5. 5 |5 |
About us…
3,029
bridges
20
ports
3,260
taxis licensed
5m
vehicles registered
3.5m
drivers licensed
256,151
recreational vessel
registrations
997,289
boat licenses
180m
in SEQ
12.1m
outside SEQ
trips taken annually on bus,
rail, ferry and light rail
33,343km
state-controlled roads
As at 30 June 2016 we manage: As at 30 June 2016: As at 30 June 2016 there were:
3.63m
customers served
face-to-face at
59
Customer Service Centres
2.5m
go cards
in use
Over 490,000
passengers travel on the
south-east Queensland
network on average
each day
Our customers conducted
6.68m
online services
Creating a single integrated transport network accessible to everyone
AITPM 2017 National Conference, Melbourne | 17 August 2017
6. 6 |6 |
• Demand
• Network
• Sub Models
• Calibration Validation
Greater Brisbane Transport Demand Model
Persons, Purpose Choice
Activity / Access Choice
Available Facilities
(Network, Services, Policy)
Suitable for Use
• Determined by Individuals
• Influenced by Facilities
AITPM 2017 National Conference, Melbourne | 17 August 2017
7. 7 |7 |
• Gravity models have fewer parameters
& tend to consider only “how much”
Form: Discrete Choice vs Gravity
$
$
$
• Discrete Choice models have a high level of complexity
and consider choices that are different or unique from
each other. Discrete choice models are developed to also
consider, “which one” across a range of variables
AITPM 2017 National Conference, Melbourne | 17 August 2017
8. 8 |8 |
Contrast: BNE vs. BSTM-MM
Converge to zonal
demand by period
More full use of HTS
Finer person/purpose
segmentation
Discrete choice form
provides sensitivity
Time, Distance, Cost
Peak & Period spreading
Internally consistent
Converge to daily
assignment road statistic
Less full use of HTS
Aggregate form reduces
segmentation
Singular gravity form is
insensitive
Generalised cost
No time choice
Prone to externalities
AITPM 2017 National Conference, Melbourne | 17 August 2017
9. 9 |9 |
Demand Segments
• 8 person types by 5
purposes
• Informed by network
conditions
• Singly constrained
• NHB trips reference
home based trip leg
• Adjustment for under-
reporting (discretionary)
State School
High School
University
Blue Collar
Worker
White Collar
Worker
Adult 18-54
Adult 55-74
Adult 75+
Home to
Primary
Primary to
Home
Home to
Secondary
Secondary
to Home
Non-Home
based Other
Home
AITPM 2017 National Conference, Melbourne | 17 August 2017
10. 10 |10 |
Demography
• Household based to Person-based
• SA1 Zones
• Zone consistency (in/out, land use,
greenfield development)
• Futures Demography defined by
TMR
• Informed by ABS Census & QGSO
(Queensland Government Statistics
Office)
MANY MORE
dwellings to
2041
AITPM 2017 National Conference, Melbourne | 17 August 2017
11. 11 |11 |
Networks
• Assign informs Demand model
– Time, Distance, Cost
• Facilities, lanes & bendiness
• Active modes, on & off road
• Light commercial vehicles (LCV)
• Network futures defined by TMR
QTRIP, TIPPS, financial constraints
S0.5 1.0 1.5 2.0 km
AITPM 2017 National Conference, Melbourne | 17 August 2017
12. 12 |12 |
Volume Delay Functions
• Modified BPR Volume Delay Function (VDF)
• Considers link end control:
Freeflow, Giveway, Signal
• Multiple (capacity based) VDF values
for each control type
• Give-way and signal inherent delay
• All VDF attributes internal to model
supports model being internally consistent
AITPM 2017 National Conference, Melbourne | 17 August 2017
13. 13 |13 |
Parking Model & Toll Model
• Logit Toll Model determines toll or no-toll
routing choice
• Time savings, average (zonal) income and
toll cost
• Logit attributes internal to model –
supports model being internally consistent
• Two car classes; do (ct) or do’t (c) use toll
• Logit Parking Model - focus on Central Traffic Area
• Capacity constrained across 4 parking types
• Remote park and walk to destination
P
AITPM 2017 National Conference, Melbourne | 17 August 2017
14. 14 |14 |
Public Transport
• PnR Lot capacity formal & informal parking
• PnR parking period accumulations
• KnR no capacity constraint
• Fares - concessions, peak/offpeak
• PT Patronage calibrated to
screenlines, board/alight
AITPM 2017 National Conference, Melbourne | 17 August 2017
15. 15 |15 |
Demand Model Process
At the highest model level, a daily composite utility of travel can be calculated
Logsums up
Composite utilities from
nested logit formulation
estimated at each stage
and fed upward
Probabilities down
Demand estimates
made using these
composite utilities
feeding down
AITPM 2017 National Conference, Melbourne | 17 August 2017
16. 16 |16 |
Overall Process
Assign to convergence
Time, Distance & Cost Skims
Seed
matrices
Mode Choice, Time of Day,
Destination Choice Generation
Demand
Trip Tables
Convergence Check
Trip Tables Previous vs Current
FEED
BACK
Big Loop
New Demand matrix via
variable blending to
¾ previous ¼ current convergence not met
convergence
met
Small Loop
New Demand matrix via
variable blending to
¾ previous ¼ current not met
convergence
Convergence Check
Trip Tables Previous vs Current
Mode Choice,
Time of Day
Demand
Trip Tables
Assign to Convergence
Time, Distance & Cost Skims
convergence met or loop count
Post
Processing
Final
Assignment
Mode Choice, Time of Day,
Destination Choice Generation
Assign to convergence
Time, Distance & Cost Skims
AITPM 2017 National Conference, Melbourne | 17 August 2017
17. 17 |17 |
Convergence Criteria
• Demand Based
• <1% mean absolute (mabs) difference in car trip demand
matrices between iterations on an zonal basis for all
scenarios
• Variable blending: an efficient convergence process
• Final loop
• Demand in is consistent with demand out
Both practically and theoretically sound
Internally consistent
AITPM 2017 National Conference, Melbourne | 17 August 2017
18. 18 |18 |
Assignment Calibration
• Screenline r2 & Y values Road and PT
R²
(>0.85)
y
(0.9x - 1.1x)
R²
(>0.85)
y
(0.9x -
1.1x)
AM6 0.976 0.911 Y 0.848 0.932 N
AM7 0.983 0.999 Y 0.86 1.006 Y
AM8 0.985 1.172 N 0.87 1.153 N
IP 0.982 0.899 Y 0.889 0.945 Y
PM3 0.983 0.921 Y 0.899 0.941 Y
PM4 0.984 0.928 Y 0.9 0.956 Y
PM5 0.983 1.066 Y 0.884 1.076 Y
RD 0.984 0.902 Y 0.9 0.966 Y
Daily (aggregated)0.989 0.95 Y 0.94 1.001 Y
Road Screenline Counts Road Individual Counts
R²
(>0.85)
y
(0.9x - 1.1x)
R²
(>0.85)
y
(0.9x -
1.1x)
AM6 0.962 2.718 N 0.884 2.433 N
AM7 0.992 1.508 N 0.905 1.495 N
AM8 0.956 0.799 N 0.924 0.832 N
AM peak Total0.985 1.284 N 0.92 1.235 N
IP 0.993 0.757 N 0.957 0.657 N
PM3 0.951 1.565 N 0.915 1.108 N
PM4 0.975 1.183 N 0.893 0.972 Y
PM5 0.972 1.086 Y 0.926 1.808 N
PM peak Total0.973 1.218 N 0.924 1.322 N
RD 0.871 0.661 N 0.933 0.88 N
Daily (aggregated)0.99 1.143 N 0.943 1.044 Y
Rail Screenline Counts Bus Screenline Counts
AITPM 2017 National Conference, Melbourne | 17 August 2017
19. 19 |19 |
Road Screenline Calibration
• Road Screenlines
AITPM 2017 National Conference, Melbourne | 17 August 2017
20. 20 |20 |
• PT Screenlines
AITPM 2017 National Conference, Melbourne | 17 August 2017
PT Screenline Calibration
21. 21 |21 |
Segment Calibration
• Observed vs Modelled Demand Comparison
AITPM 2017 National Conference, Melbourne | 17 August 2017
22. 22 |22 |
Demand based criteria
Calibration and Validation is sound
Evidence based futures
Internally consistent
Soft Launch participants satisfied
First Discrete Choice transport demand model in Australia
Suitable for use?
AITPM 2017 National Conference, Melbourne | 17 August 2017
23. 23 |23 |
Greater Brisbane Transport Demand Model
Questions?
AITPM 2017 National Conference, Melbourne | 17 August 2017
24. 24 |24 |
VDF Plot form
• Modified BPR form of Volume Delay Function
AITPM 2017 National Conference, Melbourne | 17 August 2017
25. 25 |25 |
Trip Distance by Segment
• Observed vs Modelled Trip Distance by Segment
Secondary Student Home to School
75+ Home to Other
Adult Worker WC Home to Work
Adult Worker BC Home to Work
AITPM 2017 National Conference, Melbourne | 17 August 2017
Editor's Notes
Values
In 2013 the Queensland Public Service launched new values to help revitalise the public service as seen on the slides
These are customers first, ideas into action, unleash potential, be courageous and empower people.
At TMR we are working hard to embed these in our culture and we are already working more collaboratively, more productively and smarter.
Diversity
TMR have a responsibility to ensure our employee pool reflects and better serves the diverse nature of the Queensland community. Diversity brings different perspectives into the workplace.
An inclusive workplace ensures employees feel confident and supported to contribute and participate fully in the workforce.
Diverse and inclusive teams:
generate new ideas
challenge the status-quo
introduce fresh ways of looking at problems
offer a wider range of potential solutions.
Queensland Government has four objectives for the community:
Creating jobs and a diverse economy
Protecting the environment
Delivering quality frontline services
Building safe, caring and connected communities
TMR has a key role to play in ensuring we are achieving these objectives and as such, our priorities are aligned to do so.
Advance Queensland
The Queensland Government is investing $405 million over four years to create the knowledge-based jobs of the future – this is Advance Queensland.
Advance Queensland is a comprehensive suite of programs, based on international evidence of ‘what works’, and is designed to:
create the knowledge-based jobs of the future,
drive productivity improvements and
build on our natural advantages.
It will help position the state as an attractive investment destination with a strong innovation and entrepreneurial culture.
OUR STRATEGIC PRIORITIES
TMR has responsibility for shaping the transport system, building the network, managing use of the system and providing passenger services for Queenslanders.
We need to deliver our responsibilities in the right way. Our priorities are:
Our customers
Innovation
Liveable regions and active cities
Regulation
Sustainable funding
Contemporary workforce.
In the short and long-term we continue to deliver new and upgraded infrastructure under a substantial capital program.