Evaluation of Pushback Decision-Support Tool Concept for Charlotte Douglas International Airport Ramp Operations

Transcription

1 Evaluation of Pushback Decision-Support Tool Concept for Charlotte Douglas International Airport Ramp Operations Miwa Hayashi, Ty Hoang, Yoon Jung NASA Ames Research Center Waqar Malik, Hanbong Lee Univ. of California Santa Cruz, NASA Ames Research Center Victoria Dulchinos San Jose State Univ. NASA Ames Research Center 11th USA/Europe ATM R&D Seminar June 23-26, 2015

2 1. Background 2. SARDA Problem Spot and Runway departure Advisor (SARDA) Solution Previous SARDA Studies Objectives of This Study 3. Simulator Evaluation 4. Results 5. Conclusion

3 Problem: Airport Traffic Management Inefficiency Lack of coordination The FAA, ramp, airlines First come, first served basis Pushback, spot release, takeoff Long runway queues Many stops & waits while engines are on Excess fuel burn Congestion in ramp, taxiways Poor predictability Unused slots Photo Credit: Simon_sees via Compfight cc 3

4 Spot and Runway Departure Advisor (SARDA) Solution Better coordination through sufficient data exchange Intelligent departure metering that shifts taxi delay to the gate. Photo Credit: Sifter via Compfight cc 4

5 Previous SARDA Studies SARDA-DFW (2010, 2012) Tower controller tool Dallas/Ft. Worth (DFW) airport model 2 major human-in-the-loop simulations Significant SARDA benefits demonstrated: Taxi delay reduction by 60% Fuel reduction by 33% Unrealistic ramp traffic simulation SARDA-CDM (2012) Collaborative Decision Making (CDM) 5

6 Pushback advisory Spots FAA Airport Tower Movement Area SARDA-DFW Ground controller advisory Local controller advisory Ramp Tower Ramp The Entire Airport + the National Airspace SARDA-CDM Target pushback time, target spotrelease time window assignments Ground & Local controller advisories (SARDA-DFW) 6

7 Objectives of This Study Propose a pushback metering Decision Support Tool (DST) for ramp controllers. Evaluate the tool in high-fidelity Human-in-the-Loop airport traffic simulation. 7

8 1. Background Departure Flow 2. SARDA Spot Release Planner (SRP) SARDA-CLT 3. Simulator Evaluation 4. Results 5. Conclusion

9 Departure Flow Boarding Pushback Taxi Metered Traffic Volume Spot Taxi Wheels Up Airborne Pilot: Ready to push Ready to taxi Ramp: Push approved Pushback Metering Proceed to Spot Ramp Movement Area 9

10 FAA Airport Tower Spot Release Planner (SRP) 2 Ramp Tower Spot-Release Time & Pushback Time 1 Runway-Use Sequence 10

11 Spot Release Planner (SRP) Airport snapshot Scheduled pushback times Arrivals STA Separation constraints Aircraft parameters Taxi Estimator Estimated taxi times Required separations for each pair of flights Runway Scheduler (Mixed Integer Linear Program) Optimal runway sequence Minimize the total delay for all aircraft. Pushback Estimator Spot-release times Pushback time advisories Repeated every 10 seconds 11

12 SARDA-CLT Tailored for the American Airlines Charlotte Airport (CLT) ramp tower operations. Based on SARDA-CDM. 1. No tower controller advisory. 2. No target time assignment to airlines (pushback, spot). Photo Credit: Drewski2112 via Compfight cc 12

13 SARDA-CLT Ramp controller pushback advisory Ramp Tower Data Sharing Ramp FAA Airport Tower SARDA-CLT runs the full SRP. Has all the hooks for future CDM integration. 13

14 1. Background 2. SARDA 3. Simulator Evaluation CLT & CLT Ramp Ramp Tower Simulator Facility Ramp Traffic Console (RTC) Traffic Scenarios Experiment design 4. Results Measurements 5. Conclusion

15 Arrival 18R 18C CLT Ramp Tower South Flow 18L Westbound Departure Arrival 23 Eastbound Departure 15

16 CLT Ramp North Sector L Ramp Tower East Sector 8 7 West Sector South Sector 6 Source: 16

17 Ramp Tower Simulator Facility High-fidelity ramp-tower simulator at NASA Ames 360-degree out-the-window view projection Simulated radio communication with Pseudo Pilots. Ramp Manager North East South West Researcher Station 17

18 Ramp Traffic Console (RTC) Developed for this SARDA simulation. Replaces the paper strips currently used at the CLT ramp Enables dynamic advisory updates Displayed on a 27 touchscreen monitor. Shows: Movable, zoomable map Virtual strips Radar position readings (as aircraft icons) Traffic Management Initiative constraints SARDA-CLT pushback advisories 18

19 Pushback Advisories Pushback anytime Hold at the gate for 6 minutes (Gate Hold) (Gate Hold) 7 min 21 sec to the advised pushback time 1 min 33 sec past the advised pushback time 19

20 Number of Operations (Bins: [t-10sec, t+10sec]) Traffic Scenarios Scenario 1 Scenario 2 Departure Arrival 2 traffic scenarios were created. Based on CLT traffic recorded on May 16, 2013 (South Flow). A departure push followed by an arrival push. 1-hour test duration. Traffic Management Initiative (TMI) Expect Departure Clearance Time (EDCT) 15 Miles-in-Trail (MIT) over MERIL Scenario 1 96 departures, 80 arrivals, 5 EDCT Time (Minutes) Scenario 2 84 departures, 72 arrivals, 5 EDCT 20

21 Experiment Design 16 runs in each week, 48 in total in 3 weeks 2 new CLT ramp controllers in each week, 6 in total in 3 weeks Each CLT ramp controller alternated East and South sectors 6 confederate controller participants in 3 weeks 2 ramp controllers, 4 FAA tower controllers Main comparison: Advisory vs. Baseline runs Advisory runs pushback advisories, MC taxiway bypass advisories (see the paper) Baseline no advisory, manual departure metering design in 3 weeks counterbalanced within & between the 2 Controller Seating per week: 2 Advisory conditions (Baseline, Advisory) 2 Scenarios (1, 2) 6 Controller Seating (1 through 6) 21

22 Measurements Objective Data Traffic data Scheduler performance Subjective Data Real-time workload ratings (5 minute intervals) Post-run questionnaire responses NASA Task Load Index ratings (Mental Demand, Time Pressure, Frustration) Teamwork workload ratings (Communication, Coordination) Usability Situation awareness Post-study questionnaire responses 22

23 1. Background 2. SARDA 3. Simulator Evaluation 4. Results Traffic Performance Results Workload Results Discussion: Comparison with SARDA-CDM 5. Conclusion

24 Departure Taxi Delay Boarding Pushback Taxi Spot Taxi Wheels Up Airborne Engine ON Ramp Movement Area Pilot: Ready to push Ready to taxi Ramp: Push approved Proceed to the Spot Unimpeded Taxi time Departure Taxi Delay 24

25 Taxi Delay (Seconds) Week 1 Week 2 Week 3 Baseline Advisory Scenario 1 Mean Departure Taxi Delay Observed taxi time Unimpeded taxi time In Advisory runs: On average, 1 minute reduction per aircraft. Smaller variations. Scenario 2 The 75th, median, & 25th percentiles SARDA advisories did not impact the runway throughput. 25

26 Gate Delay Boarding Pushback Taxi Pilot: Gate Delay Ready to push Ready to taxi Spot Taxi Ramp Movement Area Wheels Up Airborne Ramp: Push approved Pushback Metering Proceed to the Spot 26

27 Gate Delay Gate Delay (Seconds) Week 1 Week 2 Week 3 Actual start of movement Ready for pushback In Advisory runs, On average, 90 seconds gate delay. The 30 seconds additional gate delay caused by the Other Airlines with an MIT? Baseline Advisory Scenario 1 Mean Scenario 2 75th, median, & 25th percentiles 27

28 Fuel Burn Average Departure Fuel Usage per Run Scenario 1 Scenario 2 Baseline kg kg Advisory kg kg Reduction (kg) kg kg (%) 12.0% 10.4% ICAO s engine emission certification data % of fuel saving when SARDA was running. 28

29 Average Departure Count Departure Count Advisory Baseline Scenario 1 Departure counts in the Movement Area Advisory runs < Baseline runs. Scenario Scenario Elapsed Time (Minutes) 29

30 Histogram of Takeoff Time Deviation from EDCT Baseline EDCT Conformance Observed takeoff time Assigned EDCT Aircraft Count Advisory Tighter distribution in Advisory runs. Better EDCT conformance Time Difference from EDCT (Seconds) 30

31 Real-Time Workload Ratings Linear Mixed Model repeated-measures analysis: In Advisory runs, the FAA tower controllers ratings were reduced by 0.23 in 7 point scale (p = 0.021). The CLT ramp controllers ratings: CLT Ramp Controllers Real-Time Workload Rating (p = 0.008) Baseline In South Sector: Advisory < Baseline In East Sector: Advisory Baseline East Sector Advisory South 31

32 Discussion: Comparison with SARDA-CDM SARDA-CLT ramp DST saved 10-12% of fuel. Previous SARDA-CDM numerical simulations estimated the achievable departure fuel savings to be 23-30%. Therefore, SARDA-CLT alone demonstrated roughly 30-50% of the estimated fuel-saving benefit. 32

33 Conclusions SARDA-CLT was formulated to provide a ramp-controller DST, based on SARDA-CDM. Pushback advisories Full SRP, which utilizes data sharing with the FAA, for future CDM integration High-fidelity human-in-the-loop simulation demonstrated benefits of SARDA-CLT: On average 1-minute reduction of departure taxi time On average 10-12% fuel reduction per departure bank Improved EDCT conformance Reduced FAA tower controllers workload 33

34 Questions? Photo Credit: happyrel via Compfight cc