Data Driven Methods for Airspace Performance Analysis

Transcription

1 Data Driven Methods for Airspace Performance Analysis Lance Sherry John Shortle Students 1

2 Complexity of Interactions in Network of Distributed Agents Research and Business Opportunities Optimized Stochastic, Capacity-limited Networked Operations Aircraft Basic Aero Propulsion Barnstorming Operations Air Transportation Air Transportation - Mail Air Traffic Control Basic Airport Traffic Control Point-to-Point Scheduled Operations Air Transportation National Air Carriers Point-to-Point Service Inter-modal Air Traffic Control En-route Air Traffic Control Terminal Area Traffic Control Networked Scheduled Operation Air Transportation National/International Network airlines Civil Aviation Board Air Traffic Control Radar Precision Approach Optimized Networked Operations Air Transportation Deregulation Hub monopolies Schedule/Network optimization Overscheduling Yield Management Fuel Management airlines Air Traffic Control Radar Precision Approach Air Transportation Flexible Airline Business Models Low Cost Carriers/Regional Jet Airlines Network configurations (Hub, point-to-point) Air Traffic Control Collaborative Decision Making Revenue/Cost Synchronization Aircraft Self-separation Facility Resizing Safety/Capacity Tradeoff Years 2

3 Big Data Analytics in Air Transportation Latitude/Longitude Surveillance Data Flight Data Aircraft Performance Altitude and (Derived) True Airspeed Procedures Weather Data Model-based Engineering & Big Data Analysis Insights & Understanding for Operators, Planners, & Investors Aircraft Performance Models Automation Behavior Models Automation Behavior Models Validation 3

4 Emission Inventory 4

5 Context Airport Management is required to report Emissions Inventory for aircraft Landing and Takeoff Operations (LTO): 1. Sustainability planning 2. Climate Change Registries 3. Environmental Impact Studies Fuel Air Maximum Emissions generated during Takeoff Phase Gear-Up 100 ft AGL Thrust Reduction Altitude (Takeoff Thrust to Climb Thrust) 1500 ft AGL Accelerate to VR V2 Hold V knots at Constant Rate-of-Climb CnHm+ S + N + O CO + H O + N + O + NOx + CO + SOx + Soot + UHC Products of Ideal Combustion Products of Non-Ideal Combustion 5

6 Context Monitoring by sensors inaccurate due to dispersion effects/prohibitively expensive Inventory Models Mass of pollutants generated ICAO Reference Model Pollutant mass per flight = Number of Engines x Time in Phase of Flight (T) x Fuel Flow Rate (FFR) x Emissions Index (EI) T, FFR, EI averages from ICAO data-base 6

7 Problem Static Inventory Models over-estimate Emissions Inventory Two assumptions: 1. Average Time-in-Phase (assumed 2.9 mins takeoff) 2. Thrust Setting for Takeoff (assumed 100%) Pollutant mass per flight = Number of Engines * Time in Phase of Flight * Fuel Flow Rate * Emissions Index 7

8 Solution Improve accuracy in Emissions Inventory Estimate using 1. High-fidelity track surveillance data 2. Procedure data (i.e. navigation data-base) 3. Aircraft performance model Validated by Flight Data 4. Weather data 8

9 Flight Schedule Big Data Analytics Weather Data Aircraft Performance Model Surveillance Track Data Estimate Takeoff Thrust Setting Aircraft Type Aircraft Type Aircraft Trajectory Estimated Takeoff Thrust Emissions Data Bank Trajectory Analysis (AEDT/NIRS) Avg. Fuel Burn Rate Fuel Burn Rate for each flight Takeoff Thrust Setting Avg Time- in- Phase Time in Phase for each flight No. Engines Emissions Index Fuel Burn Rate Time in Phase Calculate Emissions Inventory (Reference Model) Emissions Inventory (tons) = # Engines Emissions Index Fuel Burn Rate Time in Phase Emissions Inventory Assume 100% Takeoff Thrust Estimated Takeoff Thrust Standalone Takeoff Center for Air Transportation Thrust Systems Models Research (CATSR) at George Mason University (ACRP-04-02) 9

10 Surveillance Track Data Latitude/Longitude V TAS = (VV GG SSSSSS θθ ) (VV WW SSSSSS φφ ) VV GG CCCCCC θθ (VV WW SSSSSS φφ ) Alternate filtering techniques developed Altitude and (Derived) True Airspeed 10

11 Aircraft Performance Equations Total-Energy model: rate of work done by forces acting on the aircraft = rate of change of potential and kinetic energy Rearranging for Thrust Aircraft performance Surveillance track data Weather (i.e. wind) 11

12 Frequency 50% 40% 30% 20% 10% 0% Results Thrust Settings for 1200 departures at ORD µ = 86% Max Takeoff Thrust, σ = 11% % Max Takeoff Weight Thrust Validation: Airline supplied takeoff thrust settings. Range in thrust reduction from 0% to 24 % An average thrust reduction of 13%, standard deviation of 8%. 12

13 Sensitivity Analysis TOW and Headwind (Marginal) Sensitivity to Headwind (~5%) MD-83 (Super MD-80) Sensitivity to Takeoff Weight (~30%) 13

14 Modernization Cost/Benefits Analysis: Metroplex Flow Deconfliction Using RNP Procedures at Midway Airport Akshay Belle 14

15 Metroplex Examples Top 10 Metroplexes in the US Sl.no Metroplex Ops per day # Airports (Year 2012) within 30NM 1 New york Chicago Los Angeles Atlanta District of 4 Columbia Dallas San Francisco Miami Denver Charlotte New York Metroplex 9NM Chicago Metroplex 14NM Total of 21 Metroplexes in the U.S serving metropolitan area that account for: 35% of the nation s population (314 M) (United States Census Bureau, 2012) 44% of the gross domestic product ($15.68 trillion) (U.S. Department of Commerce, 2012) 10 NM 17NM 8.5NM 15

16 Metroplex De-confliction Terminal Airspace Redesign (Spatial Strategy) Before After N N ORD ORD 22L Departures Airspace Conflict 22L Departures No Airspace Conflict 13C ILS MDW 13C RNP MDW Wind Direction Wind vector 16

17 Chicago Metroplex De-confliction MDW - RNP0.3 w/rf Leg approach on to 13C Chicago Metroplex Deconfliction RNP0.3 w/rf Leg approach on to 13C at MDW Safe Vertical Separation Metroplex De-confliction Vertical Profile 17

18 MDW Flows from East and West Flows From East Flows and their respective Track count Abeam Flows From West Sl.no Direction Runway Approach Count 1 ILS C RNP 87 3 Visual L Visual L Visual R Visual 70 7 E 31C ILS Visual R Visual L Visual ILS R Visual ILS C RNP Visual L Visual L Visual R Visual 56 W 19 ILS C 20 Visual R Visual 2 Abeam 22 4L Visual ILS 729 4R 24 Visual

19 TRACON Flow Track Distance & Time Performance Flows From East Track distance and time, lower the better Ranking of flow from the East Abeam - Final waypoint on STAR Flows From West Ranking of flow from the West 19

20 Benefits RNP Approaches for MDW (all Runways) RNP 31 C from West RNP 4R from the East ILS available Joliet (JOL) Chicago Heights (CGT) 20

21 Source of variance in RNP flows MDW- Flows to runway 13C from the east ILS Approach RNP Approach Vectors" up to the Start of the RNP approach (base leg) introduce as much variation in track distance/time as the ILS approaches. 21

22 Airport Surface Operations Analysis Anvardh Nanduri, Kevin Lai 22

23 ATL 2012 Jan Excess Surface Count (10 Min Periods) Day of Month Arr Dep Avg 1 Sigma 2 Sigma 23

24 Annual Surface Ops 3000 Daily Cumulative 10 Minute Surface Counts Excess number of flights Departures Arrivals 3 Sigma 2 Sigma 1 Sigma Average -500 Dates Day of Year 2012 Excess Cum Arrivals Excess Cum Departures Avg 1 Sigma 2 Sigma 3 Sigma 24

25 Daily Cumulative Surface Count ATL 2012 Frequency Daily Cumulatve Surface Count % 90.00% 80.00% 70.00% 60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% Frequency Cumulative % Normal Distribution 25

26 Causal Analysis- Reduced Departure Rate - Mar 13, 2012 (ATL) Runway Config Change 8L, 9R, 10 8R, 9L (35 ops/10 mins ) Runway Config 26R, 27L, 28 26L, 27R (34 ops/10 mins) :00-00:59 00:00-00:59 01:00-01:59 01:00-01:59 02:00-02:59 02:00-02:59 03:00-03:59 03:00-03:59 04:00-04:59 04:00-04:59 05:00-05:59 05:00-05:59 06:00-06:59 06:00-06:59 07:00-07:59 07:00-07:59 08:00-08:59 08:00-08:59 09:00-09:59 09:00-09:59 10:00-10:59 10:00-10:59 11:00-11:59 11:00-11:59 12:00-12:59 12:00-12:59 13:00-13:59 13:00-13:59 14:00-14:59 14:00-14:59 15:00-15:59 15:00-15:59 16:00-16:59 16:00-16:59 17:00-17:59 17:00-17:59 18:00-18:59 18:00-18:59 19:00-19:59 19:00-19:59 20:00-20:59 20:00-20:59 21:00-21:59 21:00-21:59 22:00-22:59 22:00-22:59 23:00-23:59 23:00-23:59 Airport Departure Rate Airport Acceptance Rate Visibility Ceiling/1000 Wind Direction/10 Wind Speed 26

27 Cumulative Surface Counts - Reduced Departure Rate - Mar 13, 2012 (ATL) Number of flights Runway Config Change 8L, 9R, 10 8R, 9L (35 ops/10 mins ) Runway Config 26R, 27L, 28 26L, 27R (34 ops/10 mins) TW1 TW4 TW7 TW10 TW13 TW16 TW19 TW22 TW25 TW28 TW31 TW34 TW37 TW40 TW43 TW46 TW49 TW52 TW55 TW58 TW61 TW64 TW67 TW70 TW73 TW76 TW79 TW82 TW85 TW88 TW91 TW94 TW97 TW100 TW103 TW106 TW109 TW112 TW115 TW118 TW121 TW124 TW127 TW130 TW133 TW136 TW139 TW142 TW145 Cumulative Actual Arr Surface Count Cumulative Sched Arr Surface Count TW Cumulative Actual Dep Surface Count Cumulative Sched Dep Surface Count 27

28 Causal Analysis- Blue Sky - May 18, 2012 (ATL) No Runway Configuration Change 8L, 9R, 10 8R, 9L (36 ops /10 mins) :00-00:59 00:00-00:59 01:00-01:59 01:00-01:59 02:00-02:59 02:00-02:59 03:00-03:59 03:00-03:59 04:00-04:59 04:00-04:59 05:00-05:59 05:00-05:59 06:00-06:59 06:00-06:59 07:00-07:59 07:00-07:59 08:00-08:59 08:00-08:59 09:00-09:59 09:00-09:59 10:00-10:59 10:00-10:59 11:00-11:59 11:00-11:59 12:00-12:59 12:00-12:59 13:00-13:59 13:00-13:59 14:00-14:59 14:00-14:59 15:00-15:59 15:00-15:59 16:00-16:59 16:00-16:59 17:00-17:59 17:00-17:59 18:00-18:59 18:00-18:59 19:00-19:59 19:00-19:59 20:00-20:59 20:00-20:59 21:00-21:59 21:00-21:59 22:00-22:59 22:00-22:59 23:00-23:59 23:00-23:59 Airport Departure Rate Airport Acceptance Rate Wind Speed Visibility Ceiling/1000 Wind Direction/10 28

29 Cumulative Surface Counts Blue Sky - May 18, 2012 (ATL) Number of flights Runway Configuration did not change 8L, 9R, 10 8R, 9L (36 ops /10 mins) TW1 TW4 TW7 TW10 TW13 TW16 TW19 TW22 TW25 TW28 TW31 TW34 TW37 TW40 TW43 TW46 TW49 TW52 TW55 TW58 TW61 TW64 TW67 TW70 TW73 TW76 TW79 TW82 TW85 TW88 TW91 TW94 TW97 TW100 TW103 TW106 TW109 TW112 TW115 TW118 TW121 TW124 TW127 TW130 TW133 TW136 TW139 TW142 TW145 TW Cumulative Actual Arr Surface Count Cumulative Actual Dep Surface Count Cumulative Sched Arr Surface Count Cumulative Sched Dep Surface Count 29

30 Airspace Risk Management - Go Around Stabilized Approaches Zhenming Wang, Houda Kerkoub 30

31 Risk Management Metrics Risk Events Metric Definitions Risk Management Metrics Risk Events and Factors Risk Events Definitions Calculate Risk Events and Factors Derived data Calculate Derived data Processed Observed Data Process Observed Data Navigation Data-base (Procedures) Aircraft Performance Models Raw Surveillance Track Data Weather Data 31

32 Missed Approach 20% Go Arounds Fit Normal (0.007, 0.005) Go Arounds are not measured/reported. Track data used to count and analyze Likelihood of a Go Around Mean = 7/1000 Std Dev = 5/1000 Range = [0 22/1000] Aborted Approach 80% local time 18:00-18:59 18:00-18:5919:00-19:59 quarter hour arrival runway configuration 9R, 22L, 28 9R, 22L, 28 9R, 22L, 28 airport acceptance rate arrival demand ceiling (ft) wind direction S S S wind speed (kts) visability (miles) temperature ( F) on-time arrival % 16.67% 26.32% 16.67% Center for Air Transportation Systems avg. Research arrival delay (CATSR) (min) at George Mason 134 University

33 19 out of 3548 Go-arounds About 5.4 per 1000 flights 33

34 Go Arounds - ASRS Taxonomy Factor % ASRS Reports Airplane Issues 54% Separation Violation 21% Weather 17% Flight/TAC Interaction 8% Runway Issues 5% No Reason provided 6% 34

35 Go Arounds - ASRS Taxonomy 1. Airplane Issue 53.74% 1.1 Unstable Approach 9.52% High and Fast 6.12% Other Approach Issue 3.40% 1.2 Alerts 4.08% 1.3 Onboard failures 40.14% 35

36 Go Arounds ATSAP Taxonomy ATSAP Data, Bayesian Network Model Firdu Bati (Dissertation) 36

37 Stabilized Approach 1000 AGL On Runway Center-line On Glidepath At Landing Speed (V Ref ) At Rate of Descent for Glide-path (< 1000fpm) 37

38 Stabilized Approaches Frequency of risk events from 1000 ft. AGL to runway threshold Fast 38

39 Stabilized Approaches Frequency of risk events from 750 ft. AGL to runway threshold 39

40 Stabilized Approaches Frequency of risk events from 500 ft. AGL to runway threshold Summary table 500 ft. AGL 40

41 Groundspeed Change No change Greater than 10 knots Rate of Descent Within limits Excessive Within limits Excessive 1000 AGL 13C Joliet (JOL) ILS available Chicago Heights (CGT Position with Position with Runway 04R 13C 31C Glidepath Centerline # % # % # % On Glidepath On Runway Centerline % % % Not On Runway Centerline % % % On Runway Centerline % % % Above Glidepath Not On Runway Centerline % % % On Glidepath On Runway Centerline % % % Not On Runway Centerline % % % On Runway Centerline % % % Above Glidepath Not On Runway Centerline % % % On Glidepath On Runway Centerline % % % Not On Runway Centerline % % % On Runway Centerline % % % Above Glidepath Not On Runway Centerline % % % On Glidepath On Runway Centerline % % % Not On Runway Centerline % % % On Runway Centerline % % % Above Glidepath Not On Runway Centerline % % % 4R 31C 20% 26% 8% 41

42 Stabilized Approaches - Factors Weight Class % Flights 750 ft. AGL Change in Speed Average Groundspeed at the Runway Threshold Heavy 20.9% knots B % 129 knots Large 12.0% 132 knots Small 47.0% knots Every approach/runway is different Procedure % dv>10kts from 1000' to THR % dv>10kts from 750' to THR % dv>10kts from 500' to THR ILS 5.97% 1.19% 0.17% RNP 4.23% 0.00% 0.00% VFR 50.15% 12.35% 0.59% 42

43 Big Data Analytics in Air Transportation Latitude/Longitude Surveillance Data Flight Data Aircraft Performance Altitude and (Derived) True Airspeed Procedures Weather Data Model-based Engineering & Big Data Analysis Insights & Understanding for Operators, Planners, & Investors Aircraft Performance Models Automation Behavior Models Automation Behavior Models Validation 43