API PUBL 4703:2001

In 1997, the United States Environmental Protection Agency (EPA) promulgated new ambient air standards for particulate matter smaller than 2.5 micrometers in diameter (PM2.5). Source emissions data are needed to assess the contribution of petroleum industry combustion sources to ambient PM2.5 concentrations for receptor modeling and PM2.5 standard attainment strategy development. There are few existing data on emissions and characteristics of fine aerosols from petroleum industry combustion sources, and the limited information that is available is incomplete and outdated. The American Petroleum Institute (API) developed a test protocol to address this data gap, specifically to: -- Develop emission factors and speciation profiles for emissions of primary fine particulate matter (i.e., particulate present in the stack flue gas including condensible aerosols), especially organic aerosols from gas-fired combustion devices; and -- Identify and characterize secondary particulate (i.e., particulate formed via reaction of stack emissions in the atmosphere) precursor emissions. This report presents results of a pilot project to evaluate the test protocol on 550,000 pounds per hour steam (approximately 650 · 106 British thermal units per hour) boiler firing refinery process gas. The tangentially fired boiler has a waterwall furnace with two rows of burners in each corner of the furnace. The unit has no controls for NOx emissions. The boiler operated at approximately 57 percent of capacity, and flue gas temperature at the stack was approximately 345 degrees Fahrenheit during the tests. The particulate measurements at the stack were made using both a dilution tunnel research test method and traditional methods for regulatory enforcement of particulate regulations. The dilution tunnel method is attractive because the sample collection media and analysis methods are identical to those used for ambient air sampling. Thus, the results are directly comparable with ambient air data. Also, the dilution tunnel method is believed to provide representative results for condensible aerosols. Regulatory methods are attractive because they are readily accepted by regulatory agencies and have been used extensively on a wide variety of source types; existing regulatory methods for condensible aerosols may be subject to significant bias, however, and sampling/analytical options are limited. Emission factors for all species measured are extremely low, which is expected for gas-fired sources. Emission factors for primary particulate, including: total particulate, PM10 (particles smaller than nominally 10 micrometers), and PM2.5; elements; ionic species; and organic and elemental carbon are presented in Table E-1. Since the boiler was firing refinery process gas with a heating value different from natural gas, emission factors are expressed in pounds of pollutant per million British thermal units of gas fired (lb/MMBtu). All tests were performed in triplicate. As a measure of the bias, precision, and variability of the results, the uncertainty and 95 percent confidence upper bound also are presented. Emission factors for semivolatile organic species are presented in Table E-2. The sum of semivolatile organic species is approximately three percent of the organic carbon. Emission factors for secondary particulate precursors (NOx SO2, and volatile organic species with carbon number of 7 or greater) are presented in Table E-3. The preceding tables include only those substances that were detected in at least one of the three test runs. Substances of interest that were not present above the minimum detection limit for these tests are listed in Table E-4. A single ambient air sample was collected at the site. In some cases, the emission factors reported in Tables E-1 to E-3 resulted from in-stack concentrations that were near ambient air concentrations. Those species concentrations that are within a factor of 10 of the ambient air concentration are indicated on the table by an asterisk (*). The primary particulate results presented in Table E-1 also may be expressed as a PM2.5 speciation profile, which is the mass fraction of each species contributing to the total PM2.5 mass. The speciation profile is presented in Figure E-1. The main findings of these tests are: -- Particulate mass emissions from the boiler were extremely low, consistent with levels expected for gaseous fuel combustion. -- Two methods for determining the average emission factor for primary PM2.5 mass gave results which differed in magnitude by a factor of 27:0.000358 lb/MMBtu using the dilution tunnel; and 0.00974 lb/MMBtu using conventional in-stack methods for filterable and condensible particulate. -- Sampling and analytical artifacts principally caused by gaseous SO2 in the stack gas were shown to produce a relatively large positive bias in condensible particulate as measured by conventional in-stack methods. Most of the difference between the dilution tunnel and conventional method results can be explained by these measurement artifacts. The results using conventional EPA methods are nominally consistent with published EPA emission factors for external combustion of natural gas (U. S. EPA, 1998). Therefore, the published EPA emission factors derived from tests using similar measurement methods also may be positively biased. -- Chemical species accounting for 74 percent of the measured PM2.5 mass were quantified. -- Organic and elemental carbon comprise 68 percent of the measured primary PM2.5 mass. -- Sulfates, iron, copper, chloride, and smaller amounts of other elements account for another 6 percent of the measured PM2.5 mass. -- Less than 26 percent of the measured PM2.5 mass is unspeciated. -- Most elements are not present at levels significantly above the background levels in the ambient air or the minimum detection limits of the test methods. -- Most organic species are not detected at levels significantly above background levels in the ambient air or field blanks. All detected organics are present at extremely low levels consistent with gaseous fuel combustion. -- Emissions of secondary particle precursors are low and consistent with levels expected for gaseous fuel combustion.