Victorian Integrated Transport Model

Henry Le

29 November 2011

Introduction

Background Model structure Main model features Other Components Validation Suggestions for further developments

Presentation Title

December 15, 2011

Page 2

Background
Recalibrated MITM 4 time periods 2272 zones Enhanced MITM for Metro - AM 2395 zones

PT & Other Enhancements

MITM for growth areas - AM 2789 zones

VITM 4 time periods 2959 zones

MITM for Melton - AM 2912 zones

MITM for Melbourne Airport - AM 2272 zones

Presentation Title

December 15, 2011

Page 3

Model structure

Presentation Title

December 15, 2011

Page 4

Main Model Features

Trip generation Mode choice and distribution model

Presentation Title

December 15, 2011

Page 5

Trip Generation - Trip purposes

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

VITM Development & Implementation

December 15, 2011

Page 6

Trip Generation - Production Trip productions are based on how many trips are made by each person, rather than by each household
Age, employment status car availability household structure geographic region

Presentation Title

December 15, 2011

Page 7

Home Based Trip Production

Segmentation by: 8 household types: the number of adults and employed adults 8 person types: based on age group and employment status 2 car availability groups Captive/competition Choice
Household Type ( ) 1 2 3 4 5 6 7 8 Adults 1 1 2 2 2 3 3 3 Employed Adults 0 1 0 1 2 1 2 3 Number of Household Vehicles ( ) 0 =0 =0 =0 =0 =0 =0 =0 =0 1 2 1 1 3+

=1 =1 =1 =1 =1 =1

2 2 2 3 3 3

=2 =2 =2

Car-ownership level segmentations by household-type segment. Shading indicates caravailability segmentation: Captive/Competition, Choice

Source: Recalibration and Revalidation of the Melbourne Integrated Transport Model (Vol1 Report, DOT, 2010)
VITM Development & Implementation December 15, 2011 Page 8

Non Home Based Trips

NHB trips are often directly related to home-based trip attractions NHB trip ends as a function of trip attractions Symmetry of trip ends NHB mode related to HB modes NHB generation model is implemented after the distribution and mode choice models
NHBO Trip Generation Model Coefficients
Mode (m) Purpose (p) HBW HBE_P HBE_S HBE_T HBSh HBSoc HBO 0.223 4 Mode (m) Region (j) Purpose (p) HBW HBE_P HBE_S HBE_T HBSh HBSoc HBO HBSh HBSoc HBO
Page 9

All

0.0620

Car

0.335 7

PT

Melbourne LGA

0.2769

Source: Recalibration and Revalidation of the Melbourne Integrated Transport Model (Vol1 Report, DOT, 2010)
VITM Development & Implementation December 15, 2011

Non-Melbourne LGA

0.1068

Mode/Destination Choices
Based on a hierarchical discrete choice model

Source: Recalibration and Revalidation of the Melbourne Integrated Transport Model (Vol1 Report, DOT, 2010)

VITM Development & Implementation

December 15, 2011

Page 10

Mode Choice/Distribution
Calculate the utility for Car & PT (with Attr. Const.) for mode choice (by origin) and destination (by matrix)

Motorised Trip Production

Mode Choice (Trip Production by mode)

Distribution (Trip matrix by mode)

Motorised Trip Attraction

Calculate Estimated Trip attraction RMSE of Trip Attraction Attraction Const

Check RMSE

Loop

Mode Choice/Distribution

VITM Development & Implementation

December 15, 2011

Page 12

Other Model Components

Network & zone system Regional Rail Demand Model Airport Model PT Model
Train & Tram Travel Time P&R PT Crowd Model

Presentation Title

December 15, 2011

Page 13

Network Developments Zone System

VITM

Internal zones

2893 including 2253 existing 123 Metro 1 & 2 517 growth areas

External zones

66 includes 19 highway zones 47 PT zones: Geelong, Ballarat, Bendigo, Gippsland and Seymour

Total

2959

Presentation Title

December 15, 2011

Page 14

Network Developments

New network includes

Detailed station platform for inner stations Extended regional rail links and zones

Presentation Title

December 15, 2011

Page 15

Regional Rail Model

Main Model Airport Model

Demographic Data

Initialization
Initial Networks

Generation Car Ownership & Household Type Model

Distribution

Mode Choice

Trip Generation

Initial Highway & PT Assignment

Network Updating

Time Periods

Combined Mode Choice& Distribution

Skims& Generalised Costs

Regional RailModel
Demographic Data Growth Parameters

Park& Ride Model RR Base Demand (AM Peak ) Special Trips : Commercial & Highway External

Model Feedback

Time Periods & Vehicle Occupancy

RR Demand (PM Peak )

+
AM IP PM OP PT Enhancement s
Travel Time Function : - Train - Tram - Bus + SkyBus PT Crowd Modelling PNR zone catchment Cost Averaging Existing Modules Cost Averaging

Highway Assignment

PT Assignment

New & Enhanced modules

Cost Blending

Presentation Title

December 15, 2011

Page 16

Regional Rail Model

Travel demand

T = train trips P = population A = attraction variable C = public transport generalized costs r1 & r2 = additional growth factors between 2008-21 and 2021-31

Presentation Title

December 15, 2011

Page 17

Regional Rail Model

Assignment

Services coded from the regional towns PT Assignment for regional & metropolitan at the same time

Presentation Title

December 15, 2011

Page 18

Airport Model
Trip generation Daily passenger trips
4 purposes: Business Resident Business non resident Non business resident Non business non resident
Time Split (AM, IP, PM) Highway Skims Demographics Passenger Trip Ends Airport Distribution GravityModel
(no assignment)

24 Hour Person Demographic Trip Matrix

Data

Trip distribution

PT GenCosts Car Gen Costs

Airport ModeChoice

(no assignment)

The trip distribution of residents based on the distribution of households

(h'way assignment)

ModeChoice

The trip distribution of non-residents - based on the distribution of total employments.

24 hrPublic 24 hr Private Vehicle Transport Matrix (Person) Matrix MITM Time & Occupancy Adjustment Factors

Time Split (AM,IP,PM,OP) Merge MITM & MATRS MITM Assignment

Mode choice
Private including taxi PT
Merge MITM into MATRS MATRS Assignment

Presentation Title

December 15, 2011

Page 19

Airport Model
Mode choice
Comparison between observed and modelled daily PT shares directions
Production (Arrivals) Modelled Observed PT share Priv. PT trips trips 6% 6107 392 12% 13% 27% 6596 13482 14950 903 2017 5549 Attraction (Departures) Modelled Observed PT share Priv. PT trips trips Demographic Data 4% 6239 262 10% 15% 28% 6745 13170 14769 755 2330 5731

Bus-Res BusNonRes NonBusRes NonBusNonRes

PT share 6% 12% 13% 27%

PT share 4% 10% 15% 28%

Assignment
Create a service for Skybus between Melbourne airport and Southern Cross station Mode 6 like express bus Headway 10 mins for all periods New fare system for mode 6 flat fare of 13$/trip Travel time function - Skybus Time (mins) = 1.2 x Car Time (mins)

Presentation Title

December 15, 2011

Page 20

PT Model - Train Travel Time

Calculate rail link distance using coordinates

Determine Rail segment speeds

Distance between two stations (km) 0-1 1 - 1.5 1.5 2 2 2.5 2.5 3 34 45 >5 Train Average Train Speed (km/hour) (excluding dwell time) 40 45 50 55 60 70 80 90

Travel Time function

Train In-vehicle Time (express/stop) = Segment Time + Dwell times (at actual stops, 0.75 mins)

Presentation Title

December 15, 2011

Page 21

PT Model - Train Travel Time

Train Travel Time: New Model vs Metlink (stop all stns)
120 y = 1.0233x R = 0.9811 100
70 90 80 y = 1.0401x R = 0.9689

Train Travel Time: New Model vs Metlink (Express)

80

60

Modelled

Modelled
0 20 40 60 80 100 120

50 40 30 20

60

40

20
10

0 0 10 20 30 40 50 60 70 80 90

Observed - Metlink

Observed - Metlink

VITM X COEFFICIENT STOP SERVICES EXPRESS SERVICES 1.02 1.04 R2 0.98 0.97

Presentation Title

December 15, 2011

Page 22

PT model - Tram Travel Time

Code Tramway speeds
Metlink Coded Travel Distance Speed Tramway Route No. (km) Time (km/h) (mins) Swanston St 3, 16, 64 10 1.4 10 3, 5, 16, 64, St. Kilda Road 67 17 4.4 15.5 Elisabeth St 57, 59 12 2.1 11 Collins Street 109, 112 12.3 2.1 10 Bourke Street 86, 96 13 2.8 13 16 Flemmington Road 59 7.5 2 16 Royal Pde 19 18.2 4.75 17.5 Victoria Pde 109 5.6 1.65 25.5 Gladstone/Station Streets 109 8 3.4 29 Ferrars Street/Cantebury Rd 96 10.5 5.1 23.5 Dandenong Rd 5, 64 13.1 5.1 15.5 Brighton Rd 67 14 3.7

Tramway network

Set Tram minimum speed: 10kph Calibrate Tram Travel Time function Tram Time (mins) = A x Car Time (mins) + B x distance (km)

Presentation Title

December 15, 2011

Page 23

PT Model - Tram Travel Time Comparison

Comparison of Tram Travel Time - AM peak Inbound
120
80 70

Comparison of Tram Travel Time - AM peak Outbound

y = 0.95x R = 0.89

100

80

y = 1.01x R = 0.77
Modelled

60 50 40 30

Modelled

60

40
20

20

10 0

0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 Scheduled Time - Metlink

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

Scheduled Time - Metlink

Tram Travel Time - AM Peak Inbound

120.0
80.0 70.0 60.0

Tram Travel Time - AM Peak Outbound

100.0

80.0

50.0 minutes 40.0 Observed 30.0 20.0 Modelled

minutes

60.0 Observed Modelled 40.0

20.0

10.0 3001 3003 3005 3006 3008 3016 3019 3048 3055 3057 3059 3064 3067 3070 3072 3075 3078 3082 3086 3096 3109

8 16 19 24 48 55 57 59 64 67 70 72 75 78 82 86 96 109 112 Tram Routes

Presentation Title

December 15, 2011

Page 24

Tram Routes

3112

PT Model - Tram Travel Time Profile

Route 112
90.00 80.00 70.00 50.00 60.00

Route 3112
70.00 60.00

40.00 30.00

Observed Modelled

minutes

minutes

50.00

40.00 Observed Modelled

30.00

20.00 20.00 10.00 5.0 10.0 15.0 20.0 10.00

5.0 10.0 15.0 20.0

Distance (km)

Distance (km)

Route 109
90.00 80.00 70.00 60.00

Route 3109
80.00 70.00 60.00 50.00

minutes

minutes

50.00 40.00 30.00 20.00 10.00 5.0 10.0 15.0 20.0 25.0 Observed Modelled

40.00 Observed 30.00 20.00 10.00 5.0 10.0 15.0 20.0 25.0 Modelled

Distance (km)

Distance (km)

Presentation Title

December 15, 2011

Page 25

PT Crowd Model
Approach PT capacity input Crowd curve
Demographic Data

Approach
Allow options to run one or both components Link Base Adjustment: Wait Time Adjustment: Set up for all 4 time periods

Parameters based on Melbourne Metro Business case Iterative process (input of no of iterations) Output: demand loading and restrained (metering) loading
Presentation Title December 15, 2011 Page 26

PT Crowd Model
PT Vehicle Capacities
Mode Bus Tram Tram Tram Tram Tram Tram Train Train Train Train V/Line Train V/Line Train V/Line Train V/Line Train V/Line Train V/Line Train V/Line Train Vehicle Type All types Type 1: A, Z, W Type 2: D1 (C1)&Z Type 3: B Type 4: C1 Type 5: C2&D2 Type 6: Z(A)&B, BA Existing 6-car NGTs 7-car NGTs 9-car NGTs 5 x 7 Car 6 Car Velocity 3x8VL, 3x7VL, 2x6VL 2x6VL, 3x3VL 1x5H Sprinter 2x3VL, 2x5N Seating Capacity 50 42 39 74 40 58 58 500 450 530 680 476 444 518 310 376 291 296 Crush Capacity 75 100 130 200 180 250 150 1,250 1,250 1,460 1,880 571 489 570 342 414 320 326 Applies to routes 3,24,30,57,64,67,70,78,82; Route 64 and 67 upgraded to Type 3 in 2021 Demographic Data Applies to routes 5,6,8,16,48, 72 Applies to routes 19,59,75,86 Applies to route 109 only Applies to route 96 only Applies to routes 1,55,112; Route 112 upgraded to Type 5 in 2021 Notes

Applies to Melton/Bacchus Marsh lines Applies to Ballarat line only Applies to Geelong line only Applies to Bendigo line only Applies to Kyneton line only Applies to Seymour line only Applies to Traralgon line only

Presentation Title

December 15, 2011

Page 27

PT Crowd Model
Crowding factor curves
2.0 1.8

Crowding Factor

1.6

1.4

Demographic Data

1.2

1.0

0.8 0% 20% Existing Trains Tram Type 3 Bus 40% Utilisation Metro Trains Tram Type 4 VLine 1 Tram Type 1 Tram Type 5 VLine 2 Tram Type 2 Tram Type 6 60% 80% 100%

Load Factor (% of Seat Occupied) <70% 70%- 100% Crush Load

Perceived In Vehicle Time Factor Seated 1 1.1 1.3 Standing 1.4 2.0

(D- LDF*Seat cap) U= ---------------------------(Crush Cap LDF*Seat cap)

Presentation Title

December 15, 2011

Page 28

Other model developments

Initialization Module

Initializing and adjusting highway network

Generating zone catchment for public transport Park and Ride model Producing initial highway and public transport skims and generalised costs for the first iteration of the model run
Demographic Data

Average cost skims

For both highway & PT skims

To improve model convergence Measures to improve model run time

Set up factor file parameter MAXFERS=1 for first 3 iterations

Presentation Title

December 15, 2011

Page 29

Model Validation

Validations Highway Am & PM

Presentation Title

December 15, 2011

Page 31

Validations PT System Boarding

Model TRAIN TRAM BUS Total Observed TRAIN TRAM BUS Total % Diff TRAIN TRAM BUS Total

AM 179,568 100,548 71,548 351,665 173,588 85,763 74,102 333,453 3% 17% -3% 5%

IP 177,984 177,293 101,135 456,412 159,219 190,657 114,459 464,335 12% -7% -12% -2%

PM 220,875 136,050 107,208 464,132 223,534 131,790 106,998 462,322 -1% 3% 0% 0%

OP 124,935 110,201 50,020 285,155 129,323 106,437 44,461 280,221 -3% 4% 13% 2%

Daily 703,362 524,092 329,910 1,557,365 685,664 514,647 340,020 1,540,331 3% 2% -3% 1%

Presentation Title

December 15, 2011

Page 32

Validations - Train boarding

Presentation Title

December 15, 2011

Page 33

Validations Train Cordon (AM peak)

Cordon Northern Cordon Line Williamstown Werribee Sydenham Craigieburn Upfield Epping Hurstbridge Ringwood Corridor Alamein Glen Waverley Dandenong Corridor Frankston Sandringham Total Observed PTD 2008 4,147 7,282 9,919 10,310 3,455 6,545 10,488 20,092 2,810 7,766 16,097 13,353 9,038 121,303 Model AM 1,502 11,224 12,306 12,619 3,785 8,788 11,990 26,065 2,708 6,422 21,932 14,021 5,655 139,017 Difference -64% 54% 24% 22% 10% 34% 14% 30% -4% -17% 36% 5% -37% 15%

Clifton Hill Burnley Caulfield

30,000 25,000 20,000 15,000 10,000 5,000 0 Observed PTD 2008

Presentation Title

Williamstown Werribee Sydenham Craigieburn Upfield Epping Hurstbridge Ringwood Corridor Alamein Glen Waverley Dandenong Corridor Frankston Sandringham

Model AM

December 15, 2011

Page 34

Validations - Tram boarding

Presentation Title

12/15/2011

Page 35

Validations Comparison of tram loading at CBD cordon

Corrid or N NW S SE E Total Observed AM IP 7,453 5,583 5,433 4,296 4,846 27,611 10,433 6,814 6,614 5,926 6,998 36,784 PM 8,109 3,510 5,577 5,168 3,795 26,158 OP 7,027 3,350 3,684 4,440 4,658 23,157 Modelled AM IP 6,132 2,430 5,192 3,665 4,055 21,474 13,168 5,751 7,315 5,359 6,217 37,809 PM 13,354 5,212 8,709 3,670 2,788 33,733 OP 9,353 3,571 5,432 2,778 2,321 23,456

120,000 100,000 80,000 60,000 40,000 20,000 0 AM IP PM OP Daily Observed Modelled

Presentation Title

December 15, 2011

Page 36

Validations - Bus boarding

Conclusions
Model is robust and comprehensive

Suggestions for further enhancements

Regional rail demand IP & OP Park and Ride model with parking capacity constraint Time choice module Taxi mode Mode choice for active mode

Acknowledgements
I wish to thank the DOT, Victoria for allowing to use their data in this presentation

Thank You

[email protected]