Date post: | 15-Dec-2015 |
Category: |
Documents |
Upload: | chris-tubman |
View: | 214 times |
Download: | 0 times |
Using value-of-time distributions in auto assignment to forecast effects of road pricing schemes with non-additive chargesLeonid Engelson & Dirk van Amelsfort
22nd Emme Users’ Conference, Portland 2011
Background
Road charges has been implemented in a dozen of cities
More cities consider
Typically designed as charge zone or charge cordon (set of links)
London, Milano, Valetta: pay as you drive or park in the zone
Singapore, Stockholm: pay each time you pass the cordon
Gothenburg (decided): pay as you pass any link of the set, once within 60 minutes (Multiple Passage Rule, MPR)
WSP Sweden was responsible for modelling in the design project of CC system for Gothenburg
How to model CC system with MPR?
22nd Emme Users’ Conference, Portland 2011
Background
22nd Emme Users’ Conference, Portland 2011
OSLO
MALMÖ
STOCKHOLM
Sweden’s second largest city500 000 inhabitantsFreight hubLand use low densityRiver with only 3 – 4 crossingsRailway station dead-end
Gothenburg suffers (some) congestion in morning peak
22nd Emme Users’ Conference, Portland 2011
…and substantial air quality problems
22nd Emme Users’ Conference, Portland 2011
Nitrogen dioxide 98-percentile in 2008
(Red = above the Environmental Quality Standards)
Politicians were inspired by success of the Congestion charging in Stockholm
Jan-July 2006 TrialSep 2006 Referendum showed: Most inhabitants of the Stockholm
commune are positive to the chargesAugust 2007 Permanent introduction
Cordon basedTime differentiated1,5-3 $ per passage
Delimited by water18 entry points sufficient
Bypass exempted from charging
22nd Emme Users’ Conference, Portland 2011
Objectives…
100 million $ net revenue in 2015.
Reduction in congestion and air polluting emissions from traffic
Reduction of traffic through central Gothenburg
and constraints…Reasonable cost increase for travellers.
Fit under the existing legislation
Logical and easy to understand for the users.
ANPR technology from Stockholm
Tax to be paid 06.00 am – 18.30 pm.
Same fare for all non-exempted vehicles
22nd Emme Users’ Conference, Portland 2011
First stage of the design :5 cordon types examined, 5 types of effects predictedSmall cordon chosen
Revenue
EnvironmentCongestion
Route choicesModal split
22nd Emme Users’ Conference, Portland 2011
Second stage of the design:
Blue and Green – two small cordons evaluated in detail
Multiple passages => one charge
22nd Emme Users’ Conference, Portland 2011
Modelling tool:
• SAMPERS – Swedish travel demand model
• 5 regions + long trips
• Four stages
• Multiclass auto assignment with generalized costs (Emme)
• Complication 1: Non-additive route cost: paying just once
• Complication 2: Discrete values of time• Too coarse representation of preferences
• Optimal discretization different for different OD-pairs
A B C
1 $ 1 $
22nd Emme Users’ Conference, Portland 2011
Solution: Hierarchic route choice with VoT distribution
2 classes auto assignment: to H and to DModule 5.25, save paths
TimeH and TimeDDistH and DistD
Paying trip matrixNon-paying trip matrix
Define modes:H = whole road networkD = links without charges
Weigh together for CBA0
1
Value of time
F
Repeat until convergence (MSA)
22nd Emme Users’ Conference, Portland 2011
Convergence with MSA applied to the demand
0.01
0.1
1
10
100
0 2 4 6 8 10 12
Iteration
Max
ab
s ch
ang
e n
um
ber
of
pay
ers
22nd Emme Users’ Conference, Portland 2011
From new Swedish VoT study: Lognormal distributions (kr/h) ($1=6.3kr)
0
0.005
0.01
0.015
0.02
0.025
0.03
0 100 200 300 400 500 600 700
Private >100 km
Private <100 km
Commuting
Business and freight low
Business and freight high
Mean log STD log Median VoT Mean VoT Estimated by-0.20 1.29 49 113 SP
-0.89 1.17 25 49 SP-0.16 1.22 51 108 SP1.16 1.22 191 403 Judgement1.64 0.36 309 330 Judgement
22nd Emme Users’ Conference, Portland 2011
Blue and Green – two small zones evaluated in detail
Multiple passages => one charge
Generally similar effectsBlue chosen
+ Stronger congestion releif+ Stronger effect on the air
quality
22nd Emme Users’ Conference, Portland 2011
Flow changes with charges, the Blue scenario
22nd Emme Users’ Conference, Portland 2011
Final scheme
Yellow sistemClosed ring+6 links
Improved Urban function and understandability compared to the blue system
(using the same method)
22nd Emme Users’ Conference, Portland 2011
Advantages of the method
• Can be applied to some road user charge systems with discounts
• Allows arbitrary distributions of VoT
• Good convergence
Limitations
• If the user pays each time, needs 2N user classes, N=number of elements of the charging system (toll rings + separate toll links)
• For example, not possible to study effect of the multiple passage rule with continuous VoT distribution for the Yellow system
22nd Emme Users’ Conference, Portland 2011
Unresolved issues
VoT for business trips and freight
Distance weight seems too high in the base scenario
• Distance weight = (gasoline and amortisation cost per km)/(median VoT)
• Too low VoT?
• The drivers don’t consider the whole distance related cost?
• In the last case (reduced distance cost), too strong (compared to the multiclass method) response to the charges
• Distangle distance cost from VoT?
Needed to further calibrate VoT and cost per km
22nd Emme Users’ Conference, Portland 2011
Conclusions
Assignment method with continuous VoT distribution worked well in the application to Gothenburg
Manages cordon charge that is independent of the number of crossings
Warning: assignment with continuous VoT may give very different results compared to the assignment with discrete VoT.
22nd Emme Users’ Conference, Portland 2011
Thank you for your attention
Questions?
22nd Emme Users’ Conference, Portland 2011
Welcome to the Nordic Emme Users’ Conference 19-20 March 2012 in
Stockholm!
22nd Emme Users’ Conference, Portland 2011