D'ARIANO ATRS 2016

Scheduling models for optimal aircraft traffic control at busy airports: tardiness, priorities,

equity and violations considerations

Dipartimento di Ingegneria

Andrea D’Ariano, ROMA TRE University, Rome, Italy124/11/2016

Junior ConsultingDipartimento di Ingegneria

� Introduction�Modeling a Terminal Control Area

�MILP formulations

Scheduling models for optimal aircraft traffic control at busy airports: tardiness, priorities, equity and violations considerations


�Computational experiments

�Conclusions

This work has been recently accepted for publication in the journal OMEGA : Reference DOI: 10.1016/j.omega.2016.04.003

2


Air Traffic Control (ATC)Air Traffic Control (ATC)

An efficient control of air traffic must ensure safe, ordered and rapid transit of aircraft on the ground and in the air resources.

With the increase in air traffic [*], aviation authorities are seeking methods (i) to better use the

[*] Source: IATA 2014

methods (i) to better use the existing airport infrastructure, and (ii) to better manage aircraftmovements in the vicinity of airports during operations.

3


Status Status ofof the the currentcurrent ATC ATC practisepractise• Airports are becoming a major bottleneck in ATC operations. • The optimization of take-off/landing operations is a key factor to improve the performance of the entire ATC system.

• ATC operations are still mainly performed by human controllers whose computer support is most often limited to a graphical representation of the current aircraft position and speed. • Intelligent decision support is under investigation in order to reduce the controller workload (see e.g. recent ATM Seminars).

4


Literature: Aircraft Scheduling Problem (ASP)Literature: Aircraft Scheduling Problem (ASP)Terminal Control Area (TCA) Terminal Control Area (TCA)

Detailed

BasicExisting Approaches Dynamic

Static

5


Literature: Research needsLiterature: Research needs

Aircraft Scheduling Problem in Terminal Control Areas:

Most aircraft scheduling models in literature represent the TCA as a single resource, typically the runway. These models are not realistic since the other TCA resources are ignored.

We present a new approach that includes both accurate modelling of traffic regulations at runways and airways.

6

This approach has already been applied to successully control railway traffic for metro lines and railway networks.


Our approach for TCAsOur approach for TCAs

Implementation and testing of:

• Detailed ASP-TCA models:incorporating safety rules at air segments, runways and holding circlesair segments, runways and holding circles

• Alternative objective functions:maximum versus average delays, delayed aircraft (violations), aircraft equity, throughput (completion time), priority tardiness

• Real-time traffic management instances:Roma Fiumicino (FCO) and Milano Malpensa (MXP) airports

7


� Introduction

�Modeling a Terminal Control Area �MILP formulations




�Conclusions

8



3 HOLDING

CIRCLES

SEVERAL AIR

SEGMENTS

1 COMMON GLIDE

PATH

3 RUNWAYS3 RUNWAYS


ASP ASP ModelModel::AlternativeAlternativeGraphGraph (AG)(AG)

Air Segments

CommonGlide Path

RunwaysHolding Circles

8

16

173

SRN

1

TOR

MBR

2 6

4 10

11

12

15

7

513

14

RWY 16R

RWY 16L

9

A

[Pacciarelli

EJOR 2002]

RWY 25

A1

tA1

10

release date αA

(w0, A1 = αA = expected aircraft entry time)Fixed constraints

tA1 = t0 + w0, A1

A1

0

αA

t0


AG AG ModelModelAir Segments

CommonGlide Path


8

16

173

SRN

1

TOR

MBR

2 6

4 10

11

12

15

7

513

14

RWY 16R

RWY 16L

9

A

A1

RWY 25

tA1

entry due date βA

(wA1,n = βA = - αA )

A1

0 n

αA

βA

11

tn = tA1 + wA1,n

tn


AG AG ModelModelAir Segments

CommonGlide Path


8

16

173

SRN

1

TOR

MBR

2 6

4 10

11

12

15

7

513

14

RWY 16R

RWY 16L

9

A

A1 A4

δ

0

-δ

RWY 25

dotted arc (A4, A1)No holding circle

dotted arc (A1, A4)Yes holding circle(δ = holding time)

A1 A4

0 n

αA

βA

-δ

0

12

Alternative constraints


AG AG ModelModel

A1 A4 A10min

Air Segments

CommonGlide Path


8

16

173

SRN

1

TOR

MBR

2 6

4 10

11

12

15

7

513

14

RWY 16R

RWY 16L

9

A

RWY 25

A1 A4 A10

0 n

αA

βA

- max

Time window for the travel time in each air segment[min travel time; max travel time]

13


CommonGlide Path

RunwaysHolding Circles Air Segments

8

16

173

SRN

1

TOR

MBR

2 6

4 10

11

12

15

7

513

14

RWY 16R

RWY 16L

9

AAAG AG ModelModel

A1 A4 A15A10 A13 AOUTA16

RWY 25Aircraft routing:A1-A4-A10-A13-A15-A16


0 n

αA

βA

γA

exit due date γA

(γA = - planned landing time)

14


CommonGlide Path


8

16

173

SRN

1

TOR

MBR

2 6

4 10

11

12

15

7

513

14

RWY 16R

RWY 16L

9

AA

BB

AG AG ModelModel


Potential conflictbetween A and B on

the common glide path (resource 15) !

RWY 25


0 n

B3 B8 B15B12 B14 BOUTB17

αA

αB

βA

γA

βBγB

Aircraft ordering problem between A and B for the common glide path (resource 15) : longitudinal and diagonal distances must be respected

15


CommonGlide Path


8

16

173

SRN

1

TOR

MBR

2 6

4 10

11

12

15

7

513

14

RWY 16R

RWY 16L

9

AA

BB

CC

AG AG ModelModel

A1 A4 A15A10 A13 AOUTA16α γ

Potential conflict between C and B on a runway (resource 17) !

RWY 25

0 n

B3 B8 B15B12 B14 BOUTB17

αA

αB

βA

γA

βBγB

COUTC17

γC

αC

16


� Introduction

�Modeling a Terminal Control Area



�MILP formulations �Computational experiments

�Conclusions

17



AG AG viewedviewed asas a MILP (a MILP (MixedMixed--IntegerInteger Linear Linear ProgramProgram))

∈∀−+≥

−−+≥∈∀+≥

AkhjiMxwtt

xMwttFmlwtt

xtf

hkijhkhk

hkijijij

lmlm

),(),,(()1(

),(

),(min

,

,C =

with m ≠ n

18

• Fixed constraintsin F model feasible timing for each aircraft on its specific route, plus α, β, γ constraints on the entrance and exit times.

• Alternative constraintsin A represent the ordering decision between aircraft at air segments and runways, plus holding circle decisions.

=

−+≥

selectediskhif

selectedisjiifx

Mxwtt

hkij

hkijhkhk

),(0

),(1,

,


InvestigatedInvestigated objectiveobjective functionsfunctionsAverage Tardiness

Priority TardinessPriority Equity

Maximum Tardiness

19

Max Completion

Avg Completion

Tardy Jobs P


� Introduction



PresentationPresentation outlineoutline


�Computational experiments�Conclusions

20



DescriptionDescription ofof the test the test casescases

21

� Each row presents 20 disturbed scenarios (ASP instances);

� Entrance delays are randomly generated with various distributions;

� Unavoidable delays cannot be recovered by aircraft rescheduling;

� ASP solutions are computed by means of CPLEX MIP solver 12.0.


A A practicalpracticalschedulingscheduling rulerule

Optimizing an objective and Optimizing an objective and looking at the other objectiveslooking at the other objectives

22



23


A A combinedcombinedapproachapproach


24


� Introduction






�Conclusions

25



AchievementsAchievements• Microscopic ASP-TCA optimization models are proposed.

• Various objective functions and approaches are investigated.

• Computational results for major Italian TCAs demonstrate the existence of relevant gaps between the objective functions.existence of relevant gaps between the objective functions.

• Combining the various objectivesoffers good trade-off solutions.

[email protected]@ing.uniroma3.it

D'ARIANO ATRS 2016

Documents