Vermont Agency of Transportation (VTrans) Statewide Travel Model Peer Review Report

Peer Review Session 2: TMIP Peer Review of the Vermont

Travel Model

• Questions for the peer review panel
• Comments from FHWA review
• Model limitations discussed in Session #1
• Addressing limitations with NCHRP 735
• Other resources for model improvements
• Peer review schedule

Questions for the Peer Review

Panel

1. FHWA identified issues in the current model, particularly as it relates to the use of different modeling platforms. How should these issues be addressed to ensure VTrans has a credible and effective model?
2. Resiliency planning has become a major focus at VTrans. Hurricane Irene caused significant damage to the state's transportation infrastructure. A major focus of our planning efforts in this area will be infrastructure design that can withstand such storms. How can VTrans use the model for resiliency planning?
3. Vermont has set ambitious objectives towards the goal of reducing energy use and emissions. These included, for example:
   1. Keep VMT annual growth rate to 1.5% (half of the national average) or less for that portion controlled by the state.
   2. Increase public transit ridership by 110%, to 8.7 million annual trips by 2030.
   3. Quadruple passenger rail trips, to 400,000 Vermont-based trips by 2030. d. Reduce share of SOV commute trips by 20% by 2030.
   4. Double bicycle and pedestrian share of commute trips, to 15.6%, by 2030.
   5. Double ride share commute trips, to 21.4% of all commute trips, by 2030. How can we use the model to monitor and evaluate progress towards goals/targets?

Questions for the Peer ReviewPanel

4. While there will continue to be a need to evaluate an occasional highway capacity project, system preservation will dominate the work of VTrans. What role, if any, could a travel demand model have in system preservation, and possibly disinvestment?
5. Performance-based planning and programming are a core component of MAP-21 requirements. How can the model assist us in developing and monitoring performance measures moving forward? Can the model play a role in an asset management system?
6. VTrans is in the process of developing a fair-share methodology; with the aim of ensuring developers pay for the proportional impact triggered by development. How can we use the model to contribute to the fair-share methodology?
7. Bicycling, walking, transit, and rail are important components of the Vermont's transportation system. Some of these modes, particularly bicycling and walking, are primarily local in scope. Is the State's travel demand model the appropriate scale to prioritize corridors for improvements (i.e. where to widen lanes for bicycle use)?

Comments from FHWA review

• Differences between the official Model platform and those used by consultants or regions
  • Complete model runs with forecast
  • Extraction of data from the Model
• Review or validation of highway network
• Reduced average trip impedances from zone refinement
• Short- and long-distance travel aggregation
• Potential discrepancy between the default speed / capacity tables and the future year assignments
• Value of flow adjustments for model fit
• Value of a comprehensive user's manual

Model Limitations Discussed in Session #1

• Assumptions regarding NHB travel
• Growth forecasting
• Feasibility of including transit
• Long- and short-distance travel
• Vehicle-ownership model
• Economic modeling

Assumptions regarding NHBtravel

• No longer includes commercial truck travel
• Updated trip-generation regression-equation coefficients:
• Households - 0.89
• Retail Jobs - 2.56
• Manufacturing Jobs - removed
• Non-Manufacturing Jobs - 0.41
• Government Jobs - 0.86
• Primary School Jobs - removed
• University Jobs - removed
• Regression-equation r-squared - 0.64
• 29% of all person-trips per day - 611,586
• Trip distribution (gamma) coefficients: a: 87,565; b: 1.338; c: 0.098
• Diagonal symmetry

Model Limitations Discussed in Session #1

• Growth forecasting
• Standard source?
• Capping minimum household growth rate at 0.0%,
• Sources of information for population growth vs. household growth
• The feasibility of including transit fully in a Model with only a daily temporal resolution: does it make sense to include only certain types of regional transit, or none at all? Can the hourly variations in transit use be reflected in a daily model?
• Long-distance and short-distance travel: which trip purposes and modes are critical in the separation of distance categories?

Model Limitations Discussed in Session #1

• Vehicle-ownership model
• For effective VMT estimation
• For effective household-level forecasting
• Economic modeling
• For effective assessments of economic impact
• For effective economic forecasts

Addressing limitations with

NCHRP 735

• Long-distance and rural trip generation:
• Urban/rural distinction: Is this suggesting that urban/rural be treated separately for all trip purposes in the trip generation module? Or just for NHB?
• What is the difference between a trip-based model with "stops" for long-distance trips, and a tour-based model?
• With only passenger-car and commercial truck travel in the model, would trip-distances over 300 miles be required?
• "Rural" trip-making is still greatest on weekdays, but "long-distance" trips are more focused around the weekend-days.
• New set of purposes: business, pleasure, and personal- business?

Addressing limitations with NCHRP 735

• Long-distance and rural trip generation (cont.):
• TAZ characteristics for generation of rural and long-distance trips:
• Population density
• Road density
• Land-use mixture
• Variation in population density
• Use these characteristics instead of households and job categories as regression-equation factors?
• Exact locations of the most popular tourist destinations to do a
• "distance-to-tourist-area" factor to support long-distance trip rates?
• Transferability of parameters
• Find similar states?
• Handle it regionally? Locally?

Other resources for model improvements

• Vehicle-ownership models and their effect on VMT estimation and household-level travel forecasting
• Travel Model Validation and Reasonableness Checking Manual provides guidelines for using vehicle ownership/availability as a validation process:
• The best state of the practice for socioeconomic models is the use of a discrete choice formulation usually a multinomial or ordered response logit model, to simulate the "choice" of the number of vehicles (or workers, children, etc.). This type of model can be readily estimated using data from a household activity/travel survey.
• Good place to start?
• Do we need good household-level income data?

Other resources for model improvements

• Economic modeling
• Economic model with a travel component or travel model with an economic component?
• From NCHRP 735: Small samples and demographic or economic models do not provide the statistical strength to make judgments about capital investment priorities or to understand travelers' decisions based on various price points.

Session #3

Peer Review Schedule

Questions for the peer review panel

Comments from FHWA review

Date: Wednesday July 10

Time: 2:00 - 4:00 PM EST

Agenda: Independent panel meeting convened to assemble comments and feedback

Participants: Expert panel members, TMIP staff

Session #4

Date: Wednesday July 31

Time: 2:00 - 4:00 PM EST

Agenda: Comments and feedback presented by peer review panel to broader group

Participants: All peer review attendees