Airport Congestion: Demand Management Strategies and Equity Issues - - PowerPoint PPT Presentation

Bressanone, July 2007 G.Andreatta (PADOVA University) 1

Airport Congestion: Demand Management Strategies and Equity Issues

Giovanni ANDREATTA (Padova University) Guglielmo LULLI (Milan “Bicocca” University)

Bressanone, July 2007 G.Andreatta (PADOVA University) 2

Background

• My Runggaldier number is =
• We have known each other for the past …
• We were together in Rome when my wallet
• Hiking and Skiing (Val Mesdì,Lagorai,

Fanes)

• Moving to Padova
• Gordon Kaufman and MIT

Bressanone, July 2007 G.Andreatta (PADOVA University) 3

Main findings: A theorem about existence and unicity

Dear Wolfgang, … grazie di esistere ! (thanks for existing)

Bressanone, July 2007 G.Andreatta (PADOVA University) 4

Air Traffic Demand vs AT Capacity

• Fast and steady increase of demand

(with exceptions: 11 September 2001, SARS, etc. ...)

• Modest increase of capacity

Need to address demand

Bressanone, July 2007 G.Andreatta (PADOVA University) 5

Bressanone, July 2007 G.Andreatta (PADOVA University) 6

How to cope with demand Peaks

• Administrative measures

(SLOT Assignment, etc.)

• Market Based Demand Management

strategies (Congestion fee, Auctions, etc.)

Bressanone, July 2007 G.Andreatta (PADOVA University) 7

This presentation will touch upon …

• Analysis of a combination of a queuing

model and a model of the elasticity of airport demand to airport charges.

• Equity issues
• Congestion pricing vs Auction mechanism

Bressanone, July 2007 G.Andreatta (PADOVA University) 8

Demand Management Strategies should

• Limit demand for access to busy airfields

and/or congested airspace during peak periods of time

• Modify temporal and/or spatial distribution
• f demand (not presented today)

Bressanone, July 2007 G.Andreatta (PADOVA University) 9

Objective of Demand Management Policies

… Enhance the efficiency of aircraft

• perations in European airspace and

airports, while interfering to the least possible extent with the overarching aim of creating a competitive, market-driven air transport environment in Europe …

Bressanone, July 2007 G.Andreatta (PADOVA University) 10

What has already been addressed

• Peak period pricing in general (widely

investigated)

• Applications to congestion-pricing of

transportation facilities (more recent)

• Applications to air transportation (fewer)

– Concentrated on airport congestion – Very limited work (unpublished) on airspace side

Bressanone, July 2007 G.Andreatta (PADOVA University) 11

Market based Demand Management (MbDM) strategies

• Congestion fee approach: the AA decides

access fees, the “market” will respond with an actual demand “lower” than the original

• ne
• Auction approach: the AA decides the level
• f acceptable demand (how many

movements), the “market” will assess the appropriate access price

Bressanone, July 2007 G.Andreatta (PADOVA University) 12

Congestion fee vs. Auction approaches

Assuming that the demand function is known:

• A congestion fee decreases the original

demand: the AA decides the fee and can “compute” the actual demand

• With an Auction: the AA decides the

“actual demand” and can compute a “threeshold” τ: the bids above τ are accepted

Bressanone, July 2007 G.Andreatta (PADOVA University) 13

Equity issues

In the presence of different categories should the imposed prices be the same for all categories?

Bressanone, July 2007 G.Andreatta (PADOVA University) 14

The cost of Congestion

Congestion costs due to any specific user have 2 components:

– Cost of delay to that user (internal) – Cost of delay to all other users caused by that user (external)

This second component can be very large!

Bressanone, July 2007 G.Andreatta (PADOVA University) 15

Optimal congestion fee

A congestion fee on a user is optimal when it is equal to the external costs that the user imposes

• n the other users.

In practice it is usually very hard to:

– Estimate external marginal delay costs

Queuing Theory has much to offer

Bressanone, July 2007 G.Andreatta (PADOVA University) 16

Objective of the Airport example

• To provide through the analysis of an

example a demonstration of the overall approach

• To offer through the example a proof of

concept, i.e., an indication of the types of benefits (and costs) that can be obtained

Bressanone, July 2007 G.Andreatta (PADOVA University) 17

Airport illustrative example

Parameter Type 1 (HEAVY) Type 2 (MEDIUM) Type 3 (LIGHT) Service rate (movements per hour) 80 90 100 Standard deviation of service time (seconds) 10 seconds 10 seconds 10 seconds Cost of delay time ($ per hour) $5,000 $3,000 $800

Bressanone, July 2007 G.Andreatta (PADOVA University) 18

Demand functions (hypothetical)

where:

x = DC + CF DC = cost of the delay time experienced by the aircraft CF = congestion fee paid by the aircraft to land or to take off from Airport

2 1

5 . 2 5 . 40 ) ( x x x ⋅ − ⋅ − = λ

2 2

. 5 5 . 1 50 ) ( x x x ⋅ − ⋅ − = λ

2 3

. 20 . 5 60 ) ( x x x ⋅ − ⋅ − = λ

40 50 60 0,001 0,003 0,01 0,00001 0,00002 0,00008 x lambda 1 lambda 2 lambda 3 40 50 60 100 39,8 49,5 58,2 200 39,4 48,6 54,8 300 38,8 47,3 49,8 400 38 45,6 43,2 500 37 43,5 35 600 35,8 41 25,2 700 34,4 38,1 13,8 800 32,8 34,8 0,8 900 31 31,1

• 13,8

1000 29 27

• 30

1100 26,8 22,5

• 47,8

1200 24,4 17,6

• 67,2

1300 21,8 12,3

• 88,2

1400 19 6,6

• 110,8

1500 16 0,5

• 135

1600 12,8

• 6
• 160,8

1700 9,4

• 12,9
• 188,2

1800 5,8

• 20,2
• 217,2

1900 2

• 27,9
• 247,8

2000

• 36
• 280

Demand Functions for three types of users

10 20 30 40 50 60 70 2 4 6 8 1 1 2 1 4 1 6 1 8 2 Total cost ($) Arriv al rate (Users/unit tim e) Type 1 Type 2 Type 3

Bressanone, July 2007 G.Andreatta (PADOVA University) 20

Case 1: No congestion fee

Parameter Type 1 Type 2 Type 3 Delay cost (DC) per aircraft $3,224 $1,935 $515 Congestion fee (CF) $0 $0 $0 Total cost of access (DC + CF) $3,224 $1,935 $515 Demand (no. of movements per hour) 12.4 28.4 52.1 Total demand (no. of movements per hour) 92.9 Expected delay per aircraft 38 minutes 42 seconds Utilization of the airport (% of time busy) 99.01%

Bressanone, July 2007 G.Andreatta (PADOVA University) 21

Congestion Pricing and Queuing Theory

Consider a queuing facility with a single type of customer in steady-state. Let: c = delay cost per unit of time/customer C = total cost of delay per unit time Wq = Expected queuing time per customer λ = demand rate Then:

q q

W c cL C λ = =

Bressanone, July 2007 G.Andreatta (PADOVA University) 22

From

it follows

where: c = (delay) cost per unit time per customer Wq = Expected queuing time per customer λ = demand rate

λ λ λ d dW c W c d dC MC

q q +

= =

Marginal Cost

q q

W c cL C λ = =

Bressanone, July 2007 G.Andreatta (PADOVA University) 23

Optimal congestion fee

A congestion fee on a user is optimal when it is equal to the external costs that the user imposes

• n the other users.

For an M/G/1 queue: Marginal Internal External cost cost cost λ λ λ d dW c W c d dC MC

q q +

= =

= +

Bressanone, July 2007 G.Andreatta (PADOVA University) 24

Optimal congestion fee

• The external cost can be explicitly

computed (under suitable assumptions) even in the presence of several categories of users

• In the presence of several categories the

external costs might be different (so are the congestion fees!)

Case 2: Optimal* congestion fee

Optimal Congestion Fee

Delay cost (DC) per aircraft $250 $150 $40 Congestion fee (CF) $1,712 $1,504 $1,341 Total cost of access (DC + CF) $1,962 $1,654 $1,381 Demand (no. of movements per hour) 29.4 33.8 14.9 Total demand (no. of movements per hour) 78.1 Expected delay per aircraft 3 minutes 0 seconds Utilization of the airport (% of time busy) 89.2%

Bressanone, July 2007 G.Andreatta (PADOVA University) 26

Congestion fee approach vs. “do nothing” approach

By charging a congestion fee equal to the external delay costs, we have:

• Greatly reduced the average delay per aircraft (3’0’’ vs.

38’42’’)

• Greatly reduced the delay costs per aircraft ($250 from

$3,224, $150 from $1,935, $40 from $515)

• Reduced the total cost of access for most aircraft ($1,962

from $3,224, $1,654 from $1,935, $1,381 from $515)

• Assuming Type I, (resp. II, III) carry 200 (resp. 100, 20)

pax on average per a/c: the total no. of pax will be increased by 50% (9,558 from 6,362)

Bressanone, July 2007 G.Andreatta (PADOVA University) 27

Demand Functions for three types of users

10 20 30 40 50 60 70 2 4 6 8 1 1 2 1 4 1 6 1 8 2 Total cost ($) Arrival rate (Users/unit time) Type 1 Type 2 Type 3

+ + +

• No Fee

+ With Fee

Bressanone, July 2007 G.Andreatta (PADOVA University) 28

Equity Issues

• By using a pure MbDM approach, some

categories might experience an untolerable burden (e.g., Type 3 aircraft). In the worse situation a category could have almost no access to the airport

• There is a need to mitigate this risk
• Who should evaluate equity?

Bressanone, July 2007 G.Andreatta (PADOVA University) 29

Proposal

Two Stage Approach:

• 1. (Administrative) The AA decides:
• the desired share of movements
• the maximum expected delay value
• 2. (MbDM) Appropriate congestion fees can

be computed so as to match previous target

Bressanone, July 2007 G.Andreatta (PADOVA University) 30

Example (cont.)

Let us suppose that the AA wants:

• 1. to obtain the same mix as the unrestricted

mix, i.e. (26.7%; 33.3%; 40.0%)

• 2. A maximum expected delay per

movement equal to 3 minutes

Case 3: Two-stage approach

Two-stage approach

Delay cost (DC) per aircraft $250 $150 $40 Congestion fee (CF) $2,000 $1,847 $1,017 Total cost of access (DC + CF) $2,250 $1,997 $1,057 Demand (no. of movements per hour) 21.6 27.0 32.4 Total demand (no. of movements per hour) 81.0 Expected delay per aircraft 3 minutes 0 seconds Utilization of the airport (% of time busy) 89.4%

Bressanone, July 2007 G.Andreatta (PADOVA University) 32

Comments

+ More equitable (according to the AA) − Economic principle is lost

Bressanone, July 2007 G.Andreatta (PADOVA University) 33

Work still in progress …

Special thanks to Professor Amedeo ODONI (MIT)