Predicting Arrival Time Uncertainty from Actual Flight ... · 6/12/2015  · AMAN: Extending the Horizon . 4 19 busiest connections from Amsterdam (2013) Stockholm . Milan . AMAN:

1

Predicting Arrival Time Uncertainty from Actual Flight Information

Maarten Tielrooij m.tielrooij@tudelft.nl

2

Flow Management Tactical ATC

24 hours 3 hours 20 minutes 0 min

boarding & departure en-route arrival

ATC

Airline

AMAN

1 hour

AMAN: Extending the Horizon

3

Surveillance error

Schedule Deviation

Weather forecast error

Performance model error

Trajectory prediction modeling error

AMAN: Extending the Horizon

4

19 busiest connections from Amsterdam (2013)

Stockholm

Milan

AMAN: Extending the Horizon

5

Aircraft 6 (on ground)

Aircraft 6 (airborne)

AMAN: Uncertainty

6

ETA

RTA

+ -

Stockholm

Milan

ETA Uncertainty # of replans

AMAN: Uncertainty

7

AMAN: Using Uncertainty

8

AMAN: Using Uncertainty

M. Tielrooij, et. al (2013) “Supporting arrival management decisions by visualising uncertainty” De Wit, et. al. (2014). “Supporting runway planning by visualizing capacity balances of arriving aircraft streams”

9

𝜎𝑡 = �𝑓 𝜎1,𝜎2, . . . ,𝜎𝑛 𝑑𝑑𝑡

0

Predicting Uncertainty

10

All flights with condition (x, y, z)

Predicting Uncertainty

11

All sets of conditions

Predicting Uncertainty

12

Flight Update Message

13

- ETA - FLTSTATE - (HoD) - (DoY)

- Format according to standards - Has ARCID, ADEP, ADES, DOF - Inbound EHAM - FPL in first message - Not CNL, DIV, AFL, (CHG,

ACH)

- Last received estimate - WRT ATA

AFTN Extract & filter

Determine ATA

Database

Determine prediction

errors Interpolate

Select common group

Calculate distribution

Store distribution sets

- To fixed ETA values

Determine FLTSTATE

- Use EDEXP FLTSTATE - If not available use DEP

14

Flight Update Message

15

341757 Flights, 1724658 messages

Number of messages per flight

Message Analysis

16

Message Analysis

17

Error Analysis

18

Error Analysis

19

Error Analysos

20

Johnson Normal

Error Distributions

21

Error Distributions

Prediction Horizon

Flig

ht S

tate

(FI,

SI,

TA, A

A)

22

Error Distributions

Prediction Horizon

Flig

ht S

tate

(FI,

SI,

TA, A

A)

23

Error Distributions

Scheduled Arrival Time in Original Flight Plan

Flig

ht S

tate

(FI,

SI,

TA, A

A)

24

Error Distributions

25

Predicting Arrival Time Uncertainty from Actual Flight Information

Maarten Tielrooij m.tielrooij@tudelft.nl

26

Effects of CDM

CDM