Development of Northwest Louisiana’s Travel Demand Model

1
Development of Northwest Louisianas Travel
Demand Model

• Model Conversion from a
• Small/Mid-Size MPO Perspective

Florida Model Task Force Meeting November 12th
13th / 2003 Chris Petro, AICP Tran. Plan. Mgr.
NW Louisiana Council of Governments
2
Presentation Outline

• What are we modeling study area background
• Where have we been and where are we going
  conversion experience
• ? Answer submitted technical questions part 1
• Static model development TransCAD (Brief)
• ? Answer submitted technical questions part 2
• Model demonstration

3
Geographic Overview
4
Geographic Overview
5
Geographic Overview
6
Study Area Transportation Network Characteristics

• Area Interstates 49, 20, 220 (a
  circumferential route) future I-state projs
  I-49 N/S and I-69
• EPA Designated - Attainment For AQ
• SporTran public trans. (approx. 3 mil.
  pass./year)
• Special Generators

Gaming
Barksdale AFB
7
Study Area Transportation Network Characteristics

• Area Interstates 49, 20, 220 (a
  circumferential route) future I-state projs
  I-49 N/S and I-69
• EPA Designated - Attainment For AQ
• SporTran public trans. (approx. 3 mil.
  pass./year)
• Special Generators

Gaming
Barksdale AFB
8
Study Area Transportation Network Characteristics

• Area Interstates 49, 20, 220 (a
  circumferential route) future I-state projs
  I-49 N/S and I-69
• EPA Designated - Attainment For AQ
• SporTran public trans. (approx. 3 mil.
  pass./year)
• Special Generators

Gaming
Barksdale AFB
9
Model Conversion Experience Where have we been

• 1990 Long Range Plan - TRANPLAN based travel
  demand model
• 200 TAZs US Census Blk. Group Geography
  (hardcopy maps)
• Physical network features were NOT spatially
  referenced (x-y grid) were drawn to scale
• 1991 links/TAZs digitized in ATLAS GIS software
  (1990 TIGER spatial accuracy)

10
1990 MPOPlanning Study Area
1990 Study Area Stats.

• Urban Pop. 284,928
• (US Census 1990)
• Land Area 499 Sq. Mi.
• 1990 US Census defined Urban Area Boundary (UZA)

11
TRANPLAN Model Details

• Base Year (1990) Socio-Economic Data
• 1990 US Census (pop. and housing)
• Dept. of Labor (employment)
• EI and EE trips estimated from multiple
  regression eq. relating 1989 travel survey data
  to the 1990 socio-econ. data
• Custom Trip Gen model (solve A solve P)
• Gravity Model
• Equilibrium assignment
• technique

12
1990 Modeled Network Features
13
1990 Modeled Network Features
14
Why Did We Convert? (Internal Justification)

• Timing Issues Long Range Plan Update
• TRANPLAN file organization was becoming
  cumbersome
• TransCAD Best At Blending Travel Model
  Development With GIS (meshed well with our GIS
  effort)
• Conversion and Import/Export Tools
• Reasonable Learning Curve

TransCAD Data Tools
15
Why Did We Convert? (External Forces)

• Louisiana Planning Council (LPC) Through LADOTD
  Purchased TransCAD For All Urban Area Models (8
  MPOs)
• LADOTD Initiated Their Long Range Plan TranCAD
  Utilized For Their Model Component
• Other State DOTs and MPOs Successful
  Implementation
• U.S. DOT Endorsement

16
Salvage Effort (Physical Network)
Since we expanded our Study Area nearly had to
start from scratch

• The 1990 TAZ boundaries were still valid
• - Define new TAZs for expanded Study Area
• 1990 links were scrapped for more geospatially
  accurate and topologically correct links
  developed from the Regional GIS centerline
  features
• No 1990 TAZ/External Sta. centroids were salvaged
• Bottom line New link/TAZ features built from
  existing GIS dataset NOT directly from TRANPLAN

17
Salvage Effort (Network Attributes /
Socio-Economic Data)

• We associated (tag within distance function) the
  1990 link attribute data to the new TransCAD base
  network, but we utilized the existing1990 link
  structure that had been already built and
  referenced in our GIS
• Reused the link capacity table (1985 HCM lane
  capacity by facility type)

Tip A form of triangulation (3 or more known
points) can be used to project unreferenced
spatial features usually through a GIS
extension or add-on projection utility
18
Salvage Effort Quality Control (QC)

• QC 1990 TAZ boundaries against the more spatially
  accurate 2000 Base Network links (centerlines)

Model development geocoding activities O-Ds from
External Travel Survey or Attraction Model
locating employment

• FYI Planning to use spatial analysis
  techniques (e.g. geocoding)
• make sure all the datasets have comparable
  spatial accuracy

19
Detailed Technical Questions Session 1

• Developing a TransCAD database from TRANPLAN
  ASCII files
• 1) What process was set in place to make sure
  the functional classification and area type
  classification of roadways was consistently and
  correctly converted?
• 2) How are speeds and capacities converted to
  maintain consistency between the TRANPLAN
  platform and the TransCAD platform?
• 3) What was done to ensure that the centroid
  loadings were put in the same locations and had
  the same number of connection for each zone?
• 4) How was the database structure set to ensure
  that future year networks were consistent with
  and correctly built upon the base year network?
• 5) How is the database maintained to preserve
  the integrity of the base network for each of the
  model years so that alternatives are never mixed
  with the base data set?

20
Model Development Philosophy Where are we going

• Limited resources Start simple and embellish
  later
• Follow proven methodology UTP Four-Step
  Modeling Process (1990 Model)
• Modeled network accurately represents
  transportation features
• Account for our unique future development
  scenarios (Statewide Plan Project Funding vs. No
  Statewide Funding)

21
Model Development Philosophy Statewide Plan
Funding Track
MPO SWP Projects
22
Model Development Philosophy Statewide Plan
Funding Track
MPO SWP Projects
23
Current MPOPlanning Study Area
Study Area Stats.

• Population 350,000 (US Census 2000)
• Land Area 1805 Sq.Mi.

24
Model Methodology Outline




Mirrors our 1990 Model Methodology
25
Development Physical Network

• Based Upon Regional GIS Project referenced
  features imported into TransCAD
• Links/Nodes - Topologically Correct (QC link
  direction and coding)
• Expanded 1990 TAZ Structure To 2000 Study Area
  (TAZUP)

26
Development Socio-EconomicData Model

• Base Year Data Sets 2000 US Census (prods.) and
  Dept. of Labor Employ. Dbase (attracts.)
• Tagged To TAZs
• Grow The Data The Delphi Method

 
Delphi Committee
Delphi Zones
Delphi Results
27
Development Socio-EconomicData Model

• Base Year Data Sets 2000 US Census (prods.) and
  Dept. of Labor Employ. Dbase (attracts.)
• Tagged To TAZs
• Grow The Data The Delphi Method

 
Delphi Committee
Delphi Zones
Delphi Results
28
Development Socio-EconomicData Model

• Base Year Data Sets 2000 US Census (prods.) and
  Dept. of Labor Employ. Dbase (attracts.)
• Tagged To TAZs
• Grow The Data The Delphi Method

 
Delphi Committee
Delphi Zones
Delphi Results
29
Development Socio-EconomicData Model

• Base Year Data Sets 2000 US Census (prods.) and
  Dept. of Labor Employ. Dbase (attracts.)
• Tagged To TAZs
• Grow The Data The Delphi Method

 
Delphi Committee
Delphi Zones
Delphi Results
30
ExternalTravel Survey

• Undertaken 11/99
• Purpose Update 1989 survey O-D data
• Quick QC use TransCADs desire line
  functionality

 
QC Ext Survey Data
31
ExternalTravel Survey

• Undertaken 11/99
• Purpose Update 1989 survey O-D data
• Quick QC use TransCADs desire line
  functionality

 
QC Ext Survey Data
32
ExternalTravel Survey

• Undertaken 11/99
• Purpose Update 1989 survey O-D data
• Quick QC use TransCADs desire line
  functionality

 
QC Ext Survey Data
33
Trip Generation Component

• Estimate TAZ Ps/As
• Trip generation rates by trip purpose will
  utilize standardized trips rates developed from
  the external travel survey
• Production Attraction models will use two way
  cross-classification

Example Prod. Rates
34
Trip Distribution Component

• Good Ol Gravity Model
• Process Factors
• Requires trips from trip generation model
• Requires travel time matrix from defined network

35
Modal Split Component

• Not Applied To Our Model
• Transit Trips Account for 0.3 - 0.4/Year Of
  Total Trips
• Cant Justify Individual Choice Models
  (multinomial Logit)

36
Trip Assignment Component

• Utilized User Equilibrium Assignment
• Process Factors
• Requires trip matrices from trip distribution
• Requires a QCd network

37
Calibration / Validation ProcessPerformance
Indicators

• Study Area Percent Error
• Total ground counts compared to total
  assignments for all screenline locations (lt5)
  our model 2.2

• Reasonable Trip Loadings

Classification Freeways Less than 7 Principal
Arterials Less than 10 Minor Arterials Less than
15 Collectors Less than 25 Source FHWA
Calibration Adjustment of System Planning
Models December 1990
1990 Model 2000 Model 6.6 6.4
1.5 2.3 1.4
1.3 5.6 6.5 Source
FHWA Calibration Adjustment of System Planning
Models December 1990
38
Calibration Toolbox

• Link penalties (Penalty.bin file)
• Link speeds
• Add and/or reconfigure placement of centroid
  connectors
• Detective Work uncover those coding errors
  (use TransCAD direction of flow and topology
  functionality)

39
The Real Test For SuccessCalibration - 1990
Model and 2000 Screenline Results

• Not comparing TransCAD to TRANPLAN
• Checking assignments for reasonableness

40
Model Uses in Support of Plan

• Performance Measures to Assess Regional
  Transportation System
• Vehicle miles traveled (VMT)
• Vehicle hours traveled (VHT)
• Average speeds
• Volume to capacity (V/C) ratios
• Screen line analysis

41
Model Status

• Finalized initial development (operational)
  10/2003
• Maintained and updated by myself and one other
  staff planner
• Future model development

Conversion Observations

• Didnt salvage too much from TRANPLAN made QC
  manageable from a smaller MPO perspective
• Very positive conversion

42
Detailed Technical Questions Session 2

• Conversion Process
• 1) Describe your experiences, touching on the
  time, effort and costs for converting and
  building your highway networks in TransCAD.
  Describe the same for the transit networks.
• 2) In hindsight are there processes and
  procedures that you would have done differently
  now having completed the conversion.
• 3) Are there programs and features in TRANPLAN
  that you could no longer perform once you
  converted to TransCAD?
•  
• Others
• 1) What is the protocol for file transfer (i.e.
  what needs to be included so that all the files
  that TransCAD creates and needs to operate are
  included when transferring datasets)?