Two-stage stochastic programming for aircraft arrival

scheduling under uncertainty

GT2L - Examination for Best PhD thesis 2020

Ahmed Khassiba
Université de Montréal - Université de Toulouse III - ENAC

Fabian Bastin Sonia Cafieri

Bernard Gendron Marcel Mongeau

Université de Montréal École Nationale de l’Aviation Civile

Ecole

April 22nd , 2021

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 1 / 25
Outline

1 Motivation

2 Two-stage stochastic programming model

3 Approach evaluation and validation

4 Solution method

5 Extension to the multiple-IAF case

6 Perspectives

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 2 / 25
 Motivation

Outline

1 Motivation

2 Two-stage stochastic programming model

3 Approach evaluation and validation

4 Solution method

5 Extension to the multiple-IAF case

6 Perspectives

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 3 / 25
 Motivation

Holding stack experience

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 4 / 25
 Motivation

Holding stack experience

Steady predicted annual air traffic

growth (despite Covid-19)

(source: IATA May 2020)

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 4 / 25
 Motivation

Holding stack experience

Steady predicted annual air traffic Limited runway capacity and a small
growth (despite Covid-19) margin of expansion

(source: IATA May 2020) (source: Wikipedia)

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 4 / 25
 Motivation

Holding stack experience

Steady predicted annual air traffic Limited runway capacity and a small
growth (despite Covid-19) margin of expansion

(source: IATA May 2020) (source: Wikipedia)

⇒ optimizing the use of the existing infrastructure

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 4 / 25
 Motivation

Aircraft Landing Problem

Wake turbulence categories: Heavy (H), Medium (M), Light (L)

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 5 / 25
 Motivation

Aircraft Landing Problem

Wake turbulence categories: Heavy (H), Medium (M), Light (L)
Minimum separations in nautical miles (NM)
Following aircraft
H M L
H 4 5 6
Leading aircraft M 2.5 2.5 4
L 2.5 2.5 2.5

H - L - M: 6 + 2.5 = 8.5 NM
M L H

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 5 / 25
 Motivation

Aircraft Landing Problem

Wake turbulence categories: Heavy (H), Medium (M), Light (L)
Minimum separations in nautical miles (NM)
Following aircraft
H M L
H 4 5 6
Leading aircraft M 2.5 2.5 4
L 2.5 2.5 2.5

H - L - M: 6 + 2.5 = 8.5 NM
M L H

L - M - H: 2.5 + 2.5 = 5 NM
H M L

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 5 / 25
 Motivation

Aircraft Landing Problem

Wake turbulence categories: Heavy (H), Medium (M), Light (L)
Minimum separations in nautical miles (NM)
Following aircraft
H M L
H 4 5 6
Leading aircraft M 2.5 2.5 4
L 2.5 2.5 2.5

H - L - M: 6 + 2.5 = 8.5 NM
M L H

L - M - H: 2.5 + 2.5 = 5 NM
H M L

Aircraft Landing Problem (ALP)

Given n aircraft about to land, find a landing schedule (sequence and times) so as to
maximize the runway throughput subject to safety constraints (runway threshold
separations mainly).

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 5 / 25
 Motivation Literature review

Related literature

Static vs. dynamic

Static / offline: a fixed set of aircraft to schedule.
Earliest papers: Dear (1976); Psaraftis (1978)
Reference MILP formulation: Beasley et al. (2000)
Dynamic programming under CPS: Balakrishnan and Chandran (2010)
Dynamic programming with aircraft classes: Briskorn and Stolletz (2014); Lieder et al. (2015)

Dynamic / online: an evolving (over time) set of aircraft to schedule.

Earliest paper: Dear and Sherif (1991)
Common approach: Rolling horizon
Reference papers: Hu and Chen (2005); Bennell et al. (2017)

Deterministic vs. under uncertainty

Deterministic: all input data known with certainty. → Most of the literature
Under uncertainty: some uncertain input data, e.g., uncertain aircraft arrival
times
Probabilistic approach: Meyn and Erzberger (2005); Chandran and Balakrishnan (2007); Lee (2008)
Two-stage stochastic programming: Sölveling et al. (2011); Sölveling (2012)
Robust programming: Kapolke et al. (2016); Heidt et al. (2016); Heidt (2017)

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 6 / 25
 Motivation Arrival Management

From Arrival Manager (AMAN)

to Extended Arrival Manager (E-AMAN)
AMAN: since 1990’s, decision support tool for Air Traffic Controllers to compute
an optimal landing schedule, 30-45 minutes ahead
E-AMAN: AMAN with an extended operational horizon to 2-3 hours before
landing (300 - 500 NM around an airport)
Start the scheduling earlier
→ More flexibility to build the sequence
→ More eco-efficient air traffic control techniques; e.g., speed reduction

AMAN screenshot at Paris Charles-de-Gaulle

airport (source: https://www.ecologique-solidaire.gouv.fr)
Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 7 / 25
 Motivation Arrival Management

From Arrival Manager (AMAN)

to Extended Arrival Manager (E-AMAN)
AMAN: since 1990’s, decision support tool for Air Traffic Controllers to compute
an optimal landing schedule, 30-45 minutes ahead
E-AMAN: AMAN with an extended operational horizon to 2-3 hours before
landing (300 - 500 NM around an airport)
Start the scheduling earlier
→ More flexibility to build the sequence
→ More eco-efficient air traffic control techniques; e.g., speed reduction
/!\ Taking into account uncertainty about arrival times
sources of uncertainty: e.g., weather, encountered traffic situation.

AMAN screenshot at Paris Charles-de-Gaulle

airport (source: https://www.ecologique-solidaire.gouv.fr)
Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 7 / 25
 Motivation Arrival Management

This thesis

New operational setup: Extended Arrival Management (2 - 3 hours

before landing)
6= Short planning horizon in most of the literature

Optimization problem under uncertainty with a “natural”

decomposition into two decision stages:
The first stage ⇒ extended operational horizon, scheduling for the
Initial Approach Fix (IAF)
The second stage ⇒ short operational horizon, scheduling for the
runway threshold
6= In the literature, two-stage stochastic models:
Sölveling et al. (2011): Sequence of aircraft class (first stage),
Sequence of individual aircraft (second stage).
Sölveling and Clarke (2014): Sequence of individual aircraft (first
stage), Target landing times (second stage).
Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 8 / 25
 Motivation Arrival Management

Contributions

1 Two-stage stochastic programming model

2 A computational study to assess the value of the modeling approach
3 Development of a Benders decomposition method
4 Model extension to the multiple-IAF case

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 9 / 25
 Motivation Arrival Management

Publications and Conferences

1 (Published) Khassiba, A., Bastin, F., Gendron, B., Cafieri, S., and Mongeau, M. (2019).
Extended aircraft arrival management under uncertainty: A computational study.
Journal of Air Transportation, 27(3):131–143
2 (Accepted in March 2020) Khassiba, A., Bastin, F., Cafieri, S., Gendron, B., and
Mongeau, M. (2020a). Two-stage stochastic mixed-integer programming with chance
constraints for extended aircraft arrival management.
Transportation Science
3 (In preparation) Khassiba, A., Bastin, F., Cafieri, S., Gendron, B., and Mongeau, M.
(2020b). Two-stage stochastic programming models for the extended aircraft arrival
management problem with multiple pre-scheduling points
Conferences:
EUROPT 2017, Montréal, Canada
INFORMS 2017, Québec city, Canada
ISMP 2018, Bordeaux, France
ROADEF 2019, Le Havre, France

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 10 / 25
 Two-stage stochastic programming model

Outline

1 Motivation

2 Two-stage stochastic programming model

3 Approach evaluation and validation

4 Solution method

5 Extension to the multiple-IAF case

6 Perspectives

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 11 / 25
 Two-stage stochastic programming model

2-3 hours from landing (E-AMAN horizon)

Decisions / E-AMAN outputs:
Target IAF times, xi
Target IAF sequence ≡ landing
sequence

 1 if aircraft i directly
δij = precedes aircraft j
0 otherwise


• Constraints: IAF safety separation (S I ),

IAF
2 h o urs
3 h o urs

IAF time windows.

Uncertain actual

IAF times: xi + ωi ,
RWY ω i ∼ N 0, σ 2 .

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 12 / 25
 Two-stage stochastic programming model

2-3 hours from landing (E-AMAN horizon)

Decisions / E-AMAN outputs:
Target IAF times, xi
Target IAF sequence ≡ landing
sequence

 1 if aircraft i directly
δij = precedes aircraft j
0 otherwise


• Constraints: IAF safety separation (S I ),

IAF
2 h o urs
3 h o urs

IAF time windows.

Uncertain actual

IAF times: xi + ωi ,
RWY ω i ∼ N 0, σ 2 .

30-45 min. from landing (AMAN horizon)

• Decisions / AMAN outputs:
r Landing times, yi
r Landing sequence ≡ Target IAF
sequence
• Constraints: Runway safety separation
(Sij ), landing time windows.
∼ Classic deterministic ALP (with a fixed
sequence)
Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 12 / 25
 Two-stage stochastic programming model

First stage Second stage

· ≥ SI · ≥ SI · ≥ S23 · ≥ S12

δ23 δ12
IAF RWY
3 2 1 3 2 1
(H) (M) (L) (H) (M) (L)

First-stage objective function

Minimize landing sequence
X length (in terms of final-approach minimum
Sij δij , e.g., Obj = S12 + S23 6= 2 × S I


separations, Sij ): min
δ,x
i6=j

Second-stage objective function

P
Minimize total time deviation w/r unconst. landing time: miny i∈A |yi − Ui |
Minimize
P total cost of time deviation w/r unconst. landing time:
miny i∈A fi (yi − Ui )
Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 13 / 25
 Two-stage stochastic programming model

Two-stage stochastic programming model

min f1 (δ, x)
δ, x

+
X X
δji = 1 , δij = 1 i ∈A In-out constraints
j∈A+ \{i} j∈A+ \{i}

I I
xj ≥ xi + S − Mij (1 − δij ) (i, j) ∈ A × A, i 6= j IAF separation constr.

I I
Ei ≤ xi ≤ Li i ∈A IAF time windows

+ +
δij ∈ {0, 1} (i, j) ∈ A ×A , i 6= j Binary nature

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 14 / 25
 Two-stage stochastic programming model

Two-stage stochastic programming model

min f1 (δ, x)
δ, x

+
X X
δji = 1 , δij = 1 i ∈A In-out constraints
j∈A+ \{i} j∈A+ \{i}

I I
xj ≥ xi + S − Mij (1 − δij ) (i, j) ∈ A × A, i 6= j IAF separation constr.

I Iα
P(xj + ωj ≥ xi + ωi + S − Mij (1 − δij )) ≥ α IAF separation probability constr.
(i, j) ∈ A × A, i 6= j

I I
Ei ≤ xi ≤ Li i ∈A IAF time windows

+ +
δij ∈ {0, 1} (i, j) ∈ A ×A , i 6= j Binary nature

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 14 / 25
 Two-stage stochastic programming model

Two-stage stochastic programming model

min f1 (δ, x)
δ, x

+
X X
δji = 1 , δij = 1 i ∈A In-out constraints
j∈A+ \{i} j∈A+ \{i}

I I
xj ≥ xi + S − Mij (1 − δij ) (i, j) ∈ A × A, i 6= j IAF separation constr.

I Iα
P(xj + ωj ≥ xi + ωi + S − Mij (1 − δij )) ≥ α IAF separation probability constr.
(i, j) ∈ A × A, i 6= j

I I
Ei ≤ xi ≤ Li i ∈A IAF time windows

+ +
δij ∈ {0, 1} (i, j) ∈ A ×A , i 6= j Binary nature

Q(δ, x, ω) = min f2 (x, ω, y )

y

yj ≥ yi + Sij − Mij (1 − δij ) (i, j) ∈ A × A, i 6= j Final-approach separation constr.

Vi ≤ yi − (xi + ωi ) ≤ Vi i ∈A Landing time windows

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 14 / 25
 Two-stage stochastic programming model

Two-stage stochastic programming model

min f1 (δ, x) + λ Eω [Q(δ, x, ω)]
δ, x

+
X X
δji = 1 , δij = 1 i ∈A In-out constraints
j∈A+ \{i} j∈A+ \{i}

I I
xj ≥ xi + S − Mij (1 − δij ) (i, j) ∈ A × A, i 6= j IAF separation constr.

I Iα
P(xj + ωj ≥ xi + ωi + S − Mij (1 − δij )) ≥ α IAF separation probability constr.
(i, j) ∈ A × A, i 6= j

I I
Ei ≤ xi ≤ Li i ∈A IAF time windows

+ +
δij ∈ {0, 1} (i, j) ∈ A ×A , i 6= j Binary nature

Q(δ, x, ω) = min f2 (x, ω, y )

y

yj ≥ yi + Sij − Mij (1 − δij ) (i, j) ∈ A × A, i 6= j Final-approach separation constr.

Vi ≤ yi − (xi + ωi ) ≤ Vi i ∈A Landing time windows

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 14 / 25
 Two-stage stochastic programming model

Model challenges

1 Computing / estimating expectation with continuous random variables:

Eω [Q(δ, x, ω)]

2 Handling chance constraints: P (. . . ) ≥ α

3 Handling convex piecewise linear objective function: f2 (x, ω, y )

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 15 / 25
 Two-stage stochastic programming model

Model challenges

1 Computing / estimating expectation with continuous random variables:

Eω [Q(δ, x, ω)]
→ Sample Average Approximation (Shapiro and Homem-de Mello, 2000)
Building an approximate problem with a fixed number of scenarios, nS
⇒ nS independent copies of the second-stage problem
Defining an appropriate number of scenarios

2 Handling chance constraints: P (. . . ) ≥ α

3 Handling convex piecewise linear objective function: f2 (x, ω, y )

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 15 / 25
 Two-stage stochastic programming model

Model challenges

1 Computing / estimating expectation with continuous random variables:

Eω [Q(δ, x, ω)]
→ Sample Average Approximation (Shapiro and Homem-de Mello, 2000)
Building an approximate problem with a fixed number of scenarios, nS
⇒ nS independent copies of the second-stage problem
Defining an appropriate number of scenarios

2 Handling chance constraints: P (. . . ) ≥ α

→ linearize by inverting a cumulative distribution function

3 Handling convex piecewise linear objective function: f2 (x, ω, y )

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 15 / 25
 Two-stage stochastic programming model

Model challenges

1 Computing / estimating expectation with continuous random variables:

Eω [Q(δ, x, ω)]
→ Sample Average Approximation (Shapiro and Homem-de Mello, 2000)
Building an approximate problem with a fixed number of scenarios, nS
⇒ nS independent copies of the second-stage problem
Defining an appropriate number of scenarios

2 Handling chance constraints: P (. . . ) ≥ α

→ linearize by inverting a cumulative distribution function

3 Handling convex piecewise linear objective function: f2 (x, ω, y )

→ linearize using additional continuous variables and adequate constraints

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 15 / 25
 Two-stage stochastic programming model

Model challenges

1 Computing / estimating expectation with continuous random variables:

Eω [Q(δ, x, ω)]
→ Sample Average Approximation (Shapiro and Homem-de Mello, 2000)
Building an approximate problem with a fixed number of scenarios, nS
⇒ nS independent copies of the second-stage problem
Defining an appropriate number of scenarios

2 Handling chance constraints: P (. . . ) ≥ α

→ linearize by inverting a cumulative distribution function

3 Handling convex piecewise linear objective function: f2 (x, ω, y )

→ linearize using additional continuous variables and adequate constraints

⇒ (Large-scale) Mixed-Integer Linear Problem

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 15 / 25
 Approach evaluation and validation

Outline

1 Motivation

2 Two-stage stochastic programming model

3 Approach evaluation and validation

4 Solution method

5 Extension to the multiple-IAF case

6 Perspectives

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 16 / 25
 Approach evaluation and validation

Two main validation experiments

1 Comparison between stochastic-programming solutions and solutions based
on first come, first served policy (FCFS) (Khassiba et al., 2019)
Key results (in high-density traffic):
less IAF conflicts,
increased runway throughput,
decreased time-to-lose closely to the terminal area.

⇒ successful validation of E-AMAN concept by two-stage stochastic prog.

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 17 / 25
 Approach evaluation and validation

Two main validation experiments

1 Comparison between stochastic-programming solutions and solutions based
on first come, first served policy (FCFS) (Khassiba et al., 2019)
Key results (in high-density traffic):
less IAF conflicts,
increased runway throughput,
decreased time-to-lose closely to the terminal area.

⇒ successful validation of E-AMAN concept by two-stage stochastic prog.

2 Computation of Value of the Stochastic Solution (VSS) (Khassiba et al., 2020a)

VSS: a standard measure to verify the relevance of taking into account

uncertainty via two-stage stochastic programming
Key results: Most benefit of two-stage stochastic program when:
high-density air traffic,
many degrees of freedom for re-scheduling,
the second-stage cost has a great importance for the stakeholder.

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 17 / 25
 Solution method

Outline

1 Motivation

2 Two-stage stochastic programming model

3 Approach evaluation and validation

4 Solution method

5 Extension to the multiple-IAF case

6 Perspectives

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 18 / 25
 Solution method

Benders decomposition
Benders decomposition applied to two-stage stochastic programming
A.k.a. L-Shaped (Van Slyke and Wets, 1969):
Benders master problem → first-stage problem
Benders subproblem(s) → second-stage problem(s)

Benders decomposition features

Variable partitioning
Level of subproblem aggregation
Acceleration techniques (single search tree, stabilization, Pareto-optimal
cuts, etc.)

Main results
CPLEX automatic partitioning unsuitable for our case.
Disaggregated Benders outperforms CPLEX default Branch-and-Cut.
Partially-aggregated Benders performs best.
Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 19 / 25
 Extension to the multiple-IAF case

Outline

1 Motivation

2 Two-stage stochastic programming model

3 Approach evaluation and validation

4 Solution method

5 Extension to the multiple-IAF case

6 Perspectives

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 20 / 25
 Extension to the multiple-IAF case

Operational setup with 2 IAFs feeding a single RWY

First stage: several IAFs, schedule for each Second stage: schedule for the runway Two problem variants
IAF
Target landing times Variant 1: IAF assignment as
[ additional decision? ] IAF Target landing sequence a first-stage decision
assignment
(inherited from the first stage) Variant 2: IAF assignment as
Target IAF times an input (fixed)
Target IAF sequence
(A single!) target landing sequence
Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 21 / 25
 Extension to the multiple-IAF case

Some modeling highlights and first results

Definition of the sequencing variables updated (as in scheduling theory 6= TSP):


1 if aircraft a lands before aircraft b
δab =
0 otherwise

Example:
− 1 1 1 1
 
IAF 1 sequence: 1 - 3 - 5
0 − 1 1 1
IAF 2 sequence: 2 - 4 δ = 0

0 − 1 1

Runway sequence: 1 - 2 - 3 - 4 - 5
0 0 0 − 1
0 0 0 0 −

Variant 1: additional binary decision variables:

assignment variables (aircraft to IAF),
variables capturing whether two aircraft are assigned to the same IAF

First results: slight increase in the expected makespan (last landing time)

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 22 / 25
 Perspectives

Outline

1 Motivation

2 Two-stage stochastic programming model

3 Approach evaluation and validation

4 Solution method

5 Extension to the multiple-IAF case

6 Perspectives

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 23 / 25
 Perspectives

Perspectives

More operational aspects

flight initial status: on-ground or airborne,
aircraft-type and flight-phase dependent delay cost,
dynamic case, multiple runways.

Uncertainty
Dependent random deviations,
Quantification (Tielrooij et al., 2015).

Solution method
Benders decomposition acceleration: e.g., solving second-stage (primal)
problem by dynamic programming (Faye, 2018), and generate
Pareto-optimal cuts by solving M-W problem (Magnanti and Wong, 1981).
Scenarios generation (Crainic et al., 2016) and clustering.

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 24 / 25
 Thank you for your attention!

Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 25 / 25
 References

—————– —————–
Balakrishnan, H. and Chandran, B. G. (2010). Algorithms for scheduling
runway operations under constrained position shifting. Operations
Research, 58(6):1650–1665.
Beasley, J. E., Krishnamoorthy, M., Sharaiha, Y. M., and Abramson, D.
(2000). Scheduling aircraft landings: The static case. Transportation
Science, 34(2):180–197.
Benders, J. F. (1962). Partitioning procedures for solving mixed-variables
programming problems. Numerische Mathematik, 4(1):238–252.
Bennell, J. A., Mesgarpour, M., and Potts, C. N. (2017). Dynamic
scheduling of aircraft landings. European Journal of Operational
Research, 258(1):315–327.
Briskorn, D. and Stolletz, R. (2014). Aircraft landing problems with
aircraft classes. Journal of Scheduling, 17(1):31–45.
Chandran, B. G. and Balakrishnan, H. (2007). A dynamic programming
algorithm for robust runway scheduling. In Proceedings of the 2007
American Control Conference, pages 1161–1166. IEEE.
Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 25 / 25
 References

Crainic, T. G., Hewitt, M., Maggioni, F., and Rei, W. (2016). Partial
Benders decomposition strategies for two-stage stochastic integer
programs. Technical Report 2016-37, CIRRELT, Université de Québec à
Montréal, Canada.
Dear, R. G. (1976). The dynamic scheduling of aircraft in the near
terminal area. Technical Report R76-9, Massachusetts Institute of
Technology. Flight Transportation Laboratory.
Dear, R. G. and Sherif, Y. S. (1991). An algorithm for computer assisted
sequencing and scheduling of terminal area operations. Transportation
Research Part A, 25(2):129–139.
Faye, A. (2018). A quadratic time algorithm for computing the optimal
landing times of a fixed sequence of planes. European Journal of
Operational Research, 270:1148–1157.
Heidt, A. (2017). Uncertainty Models for Optimal and Robust ATM
Schedules. PhD thesis, Friedrich Alexander University, Erlangen,
Germany.
Heidt, A., Helmke, H., Kapolke, M., Liers, F., and Martin, A. (2016).
Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 25 / 25
 References

Robust runway scheduling under uncertain conditions. Journal of Air

Transport Management Part A, 56:28–37.
Hu, X.-B. and Chen, W.-H. (2005). Receding horizon control for aircraft
arrival sequencing and scheduling. IEEE Transactions on Intelligent
Transportation Systems, 6(2):189–197.
Kapolke, M., Fürstenau, N., Heidt, A., Liers, F., Mittendorf, M., and
Weiß, C. (2016). Pre-tactical optimization of runway utilization under
uncertainty. Journal of Air Transport Management Part A, 56:48–56.
Khassiba, A., Bastin, F., Cafieri, S., Gendron, B., and Mongeau, M.
(2020a). Two-stage stochastic mixed-integer programming with chance
constraints for extended aircraft arrival management. Transportation
Science.
Khassiba, A., Bastin, F., Cafieri, S., Gendron, B., and Mongeau, M.
(2020b). Two-stage stochastic programming models for the extended
aircraft arrival management problem with multiple pre-scheduling points.
Khassiba, A., Bastin, F., Gendron, B., Cafieri, S., and Mongeau, M.
(2019). Extended aircraft arrival management under uncertainty: A
computational study. Journal of Air Transportation, 27(3):131–143.
Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 25 / 25
 References

Lee, H. (2008). Tradeoff evaluation of scheduling algorithms for

terminal-area air traffic control. Master’s thesis, Massachusetts Institute
of Technology.
Lieder, A., Briskorn, D., and Stolletz, R. (2015). A dynamic programming
approach for the aircraft landing problem with aircraft classes. European
Journal of Operational Research, 243(1):61–69.
Magnanti, T. L. and Wong, R. T. (1981). Accelerating Benders
decomposition: Algorithmic enhancement and model selection criteria.
Operations Research, 29(3):464–484.
Meyn, L. A. and Erzberger, H. (2005). Airport arrival capacity benefits due
to improved scheduling accuracy. In Proceedings of the 5th Aviation,
Technology Integration and Operations and the 16th Lighter-Than-Air
Systems Technology and Balloon Systems Conferences.
Psaraftis, H. N. (1978). A dynamic programming approach to the aircraft
sequencing problem. Technical report, Massachusetts Institute of
Technology.
Shapiro, A. and Homem-de Mello, T. (2000). On the rate of convergence
Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 25 / 25
 of optimal solutions of Monte Carlo approximations of stochastic
programs. SIAM Journal on Optimization, 11(1):70–86.
Sölveling, G. (2012). Stochastic programming methods for scheduling of
airport runway operations under uncertainty. PhD thesis, Georgia
Institute of Technology, Atlanta, GA.
Sölveling, G. and Clarke, J.-P. (2014). Scheduling of airport runway
operations using stochastic branch and bound methods. Transportation
Research Part C, 45:119–137.
Sölveling, G., Solak, S., Clarke, J.-P., and Johnson, E. (2011). Runway
operations optimization in the presence of uncertainties. Journal of
Guidance, Control, and Dynamics, 34(5):1373–1382.
Tielrooij, M., Borst, C., Van Paassen, M. M., and Mulder, M. (2015).
Predicting arrival time uncertainty from actual flight information. In
Proceedings of the 11th USA/Europe Air Traffic Management Research
and Development Seminar, pages 577–586. FAA/EUROCONTROL.
Van Slyke, R. M. and Wets, R. (1969). L-Shaped linear programs with
applications to optimal control and stochastic programming. SIAM
Journal on Applied Mathematics, 17(4):638–663.
Ahmed Khassiba (UdeM-ENAC) Stochastic programming for E-AMAN April 22nd , 2021 25 / 25