Ramboll Environ
July 8, 2016
Executive Summary
In 2012, the Canadian Energy Partnership for Environmental Innovation (CEPEI) published a technical memorandum1 providing alternative PM2.5 (particles with aerodynamic diameter of 2.5 micrometers and smaller) air emission factors and species profiles for natural gas-fired boilers, process heaters, a diesel engine and gas turbine combined cycle/cogeneration power plants. This document provides updated emission factors for natural gas-fired gas turbines/combined cycle/cogeneration units and new emission factors for natural gas-fired spark-ignited reciprocating engines.
Previously reported PM2.5 emission factors and species profiles2 are based on tests conducted in the United States (U.S.) from 1998 to 2003 during an industry-government collaboration led by GE Energy and Environmental Research Corporation (GE EER) using a dilution sampling methodology. Dilution sampling is thought to provide more accurate measurements of PM2.5 from gas combustion than traditional hot filter/cooled impinger test methods. The California Energy Commission, New York State Energy and Research Development Authority and U.S. Department of Energy co-sponsored the work along with the American Petroleum Institute (API) and Gas Research Institute. The U.S. Environmental Protection Agency (U.S. EPA) served as external peer reviewers during the study, contributing to the study design and results review, eventually adopting PM2.5 and PM10 emission factors for natural gas combustion derived from the results for use in its tri-annual air pollutant National Emission Inventories (NEIs). Those tests included two natural gas-fired heavy duty gas turbine combined cycle power generation units with lean premix combustors and a refinery gas-fired aeroderivative gas turbine cogeneration system, all three with supplementary firing capability and with post-combustion emission controls (oxidation catalyst and selective catalytic reduction, SCR). No natural gas-fired reciprocating engines were included in those collaborative tests. API separately sponsored tests of three natural gas-fired spark-ignited reciprocating engines operating as natural gas production compressor drives in 20033. In 2008, GE Energy subsequently conducted method evaluation tests on a natural gas-fired combined cycle power generation unit with SCR using a similar dilution methodology based on modified U.S. EPA Conditional Test Method 39 (CTM 39), with external peer review participation from U.S. EPA, California Air Resources Board and the South Coast Air Quality Management District. In 2014, the Utah Department of Environmental Quality (in consultation with U.S. EPA) approved PM10 tests using modified CTM 39 to demonstrate compliance with PM10 emission limits based on U.S. EPA’s NEI PM10 emission factor. Thus, modified CTM 39 has been applied with consent of regulatory agencies for PM2.5/10 emission factor development and for regulatory compliance demonstration.
In 2015, CEPEI and the Petroleum Technology Alliance of Canada (PTAC), including the British Columbia Oil and Gas Research and Innovation Society (BC OGRIS), sponsored new tests on two natural gas-fired engines in Canada: a gas turbine engine and a spark-ignited reciprocating engine, both operating as natural gas pipeline compressor drives and with no post-combustion controls. The engines are typical of Canadian natural gas pipeline engines in terms of size, configuration, emission controls and operation. The objective of the tests was to provide data for developing updated PM2.5 emission factors and species profiles representative of engines in Canada without post-combustion controls applicable to upstream and downstream oil and gas operations and natural gas end users. The tests were conducted using a stationary source dilution sampling method combined with ambient air sample collection and analysis methods to determine both the mass and chemical speciation of combined filterable plus condensable PM2.5 emissions. The methodology is similar to that used for the earlier GE EER test program. The chemical composition of the collected aerosols also was determined (elements, selected ions and organic and elemental carbon). Detailed test results obtained in this study are provided in a separate Test Report.
Updated PM2.5 emission factors and species profiles were derived from the CEPEI test results and data from the earlier tests noted above. PM2.5 mass emission factors for gas turbines, gas turbine combined cycle/cogeneration units and for four-stroke reciprocating internal combustion engines, expressed as kilograms of PM2.5 per gigajoule of fuel heat input (kg/GJ), are provided in Table E-1. The maximum and 95% confidence upper bound provide an indication of the upper limits of the data set. The 99% confidence upper prediction limit provides an indication of an upper limit for the average for the next unit tested.
U.S. EPA’s Compilation of Air Pollutant Emission Factors (“AP-42”) is a widely-referenced source of emission factors. The published AP-42 filterable and condensable particulate matter emission factors for natural gas-fired gas turbine and four-stroke reciprocating engines, also shown in Table E-1, are based on tests of three gas turbines4, two four-stroke lean burn and three four-stroke rich burn reciprocating engines.
The average CEPEI PM2.5 emission factor of 0.000101 kg/GJ for gas-fired gas turbines and cogeneration/combined cycle units based on dilution sampling methods is 1/28 (3.5%) of the combined AP-42 gas turbine emission factor for filterable and condensable particulate matter (0.00285 kg/GJ).
The average emission factor of 0.00150 kg/GJ for four-stroke reciprocating engines is 1/6 (16%) of the combined filterable and condensable particulate matter emission factor for all four-stroke engines derived from the AP-42 data set (0.00673 kg/GJ). Further, there are no condensable particulate matter test data for four-stroke rich burn engines in the AP-42 data set; the condensable particulate matter emission factor reported in AP-42 for four-stroke rich burn engines is based on two four-stroke lean burn engine tests.
Although AP-42 does not report uncertainty associated with the emission factors, the total particulate matter emission factor uncertainty calculated from the underlying data is 85% for the gas turbine and 270% and 438% for four-stroke rich burn and lean burn engines, respectively. The very large uncertainties for the AP-42 four-stroke engine emission factors are due to both the wide range of emissions among the units and the small number of units tested. Although the CEPEI and AP-42 data sets are similar in size, the improved precision of measurements in CEPEI’s data set results in lower uncertainties. Although the data sets are small in both cases, we consider the CEPEI emission factors more robust than the AP-42 factors because of much lower uncertainty (in terms of both relative percent and absolute magnitude).
Average PM2.5 chemical species are measured primarily as organic carbon, with minor amounts of sulfate, ammonium, elemental carbon, chloride, nitrate and other elements. Iron and silica were more prevalent in PM2.5 from the reciprocating engines than the gas turbines and combined cycle/cogeneration units. Sulfate and ammonium were not detected in the samples from the CEPEI 2015 test program. This likely reflects low natural gas sulfur content and the absence of postcombustion catalysts (e.g., selective catalytic reduction5 or CO oxidation catalysts) on this unit, as compared with the refinery gas and natural gas-fired units tested previously in the U.S which did have post-combustion catalysts.
The CEPEI PM2.5 emission factor for gas turbines and combined cycle/cogeneration units is based on six tests of five units6 including one unit firing refinery gas and four units firing natural gas. This includes simple and combined cycle/cogeneration units with and without post-combustion catalysts for NOX and CO emissions reduction. In contrast, the AP-42 PM emission factors include data for five tests of three natural gas-fired gas turbines with water injection (for NOX emissions control) but without post-combustion catalysts.
The CEPEI PM2.5 emission factor for four-stroke reciprocating engines is based on tests of three units: one four-stroke rich burn engine with non-selective catalytic NOX reduction and two four-stroke lean burn engines with no post-combustion emission controls. The number of units tested is comparable with the number of units included in the AP-42 data sets. The PM2.5 emission data in the CEPEI data set for the rich burn engine is approximately four times greater than the PM2.5 emission data for the two lean burn engines. The small number of units and range of PM2.5 emissions contribute to large relative uncertainty – 186% – in the average CEPEI PM2.5 emission factor. Nevertheless, the uncertainty associated with the CEPEI PM2.5 emission factor is considerably lower than the uncertainties for the AP-42 filterable and condensable particulate matter (and by summation, total PM) emission factors, as noted above. Previous studies showed that the dilution sampling test methodology on which the CEPEI PM2.5 emission factor is based is more accurate and precise than the hot filter/cooled impinger test methods used for the AP-42 filterable and condensable particulate matter emission factors. Therefore, the CEPEI PM2.5 emission factor for natural gas-fired four-stroke engines is considered more robust than the respective AP-42 emission factors.
As a general precaution, an average or median emission factor should not be used to establish emissions limits or standards because emissions from half of the units will be higher than the average and half will be lower (assuming a normal distribution). However, an average or median emission factor is appropriate to estimate average emissions from a population of similar units. Additional testing of natural gas-fired gas turbines and/or combined cycle/cogeneration units over time could further reduce uncertainty and improve emission factor quality.
# 15-ARPC-09