attaches immediately at the outlet of the probe (or cyclone, if used). The tare weight of the filter, Borosilicate or quartz glass, stainless steel filter support, silicone rubber or Viton O-ring and Teflon tape seal will not exceed 31 grams. The filter holder is designed to use a 47 mm glass fiber filter meeting the criteria in section 3.1.1 of Method 5. Figure 5I- 1 presents a schematic of the filter holder system. These units are commercially available. 2.1.3 Glass Plugs and Clamps. Once the filter holder has been assembled, desiccated and tared it is critical that the filter be isolated from any external sources of contamination. This can be accomplished by covering the leak-free ground glass or O-ring socket on the front half glass filter cover with a Borosilicate or quartz ground glass plug. The plug shall be secured in place with the appropriate sized laboratory impinger clamp or any system that can ensure a leak-free fitting. It is beneficial to place the glass plug on the inlet socket as soon as the unit is assembled, however do not tare the assembly with the plug in place, as this will increase the tare weight introducing additional error into the final weighings. 2.2 Sample Recovery. Is the same as Method 5 for: Glass Sample Storage Containers, Graduated Cylinder and/or Balance, Plastic Storage Containers, Funnel and Rubber Policeman (Method 5 sections 2.2.3, 2.2.5--2.2.8, respectively) with the following exceptions: 2.2.1 Probe-Liner and Probe-Nozzle Brushes. Teflon and nylon bristle brushes with stainless steel wire handles, should be used to clean the probe. The probe brush shall have extensions (at least as long as the probe) of Nylon, Teflon, or similarly inert material. The brushes shall be properly sized and shaped to brush out the probe liner and nozzle. 2.2.2 Wash Bottles--Two. Teflon wash bottles are recommended however, polyethylene wash bottles may be used at the option of the tester. It is recommended that acetone not be stored in polyethylene bottles for longer than a month. 2.2.3 Sample Holder: A portable carrying case with clean compartments of sufficient size to accommodate each filter assembly. The filters shall be able to lay flat with the stainless steel filter support placed down in the compartment. This system should have an air tight seal to prevent contamination to the filters during transport to and from the field. It is recommended that desiccant be used in this case. The desiccant, if used, is housed in a container that is capped with a 0.1 micron screen to ensure that no dust particles can contaminate the outside of the filter h [Text truncated at 120,000 characters. The full text is on the page linked above.] --- Source: Frix Law Library, https://www.frixlaw.com/law-library/documents/fr%3A97-33740. Public record. Not legal advice.