Do En Route Delays Matter?
Some Preliminary Evidence
Mark Hansen
Huifang Yin
NAS Performance Workshop
3/16/06
1
Overview
‰ Assess the effect of en route delay on
system delay
‰ Determine effect of en route delay to a
particular flight on queuing delay at its
destination airport
‰ Define and calculate a delay multiplier
for flights delayed in ZID
2
Motivating Question (from Arnie
Barnett)
TSA no longer requires passengers
boarding flights to show ID. How much
flight boarding delay does this save?
3
Answer to Motivating Question
~0. (The bottleneck is at the entrance to
the plane.)
4
General Principle
‰ In a queuing situation the effect of an
upstream delay on total delay is not
obvious
‰ The effect may be
¾0
¾A lot less than the upstream delay
¾About the same as the upstream delay
¾A lot more than the upstream delay
5
Queuing Diagram
6
Queuing Diagram Cont.
7
Queuing Diagram Cont.
8
Queuing Diagram Cont.
9
System Delay Impact of En
Route Delay (example 1)
10
System Delay Impact of En
Route Delay (example 1) Cont.
11
System Delay Impact of En
Route Delay (example 1) Cont.
12
System Delay Impact of En
Route Delay (example 1) Cont.
13
System Delay Impact of En
Route Delay All Cases
14
Delay Multiplier
∆d S
M=
dE
Change in system
delay caused by en
route delay
En route delay
duration
15
Measuring M
‰ Simulation
¾Introduce an “exogenous” en route delay
to a specific flight
¾Observe change in system delay that
results
¾Somewhat tedious
‰ Estimation from operational data
16
Estimation Procedure
‰ Measure an en route delay, df, for flight f
‰ Determine when flight f arrived at destination
‰ Estimate when f joined the arrival queue
assuming FIFO
‰ Assume that without the en route delay f
would have joined the queue df minutes
sooner
‰ Calculate how earlier arrival would have
affected total arrival delay at destination
airport
17
System Delay Impact of En
Route Delay
From ASPM
4
From
MITRE en
route delay
study.
(Salomos et
al.)
3
From ASPM
1
2
18
Results
‰ En route delays in ZID for a specific
day
‰ Seven destination airports: ATL, CVG,
DFW, DTW, IND, ORD, SDF
‰ ~2000 flights
19
M-Value Distributions
100%
90%
80%
70%
Grand Total
ATL
60%
CVG
DFW
50%
DTW
IND
40%
ORD
SDF
30%
20%
10%
0%
0
0~1
1
>1
M value
20
M Value Averages
1.60
Average of M
1.38
1.40
Total system delay/Total
center delay
1.21
1.20
1.00
0.99
1.02
1.02
0.97
1.05
1.00
1.00
1.01
0.99
0.91
0.92
1.00
0.93
0.82
0.80
0.60
0.40
0.20
0.00
ATL
CVG
DFW
DTW
IND
ORD
SDF
Grand Total
21
Targeting En Route Delay
Reductions
‰ Identify flights with high and low M
values in real time
‰ Give priority to high M value flights
22
Potential for Targeting En Route
Delay Reductions
Cumulative System Delay Savings
100.00%
90.00%
80.00%
70.00%
60.00%
40% system delay
impact reduction
from 3% en route
delay reduction.
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
80.00%
90.00%
100.00%
Cumulative En Route Delay Savings
23
Conclusions
‰ Not a 1-to-1 mapping of en route delay
to system delay
‰ On average 1 minute of en route delay
causes 1 minute of system delay
‰ But there is a large variation in impact
‰ Implications for
¾Measuring sector performance
¾Prioritizing flights in real time
24
Thanks to
‰ Steve Bradford, Rich Jehlen, Diana
Liang at FAA
‰ George Solomos at MITRE
‰ Jasenka Rakas at UCB
25