EMME Users’ Group Meeting

Recent toll patronage forecasting using EMME

27 May 2011

General Modelling Methodology

Toll choice model is typically an “add-on” Traffic Assignment : Vehicle demand from other models More detailed route assessment One or more toll facilities Can be project specific Projections for average weekday Demands expanded externally for annual

revenue and patronage Two general approaches:

– Logit demand segmentation– Distributed VOT MC equilibrium assignment

BaseVehicle

DemandModel

* Travel Surveys* Demographics* Transport Networks

Vehicle Tripsby Purpose

and Toll Class

Toll Choice Model* Value of Time(SP/RP Surveys)

Patronage and RevenueModel

Road NetworkPerformance

Route combinations

A

B

T1

T2

T3

T4Tolled Segment

Access Route (untolled)

Egress Route (untolled)

Full Toll Route

Possible Alternative Routes

T3

Logit-based demand segmentation model

HCVLCV

Cars

VehicleDemandby Class

Non-TollRoute

Demand

TollRoute ADemand

TollRoute BDemand

TollRoute CDemand

TollRoute DDemand

TollRoute EDemand

Total Flows

Logit Based Demand Segmentation Processfor Single Time Period

Convergence

No

Results

Market Trip Assignment

Toll Route AAssignment

Toll Route BAssignment

Toll Route CAssignment

Toll Route DAssignment

Toll Route EAssignment

Non-TollAssignment

Route Market Skimming

Non-TollRouteSkims

TollRoute ASkims

TollRoute BSkims

TollRoute C

Skims

TollRoute D

Skims

TollRoute ESkims

Demand Segmention(LOGIT)

Logit based demand segmentationBenefits of logit based demand segmentation assignment: Most common method in Australian context Strong financial market acceptability Can address toll capping or user budget limits

Disbenefits of the logit based demand segmentation assignment: Skimming and assignment to toll routes can be complex and

error prone Limitations on toll route and vehicle class markets

combinations Often used project specific application Difficult to adopt for general city wide use or use for many toll

facilities

Distributed VOT MC Assignment

HCV

LCV

Cars (3 Trip Purposes)

VOTGroup 5

VOTGroup 4

VehicleDemand

VOTGroup n

VOTGroup 1

VOTGroup 2

VOTGroup 3

Value ofTime

Segmention

Multi-class EquilbriumAssignment

VTTS Parameters by Class

Total Flows

Distributed VOT Multi-class Equilibrium Assignmentfor Single Time Period

Convergence No

Results

RedoAssignments

0

2

4

6

8

10

12

14

16

1 2 3 4 5 6

Prop

orti

on o

f D

Eman

d M

Ark

et

Increasing Value ot Time (Higher willingness to pay toll)

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0 12 24 36 48 60 72 84

Prob

abilit

y Den

sity

Value of Time ($/hr)

BusinessCommuteOtherLCVHCV

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0 12 24 36 48 60 72 84

Prob

abilit

y Den

sity

Value of Time ($/hr)

BusinessCommuteOtherLCVHCV

Distributed VoT multi-class assignment

Benefits of Distributed VOT multi‐class equilibrium assignment approach: All possible toll route combinations are assessed at similar level of detail. Use of transport software built‐in equilibrium assignment algorithms Potential for reduced model run times and more stable outputs Can address many toll roads together Model could be applicable for general planning use Less potential for user specification error

Disbenefits of Distributed VOT multi‐class equilibrium assignment approach: Less commonly used and possibly not as well accepted Cannot handle toll capping easily Number of classes may limit VOT segmentation Requires more innovative SP/RP survey analysis

Conclusions

Toll roads are an increasing feature of large Australian city road networks

Methods for patronage forecasting (i.e. for bid teams) have been complex and unwieldy

Distributed VOT MC assignment techniques may be adaptable for general planning agency use