triangle-modeling-and-analytics/trmg2

Purpose and Entry Points

TRMG2 is the Triangle Regional Model’s second-generation regional travel-demand model. Its executable workflow is a sequenced set of Caliper/TransCAD-style resource scripts, compiled through src/_TRMCompile.lst, with model/scenario definitions in trmg2.model and trmg2.scenarios. The published technical documentation is rendered under docs/, notably docs/index.Rmd and docs/index.html.

Scenario, Network, and Socioeconomic Foundation

Scenario creation establishes a model-year network and inputs from baseline master data plus project lists. src/01 - CreateScenario.rsc consumes year-specific roadway/transit changes under scenarios/2035/ (and corresponding model years) and network assets under master/networks/. Socioeconomic data and TAZ geography are versioned separately in master/sedata/ and master/tazs/; these are core boundaries between scenario inputs, geography, and demand modeling.

Supply Preparation and Accessibility

Network calculations and accessibility prepare the impedance and land-use measures required by downstream demand models. src/02 - Network Calculations.rsc precedes src/03 - Accessibility.rsc; accessibility parameters and outputs are represented in master/accessibility/accessibilities.csv and master/accessibility/attraction_rates.csv. Skimming is an explicit later supply-demand feedback boundary in src/10 - Skimming.rsc, producing travel impedances after network conditions are available.

Resident Travel Demand Pipeline

The resident model follows a conventional person/household-to-trip pipeline: synthetic population, productions, nonmotorized processing, aggregation, time-of-day, parking, mode choice, and destination choice. Principal stages are src/04 - PopulationSynthesis.rsc, src/05 - Resident Productions.rsc, src/08 - Time of Day.rsc, src/12 - Mode Choice.rsc, and src/13 - Destination Choice.rsc. Their calibrated model inputs are versioned by purpose under master/resident/, including population seeds, generation rates, mode coefficients, destination-choice size terms, and time-of-day factors.

Specialized Demand Submodels

Separate scripts add demand not handled by the resident home-based pipeline: non-home-based travel (src/14 - NHB Models.rsc), airport trips (src/Airport.rsc), commercial vehicles (src/Commercial Vehicles.rsc), external/internal-external travel (src/External Models.rsc), and university travel (src/University.rsc). These models maintain distinct calibration/configuration boundaries in master/airport/, master/cv/, master/external/, and master/university/, while their R-based estimation/preprocessing artifacts and validation outputs reside in docs/R/ and docs/data/output/.

Assignment and Feedback Outputs

Demand is converted into assignment-ready matrices by src/16 - Create Assignment Matrices.rsc, then loaded onto roadway and transit networks by src/17 - Roadway Assignment.rsc and src/18 - Transit Assignment.rsc. src/19 - Summaries.rsc is the terminal reporting stage. Assignment depends on Caliper binary network, route, matrix, and scenario formats in master/networks/, rather than a database or web service.

R Documentation and Calibration Boundary

R is used primarily for data preparation, estimation, calibration, and static documentation—not as the runtime orchestrator. Scripts such as docs/R/create_dc_est_tables.R, docs/R/estimate_nhb.R, and docs/R/university_productions.R transform survey, census, GIS, and observed-count inputs under docs/data/input/ into calibrated tables and diagnostic outputs under docs/data/output/. The R environment is pinned with docs/renv.lock; rendered HTML documentation is deployable as a static site.