|Home | About | Journals | Submit | Contact Us | Français|
We present ground-based, advected aircraft engine emissions from flights taking off at Los Angeles International Airport. 275 discrete engine take-off plumes were observed on 18 and 25 May 2014 at a distance of 400m downwind of the runway. CO2 measurements are used to convert the aerosol data into plume-average emissions indices that are suitable for modelling aircraft emissions. Total and non-volatile particle number EIs are of order 1016–1017 kg−1 and 1014–1016 kg−1, respectively. Black-carbon-equivalent particle mass EIs vary between 175–941mgkg−1 (except for the GE GEnx engines at 46mgkg−1). Aircraft tail numbers recorded for each take-off event are used to incorporate aircraft- and engine-specific parameters into the data set. Data acquisition and processing follow standard methods for quality assurance. A unique aspect of the data set is the mapping of aerosol concentration time series to integrated plume EIs, aircraft and engine specifications, and manufacturer-reported engine emissions certifications. The integrated data enable future studies seeking to understand and model aircraft emissions and their impact on air quality.
Aircraft engine particle emissions are important contributors to local air quality near airports1–5, and these downstream environmental impacts are likely to increase in concert with the projected growth of the aviation sector over coming decades. Emissions relevant to local impacts occur during multiple stages of aircraft movement including idle, taxi, take-off, and the portion of climb out and landing below 3,000 feet altitude above field elevation. Collectively these stages are referred to as the landing and take-off (LTO) cycle. Emissions standards for each phase of the LTO cycle are set by the International Civil Aviation Organization (ICAO) pursuant to Volume II of Annex 16 of the Convention of International Civil Aviation6, which recommends standard engine emissions testing methods for gas-phase and aerosol species. Aerosol emissions are quantified by engine manufacturers prior to certification and operation in terms of a smoke number metric that describes how soot particles collected by a filter change the reflectance of the filter over a defined sampling time, and which is known to be dependent on sampling conditions and soot properties.
Recognizing the significant limitations of the smoke number metric, current and future efforts are underway to measure engine LTO aerosol emissions in a more rigorous fashion by measuring particle number and/or mass emission indices. While these data will inform transportation modelling of the next generation of aircraft engines, there are currently no plans to recertify older engines that are in service now (and are likely to remain in service in the coming decades). In addition, the LTO emissions certification process is idealized as engine conditions are measured under discrete, steady thrust settings that may differ from the thrust actually applied by pilots. For example, thrusts applied during taxi and idle may be lower than the 7% of maximum thrust that is commonly assumed. Similarly, pilots frequently take off under reduced thrust of up to 25% below maximum, depending on runway conditions and aircraft weight and specifications7,8. Congestion on taxiways may lead to sudden engine accelerations and decelerations as aircraft taxi to and from runways and thrust reversers may be applied during landings, all of which are non-standard conditions5. In addition, aerosol particles undergo modifications by condensation and coagulation in the airport environment. These modifications lead to effective emissions that impact airport air quality not considered in the ICAO certification process. Finally, the amount of time spent in each phase of the LTO cycle (referred to as time-in-mode, or TIM) drives the overall amount of engine emissions and is not well constrained.
Given these myriad sources of variability, there is a need to understand the emissions from currently in-service engines under real-world conditions. Here, we investigate particles emitted by aircraft during take-off operations at Los Angeles International Airport and synthesize these emissions measurements with information on the aircraft and engine specifications as well as the ICAO certification emissions values for each engine model. The work flow of this study is shown in Fig. 1, leading from the base data files generated by the instruments, field notes, and existing emissions databank, through the intermediate analysis and processing steps, to arrive at synthesized output data files that form the two data records described by this data descriptor. The environmental, aircraft, engine, and emissions data parameters are listed in Table 1.
This comprehensive dataset informs future studies seeking to model the impacts of take-off engine emissions at and in the vicinity of airports, to evaluate the performance of current approximation methods for estimating emissions from smoke number measurements, to explore plume-averaged and transient emissions profiles during many aircraft take-off events, and to understand the background aerosol concentrations and properties downwind of a major airport in between the take-off events, among other uses. A powerful and unique feature of this data set is that ambient concentrations of both aerosols and carbon dioxide are used to compute EIs and link these EIs to specific aircraft and engines as well as engine specifications and emissions standards.
Data were collected at 400-m distance downwind of the northern take-off runway (24L) of Los Angeles International Airport (33.9509°N, 118.398°W) on two days: 18 May and 25 May 2014. A schematic of the airport and aerial view of the sampling location near the runway are shown in Fig. 2. Runway 24L has a declared length of 3,135m, 263° true bearing, and is at an elevation of 38m. A predominantly onshore sea breeze of 0–10ms−1 was oriented down the runway (±20°) during both measurement days, which advected the aircraft take-off plumes to the sampling inlet of the NASA Langley Aerosol Research Group (LARGE) Mobile Laboratory. The height of the inlet was approximately 3m above the ground. Ambient temperature and relative humidity varied between 20–25°C and 45–65%, respectively. Table 1 shows the complete list of environmental data.
Ambient air was drawn through a 1.3cm outer diameter stainless steel tube and distributed to instruments inside the mobile laboratory with varying transport tubing lengths of 6–10m depending on where each instrument was mounted. Given the uncertainty in these varying transmission lengths, the data are not corrected for size-dependent diffusional losses to the tubing walls. As a constraint on these impacts, assuming a nominal flow rate of 50lmin−1, calculations suggest that 13–22% of 20-nm-diameter particles are lost across the full range of transport tubing lengths9.
Laboratory analyses of fuel composition were obtained from LAXFUEL for the tanks on issue on May 18th and 25th, which describe the fuel sulphur content, aromatic content, naphthalenes content, and the net heat of combustion. Mean fuel properties (±1 standard deviation) are reported for each day in Table 2, assuming equal weighting across the two fuel batches analysed for the 18th and across the four fuel batches analysed for the 25th. On the 18th only a single tank was on issue, while on the 25th, a tank fuel containing moderate sulphur and low aromatics (710ppmmS and 12 % by volume, respectively) was initially on issue and was followed by another tank containing a blend of domestic and foreign fuels whose sulphur and aromatic contents varied considerably (620–1,780ppmmS and 17.6–23% by volume, respectively). Since it is not possible to determine when a particular aircraft would have fuelled prior to departure, it is not possible to draw composition-specific conclusions on emissions parameters. For completeness, we report here the individual batch analyses and the daily-averaged fuel properties, which are indicative of moderate sulphur content and typical aromatic content as compared to previously reported jet fuel properties10,11.
Environmental and emissions measurement parameters are summarized in Table 1, and a brief description of each instrumental measurement method is given below.
The time of each aircraft take off event and the aircraft tail number were recorded by an observer in the field. These data were then matched to a plume peak in the measurement time series. Of these pieces of information, the time recorded in the field when the observer noticed the aircraft begin to take off (TakeoffStartTime_UTC) can be uncertain because the observer was distracted, troubleshooting instruments, or missed the start of take off and instead wrote down the time when the measurement peak appeared. Consequently, while these times are used to roughly match the aircraft tail numbers to observed concentration peaks, we also report PlumeStartTime_UTC as the start of the plume as determined during data post-processing. In a few instances, TakeoffStartTime_UTC is after PlumeStartTime_UTC, and the latter is the more reliable data parameter.
Temperature, pressure, wind speed, and wind direction were measured continuously with a WeatherHawk 232 weather station mounted on top of the mobile laboratory near the sampling inlet. Air temperature is measured to within ±0.5°C accuracy with a thermistor, while the piezoresistive barometric pressure transducer is accurate to within ±1.5kPa. Wind speed is measured with a cup anemometer with a starting threshold of 0.78ms−1, and wind direction is determined from a vane sensor with 1% linearity and ~1ms−1 sensitivity.
CO2 mixing ratio was measured at 1Hz with a Licor LI-7000 CO2/H2O Gas Analyser. The instrument measures the differential absorption of light by carbon dioxide and water vapour at 4.255 and 2.595μm wavelengths, respectively. Only the CO2 data were calibrated and are included in this data set.
BC mass concentration was measured with a Thermo Scientific Multi-Angle Absorption Photometer (MAAP)12. The MAAP continuously measures the amount of light transmitted through a particle-loaded glass fiber filter material as well as light backscattered off of the filter, both at 670nm wavelength. These measurements are then used to determine a black carbon equivalent aerosol mass concentration by assuming a mass absorption coefficient of 6.6m2 g−1. The instrumental uncertainty is estimated to be ±12%13.
Particle number concentration was measured at 1Hz with a TSI Condensation Particle Counter (CPC; Model 3775). For particle concentrations less than 5×104 cm−3, individual particles are detected as pulses of laser light (‘single-counting mode’), while for higher particle concentrations (5×104 to 1×107 cm−3) the total amount of light scattered by the particle population is used to determine concentration (‘photometric mode’). Particle concentration accuracy is reported as±10% in single-counting mode and±20% in photometric mode. The minimum detectable particle size of the CPC is 4nm diameter and it has a response time of 4s.
Non-volatile particle number concentration was measured at 1Hz with a TSI CPC (Model 3022A) located downstream of a thermal denuder. The thermal denuder is a stainless steel tube heated at 350°C in order to evaporate volatile material on the aerosol (e.g., sulphur and nitrate species and organics). The 3022A CPC is similar to the 3775 Model, except that the minimum detectable particle size is 7nm diameter and the response time is <13s. This difference in lower detection size does not preclude direct comparison of total and non-volatile particle number because non-volatile particles, such as black carbon, are known to be greater than 10–20nm in diameter.
The particle number size distribution between 6 and 575nm diameter was measured at 1Hz with a TSI Engine Exhaust Particle Sizer (EEPS; Model 3090). Particles are drawn into the EEPS at 10lmin−1 flow rate and are given a positive charge via corona charging before entering a measurement region of two concentric cylinders on which an electric field is applied. The positively-charged particles move toward the outer electrode at a rate proportional to their size-dependent electrical mobility until they ultimately impact on one of several sensitive electrometers. An inversion algorithm is applied to the time-dependent electrometer currents to retrieve the aerosol size and number concentration at 1Hz.
Past comparisons between an EEPS (or similar instruments such as the Cambustion DMS500 and TSI FMPS) and a Scanning Mobility Particle Sizer (SMPS) have shown that the EEPS slightly undersizes soot particles relative to the SMPS14–18. This undersizing is particularly problematic for large, diesel soot agglomerates, and is less of an issue for aircraft soot that tend to be sub-100-nm in diameter and closer to compact spheres than fractal agglomerates. In order to better understand the performance of the EEPS relative to the state-of-the-art SMPS, we examined data from both instruments during the NASA Alternative Fuel Effects on Contrails and Cruise Emissions (ACCESS) project, which sampled the exhaust of the NASA DC-8 CFM56 engines. This comparison showed that the EEPS undersizes particles by about 10% relative to the SMPS for intermediate and high engine thrust settings, which is similar to the size discrepancy reported by Hagen et al.14 for aircraft soot sampled during the 2004 NASA APEX project. Consequently, we have corrected the size bins by a scaling factor of 1.1 for all EEPS data, but also report the uncorrected EEPS size bin diameters in the data record files. The EEPS-SMPS comparison using ACCESS data shows good agreement between the number concentrations reported by both instruments, which are well within the previously reported instrumental uncertainty of ±20%16.
CCN concentration at (2.6±0.2)% water vapour supersaturation was measured at 1Hz with a Droplet Measurement Technologies CCN Counter19,20. The instrument was operated at an elevated flow rate of 1lmin−1 and elevated temperature gradient of 16°C to effect this high supersaturation, which was calibrated using size-classified, dry ammonium sulfate aerosols and Scanning Mobility CCN Analysis21. The uncertainty in supersaturation of 0.2% is propagated from the scatter in the calibration critical activation diameters using Köhler theory with corrections for incomplete solute dissociation following the ion-interaction approach of Pitzer and Mayorga with parameters obtained from Clegg and Brimblecombe22–24. The uncertainty in the CCN concentration is estimated to be 7–16%25.
Aerosol light extinction is measured at 530nm wavelength with a cavity attenuated phase shift extinction (CAPS PMex) monitor26. The instrument sensitivity is 2.5Mm−1 with a response time of less than two seconds. The CAPS PMex instrument is also sensitive to the presence of absorbing gases (e.g., NO2), which were not measured during this project. Yu et al.27 found this correction to relatively minor for aircraft engine idle conditions, and the emissions index of NO2 at higher engine thrusts is even lower than at idle28.
Here, we report engine emissions parameters in terms of a plume-average emissions index that is normalized to the rate of engine fuel burn. This normalization process takes into account differences in plume dilution that can be affected by turbulence, varying wind speed and direction, as well as differences in sampling instrument response time constants. For example, the MAAP response time is much slower than that of the particle concentration and size distribution measurements owing to the internal averaging and smoothing algorithms applied by the instrument firmware.
The emission index of particle species X is determined following Moore et al.29 as
ΔX and ΔCO2 are the background-subtracted peak areas of the measured concentrations of species X and CO2 at standard temperature and pressure, respectively; EICO2 is the emissions index of CO2, assuming that the carbon content in the fuel is constant and is completely converted to CO2; R is the ideal gas constant; T is the temperature at STP (273.15K); P is the pressure at STP (1 atm); Vm is the molar volume of ideal gas at STP (22.4lmol−1); α is the fuel hydrogen-to-carbon molar ratio (assumed to be 1.92); and MCO2, MC, MH are the molar masses of CO2, carbon, and hydrogen, respectively.
Figure 3 shows example time series of CO2, black carbon mass concentration (BC), and particle number concentration (CN), where the shaded regions represent the background-subtracted peak areas (ΔCO2, ΔBC, and ΔCN). Bounding points were visually set by the authors on either side of the peak and a linear fit between those points establishes the background baseline. The background-subtracted peak area is then the difference between the integrated area under the concentration time series and the area under the background baseline between those two points. This was determined using the areaXY function in Igor Pro (Wavemetrics, https://www.wavemetrics.com/).
The United States Federal Aviation Administration (FAA) and similar national regulatory agencies maintain civil aircraft registration records that describe the general specifications of the airframe and engines as well as attesting to the ownership of the aircraft and its airworthiness. Each aircraft has a unique tail number identifier that maps to the various national registration databases, with the first letter denoting the nationality of the aircraft. For example, the United States nationality designator is ‘N’, and U.S. flag aircraft have tail numbers that begin with ‘N’ followed by a unique alphanumeric identifier. The aircraft tail numbers were photographed with a telephoto lens prior to take-off as the plane taxied to the runway. The tail number was then used with aircraft registration databases as in Fig. 1 to obtain the detailed specifications of the aircraft and engine. These data include the airline (also visible from the aircraft markings), the aircraft manufacturer and year of manufacture, and the aircraft model and series. In addition, the number of engines, their manufacturer, and their model and series are contained within the registration database. A list of compiled aircraft and engine parameters is given in Table 1, and the emissions data are summarized by engine type in Table 3.
The International Civil Aviation Organization maintains a database of the engine exhaust emissions parameters of production aircraft engines. Emissions are characterized by the engine manufacturers following ICAO Annex 16 Vol II and reported on a voluntary basis30. For each engine model and series, the EDB lists specifications including the engine type (for example, turbofan, mixed turbofan, and turboprop), the engine bypass and pressure ratios, and the engine maximum rated thrust. Emissions parameters reported in the EDB include carbon monoxide, nitrogen oxides, hydrocarbons, and smoke number for four different engine power conditions that correspond to different point in the LTO cycle (idle, approach, climb out, and take off conditions). Detailed information about the test-specific fuel, its properties, and fuel flow rates at each power setting are also provided. Since this data descriptor focuses on aerosol emissions, we only include the EDB smoke number and associated test and fuel information in the synthesized data set (Data Citation 1).
Data were analysed with commercially-available software including Igor Pro 6.37 and Microsoft Excel 2013. Summary statistics reported in the Technical Validation Section in this data descriptor are generated with custom code in ‘R’, which is available without restriction in the data records as an HTML file: LAX-Ground-ProcessingCode_R01.html.
Two identical data records are associated with this work: a set of data archived in the NASA Aeronautics Field Projects database (https://aero-fp.larc.nasa.gov/projects/lax) and a Dryad data set (Data Citation 1). While the NASA database is the primary repository for NASA Aeronautics emissions research, the Dryad database is indexed with a digital object identifier (DOI) ensuring a persistent identifier is assigned to the data set. Each data record contains two files corresponding to the Output Data Files in Fig. 1. Time series of synthesized aerosol and carbon dioxide concentration data on a common time base are found in the LAX-Ground-Summary_TS_20140518_R01_thru20140525.xlsx Microsoft Excel workbook, where TS denotes time series data. Meanwhile, the calculated emissions indices across both days are assimilated with aircraft- and engine-specific parameters for each of the 275 take-off plume test points in the Microsoft Excel workbook entitled LAX-Ground-Summary_TP_20140518_R01_thru20140525.xlsx, where TP denotes test point data. Also included in both data records is an HTML file containing code used to process the data and figures in ‘R’ for this data descriptor named LAX-Ground-ProcessingCode_R01.html.
Archived measurement parameters and associated aircraft and engine characteristics are detailed in Table 1. These data map to the README tabs in the.xlsx workbooks.
The test points cover 29 different aircraft engine master model combinations that correspond to a wide variety of aircraft as shown in Table 3. The majority of aircraft plumes sampled come from CFM56-3B and CFM56-7B class engines that are popular in the Southwest fleet of 737s, from the CFM56-5B class engines on Virgin America Airbus A319/320/321 aircraft, and from GE CF34-8 class engines on Delta Connection CRJ and ERJ aircraft. This distribution is due, in part, to the close proximity of Runway 24L to Terminal 1, which serves Southwest Airlines.
Data were compiled into the data records described above, and data integrity is verified as follows:
Summary statistics reported in this data descriptor are generated with custom code in ‘R’, which is available in the data records as an HTML file: LAX-Ground-ProcessingCode_R01.html. All files are self-explanatory with metadata in the README tabs of the .xlsx data files.
Some possible uses of this dataset include:
A key feature of this data descriptor is that ambient concentrations of both aerosols and carbon dioxide are used to compute EIs and link these EIs to specific aircraft. This level of specificity is uncommon in the past literature, where only ambient aerosol concentrations are often reported. Given the usefulness to modellers of reporting results in terms of an EI (in other words, in terms of the mass of fuel burned), we suggest that future studies should follow this approach. In addition to the methodology employed in this data descriptor, it is recommended that future aircraft emissions sampling studies should also
How to cite this article: Moore, R. H. et al. Take-off engine particle emission indices for in-service aircraft at Los Angeles International Airport. Sci. Data 4:170198 doi: 10.1038/sdata.2017.198 (198).
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank Jim Moses at LAXFUEL for providing the fuel analysis information and Robert Freeman and Steve Baughcum for discussions. This work was supported by the NASA Advanced Air Vehicles Program, Advanced Air Transport Technology Project and the DLR Aeronautics Research Programme. R.H.M., E.C., and T.S. were supported, in part, by NASA Postdoctoral Program fellowships. T.J. was supported by the Helmholtz Association (grant number W2/W3-060) and the German Science Foundation (DFG grant number JU3059/1-1).
The authors declare no competing financial interests.