1. joint work with James Jones and David Lovell
From Air Traffic Management to Finding a Parking Space: Dynamically Assigning Destination Resources while Enroute
Michael Ball
Robert H Smith School of Business & Institute for Systems Research
University of Maryland
joint work with James Jones and David Lovell

2. Problem Scenario: airplane or auto seeks to obtain or adjust necessary arrival resources while enroute
Sample Resources:
Arrival time slot
Bay at loading dock
Arrival gate
Parking space
Enroute adjustments:
Change in speed
Change in route (incl destination)

3. Recent research on speed control in air traffic management (Jones, Ball & Lovell)
Stochastic Optimization Models for Transferring Delay along Flight Trajectories in Order to Reduce Fuel Usage
Combining Control by CTA and Dynamic Enroute Speed Adjustment to Improve Ground Delay Program Performance

4. The Players:
Pilot / Flight Deck
Airline Operational
Control Center
FAA -- CAAC/ATMB:
Air Traffic Control

5. Air Traffic Control (ATC)
The Players:
Pilot / Flight Deck
Airline Operational
Control Center
FAA:
Air Traffic Control (ATC)
Collaborative decision making initiatives, starting in the mid-90s, have sought to improve FAA-airline joint decision-making.

6. Aircraft Fuel Usage Characteristics
Specific Range
NM/ton
Fuel efficiency is a concave function of speed
General Characteristics:
Cost curves are relatively flat
Cruise speeds can often exceed the maximum range
Slowing down during cruise can increase specific range
Maximum Range
Range of Operation
vmin
vmax
Mach
Number

7. How can aircraft adjust their travel time today?
Reduce or increase speed within certain window.
Take “short-cuts” allowed by air traffic control (ATC)
Path extension techniques to increase time, e.g. vectoring, circular holding patterns, “tromboning”
… 1.) may be restricted in congested areas (more likely to be available on long-haul flights).
… in future (NextGen), new navigational technologies may make individual flights more independent and allow greater freedom in speed determination; more precise and flexible flight planning will decrease 2.) .

8. Insights on arrival delays:
Tromboning and vectoring:

9. Circular holding patterns:
Jan 15, :30-12:45pm
Circular holding patterns:
40 nm
100 nm

10. Jan 15, :30-12:45pm
Two ways to get from SFO to PHL in 4 hrs, 45 min:
40 nm
30 min on green path
4 hrs, 30 min departure and en-route time:
4 hrs, 15 min departure and en-route time:
15 min on red path
100 nm
Fuel usage comparison:
fuel(4 hrs 30 min departure + enroute) ~ fuel (4 hrs 15 min departure + enroute)
fuel(15 min on red path) << fuel (30 min green path)  2nd option uses much more fuel

11. Europe main 34: ave: min
US OEP 34: ave: min
From (Knorr et al 2011)

12. Approach
Develop a process for transferring delay away from the terminal by assigning a Controlled Time of Arrival (CTA) at 500 nm

13. Multiple performance criteria / objectives: Throughput
Multiple Integer programming and stochastic integer programming models developed:
Multiple performance criteria / objectives:
Throughput
Delay transfer / fuel usage
Equitable treatment of airlines
Stochastic aspects:
Unmanaged flights
Airport capacity

14. Sample results: delay transfer graph
On average ~ 19% of delay is transferred;
Why can’t more delay be transferred?
Limit on speed ranges / time that can be absorbed within 500 nmi
Congestion due to unmanaged flights
Test case: ATL, 5/1/2012

15. Research described today:
Stochastic Optimization Models for Transferring Delay along Flight Trajectories in Order to Reduce Fuel Usage
Combining Control by CTA and Dynamic Enroute Speed Adjustment to Improve Ground Delay Program Performance

16. Ground Delay Programs control = flight departure time
delayed
departures
delayed departures
control = flight departure time
decision variable = flight arrival time (slot)
delayed arrivals/
no airborne holding
delayed departures 16

17. GDP Planning Process
These processes adjust the arrival slots assigned to flights, which in turn are used to adjust the controlled time of departure (CTD) of each flight.

18. Motivation: flight exemptions
Ground delay SFO EWR flight vs ORD  EWR flight??
Choice of flights to reassign arrival slot from 4:00  4:45 (EST)
SFO flight  departure time change: 10:50  11:35 (EST)
ORD flight  departure time change: 1:50  2:35 (EST)
5 hr, 10 min flight
Figure 10: United States National Airspace System
Suppose weather at EWR clears at 12:45 (EST)  no delays are necessary!!
EWR
ORD
SFO

19. Motivation: flight exemptions
Ground delay SFO EWR flight vs ORD  EWR flight??
Choice of flights to reassign arrival slot from 4:00  4:45 (EST)
SFO flight  departure time change: 10:50  11:35 (EST)
ORD flight  departure time change: 1:50  2:35 (EST)
5 hr, 10 min flight
Figure 10: United States National Airspace System
Suppose weather at EWR clears at 12:45 (EST)  no delays are necessary.!!
EWR
ORD
SFO
At 12:45 SFO flight would already be en-route  delay already taken (unecessarily).

20. Motivation: flight exemptions
Ground delay SFO EWR flight vs ORD  EWR flight??
Choice of flights to reassign arrival slot from 4:00  4:45 (EST)
SFO flight  departure time change: 10:50  11:35 (EST)
ORD flight  departure time change: 1:50  2:35 (EST)
5 hr, 10 min flight
Figure 10: United States National Airspace System
Suppose weather at EWR clears at 12:45 (EST)  no delays are necessary.!!
EWR
ORD
SFO
At 12:45 ORD flight would still be on the ground  delay could be rescinded.

21. Lesson: it is better to “focus” delays on short-haul flights; this is typically done by exempting flights outside an exemption radius from the ground delay program
Figure 10: United States National Airspace System
EWR
ORD
SFO

22. Speed Control in Ground Delay Programs and a New Control Architecture
Current practice: while planning is based on controlled times of arrival (CTAs), actual control is based on a controlled time of departure (CTD) to flights; the CTD is obtained from the CTA by subtracting the estimated flight time.
Controlling based on CTAs (rather than CTDs) may offer a more attractive means for controlling flights:
Provides carriers more flexibility and control
Allows for system-wide trade-offs
Delay allocation under conventional GDP Planning
Travel Time GD
STD
CTD
STA
CTA

23. Delay allocation under CTA based GDP Planning
Speed Control in Ground Delay Programs and a New Control Architecture
Current practice: while planning is based on controlled times of arrival (CTAs), actual control is based on a controlled time of departure (CTD) to flights; the CTD is obtained from the CTA by subtracting the estimated flight time.
Controlling based on CTAs (rather than CTDs) may offer a more attractive means for controlling flights:
Provides carriers more flexibility and control
Allows for system-wide trade-offs
Flexibility in departure time
Delay allocation under CTA based GDP Planning
Variable
Delay
Travel Time
GDmin
STD
CTDmin
CTDmax
STA
CTA

24. Delay allocation under CTA based GDP Planning
Speed Control in Ground Delay Programs and a New Control Architecture
Current practice: while planning is based on controlled times of arrival (CTAs), actual control is based on a controlled time of departure (CTD) to flights; the CTD is obtained from the CTA by subtracting the estimated flight time.
Controlling based on CTAs (rather than CTDs) may offer a more attractive means for controlling flights:
Provides carriers more flexibility and control
Allows for system-wide trade-offs
Dynamically adjust arrival time
Delay allocation under CTA based GDP Planning
Variable
Delay
Travel Time
GDmin
STD
CTDmin
CTDmax
STA
CTA

25. FAA procedural modifications
New GDP Architecture
FAA procedural modifications
Replace the use of CTDs with CTAs in GDP planning
Remove the exemption radius
Allow en route speed changes by carriers
Airline decision making modifications
Incorporate speed changes into substitution and cancellation process
Introduce decision-support models to support airline decision-making with substitutions and cancellations (including airborne flights and new exemption policy)  airlines take over weather “hedging” formally done by FAA

26. Airline cancellation and substitution problem (w CTD control):
assignment fixed
Delta flights:
4:00
4:04
4:08
alternate assignments
Slots assigned to Delta
4:12
4:16
4:20
4:24
4:28
4:32
airborne flights
Cancel
flights on ground 26

27. Airline cancellation and substitution problem (w CTA control):
Delta flights:
4:00
4:04
4:08
alternate assignments
Slots assigned to Delta
4:12
4:16
4:20
4:24
4:28
4:32
airborne flights
Cancel
flights on ground 27

28. Airline cancellation and substitution problem (w CTA control):
Delta flights:
GDP canceled: newly avail slot
4:00
4:04
4:08
alternate assignments
Slots assigned to Delta
4:12
4:16
4:20
4:24
option to adjust speed and arrive in “new” slot
4:28
4:32
airborne flights
Cancel
flights on ground 28

29. System modifications:
FAA role / problem becomes easier
Changes required in the manner in which slot allocations are dynamically adjusted over time
Airlines’ problem becomes harder:
must dynamically adjust flight speeds and coordinate airborne flight arrival times with departure times of flights on the ground
stochastic integer programming models support airline flight planning and dynamic flight adjustments

30. Back to Original Problem Scenario:
Parking problem:
New technology – info distributed dynamically to drivers on which spaces are available, when will spaces become available; drivers can reserve enroute, etc

31. Key Issues for both cases
How are resources (slots / parking spaces) allocated:
administrative rules
market mechanism, e.g. auction
2 level approaches: long-term allocation of reservation or priority via admin or market mechanism / short term allocation based on assigned priority
Stochastic aspects:
travel times
airport capacity (number slots available)
when parking space will be vacated
Dynamics:
trading and dynamic reallocation
User vs system optimal; system vs profit maximizing
Periodic optimization / re-optimization vs real-time / transaction-oriented control

32. Air traffic mgmt. scenarios
Strong winds encountered; will arrive 10 min late.
AOC
ATC

33. Air traffic mgmt. scenarios
Some winds encountered; prefer to arrive 10 min late.
AOC
ATC

34. Air traffic mgmt. scenarios
New slot available you can arrive 10 min earlier.
AOC
ATC

35. Air traffic mgmt. scenarios
Several new slots available; how can you make use of them
AOC
All of these scenarios will require optimization / coordination / trading among multiple airlines
ATC

36. Auto / parking scenarios
My estimated arrival time is 8:30 AM, I want a nearby parking space.
Parking coordination center

37. Auto / parking scenarios
I am willing to park further away than previously indicated.
Parking Coordination Center

38. Auto / parking scenarios
I am willing to pay more than previously indicated.
Parking Coordination Center

39. Auto / parking scenarios
Don’t need space – I have decided to skip work and go play golf.
Parking Coordination Center

40. Auto / parking scenarios
A space likely will be available near location X.
Parking Coordination Center

41. Auto / parking scenarios
Your assigned space has not been vacated it most likely will not be available.
Parking Coordination Center

42. Auto / parking scenarios
A better space is available for a higher price.
Parking Coordination Center

43. Issues of user vs system optimal
Final Thoughts
Problems discussed are stochastic, dynamic and have multiple decision makers.
Solution architecture can employ iterative optimization / re-optimization or real-time control.
Issues of user vs system optimal
Good solution approaches should take mechanism design perspective.

44. Some References
Zou, B, N Kafle, O Wolfson & J Lin (2015) A mechanism design based approach to solving parking slot assignment in the information era, Transportation Research Part B, 81, 631 – 653.
Gang Y & C Cassandras (2012) A new “Smart Parking” System Infrastructure and Implementation, Procedia: Social and Behavioral Sciences, 54, 1278 – 1287.
Jones, J, D Lovell & M Ball (2017) Stochastic Optimization Models for Transferring Delay along Flight Trajectories in Order to Reduce Fuel Usage, Transportation Science, Articles in Advance, Published Online: February 24,
Jones, J., D. Lovell and M. Ball (2015) Combining Control by CTA and Dynamic Enroute Speed Adjustment to Improve Ground Delay Program Performance, in Proceedings of the 11th USA/Europe Air Traffic Management R&D Seminar.