January 10, 2007
Dilution-Based Emissions Sampling From Stationary Sources: Part 2- Gas-Fired Combustors Compared With Other Fuel-Fired Systems
By England, Glenn C; Watson, John G; Chow, Judith C; Zielinska, Barbara; Et al
Regulatory management for fine particulate matter (PM^sub 2.5^) reductions has created a growing need for emission factor data and source chemical profiles that go beyond the database available in current emission inventories. These kinds of data are needed to support investigations of PM^sub 2.5^ health and environmental effects and provide improved resources for air quality modeling. Evaluation of approaches for sampling PM^sub 2.5^ from combustion systems indicates that the emissions estimates are method dependent and somewhat ambiguous. These ambiguities result from the use of certification methods that filter particles from stack gas at high temperatures. Then they attempt to account for additions to PM^sub 2.5^ mass by condensable vapors in the atmosphere at lower temperature by passing filtered stack gas vapor through iced water impingers. An alternate approach, adopted for mobile source test methods, uses clean filtered air to dilute the hot moist stack gas in a mixing chamber that simulates cooling to ambient conditions forming condensable particulate matter (CPM) before actual sampling for mass and chemical composition.1,2 Differences in these methods arise from taking into account the PM^sub 2.5^ mass after the partitioning at ambient conditions between stable condensable species and semivolatile material, including carbon compounds. The dilution method simulates the entry of hot emissions into the ambient air and provides a means of establishing a reproducible benchmark estimate of PM^sub 2.5^ at the point of mixing and cooling at the stack.1-5 The dilution method is widely used to characterize mobile source emissions as a self-consistent, reproducible method for compliance applications. U.S. Environmental Protection Agency (EPA) has developed a dilution method for stationary sources (CTM- 039).1,6 ASTM International7 also is considering a voluntary consensus standard for stationary source dilution samplers.
Until recently, one of the main problems in using dilution samplers for stationary sources was the bulkiness and/or operational complexity of the samplers available. The current availability of low-cost, lightweight, computer- based instrumentation and controls has considerably simplified operation. Previously, the sampler size and weight had made them difficult to deploy for stationary sources, especially those having elevated stack access. To minimize this problem, compact dilution samplers (CDSs) were designed and tested.2,6 England et al.3 describe such a sampler, its sampling methodology and performance, and establish an intercomparison between the CDS and the Desert Research Institute (DRI) reference sampler.2,8 In the comparison tests, the two dilution samplers were found to produce essentially the same results for the stationary sources examined.3 The methodology follows that developed by several researchers and provides a means of comparison between the dilution method and certified stationary source test methods used for regulatory purposes, such as EPA Methods 5, CTM- 040, 201A, and 202.
One of the major areas where knowledge is limited for PM^sub 2.5^ emission factors concerns gas-fired combustion systems. An important aspect of the CDS development foresaw its application to sampling these combustors, of which particulate emission rates are very low relative to other combustion systems. In this paper, the results reported previously2,3,8,9 are extended, describing emissions characterization for a series of combustors using a dilution sampler technique. The characterization includes the mass emission rates, particulate matter (PM) precursor gas emissions, and PM source chemical profiles. The gas-fired combustor results are compared with tests of a stationary diesel engine source and other literature data for combustion systems using coal and biomass to provide a perspective on the different PM^sub 2.5^ emissions and their source profiles.
The sampling program used the CDS described earlier,3 with parallel testing in two cases using the DRI dilution sampler.8,9 The DRI sampler is a large bulky unit compared with the CDS, following the design of Hildemann et al.9 and Chang and England.10 It was used for sampling of a number of the combustion units, and later the CDS application completed the program. The combustors tested are listed in Table 1. They include different gas-fired process heaters, an institutional boiler, and electricity generators selected over a range of EPA source classification code (SCC) numbers. Listed in the table are the general specifications for the units studied, including emission controls, along with the samplers used in each test and references to reports giving details of the tests.3 A diesel fuel-powered backup electrical generator sampled is included in the table. Multiple sampling runs were taken at each site for uncertainty analysis as indicated in the table. The data are reported in terms of a PM^sub 2.5^ emission factor in pounds per million Btu (MMBTU) and in terms of PM^sub 2.5^ mass abundances (concentrations of each species divided by PM^sub 2.5^ mass) for source profiles. Data also were acquired for emission factors of (1) inorganic reactive gases (SO^sub 2^, nitrogen oxide [NO^sub x^], and NH^sub 3^) relevant to PM^sub 2.5^ formation, (2) volatile organic vapors with carbon number >7 (volatile organic compound [VOC]^sup 8+^) of potential interest as PM^sub 2.5^ precursors, and (3) coemissions of semi-VOCs (SVOCs) as potential PM^sub 2.5^ tracers. The emissions of SVOC, including polycyclic aromatic compounds (PAHs), involved identification of a large number of trace compounds found at or near ambient concentrations or near sampler blank levels. Because these results are exploratory in nature and complicated in their detail, they are reported elsewhere in the references cited in Table 1 and are not included here.
Sampling and Testing
The sampling procedure followed the general stack sampling approach given for EPA Method 5.21 The samplers were first cleaned with a rinse of distilled, deionized water (DDW) followed with an acetone rinse to remove visible deposits. Samplers were then wrapped in a heating blanket and baked out to remove residual organics. During bakeout, a small flow of purified air was passed through the system for 2-4 hr. High-efficiency particulate air (HEPA) filters and charcoal dilution air filters were used to purify the dilution air. The HEPA filters were replaced occasionally throughout the program but never exceeded the maximum rated pressure drop across the filter during sampling. Activated charcoal filters for cleaning the inlet dilution air were regenerated or replaced before each test series.
Sample media for specific tests, including filters and adsorbers, were prepared in the laboratory before sampling. Samples for mass and elemental compos\ition were collected on 47-mm polytetrafluoroethylene (PTFE)- membrane filters, and those for carbon and water-soluble ion analysis were collected on prebaked 47- mm quartz fiber filters (QFFs). Representatives from each batch of filters were tested for contaminants before sampling.
VOC samples were collected in stainless steel canisters, and SVOCs were collected with a filter pack consisting of PTFE- impregnated glass fiber filters followed by a polyurethane foam and Amberlite XAD-4 sorbent resin. Samples for NH^sub 3^ and SO^sub 2^ were collected on citric acidimpregnated and potassium carbonate- impregnated filters, respectively, which were placed behind the PTFE and quartz filters. The filters were assembled in holders capped and stored appropriately.
The test instrumentation, including pressure transducers, Pitot tubes, relative humidity (RH) sensors, and thermocouples, was recalibrated periodically. The venturi flow and orifice elements were calibrated once but were inspected routinely for damage before testing and were recalibrated as required. Thermal mass flow meters were calibrated annually by the manufacturer; thermocouples were calibrated annually, with a mercury-in-glass thermometer as a reference. The RH sensors and the rotameters were factory calibrated. A 2.5-μm aerodynamic diameter cut point in-stack cyclone (Anderson CASE-PM^sub 2.5^), flow rate 25 actual L/min, was added to the probe head to remove large and coarse particles. Typically run times of 6 hr were selected for the gas-fired units to collect enough material for maximizing the number of chemical components detectable within a single test day (shorter run times are considered feasible when sampling for particulate mass only). Run times of 20 and 120 min provided sufficient sample mass for chemical analysis from the diesel generator without and with the diesel particulate filter, respectively.
The dilution sampler was assembled on the ground, heated to 150 C for bakeout, and raised fully assembled to the stack catwalk (typically using a conventional pulley and rope for the CDS and a light crane for the larger DRI sampler) along with ancillary sample collection equipment (filters, pumps, etc.). Before each sampling run, measurements of stack gas velocity, temperature, and oxygen concentration along two orthogonal axes in the stack were made before sample collection to determine stack gas conditions and identify representative sampling locations. The dilution sampler then was loaded with sample media and leak tested. A maximum of 2% of the total flow through the samplers was allowed for leakage, although actual leakage typically was
After completion of a sample run, the probe was removed, and the heated components were allowed to cool off. Sample media were removed from the manifold and stored in a cooler. The undiluted portions of the sampler (the probe and sample venturi flow meter) were rinsed with DDW followed with acetone to recover wall deposits. Because previous studies showed that deposits in the tunnel and residence time chamber of the sampler were expected to be
The samples collected were submitted to the DRI Environmental Analysis Facility for chemical analysis,22,23 which included: (1) mass by gravimetry; (2) elements by X-ray fluorescence24; (3) organic carbon (OC) and elemental carbon (EC) fractions by the Interagency for Protected Visual Environment (IMPROVE) Thermal Optical Reflectance method25,26,28; (4) water-soluble anions and cations and adsorbed gases by automated colorimetry and ion chromatography27; (5) VOCs by gas chromatography (GC)29; and (6) SVOCs by solvent extraction GC with mass spectroscopic and Fourier infrared detection.30
Emission factors were calculated for each test run from mass collected, measurements of volume flow of the stack gas, and fuel flow rate; the results for the runs for each combustor were averaged. Volume flow was calculated in dry standard cubic meters per unit time. Data were collected to include mean values, as well as standard deviations, based on the number of samples taken. The uncertainties in the emission factors are based on gravimetric mass determinations and flow rates. Uncertainties in the source profile chemical abundances include the variations from the samples and the analytical determination errors expected in the analytical methods. Total uncertainty and uncertainty bounds were calculated for each dataset by propagating systematic measurement errors from each measurement element using addition in quadrature and using the standard deviation to calculate the random uncertainty.
Field blanks typically included "trip blanks" (media samples shipped from laboratory to field and back for determining background levels associated with transport and laboratory sample handling) and "dilution system blanks (DSBs)" (samples collected from the dilution sampler but with only filtered dilution air flowing through the system for determining total background levels in the sample). Generally, trip blanks for media used to sample PM^sub 2.5^ mass and composition were not significant relative to sample and DSB concentrations. Therefore, the results are not blank corrected. The significance of DSBs for PM^sub 2.5^ mass and composition determination varied among the field sites depending on the component of interest. DSBs for some of the gas combustion tests showed that PM^sub 2.5^ mass and species background levels in the dilution air were significant for some of the units. Relative concentrations of iron, chromium (not detected), nickel, molybdenum (not detected), and manganese (not detected) in the blanks did not indicate significant contamination from the stainless steel surfaces in the sampler. This indicates that certain low-level component concentrations were masked by the dilution air rather than background in the sampling media or from the sampler itself and suggested the likelihood of a bias in the average results for those components. Comparison of DSBs and nearby ambient air PM^sub 2.5^ mass concentrations at one field site suggest a probable PM^sub 2.5^ mass removal efficiency of ~80% across the HEPA and charcoal filter systems in both the DRI and CDS samplers. Because the PM^sub 2.5^ mass removal efficiency is similar across the different HEPA filter designs used in the two samplers, and because HEPA filter removal efficiency is rated at 99.97% for particles >0.3 μm, this suggests that much of the apparent particle penetration may be because of ultrafine (
An example of CDS stack and pretest DSB results are shown in Figure 1 for the gas-fired unit Site Echo. This site had among the lowest PM^sub 2.5^ stack concentrations by mass of the gas-fired units tested in this study. Results for the DRI dilution sampler indicated similar characteristics. A ratio of sample to DSB of ~1 in the figure indicates that the estimated PM^sub 2.5^ composition was influenced by PM^sub 2.5^ found in the filtered dilution air. At values of this ratio increasingly
The data in the figure indicate that many of the major components and PM^sub 2.5^ mass are well above the DSBs, but a number of the trace elements and carbon components are at sample/DSB ratios of ~1 or less. Composition data in this range, where system blanks are similar to or larger than sample concentrations, have a large and ill-defined uncertainty.
The data generated for emissions factors and for source profiles give a cross-section of the combustors that are incorporated in a "standard" portfolio, such as EPA's AP-42 guidelines31 and the SPECIATE source profile library.32
RESULTS AND DISCUSSION
PM^sub 2.5^ and Gas Emission Factors
The results for the estimation of average PM^sub 2.5^ mass emission factors and uncertainty estimates based on the dilution samplers are summarized in Table 2 for the gas- and oil-fired units. Detailed test data by run and unit are given by England.2 These results indicate that the gas-fired units have substantially lower PM^sub 2.5^ emissions than the oil- fired combustor and the diesel engine tested. Comparison with diesel powered truck data or data from coal-based units indicates that both the gas and the diesel units produce lower emissions than coal burning, even in emission- contro\lled units. For example, data for eastern coal combustion in a pilot unit indicate that the average coal- fired emissions are ~0.36 lb/MMBTU and 0.068 lb/ MMBTU before and after a baghouse filter, respectively, for PM^sub 2.5^ removal.33 The average mass emission rates are considerably higher than the gas-fired units tested (
Gas-phase PM^sub 2.5^ precursor emissions from the combustors also are of interest. These are summarized in Table 3; the detailed data are given by England.2
The precursors of concern for ambient particles include SO^sub 2^, nitrogen oxides (NO^sub x^ = NO + NO^sub 2^), NH^sub 3^, and VOC, particularly of carbon number 8 and higher (VOC^sup 8+^). The VOC focus on higher molecular weight species recognizes that lighter molecular weight species do not contribute to particle formation in the atmosphere. 34,35 The emissions listed in the table are obviously dependent on the postcombustion controls listed in Table 1.
Some generic combustion unit comparisons can be made for units with and without controls, as noted in Table 3. For example, the SO^sub 2^ emissions were found to be fuel sulfur dependent, as also found for coal combustors. 33 This study found that the SO^sub 2^ emissions for gas- fired units were directly proportional to fuel sulfur content, as expected from many previous studies.
The NO^sub x^ emissions were lowest for units with selective catalytic reduction of NO^sub x^ emissions (SCR). For example, SCR decreased the NO^sub x^ emissions compared with uncontrolled units typically by ~0.1-0.2 lb/MMBTU to 1 order of magnitude lower levels for the gas-fired units. The oil-fired combustor displayed much higher NO^sub x^ emissions than the gas-fired units. NH^sub 3^ emissions generally were very low in uncontrolled units and increased with SCR, evidently as a result of NH^sub 3^ slippage through the catalyst unit. The data reported for SO^sub 2^ and NO^sub x^ emission factors are generally consistent with a large number of tests done previously on gas- and oil-fired units.31 The NH3 data and perhaps the SO^sub 2^ data should be used with discretion, because they are potentially reactive on surfaces, potentially decreasing the reliability of the samples in the CDS.
As expected, the VOC^sup 8+^ emissions are generally low compared with oil-fired units. A sample of VOC^sup 8+^ speciation data acquired using the canister or Tenax-based data is listed in Table 4. The data shown mainly correspond with the fraction identified, which were at concentrations of a factor of five or more above the ambient concentrations or the DSBs. The average emissions are highly uncertain and varied substantially from combustor to combustor. The results indicate that a variety of trace species are present in the emissions. Prevalent among the species identified are a series of alkanes and substituted aromatics. Emphasized in the table are compounds, including some isomers that were observed in emissions from at least four of the six units studied. These include two alkanes, nonane and 3(2),6-dimethyl octane, and the substituted aromatics, the xylenes, styrene, trimethylbenzene, ethyl- and propyl- benzene. The aromatic compounds have emerged as having particularly substantial secondary OC-forming potential in the atmosphere.34,35
Emissions from Different Methods. A perspective for the dilution sampler measurements relative to other data in the literature is important. This is facilitated in Table 5 with a comparison between standard certification methods for reporting emissions as the sum of filterable particulate matter (FPM) and CPM. The comparison is made assuming that the literature reports of PM^sub 10^ from gas-fired units are approximately equivalent to PM^sub 2.5^; that is, the gas- fired combustors emit particles principally in the PM^sub 2.5^ range. As indicated in the table, the PM^sub 10^ emission rates, including those in the standard AP-42 reference,36 are consistently much larger than those found for the dilution samplers. The principal contribution to total PM^sub 10^ emissions by the standard reference methods is the CPM fraction. The dilution sampler results from the units investigated in this study are consistent with those reported for gas-fired household appliances reported by Hildemann et al.37 This leads to the conclusion3 that the PM mass emission data reported for gas-fired combustors are method dependent. This indicates further the need to review and update the present reference methods at the regulatory level to incorporate and account for recent sampler and analytical developments.
Of interest are composition data for emitted PM^sub 2.5^, which make up source profiles obtained in the study. Tables 6-10 list the average results obtained for the combustors studied. The results for these profiles include the watersoluble species as ions of interest, and the quantifiable elements.
The estimates of EC and OC are also shown; the estimates of SVOC from the backup filters are included for comparison with the stable condensed material found on the sampling filters. The potential importance of the volatility of the OC component is illustrated in the data shown in the tables. Typically for gas- and oil-fired units, the backup filter OC is as large as the primary OC fraction. This implies large uncertainty in the OC measurements because of the potential for measurement artifacts resulting from adsorption of VOCs on the QFFs. The data are important for interpreting the carbon emissions from any combustion source. In the tables, the elements that have an uncertainty in estimate exceeding 100% (i.e., cannot be distinguished from zero at the 95% confidence level) are flagged. These fractions of the total PM^sub 2.5^ mass emitted should be used with caution for quantitative analyses, including receptor modeling. By mass fraction, the carbon components and sulfate are generally the largest portion of the PM^sub 2.5^ samples taken, except for the No. 6 oil-fired boiler, for which sulfate is by far the largest fraction. Ammonium and nitrate are also found in significant amounts. In one case in particular (Table 8), located in a coastal area, Cl^sup -^ was found to be significant in the PM^sub 2.5^ composition, although this observation is not consistent among all of the tests conducted in such areas.
The trace elements generally make up a very small fraction by mass of the PM^sub 2.5^, except for the No.6 oil-fired boiler. The trace metal concentrations in the stack gas from the gas-fired units were similar to or less than the system blanks in some cases, as illustrated in the example shown in Figure 1. Thus, the trace element data in these cases are considered questionable. In the oil- fired boiler case, trace elements from fly ash make up a substantial fraction of the PM^sub 2.5^ sampled. The diesel engine-powered generator samples showed essentially all carbon with small amounts of SO^sup -^^sub 4^ and NO^sup -^^sub 3^.
From the sampling and chemical analysis in this study, there does not appear to be any single unique chemical marker for gas combustion emissions. However, some insight can be gained about the nature of the trace metals by comparison with soil dust composition. Comparison of the ratio of metals to Si shown in Tables 6-8 illustrates similarities and differences in the metal ratios compared with typical soil dust ratios. These results are qualitatively consistent with the hypothesis that the trace element composition is similar to soil dust on which is superimposed enriched metals related to steel, such as, Ti, Mn, Mo, and Zn. Enrichment of V and Ni may be related to steel manufacture or simply to concentrations in fuel oil residue in a dual fuel-fired boiler. Given the level of dilution systems blank data relative to the trace elements, the apparent soil dust component may reflect ambient air conditions. This may imply that small amounts of soil dust in the gas-fired emissions are present from the combustor inlet airstream or in the fuel from production residue or transportation contamination combined with eroded metallic material from the process equipment.
The oil-based data indicate that the presence of V or Ni is usable as an oil combustion marker, as determined in a variety of historical work. In the case of oil or diesel oil firing, the actual profiles will depend quantitatively on the trace chemical composition of the fuel burned. Thus, the data found in this study require fuel use qualification if they are to be used for source apportionment.
Illustrative data37-43 for comparison with the gas-, oil-, and coal-fired profiles from this study are shown in Table 11. Included is a sample of a heavy-duty dieselpowered vehicle; representative coal supplies of the Western United States, Texas, and the East; and two types of biomass, soft wood and grass, as surrogates for material that might be used in energy-producing facilities. Along with the data from this study, the tabulated data clearly indicate the variability in emissions as a function of the fuel type.
The results in Table 12 for the heavy-duty dieselpowered vehicle are similar to the diesel generator in this study, except more trace metals were reported. These may be associated with lubricating oil, fuel quality, or engine wear. Evidently reciprocating internal combustion diesel engines will produce in combustion more EC than OC,42,43 unlike continuous gas-fired combustion processes. The coal combustion data indicate that some fraction of carbon is present in the emissions, which evidently varies considerably in these examples, with coal source, combustor, and emission \controls. For the coal facility samples, the carbon fraction is ~5% except for Texas lignite. There is also a significant presence of acid species in coal PM^sub 2.5^, including sulfate and nitrate in the emissions. The trace elements found in the coal samples are generally similar to one another as fly ash. The enrichment of Se is clear from the data and is commonly used as a coal combustion tracer. The variability of As in the samples suggests that this element is not a particularly good tracer for coal combustion.
The results for the biomass illustrate that, again, OC and EC can be expected as a relatively large fraction of the PM^sub 2.5^ emissions. The acid species are variable depending on the S, N, and Cl content of the biomass and the combustion process. The trace elements in biomass typically are lean in metals, especially the heavier trace elements, such as V, Ni, Se, and Pb. As is commonly known, biomass combustion is well marked with enriched K relative to other elements. Yet, the fraction is not necessarily greater than coal combustor samples. Comparison of the gas-fired and oil-fired combustor samples with the coal and biomass results do not suggest a unique elemental tracer for gas or diesel oil combustion.
The "conventional" particle source profiles for the tested gas- fired units did not yield unambiguous markers for such combustors. As an alternative, the VOC and SVOC speciated data were examined for possible markers. The organic species are also of interest in their own right for characterizing components potentially of interest in human exposure. As illustrated for VOC^sup 8+^ in Table 5, the composition of VOC varied substantially from unit to unit. This also was the case for the SVOC, including the PAHs (not shown here). The organic vapor composition did not provide added insight for unique tracers of gas- fired combustors.2,11-20
In the aggregate, the carbon components observed in the tests of the gas-fired units are listed in Table 12 for comparison with the Hildemann et al.37 work characterizing natural gas-fired household appliances in Southern California. Their results indicated lower OC concentrations than found for industrial combustors, except for two process heaters. The EC emissions tend to be less than OC except in one case (API/B).12 The VOC^sup 8+^ emissions tend to be similar to total particulate carbon (TC), but SVOC was much smaller than TC or VOC^sup 8+^. The results of this exploratory work11-20 indicate that the major portion of the PM^sub 2.5^ emissions consisted of empirically aggregated material as OC. Further study of VOC and SVOC emissions from gas-fired combustors is warranted with continuing efforts to characterize emissions from these units.
Evaluating Uncertainty and Variability
There are two useful measures of ambiguity in emissions determinations. There is uncertainty in the reproducibility of emission measurements from the same source, resulting from operating conditions, ambient conditions, measurement protocol, and so forth. There is variability in emissions measurement across sources within the same category or SCC code resulting from the age and maintenance of the source, as well as modifications over time to modify performance, and so forth. The uncertainty in emission factors reported in the tabulations in this study applies to the reproducibility of measurements for the same source. Nominally, uncertainty is designated in terms of the standard deviations of the data and confi- dence limits, as well as the range of estimates from individual runs at the units sampled. The confidence limits and standard deviations also include the precision and accuracy estimates from the sampler evaluation described by England and colleagues.2,3 Variability can be estimated in an exploratory way at least from the combustors by SCC as indicated in Table 1. The variability levels can include random uncertainties (precision) and systematic uncertainties (bias) associated with differences in the combustion units or their operation and inherent design and construction differences in similar units placed in the same category, as well as random inconsistencies in the sampling or analytical techniques. The results cited are considered minimum uncertainty or variation, because they do not include allowance for fuel variations, equipment or process variations, or maintenance practices between combustion systems.
The use of the emission factors found from this study and values published in the EPA AP-42 guidelines31 take into account measurement reproducibility and measurement bias, along with operational deviations in time from a single unit expected in the source classifications examined. In contrast, the published literature does not generally account for across or interunit variability of different combustors and their condition, which are present in the real world situations. Without any extensive background knowledge, it is often assumed that combustors in a given SCC category are equivalent in emissions without consideration for the influence of maintenance and fuel variations, as well as process modifications, and so forth. Because testing of a representative number of units within SCC codes used in the United States, for example, is expensive, an average emission estimate based on a statistical model has not been achieved for most combustion systems within or across SCC categories. The exception to this situation is the extensive, sustained collection of data for gaseous emission rates from large utility boilers equipped with continuous monitoring instruments. The methodology for addressing the issue of representativeness within or across SCC codes has yet to be worked out (e.g., NARSTO44).
Applications in Air Quality Management
Although there are limitations to this study, the results represent an important contribution to knowledge about stack effluents from gas-fired combustors used in the United States today. The first part of this project determined that the results reported for PM^sub 2.5^ emissions from stationary sources are method dependent in that standard certification methods give two different estimates of PM^sub 2.5^ emissions, which include FPM and CPM. The sum of these two estimates is not necessarily that characterizing a source after the stack effluent has mixed and cooled with ambient air. Dilution sampling is an alternate estimate, which accounts for cooling and mixing in "chemically" clean air; this estimate of PM is said to be closer to the emissions as they mix with ambient air than the other two standard methods.
Dilution sampler methodology already used in sampling mobile sources for compliance provides a credible, self-consistent sampling approach by simulating the entry of hot stack gases containing particles and condensable material into air at ambient temperature conditions for a range of sources important to air quality. Furthermore, the method allows for the application of the same sampling and analytical methods to characterize PM^sub 2.5^ chemistry as adopted for ambient air sampling.
The data obtained for PM^sub 2.5^ emissions discussed in the companion paper3 indicate that the FPM estimates from in-stack methods are systematically lower than those reported for dilution sampling. In contrast, the sum of FPM and CPM from in-stack methods yields much larger apparent emission values relative to dilution sampling for the sources tested in this program. The most likely explanation3 for this apparent bias in the in-stack method for CPM results (by which the sample is collected in iced impingers charged with water as a solvent) is measurement artifacts that convert some of the dissolved noncondensable acid gases and VOCs to residues that are counted as CPM.45-49 Although EPA's implementation of PM10 National Ambient Air Quality Standards includes CPM in its definition of PM^sub 10^ emissions, many of the state regulations do not require measurement of CPM in their definition of PM^sub 10^ emissions enforcement, because EPA has not consistently implemented its guidance on this issue.4 As a result of recent EPA guidance for emissions reporting, several of the states that did not previously require CPM measurements recently began adding CPM test requirements to attempt to improve PM^sub 2.5^ emission inventories.
The use of FPM measurements for PM^sub 10^ emissions, extrapolated to PM^sub 2.5^ primary emissions for facility permitting or for regulatory analysis, would underestimate emissions rates compared with rates relative to dilution sampling results. The method dependence of emission estimates presents an important issue to using such data for air quality management practice. Regulatory guidance is needed for states to use the dilution-based data versus the sum of FPM and CPM for emissions characterization. There is a parallel need for regulatory leadership to bring consistency to the emission sampling and analysis methodology for mobile and stationary sources consistent for compliance assessment.
The mean values of emission and control factors found using dilution sampling in this study are useful updates of the EPA AP-42 guidelines, and the uncertainties stated in the study will assist in estimating uncertainties in stationary source emission inventories developed from such data. In keeping with recommendations of recent emission inventory assessments,44 this study has attempted to address at least a contemporary range of gasand oil-fired stationary sources, with improved knowledge about uncertainties associated with the mean emission factors. This and other recent work on emissions characterization should be used preferentially to establish the quality of or the updates of default emission factor data for corresponding source classification in AP-42. Efforts are needed to systematically expand such measurements to evaluate high uncertainty source categories, which will enhance the opportunities of ai\r quality management programs to optimize emission reduction strategies.
The source-composition profiles obtained from this study indicate the relative importance of particulate carbon emissions in fossil fuel combustors (although the absolute values of particulate carbon emissions from the gas-fired sources are extremely low relative to diesel, fuel oil, and coal combustors). As sulfate and nitrate in particles decline with implementation of NO^sub x^ and SO^sub 2^ emission reduction strategies, carbon components in emissions from fires (wildfires, residential wood combustion, open burning, etc.), mobile sources (on-road and off-road), and certain types of stationary sources will become an increasingly the major source of PM^sub 2.5^ in the ambient air.
CONCLUSIONS AND RECOMMENDATIONS
The following conclusions derive from this study. First, PM^sub 2.5^ emission rates for gas-fired units based on dilution sampling are very low, probably near ambient concentration levels in many cases. These levels are a challenge to quantify with current samplers and sampling protocols. Second, PM^sub 2.5^ emission factors derived for a boiler using No. 6 fuel oil and for a diesel engine back-up generator were 1 order of magnitude larger than those found for gas-fired sources. Third, analyses of filter samples from the dilution sampler to characterize the chemical composition of PM^sub 2.5^ indicate that most of the material from gasand oil- fired units evaluated is carbonaceous, with substantial amounts of OC evidently present. Sulfate and NO^sup 3-^ also are present in measurable amounts; SO^sup 4-^ varies approximately linearly with fuel sulfur content. Metals are present in trace amounts in the particles, except for the case of diesel engine PM^sub 2.5^, which did not have a detectable metal content (most likely a result of much smaller stack gas sample volumes and consequently higher in- stack detection limits for the diesel engine tests compared with other tests). Fourth, a substantial fraction of the OC found in particles characterized using a dilution sampler, especially the gas- fired units, appears to be volatile in character so that the estimates of OC in the particles characterized the dilution sampler protocol or standard reference methods is highly uncertain. Fifth, particulate composition (source) profiles or profiles based on speciation of organic vapors from various sources could be used as the basis for future studies but did not add new source markers for receptor modeling of gas-fired combustors. Lastly, uncertainties in the emission factor estimates are characterized for the sample combustors in the study using formal statistical measures for each unit and consequent estimates of the (relative) accuracy and precision of the dilution sampler method. These are useful as minimum uncertainty statements but do not include the variability in fuel use design, location, seasonal variability, or other process variation among sources in the classifications studied.
The following are recommendations are derived from the study. First, further investigation of the methodology as applied to sampling carbonaceous material, especially OC and its vapor counterparts, is recommended, because this component is the principal material in the particulate emissions from gas- and oil- fired combustors. Second, the investment in systematic, self- consistent source characterization measurements is substantial but enables investigators to obtain emissions rate data for use in inventory development with much greater confidence than in using engineering calculations or outdated, pre-1990s literature. The methodology should be considered for application to other sources that appear to be high-priority contributors to PM^sub 2.5^ concentrations in ambient air. Third, scientifi- cally sound guidance concerning interpretation and use of dilution sampler- based and hot filter/ice impingerbased emissions data for PM^sub 2.5^ mass emission rates is needed to support air quality management objectives. Lastly, efforts to develop a consensus among stakeholders is warranted to unify methodologies for characterizing stationary and mobile sources.
This work was sponsored jointly by the American Petroleum Institute (Contract No. 00-0000-4303), the California Energy Commission (CEC), the U.S. Department of Energy (Contract No. DE-FC- 26-00BC15327), the New York Energy Research and Development Authority (NYSERDA), and the Gas Research Institute (Contract No. 8362). The unpublished test reports for the combustors and the final reports for the study are available from NYSERDA [http:// www.nyserda.org/programs/Environment/ EMEP/finalreports.asp (listed under "Air Quality and Related Health Research: Particulate Matter (PM), Ozone and Co-Pollutants)]. The reports can also be obtainedfromtheCEC( http://www.energy.ca.gov/pier/final_ project_reports/CEC-500-2005-032_to_44.html) or from American Petroleum Institute. We are indebted to a number of investigators who participated in the program, including: Stephanie Wien, Lynn Hildemann, James Schauer, Phillip Hopke, Glen Cass, Praveen Amar, Ron Myers, Thomas Logan, and Karen Magliano. We are also indebted to representatives of sponsors for support of this work, including Barry Liebowitz, Janet Joseph, Karin Ritter, Marla Mueller, Kathy Stirling, and Paul Drayton.
With the establishment of a National Ambient Air Quality Standard for PM^sub 2.5^, source emissions data for these fine particles are found to be seriously limited. This work responds to the need to add to the quantitative knowledge of PM^sub 2.5^ emissions and provide chemical composition of the particles for gas-fired combustion source categories. The comparison with firing with other fuels illustrates how low emissions from gas-fired systems are relative to other fossil fuel combustors. The compact dilution sampler facilitates the efficient, self-consistent sampling for stationary source emissions.
1. Measurement of PM2.5 and PM10 by Dilution Sampling (Constant Sampling Rate Procedures). Report CTM-039; U.S. Environmental Protection Agency: Research Triangle Park, NC, 2003.
2. England, G.C. Development of Fine Particulate Emission Factors and Speciation Profiles for Oil- and Gas-Fired Combustion Systems. Report No. 04-05; New York Energy Research and Development Authority: Albany, NY, 2004.
3. England, G.C.; Watson, J.G.; Chow, J.C.; Zielinska, B.; Chang, M.O.; Loos, K.; Hidy G.M. Dilution-Based Emissions Sampling from Stationary Sources: Part 1-Compact Sampler Methodology and Performance; J. Air & Waste Manage. Assoc. 2007, 57, 65-78.
4. Chang, M.-C.O.; Chow, J.C.; Watson, J.G.; Hopke, P.K.; Yi, S- M.; England, G.C. Measurement of Ultrafine Particle Size Distributions from Coal, Oil and Gas-Fired Stationary Combustion Sources; J. Air Waste & Manage. Assoc. 2004, 54, 1494-1505.
5. Chang, M.-C.; England, G.C. Development of Fine Particulate Emission Factors and Speciation Profiles for Oil and Gas-Fired Combustion Systems, Update: Critical Review of Source Sampling and Particulate Source Emission Profiles. Prepared for U.S. Department of Energy, National Energy Technology Laboratory: Pittsburgh, PA; Gas Research Institute: Des Plains, IL; American Petroleum Institute: Washington, DC, 2004.
6. Myers, R.E.; Logan, T. Progress on Developing a Federal Reference PM Fine Source Test Method; Unpublished Report; U.S. Environmental Protection Agency: Research Triangle Park, NC, 2003; available at: www. epa.gov/ttnchiel/conference/eill/pm/invers.pdf (accessed 2004).
7. Dilution Test Method for Determining PM2.5 and PM10 Mass in Stack Gases; Proposed Method D22.03-W1752; American Society for Testing and Materials; West Conshohocken, PA, 2005.
8. Chow, J.C.; Watson, J.G.; Kuhns, H.; Etyemezian, V.; Lowenthal, D.; Crow, D.; Kohl, S.; Engelbrecht, J.; Green, M. Source Profiles for Industrial, Mobile and Area Sources in the Big Bend Regional Aerosol Visibility and Observational Study; Chemosphere 2004, 54, 185-208.
9. Hildemann, L.; Cass, G.; Markowski, G. A. Dilution Stack Sampler for Organic Aerosol Emissions: Designs, Characterization, and Field Tests; Aerosol Sci. Technol. 1989, 10, 193-204.
10. Chang, M.-C.O.; England, G.C. Development of Fine Particulate Emission Factors and Speciation Profiles for Oil and Gas-Fired Combustion Systems; Other Report: Pilot Scale Dilution Sampler Design and Validation Tests (Laboratory Study). Unpublished Report prepared for the U.S. Department of Energy, National Energy Technology Laboratory: Pittsburgh, PA; The Gas Research Institute: Des Plains, IL; and the American Petroleum Institute: Washington, DC, 2004.
11. American Petroleum Institute. Gas Fired Boiler-Test Report Site A: Characterization of Fine Particulate Emission Factors and Speciation Profiles from Stationary Petroleum Industry Combustion Sources; Publication No. 4703; American Petroleum Institute: Washington, DC, 2001.
12. American Petroleum Institute. Gas Fired Heater-Test Report Site B: Characterization of Fine Particulate Emissions and Speciation Profiles from Stationary Petroleum Industry Combustion Sources; Publication No. 4704; American Petroleum Institute: Washington, DC, 2001.
13. American Petroleum Institute. Gas-Fired Steam Generator-Test Report Site C: Characterization of Fine Particulate Emission Factors and Speciation Profiles from Stationary Petroleum Industry Sources; Publication No. 4712; American Petroleum Institute: Washington, DC, 2001.
14. Wien, S.; England, G.; Chang, M. Development of Fine Particulate Emission Factors and Speciation Profiles for Oil and Gas Fired Combustion Systems. Topical Report: Test Results for a Gas- Fired Process Heater (Site Alpha); Prepared for the U.S. Department of Energy, National Energy Technology Laboratory: Pittsburgh, PA; the Gas Research Institute: Des Plains, IL; and the American Petroleum Institute: Washington, DC, 2003.
15. Wien, S., En\gland, G.; Chang, M. Development of Fine Particulate Emission Factors and Speciation Profiles for Oil and Gas Fired Combustion Systems. Topical Report: Test Results for a Combined Cycle Power Plant with Supplementary Firing, Oxidation Catalyst and SCR at Site Bravo; Prepared for the U.S. Department of Energy, National Energy Technology Laboratory: Pittsburgh, PA; the Gas Research Institute: Des Plains, IL; and the American Petroleum Institute: Washington, DC, 2004.
16. Wien, S.; England, G.; Chang, M. Development of Fine Particulate Emission Factors and Speciation Profiles for Oil and Gas Fired Combustion Systems. Topical Report: Test Results for a Gas- Fired Process Heater with Selective Catalytic Reduction (Site Charlie); Prepared for the U.S. Department of Energy, National Energy Technology Laboratory: Pittsburgh, PA; the Gas Research Institute: Des Plains, IL; and the American Petroleum Institute: Washington, DC, 2004.
17. Wien, S.; England, G.; Chang, M. Development of Fine Particulate Emission Factors and Speciation Profiles for Oil and Gas Fired Combustion Systems. Topical Report: Test Results for a Dual Fuel-Fired Commercial Boiler at Site Delta; Prepared for the U.S. Department of Energy, National Energy Technology Laboratory: Pittsburgh, PA; the Gas Research Institute: Des Plains, IL; and the American Petroleum Institute: Washington, DC, 2004.
18. England, G.C.; Wien, S.; McGrath, T.; Hernandez, D. Development of Fine Particulate Emission Factors and Speciation Profiles for Oil and Gas Fired Combustion Systems. Topical Report: Test Results for a Combined Cycle Power Plant with Oxidation Catalyst and SCR at Site Echo; Prepared for the U.S. Department of Energy, National Energy Technology Laboratory: Pittsburgh, PA; the Gas Research Institute: Des Plains, IL; and the American Petroleum Institute: Washington, DC, 2004.
19. Hernandez, D.; Nguyen, Q.; England, G.C. Development of Fine Particulate Emission Factors and Speciation Profiles for Oil and Gas Fired Combustion Systems. Topical Report: Test Results for a Diesel- Fired Compression Ignition Reciprocating Engine with a Diesel Particulate Filter at Site Foxtrot; Prepared for the U.S. Department of Energy, National Energy Technology Laboratory: Pittsburgh, PA; the Gas Research Institute: Des Plains, IL; and the American Petroleum Institute: Washington, DC, 2004.
20. England, G. C.; McGrath, T. Development of Fine Particulate Emission Factors and Speciation Profiles for Oil and Gas-Fired Combustion Systems, Topical report: Test Results for a Cogeneration Plant with Supplemental Firing, Oxidation Catalyst and SCR at Site Golf; Prepared for the U.S. Department of Energy, National Energy Technology Laboratory: Pittsburgh, PA; the Gas Research Institute: Des Plains, IL; and the American Petroleum Institute: Washington, DC, 2004.
21. U.S. Environmental Protection Agency. Method 5-Determination of Particulate Matter Emissions from Stationary Sources. In Code of Federal Regulations, Title 40, Part 60, Appendix A, 1996; available at: http://www.epa.gov/ttn/emc/progate.html (accessed 2004).
22. Chow, J.C. Critical Review: Measurement Methods to Determine Compliance with Ambient Air Quality Standards for Suspended Particles; J. Air & Waste Manage. Assoc. 1995, 45, 320-382.
23. Chow, J.C.; Watson, J.G. Guideline on Speciated Particulate Monitoring. Unpublished Report; Prepared for the U.S. Environmental Protection Agency, Desert Research Institute: Reno, NV, 1998.
24. Watson, J.G.; Chow, J.C.; Frazier, C.A. X-Ray Fluorescence Analysis of Ambient Air Samplers. In Elemental Analysis of Airborne Particles, Vol. 1, Landsberger, S.; Creatchman, M. Eds.; Gordon and Breach Science: Amsterdam, the Netherlands, 1999; pp 67-96.
25. Chow, J.C. Watson, J.G.; Pritchett, L.C.; Pierson, W.R.; Frazier, C.A.; Purcell, R.G. The DRI Thermal/Optical Reflectance Carbon Analysis System: Description, Evaluation, and Applications in U.S. Air Quality Studies; Atmos. Environ. 1993, 27A, 1185-1201.
26. Chow, J.C.; Watson, J.G.; Crow, D. Lowenthal, D.H.; Merrifield, T. M. Comparison of IMPROVE and NIOSH Carbon Measurements; Aerosol Sci. Technol. 2001, 34, 23-34.
27. Chow, J.C.; Watson, J.G. Ion Chromatography in Elemental Analysis of Airborne Particles. In Elemental Analysis of Airborne Particles, Vol. 1; Landsberger, S.; Creatchman, M., Eds., Gordon and Breach: Amsterdam, the Netherlands, 1999; pp 97-137.
28. Chow, J.C.; Watson, J.G.; Chen, L.-W.A.; Arnott, W.P.; Moosmuller, H.; Fung, K.K. Equivalence of Elemental Carbon by Thermal/Reflectance and Transmittance with Different Temperature Protocols; Environ. Sci. Technol. 2004, 38, 4414-4422.
29. Zielinska, B.; Sagebiel J.C.; Harshfield, G.; Pasek, R. Volatile Organic Compound Measurements in the California/Mexico Border Region during SCOS97; Sci. Total Environ. 2001, 276, 19-32.
30. Zielinska, B.; Sagebiel, J.C.; Harshfield, G.; Gertler, A.; Pierson, W.R. Volatile Organic Compounds up to C20 Emitted from Motor Vehicles: Measurement Methods; Atmos. Environ. 1996, 30, 2269- 2286.
31. U.S. Environmental Protection Agency. Compilation of Air Pollution Emission Factors. Report AP-42. Internet Edition. Office of Research and Development: Research Triangle Park, NC, 2004; available at: http://www.ehso.com/Air_AP42.htm (accessed 2004).
32. U.S. Environmental Protection Agency. SPECIATE: EPA's Repository of Total Organic Compound and Particulate Matter Speciated Profiles for a Variety of Sources for Use in Source Apportionment Studies; Office of Air Quality Planning and Standards: Research Triangle Park, NC, 1994; available at: http://www.epa.gov/ ttn/chief/software/speciate/ (accessed 2004).
33. Lipsky, E.; Pekney, N.J.; Walbert, G.F; O'Dowd, W.J.; Freeman, M.C.; Robinson, A. Effects of Dilution Sampling on Fine Particle Emissions from Pulverized Coal Combustion; Aerosol Sci. Technol. 2004, 38, 574-587.
34. Grosjean, D.; Seinfeld, J. Parameterization of the Formation Potential of Secondary Organic Aerosols; Atmos. Environ. 1989, 23, 1733-1747.
35. Seinfeld, J.; Pandis, S. Atmospheric Chemistry and Physics from Air Pollution to Climate Change. John Wiley & Sons: New York, NY, 1998.
36. Office of Air Quality Planning and Standards. Compilation of Air Pollutant Emission Factors, Volume 1-Stationary Point and Area Sources. Supplement F. Report AP-42, 5th ed.; U.S. Environmental Protection Agency: Research Triangle Park, NC, 2000.
37. Hildemann, L.M.; Klinedinst, D.B.; Klouda, G.A.; Ciurrie, G.R. Chemical Composition of Emissions from Urban Sources of Fine Organic Aerosol; Environ. Sci. Technol. 1991, 24, 744-759.
38. Labban, R.; Veranth, J.M.; Chow, J.C.; Engelbrecht, J.L.; Watson, J.G. Size and Geographical Variation in PM1, PM^sub 2.5^ and PM^sub 10^: Source Profiles from Soils in the Western United States; Water Air Soil Pollut. 2004, 157, 13-71.
39. Zielinska, B.; McDonald, J.; Hayes, T.; Chow, J.; Fujita; E.; Watson, J. Northern Front Range Air Quality Study. Final Report, Volume B: Source Measurements; Prepared for the Office of the Vice President for Research and Information Technology, Colorado State University: Fort Collins, CO; Desert Research Institute: Reno, NV, 1998.
40. Watson, J.; Chow, J.; Houck, J. PM^sub 2.5^ Chemical Source Profiles for Vehicle Exhaust, Vegetative Burning, Geological Material and Coal Burning in Northwestern Colorado during 1995; Chemosphere 2001, 43, 1141-1151.
41. Chow, J.C.; Watson, J.G.; Kuhns, H.D.; Etymezian, V.; Lowenthal, D.H.; Crow, D.J.; Kohl, S.D.; Engelbrecht, J.P.; Green, M.C. Source Profiles for Industrial, Mobile and Area Sources in the Big Bend Regional Aerosol Visibility and Observational (BRAVO) Study; Chemisphere 2004, 54, 185-208.
42. Watson, J.G.; Chow, J.C.; Lowenthal, D.H.; Pritchett, L.C.; Frazier, C.A.; Neuroth, G.R.; Robbins, R. Differences in the Carbon Composition of Source Profiles for Diesel- and Gasoline-Powered Vehicles; Atmos. Environ. 1994, 28, 2493-2505.
43. Zielinska, B.; Sagebiel, J.; Whitney, K.; Lawson, D. Emission Rates and Comparative Chemical Composition from Selected In-Use Diesel and Gasoline-Fueled Vehicles; J. Air & Waste Manage. Assoc. 2004, 54, 1138-1150.
44. NARSTO. Improving Emission Inventories for Effective Air Quality Management across North America; Report NARSTO-05-001; NARSTO: Kennewick, WA, 2005.
45. Corio, L.A.; Sherwell, J. In-Stack Condensible Particulate Matter Measurements and Issues; J. Air & Waste Manage. Assoc. 2000, 50, 207-218.
46. Dewees, W.G.; Steinberger, S.C.; Plummer, G.M.; Lay, L.T.; McAlister, G.D.; Shigehara, R.T. Laboratory and Field Evaluation of the EPA Method 5 Impinger Catch for Measuring Condensable Matter from Stationary Sources; In Proceedings of the EPA/A&WMA Symposium on Measurement of Toxic and Related Air Pollutants. Publ. VIP-13 (EPA Report No. 600/89-060); A&WMA: Pittsburgh, PA, 1989.
47. Filadelfia, E.J.; McDannel, M.D. Evaluation of False Positive Interferences Associated with the Use of the EPA Method 202; In Proceedings of the 89th Annual Conference and Exhibition, Nashville, TN; A&WMA: Pittsburgh, PA, 1996.
48. Holder, T.E.; Goshaw, D.G.; Richards, J.R. Artifact Formation in Method 202 Sampling Trains Used to Measure Condensible Particulate Matter Emissions from Portland Cement Kilns; Paper 451; In Proceedings of the 94th Annual Conference & Exhibition, Orlando, FL; A&WMA: Pittsburgh, PA, 2001.
49. Wien, S.E.; England, G.C.; Loos, K.R.; Ritter, K. Investigation of Artifacts in Condensible Particulate Measurements for Stationary Combustion Sources; Paper 536; In Proceedings of the 94th Annual Conference and Exhibition, Orlando, FL; A&WMA: Pittsburgh, PA, 2001.
Glenn C. England
GE Energy, Santa Ana, CA
John G. Watson, Judith C. Chow, and Barbara Zielinska
Desert Research Institute, Reno, NV
M.-C. Oliver Chang
California Air Resources Board, El Monte, CA
Karl R. Loos
Shell Global Solutions, Houston, TX
George M. Hidy
Envair/Aerochem, Placitas, NM
About the Authors
Glenn England is manager o\f air quality technology with GE Energy's Environmental Services business unit. John Watson, Judith Chow, and Barbara Zielinska are research professors at the Desert Research Institute. M.-C. Oliver Chang is a research scientist at the California Air Resources Board. Karl Loos is recently retired as science advisor for Shell Global Solutions, and George Hidy is a principal of Envair/Aerochem. Address correspondence to: George Hidy, 6 Evergreen Dr., Placitas, NM 87043; phone: +1-505- 771-4083; fax: +1-505-771-4083; e-mail: dhidy113@ comcast.net.
Copyright Air and Waste Management Association Jan 2007
(c) 2007 Journal of the Air & Waste Management Association. Provided by ProQuest Information and Learning. All rights Reserved.