Article Cite This: Environ. Sci. Technol. 2019, 53, 8976−8984
pubs.acs.org/est
Observations of Methane Emissions from Natural Gas-Fired Power Plants Kristian D. Hajny,† Olivia E. Salmon,†,¶ Joseph Rudek,‡ David R. Lyon,‡ Andrew A. Stuff,† Brian H. Stirm,§ Robert Kaeser,† Cody R. Floerchinger,∥ Stephen Conley,⊥ Mackenzie L. Smith,⊥ and Paul B. Shepson*,†,# †
Purdue University, Department of Chemistry, West Lafayette, Indiana 47907, United States Environmental Defense Fund, Austin, Texas 78701, United States § Purdue University, School of Aviation and Transportation Technology, West Lafayette, Indiana 47906, United States ∥ Harvard University, Department of Earth and Planetary Sciences, Cambridge, Massachusetts 02138, United States ⊥ Scientific Aviation, Inc., Boulder, Colorado 80301, United States # Stony Brook University, School of Marine and Atmospheric Sciences, Stony Brook, New York 11794, United States
Downloaded via RUTGERS UNIV on August 7, 2019 at 00:26:27 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
‡
S Supporting Information *
ABSTRACT: Current research efforts on the atmospheric impacts of natural gas (NG) have focused heavily on the production, storage/transmission, and processing sectors, with less attention paid to the distribution and end use sectors. This work discusses 23 flights at 14 natural gas-fired power plants (NGPPs) using an aircraft-based mass balance technique and methane/carbon dioxide enhancement ratios (ΔCH4/ΔCO2) measured from stack plumes to quantify the unburned fuel. By comparing the ΔCH4/ΔCO2 ratio measured in stack plumes to that measured downwind, we determined that, within uncertainty of the measurement, all observed CH4 emissions were stack-based, that is, uncombusted NG from the stack rather than fugitive sources. Measured CH4 emission rates (ER) ranged from 8 (±5) to 135 (±27) kg CH4/h (±1σ), with the fractional CH4 throughput lost (loss rate) ranging from −0.039% (±0.076%) to 0.204% (±0.054%). We attribute negative values to partial combustion of ambient CH4 in the power plant. The average calculated emission factor (EF) of 5.4 (+10/−5.4) g CH4/million British thermal units (MMBTU) is within uncertainty of the Environmental Protection Agency (EPA) EFs. However, one facility measured during startup exhibited substantially larger stack emissions with an EF of 440 (+660/−440) g CH4/MMBTU and a loss rate of 2.5% (+3.8/−2.5%).
■
INTRODUCTION Natural gas (NG) usage has been growing since the large-scale implementation of hydraulic fracturing and horizontal drilling technologies to take advantage of shale resources.1,2 Along with market factors and stricter environmental regulations, this has led to a near doubling of the U.S. electricity generation from NG since 2008, with electricity generation matching or surpassing that of coal since 2016.3 NG offers improved efficiency and availability and only produces 56% the amount of carbon dioxide (CO2) per unit energy as coal, making it a potential “bridge fuel” in the transition toward renewable energy.4 However, NG is primarily composed of methane (CH4), the second most important anthropogenic greenhouse gas (GHG) accounting for 9% of all U.S. GHG emissions in 20175 based on its 100-year Global Warming Potential (GWP) (calculations are detailed in the SI). CH4 is also a short-lived gas with a GWP of 84 over a 20year period compared to a GWP of 28 over a 100-year period.6 When using the 20-year GWP, CH4 emissions are equivalent to © 2019 American Chemical Society
22% of annual U.S. GHG emissions (calculation detailed in SI). The short-term impact of reducing CH4 emissions makes it an important focus of climate change mitigation efforts. Over 50% of global CH4 emissions are related to human activity, and losses from the energy sector are the largest anthropogenic source in the U.S.7 On the basis of a recent synthesis of CH4 emissions from well to end user, Alvarez et al.8 estimated that 2.3% of U.S. gross production of NG is emitted to the atmosphere. At this loss rate, supply chain CH4 emissions nearly double the short-term climate impact of the combustion of NG for energy. Therefore, quantifying losses along the NG supply chain from production to end use is essential. To realize Received: Revised: Accepted: Published: 8976
March 28, 2019 June 14, 2019 June 24, 2019 June 24, 2019 DOI: 10.1021/acs.est.9b01875 Environ. Sci. Technol. 2019, 53, 8976−8984
Article
Environmental Science & Technology Table 1. Key Parameters Describing the NGPPs Studieda ID
state
firing method
max capacity (MW)
operation type
commercial operating date
P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14
UT UT IL IL IN FL FL FL FL FL OH OH OH MI
combined cycle combined cycle combined cycle combined cycle combined cycle combined cycle simple thermal combined cycle combustion turbine 50% combined cycle, 50% combustion turbine combined cycle combined cycle combined cycle 50% combined cycle, 50% combustion turbine
560 1180 1310 680 730 3530 1070 1190 790 2100 760 590 940 1000
intermediate intermediate baseload peaking baseload baseload baseload baseload intermediate intermediate baseload baseload baseload baseload
2005 2007, 2014 2002 2002 2002 2009, 2011 2013 2014 2002 2003, 2009 2012 2012 2003 2001, 2002
a
Operation types are based on the most active units (e.g., a plant with 1 baseload and 3 intermediate units is labeled baseload). Maximum capacities are calculated from the maximum hourly heat input from AMPD,23 and the heat rate from the U.S. Energy Information Administration.22 Commercial operating dates are from AMPD data and news articles for the plants. Multiple years are listed when additional units were added, and conversion dates are used if plants were converted to gas.
units (MMBTU) for all NGPPs based on the Intergovernmental Panel on Climate Change (IPCC) recommendations.24−26 The Greenhouse Gas Inventory (GHGI), a comprehensive bottomup inventory used to estimate national emissions by source category, instead uses an EF of 3.9 g CH4/MMBTU for CC and combustion turbine NGPPs based on both the IPCC and the EPA’s Compilation of Air Pollutant Emission Factors (AP42).27−29 In this work, we present statistically meaningful results from 5 NGPPs and discuss the magnitude and variability of stack emissions. A companion paper is being prepared to investigate the measured CO2 emission rates (ERs) as compared to AMPD reported CO2 ERs.30
the climate benefit of NG, it must be efficiently handled and combusted. There have been multiple studies focused on the production, storage, and processing of NG, but there has been little work on end users, such as natural gas-fired power plants (NGPP).8−21 Lavoie et al.17 studied three NGPPs and saw emissions of unburned CH4 from the stacks and relatively large CH4 leaks attributed to nonstack sources on-site. The Lavoie et al. study was based on a small sample size of 3 combined cycle power plants (CC), which use the combustion gases to turn a turbine. Excess heat is then used to generate steam to turn a second turbine. CC plants are the most efficient NGPPs, producing 46% more energy per energy content of fuel consumed than a simple combustion turbine.22 Because of this higher efficiency, CC facilities are the most common type of NGPP, providing 89% of the electricity produced by NG.23 As such, we focus largely on CC NGPP emissions in this study. We expand on the work of Lavoie et al. by sampling a larger set of NGPPs to thoroughly investigate the prevalence of on-site CH4 leaks and to gather more robust emissions data to compare to the Environmental Protection Agency (EPA) estimates. We studied 14 NGPPs, but only 5 showed downwind ΔCH4, while all showed ΔCO2 downwind. This suggests CH4 emissions were too low to be detected above atmospheric variability at most NGPPs. This work focuses on the 5 NGPPs that showed downwind ΔCH4. We calculate ΔCH4/ΔCO2 (ppm/ppm) ratios when flying through the stack emission plumes with/ against the mean wind direction or circling near the stacks. We quantify the facility (nonstack) CH4 leaks by comparing this to the same ratio from downwind aircraft-based mass balance experiments (MBE), which would capture all plant emissions. Although the EPA calculates CH4 emissions from NGPPs, it does so using emission factors (EF) that have not been welltested and may underestimate emissions based on previous work in the NG sector.2,9,10,17 The EPA requires that facilities report hourly averaged CO2 emissions through the Air Markets Program Data (AMPD) using Continuous Emissions Monitoring Systems (CEMS) as described in Title 42 of the U.S. Code of Federal Regulations.23,24 As for CH4, the Greenhouse Gas Reporting Program (GHGRP), a reporting program for GHG point sources, uses an EF of 1 g CH4 per million British thermal
■
MATERIALS AND METHODS Instrumentation. Flights were conducted using Purdue’s Airborne Laboratory for Atmospheric Research (ALAR),11,31−34 which is a modified twin-engine Beechcraft Duchess aircraft. ALAR is outfitted with a global positioning and inertial navigation system, a Best Air Turbulence probe for high precision 3-dimensional wind measurements,35 and a model G2301-m Picarro Cavity Ring Down Spectrometer designed for 0.5 Hz airborne measurements of CO2, CH4, and H2O.36 We conducted multiple three-point calibrations each flight using NOAA-certified standard cylinders containing CO2 and CH4, with concentrations that bracket the range of typical observations. A direct absorption ethane (C2H6) spectrometer designed by Aerodyne Research and modified at Harvard University was added to ALAR for a subset of flights (see SI for instrument details). Flight Design and Site Selection. We conducted a total of 23 flights at 14 NGPPs to quantify NGPP CH4 emissions. Combined, these plants represent 3.4% and 1.5% of NG and total U.S. nameplate capacity, respectively.37 A principal goal of this study was to produce a representative data set; thus, NGPPs were chosen to include a variety of regions, firing methods, maximum capacities, ages, and operation types, as shown in Table 1.17 These NGPPs are relatively new, but this is to be expected as ∼70% of NGPP capacity comes from units ≤20 years old.37 Operation type refers to a unit’s typical generation and is defined here as baseload units operating >70% of the year, 8977
DOI: 10.1021/acs.est.9b01875 Environ. Sci. Technol. 2019, 53, 8976−8984
Article
Environmental Science & Technology Table 2. Conditions and Date of Each Mass Balance Flighta ID
date (MM/DD/YY)
time (local)
wind direction (deg)
wind speed (m/s)
flight method
P6 P6 P6 P8 P8 P4 P4 P4 P3 P3 P3 P2 P2
11/12/16 11/13/16 11/19/16 11/14/16 11/17/16 6/14/17 6/21/17 7/24/17 5/18/17 7/7/17 7/24/17 10/12/17c 10/13/17c
13:50−15:40 14:35−15:33 13:14−15:14 14:05−14:50 13:12−14:19 12:40−13:11 14:50−15:40 16:05−17:26 13:46−15:44 14:40−15:16 12:01−13:38 13:26−13:53 12:23−12:51
145 ± 70 130 ± 10 335 ± 30 250 ± 10 38 ± 8 220 ± 13 250 ± 20 50 ± 10 260 ± 10 340 ± 10 30 ± 20 290 ± 50 NAb
2±1 4.1 ± 0.9 4±2 3.2 ± 0.7 8±2 7±1 4±1 4.2 ± 0.9 13 ± 3 11 ± 1 5.5 ± 2 2±1 3b
DTd DT DT DT DT spirald spiral DT DT DT DTe DTf spiral
a
Uncertainties are 1σ. DT = downwind transects. bComplex wind conditions, see SI for details. cAMPD data shows the facility to be in startup, having begun producing within the past 4 h. dCH4 emissions were too low to be quantified. eEmissions cannot be quantified because of contamination from an upwind source. fEmissions could not be quantified due to incomplete plume capture downwind.
(stack ER) according to eq 2. Stack ratios were calculated in a similar manner to previous studies17,41,42 with details provided in the SI. Any MBE ERs that are substantially greater than the stack ER would indicate potential facility-scale CH4 leaks.
intermediate units operating 30−70%, and peaking units operating
