TRM Information

The Triangle Regional Model (TRM) is a travel demand forecasting model that is specifically tuned to the Triangle. The model is a mathematical representation of the travel demand and capacity for every road in the region. The travel demand is represented using highway and transit networks. The highway network represents all major roads in the Triangle, while the transit network includes all public transportation services. In addition to the transportation networks, the TRM also incorporates land activity data for each transportation analysis zone (TAZ). The TAZs and transportation networks are used together to create a “four-step” trip-based travel demand model.

The “four-step” model is the demand model that is used most frequently by MPOs. The four steps of the “four-step” model are:

  • Trip generation
  • Trip distribution
  • Mode choice
  • Trip assignment


Step 1: Trip generation

The first step of developing the TRM is to estimate the number of daily trips that take place in the Triangle. Each trip begins in one TAZ and ends in another. The number of trips to and from each TAZ is calculated using some assumptions about the number and types of trips the households in the TAZ are expected to take on a standard day. Step 1 produces an estimate of the number of motorized trips starting and ending in each TAZ.

The TRM uses eight trip purposes: Home – based work, Home – based shop, Home – based other, Non – home – based work, Non – home – based other, Medium truck, Heavy – duty truck, and Non – freight commercial. These trip ends assume land activity, residential, and employment characteristics, producing “production” and “attraction” zones.


Step 2: Trip distribution

In step two, the trips generated in step one are linked geographically into complete trips between a production zone, where the trip originates, and an attraction zone, where the trip ends. The connectors of a complete trip can be located within the same TAZ, adjacent TAZs, or between TAZs that are some distance apart.

When modeling these trips, it is assumed that the closer a production and attraction zone are, the more likely the trip will occur. Therefore, most trips generated in a zone will be attracted to nearby attractions, rather than attractions that are further away. The exception to this assumption is commuter trips, where people are more willing to travel longer distances for work than they normally would for commercial trips.

Once this modeling is complete, a set of motorized person trip tables are made to and from each TAZ in the Triangle for each type of trip.


Step 3: Mode choice

Step three uses the nested choice structure shown below to assign a mode type for each trip. The model makes assumptions about mode choices based on the relative availability of each mode. Factors in these assumptions include: travel time, cost, accessibility, automobile ownership, and group travel incentives (such as carpool lanes). The mode choice factors are arrayed in an equation that weighs the probability of each traveler selecting each mode, given the characteristics of both the mode and the traveler.

Step 4: Trip assignment

The fourth and final step of the TRM is determining what route is taken for each trip. The first step of trip assignment is building least – cost paths between all zone pairs in the Triangle. There are 2857 TAZs in the Triangle, resulting in 8.2 million least – cost paths. Next, depending on the travel mode, trips are assigned to either the highway network or the transit network.

For highway path building and trip assignment, the effects of congestion are considered using capacity and congestion metrics that differ per road segment. Once a predefined number of trips have started on a road, causing congestion, the model re-evaluates the travel times between TAZs, and re-assigns trips as necessary to establish an equilibrium of transit (see the Four – Step Regional Forecasting Model graphic).

The TRM includes projected highway and transit networks for 2025, 2035, and 2045. The highway network includes approximately 22,000 road segments, and the transit network includes approximately 300 transit routes. As a result, each run of the model is quite costly in computing power and time.

Source: Metropolitan Washington Council of Governments