City of Bellevue, City of Kirkland, City of Redmond (BKR) Peer Review

3.0 Expectations of a New BKR Model

3.1 Existing Model

The current BKR model is a four-step, trip-based model modeled in EMME. It includes a fixed rate cross classification trip generation model, a gravity model for trip distribution, and a mode choice model that focuses on vehicle travel, with the following mode options: Single occupant vehicle (SOV), high occupancy vehicle with two occupants (HOV 2), high occupancy vehicle with three or more occupants (HOV 3+), and park and ride. Non-motorized modes are not explicitly represented. Highway assignment includes a multi-class equilibrium assignment using EMME. The EMME model results, after post processing using software developed in house, are frequently used to support traffic operation analysis in conjunction with Dynameq, a dynamic traffic assignment model, and Synchro, a traffic signal optimization and intersection analysis tool.

The BKR model is closely tied to the Puget Sound Regional Council (PSRC) regional model. It covers the same area as the PSRC regional model with much more refined traffic analysis zones (TAZ) and transportation networks within the three city area. TAZs outside the three city area are aggregated and treated as external zones. Trip ends for the external zones are taken directly from the PSRC model, which is combined with trip ends from the zones within the BKR area for trip distribution, mode split and traffic assignment.

3.2 Future Model Update

Several new developments have prompted BKR to initiate a major update to their model. First, the City of Kirkland and City of Redmond have either completed or are in the process of migrating from vehicle oriented performance measurement to multimodal performance measurement. The City of Bellevue is poised to develop multimodal measures as well. The existing vehicle focused model cannot effectively support multimodal analysis and planning functions. BKR is interested in having a model that can more explicitly model non-motorized travel.

Second, the BKR model is often used to assess the transportation impacts of mixed use developments and land use scenarios with different densities. The existing BKR model is not sensitive to changes in certain aspects of these scenarios. For example, the current model assumes that similar households located in downtown Bellevue generate the same number of trips as households located in the fringe areas of Kirkland or Redmond. This “static” and “fixed” trip generation approach may not accurately reflect the complex and dynamic nature of the human travel decision making process in the real world.

Third, newly collected survey data and recent advances in the PSRC regional model make it a logical time to develop the next BKR model. The City of Bellevue joined with PSRC to conduct a new household activity survey. The survey provided new information on regional travel behavior. PSRC has completed major updates to its 4-step model and has recently released their activity based model for use in application. BKR would like to leverage these successes in the development of the new model.

BKR identified four primary expectations for the new BKR model:

1. The model must have a multimodal focus and have the ability to generate multimodal (i.e. highway, transit, bike) performance metrics.
2. The model must be sensitive to changes in land use, accessibility, transportation policies, and provide output that better supports traffic operational analysis. For example:
   1. The model should have the ability to examine mixed use and high density developments within the urban area,
   2. The model should be sensitive to change in access or cost by different modes,
   3. The model should have the ability to test different tolling and parking surcharges, and
   4. The BKR model should provide improved input to the BKR region-wide dynamic traffic assignment (DTA) model in Dynameq to better support a variety of traffic operational analyses.
3. The results of the model should be intuitive and repeatable. The results should make sense and the results should be easy to explain and interpret. The model will be routinely used in development impact review and concurrency analyses. Thus, it is important that the new model is capable of performing consistently and that the results can be repeated.
4. The model should be practical. It should have reasonable run times (i.e. take no longer than overnight to complete), should not be too complex to develop and calibrate, and should be easy to maintain and update.

3.3 Goals for the Current Peer Review

The primary goal of the Peer Review was to review the draft model design that was developed by BKR with the help of RSG by:

• Reviewing model components, their subcomponents and data requirements,
• Identifying opportunities for improvement,
• Discussing uncertainties of inputs and assumptions and advise on how they can be managed, and
• Estimating resources required for implementation.

In reviewing the draft model design, BKR desired for the panelists to focus their attention on answering the following question: “What are proven, best practice modeling techniques that can be cost effectively incorporated into the next BKR model?”