Almost two years ago, the Spokane Regional Transportation Council (SRTC) was charged with the task of coordinating a major-investment study. The focus of the study was to analyze a number of high-capacity transportation (HCT) alternatives for an east-west corridor through the South Spokane Valley currently in public ownership. Among the alternatives selected were: light-rail transit, exclusive busways, and high-occupancy vehicle (HOV) lanes. In order to complete the study, SRTC undertook an extensive travel demand model enhancement. Working closely with TModel Corporation, we integrated post-distribution mode-choice modeling into the major-investment study process.

The first step was selecting the process for modeling mode-choice. After assessing data and staff requirements, as well as the project timeline, a post-distribution process was chosen. Once the methodology was selected, it was necessary for SRTC to evaluate the adequacy of its regional travel demand model. This part of the process was crucial since the entire post-distribution methodology depended on the distribution of trips in the vehicle model. With the assistance of TModel, enhancements such as multi-point assignment and the integration of land-use type two files became a part of the SRTC model.

After making these improvements and calibrating the vehicle model, the development of a model for each mode began. The light-rail and busway models were created by taking the vehicle model and globally assigning all of the links into pedestrian links. Then secondary transit routes were added using the buffer function in the network graphics editor. Finally, the HCT facility was inserted for each alternative being evaluated. By creating the HCT models this way, transportation analysis zone structure is consistent between modes, transit routes are linked to the HCT facility being studied, and pedestrian features are incorporated into each network. The high-occupancy vehicle model was developed by adding the facility and using link delay coefficients to separate single-occupant trips from access to HOV links.

At this point the modeler needs to make a series of choices for input values that affect model generated data, among them are: utility factors, average vehicle occupancy data, auto and transit captive trips, logit and TMODEL2 exponentiation, and mode capacity characteristics. The SRTC modeling process depended on a number of sources in order to make these decisions: NCHRP 187 (chapter 4), local studies, and other studies from around the country.

The next step entailed developing trip tables for each land-use and forecast year alternative being evaluated. The vehicle trip table was created by factoring to account for auto captive trips, this is done in the TMODEL2 trip table manipulator. As soon as these trip tables are developed, you enter the post-distribution mode-choice module and compare impedances (weighted with your utility factor inputs) between modes by 'skimming trees'. These travel time files determine the mode split. The impedance data is then used to develop mode share percentages that will be used to factor trips between modes. Next, files are created for each mode and trip type being analyzed. Trip tables are created for each mode by trip type. Vehicle occupancy or person occupancy factors are entered in this step.

The trip tables are accumulated by mode in the trip table manipulator and a transit captive trip table is added to light-rail and busway alternatives. The final step is assigning the a mode-choice trip table to its corresponding network to arrive at link volumes and other operational data. TModel developed a multi-point assignment HOV/SOV module in order to meet our modeling needs on this project. We at SRTC appreciate all of the time and effort everyone at TModel has made in order to help make our project a success.

Ed Hayes & Bill Bennett
Spokane Regional Transportation Council