FAA CENTER OF EXCELLENCE FOR ALTERNATIVE JET FUELS & ENVIRONMENT

Project manager: Hua He and Aniel Jardines, FAA Lead investigators: Hamsa Balakrishnan (MIT) and Tom Reynolds (MIT LL)

Surface Analysis to Support AEDT Aircraft Performance Module Development

Project 46

April 18-19, 2017 Alexandria, VA

Opinions, findings, conclusions and recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of ASCENT sponsor organizations.

2

Motivation

•  Taxi phase in the Aviation Environmental Design Tool (AEDT) is currently modeled using default or user-specified taxi times, coupled with engine idle fuel and emissions assumptions from ICAO Aircraft Engine Emissions Databank

•  These assumptions reduce the accuracy of the taxi performance modeling

•  For some applications, higher fidelity surface modeling approaches may be needed

3

Objectives

•  Identify needs and evaluate methods for improving taxi performance modeling in AEDT in order to better reflect actual operations

•  Develop and validate enhanced taxi models by combining ASDE-X surface track data with statistical models of engine performance (using Flight Data Recorder information)

•  Make recommendations for enhancements to AEDT Aircraft Performance Module (APM) based on knowledge gained

4

Outcomes and Practical Applications

AEDT APM

Total fuel burn

per time-period, per airport

Fuel flow rate profiles

ASDE-X data

Current scope

5

Outcomes and Practical Applications

AEDT APM

Total fuel burn, noise and emissions

per time-period, per airport

Fuel flow rate profiles

Thrust profiles

ASDE-X data

Potential extensions

6

Approach

  AEDT APM

surface model needs

assessment

  Aircraft surface

performance modeling

enhancements

  Aircraft taxi performance

model validation

  AEDT APM

enhancement recommendations

Stakeholder input,

supporting documents & prior research

ASDE-X data

FDR data

Model development data Validation data

7

Schedule and Status

•  Assess AEDT aircraft surface performance modeling needs

•  Develop enhanced aircraft surface performance models

•  Validate enhanced aircraft surface performance models

•  Recommend AEDT APM enhancements

[Sept.-Nov. 2016]

[Oct. 2016-present]

[Jan. 2017-present]

[~Aug. 2017]

8

Recent Accomplishments [1]

•  Assessment of AEDT surface APM modeling needs –  Synthesis of findings from AEDT documentation, ACRP studies

02-45 and 02-27, and stakeholder input •  Airport average taxi-time is either default (7 min taxi-in/ 19 min taxi-out)

or airport-specific for 75 airports –  Default taxi database is outdated –  Analysis in ACRP 02-45 suggests 16 min for taxi-out/ 7 min for taxi-in –  Airport specific taxi-time estimates off by 4-50% [recent ASPM data]

•  Taxiway network model of airport; taxi-path for each flight –  Taxi-time based on taxi-path length and speed (AEDT default: 15 knots)

»  ACRP report suggests ~13 knots –  Taxi-in time is considered to be unimpeded; taxi-out time includes a queuing

delay at the runway. Queuing delay is sum of runway occupancy times of aircraft already in the queue

»  Queues are assumed only at the runway; effects of acceleration not included

»  Deterministic runway occupancy times. Variation by aircraft type/weather condition not known

•  No surface-specific regression model for fuel flow rates

9

Recent Accomplishments [2]

•  Assessment of AEDT surface APM modeling needs

•  Identified needs: –  Models that are representative of a wide range of taxi

conditions, aircraft types, airports, airlines, and weather conditions

–  Identify key locations and events at major airports (e.g., non-movement area, taxiway intersections, spot and departure runway queues, runway crossings, etc.)

–  Need to model operational variability •  Assess uncertainty associated with fuel burn estimates •  Evaluate sensitivity to various factors (e.g., takeoff mass,

ambient conditions) –  Need data-driven validation of models

10

Recent Accomplishments [3]

•  Surface APM model enhancements –  In order to determine the “full” fuel burn

profile, need to consider both the pre-ASDE-X (frequently the non-movement area) portion and the ASDE-X portion of surface operations

–  The pre-ASDE-X portion may include engine start and ramp area movements

–  Can vary by airport; needs characterization by airport

–  For the ASDE-X portion, prior work suggests that the taxi time and number of acceleration events are significant

Extract taxi time & # of

accelerations Fuel burn estimate

Regression model

ASDE-X

Gate

Enginestart

Taxi

Push-back

FDR dataASDE-X data

Pre-ASDE-X contribution

11

Recent Accomplishments [4]

•  Initial identification of Non-Movement Area (NMA) & “pre-ASDE-X” area (PAXA) for US airports in FDR data sets of European carrier (BOS, JFK, EWR, MIA, ORD, LAX) –  Determine NMA (based on airport markings) & PAXA (based on

ASDE-X data analysis) polygons for each of the US airports

BOS JFK

MIA ORD

NMAPAXA

12

Recent Accomplishments [5]

•  Determining PAXA regions

ASDE-X Archives

Select airport, analysis days & plot

data

Generate coverage grid

Identify PAXA polygon based on interior edges of largest cluster

Latit

ude

Longitude

Latit

ude

Longitude

e.g. MIA

e.g., MIA

13 750 800 850 900 950Time (s)

-2

0

2

4

6

8

10

Spee

d (m

/s)

DataFilterSmoother

Recent Accomplishments [6]

•  Smoothing/filtering algorithms for FDR and ASDE-X data –  Need to “match” events (e.g., turns and accelerations) –  Resolution and noise characteristics vary between datasets –  Infer ground truth data (equivalent tracks) for algorithm validation

100 200 300 400 500 600 700Time

0

2

4

6

8

10

12

Spee

d (m

/s)

DataFilterSmootherASDE-XFDR

(Not the same flight)

14

Summary

•  Development and validation of enhancements to AEDT surface APM

•  Next steps –  Model range of taxi conditions, aircraft types, airports, airlines, and

weather conditions –  Analyze ASDE-X data from major airports to identify key locations

and events (e.g., non-movement area, taxiway intersections, spot and departure runway queues, runway crossings, etc.)

–  Model operational variability and evaluate sensitivity –  Develop and validate models by synthesizing data from

•  2006 FDR (European carrier) •  2016 FDR (European carrier, including 123 arrivals and 368 departures

at six US airports) •  Aggregate (i.e., total over taxi phase) fuel use data from A4A •  2016 ASDE-X data

15

References

•  ACRP 02-45, “Methodology to Improve EDMS/AEDT Quantification of Aircraft Taxi/Idle Emissions”, Transportation Research Board, 2016.

•  ACRP 02-27, “Aircraft Taxi Noise Database for Airport Noise Modeling”, Transp. Research Board, 2013.

•  Chati, Y.S., and Balakrishnan, H., “Analysis of Aircraft Fuel Burn and Emissions in the Landing and Takeoff Cycle using Operational Data," Intl. Conf. on Research in Air Transportation, Istanbul, 2014.

•  DOT-VNTSC-FAA-16-11, “Aviation Environmental Design Tool – Technical manual, Version 2b”, U.S Department of Transportation (FAA), 2016.

•  H. Khadilkar and H. Balakrishnan, “Estimation of Aircraft Taxi Fuel Consumption using Flight Data Recorder Archives”, Transportation Research Part D: Transport and the Environment, 17, (7), 2012.

•  Särkkä, Simo. "Unscented Rauch-Tung-Striebel Smoother." IEEE Transactions on Automatic Control 53.3 (2008): 845-849.

•  Lopez, Remy, and Patrick Danes. "Low-Complexity IMM Smoothing for Jump Markov Nonlinear Systems." IEEE Transactions on Aerospace and Electronic Systems (2017).

Contributors •  PIs: Hamsa Balakrishnan and Tom Reynolds •  MIT Lincoln Laboratory: Emily Clemons •  MIT Students: Sandeep Badrinath, Yashovardhan Chati •  FAA: Chris Dorbian

16