Emmanuel Sunil, Joost Ellerbroek and Jacco Hoekstra...• CAMDA uses the notion of airspace...

11

June 26, 2018

Emmanuel Sunil, Joost Ellerbroek and Jacco Hoekstra

2

OutlineIntroduction

Airspace Stability & CAMDA Framework

Fast-Time Simulation Experiments

Results

Conclusions

3

1.Introduction

4

Centralized vs. Decentralized AirspaceCentralized ATC Decentralized ATC

Sepa

ratio

nRo

utin

g

Decentralized ATC is expected to increase capacity

5

Previous Research

6

Open Problems

1. Airspace Design/Structure– How should traffic be organized in the context of

decentralized separation?

2. Safety Modeling – How does the organization of traffic affect the number of

conflicts in decentralized airspace?

3. Capacity Modeling– How does airborne CD&R affect the capacity of

decentralized airspace?

7

Safety ≠ Capacity

Increasing separation between tracks improves airspace safety,but it can also decrease airspace capacity

North Atlantic Organized Track System

8

Capacity for Decentralized ATC

Capacity

Safety

Efficiency

What weighting should be given to safety and efficiency?

9


Safety

Efficiency

Capacity?

10


Safety

Efficiency

CapacityStability

Airspace stability considers the propagation ofConflict Chain Reactions

11

What is CAMDA?

• CAMDA uses the notion of airspace stability to measure the capacity of decentralized airspace concepts

• It can be applied to simulation data to determine capacity semi-empirically– Conflict chain reactions are unpredictable

• It uses measurable airspace properties as inputs

• It calculates the maximum theoretical capacity of an airspace as its main output

12

Research Goals

1. Develop the underlying models of the CAMDA framework

2. Demonstrate CAMDA for an unstructured direct-routing airspace concept that uses decentralized separation

13

2.Airspace Stability & CAMDA

14

Intrusions vs. Conflicts

2𝑆𝑆ℎ

Intrusion Conflict

15

Airspace Stability

A

B

1

C

# Conflict

1 A-B

16

Airspace Stability

A

B

1

C

2

# Conflict Type

1 A-B Primary

2 A-C Secondary

17

Airspace Stability

A

B

1

C

2

3

# Conflict Type

1 A-B Primary

2 A-C Secondary

3 C-B Secondary

Conflict Chain Reactions

Domino Effect Parameter

18

Domino Effect Parameter (DEP)

𝑅𝑅𝑅 𝑅𝑅2 𝑅𝑅3

𝑪𝑪𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒍𝒍𝒏𝒏𝒏𝒏 𝑪𝑪𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒍𝒍𝒘𝒘𝒏𝒏

𝐷𝐷𝐷𝐷𝐷𝐷 =𝑅𝑅𝑅 − 𝑅𝑅𝑅𝑅𝑅𝑅 + 𝑅𝑅𝑅 =

𝐶𝐶𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑙𝑙𝑤𝑤𝑤𝑤𝐶𝐶𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑙𝑙𝑛𝑛𝑤𝑤

− 𝑅

𝐶𝐶𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑙𝑙𝑛𝑛𝑤𝑤 = Total conflicts no resolution

𝐶𝐶𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑙𝑙𝑤𝑤𝑤𝑤 = Total conflicts with resolution

DEP is the number of secondary conflicts per primary conflict

19

DEP Example

A

B

1

C

2

3

# Conflict Type

1 A-B Primary

2 A-C Secondary

3 C-B Secondary

20

DEP Example𝑪𝑪𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒍𝒍𝒏𝒏𝒏𝒏

𝑪𝑪𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒍𝒍𝒘𝒘𝒏𝒏

𝑪𝑪𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒍𝒍𝒏𝒏𝒏𝒏 𝑪𝑪𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒍𝒍𝒘𝒘𝒏𝒏

𝑪𝑪𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒍𝒍𝒏𝒏𝒏𝒏 = 𝟏𝟏

𝑪𝑪𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒍𝒍𝒘𝒘𝒏𝒏 = 𝟑𝟑

𝑪𝑪𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒍𝒍𝒏𝒏𝒏𝒏 = 𝟑𝟑

𝑪𝑪𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒍𝒍𝒘𝒘𝒏𝒏 = 𝟔𝟔

𝐷𝐷𝐷𝐷𝐷𝐷 =3𝑅− 𝑅 = 2

𝐷𝐷𝐷𝐷𝐷𝐷 =63 − 𝑅 = 𝑅

Higher DEP Lower Stability

21

Capacity Measurement Using DEP

Conflict Chain Reactions:– Reduce airspace safety by causing more conflicts– Reduce airspace efficiency by increasing flight

distances

Safety

Efficiency

CapacityStability

Higher DEP Lower Stability Lower Capacity

22

CAMDA Capacity Definition

23


24


When 𝜌𝜌 → 𝜌𝜌𝑚𝑚𝑡𝑡𝑚𝑚, the resolution of any primary conflict causes an infinite number of secondary conflicts

𝐷𝐷𝐷𝐷𝐷𝐷(𝜌𝜌) =𝐶𝐶𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑙𝑙𝑤𝑤𝑤𝑤(𝜌𝜌)𝐶𝐶𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑙𝑙𝑛𝑛𝑤𝑤(𝜌𝜌)

− 𝑅

25

CAMDA FrameworkInstantaneous Conflict

Count Without CR

Total Conflict Count Without CR

Local Conflict Count Per Unit Distance Without CR

Assumption 1:Local Conflict Rate

Total Conflict Count With CR

DEP and Capacity


+Assumption 2:Total A/C Count

WithoutConflict Resolution

WithConflict Resolution

+

26


Count Without CR





DEP and Capacity





+

27

Instantaneous Conflict Count Without CR

Inst. Conflict Count =Number of combinations of 2 aircraft

Conflict probability between any 2 aircraft

X

28


1

3

24 5 6

29


𝑝𝑝 =𝐵𝐵𝑐𝑐

𝐵𝐵𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑙𝑙

30


31


Count Without CR





DEP and Capacity





+

32

Conflict Resolutions Increase Flight Distances

A

B

WA

33

Extra Distance Searched and Flown Due to CD&R, 𝐷𝐷𝑐𝑐𝑐𝑐𝑐𝑐

• 𝐷𝐷𝑐𝑐𝑐𝑐𝑐𝑐 is influenced by conflict chain reactions

• Conflict chain reactions are unpredictable, and they are affected by:– Traffic Density– Interactions between the selected airspace design and

CD&R algorithm

• The value of 𝐷𝐷𝑐𝑐𝑐𝑐𝑐𝑐 needs to be determined from simulation data for a particular airspace concept– CAMDA is semi-empirical

34


Count Without CR





DEP and Capacity





+

35

DEP & Airspace Capacity


− 𝑅

36

DEP & Airspace Capacity

From Simulation Data

37

CAMDA Quick Start Guide

1. Perform two sets of simulations at multiple densities, 𝜌𝜌:a. Without conflict resolution to determine 𝑪𝑪𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒍𝒍𝒏𝒏𝒏𝒏b. With conflict resolution to determine 𝑪𝑪𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒍𝒍𝒘𝒘𝒏𝒏

2. Use simulation data to compute empirical Domino Effect Parameter for each density, 𝐷𝐷𝐷𝐷𝐷𝐷 𝜌𝜌 :

𝐷𝐷𝐷𝐷𝐷𝐷 𝜌𝜌 𝑠𝑠𝑠𝑠𝑚𝑚 =𝐶𝐶𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑙𝑙𝑤𝑤𝑤𝑤(𝜌𝜌)𝐶𝐶𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑙𝑙𝑛𝑛𝑤𝑤(𝜌𝜌)

− 𝑅

3. Determine max theoretical capacity, 𝜌𝜌𝑚𝑚𝑡𝑡𝑚𝑚, using Least-Squares regression

𝐷𝐷𝐷𝐷𝐷𝐷 𝜌𝜌 𝑠𝑠𝑠𝑠𝑚𝑚 =𝜌𝜌

𝝆𝝆𝒎𝒎𝒕𝒕𝒎𝒎 − 𝜌𝜌

38

3.Fast-Time Simulation Experiments

39

BlueSky Open ATM Simulator

https://github.com/ProfHoekstra/bluesky

40

Conflict Detection: State Based

2𝑆𝑆ℎ

2𝑆𝑆𝑣𝑣

𝑉𝑉𝐴𝐴

𝑉𝑉𝐵𝐵A

B

41

Conflict Resolution: MVP

Trajectory Recovery• Horizontal: Direct to sector-exit waypoint • Vertical: Return to pre-conflict FMS altitude

42

Three Experiments

1. Effect of Conflict Detection Parameters– Horizontal and vertical separation requirements– Look-ahead time

2. Effect of Conflict Resolution Dimension– Horizontal only (speed + heading)– Vertical only (vertical speed)

3. Effect of Ground Speed Distribution

More than 2 million flights simulated for all three experiments

43

4.Results

44

Conflict Detection Experiment

Condition Name Horizontal Separation [NM]Vertical

Separation [ft]Look-Ahead Time [mins]

Baseline 2.5 500 5

Double Separation 5.0 1000 5

Double Separation + Half Look-Ahead 5.0 1000 2.5

45


Separation requirements has a greater impact than look-ahead time on maximum airspace capacity


− 𝑅

46


• A doubling of separation requirements leads to a quadrupling of conflict probability

• A doubling of look-ahead time leads to a doubling of conflict probability

• Therefore separation requirements have a greater effect on capacity than look-ahead time

47

Conflict Resolution Experiment

Horizontal (Speed + Heading)Vertical (Vertical Speed)

48

Conflict Resolution Experiment

Vertical conflict resolution reduces capacity as traffic is more closely packed in the vertical direction in en-route airspace

49

Speed Distribution Experiment

400kts

350kts 450 kts400kts350kts 450kts400kts

50

Speed Distribution Experiment

No substantial effect of speed distribution on the maximum capacityof unstructured direct-routing airspace

51

5.Conclusions

52

CAMDA Pros and Cons

Pros Cons•Provides a physical definition of capacity in terms of conflict chain reactions

•Can be used to understand the effect of airspace design parameters on capacity

•Unbiased and objective metric to compare different airspace designs and CD&R algorithms

•No weighting/tuning required

•Reduce the number of simulations needed to evaluate an airspace design

•Semi-empirical: Conflict chain reactions are difficult to predict analytically

53

Capacity of Unstructured Airspace

• Traffic separation requirements have a greater impact on capacity than look-ahead time– Separation requirements have a larger impact on conflict

probability

• Restricting conflict resolutions to the vertical direction reduces airspace capacity– Aircraft are more closely packed in the vertical direction

• Ground speed distribution does not have a substantial effect on the capacity of unstructured direct-routing airspace

54

“Analyzing the Effect of Traffic Scenario Properties on Conflict Count Models”

Effect of Speed, Heading, Altitude and Spatial Distribution of Traffic on Safety

When: 11:40, Friday, Classroom C4-026V

Related Work

55

Thank You For Your Attention!

[[email protected]][https://www.researchgate.net/profile/Emmanuel_Sunil]

Slide Number 1OutlineSlide Number 3Centralized vs. Decentralized AirspacePrevious ResearchOpen ProblemsSafety ≠ CapacityCapacity for Decentralized ATCCapacity for Decentralized ATCCapacity for Decentralized ATCWhat is CAMDA?Research GoalsSlide Number 13Intrusions vs. ConflictsAirspace StabilityAirspace StabilityAirspace StabilityDomino Effect Parameter (DEP)DEP ExampleDEP ExampleCapacity Measurement Using DEPCAMDA Capacity DefinitionCAMDA Capacity DefinitionCAMDA Capacity DefinitionCAMDA FrameworkCAMDA FrameworkInstantaneous Conflict Count Without CRInstantaneous Conflict Count Without CRInstantaneous Conflict Count Without CRInstantaneous Conflict Count Without CRCAMDA FrameworkConflict Resolutions Increase Flight DistancesExtra Distance Searched and Flown Due to CD&R, 𝐷 𝑐𝑑𝑟 CAMDA FrameworkDEP & Airspace CapacityDEP & Airspace CapacityCAMDA Quick Start GuideSlide Number 38BlueSky Open ATM SimulatorConflict Detection: State BasedConflict Resolution: MVPThree ExperimentsSlide Number 43Conflict Detection ExperimentConflict Detection ExperimentConflict Detection ExperimentConflict Resolution ExperimentConflict Resolution ExperimentSpeed Distribution ExperimentSpeed Distribution ExperimentSlide Number 51CAMDA Pros and ConsCapacity of Unstructured Airspace Related WorkThank You For Your Attention!

Emmanuel Sunil, Joost Ellerbroek and Jacco Hoekstra...• CAMDA uses the notion of airspace...

Documents

Transcript of Emmanuel Sunil, Joost Ellerbroek and Jacco Hoekstra...• CAMDA uses the notion of airspace...