Model Applications in Australia

Sylvia Tesche
31 October 2011
Introduction
Modelling in Australia
Examples
Greater Hobart Urban Travel Demand Model
Dynon Port Precinct Avenue Model
GHUTDM Page 2 November 18, 2011
Transport Models in Australia
The Dynon Port Precinct Model Page 3 November 18, 2011
Source: Google Earth
Greater Hobart Urban Travel Demand
Model
31 October 2011
11/18/2011 Presentation Title Page 5
11/18/2011 Presentation Title Page 6
11/18/2011 Presentation Title Page 7
0
50000
100000
150000
200000
250000
1803 1825 1850 1875 1900 1950 1975 2000
Relatively low population growth in late 20
th
C
Aging population structure
Background and scope
In April 2011, DIER commissioned AECOM to develop a four-step travel
demand model for Greater Hobart
The model would have capabilities to forecast traffic volumes & PT
patronage under a range of scenarios:
- Transport infrastructure (new roads, increased road capacity, light rail)
- Land use (e.g. new residential development)
- Demographics (changes in population & employment levels)
- Economic conditions (increased vehicle operating costs)
- PT service characteristics (improved frequencies, fare changes, etc>)
The project was scheduled to complete in October 2011
Presentation Title Page 8 November 18, 2011
Hobart Model
Structure
Presentation Title Page 9 November 18, 2011
MODELOUTPUTS
Generalisedcostskims
Dailypersontripends
DailyPAtripsbypurpose
AM,PM,OPODmatrices
bymode
Traffic/passengerflows
Networkperformance
Generalisedcostskims
DailyPAtripsbypurpose
andmode
Report
DATAINPUTS MODELPROCESS
Initialization
Hwy&PTassignment
Initialnetwork
ExternalModel
CommercialVehicleModel
Hwy&PTassignmentby
timeperiodassignment
Demographicdata
HHsegmentation
Carownershipmodel
Tripgeneration
TripProduction/
Tripends
Generalisedcosts
Gravitymodel
Parametersby
DailyPAtrips
Modesplitfactors
bypurpose&sector
Tripdistribution
Modechoice
Timeperiodmodel/CBD
parkingdelaymodel
Blendedcostskims
Timeperiodfactors
bypurpose&sector
Vehicleoccupancy
factorsbypurpose
Network
development
(zones,roads,etc)
Speed/flowcurves
(linkcosts,tolls)
Convergence
Yes
No

A
n
a
l
y
s
e

H
o
u
s
e
h
o
l
d

S
u
r
v
e
y

D
a
t
a

Model Inputs Zone System
November 18, 2011 Presentation Title Page 10
The model has
500 internal zones
9 external zones
Model Inputs Road Network
November 18, 2011 Presentation Title Page 11
Model Inputs Public Transport Services
November 18, 2011 Presentation Title Page 12
312 bus routes
coded into the
model.
44 were express
services
Model inputs Demographic data
Number of households
Population
Number of workers
Dependants
- Dependants (0-11)
- Dependants (12-17)
- Dependants (18-65)
- Retirees (65+)
Enrolments categorised by:
- Kindergarten/primary school enrolments
- Secondary school enrolments
- Tertiary institution enrolments which include Universities and TAFE.
Employment categorised by 5 types:
- Education
- Entertainment
- Industry
- Retail
- Service
November 18, 2011 Presentation Title Page 13
Model inputs Demographic data
November 18, 2011 Presentation Title Page 14
Area
Population Employment
Total Education
Enrolments
2009 2031 2009 2031 2009 2031
Hobart LGA 48,906 58,058 49,059 58,240 24,684 29,303
Kingborough
Part A SLA 29,825 34,825 6,691 7,813 4,927 5,753
Glenorchy LGA 43,869 51,375 19,137 22,744 6,433 7,645
Brighton LGA 15,139 18,753 1,927 2,291 2,043 2,428
Clarence LGA 50,029 58,750 12,406 14,465 10,317 12,029
Sorell Part A
SLA 11,380 12,849 1,998 2,330 1,165 1,358
Total 199,149 234,609 91,220 107,881 49,569 58,516
Model inputs Greater Hobart Household Travel Survey
November 18, 2011 Presentation Title Page 15
The GHHTS provides the most important source of travel behaviour data for the
development of model parameters.
The survey undertaken between 2008-2009:
- Covered the Greater Hobart metropolitan area encompassing Derwent Valley
Part A, Brighton ,Glenorchy, Clarence ,Hobart, Kingborough Part A, and
Sorell Part A.
- Interviewed over 2,400 households. All persons aged 5+ required to complete
a travel diary for two consecutive days. The sample data include:
Household table: 2,433 records on household characteristics such as address, household
size, number of vehicles and area.
Person table :5,385 records on the members of household such as age group, sex, and
employment status.
Trip table: 27,131 records on origin, destination, mode, time of departure, time of arrival,
purpose
Model Validation Comparison of Trip Generation
November 18, 2011 Presentation Title Page 16
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
Modelled
HTS
Comparison of Trip Production (Daily Person Trips)
Model Validation Comparison of Trip Distribution
November 18, 2011 Presentation Title Page 17
0
2
4
6
8
10
12
14
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
A
v
e
r
a
g
e

T
r
i
p

L
e
n
g
t
h

(
k
m
)

Travel Market Segments
Observed
Modelled
Comparison between modelled and observed average trip length
Model Validation Comparison of Mode Choice
November 18, 2011 GHUTDM Page 18
Comparison between modelled and observed mode share by purpose
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
HBW HBEP HBES HBET HBS HBR HBV HBO NHBO NHBW Total
Car Modelled
Car Observed
PT Modelled
PT Observed
Cycle Modelled
Cycle Observed
Walk Modelled
Walk Observed
Model Validation Assignment
Comparison of Highway Volumes
November 18, 2011 Presentation Title Page 19
-100
-80
-60
-40
-20
0
20
40
60
80
100
0 10 20 30 40 50 60
%

D
i
f
f
e
r
e
n
c
e
Screenline Traffic Volume ('000s)
Inbound Direction
Outbound Direction
Percentage of difference between
modelled and observed daily
traffic volumes (cars only)
Model Validation Assignment
Comparison of Highway Volumes
November 18, 2011 Presentation Title Page 20
y = 0.93x
R = 0.96
0
5,000
10,000
15,000
20,000
25,000
30,000
35,000
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000
O
n
e
-
w
a
y

t
r
a
f
f
i
c

v
o
l
u
m
e

(
m
o
d
e
l
l
e
d
)
One-way traffic volume (observed)
Observed and modelled daily traffic volumes (cars only) for all sites
Model Validation Assignment
Comparison of PT Boardings
November 18, 2011 Presentation Title Page 21
Observed and modelled daily bus boardings by corridor
y = 1.08x
R = 0.98
-
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
4,500
5,000
- 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 4,500 5,000
O
n
e
-
w
a
y

b
o
a
r
d
i
n
g
s

b
y

c
o
r
r
i
d
o
r

(
m
o
d
e
l
l
e
d
)
One-way boardings by corridor (observed)
Model Sensitivity Future base year model
November 18, 2011 Presentation Title Page 22
Network performance indicators for 2009 Base and 2031 Base
Network Performance Indicator 2009 Base 2031 Base % Change 2009-31
Daily trips
Car 405,464 474,526 +17.0%
Public transport 20,992 26,809 +27.7%
Bicycle 4,312 5,112 +18.5%
Walk 88,365 103,764 +17.4%
Total 519,134 610,210 +17.5%
Daily mode share
Car 78.1% 77.8% -0.3%
Public transport 4.0% 4.4% +0.3%
Bicycle 0.8% 0.8% 0.0%
Walk 17.0% 17.0% 0.0%
Total 100.0% 100.0% 0.0%
Daily car person km and hours
Total car person km 4,060,304 4,698,802 +15.7%
Total car person hours 88,007 110,252 +25.3%
Average trip distance (km) 10.0 9.9 -1.1%
Average travel time (min) 13.0 13.9 7.1%
Daily public transport boardings
Total 28,673 35,529 +23.9%
Average boardings 1.37 1.33 -3.0%
Daily public transport km
Total 148,992 189,446 +27.2%
Average trip distance (km) 7.10 7.07 -0.4%
Daily public transport hours
Total 361,960 478,390 +32.2%
Average travel time (min) 17.24 17.84 +3.5%
Model Output Comparison of Screen-line volume
November 18, 2011 Presentation Title Page 23
Screen-
line
Base Base Difference
2009 2031 Abs %
1 90,568 103,228 12,660 14.0%
2 51,471 56,138 4,668 9.1%
3 20,619 23,132 2,513 12.2%
4 43,728 50,163 6,435 14.7%
Model Output Comparison of v/c ratios
November 18, 2011 Presentation Title Page 24
2009 2031
Model Sensitivity Tests
November 18, 2011 Presentation Title Page 25
Test 1: Reduction in the supply of parking in the Hobart CBD
Test 2: Increase in the capacity of key roads
Test 3: No Greenfield development (all infill)
Test 4: Introduction of high frequency passenger transport network
Test 5: Higher growth in population, employment and educational
enrolments
Test 6: Increase in the cost of car transport
Test 7: Proposed LRT
Model Sensitivity Test 1
November 18, 2011 Presentation Title Page 26
Reduction in the supply of parking by 50% in the Hobart CBD
To/From
Number of Car Trips
2031 Base
Sensitivity
Test 1
% Change
CBD
64,377 58,496 -9.1%
CBD Fringe
50,828 47,550 -6.4%
Total
115,206 106,047 -8.0%
Network Performance Indicator 2031 Base Sensitivity Test 1 % Change
Daily trips
Car 474,526 467,126 -1.6%
Public transport 26,809 27,516 +2.6%
Bicycle 5,112 5,204 +1.8%
Walk 103,764 110,373 +6.4%
Total 610,210 610,219 0.0%
Daily mode share
Car 77.8% 76.6% -1.2%
Public transport 4.4% 4.5% +0.1%
Bicycle 0.8% 0.9% 0.0%
Walk 17.0% 18.1% +1.1%
Total 100.0% 100.0% 0.0%
Sensitivity Test 4 High Frequency of Bus Network
November 18, 2011 Presentation Title Page 27
Characteristic Platinum Gold Silver
Routes
Claremont to Sandy
Bay, Hobart to
Shoreline and
Kingston Central
Northern Orbiter,
Western Connector,
The Derwent, Eastern
Connector, The Ferry,
Uni Bus
N Town, Lutana
Goodwood Glen,
South Hobart, Mt
Nelson, Southern
Connector
AM/PM peak
frequency (7am-9am,
4pm-6pm)
5 minutes 10 minutes 15 minutes
Inter peak frequency
(9am-4pm)
10 minutes 15 minutes 30 minutes
Network Performance
Indicator
2031 Base
Sensitivity Test
4
% Change
Daily trips
Car 474,526 461,343 -2.8%
Public transport 26,809 41,724 +55.6%
Bicycle 5,112 4,939 -3.4%
Walk 103,764 101,766 -1.9%
Total 610,210 609,771 -0.1%
Daily mode share
Car 77.8% 75.7% -2.1%
Public transport 4.4% 6.8% +2.4%
Bicycle 0.8% 0.8% 0.0%
Walk 17.0% 16.7% -0.3%
Total 100.0% 100.0% 0.0%
Network performance indicators for 2031 Base and Sensitivity Test 4
Sensitivity Test 4 High Frequency of Bus Network
November 18, 2011 Presentation Title Page 28
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
16,000
2031 Base
Sensitivity Test 4
Change of Daily PT boardings
Sensitivity Test 7 LRT
November 18, 2011 Presentation Title Page 29
LRT from Claremont to Sandy Bay via Mawsons Place
Network Performance Indicator Y2031Base
Y2031Base_Scen
7
Daily trips
Car 474,526 468,473
Public Transport 26,809 33,649
Bicycle 5,112 5,010
Walk 103,764 102,609
Total 610,210 609,741
Daily mode share
Car 77.8% 76.8%
Public Transport 4.4% 5.5%
Bicycle 0.8% 0.8%
Walk 17.0% 16.8%
Total 100.0% 100.0%
Public transport boardings DAILY DAILY
Bus 35,497 30,933
LRT - 11,500
Total 35,497 42,433
Average PT boardings 1.33 1.26
Sensitivity Test 7 LRT
November 18, 2011 Presentation Title Page 30
Boarding,
Alighting and
Loading
Profile
Sensitivity Test 7 LRT
November 18, 2011 Presentation Title Page 31
Boarding, Alighting and Loading Profile
Most strategic transport models use a four-step
process.
Presentation Title Page 32 November 18, 2011
Generation
Distribution
Mode choice
Assignment
Land use
Travel costs
Networks
Iterate until
convergence
Volumes
Travel costs
The original MITM used a standard four-step process,
with mode choice and assignment applied separately for
each time period
Presentation Title Page 33 November 18, 2011
Generation
Weekday
Distribution
Weekday
Assignment
AM IP
External trips
Freight
Special generators
Park-and-ride
Mode choice
AM IP
Why update the model?
- Calibrate with recent travel data (VISTA)
- Four time periods to represent a full days travel
- Simpler trip purposes (14 to 9)
- Improved sensitivity of mode and destination decisions
- Rail journeys from regional Victoria
- Improved car ownership and park-and-ride models
- Redesigned software implementation
Presentation Title Page 34 November 18, 2011
Errors in Models
0
2 0
4 0
6 0
8 0
1 0 0
1 2 0
Low Complexity High Complexity
Level of Complexity
M
o
d
e
l

E
r
r
o
r
Specification Error
0
2 0
4 0
6 0
8 0
1 0 0
1 2 0
Low Complexity High Complexity
Level of Complexity
M
o
d
e
l

E
r
r
o
r
Specification Error Measurement Error
0
2 0
4 0
6 0
8 0
1 0 0
1 2 0
Low Complexity High Complexity
Level of Complexity
M
o
d
e
l

E
r
r
o
r
Specification Error Measurement Error
Total Error
0
2 0
4 0
6 0
8 0
1 0 0
1 2 0
Low Complexity High Complexity
Level of Complexity
M
o
d
e
l

E
r
r
o
r
Specification Error Measurement Error
Total Error
0
2 0
4 0
6 0
8 0
1 0 0
1 2 0
Low Complexity High Complexity
Level of Complexity
M
o
d
e
l

E
r
r
o
r
Specification Error Measurement Error
Total Error
What are the main changes?
- Trip productions are based on how many trips are made
by each person, rather than by each household
- Old model
Household size
cars owned
workers and dependants
- New model
Age,
employment status
car availability
household structure
geographic region
Presentation Title Page 36 November 18, 2011
What are the main changes?
Presentation Title Page 37 November 18, 2011
What are the main changes?
Presentation Title Page 38 November 18, 2011
Trip purposes have been simplified in the new model
- Old
Home-Based Work (Blue Collar)
Home-Based Work (White Collar)
Home-Based Education (Primary)
Home-Based Education (Secondary)
Home-Based Education (Tertiary)
Home-Based Shopping
Home-Based Recreation
Work-Based Work
Work-Based Shopping
Work-Based Other
Shopping-Based Shopping
Shopping-Based Other
Other Non Home-Based
Presentation Title Page 39 November 18, 2011
- New
Home-Based Work
Home-Based Education (Primary)
Home-Based Education (Secondary)
Home-Based Education (Tertiary)
Home-Based Shopping
Home-Based Social
Home-Based Other
Employers Business
Non Home-Based Other
Distribution and mode choice from MITM are combined
in the new model (VITM).
Presentation Title Page 40 November 18, 2011
Generation
Weekday
Distribution
Weekday
Assignment
AM IP
Mode choice
AM IP
Generation
Weekday
Distribution and mode
choice
Weekday
Assignment
AM OP PM IP
Old New
A hierarchical discrete choice model for both
destination and mode choices.
November 18, 2011 Presentation Title Page 41
Trips by origin
Split by destination
Split by mode
Trips by origin
Split by mode
Split by destination
Trips by origin
Split by destination and mode
Model Software
Cube Voyager
Example Highway assignment programs
Inputs - Network
Inputs - Network
VITM Highway Network Statistics
- 2,912 zones
- Over 26,000 nodes
- Over 50,000 links
- base year 2008, and future years 2011, 2016, 2021,
2026, 2031, 2036, 2041 and 2046
- Four time periods: AM, IP, PM & OP
Up to 100 new road projects must be implemented at
each incremental modelled year
NETWORK program in Cube
Source: Mansel Rogerson, VicRoads
Network project list
Source: Mansel Rogerson, VicRoads
Inputs PT Services
Inputs PT Services
Old format
New format
Source: Craig Somerville, DOT
Inputs Demographic Data
Demographic Description
POP Population
HH Households
AGE0_4 Population aged 0-4 years
AGE5_11 Population aged 5-11 years
AGE12_17 Population aged 12-17 years
AGE18_25 Population aged 18-25 years
AGE26_64 Population aged 26-64 years
AGE65PLUS Population aged 65 years and over
DEP0_4 Dependents (non-workers) aged 0-4 years
DEP5_11 Dependents (non-workers) aged 5-11 years
DEP12_17 Dependents (non-workers) aged 12-17 years
DEP18_25 Dependents (non-workers) aged 18-25 years
DEP26_64 Dependents (non-workers) aged 26-64 years
DEP65PLUS Dependents (non-workers) aged 65 years and over
EMP_RETAIL Retail jobs
EMP_TOTAL Total jobs
ENROL_PR Primary school enrolments
ENROL_SEC Secondary school enrolments
ENROL_TER Tertiary school enrolments
Inputs Demographic Data
Key processes Utility & Choice theory
U = V + e
Parameters updated in VITM
Choice hierarchy
November 18, 2011 Presentation Title Page 53
Choice
Choice Choice
Choice
Choice
0% -1000
l
1
l
2
0% -1000
l
2
> l
1
0.00
20.00
40.00
60.00
80.00
100.00
120.00
0 0.2 0.4 0.6 0.8 1 1.2 1.4
S
p
e
e
d

(
k
m
/
h
r
)
Volume Capacity Ratio
Freeway Curve Highway Curve
Key processes Speed flow curves
Key processes Highway assignment
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
0 20 40 60 80 100 120 140 160 180 200
R
e
l
a
t
i
v
e

G
a
p
Highway Iteration
Frank Wolf e Bi Conjugate
A Relative Gap of 0.0001 (1e-4) is
required to assure that the assignment
is sufficiently converged to achieve
stable link flows. (Boyce, et al., 2004)
Key processes PT capacity constraint
Source: Liem Huynh, AECOM
0.8
1.0
1.2
1.4
1.6
1.8
2.0
0% 20% 40% 60% 80% 100%
C
r
o
w
d
i
n
g

F
a
c
t
o
r
Utilisation
Train Tram_type1 Bus Tram_type2
Tram_type3 Tram_type4 Tram_type5
Model outputs Public transport loads
Model outputs Public transport validation example
Model outputs Car ownership
Model outputs Public transport utility
Four step model
Trip generation
Presentation Title Page 62 November 18, 2011
- Transport is a derived demand
- Different land-uses will have different trip generation rates
- Influences on trip generation include:
Socio-economic (higher income households produce more trips)
Car ownership (households with more than one car produce more
trips)
Land-use factors
- In Melbourne, demographic inputs supplied by DPCD
(Victoria in Future)
Updated forecasts due November 2011
Gravity models have traditionally been used for
distribution. Logit models are usually used for mode
choice.
Presentation Title Page 63 November 18, 2011
Origin
Trip distribution
Presentation Title Page 64 November 18, 2011
- Forecasting the demand for travel
between each origin and
destination based upon the trip
generations and an impedance
function
- Gravity model generally used
- Impedance function generally
some function of generalised cost
(time, distance) between i and j
- Trip distribution can change with
new infrastructure
Image here created as a fill
Original must be proportionate
to 8.5cm x 14.2cm
Tij = Pi Aj Fij
i
j
Mode choice
Presentation Title Page 65 November 18, 2011
- The probability of using each
mode (e.g. car and PT) based
upon the relativity of the
generalised cost of travel by each
mode for each origin-destination
- A logit model is generally used
an s-shaped curve
- Steep curves more sensitive to
change i.e. a small change in gen.
costs leads to a large change in
number of trips
- Incorporates a mode constant for
non-quantifiable effects e.g.
comfort
Image here created as a fill
Original must be proportionate
to 8.5cm x 6.9cm
Image here created as a fill
Original must be proportionate
to 8.5cm x 6.9cm
i
j
Tij
m1
Tijm2
Trip assignment
Presentation Title Page 66 November 18, 2011
- Finding the least generalised cost
(travel time, distance, toll) path
between each origin and
destination in the network
- Iterated until nobody can change
their route between i and j and
reduce their generalised cost
- Time consuming in models with
large numbers of zones (MITM has
2912 zones = 2912
2
= 8.5 million
movements)
- Level of convergence is important
can affect measurement of
project benefits
Image here created as a fill
Original must be proportionate
to 8.5cm x 6.9cm
Image here created as a fill
Original must be proportionate
to 8.5cm x 6.9cm
v
j
i
v = travel demand from i to j
Generalised cost
November 18, 2011 Presentation Title Page 67
- Generalised costs are the costs perceived by the traveller as they decide which mode to
use.
- These sum time and monetary costs and using the value of time convert to a single unit
- Car generalised costs
In-vehicle time (generally no waiting time)
Vehicle operating costs (fuel)
Tolls
Parking charges
- PT generalised costs
Walk time
Waiting time
In-vehicle time
Fare
Interchange penalties
- PT in-vehicle time can be weighted to reflect crowding, punctuality of services etc.
Choice modelling
Choice theory & Utility
November 18, 2011 Presentation Title Page 69
Random Utility Models
- Mode split in four step models is generally based on
economic choice theory that uses Random Utility
Models (RUM)
- Most commonly expressed as a multinomial logit model
(MNL)
- Based on the premise that we make choices that
maximise the utility that we derive from the choices
Utility
November 18, 2011 Presentation Title Page 70
Utility
- The utility we derive from travel is the activity we want to go to
- If the cost of getting to the destination (the disutility) is greater than the
utility gained by the activity at the destination then the trip will not take
place
- Where a destination is chosen and the choice is between which mode to
use, choice theory states that the mode with the lowest disutility will be
chosen
Mode choice utility difference
November 18, 2011 Presentation Title Page 71
Percentage Public Transport
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
-1000 -800 -600 -400 -200 0 200 400 600 800 1000
Disutility Difference (Cents)
Favours Cars
Favours Public Transport
Mode choice random utility
November 18, 2011 Presentation Title Page 72
However
- economists express the theoretical utility as:
Theoretical utility = observed utility + random term
U = V + e
- The random term is included to reflect the uncertainty associated with the
estimated utility value
- If it is assumed that the random terms are independently and identically distributed
then analysts tend to use the multinomial logit model to model the uncertainty
Mode choice logit models
November 18, 2011 Presentation Title Page 73
Multinomial logit model
Where:
= the probability of using mode K between points i and j;
= the number of trips between i and j on mode K;
= the total number of trips between i and j;
= the generalised cost of travel between i and j on mode K;
= a scaling factor that measures the sensitivity to the utility
( )
( )

= =
k
k
ij
k
ij
ij
K
ij
K
ij
V
V
T
T

exp
exp
Pr
Mode Choice Multinomial Logit.
Choice
Walk Car PT
Mode Choice Binomial Logit.
Choice
Walk Motorised
Mode Choice Nested Logit.
Choice
Walk PT Car
Toll No Toll Bus
Train Walk
Access
Train Car
Access
Mode Choice Binomial Logit Example.
Where:
GC
PT
= the generalised cost (Utility) of public transport travel
GC
car
= the generalised cost (Utility) of car travel
Pr
PT
= the probability of using public transport
l = a scaling factor for the sensitivity of generalised cost
D = an alternative specific constant
( ) ( ) A + +
=
Car PT
PT
GC GC exp 1
1
Pr
Mode choice functions
Mode choice depends upon steepness of curve and the
delta offset (the mode constant)
Mode choice functions
Mode choice depends upon steepness of curve and the
delta offset (the mode constant)
Mode Choice Curves - Lambda.
Percentage Public Transport
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
-1000 -800 -600 -400 -200 0 200 400 600 800 1000
Disutility Difference (Cents) Favours Cars
Favours Public Transport
Low
High
Mode Choice Curves - Delta.
Percentage Public Transport
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
-1000 -800 -600 -400 -200 0 200 400 600 800 1000
Disutility Difference (Cents)
Favours Cars Favours Public Transport
D Favours Cars D Favours PT
Mode Choice Hierarchical Structure.
Choice
Walk Motorised
PT
Car
0% -1000
l
1
l
2
l
2
> l
1
0% -1000
Mode Choice Blended Cost.
All
Walk Motorised
PT
Car
| |
PT Car
C C
Mot
e e
C
2 2
log
1
2

=
| |
Mot Walk
C C
All
e e
C
1 1
log
1
1

=
Thank You