The panel convened separately in Session #3 to discuss specific recommendations, feedback and model development goals. Following this independent panel caucus, the panel presented a summary of their recommendations during Session #4 to CCMPO staff and the other attendees at the peer review.
The panel utilized the stated modeling priorities as a template for making their final recommendations. Each panel member was charged with developing content for a specific topic area following a 10-15 minute discussion of each of the short and long term priorities among all the panel members in Session #3.
Modeling mixed land uses to support additional non-motorized and transit use
The peer review panel recommended that CCMPO refer to two separate studies pertaining to the modeling of mixed land uses to support additional non-motorized and transit use.
The first is the "Statistical Analysis of Urban Design Variables and Their Use in Travel Demand Models" prepared by Portland Metro and the Oregon Department of Transportation in November 2003. The executive summary states:
"Many travel demand models do not account for land use mixing and urban design effects. The purpose of this study is to further the understanding of how aspects of urban design influence transportation choices. This research identifies where it is important for models to account for urban design issues and where there would be minimal or no effect. It is intended to show how much land use change is necessary to significantly affect travel behavior."
The second reference cited by the panel was the report entitled "Non-Vehicle Accessibility in the Atlanta Region" (D'Onofrio and Kim). The report identifies a number of non-motorized performance measures that can be used to assist in spatially assessing current conditions and establishing project prioritization. Specifically, the Potential Walking Demand Measure and the MultiModal Accessibility Measure are identified, described, and illustrated.
Finally, the panel recommended that robust and simple mixed used variables like the ones present in the pre-distribution non-motorized mode choice model should be considered for the post-distribution motorized mode choice model.
Improving air quality and climate change modeling
The panel identified a number of resources pertaining to the Motor Vehicle Emissions Simulator (MOVES) emissions model including the resources available on the Office of Transportation and Air Quality (OTAQ) website. The panel also recommended that CCMPO and State agencies pursue formalized MOVES training with the software just prior to beginning the preparation of inputs and testing.
The panel also stressed the importance of building partnerships and "making friends" with other critical agencies such as State environmental planning agencies, VTrans, and the regional EPA and FHWA Division offices. The panel also recommended reaching out to other MPOs in New England who have or will soon make the transition to MOVES to get their feedback and insights on their own experiences.
The panel highlighted some of the data requirements and information required by MOVES and the additional spatial, temporal and seasonal detail that is required.
Finally, the panel described some of the Sketch Planning tools which are now available that can be used to quickly evaluate different alternatives. Examples include Rapid Fire, Envision Tomorrow, and GreenSTEP, among others.
Reviewing sensitivity to increased travel costs (e.g. gasoline) in the model
The panel was sympathetic to the question often asked by stakeholders about whether or not the regional travel model is responsive to the cost of gasoline. Especially in light of current increases in the cost of gasoline as $4 per gallon is approached and will perhaps be exceeded.
The panel stressed the short and long term effects associated with increases in travel costs such as gasoline. Modal shifts may be short-lived while long term effects could include change in urban form. Form changes might include people living closer to their jobs or jobs moving to the suburbs.
In addition, there are traveler responses where a household might elect to purchase a more fuel efficient vehicle as well as government and/or industry responses based on more stringent standards and/or consumer demands.
All of these factors make it very challenging to capture these effects in traditional regional demand models. Auto operating cost is a typical model application variable that assumes a per mile cost such as $0.12/mile which is applied in the current CCMPO travel model. However, the panel presented research showing the auto operating cost as being fairly stable historically.
The panel pointed to the emerging research of household incomes and household travel budgets. Disposable income drives travel behavior and choices when costs increase. Households with less disposable income will be forced to make notable changes while households with more disposable income will make fewer behavorial changes.
The panel concluded that there is no easy fix and sound theoretical approaches to incorporating these factors into the existing travel model would be expensive. In a nutshell, these tools do about as good a job as we can expect until more is learned from the research.
Refinements to 1996 FHWA Quick Response Freight Manual truck modeling
The panel acknowledged that the QRFM is generally the most widely used and is generally thought to be acceptable.
The panel identified a number of resources which they suggested CCMPO review and consider (NCHRP 384, NCHRP 570, NCFRP 8, NCHRP 594, NCHRP 410).
Visualization techniques and performance measures for better communication to stakeholders
The panel agreed that presenting performance measures and visualizations of model output data to stakeholders can be very challenging, but is also critically important. Effective communication of key issues often requires that simple and clear graphics that make intuitive sense be developed.
The panel presented a number of examples from their own project work and modeling experience as reference for CCMPO. The samples included tabular data, maps, charts, and other visualizations of travel model input and output data.
Recommendations regarding interaction/integration between model and microsimulation tools
The panel recommended that CCMPO consider Dynamic Traffic Assignment (DTA) for inclusion in the regional travel demand model citing the increased temporal resolution of minute-by-minute link flows, queues and congested speeds. A discussion of experienced travel time and time-dependent shortest paths which are a hallmark of DTA were discussed.
The panel cautioned that implementing DTA in a regional context, even in a small to mid-size geography like CCMPO will not be a trivial exercise. The panel believes the benefits are significant and therefore warrant the investment.
The panel recommended that CCMPO consider hiring a contractor for the initial DTA implementation work with a focus on network preparation, establishing the flow models, and validation. Once in place, the panel felt that CCMPO could easily continue to work with and manage the DTA implementation moving forward.
Finally, the panel suggested that DTA might be powerful enough in and of itself and provide answers to the kinds of policy questions being asked by CCMPO and the region's stakeholders to potentially eliminate the need for microsimulation.
The region might also benefit from the lessons learned and experiences gained from past and current research being conducted with TRANSIMS in the Burlington, Vermont region. A Track 1 TRANSIMS implementation was conducted in the Burlington region in 2006-2007 and current work as part of the SHRP2-C10A project is seeking to integrate an activity-based demand model and the TRANSIMS microsimulator using Burlington as a test bed.
Transition to activity-based demand models
The panel strongly recommends that CCMPO transition to activity/tour based demand models in the long-term because of the increased policy sensitivities. The panel felt CCMPO is in a very unique position in light of the fact that it may be able to pivot off research work currently being conducted by RSG as part of the SHRP2-C10A project. In this research work being funded by the Transportation Research Board (TRB), the DaySim activity-based demand model has been implemented in Burlington. In many ways, CCMPO would not be starting from scratch since many of the required data inputs would be the same even in the event an alternative activity-based software package was preferred.
The panel understands the concerns of CCMPO staff when it comes to the transition to an activity-based disaggregate model framework. The increased complexity of the model system and the additional time and resources needed to develop and maintain future year inputs are the primary concerns. The panel suggested that the implementation need not be overly complex and stressed that once the model system is configured and built any additional costs to maintain and run the model system would be minimal as compared to the existing aggregate trip-based model.
Expanding the model geography and integration with State-wide model
In the existing CCMPO regional travel model, external trips account for 12 percent of the total demand and 40 percent of the total network vehicle miles of travel. As such, expanding the model boundary outside of Chittenden County has been considered for some time.
The panel recommended that now would be a good time to expand the model geography with the release of the 2010 Census data products. The panel also recommended that concave areas in the model geography should be eliminated, and that traffic analysis zone boundaries need not conform to Census zone boundaries.
The panel also suggested that CCMPO consider what the eventual non-attainment area might be in the event CCMPO becomes non-attainment for ozone and expand the model boundary to include the affected areas.
Finally, the panel provided reference material pertaining to the expansion of travel model systems including a report entitled, "Update and Expansion of Lima Travel Demand Model To Cover All of Allen County" by the Ohio Department of Transportation that provides technical documentation on the geographic expansion of the Lima, Ohio travel model.
In terms of integration with the Statewide Travel Model the panel recommended that it is important to determine what data should and can be extracted from the broader model, whether that is commercial vehicle trip tables, freight trip tables, and/or external trip tables.
The VTrans Statewide Travel Model is currently being reviewed and updated by the University of Vermont Transportation Research Center (TRC). The panel recommended that CCMPO staff stay involved in the Statewide model development efforts especially when it comes to defining zone geographies.
Finally, the panel suggested that in general, freight and commodity flow modeling is typically handled within the statewide model as freight movement is national and in the case of Vermont international in scope. When and if the Statewide model can provide useful data related to freight movements these should be incorporated into the CCMPO regional travel model where possible.
Recommendations regarding land use forecasting tools and existing LUAM
The panel recommended that CCMPO perform a detailed review of the LUAM and identify the strengths and weaknesses of the existing land use allocation model to help narrow the research into new, better tools and more complex tools.
CCMPO also must determine 'needs' and 'wants' for the land use allocation model. Are tools needed to help determine population, household and employment control totals? Are tools needed to allocate pre-determined demographic controls totals? Are tools needed to develop multiple land use scenarios for comparisons? The answers to these kinds of questions will help determine what kind of land use modeling tool should ultimately be selected.
The panel identified a number of resources for CCMPO to review including "A Study of Alternative Land Use Forecasting Models" (Zhao, 2006) prepared for Florida DOT. The panel also highlighted two upcoming TMIP webinars dedicated to the I-PLACES and PECAS modeling frameworks.
Finally, the panel members described the tools and processes in place at their own planning organizations. COMPASS uses committee and workgroups of city planners, developers, and public works employees to ensure as much good local knowledge as possible. Tools utilized by COMPASS in their own land-use allocation process include GIS, Excel, and UPlan. The process in Ohio is also generally demographer based with an Excel macro utilized to allocate development.
To our knowledge this is the first TMIP Peer Review which has been conducted remotely via web-conferencing instead of as an in-person on-site meeting. CCMPO staff was mindful of the resources and scheduling difficulties associated with an in-person peer review process. Therefore, the agency proposed initiating a 'virtual' peer review which was completed over the course of four online webinar sessions. The intent was to remove potential barriers to participation for panelists, decrease costs, and still provide thoughtful reviews to assist in determining future model investments.
The on-line sessions were hosted by FHWA using Adobe Connect Pro. In addition, the visual and audio portions of the web-meetings were recorded by support staff at the Volpe Center. No information technology issues or snafus were encountered. It was very easy to upload content to the web-meeting room such as the powerpoint slides, as well as supporting documentation such as Word and PDF documents. Panelists did not encounter any difficulties with the conference calling number dial-ins and the fidelity of the audio recordings was surprisingly good.
The meetings were scheduled using Doodle Poll results in order to select meeting dates and times most convenient for the panel members and all other meeting participants.
The panel agreed that something is lost when face-to-face meetings are eliminated. The most compelling advantage is that the agency was able to invite panelists that might not have likely been able to participate in a more typical review format given time and resources constraints. The panelists agreed that four roughly two-hour web-meetings were more than sufficient to cover the material and enabled them to make substantive comments and recommendations.
The panel concluded that the virtual peer review format is a very good approach for small and mid-sized MPOs with relatively simple and straight forward travel demand model implementations. The panel cautioned though that large regions and those with more complex model systems should likely still opt for the traditional on-site in-person TMIP peer review format.
