Download
# Charles N. Glover PowerPoint Presentation, PPT - DocSlides

jane-oiler | 2016-04-20 | General
** Tags : **
charles-n-glover
fca flight
flights delay
flight
fca
delay
flights
model
weather
cost
operator
ansp
carrier
cdm
expected
time
ground
operators
airlines
arrival
capacity
### Presentations text content in Charles N. Glover

Show

Michael O. Ball. David J. Lovell. Collaborative Approaches to the Application of . Enroute. Traffic Flow Management Optimization Models. Weather in the NAS. 2. Weather is unpredictable. How long will it last? . ID: 285705

- Views :
**41**

**Direct Link:**- Link:https://www.docslides.com/jane-oiler/charles-n-glover
**Embed code:**

Download this presentation

DownloadNote - The PPT/PDF document "Charles N. Glover" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.

Slide1

Slide24

Slide25

Slide26

Charles N. GloverMichael O. BallDavid J. Lovell

Collaborative Approaches to the Application of

Enroute

Traffic Flow Management Optimization Models

Slide2Weather in the NAS

2

Weather is unpredictableHow long will it last? Where is it headed? How dangerous is it? These are research questions for meteorologists. These questions also affect the NAS.

Slide3The Impact on Flights

3

Suppose many flights are scheduled to travel through a region of airspace. A weather disturbancein this region can greatly reduce capacity.This leads to a problem of congestion in the NAS.

Slide4The Implementation of AFPs

4

Airspace Flow Programs (AFP’s) are generally implemented over volumes of airspace where a capacity reduction is predicted over an extended period of time. Flights are given the choice of flying around the flow constrained area (FCA) (with or without ground delay), or flying through the FCA with a given amount of ground delay.

Primary Route

Bad Weather in Airspace

Secondary Route

Slide5Weather Uncertainty

5

Given the uncertainty of weather, it is entirely possible that the FCA clears earlier than anticipated. In such a case, flights still grounded have the possibility of reducing their delays by departing sooner. flights that have rerouted around the FCA can possibly have their route length shortened by now flying through the FCA.

Primary Route

Bad Weather in Airspace

Secondary Route

Hybrid

Slide6Approaches to this Problem

6

The problem of how best to allocate flights to FCA arrival slots in under a reduced capacity situation was studied by

Ganji

and later by Glover.

Two different IPs were proposed which seek to minimize the total expected delay of all flights scheduled to fly through the FCA.

Both IPs assume a probability distribution of possible weather scenarios and give outputs which include courses of action based on the weather outcomes.

Unfortunately, neither of these models take CDM into account.

Slide7Collaborative Decision Making

7

Collaborative Decision Making (CDM) originates from an understanding that neither side (the FAA/EUROCONTROL, nor the airlines) has all the information necessary to solve air traffic flow problems.

Most, if not any new major ATFM system proposed either in the US or Europe must adhere to Collaborative Air Traffic management principles.

Slide8GG Model and CDM

8

In seeking to minimize the total expected delay, the GG model assumes that the air navigation service provider (ANSP) can exercise universal control, which is a rare occurrence.

So although the model they propose is powerful from a mathematical point of view, they are unlikely to be accepted as currently proposed.

Slide9Accounting for CDM

9

Here, we analyze two alternative approaches for incorporating these models into a CDM-like setting.

In both cases, the ANSP allocates certain resources to the flight operators, and the flight operators then optimize the use of the resources they are given.

This gives the flight operators the final decision on which flights are rerouted and to adjust the ground delay amounts.

Slide10Overview of IP

10

Here we will describe the GG model.

This is a two-stage stochastic IP.

Stage one represents the initial assignment of delay and reroutes under the assumption that the weather lasts the entire duration.

One constraint set says that

each flight must be given an initial assignment

The other says that

the FCA capacity must be respected.

Slide11Overview of IP

11

Here we will describe the GG model.

This is a two-stage stochastic IP.

Stage

two represents a set of weather clearance times (called scenarios) addressing the possibility that the weather clears early.

In each scenario,

each flight must be reassigned

In each scenario,

the FCA capacity must be respected.

In each scenario,

we cannot violate the laws of geometry or physics.

Slide12Modification to the GG Model

12

The adjustments we made to the GG model were based on bringing in line with CDM principles.

Each flight now has a flight operator.

Instead of assigning FCA arrival times to flights, these assignments are made to flight operators, who then have the ability to distribute the delay as needed.

The basic fairness mechanism is still Ration-by-Schedule, which is accepted.

Slide13The First Mechanism

13

The ANSP executes RBS and allocates to each flight operator a bundle of FCA arrival times.

Each flight operator then makes decisions regarding which flights will stay on their primary routes and experience ground delay and which will be rerouted around the FCA.

This is functionally equivalent to the CDM cancellation and substitution process.

Slide14The Second Mechanism

14

The

ANSP executes RBS and

allocates to each flight operator a bundle of FCA arrival times.

Each flight operator then makes decisions regarding which flights will stay on their primary routes and experience ground delay and which will be rerouted around the FCA.

The ANSP then executes an inter-airline exchange.

This is functionally equivalent to the CDM compression algorithm.

Slide15Flight Operators Using the GG Model

15

The flight operator decision step can take advantage of the GG model.

However, because the flight operator only knows about their flights, this gives a limited viewpoint of the utilization of the NAS.

This implies that the flight operators are using an approximate objective function, where they are estimating future ANSP behavior.

Slide16Experiment

16

We tested the two mechanisms versus the system optimal (provided by the GG model). The input to the model consisted of 400 flights and an FCA whose duration was expected to last 5 hours. We also assumed seven airlines and four aircraft types:

ERJ-170

737-300

757-200

767-400

Carrier 1

64

64

16

16

Carrier 2

30

12

6

12

Carrier 3

0

70

0

0

Carrier 4

20

8

0

12

Carrier 5

0

10

10

0

Carrier 6

20

0

0

0

Carrier 7

12

9

9

0

Slide17Airline Costs

17

Flight operator decisions are based on minimizing expected airline cost of delay.

The aircraft operation cost per minute of airborne delay is $64 and $32 for ground delay.

We used a cost per minute of delay per passenger of $3.00, of which airlines consume 1/6

th

of this cost.

Thus in our model for the cost of ground delay for an airline is 32 + 0.5

Psn

(

f

)

The cost for airborne delay for an airline is

64 + 0.5

Psn

(

f

)

Psn

(

f

) is the number of passengers on flight

f.

Slide18Thank You!

18

Questions?

Slide19ANSP Costs

19

To determine the efficiency of the solutions produced, we executed the GG model on this input, ignoring the carriers.

This gives a system optimal.

We assumed that the ANSP’s objective is to reduce the overall expected cost of delay, using the same figures from the previous slide.

Slide20Modeling Compression

20

Compression can also be modeled by an objective function (after the airlines have individually run the algorithm themselves) via a

superlinear

cost function:

Let md =

Max

f

{time (goal(

f

)) – earliest(

f

)}.

Then the cost of assigning flight

f

to the FCA arrival time

i

is

cst

(

f

,

i

) = (time(

i

) – (time(goal(

f

)) +

md

))

1+e

Where goal(

f

) is the time that the

f

’s carrier wishes to assign to flight

f

and earliest

f

is the earliest time that

f

can arrive at the FCA.

Slide21Total Expected Costs of Delay

21

Slide22Results & Conclusions

22

As expected, neither MECH1 nor MECH2 received the optimal cost, but both provided within 10% of the system optimal.

MECH2 does achieve a small, but noticeable advantage over MECH1.

Although MECH1 provides airlines with more power in controlling their flights, it also requires more information, particularly when these flights will be at the FCA boundary which can be difficult to compute.

MECH2 only requires a priority listing of the flights.

Both mechanisms reduce the role of the ANSP to one of allocating a set of FCA arrival times to airlines and allowing the flight operators to optimize the resources they are given.

Slide23Slide24

Slide25

Slide26

Next Slides