Dynamic Airline Slot Exchange
during Ground Delay Programs
Thomas Vossen,
University of Colorado, Boulder
Michael Ball, Philippe Montebello
University of Maryland, College Park
Ground Delay Programs:
Motivation:
airline schedules “assume” good weather
SFO: Scheduled Arrivals
18
Number of Arrivals
16
VMC airport
acceptance rate
14
12
10
8
IMC airport
acceptance rate
6
4
2
0
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Ground Delay Programs
delayed
departures
delayed departures
delayed arrivals/
no airborne holding
delayed departures
GDPs under CDM
Resource Allocation Process:
• FAA: initial “fair” slot allocation
[Ration-by-schedule]
• Airlines: flight-slot assignments/reassignments
[Cancellations and substitutions]
• FAA: periodic reallocation to maximize slot utilization
[Compression]
Compression Example
Slot released
by canceled/
4:05
delayed
flight
4:05
4:20
USA 51
UAL 234
USA 345
4:50
AAL 672
4:50
AAL 672
5:10
AAL 95
5:10
AAL 95
Earliest time
of arrival = 4:20
Slot Exchange Alternatives
• Compression as Reallocation
– Dynamic changes to airline “demand profiles”
necessitate (re)rationing
• Compression as Slot Trading
– e.g., Slot Credit Substitutions:
“I am willing to cancel flight f1 if I can move up
flight f2”.
Slot Trading Opportunities
160
40
140
35
Cum. cnx (% of flts in gdp)
Cum. Subs (% of flts in gdp)
Airline Substitution/Cancellation Patterns
120
100
80
60
40
20
0
1
75 149 223 297 371 445 519 593 667 741 815
Cumulative minutes (from start of program)
30
25
20
15
10
5
0
1
51 101 151 201 251 301 351 401 451 501 551 601 651 701 751
Cum ulative m inutes (from start of program )
Consider potential benefits of extending slot
trading framework
– e.g., Increase offers submitted by airlines
From 1-for-1 to 2-for-2 trades
• Compression and/or slot credit substitution can be
interpreted as a 1-for-1 trading system, i.e. offers involve
giving up one slot and getting one in return (many offers
are processed simultaneously)
• What about k-for-k or k-for-n offers, e.g. 2-for-2:
Trade??
Value proposition for
compression & SCS
Value(slot) = $0
unusable because of
cancellation or flight delay
usable slot
Value(slot) > $0
SCS/Compression “trades” are always
driven by the exchange of a slot with
value 0 and a slot with value > 0!!
2-for-2 trades enable airlines to profit by
exchanging pairs of usable slots that result
in an increase in overall value to the carrier.
Airline B
Airline A
B20 (low priority)
A25 (high priority)
s3
s1
s2
A10 (low priority)
s4
B35 (high priority)
A’s value proposition: valA(s3) – valA(s1) + valA(s4) – valA(s2) =
2000 - 1500 + 300 - 500 = $300
B’s value proposition: valB(s1) – valB(s3) + valB(s2) – valB(s4) =
500 - 800 + 2500 - 1800 = $400
Another view of 2-for-2 trading:
generalized substitutions
Normal Substitution
Generalized Substitution
Issues
• System Design:
– How do airlines represent and generate offers?
– Formulation and solution of FAA mediation problem
• System Evaluation:
– Airline objectives and strategies
– Performance Measurement:
• comparison with optimal centralized solution
(system efficiency)
% improvement
30
25
20
15
%Improvement
Compression Benefits
2-for-2 Trading Model
•
•
compression executed after flts
with excessive delay (>2hrs) are
canceled
Global Max.
35
10
5
5
0
0
GDP
bo
s0
106
-0
1
bo
s0
115
-0
1
bo
s0
119
-0
bo
1
s0
130
-0
1
bo
s0
208
-0
1
bo
s0
214
-0
bo
1
s0
221
-0
1
bo
s0
226
-0
1
bo
s0
310
-0
bo
1
s0
313
-0
1
bo
s0
321
-0
1
bo
s0
323
-0
bo
1
s0
330
-0
1
bo
s0
408
-0
1
bo
s0
418
-0
1
bo
s0
106
-0
1
bo
s0
115
-0
1
bo
s0
119
-0
1
bo
s0
130
-0
1
bo
s0
208
-0
1
bo
s0
214
-0
1
bo
s0
221
-0
1
bo
s0
226
-0
1
bo
s0
310
-0
1
bo
s0
313
-0
1
bo
s0
321
-0
1
bo
s0
323
-0
1
bo
s0
330
-0
1
bo
s0
408
-0
1
bo
s0
418
-0
1
Initial Results
• Airline Objective: On-time Performance
proposed offers: all at-least, atmost pairs that improve on-time
performance
Compression
Global Max
GDP
Trading Model
40
40
35
30
25
20
15
10
Initial Results
• Airline Objectives: Passenger Delay Costs
Objective Function 1
Objective Function 2
•
•
Imposed “Staircase” structure
80
70
70
60
60
% Im p ro v em e n t
80
50
40
30
50
40
30
20
20
10
10
0
0
1/
6/
20
0
1/
13 1
/2
00
1/
20 1
/2
00
1/
27 1
/2
00
1
2/
3/
20
0
2/
10 1
/2
00
2/
17 1
/2
00
2/
24 1
/2
00
1
3/
3/
20
0
3/
10 1
/2
00
3/
17 1
/2
00
3/
24 1
/2
00
3/
31 1
/2
00
1
4/
7/
20
0
4/
14 1
/2
00
4/
21 1
/2
00
1
% Im p ro v e m e n t
“Standard” Passenger Delays
GDP
0 1 0 0 1 0 0 1 0 0 1 00 1 0 0 1 0 0 1 0 0 1 00 1 0 0 1 0 0 1 0 0 1 0 0 1 00 1 0 0 1 0 0 1
20
/2
/2
/2
/2
/2
/2
/2
/2
/2
/2
/2
/2
/2
/2
/2
6/
1 0 /1 7 /2 4
1 7 /2 4 3 /3
1 3 /2 0 /2 7 2 /3 /1 0
3 1 4 /7 /1 4
21
1/
3/
2
2/
2
1
1/
3
3
1
3/
4
4/
GDP
Towards a practical system:
offer structure
high priority flights
“move up” range
low priority flights
current position
“move down” range
current position
Offers:
• Airlines willing to accept high priority moves up in exchange
for low priority moves down
Data Requirements:
• 1 new data item per flight -- LET: latest exchange time
Towards a practical system:
mediation problem
• IP formulation that assigns flights to slots in a
manner consistent with offers
– Allows airlines to express relative per unit value of upmoves vs down-moves
• Objective Function
– Efficiency: maximize total distance or number of up
moves
– Equity: aims to distribute benefits in proportion to
offers submitted
Towards a Practical System:
Airline Cost Function for Simulation
3.5
cancellation
cost
x 10
4
3
2.5
2
1.5
1
0.5
0
0
20
40
60
80
100
120
Time (delay)
140
160
180
2 to 3 hrs
Towards a Practical System:
Initial Results
Objective:
•
•
Maximize total distance of Up
moves
Maximize total distance of Up
moves
+ sum of equity components
80%
60%
70%
% Improvement
70%
50%
40%
30%
20%
60%
50%
40%
30%
20%
10%
0%
0%
0 00
01 00 00
00 00 00 -0
00 00 00 00 01
0 0 01
1- 09- 21- -03 - -21 - 21- 17- 21- 04- 21- -2 1 21- 29- 04- 031
6
6
1
6
11
6
6
1
11
1
6
1
6
-0
-0
-0
-0
-0
-0 K -0 A-0 A-0 P-0 L-0 A-0 O-0 L-0
-0
G OS OS W R W R AD AH
S
F
H
T
F
E
V
G
G
J
I
I
P
S
M
S
S
L
B
L
B
E
E
C
GDP
Compression
VG
-0
BO 1-1
S- 1-0
0
BO 1-0 0
S- 9-0
EW 06- 1
2
R 1-0
EW 01 - 0
03
R
-0 -00
6
IA
D 21 -0
6- 00
IA
21
H
-0 -00
1
JF -1
7
K
-0 -0
LG 6 -2 1
A- 1-0
0
LG 1-0 0
A- 4-0
0
M 6-2 0
S
P- 1-0
0
0
PH 6-2
1L00
SE 06A- 210
0
SF 1-2 0
9O
01
-0
ST 1 -0
4
L01 -00
-0
300
10%
C
% Improvement
Objective:
GDP
Trading Model
Summary and Conclusions
• Results illustrate potential benefits of slot trading
framework
• Trading benefits may be limited by carriers
which operate smaller aircraft
– Introducing side payments may “induce” small
carriers to accept delays
• Results depend significantly on airline bid
generation strategies.
– Best airline approach depends on internal schedule
and cost structure, attitude toward risk and strategy of
competitors.
– Competitive simulations currently being constructed.