During the later half of the meeting, the peer review panel spent about one hour in an executive session, closed to all other participants of the meeting. The reason for this closed session was to allow panel members to speak freely and openly among themselves while developing formal recommendations. This section details those panel recommendations.
The panel was supportive of APC’s use of Cohort analysis (i.e. birth rate, survival rate, in-out migrate rate) and land use planning (i.e. housing construction permit approach) for developing demographic forecasts. They emphasized the importance of taking into account migration rates since the region is growing rapidly. APC should consider forecasting additional socioeconomic variables in their dataset to support the development of the trip generation rates (i.e. household income, number of household vehicles), as discussed in the next section.
The panel also commended APC on their validation of the 150,000 businesses within the region. They recommended using more detailed employment categories, rather than just retail and non-retail. For example, categorizing employment into Industry, Office, Service and Retail is common in other MPOs, as they all attract individual (and commercial vehicle) trips by purpose differently.
The panel recommended revising the trip rates that are used in the model. Ideally, the trip rates would be calculated from a new household travel survey within the region. However, if a household travel survey is not conducted, then the panel recommended borrowing trip rates from another region in Louisiana or surrounding states that has similar land-use and travel patterns (i.e. low density area dominated by auto travel).
The panel also recommended generating person trips at the household level. In addition, the model should include trip rates by more demographic characteristics besides household size. For example, household vehicles and income level are also correlated with trip rates.
The panel recommends moving to a doubly constrained gravity model for home based work (HBW) trips where there is a fixed set of workers and a fixed set of work locations. They also recommended that the gravity model be recalibrated using congested travel time skims, rather than free flow skims. K-factors should be validated to count data by time-of-day. At the district-to-district or county-to-county level, HBW trip distribution should be compared to Census Transportation Planning Products (CTPP) Journey-to-Work distributions. In addition, the model should be analyzed by sub-area distribution (e.g., at district and/or county level), in addition to trip length (minutes) and distance calibration by trip purpose and by time-of-day.
If a new household travel survey is conducted, then APC should consider replacing the gravity model with a destination choice model.
The panel stressed the importance of moving from a daily model to a peak directional model, and thus the importance of developing separate skims by time-of-day. Hourly time-of-day factors should be used to split the trips by time-of-day before assignment. The panel recommended that APC research the peak periods in their region (i.e. via examination of counts and professional judgment) and use that to determine the time periods for each set of skims. Feedback loops should be implemented between assignment and trip distribution to ensure consistency between these model components. The panel recommends following the guidance discussed in “Traffic Assignment and Feedback Research to Support Improved Travel Forecasting” prepared for FTA by Caliper Corporation.
As mentioned above, it is critical for APC to develop separate congested skims by time-of-day. The panel also suggested that APC explore different assignment algorithm approaches. They recommended improving the volume delay function to match observed speeds at different levels of traffic volume. APC can “play around with” the volume delay parameters. Capacity should not differ by time-of-day (i.e., jam capacity should be used for all time periods).
The panel also recommended the following improvements to the highway network:
The panel has put together recommendations for alternative approaches to transit planning and forecasting that do not rely on the regional model:
Similar to transit planning, the panel recommends that APC take an off-model approach to non-motorized planning. The panel recommends beginning the analysis by examining the existing household travel survey to gain an understanding of the existing local non-motorized travel patterns and the demographics of the individuals making these types of trips. The panel provided recommendations for alternative approaches to non-motorized trip planning and forecasting at the TAZ-level, without incorporating a formal non-motorized mode choice component directly within the regional model:
The panel also recommended the following reports and guidelines on non-motorized planning:
The peer review panel made additional general recommendations to APC related to improvements of their travel demand model, such as placing additional emphasis on validation, including a comparison of modeled volumes to time of day directional traffic counts and class counts, by federal functional class and volume group; conducting reasonableness and sensitivity checks; and testing the model on observed “forecast” year (i.e. Year 2015).
They also suggested implementing all model improvements by hiring a travel demand modeling consultant that has a strong reputation for developing state-of-the practice travel demand models, through a competitive procurement process. Selection criteria should be weighted toward merit and what they will deliver for the cost. The consultant should discuss pros and cons of different potential approaches to model improvement before deciding on improvements to implement and data to collect, if applicable.
The panel recommended the following reports for use in guiding their model update efforts:
