Tutorial TPAS - ICRAT 2014 11/06/14 1 2 INDEX 1. - - PDF document

SLIDE 1

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 1 ¡

1

TUTORIAL SESSION

TPAS: TEST-BED PLATFORM FOR ATM STUDIES

Speaker: Dr. Sergio Ruiz

Scientific supervisor: Dr. Miquel A. Piera sergio.ruiz@uab.es miquelangel.piera@uab.es

INDEX

1. Introduction: need for a micro-model 4D view of the ATM 2. Test-bed Platform for ATM Studies (TPAS) 3. Micro-scale data framework to store and manage ATM state- space information 4. TPAS functionalities and case studies 5. How to use TPAS (APIs and functionalities) 6. Practical example: temporal looseness for ground delays 7. Other ATM studies and Future developments

2

• 1. INTRODUCTION

3

• 1. INTRODUCTION: NEED FOR

A 4D MICRO-MODEL ATM VIEW

“A lack of a proper coordination among the main air traffic management (ATM) actors through their different decision making processes limits the actual system capacity and leads to an inefficient and congested air transport system.”

4

Need for a shared 4D micro-model view of the ATM TBO framework

(SESAR, NextGen…)

ATM CAPACITY AND MAIN ATM ACTORS

ATM capacity = “amount of aircraft that can be managed safely by Air Traffic Management system in a period of time” Actors:

• Demand: Airspace Users (AUs), e.g., airlines
• Supply: Air Traffic Control (ATC) centers and Airport Operators (AOs)
• Network Manager (in Europe, EUROCONTROL, in charge of ATFCM

to ensure Demand and Capacity Balancing)

5

Introduction

CURRENT ATM MODEL

“A lack of a proper coordination among the main air traffic management (ATM) actors through their different decision making processes limits the actual system capacity and leads to an inefficient and congested air transport system.”

• Focused on capacity management

Introduction DEMAND CAPACITY

6

SLIDE 2

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 2 ¡

CURRENT ATM OPERATIONS

Assume NO conflicts Introduction

7

ATFCM VIEW OF THE ATM

1/5 2/5 2/5 1/5 1/5 0/5 0/5 0/5 Assumed maximum sector pre-declared capacity: 5/5 PREDICTION:

SIMULATION TIME 1

Introduction

8

ATFCM does not see what is happening at micro-scale/trajectory level

ATFCM VIEW OF THE ATM

0/5 0/5 0/5 0/5 0/5 6/5 1/5 0/5

Capacity problem Regulation required

Assumed maximum sector pre-declared capacity: 5/5

ATFCM does not see what is happening at micro-scale/trajectory level

Introduction PREDICTION:

SIMULATION TIME 2

9

ATC VIEW OF THE ATM

Conflict predicted Regulation Applied (delay) ATC intervention required Sector capacities satisfied

Introduction ATC PREDICTION

10

ATC VIEW OF THE ATM

Conflict (emergent dynamic) Conflict resolution applied (local view) Sector capacities satisfied

Introduction ATC PREDICTION

11

A network micro-scale view is required to identify the Emergent Dynamics of local decisions and to coordinate the main ATM actors

SESAR ATM MODEL

New ¡Concept ¡of ¡Operations ¡(SESAR ¡programme): ¡

• Trajectory-­‑based ¡operations ¡(4D ¡Trajectory) ¡
• Collaborative ¡planning ¡(sharing ¡of ¡information ¡and ¡decisions) ¡

¡ ¡

• Dynamic ¡airspace ¡(flexible ¡route ¡structures ¡and ¡sectors) ¡
• New ¡and ¡innovated ¡technologies ¡(automation, ¡precision, ¡

reliability, ¡efficiency…) ¡

4D contract: AU assume the compromise

• f flying the trajectory planned for the full

flight with enough precision in the 4 dimensions (3D + time), and the ATM services agree to facilitate.

Introduction

12

Good trade-off between efficiency and capacity Anticipation of the Emergent Dynamics (safety)

SLIDE 3

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 3 ¡

DESIRABLE ATM MODEL: TBO (SESAR, NEXTGEN...)

13

Introduction

• A microscopic 4D trajectory model of the traffic flows
• Automated and coordinated stakeholders’ DSTs to ensure a

more precise and stable traffic synchronization along the network

• Allow the participation of the Airspace Users through

arbitrated negotiation processes during the entire network planning process

• A common overall sight of the ATM current and predicted

states

• The anticipation of the potential emergent dynamics at the

network due to local decisions shared among all the ATM stakeholders DSTs

4D Trajectory

TBO-BASED ATM MODEL (SESAR, NETXGEN…)

• Focus on synchronized 4D trajectories (arbitrated by Network Manager)

Introduction

14

COORDINATION Coordination requires all the actors having the same traffic view at micro-scopic/ trajecotry level

2.TEST-BED FOR ATM STUDIES (TPAS)

15

• 2. TEST-BED PLATFORM

FOR ATM STUDIES (TPAS)

Available now for the ATM community

16

ATM Micro-model framework Basic TPAS functionalities Advanced TPAS functionalities User interaction

TPAS

Useful for: Fast prototyping Test DSS Benchmarking

EXAMPLE: STREAM WP-E

17

Strategic de-confliction of 4000 trajectories over Europe in 2 hour look-ahead Computational time to find a solution < 60 sec. (with a regular computer) Processing about 20 million of waypoints (including what-if trajectories) TPAS

• 3. MICRO-SCALE DATA FRAMEWORK

18

SLIDE 4

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 4 ¡

• 3. SPATIAL DATA

STRUCTURE (SDS)

A micro-scale data framework to support the representation

• f the air traffic demand on the different airspace sectors at

aircraft trajectory (microscopic) level è è “ATM 4D snapshot”

Safety envelope in a given instant Turbulence generated in a given instant

19 Δt = 1sec. Δt = 1sec. Δt = 1sec. Δt = 1sec.

SDS Physical concept of a SDS Logical concept of a SDS

• Spatial Data Structure (SDS): is a database that represents a spatial region (e.g. an air

sector) by using individual memory positions to represent each of the discrete (3D) coordinates

• f the sector.
• Spatial data: spatial information (e.g., discrete trajectory representation) and non-spatial

information (e.g., Flight Number id)

• Such memory positions are sorted in a way that, given a certain coordinate, the information

stored inside the SDS is easily recoverable applying linear mathematical formula: Y, Z: order or size of dime sion Y and Z.

SPATIAL DATA STRUCTURES

20

SDS Other non-spatial state-space info can be added

SPATIAL DATA STRUCTURES

21

SDS

SPATIAL DATA STRUCTURES

Similar (but different) types of data structures:

• Occupancy grids: specialized usage of SDSs used in robotics

to build in real time maps (for navigation purposes)

• Hash tables: to store keys/pointers –and only keys/pointers– to

the data values of interest (not necessarily related with any kind of spatial data)

• Look-up tables: to store pre-calculated values for a given

function in order to avoid online –time consuming– calculations

22

SDS

SPATIAL DATA STRUCTURES

• Granularity or resolution: distance between discrete points of

the SDS

• Similarly to digital cameras, the resolution determines both

the “quality” of the SS stored and the “efficiency” at processing and managing the spatial data.

• The excess of resolution may lead to a loss of computer

performance as well as to an inoperable amount of memory requirements

• A lack of resolution may lead to lose some important
• bjects of the space (thus missing the detection of some

existing conflicts in the CD process, i.e. false negative errors) or to a lose of filtering performance.

23

SDS

ANY DECISION MUST PRESERVE SAFETY

Air Traffic Management (ATM)

Safety Capacity Cost efficiency Environ. impact

ATM services: facilitate orderly and safe air transportation system Avoid accidents & incidents Allocate demand Minimum cost Be sustainable

24

SDS

SLIDE 5

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 5 ¡

Two methods of Confliction Detection based on SDS:

• Build and store the safety envelope (4D tube)
• perform

Conflicts are detected at the moment of storing the 4D envelope points by comparing time-windows Conflicts are detected at the moment

• f storing the position by comparing

distances at each time-step

• Store the point-mass position and pairwise comparisons (only between near cells)

SDS FOR CONFLICT DETECTION

25

SDS

Each coordinate of the grid has memory positions in the database to store information Each trajectory processed is stored and compared with trajectories using same spatial resources

Time-windows comparison is

• nly performed

here No comparison is needed!!!

1 1 2 3 4 5 6 7 8 9 2 3 4 5 6 7 8 9 10 Y X

MICRO-SCALE DATA FRAMEWORK FOR ATM

26

SDS

CASE STUDY: TIME- BASED MTCD

Aircraft must avoid the turbulences generated by other traffic

27

SDS

S.Ruiz, M.A.Piera, I. del Pozo, “A Medium Term Conflict Detection and Resolution system for Terminal Maneuvering Area based on Spatial Data Structures and 4D Trajectories”, Journal of Transportation Research part C: Emerging Technologies, Elsevier, 2012

28

SDS FOR A LARGE ATM AND A LARGE NUMBER OF 4DT

28

Physical concept of a SDS

Time-Space Data Structure (TSDS)

Faster Conflict Detection

Logical concept of a SDS

Information is stored in a way that saves 98% of memory

Relational SDS

SDS

SDS 3D VS. TSDS

A B A B A B A B SDS 3D SDS 4D t0 t1 Case 1: A è t0 B è t0 Case 2: A è t0 B è t1 Lot of comparisons saved Comparison always needed

29

SDS

RTSDS EXAMPLE

30

Tr1, Tr2, Tr3, Tr4 and Tr5 processed in sequential order è

SDS

SLIDE 6

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 6 ¡

0.2 º 0.5º

Constant bin size Bin-size variable with latitude 0.2º in latitude ≈ 12NM 0.5º in longitude ≈ 30NM in Equator 26NM in lat 30º 10NM in lat 70º Optimal size: 10NM (≈20Km) Planar SDS Geodesic SDS

31

SDS Current physical concept of the SDS Variable bin-size due to the curvature of the Earth

32

SDS Current physical concept of the TSDS t0 t1 tT … t… Time discretization: 1 sec.

è For 2-hour prediction look-ahead: 7200 SDSs 33

SDS TSDS allows storing current and future states

S.Ruiz and M.A.Piera, “Relational time-space data structure to enable strategic de-confliction with a global scope in the presence of a large number of 4D trajectories”, Journal of Aerospace Operations, IOS press, 2013.

SDS FOR CONFLICT DETECTION (PERFORMANCE)

1000 2000 3000 4000 5000 50 100 150 200 250 300 350 400 450 500 Number of Aircraft, n Search Time, T (sec.) Pairwise CD without SDS RTSDSbased CD TSDS-based performance (clustering pairwise) vs. pure pairwise CD algorithms

More than 8 minutes Less than 8 seconds

34

5000 trajectories è Ideal for real-time Strategic De-confliction SDS

• 4. TPAS FUNCTIONALITIES

35

TPAS FUNCTIONALITIES

BASIC FUNCTIONS OF THE INFORMATION MANAGER:

• Management of aircraft/flight information
• Management of original trajectories/flight plans
• Management of what-if trajectories
• Generate trajectories from the routes/flight plans
• Add ATM information to the SDS (e.g. 4D trajectories)
• Delete ATM information from the SDS (e.g. 4D trajectories)
• Extract ATM State Space information from SDS (e.g. conflicts,

temporal looseness, complexity map…)

• Management and classification of ATM State Space information

(e.g. temporal sorting of conflicts, computation of basic metrics…)

• Coordination of modules functionalities

36

TPAS FUNCTIONALITIES

SLIDE 7

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 7 ¡

TPAS FUNCTIONALITIES

Basic functions to manage:

• Coordinates (geodesic and UTM)
• Also includes:
• Computation of 2D/3D distances between 2 coordinates (great circle or loxodromic)
• Conversion from Geodesic to UTM and vice versa
• Routes
• Trajectories
• Flights
• Also includes: Parameterization of BADA models
• Spatial Data Structures
• Input-Output operations
• Including: G.Earth, FACET, GNUPlot, AIDL, cvs, txt, among others

37

TPAS FUNCTIONALITIES

TPAS FUNCTIONALITIES

Advanced functions (modules):

• Conflict Detection (time-based or spatial-based)
• Conflict Resolution (with Geometric Optimization Approach)
• Clusterizer (return sets of trajectories in conflict)
• Interaction Causal Solver (CR considering domino effects)
• Temporal looseness calculation (No-go zones and First Range

Looseness)

• Ground Delays
• Statistics, metrics and tools for ATM uncertainty

38

TPAS FUNCTIONALITIES WPT Conflict Avoidance Recovery Aircraft A (Ownship) Aircraft B (Intruder)

2 problems: Conflict avoidance and recovery

Geometric Optimization Approach

CONFLICT RESOLUTION

• Analytical method
• Allows different maneuvers
• Optimal solutions (local)

è Given a conflict and a kind of maneuver we can find an optimal solution (if any exist)

ADVANTAGES:

• Dr. Bilimoria (NASA)

39

TPAS FUNCTIONALITIES

CONFLICT RESOLUTION

EXEMPLE: 2 trajectories in conflict and 4 possible solutions

40

TPAS FUNCTIONALITIES This example also shows the Output features (.kml) Solutions computed using Geometric Optimization Approach (GOA)

W i t h

• u

t C D & R ( A

n

• m

)

Stabilizing Effect (S) Destabilizing Effect (D) Deviated aircraft with conflicting nominal trajectories

W i t h C D & R ( A

C D R

)

* Bilimoria (NASA) performed several experiments that confirms the importance of taking into account the Domino effects of the resolution trajectories

DOMINO EFFECTS

The State-Space stored SDS can be used to explore Domino Effects

41

Domino Effects may appear during conflict resolution in the presence of multiple trajectories (surrounding traffic) TPAS FUNCTIONALITIES

Tr1 Tr2 Tr3 Tr11 Tr2 Tr3 Tr1 Tr21 Tr3 C=1 C=1

Total cost = 1

(a) (c)

Tr1 Tr22 Tr3 C=1.4

(d)

Tr12 C=1.2

(b) (f)

Tr2 Tr3 Tr11 Tr2 Tr31 C=1

(g)

C=1 Tr11 Tr2 Tr32 C=1

(h)

C=2

Total cost = 2 Total cost = 1.2 Total cost = 1.4 Total cost = 3 Non-feasible scenario

(e)

Tr111 Tr2 Tr3 C=1.5

Total cost = 1.5

Conflict Resolution with multiple trajectories is a highly combinatorial problem

42

SLIDE 8

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 8 ¡

CLUSTERIZER

• Reduce the solution space combinatorial exploration

43

TPAS FUNCTIONALITIES

INTERACTION CAUSAL SOLVER

44

• Reduces the search to the

Pareto Frontier of the feasible solutions

• Causal model with a

discrete event approach RESOLUTION TRAJECTORY GENERATOR (RTG) CONFLICT DETECTION (CD) FLIGTH PLANS

(ORIGINAL TRAJECTORIES)

CONFLICTS

TRIAL TRAJECTORIES

SDS

WPT Conflict Avoidance Recovery Aircraft A (Ownship) Aircraft B (Intruder)

GEOMETRIC OPTIMIZATION APPROACH

STATE SPACE GENERATION

CONFLICTS

INTERACTION CAUSAL SOLVER (ICS)

CAUSAL MODEL 45

Alternate trajectories are sorted by preferences of the Airspace Users

Most preferred Less preferred

Ideal solution [11, 22, 33, 44, 55 …, N-1N-1, NN] Non feasible solution

46

New Pareto-efficient solution:

[7601, 22, 33, 44, 55 …, N-1N-1, NN]

New Pareto-efficient solution:

[11, 22, 33, 44, 3245 …, N-1N-1, NN]

7601 5201

1

3245 1685

5 5201 1 3245 1685 5 7601

47 48

SLIDE 9

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 9 ¡

49 50

SIMULATIONS WITH FACET TOOL

51

SIMULATIONS WITH FACET TOOL

52

STRATEGIC DE- CONFLICTION

• Strategic ¡De-­‑confliction: ¡planning ¡actions ¡taken ¡from ¡several ¡hours ¡in ¡

advance ¡up ¡to ¡few ¡minutes ¡before ¡to ¡the ¡execution ¡phase, ¡in ¡order ¡to ¡ anticipate ¡the ¡separation ¡of ¡flights ¡even ¡before ¡they ¡takeoff, ¡and ¡also ¡while ¡ they ¡are ¡airborne, ¡but ¡always ¡with ¡enough ¡anticipation ¡to ¡allow ¡a ¡Collaborative ¡ Flight ¡Planning ¡subject ¡to ¡the ¡applicable ¡safety ¡standards ¡

Layers of safety in ATM (separation of flights):

• Operational: avoid imminent crashes (e.g., TCAS)
• Tactical: medium term planning of the traffic within a sector (i.e.,

ATC separation tasks 20-30 min. in advance)

53

4010 direct route trajectories over European ATM 311 conflicts 286 resolution maneuvers

54

Results of WP-E STREAM

S.Ruiz, J.Nosedal, M.A.Piera, A.Ranieri, “Strategic de-confliction in the presence of a large number of 4D trajectories using a causal modeling approach”, Journal of Transportation Research part C: Emerging Technologies, Elsevier, 2014.

SLIDE 10

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 10 ¡

TEMPORAL/LONGITUDINAL LOOSENESS

55

Yellow zones indicate potential conflicts if the temporal dimension changes

TEMPORAL/LONGITUDINAL LOOSENESS

56

Delay applied to flight A Departure times of B and D remain constant

TEMPORAL/LONGITUDINAL LOOSENESS

57

Early departure applied to flight A Departure times of B and D remain constant

TEMPORAL/LONGITUDINAL LOOSENESS

58

TPAS FUNCTIONALITIES λ =First Range Looseness (FRL)

GROUND DELAYS (GD)

59

hot-spot identification Looseness distribution pre-GD Looseness distribution post-GD

J.Nosedal, M.A.Piera, S.Ruiz and A.Ranieri, “An efficient algorithm for smoothing the airspace congestion by fine tuning of departure times”, Journal of Transportation Research part C: Emerging Technologies, Elsevier, 2014

• 6. HOW TO USE TPAS

60

SLIDE 11

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 11 ¡

HOW TO USE TPAS?

• Obtain TPAS: email to sergio.ruiz@uab.es
• Installation

1. Static method è requires .dll, .lib and .h files

• Add .lib to “additional dependencies” of the linker
• Add the .h to your code
• Call the .dll from your code

2. Dinamic method è Only .dll file

• Use the method loadLibrary()
• Call the functions as given in the APIs documentation

61

The static method is easier in C++ and the dynamic is easier in Java and C#

HOW TO USE TPAS?

62

TPAS_Interface TPAS; // Create instance of TPAS int idOriginATM = TPAS.Coord_createCoord(); TPAS.Coord_setGeoandConvert(idOriginATM, 25, -25); // lat/long TPAS.Coord_setZ(idOriginATM, 2400); // in meters TPAS.Coord_setTime(idOriginATM, 0); // in seconds int idSDS = TPAS.SDS_create(); // Create the micro-scale data framework TPAS.SDS_setOriginScenarioATM(idSDS, idOriginATM); TPAS.SDS_setLengthX(idSDS, xLengthATM); TPAS.SDS_setLengthY(idSDS, yLengthATM); TPAS.SDS_setLengthZ(idSDS, zLengthATM); TPAS.SDS_setLengthT(idSDS, tLengthATM); TPAS.SDS_setSizeBins(idSDS, spatialDiscretization); TPAS.allocateMemory(idSDS); // SDS is ready

HOW TO USE TPAS?

63

// Generate waypoints int idTERTO= TPAS.Coord_createCoord_Geod(lat, long); int idRUSIK= TPAS.Coord_createCoord_Geod(lat, long); int idCANIS= TPAS.Coord_createCoord_Geod(lat, long); int idENETA= TPAS.Coord_createCoord_Geod(lat, long); int idRWY= TPAS.Coord_createCoord_Geod(lat, long, z); // Generate routes int idTERTO3C = TPAS.Route_create(); // TERTO3C TPAS.Route_addCoord(idTERTO3C, idTERTO); TPAS.Route_addCoord(idTERTO3C, idCANIS); TPAS.Route_addCoord(idTERTO3C, idENETA) TPAS.Route_addCoord(idTERTO3C, idRWY); int idRUSIK3C = TPAS.Route_create(); // RUSIK3C TPAS.Route_addCoord(idRUSIK3C, idRUSIK); TPAS.Route_addCoord(idRUSIK3C, idCANIS); TPAS.Route_addCoord(idRUSIK3C, idENETA); TPAS.Route_addCoord(idRUSIK3C, idRWY);

HOW TO USE TPAS?

64

// Create flights vector <int> vFlights; vFlights[0] = TPAS.Flight_create(); TPAS.Flight_setRoute(vFlights[0], idRUSIK3C); TPAS.Flight_setEntryTime(vFlights[0], 150); // in seconds TPAS.Flight_setEntrySpeed(vFlights[0], 230); // in mps TPAS.Flight_generatePlannedTrajectory(vFlights[0]); // 4DT is generated vFlights[1] = TPAS.Flight_create(); TPAS.Flight_setRoute(vFlights[1], idTERTO3C); TPAS.Flight_setEntryTime(vFlights[1], 50); // in seconds TPAS.Flight_setEntrySpeed(vFlights[1], 232); // in mps TPAS.Flight_generatePlannedTrajectory(vFlights[1]); // 4DT is generated TPAS.CD_detectConflicts(); //Perform CD and CR TPAS.CD_showConflicts(); // Output conflicts information

HOW TO USE TPAS?

65

// An example even easier TPAS_Interface TPAS; // Create instance of TPAS … // Configure originATM and SDS just as before Vector <int> vFlights; // Create vector to store the ids/pointers of flights vFlights = TPAS.Parsers_intputFreeFlight_UTM2Geod(“traffic.txt”, timezone=28); TPAS.CD_detectConflicts(); //Perform Conflict Detection TPAS.CD_showConflicts(); // Output conflicts information Note that by sharing scenarios we could apply different CR modules and perform benchmarking comparisons

x_ini y_ini h_ini x_end y_end h_end speed (in mps)

HOW TO USE TPAS?

66

// An example coupling an external CR TPAS_Interface TPAS; // Create instance of TPAS … // Configure originATM and SDS just as before Vector <int> vFlights; // Create vector to store the ids/pointers of flights vFlights = TPAS.Parsers_intputTrajectories(“C//inputTrajectories/TR_”); TPAS.CD_detectConflicts(); //Perform Conflict Detection TPAS.CD_outputConflicts2txt(); // Write conflicts in a .txt file …// Process the conflicts in your own CR vFlights = TPAS.Parsers_changeTrajectories(“C//newTrajectories/TR_”); TPAS.CD_detectConflicts(); // Verify the new scenario is clean of conflicts TPAS.Parsers_outputTrajectories2kml(vFlights ); // Output your trajectories to kml TPAS.Parsers_outputTrajectories2gnuPlot(vFlights ); // Output your trajectories to gnuPlot

Format: FlightNumber entryTime (in s.) Long Lat Height (in m.) speed (in mps)

SLIDE 12

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 12 ¡

6.HANDS-ON EXAMPLE

67

• 7. HANDS-ON

EXAMPLE

68 Change entryTime of flight 1

FRL flight 1: [-61, 2]

2 1 3 2 1 3

100

1

103 188

• 47

38

2 3

8.OTHER ATM STUDIES AND FUTURE DEVELOPMENTS

69

INTEGRATION WITH SWIM

70

AIRPORT INTEGRATION (NOP+AOP)

71

System presented in the Eurocontrol’s challenge of SWIM Master Class 2013

STRATEGIES FOR ATM UNCERTAINTY

72 Contribution to the wiki of ComplexWorld (Uncertainty in ATM)

SLIDE 13

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 13 ¡

è A strategic buffer of 10NM/12NM could absorb big part of the WPE and ETD uncertainties, thus providing with a more robust network

TRAFFIC ROBUSTNESS ANALYSIS AND ASSESSMENT OF UNCERTAINTY IMPACT

!! "#$%&'(! )*+,! )*+-! +./,! +./-! 0"1!23/4!

• ,,!
• 56!
• 57!

,89! ,:5! 9"1!23/4!

• 06!
• 95!
• 0,!
• ;9!
• 6-!

7"1!23/4!

• 87!
• 86!
• ::!
• 77!
• :9!

,5"1!23/4! 6;5! 6-9! 6,0! 6-9! 6;5!

è Trade-off: More stable/robust network (predictability) vs. additional CR amendments required (efficiency) and less space available due to the extra buffers (capacity)

73

t0 t0 + 30' t0 + 60' t0 + 90'

p = 1 If few hours before (e.g. 60’) flight execution severe weather is predicted, the optimal 4D trajectory is updated

• 9. FUTURE WORK

Network Disruptions

74

t0 t0 + 30' t0 + 60' t0 + 90'

p = 0.5 Since p=0.5 a new trajectory can be computed considering new waypoint in the halfway

75

t0 t0 + 30' t0 + 60' t0 + 90'

p = 0 Weather predictions are updated and RBT is modified in consequence Severe weather is definitely not going to happen (p=0)

76

t0 t0 + 30' t0 + 60' t0 + 90'

p = 1 Weather predictions are updated and RBT is modified in consequence Severe weather is definitely going to happen (p=1)

77

t0 t0 + 30' t0 + 60' t0 + 90'

p = 0.5 ATM planning should consider at t0 (and before) the probability of both probable trajectories diverging at t0+30’ Note: weather predictions and SBT/RBT updates will be continuous and not in 30’ steps a b c d

NEW UNCERTAINTY MODELS

Network Disruptions

78

SLIDE 14

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 14 ¡

PUBLICATIONS

• 1. T.Jung, M.A.Piera, S.Ruiz, “A causal model to explore the ACAS induced collisions”,

Journal of Aerospace Engineering, SAGE, 2014.

• 2. J.Nosedal, M.A.Piera, S.Ruiz and A.Ranieri, “An efficient algorithm for smoothing

the airspace congestion by fine tuning of departure times”, Journal of Transportation Research part C: Emerging Technologies, Elsevier, 2014

• 3. S.Ruiz, J.Nosedal, M.A.Piera, A.Ranieri, “Strategic de-confliction in the presence of

a large number of 4D trajectories using a causal modeling approach”, Journal of Transportation Research part C: Emerging Technologies, Elsevier, 2014.

• 4. S.Ruiz and M.A.Piera, “Relational time-space data structure to enable strategic de-

confliction with a global scope in the presence of a large number of 4D trajectories”, Journal of Aerospace Operations, IOS press, 2013.

• 5. S.Ruiz, M.A.Piera, I. del Pozo, “A Medium Term Conflict Detection and Resolution

system for Terminal Maneuvering Area based on Spatial Data Structures and 4D Trajectories”, Journal of Transportation Research part C: Emerging Technologies, Elsevier, 2012

• 6. C.A. Zúñiga a,⇑, M.A. Piera a, S. Ruiz a, I. Del Pozo, “A CD&CR causal model

based on path shortening/path stretching techniques”, Journal of Transportation Research part C: Emerging Technologies, Elsevier, 2012

79

FEEDBACK AND QUESTIONS ARE WELCOME

Thanks for your attention Email: Sergio.ruiz@uab.es

80

ATM UNCERTAINTY

• Navigational imprecision and tracking errors
• E.g., relatively little trajectory deviations
• Individual-level perturbations
• E.g., delays, deviations…
• Network-level perturbations
• E.g., convective weather, sudden loss of airport/sector capacity…

Uncertainty grows with time è it seriously affects Strategic De-confliction Typical lateral buffer: 1NM è Little re-planning, but “the one that deviates the one that pays” è Full re-planning: need for real-time Strategic De-confliction

81

OK!

T0 T0+20’ T0+40’ T0+60’

Risk-of-deviation

NEW UNCERTAINTY MODELS

82

PLANNED

COLLISION! DEVIATION!

T0 T0+20’ T0+40’ T0+60’

Risk-of-deviation

NEW UNCERTAINTY MODELS

83

EXECUTION

CONFLICT!

Nominal trajectories are not really in conflict, but the risk of collision in case of deviation Is too high (i.e. tactical ATC will consider it as a conflict) Consider a risk-of-deviation model T0 T0+20’ T0+40’ T0+60’

Risk-of-deviation

NEW UNCERTAINTY MODELS

84

SLIDE 15

Tutorial ¡TPAS ¡-­‑ ¡ICRAT ¡2014 ¡ 11/06/14 ¡ 15 ¡

OK!

Now the route is sub-optimal but it is more robust and will reduce the workload of tactical controllers T0 T0+20’ T0+40’ T0+60’

Risk-of-deviation

NEW UNCERTAINTY MODELS

85

More state-space info can be added

• More options:

NEW UNCERTAINTY MODELS

TP uncertainty

86