www'metsci'comcdm

1
www.metsci.com/cdm
2
Schedule Delay
Propagates into Large Delay
3
Non-linear Response to Capacity Reduction
Reducing the airport capacity to 50, generates
over 65,000 minutes of delay at the airport and
139,000 minutes system wide.
300000
System delay
250000
200000
Delay
150000
Airport delay
100000
50000
0
100
96
92
88
83
80
75
72
67
63
58
55
50
47
42
Available Airport Capacity (Percentage)
4
Current CDM GDP Process
FAA/Airline
Evaluation Demand Vs.
Capacity
Issue GDP
Airline Response
GDP Modeling
GDP Revision
Send Proposed GDP
(Substitutions
Compression
(Ration by schedule)
/Extension
Cancellations)
Advisory
Airline Response
(cancellations)
Yes
Is GDP still
required?
Exit loop when
program expires
or is cancelled.
No
End
5
TFM Dependencies
Although, the capacity at an airport is reduced
and ATC delay has been applied, users may cancel
and delay flights as well as notify the FAA of
earliest feasible departure times of flights.
FAA Actions
Issue ATC delay
Reduce ATC delay
Airline Actions
Generate Demand
Cancellations
Airline Delays beyond ATC delay
Earliest Departure/Arrival Time updates
6
P-DELAY PSGR X min DLYD
Passengers
Passengers
0 CNX
60
176 (60 late)
116
93
93
87
87
87
87
P-Delay min 3600
Original P-Delay is 738 greater than revised
Total 443
P-Delay min 26580
7
Airline and FAA Actions can Reduce Delays
8
What Do We Measure?

• Event Predictability (also referred to as
  compliance) ETA predictability
• Departure Compliance
• ETE Fluctuation
• Arrival Compliance
• Arrival rate delivery
• Rate Control Index (RCI)
• What could have been done to improve?
• What are the impacts of new changes (e.g., _5
  minute window)?

9
What Do We Measure?

• Departure Compliance

10
What Do We Measure?

• ETE Fluctuation

11
What Do We Measure?

• Arrival Compliance

12
What Do We Measure?Arrival Rate Delivery

• Rate Control Index (RCI)
• Used to quantify the deviations in the actual
  hourly arrival counts from the planned hourly
  arrival counts
• Can be used to compare one program to any other
• Perfect score 100
• Aggregate RCI
• Looks only at total number of flights in each
  hour
• Nominal RCI
• Looks at which flights arrived in each hour
• Nominal RCI lt Aggregate RCI

13
What Could Have Been Done to Improve?

• More frequent revisions
• Tighter enforcement of departure times
• Higher standards for data quality to remove
• Time-Out cancellations
• Time-Out delays
• Cancel but flew
• Re-evaluated treatment of Pop Up flights

14
What Do We Measure?/- 5 minute Compliance Test

• Controlled flights departing to a GDP are
  required to depart within /- 5 minutes of their
  Expected Departure Clearance Time (EDCT)
• Flights that fail to meet the compliance window
  must request a new EDCT from the ATCSCC.
• Miles In Trail restrictions are lifted for
  flights destined for 7 airports ATL, DFW, ORD,
  EWR, STL, SFO, PHL.
• Goal of improved departure compliance.

15
Impact of New Changes

• Departure

Standard Deviation Before test 32.14 During
test 15.90
16
Impact of New Changes

• En Route

Standard Deviation Before test 12.85 During
test 10.00
17
Impact of New Changes

• Arrivals

Standard Deviation Before test 33.32 During
test 38.12
18
Impact of New Changes

• RCI

Average Aggregate RCI 1/1/01-4/2/02 87.94 Averag
e Aggregate RCI 4/3/02-4/21/02 96.75 RCI has
improved by 10
19
The Need for Revision

• As the demand changes, the need for revisions
  increases.
• The arrival demand would have been smoothed out
  if a revision had been run at 1758Z during this
  PHL program.
• Solid bars current demand
• Hashed bars revised demand

20
The Fallacy of Binning

• Measuring demand in one-hour bins does not
  describe how that demand is distributed
  throughout the hour.
• Each hour bin below has the same total demand.
  However, the demand distributions vary
  dramatically.

s 19.4
s 8.66
s 0
21
Solution to BinningArrival Flow Rates
Viewing the demand as a Flow Rate shows the
actual demand fluctuations, forecasting the
potential for airborne holding viewed as a
moving window average
Viewing the demand in one-hour bins does not
show the underlying fluctuations that exist.
Images from the Flow Rate Analysis Tool
22
Equity Measures

• Affected Average Delay
• Proportions from carrier statistics
• EMF/EMA from power run

23
Equity Measures

• Affected Average Delay

24
Equity Measures

• Proportion of Flights Affected by Carrier

25

• Equity Metric for Airlines (EMA) and Equity
  Metric for Flights (EMF)

• EMA/EMF Compares the amount of imposed ground
  delay to the baseline of airborne holding
• 2-8 Good Equity
• 9-16 Significant deviations from good equity
• 16 Poor equity

26
The Many Facets of Equity

• Long haul Vs. Short haul flights
• GA Vs. Scheduled Carriers
• Regional Vs. Majors
• Overhead stream Vs. Departures
• Equity across flights, carriers, ARTCCs,
  airports,...

• No single vantage point serves for all occasions

27
Distance-based GDPs

• Set inclusion for ground delay based on distance
  from GDP airport
• As GDP radius is increased, delay per flight goes
  down but (risk of) unrecoverable delay goes up
• Find optimal tradeoff

28
Delay Graphic
Air Hold
Unrecov Delay
Average Delay
Max Delay
Minutes
Air Hold
Ouch
220
200
160
180
160
Unrecoverable Delay
140
120
100
Average Delay
80
58
60
40
37
20
200 400 600 800 1000 1200 1400 1600
1800 2000 2200 2400 2600 2800
Miles
29
What Do Airlines Measure?

• Air En Route Times
• Fuel Burn
• Diversions
• Taxi-Out Times
• Passenger Load Factor

30
There is Always a Trade-off in Delaying Actions

• As the time of the event approaches, certainty
  goes up but possible options are reduced.

31
Options Decrease While Knowledge Increases
Increase knowledge earlier
Options
Keep options open longer
Certainty

• Objective increase knowledge while options
  remain open.
• Essential to explore tactical relationships