An Update On Advanced Model Development

1
An Update On AdvancedModel Development

• Ken Cervenka, P.E., AICP
• North Central Texas Council Of Governments
• 11th TRB National Transportation
• Planning Applications Conference
• Daytona Beach, Florida
• May 8, 2007

2
Agenda

• TRB Innovations in Travel Demand Modeling
• Conference
• Activities in the U.S.
• Activities Outside the U.S.
• What Still Needs to Happen

2
3
TRB Innovations In Travel Demand Modeling
Conference

• Austin, Texas, May 21-23, 2006
• Interaction Between Researchers and
  Practitioners
• Sunday Workshops
• Innovations in Practice
• FTA Findings for Meaningful Forecasts
• Monday Plenary Sessions
• Overview of the Policy Issues
• Moving Innovative Models Into Practice

3
4
Innovations Conference (Cont.)

• Monday Breakout Session 1
• Tour Based Models
• Data and Synthetic Populations
• Land Use Forecast Models
• Monday Breakout Session 2
• Activity Based Models
• Survey Methods
• Assignment Advances

4
5
Innovations Conference (Cont.)

• Tuesday Breakout Session 3
• Education and Outreach
• Emerging Modeling Considerations
• Validation
• Tuesday Breakout Session 4
• Transitioning to a New Model Framework
• The Future of Travel Behavior and Data
  Collection

5
6
Innovations ConferenceClosing Session
Institutional Issues

• MPO Roundtable Discussion
• What Did You Learn?
• What Questions Were Not Fully Answered?
• Obstacles to Overcome to Move Forward
• Agency Expectations for Moving Forward

6
7
Innovations ConferenceObservations (1)

• The More You Know, The More You Know You Dont
  Know
• One Size Does Not Fit All
• Better Models Should Inform Us of a Lot More and
  Make Us
• Think Differently
• The Driving Force for New Models Should be the
  Scope of Real-
• World Issues Needing Decisions

7
8
Innovations ConferenceObservations (2)

• Better Models For
• Investment Analysis
• Goods Movement Planning
• Operating Decisions and Policies
• Impacts of Pricing Mechanisms on Both Revenue
  Generation and Behavioral Change
• Environmental/Energy Impacts
• Supply/Demand Relationships Between
  Transportation and Land Use/Economic Development

8
9
Innovations ConferenceObservations (3)

• Positive Push Towards Understanding the True
  Causality of
• Behaviors, Rather Than Simple Statistical
  Representations
• But How Different Will the Forecast Results
  Obtained From
• Theoretically Stronger Models Be, and Will They
  Lead to
• Different Choices by Decision Makers?
• While Several MPOs are Close to State-of-the-Art,
  Average
• Practice is Well Below Yet Most Agencies Appear
  Satisfied
• with Their Current Tools

9
10
Innovations ConferenceObservations (4)

• Strong Desire by Practitioners to See the More
  Advanced
• Tools Put Through Rigorous Sensitivity and
  Forecast
• Validation Checks
• But Existing Four-Step Models Should Also be
  Subjected
• to Critical Review
• The Travel Model Industry Does Not Have the
  Same
• Commitment to Review as Other Industries

10
11
Advanced Model Structures In U.S.

• In Use (Applications Studies)
• Columbus, Ohio (MORPC)
• Lake Tahoe Region, Nevada
• New York City (NYMTC)
• Sacramento, California (SACOG)
• San Francisco County
• St. Louis, Missouri (EWGCOG)

11
12
Advanced Model Structures In U.S.

• In Development or Testing
• Atlanta, Georgia (ARC)
• Dallas-Fort Worth, Texas (NCTCOG)
• Denver, Colorado (DRCOG)
• Michigan DOT
• Ohio DOT
• Oregon DOT
• Portland, Oregon (METRO)
• San Francisco/Oakland, California (MTC)

12
13
Advanced Modeling Outside The U.S.

• CalgaryWeekend Activity Model
• Copenhagen, DenmarkTour-Based
• France (Several Cities) Simultaneous Destination
• And Mode Choice Models
• JakartaActivity-Based Microsimulation Model
• Netherlands (ALBATROSS)
• SantiagoConsistent Equilibration Between Supply
• and Demand
• SydneyTour-Based Model

13
14
What Still Needs To Happen (1)

• Major Focus Needed on Data Quality and Asking
• the Right Questions, and Not Just Quantity
• Weaknesses in the Survey Data and
• Networks Needed to Drive New Models
• For Estimation (e.g., Trip Records)
• For Calibration (e.g., Transit Shares)
• For Validation (e.g., Time-Of-Day Counts)
• For Forecasting (Realistic/Balanced Networks
• And Demographics)

14
15
What Still Needs To Happen (2)

• Even Theoretically Correct Advanced Models
• Must Still be Demonstrated to be Useful
• Estimated Coefficients Look Reasonable
• Validates Well to the Calibrated Base Year
• Forecast Validation Checks Look Good
• Sensitivity Tests and Other Integrity Checks
  Make Sense
• Long-Range Forecasts Look Reasonable
• Decision Makers Accept the Results

15
16
What Still Needs To Happen (3)

• Need More Clear-Cut Examples of Success
• Stories
• Different Transportation Decisions Were Made
• Model Run Times are Worth the Effort
• Management Recognition of the Costs for Good
• Data Collection/Analysis, and Model Development
• Be Cautious About Direct Cost Comparisons
  Against
• Typical Model Development Efforts

16
17
What Still Needs To Happen (4)

• Clearinghouse of Information
• International Knowledge Base
• Critical Review of Modeling Procedures
• Existing In Place Models
• Use of Same Testbed Data (e.g, NGSim)
• Critical Reviews of Available Software
• Creation of Transit Skims
• Road/Transit Assignment Details

17