Methodology to Improve AEDT Quantification of Aircraft Taxi/Idle Emissions (2016)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

3  Do the current databases, input variables, algorithms and sub-models of AEDT provide a reasonable estimate of taxi-related emissions? To answer these questions, the literature review focused on the following four categories of information: Information Categories  Aircraft Performance Characteristics – including aircraft taxi/idle times-in-modes under alternative airfield conditions, single-engine taxi procedures, flight data recorder (FDR) data, etc.;  Aircraft Engine Emissions – including ICAO reference fuel flows and emission indices, aircraft engine emissions and ambient measurements, etc.;  AEDT Performance and Development – including model architecture, development programs and timeframes, model accuracy and sensitivity tests, etc.; and  Regulatory Framework – including the FAA’s rules/regulations that relate to the taxi process and the requirements of the National Environmental Policy Act (NEPA), the federal Clean Air Act (CAA), and other state/local requirements as they pertain to airport emissions and ambient air quality. Based on the findings of the literature search and review of model inputs in Tasks 1 and 2, some additional issues, or research “gaps,” in need of resolution were also identified. In addition to these research “gaps,” the Task 3 White Paper identified potential model inaccuracies revealed in the review. Table 1 below summarizes those findings relative to potential model inaccuracies and their possible impact on taxi/idle emission predictions. For detailed discussion of those issues, the reviewer is referred to the complete Task 3: (Working Paper) included as Appendix A. Table 1 Summary of AEDT Shortcomings Related to Taxi/Idle Emissions Shortcoming Impact AEDT assumptions regarding the duration of taxi/idle modes are not representative of actual conditions. Differences in actual emissions and AEDT emissions are (at least) directly proportional to differences in the duration of taxi/idle modes. AEDT assumes fixed fuel flow rates during taxi/idle based on seven percent thrust. Actual fuel flow rates vary considerably and emission indices are a function of fuel flow rate. AEDT uses one emission index value for each pollutant. Actual emission indices are complex functions of fuel flow rate, ambient temperature, and other factors. AEDT does not account for variations in operational practice, including tug assisted single-engine (i.e., reduced engine) or electric taxiing. Discrepancies between actual and AEDT assumed operating patterns directly impact the accuracy of model estimates. 3. Research Approach and Findings Based largely upon the outcomes of the Task 1 Literature Search and Task 2 AEDT Review, the subsequent research tasks were entitled and performed as follows: Subsequent Tasks  Task 4: Analysis of In-service Engine Performance Data  Task 5: Evaluate the Implications of Model Inaccuracies

4  Task 6: Develop List of Potential AEDT Improvements  Task 7: Stakeholder Outreach  Task 8: Interim Report  Task 9: Identification of Near-Term Model Improvements  Task 10: Steps Needed for Implementation  Task 11: Final Report Notably, for the interrelated Tasks 4, 5 and 6 designed to develop the initial list of potential model improvements, the Research Team targeted three fundamental elements (i.e., factors) within the current AEDT framework that the model uses to calculate aircraft engine emissions during the taxi/idle mode. These factors are identified and represented as follows: Targeted Aircraft Taxi/Idle Emission Variables Taxi/idle emissions = TIM x FFR x EI Where:  Time-in-mode (TIM) = the time aircraft engines are operating in the taxi/idle mode (seconds);  Fuel flow rate (FFR) = the rate of fuel intake (kilograms (kg) per second); and  Emission index (EI) = the emissions generated per mass of fuel burned (grams (g) per kg of fuel burned). The remainder of this section discusses each of these factors in detail, including how the Research Team evaluated options for improving the model’s treatment or consideration of each one. There is also a discussion of potential model improvements not directly related to the three factors which are discussed under a fourth category termed “Additional Considerations.” For ease of review, Table 2 provides a summary of this information. 3.1 Time-in-Mode In its current configuration, AEDT considers aircraft taxi/idle time-in-mode two different ways, depending on whether the model is being used to (i.) only prepare an emissions inventory or to (ii.) perform atmospheric dispersion modeling (which also provides an emissions inventory). These two methods are summarized as follows:  Emissions Inventory – When only an emissions inventory is being prepared, model users are instructed to input airport/aircraft-specific values (or use default values) of taxi times (i.e., taxi- in and taxi-out), defined as the amount of time required for an aircraft to taxi to/from an airport’s terminal gate and/or runway. These values include periods of aircraft delay encountered along the way.  Dispersion Modeling – When dispersion analysis is performed, model users are required to input a value (or use default) for the aircraft taxi speed. Under this option, the model performs Delay and Sequence Modeling which uses airport operational schedules, runway configurations, and the airport’s capacity to estimate taxi/idle times and delay periods. Because the AEDT default taxi/idle time-in-mode represents a potential source of inaccuracy in computing aircraft taxi/idle emissions, this parameter was the focus of further analysis, as discussed below. 3.1.1 Default Taxi Times Presently, the AEDT default aircraft taxi/idle (i.e., taxi-in, taxi-out and delay) times-in-mode are set to 7 minutes for taxi in and 19 minutes for taxi out (a total of 26 minutes). These values, dating back to the

5 1980s, were derived from the U.S. Environmental Protection Agency’s (EPA’s) AP-42, Compilation of Air Pollutant Emission Factors as being typical durations for civil aircraft (i.e., commercial jumbo, long- range, and medium-range jets) operating within the standard landing/take-off (LTO) cycles at large congested metropolitan airports. In search of an alternative to these default values, it is instructive to know that AEDT also has an internal database containing airport-specific taxi-in and taxi-out times. While there are a substantial number of airports included in the list (i.e., 75 airports), the number of years for which taxi data are provided varies by airport and the values are also somewhat out of date, with the most recent data being from 2006. To improve modeling taxi times, the Research Team extracted five years of taxi data for the 74 commercial service airports contained in the FAA’s Aviation System Performance Metrics (ASPM) Database that have these data. For the purposes of this analysis, the ASPM’s actual taxi times from this database, which include unimpeded taxi times and delay periods (i.e., impeded taxi times), were used. Based on the outcome of this analysis, three options were developed that, if implemented, could improve the simulation of aircraft taxi/idle times in AEDT and thus also improve the accuracy of an emissions inventory computed using these data.5 These taxi time options are described below. Aircraft Taxi Time Options  Option 1 – Update the AEDT default values using the average taxi in/out times computed from the ASPM data. As shown in Table 3, these values vary somewhat by airport size (i.e., large, medium, small) and number of runways, but trend toward overall averages of 7 minutes for taxi in and 16 minutes for taxi out. This option would reduce the default taxi/idle times (and the resultant emissions) by 12 percent when compared to the current default values and would apply to all aircraft, not just those equipped with commercial jet engines.  Option 2 – Under this option, AEDT users would select default taxi in/out times based on an array of airport-specific design characteristics, including: (i.) the number of runways, (ii.) whether an airport is a hub or non-hub, and/or (iii.) for hub airports, the size of the airport (i.e., large, medium, and small). Again, as shown in Table 3, while taxi/idle times and the resultant emissions would increase at several airports (i.e., 5 of the 75 airports) and remain the same at several more (5 to 6, depending on the selected method), taxi/delay times and the resultant emissions would decrease from 4 percent to 50 percent at the majority of the airports when compared to the current AEDT default values.  Option 3 – The current version of AEDT apparently does not use the airport-specific database of taxi in/taxi out times from the EDMS database that are out-of-date. Under this option, that database would be expanded to include more airports and updated using average taxi times (including delay periods) obtained from the most recent five years of data from the ASPM. These data are listed in Table 4, where they are compared with the current default values. With the exception of six airports, these data would decrease aircraft taxi/idle times (and the resultant emissions predictions) from 4 percent to 50 percent at the majority of the airports, when compared to the current default values. Notably, although the 02-45 Research Project is only addressing taxi/idle emissions that result from the operation of commercial jet aircraft, the taxi time options described above can be assumed for all aircraft operating at an airport (i.e., jets, turboprops, and props). 5 When conducting dispersion modeling using AEDT, aircraft taxi times are computed based on input values of taxi speed and aircraft delay which is calculated using an internal Delay and Sequence Modeling mechanism.

6 Table 2 AEDT Taxi/Idle Emission Improvement Research Approach Taxi/Idle Emissions Computational Factors Research Parameter(s) Research Accomplishments Improvement Option(s) Time-in-mode Default taxi time (criticall for AEDT users that use the default time-in- mode option) - Currently 7 minutes taxi in and 19 minutes taxi out. Extracted five years of taxi data for the 74 commercial service airports for which there is data in the ASPM. Using this data, the following were derived:  Averages for taxi in and out for 1) all airports (hub plus non-hub) and 2) based on the number of runways.  Averages by 1) type of airport (hub, non-hub) and 2) type of airport based on the number of runways.  Averages by size of hub airport (large, medium, and small).  Five year averages for each of the 74 airports. 1) Change default taxi in and taxi out to values derived for all airports. --7 minutes taxi in / 16 minutes out 2) Provide user query for type of airport and number of runways in GUI – link to new database. -- Values in Table 3 3) Default to average of five years of data for specific airport being evaluated. -- Values in Table 4 Default taxi speed (important for AEDT users that evoke the dispersion modeling option ) - Except for queue area before departure, AEDT assumes aircraft taxi at one speed (default 15 knots (17.26 mph)). Users may also enter aircraft specific speeds. Reviewed 258,824 data samples from FDR data (1,800 aircraft operations—mostly Airbus from European carrier). 1) Change default assumption to weighted average value based on FDR data. -- 11 knots (12.66 mph) 2) Allow users to indicate (or the model to distinguish) whether a taxiway is used for aircraft taxiing in or out. -- 13 knots (14.96 mph) for taxi in taxiways and 10 knots (11.51 mph) for taxi out taxiways Fuel Flow Rate (FFR) FFR - Actual FFRs can be higher or lower than those listed in engine- specific ICAO datasheets for operation at 7 percent thrust. Rates are positively correlated with thrust setting and bleed flow. Furthermore, a range of FFRs is used during idle, not just a fixed single value. Reviewed 258,824 data samples from FDR data (1,800 aircraft operations--mostly Airbus from European carrier) and data from APEX2, APEX3, and ACRP 02-03a. 1) Adjust FFRs in databases only for those aircraft for which there are FDR data (varies between 80 and 111 percent of ICAO idle value for taxi in and between 90 and 113 percent for taxi out). -- Values in Table 6 2) Adjust the FFRs for those aircraft for which there are FDR data or for which the data are representative. -- Values in Table 6 and list of engines in Table B-4 3) Use a single, global adjustment to all commercial jet engines. -- 92 percent of an engine’s ICAO idle value Emission Indices CO and HC - Note: EIs should be adjusted only if FFRs are adjusted. Reviewed studies that address dependence of CO and HC on FFR and ambient temperature including: ACRP Report 63, Project 02-03a. (Measurement of Gaseous HAP Emissions from Idling Aircraft as a Function of Engine and Ambient Conditions). Yelvington, P. E.; Herndon, S. C.; Wormhoudt, J.; Jayne, J. T.; Miake- Lye, R. C.; Knighton, W. B.; Wey, C., Chemical Speciation of Hydrocarbon Emissions from a Commercial Aircraft Engine. Journal of Propulsion and Power 2007, 23 (5), 912-918. 1) Apply a global adjustment factor assuming all engines CO and HC EIs follow same temperature/FFR dependence as the CFM56- 7B family of engines. -- Factor varies depending on ambient temperature (Figure 1) 2) Apply an engine specific adjustment factor. -- Factors vary depending on engine (Table 6 and Figure 1) 3) Apply adjustment factors only to the CFM56 family of engines. -- Factor varies depending on ambient temperature (Figure 1) NOx Reviewed studies that address the dependence of NOx on fuel flow and ambient temperature, including documentation for Boeing Fuel Flow Method 2. There is only one option to adjust emissions of NOx. To what engines it would be applied would depend on the FFR adjustment option described above. Additional Assumptions regarding single/reduced engine taxi procedures to be Discovered very few documents or sources to address the extent to When selected by user, apply factor of 0.995 to FFRs for taxi in

7 considerations included in modeling. which reduced taxi procedures are in use or provide data to indicate the percentage of time or frequency when aircraft are taxied with fewer than all engines. operations and 0.96 to FFRs for taxi out operations. Allow for e-taxi procedures to be included in modeling. Discovered very few documents to address the emissions benefit, or provide data to support the emissions benefit, of alternative taxi systems. Allow users to specify the percentage that taxi-related emissions should be reduced. Emission distribution across airfield. Constant thrust assumption (i.e., should the taxi/idle thrust values vary across airfield idle/taxi phase (e.g., x min @ 4 percent thrust and y min @ 12 percent thrust) or is a single thrust assumption sufficient?) Not applicable Allow users to define areas other than the runway queue area where aircraft are delayed (e.g., crossing active runways, ramp area where aircraft are held waiting for gate, deicing area).

8 Table 3 ASPM Aircraft Taxi Times Airport Size No. of Runways No. of Airports Average Time (min.) Percent Change in Taxi/Idle Times (and Emissions)a Taxi In Taxi Out All 1-8 74 7 16 -12 1 2 4 13 -35 2 20 5 15 -23 3 21 6 14 -23 4 21 7 16 -12 5 5 9 17 0 6 3 8 17 -4 7 1 9 15 -8 8 1 9 17 0 Hub (All) 1-8 71 7 16 -12 1 2 4 13 -35 2 17 5 15 -23 3 21 6 14 -23 4 21 7 16 -12 5 5 9 17 0 6 3 8 17 -4 7 1 9 15 -8 8 1 9 16 -4 Hub – Large Airportb 1-8 29 8 17 -4 1 1 4 13 -35 2 2 6 21 +4 3 5 7 16 -12 4 13 7 17 -8 5 3 9 18 +4 6 3 8 17 -4 7 1 9 15 -8 8 1 9 16 -4 Hub – Medium Airportc 1-5 32 5 12 -35 1 1 5 12 -35 2 10 5 11 -38 3 13 5 12 -35

9 4 7 6 13 -27 5 1 5 13 -31 Hub – Small Airportd 2-5 10 5 12 -35 2 5 4 12 -38 3 3 5 12 -35 4 1 4 9 -50 5 1 5 13 -31 All Non-hub 2 3 5 12 -35 a When compared to the current default taxi in/out values of seven and 19 minutes, respectively. b Large airport = Airports that account for at least one percent of the total U.S. passenger enplanements (see Table 4 for a list of these airports). c Medium airport = Airports that account for between 0.25 percent and one percent of the total U.S. passenger enplanements (see Table 4 for a list of these airports). d Small airport = Airports that account for between 0.05 percent and 0.25 percent of the total U/S. passenger enplanements (see Table 4 for a list of these airports). Table 4 ASPM Aircraft Taxi Times by Airport Airport Code Airport Name Hub Type No. of Runways Average Taxi Time (min.) Percent Change in Taxi/Idle Times (and Emissions)a In Out ABQ Albuquerque Intl. Sunport Medium 3 5 10 -42 ANC Ted Stevens Anchorage Intl. Medium 3 4 12 -38 ATL Hartsfield-Jackson Atlanta Intl. Large 5 11 20 19 AUS Austin-Bergstrom Intl. Medium 2 5 12 -35 BDL Bradley Intl. Medium 3 5 13 -31 BHM Birmingham-Shuttlesworth Intl. Small 2 4 12 -38 BNA Nashville Intl. Medium 4 6 12 -31 BOS General Edward Lawrence Logan Intl. Large 6 7 18 -4 BUF Buffalo Niagara Intl. Medium 2 4 12 -38 BUR Bob Hope Medium 2 3 11 -46 BWI Baltimore Washington Intl. Large 4 6 13 -27 CLE Cleveland-Hopkins Intl. Medium 3 6 13 -27 CLT Charlotte/Douglas Intl. Large 4 8 18 0 CVG Cincinnati/Northern Kentucky Intl. Medium 4 6 15 -19 DAL Dallas Love Field Medium 3 4 10 -46

10 DAY James M Cox Dayton Intl. Small 3 5 13 -31 DCA Ronald Reagan Washington National Large 3 5 16 -19 DEN Denver Intl. Large 6 8 14 -15 DFW Dallas/Fort Worth Intl. Large 7 9 15 -8 DTW Detroit Metropolitan Wayne County Large 6 9 19 8 EWR Newark Liberty Intl. Large 3 9 22 19 FLL Fort Lauderdale/Hollywood Intl. Large 2 5 16 -19 GYY Gary/Chicago Intl. Non-hub 2 4 11 -42 HNL Honolulu Intl. Large 4 6 13 -27 HOU William P Hobby Medium 4 6 9 -42 HPN Westchester County Small 2 5 13 -31 IAD Washington Dulles Intl. Large 4 7 16 -12 IAH George Bush Intercontinental Large 5 8 17 -4 IND Indianapolis Intl. Medium 3 6 13 -27 ISP Long Island MacArther Small 4 4 9 -50 JAX Jacksonville Intl. Medium 2 5 13 -31 JFK John F Kennedy Intl. Large 4 10 27 42 LAS McCarran Intl. Large 4 6 14 -23 LAX Los Angeles Intl. Large 4 9 15 -8 LGA La Guardia Large 2 7 24 19 LGB Long Beach Daugherty Field Small 5 5 13 -31 MCI Kansas City Intl. Medium 3 5 11 -38 MCO Orlando Intl. Large 4 7 13 -23 MDW Chicago Midway Intl. Large 5 6 11 -35 MEM Memphis Intl. Medium 4 7 16 -12 MHT Manchester Small 2 4 12 -38 MIA Miami Intl. Large 4 8 16 -8 MKE General Mitchell Intl. Medium 5 5 13 -31 MSP Minneapolis-St Paul Intl Large 4 7 17 -8 MSY Louis Armstrong New Orleans Intl. Medium 2 5 11 -38 OAK Metropolitan Oakland Intl. Medium 4 6 10 -38 OGG Kahului Medium 2 6 8 -46 OMA Eppley Airfield Medium 3 4 12 -38 ONT Ontario Intl. Medium 2 4 10 -46 ORD Chicago O’Hare Intl. Large 8 9 16 -4

11 PBI Palm Beach Intl. Medium 3 4 13 -35 PDX Portland Intl. Medium 3 4 11 -42 PHL Philadelphia Intl. Large 4 6 19 -4 PHX Phoenix Sky Harbor Intl. Large 3 7 15 -15 PIT Pittsburgh Intl. Medium 4 5 14 -27 PSP Palm Springs Intl. Small 2 5 12 -35 PVD Theodore Francis Green State Small 2 4 12 -38 RDU Raleigh-Durham Intl. Medium 3 5 14 -27 RFD Chicago/Rockford Intl. Non-hub 2 4 10 -46 RSW Southwest Florida Intl. Medium 1 5 12 -35 SAN San Diego Intl. Large 1 4 13 -35 SAT San Antonio Intl. Medium 3 4 11 -42 SDF Louisville Intl-Standiford Field Small 3 5 12 -35 SEA Seattle-Tacoma Intl. Large 3 6 15 -19 SFO San Francisco Intl. Large 4 6 17 -12 SJC Norman Y Mineta San Jose Intl. Medium 3 4 10 -46 SJU Luis Munoz Marin Intl. Medium 2 6 13 -27 SLC Salt Lake City Intl. Large 4 6 18 -8 SMF Sacramento Intl. Medium 2 4 10 -46 SNA John Wayne Airport-Orange County Medium 2 6 12 -31 STL Lambert-St Louis Intl. Medium 4 5 11 -38 SWF Stewart Intl. Non-hub 2 5 14 -27 TPA Tampa Intl. Large 3 5 12 -35 TUS Tucson Intl. Small 3 4 11 -4 a Percent increase/decrease when compared to the current default value. 3.1.2 Default Taxi Speed When AEDT computes aircraft taxi/idle times for the purpose of performing dispersion analysis, model users are instructed to input an airport-specific value (or use a default value) for the aircraft taxi speed. Under this selection, the model performs Delay and Sequence Modeling which uses airport operational schedules, runway configurations, and the airport’s capacity to estimate taxi/idle times and delay periods. In its current configuration, the AEDT default aircraft taxi speed is set to 15 knots (17.26 miles-per-hour (mph)) for these computations. For the purposes of this assessment, the Research Team evaluated available Flight Data Recorder (FDR) data for a number of aircraft types. Notably, the FDR dataset is similar, but not identical, to the dataset used for ACRP Project 11-02 /Task 8 (Enhanced Modeling of Aircraft Taxiway Noise – Scoping). Among the 100+ available parameters contained in the FDR dataset is the aircraft groundspeed data during taxi operations.

12 Listed in Table 5, these FDR groundspeed data were evaluated for seven high-bypass turbofan jet aircraft engines from four engine families, which included the following:  Rolls-Royce Trent series,  Rolls-Royce RB211 series,  PW4000 by Pratt & Whitney, and the  CFM56 by CFM International. As shown in Table 5, the FDR data for these “Test Data Engines” (representing approximately 68,000 taxi in and 111,000 taxi out events) indicate that the average unimpeded aircraft taxi in and taxi out aircraft ground speeds are 13 knots (14.96 mph) and 10 knots (11.51 mph), respectively. Importantly, when aircraft were stationary (i.e., when groundspeed was equal to zero), these data were not considered. Table 5 FDR Unimpeded Aircraft Taxi Speeds (knots) Engine Taxi In Taxi Out No. of Samples Average Standard Deviation No. of Samples Average Standard Deviation Trent 772 389 13 9.0 641 10 5.4 RB211-535E4 7,079 13 6.7 12,101 9 4.4 PW4168A 15,382 12 7.3 25,162 10 6.2 CFM56-5C4 14,564 12 6.8 21,367 9 5.9 CFM56-5B5/P 10,675 14 7.8 16,917 10 7.0 CFM56-5B4/2P 13,835 14 7.8 23,131 9 6.1 CFM56-5B1/2P 6,324 13 7.4 11,944 10 6.8 All Operations 68,248 13 7.4 111,263 10 6.2 The relevance of these “Test Data Engines” to the aircraft fleet being operated in the U.S. by international and U.S. domestic airlines is discussed in Appendix B of this document. Based on the outcome of this analysis, two options were developed that, if implemented, could improve the simulation of aircraft taxi speeds in AEDT, thereby improving the accuracy of an emissions inventory computed using these data. These options are described below. Taxi Speed Options  Option 1 – As stated above, currently the AEDT default aircraft taxi in and out speed is 15 knots (17.26 mph). Based on FDR data (Table 5), a weighted average speed of 11 knots (12.66 mph) could replace the current value. This change would decrease predicted emissions associated with unimpeded taxi times by 27 percent, as compared to the current AEDT.  Option 2 – Under this option, AEDT users would indicate (or the model would distinguish) whether a taxiway is used for aircraft that are taxiing in, taxiing out, or both. For example, if a taxiway is used by aircraft arrivals, the model would default to an aircraft speed of 13 knots (14.96 mph). Similarly, if a taxiway is used for departures, the model would default to a speed of 10 knots (11.51 mph). In this case, the revision in aircraft speeds would also decrease the predicted non-delayed, taxi-related emissions, with the level of decrease dependent on the

13 number and length of the taxiways and whether they are assigned for arrivals, departures, or both. 3.2 Fuel Flow Rates (FFRs) The FFRs currently used by AEDT during the idle/taxi phase of an LTO cycle are those listed in engine- specific ICAO certification sheets. These rates nominally correspond to an engine thrust equal to seven percent of each engine’s maximum rated thrust. Several studies, however, have demonstrated that actual FFRs can differ significantly from the ICAO values. Besides the direct impact of these deviations on calculated emissions (the 2nd term in TIM × FFR × EI equation), there are also indirect impacts due to the dependence of EIs on FFR. This is especially the case for CO and VOCs EIs, which increase measurably with corresponding decreases in thrust settings. The Research Team consulted the following sources of information that discuss/document FFRs during actual aircraft operations:  The FDR data discussed previously in this document that includes data for 1,800 aircraft operations of aircraft equipped with engines from four engine families; 6  Published papers that document FFRs established directly from FDR data or indirectly by comparing EIs observed downwind of aircraft to those determined during staged engine tests; and  JETS/Aircraft Particle Experiment2 (APEX2) and APEX3 studies data.7 The FFR data from the above sources are summarized in Table 6. Notably, the data emphasis is on FFRs rather than engine thrust setting because, as stated previously, FFRs are one of the three fundamental elements within the current AEDT framework that are involved in the calculation of aircraft engine emissions during the taxi/idle mode. FFRs are also used by the Research Team to suggest adjustments to pollutant and pollutant precursor EIs (discussed in the next section). Additional data/information (e.g., histograms of FFRs from the FDR dataset) are included in Appendix C. 6 The relevance of these test data engines to the fleet of aircraft being operated in the U.S. by international and domestic airlines is discussed in Appendix B of this document. 7 JETS/APEX2 and APEX3 were studies of aircraft emissions sponsored by the National Aeronautics and Space Administration (NASA), the Federal Aviation Administration, the California Environmental Protection Agency (EPA), and the California Air Resources Board, conducted by researchers from the U.S. EPA, NASA, Aerodyne Research, Inc., the University of California- Riverside, University of Missouri-Rolla, and Arnold Air Force Base.

14 Table 6 Fuel Flow Rates (FFRs) Method Study Engine (Airframe When Provided) FFR/ICAO FFR (Percent) FDR Data ACRP 02-45 CFM56-5B1/2P 91 (in), 94 (out) CFM56-5B4/2P 87 (in), 89 (out) CFM56-5B5/P 111 (in), 113 (out) CFM56-5C4 85 (in), 91 (out) PW4168A 95 (in), 98 (out) RB211-535E4 80 (in), 84 (out) Trent772 87 (in), 85 (out) ACRP 02-03a, Figure V-4 of Final Report (ACRP Report 63) CFM56-5B4-2 (A320) 90 (83 percent of time 0.1 kg/s, 17 percent of time 0.15 kg/s) ICAO = 0.121 Turgut et al, 2013 CFM56-7B 89 Khadilkar 2012, based on 2300 flights CFM56-5 (A319) 82 CFM56-5 (A320) 92 CFM56-5 (A321) 92 GE CF6-80E1A4 (A330- 202) 91 RR Trent 772B-70/772C- 60 (A330-243) 87 Trent 553 (A340-500) 74 ARJ85 100 RB211 (B757) 94 GE90/PW4084/Trent877 (B777) 98 Comparison of Staged and in-use NOx EIs JETS/APEX2, Wood et al. 2008 CFM56-3B1, -7B22 ~80 to 85 percent (taxi/idle), >100 percent during acceleration Comparison of Stated and in-use HCHO EIs JETS/APEX2, Herndon et al. 2009 CF6-50C2, -80C2A5F; CFM56-7B22, -7B26, - 3C1; JT8D-15; V2527- A5 HC EI 1.5 – 2.2 times higher at real-world operation compared to 7 percent Staged Engine Tests, FFR JETS/APEX2 CFM56-7B22 67 (aircraft 1), 79 (aircraft 2)

15 Read from Cockpit CFM56-3B1 84 CFM56-3B2 87 APEX3 CFM56-3B1 73 (aircraft 1), 88 (aircraft 2) AE3007-A1E 106 AE3007-A 110 PW-4158 NA RB211-535E4-B 61 (aircraft 1), 73 (aircraft 2) Average Normalized FFR 92 Rather than list the absolute FFRs (in kg/s), the values in Table 6 provide the ratio of the actual FFR to the engine-specific ICAO idle/taxi FFR. An important distinction between the FDR data and the JETS/APEX2-3 listings relates to the scope of the two sets of data. The FDR data can be considered “complete” for the engines studied, in that the full distribution of FFRs (comprising operations while stationary, while accelerating, and while moving at a constant speed, etc.) is accounted for in the average. These distributions of rates are provided in Appendix C. In contrast, the data from the JETS/APEX2 and APEX3 studies only reflect the FFR used during “ground idle” operation, and thus do not reflect FFRs at any other setting (e.g., acceleration). For these reasons, the average normalized FFR listed was calculated using only the FDR. The average normalized actual FFR for the engines in the ACRP 02-45 FDR dataset is approximately 92 percent of the ICAO FFR. As extensive as this list is, there are some important data gaps. For example, there are no entries for several types of engines (e.g., CF6, PW-x, etc.). Furthermore, it is not known how representative these data are of the seasonal dependence of emissions, since the FDR data evaluated for the purpose of the ACRP 02-45 project were only acquired during the spring and fall. Use of air conditioning or de-icing also increases the FFR due to bleed air demand and may be unaccounted for in this dataset.8 From the Research Team’s evaluation and analysis of actual rather than test rates, three options were developed to improve the FFRs in AEDT: Fuel Flow Rate Options  Option 1 – Use the engine-specific FFR adjustment only for the engines listed in Table 6.  Option 2 – Use the engine-specific FFRs for the test data engines and engines determined to have “like” fuel flows and EIs (discussed and listed in Appendix B).  Option 3 – Use a single, global adjustment to all commercial jet engines based on the average normalized FFR of approximately 92 percent, as reported in Table 6. 3.3 Emission Indices (EI) Similar to FFR, the EIs that are used to calculate emissions of CO, total hydrocarbons (HC), and NOx are from the ICAO certification data for each engine. These EIs are affected by FFR and ambient conditions (especially temperature) and respond differently, depending on the pollutant or pollutant precursor. Since the vast majority of total CO and HC emissions occur during the idle/taxi phase of an LTO, use of appropriate EIs during idle/taxi is considered important for accurately calculating emissions. 8 The FDR data was collected during the months of April and October.

16 The Research Team reviewed studies that address the dependence of CO, HC, and NOx EIs on FFR and ambient temperature. Among others, this dependence has been studied as part of the following projects: APEX (1), JETS/APEX2, APEX3, ACRP Project 02-03a (Measurement of Gaseous HAP Emissions from Idling Aircraft as a Function of Engine and Ambient Conditions). Figure 1 summarizes the CO and HC results for the CFM56-7B22 engine from ACRP 02-03a, a study that focused on the CFM56-7B family of engines. Calculated EIs/temperature adjustment factors (defined below) to the ICAO EIs are plotted on Figure 1, based on ambient temperature. Six variants in the FFR are provided, expressed in both absolute values (kg/s) and as percentages of the ICAO idle/taxi FFR for the CFM56-7B22 engine (0.105 kg/s). Using the values in Figure 1, the ICAO EIs in AEDT for CO and HC would be adjusted as shown below: EI (temperature, FFR) = Adjustment factor × EI (ICAO idle) The following provides an example of how the EI for HC emitted from a CFM56-7B22 engine would be adjusted using the values in Figure 1.  A CFM56-7B22 engine is operating at 0.095 kg/s and the ambient temperature is 283 K (10 °C). At this FFR, the engine is idling at 90 percent of the ICAO FFR for seven percent thrust. In Figure 1, the adjustment factor is obtained from the blue line with triangles (i.e., the 90 percent line). At the ambient temperature of 283 K, the adjustment factor is 1.8. The ICAO HC EI for the CFM56-7B22 is 2.5 g/kg. Therefore, the adjusted HC EI is 4.5 g/kg (2.5 g/kg x 1.8 = 4.5 g/kg). Figure 1 CFM56-7B22 Fuel Flow/Temperature Adjustment Factors for CO and HC EIs From the Research Team’s evaluation and analysis of actual rather than test rates, three options were developed to improve the CO and HC EIs in AEDT: CO/HC EI Options  Option 1 – Use a global FFR adjustment factor and assume that all engines’ (CFM56s, RB211s, CF6s, etc.) CO and HC EIs follow the same temperature and FFR dependence as the CFM56-7B family of engines. In other words, assume that all engines operate at the same average normalized FFR (approximately 92 percent of the ICAO idle FFR), and that the CO and HC EIs can be adjusted based on this FFR just as they can for the CFM56-7B engines.  Option 2 – Apply an engine-specific FFR adjustment factor (using values from Table 6), and then use the appropriate curve from Figure 1, again assuming the EI adjustment factor is the

17 same as that observed for the CFM56-7B family of engines. For engines that idle at intermediate FFRs (e.g., 88 percent of the ICAO FFR instead of 85 percent or 90 percent), interpolated traces have been added to the figure in Appendix C. These interpolated lines are based on the observed FFRs from the FDR dataset. Explicit formulas for the specific engines from the FDR dataset are also listed in Appendix C.  Option 3 – Only apply adjustment factors to CFM56 family of engines. In contrast to CO and HC, NOx EIs increase with FFR and ambient temperature. In AEDT, the ICAO NOx EIs are adjusted to ambient conditions from the standard ICAO conditions of 15°C, one atmosphere pressure using the Boeing Fuel Flow Method 2. The reduced FFRs observed in actual aircraft operation (summarized in Table 6) warrant a concomitant reduction in the NOx EI. The relationship between FFR and NOx EI is approximated as being linear as demonstrated in Figure 2, which displays the relative NOx EI at “ground idle” versus the relative FFR during the JETS/APEX2 and APEX3 engine tests. Relative NOx EI is computed by the ratio NOx EI (ground idle) / NOx EI (ICAO idle setting). Relative FFR is computed by the ratio FFR (actual) / FFR (ICAO idle setting). The blue dots are the JETS/APEX2 and APEX3 data, for the following engines: CFM56-7B22, CFM56-3B1, CFM56-3B2, and RB211-535E4-B. The grey line is for comparison and has a slope of 1. Figure 2 Relationship of NOx Emissions to FFR From the Research Team’s evaluation and analysis, the following model improvement option was developed to improve the NOx EIs in AEDT:  NOx EI Option – In addition to applying the ambient condition (temperature and relative humidity) adjustment to NOX EIs according to the Boeing Fuel Flow Method 2, an additional FFR adjustment would be applied as follows: Adjusted EINOx = (FFR/FFRICAO) × EINOx (ICAO idle) Where:  FFR/FFRICAO = Ratio of actual FFR to ICAO FFR and  EINOx (ICAO idle) = ICAO Idle NOx EI

18 For example, consider a CFM56-7B22 operating at 0.095 kg/s. The ICAO FFR is 0.105 kg/s. The EI adjustment factor is equal to FFR/FFRICAO = 0.095/0.105 = 0.9, and so the ICAO NOX EI of 4.5 g/kg is reduced to 4.05 g/kg. Note that the final NOX emission rate (in g/s) - which is equal to the product of the FFR and the NOX EI - is 81 percent of the ICAO emission rate, since both the FFR and the EI decrease to 90 percent of their original values (and 0.9 × 0.9 = 0.81). Importantly, although there is only one option to adjust emissions of NOx, to what engines it would be applied would depend on which FFR adjustment was assumed (see Section 3.2 of this Report). 3.4 Additional Considerations In addition to the assessment of EIs to improve how AEDT computes aircraft taxi/idle emissions, the ACRP 02-45 research also addressed how the model could be improved to include consideration of some aircraft operational parameters that could reduce emissions, including reduced engine taxiing and alternative-power aircraft taxi systems. Also considered were potential improvements that could enhance AEDT use for dispersion analyses of aircraft emissions. Each of these is discussed further below. 3.4.1 Reduced Engine Taxiing As discussed in the Task 3: (White Paper) of the Task 1 Literature Search results (see Appendix A), many airlines promote the practice of taxiing with less than all of an aircraft’s engines operating as a method of reducing fuel-burn (and emissions). Although this practice is generally referred to as “single engine taxiing,” it can involve more than one of an aircraft’s engines. For this reason, the practice is referred to in this document as “reduced engine taxiing.” Notably, reduced engine taxiing is performed only at the discretion of an aircraft’s pilot. Factors considered by pilots include meteorological conditions (if wet, the surface of the taxi path could be slippery) and the need for full power for runway crossings. Very few documents address the extent to which reduced taxi procedures are in use or provide data to indicate the percentage of time or frequency when aircraft are taxied with fewer than all engines. One source of information is a document published from the ACRP 11-02 project (Task 8: Enhanced Modeling of Aircraft Taxiway Noise). This document provides the percentage of time each engine on that project’s test data aircraft was operating. The percentages of time were reported for periods when the aircraft were both stationary and moving and when they were arriving and departing. Most of the aircraft evaluated for this project were equipped with two engines (e.g., A319, B757). Because it is not known how many samples were considered in the evaluation of aircraft with more than two engines, these aircraft were not included in this evaluation. When considering just the two-engine jets and an equal amount of time when the aircraft were stationary and moving, pilots taxied aircraft in with just one engine operating an average one percent of the time and taxied out with just one engine operating an average eight percent of the time. Presently, AEDT does not provide a direct method of considering any reduction in taxi-related emissions that would be realized by the practice of reduced engine taxiing (there are ways to do so, but they are time consuming and data intense). As such, the following model improvement was considered by the Research Team to allow such an option:  Reduced Engine Option – Allow users to indicate whether reduced engine taxiing procedures should be considered in the calculation of emissions from jet aircraft equipped with two (or more) engines. If they should be considered, apply emission reduction factors to the calculated emissions. For taxi in operations, the factor would be a minimal value of 0.995 (a 0.5 percent reduction). For taxi out operations, the factor would be 0.96 (a four percent reduction).

19 3.4.2 Alternative Taxi Systems Recognizing that the reduced use of an aircraft’s engines on the ground could result in significant fuel savings, at least two manufacturers/joint ventures are developing methods to taxi aircraft on the ground with no engines operating. Honeywell, and the aerospace firm Safran, have been working together to develop an electric motor that would power an aircraft’s main wheels, the motor being powered by an aircraft’s auxiliary power unit (APU). The Honeywell/Safran product is known as the Electric Green Taxiing System (EGTS) and it is reported by the joint venture that the product will be available in 2017.9 Wheeltug, a system being developed by a subsidiary of Borealis Exploration Limited, involves an electric motor in the hub of the nose wheel.10 This motor would also be powered by an aircraft’s APU. As recently reported by Wheeltug, they expect to begin installing their system onto aircraft belonging to eleven airlines, including six flagship carriers, in the near term.11 As stated above, very little documentation is available addressing reduced engine taxiing. Even fewer address the emissions benefit, or provide data to support the emissions benefit of these alternative taxi systems. Currently, AEDT does not provide direct methods to consider/account for any reduction in taxi-related emissions that would be realized by the use of alternative taxi systems. As such, the following improvement was considered by the Research Team to allow for such a consideration:  Alternative Taxi Option – Because the benefits of these systems are not yet well defined but data may be available in the near future, allow users to enter a percentage by which taxi-related emissions would be reduced to account for the type of alternative taxi system in use at an airport. 3.4.3 Airfield Emission Distribution (Dispersion Analysis Only) When preparing an emissions inventory, the emission totals are not location-specific; thus, it is not necessary for an analyst to know, or for the computer model to simulate, where delay periods (or delay areas) occur on an airfield. However, when performing dispersion analysis, the location at which emissions are generated, as well as the distance between the emission source(s) and a receptor (i.e., locations for which the model will derive estimated concentrations of pollutants), will have a direct impact on the modeled pollutant concentrations. When performing dispersion analysis, the AEDT Delay and Sequence Modeling is used to simulate taxi operations. The Delay and Sequence Modeling takes into account aircraft operational schedules, active runway configurations, and the delays that are associated with airport capacity. To use the Delay and Sequence Modeling, analysts must define the location(s) of receptors, airport gates, taxiways, runways, taxi paths (i.e., defined taxiway/runway connections), and runway configurations (i.e., weather dependent runway usage). The following was considered by the Research Team as an option to more realistically model aircraft taxiing at an airport:  Airfield Emission Distribution Option – The modeling of aircraft queues (i.e., areas at which aircraft are delayed) are only considered for departing flights. All arrivals are assumed to have unimpeded taxi conditions from an airport’s runway to a gate. In reality, and for reasons that may include an insufficient number of gates, arrivals could be “held” (i.e., delayed) in ramp areas before parking at a gate. To better define these areas, the computer model could be modified to provide an option for analysts to define where areas of delay occur and the extent of the delay. 9 Electric Green Taxiing System overview - www.greentaxiing.com/overview.html 10 Wheeltug - www.wheeltug.gi/ 11 Royal Aeronautical Society, Toulouse, 13 September 2013. Delivered by Isaiah Cox, WheelTug Chief Executive Officer