



MEMORANDUM

TO:	Mary Johnson, U.S. Environmental Protection Agency, OAQPS/SPPD

FROM:	Eastern Research Group, Inc.

DATE:	June 30, 2016

SUBJECT:	Review of PM and Urban Metal HAP Data Variability for Boilers



1.0	INTRODUCTION
      The EPA received a petition asserting that the public lacked an opportunity to comment on a new provision in the February 2013 final amendments to the Area Source Boilers Rule (40 CFR part 63 subpart JJJJJJ) that eliminated further performance testing for particulate matter (PM) at certain boilers, if their initial compliance test shows PM emissions are equal to or less than half of the PM emission limit. The EPA granted reconsideration on this issue to provide an opportunity for comment on the provision. The EPA requested comment, along with supporting information, on the magnitude and range of variability in PM and urban metal hazardous air pollutant (HAP) emissions from individual boilers. This memorandum summarizes the data used to inform the EPA's decision on the PM testing provision in its final action on reconsideration.
      To assess the magnitude and range of variability in PM and urban metal HAP emissions from individual boilers and whether the emissions variability at an individual boiler within a specific subcategory could result in an exceedance of the applicable PM limit by such boiler whose PM emissions are demonstrated to be equal to or less than half of the applicable PM emission limit, data submitted as a result of two Information Collection Requests (ICRs) (approved under OMB Control No. 2060-0616) and subpart JJJJJJ compliance reporting requirements were reviewed. 
      

2.0	URBAN METAL HAP EMISSION VARIABILITY
      The variance of total selected non-mercury metallic HAP (TSM8, which is composed of the metals As, Be, Cd, Cr, Pb, Mn, Ni, and Se) emissions was examined. Since PM serves as surrogate for all urban metal HAP (including Hg) for biomass and oil boilers, Hg variability was also assessed. The data assessed was obtained through the two ICRs.
      Twenty-two area source boilers reported data on one or more of the TSM8 pollutants in the ICR database; of those units seven had more than one emission test result to evaluate intra-unit variability and none tested for all eight metals. An eighth data point was available when we derived an emission rate from a lead fuel analysis. Cadmium and lead were the most prevalent pollutant. Six of these eight units fired biomass and each of these biomass-fired boilers co-fired a variety of fuels. Specific biomass fuels included tire derived fuel, unadulterated timber, and plant-based agricultural residue. The oil-fired boilers simply fired No. 2 distillate fuel oil. Table 1 below shows intra-unit variability, specifically a ratio of maximum to minimum emission rates from each performance test conducted at each of the seven boilers. The emission ratios for biomass-fired boilers ranged from 1.00 to 7.28 for cadmium with two ratios greater than 1.00; and for lead the ratios ranged from 1.00 to 6.40 with four ratios greater than 1.00. For the oil-fired units, the lead emission ratios were 1.00. The details of the tests for each of the eight boilers are in Appendix A-1 and Appendix A-2.  No TSM8 data were available at area source coal-fired boilers.

Table 1. Area Source Metals Variability in ICR Data
Source Type
                       Facility / Boiler (Fuel Details)
                             Measure of Variance; 
                             Ratio of Max over Min

                                       
                                    Cadmium
                                     Lead
Area - Biomass
CAPacificOroville / Boiler 1
(Biomass; Ag. Residue; Unadulterated Lumber &Timber)
                                     1.00
                                     1.00
Area - Biomass
CAPacificOroville / Boiler 2
(Biomass; Ag. Residue; Unadulterated Lumber &Timber)
                                     1.00
                                     1.00
Area - Biomass
MIGraylingGeneratingStation / Boiler1 
(Reinjection char.; TDF; Plywood, Particleboard with glues or resins; Unadulterated Lumber and Timber)
                                     1.00
                                     1.00
Area - Biomass
MIGraylingGeneratingStation / Boiler1 
(Secondary Mill Residues; TDF; Unadulterated Timber)
                                     7.28
                                     6.40
Area - Biomass
MIGraylingGeneratingStation / Boiler1 
(Secondary Mill Residues; TDF; Unadulterated Timber)
                                       -
                                     1.00
Area - Biomass
MIHillmanPower / EUBoiler
(Unadulterated Lumber; Secondary Mill Residuals; TDF)
                                       -
                                     1.00
Area - Biomass
MIHillmanPower / EUBoiler
(Unadulterated Lumber; TDF)
                                     3.06
                                     1.35
Area - Biomass
MIVikingEnergyofLincoln / Boiler1
(Hog Fuel; Reinjection char; TDF; Unadulterated Timber)
                                     1.00
                                     1.00
Area - Biomass
MIVikingEnergyofLincoln / Boiler1
(Reinjection char; TDF; Treated Wood; Unadulterated Timber)
                                       -
                                     1.00
Area - Biomass
MIVikingEnergyofLincoln / Boiler1
(Reinjection char; TDF; Unadulterated Timber)
                                       -
                                     1.07
Area - Biomass
MIVikingEnergyMcBain
(TDF / Wood: Unadulterated Timber)
                                       -
                                   1.39[(a)]
Area - Oil
VAMerckElkton / B-7
(No. 2 Distillate)
                                       -
                                     1.00
Area - Oil
VAMerckElkton / B-8
(No. 2 Distillate)
                                       -
                                     1.00
Footnote:
(a)	These tests include three stack test events and a fourth data point derived from a metals fuel analysis.

      Intra-unit TSM8 emission variability from the best performing major sources was also examined. There are several best performing major source coal and biomass units with multiple emissions tests for TSM8. Table 2 below shows average intra-unit variability for subcategories of major source boilers. In total, 18 performance tests were available from eight biomass-fired boilers. The average intra-unit ratio of maximum to minimum emission rate was 2.79. There were less TSM8 data for coal-fired boilers, at two different boilers, each with two TSM8 tests, but variability was comparable: the average intra-unit TSM8 emission ratio was 2.55. There were no oil-fired boilers with multiple tests for TSM8. The details of these tests are shown in Appendix A-3.
      There were three major source units in the best performers list (one coal, and two oil) that also had more than one sample date for fuel analysis. This small dataset shows little variation in fuel, each had a ratio of less than 2. It should be noted that the number of days between samples at each of these three boilers were 1-day, 2-days, and 25-days, respectively. Table 2 shows the intra-unit variability for these three units. The details of these fuel analyses are shown in Appendix A-4.

Table 2. Major Source Metals Variability in ICR Data
                                  Source Type
                              Boiler Subcategory
                                      ICR

                                       
                                   Data Type
                 Count of Boilers; Count of Total Test Events 
                  Measure of Variance; Average Ratio of Max 
                                   over Min
Major
Biomass
                                     TSM8
                                     8; 18
                                     2.79
Major
Coal
                                     TSM8
                                     2; 4
                                     2.55
Major
Coal
                                   TSM8[(b)]
                                     1; 2
                                     1.14
Major
Oil (Heavy)
                                  TSM8[(b)] 
                                     1; 2
                                     1.18
Major
Oil (Light)
                                  TSM8[(b)] 
                                     1; 2
                                     1.73
Footnote:
(b)	Based on fuel analysis instead of stack test data.

      The tables above show that emissions from a given source are not constant but can vary. In particular one unit (i.e., MIGraylingGeneratingStation - Boiler 1) in Table 1 was tested with different fuels. One test with secondary mill residue had much higher cadmium and lead variances. 
      Mercury variability data were limited at area source oil-fired and biomass-fired boilers. Only one oil-fired boiler reported variability in Hg emissions. There were eight different fuel samples across a 12-year period. The ratio of the maximum fuel sample to the minimum fuel sample was over 100. However, all of these results were reported as less than the detection limit. Therefore, the variation in emissions does not reflect the actual measured emissions, but instead variations in detection limits from different sampling companies or test methods. The details of these tests are shown in Appendix A-5. For the biomass-fired boilers, nine different facilities had multiple Hg fuel analysis samples and eight of the nine of these facilities showed limited variability, with emission ratios less than 2. One facility had 20 different samples over a two week period and had an emission ratio of 4.6. The details of these tests are shown in Appendix A-6. Three biomass-fired boilers had multiple Hg performance tests. The ratios of the maximum performance test emission rates were: 37.1, 3.2 and 16.2. The boiler with the 37.1 ratio can be partially attributed to changes in fuel blend. Specifically, heat input for the 1999 test was 93% biomass while the 2005 test fired TDF. The exact percent contribution of TDF in 2005 is unknown. The details of these tests are shown in Appendix A-7.
3.0	PM EMISSION VARIABILITY
      Variance of PM performance test data was examined using compliance data reported for area source boilers in WebFIRE and the two ICRs.
3.1	Area Source Boiler Data
      40 CFR part 63 subpart JJJJJJ requires facilities to submit results of performance tests to the Compliance and Emissions Data Reporting Interface (CEDRI). Sixteen total PM test reports at area source boilers were identified in WebFIRE. Intra-unit PM variability could not be assessed, however, because only one test result was available for each boiler. The details of these tests are shown in Appendix B-1. PM emission test data was reported in the ICR databases from area source boilers. There were 19 different test results from coal-fired boilers and 23 different test results from biomass-fired boilers. The details of these tests are shown in Appendices B-2 and B-3. Again, intra-unit PM variability could not be assessed because only one test result was available for each boiler.
3.2	Major Source Boiler Data
The variability of PM emissions at major source boilers was evaluated given that there was not multiple PM test data at area source boilers to assess intra-unit variability. The 83 best performing coal-fired boiler test run averages were compared to half of the appropriate PM limits in the area source rule. Boilers with multiple performance tests were used to measure intra-unit PM variability. In total, 40 coal-fired boilers had multiple test events and four of those units (10%) had an instance where the minimum test average was below half of the coal-fired area source PM emission limit (i.e., 0.03 for units 30 MMBtu/hr or greater, 0.42 for units between 10-30 MMBtu/hr) and the maximum test average at the same unit was above the emission limit. (See Appendix B-4). The ratios of the maximum performance test emission rates to the minimum emission rates were: 14.4, 38.5, 13.7, and 37.8.
      The 39 best performing biomass-fired boiler test run averages (across all biomass subcategories) were compared to half of the appropriate PM limits in the area source rule. Out of the 39 tests, 17 biomass units had multiple test events and there were no cases where one test run average was below half of the limit but another test run average was above the limit. (See Appendix B-5).
      The best performing major source oil-fired boiler test run averages were compared to half of the appropriate PM limits in the area source rule (0.03 for all liquid units). There were 8 test run averages, and three units had more than one test. There were no cases where one test run average was below half of the limit but another test run average was above the limit. (See Appendix B-6.)


