Assignment
Highway Assignment
For a more basic introduction to highway assignment, see the Travel Forecasting Resource page on the topic. In short, roadway assignment is where the predicted trips and roadway network are combined to predict volumes and travel times.
TRMG2 uses a static, multi-class assignment with the following basic vehicle classes:
- SOV: single occupancy vehicles
- HOV2: vehicles with two occupants
- HOV3+: vehicles with three or more occupants
- CV: commercial vehicles
- SUT: single-unit trucks
- MUT: multi-unit trucks
In addition, the classes are further segmented by value of time. This is done to accurately capture toll road usage. Finally, vehicles are assigned separately by time period. More details are provided in the sections below.
Assignment by time period
A defining feature of TRMG2 is the independent treatment of each time period. Traditionally, travel models used the AM period to represent both AM and PM periods. The primary benefit of this approach was reduced computational burden, and as long as the AM and PM periods were nearly symmetrical, errors introduced by this simplification were small.
In the Triangle, the AM and PM periods are not symmetrical. The trips-in-motion analysis performed using the travel survey shows a large disparity between the peaks driven largely by non-work travel. The PM peak lasts longer, and places more demand on the transportation system. In order to capture these differences, model estimation and application must treat the PM peak separately.
Matrix creation
TRMG2 has models to predict demand for many markets including residents, commercial vehicles, university students, and external travelers. The output of these models is combined into a single trip matrix for each time period. During this process, directional and occupancy factors are applied to arrive at vehicle trip matrices in Origin-Destination format (click here for resident factors).
A separate note about the “auto pay” mode (Uber, Lyft, etc.) is warranted. When converting person trips to vehicle trips, a person traveling alone is a single vehicle trip (1 person trip = 1 vehicle trip). However, by definition, they have a driver. This means they are eligible for any HOV lanes the region might construct in the future. For this reason, a person traveling alone is added to the HOV2 matrix while a person traveling with one or more family members is added to HOV3+.
Value of time
Once trips are collapsed into the primary vehicle classes, they are assigned values of time. The base values were borrowed from the NC Statewide Model, but were then calibrated to match NC 540 traffic counts.
| Name | VOT ($/hr) |
|---|---|
| Autos/CV (PK) | 17.11 |
| Autos/CV (OP) | 17.00 |
| SUT | 26.82 |
| MUT | 40.48 |
| Calibration Factor | 2.00 |
Assignment parameters
Relevant assignment details are listed below:
- Algorithm: N-conjugate Frank-Wolfe
- Volume-Delay Function (VDF): BPR
- Relative Gap: .0001 (1e-4)
- Maximum Iterations: 500
VDF parameters
The table below shows the alpha and beta values used for the BPR VDF.
Alphas
| HCMType | Downtown | Rural | Suburban | Urban |
|---|---|---|---|---|
| Arterial | 1.15 | 1.15 | 1.15 | 1.15 |
| Collector | 1.60 | 1.75 | 1.60 | 1.60 |
| Freeway | 0.90 | 0.65 | 0.65 | 0.65 |
| Local | 2.00 | 4.00 | 1.90 | 1.90 |
| MajorArterial | 1.00 | 1.00 | 1.00 | 1.00 |
| MajorCollector | 1.25 | 1.25 | 1.25 | 1.25 |
| MLHighway | 0.90 | 0.70 | 0.70 | 0.80 |
| Superstreet | 0.75 | 0.75 | 0.75 | 0.75 |
| TLHighway | 1.15 | 1.00 | 1.15 | 1.15 |
Betas
| HCMType | Betas |
|---|---|
| Arterial | 5 |
| Collector | 4 |
| Freeway | 7 |
| Local | 3 |
| MajorArterial | 5 |
| MajorCollector | 4 |
| MLHighway | 5 |
| Superstreet | 4 |
| TLHighway | 5 |
Transit Assignment
TRMG2 uses Caliper’s Pathfinder transit assignment algorithm. This is the same algorithm used for skimming and assignment uses identical networks. Separate assignments are done for each combination of access mode (e.g. walk, PNR, KNR) and primary mode (e.g. local bus, express bus, or rail in future scenarios). This ensures consistency between mode choice and assignment. As with roadway assignment, each time period is also assigned separately. Transit assignment is not part of model feedback, but happens after highway assignment has converged.
TransCAD GIS Software, 2022