Model Description
Overview of the Gen3/Ver. 1.1.0 model, including development phases, platform, and distinguishing features.
Read More
Current Model Documentation
Access current user guides, transmittal materials, network documentation, and calibration or validation references.
Read More
Past Model Documentation
Review archived Phase 1 and Phase 2 Gen3 reports, design plans, calibration documents, and sensitivity testing results.
Read More
As of 2026, COG/TPB currently has two production-use travel models: the trip-based Gen2/Ver. 2.4.6 Travel Model and the activity-based Gen3/Ver. 1.1.0 Travel Model. This page focuses on the Gen3/Ver. 1.1.0 Model. The Gen3 Model has not yet been used for the air quality conformity (AQC) analysis of an LRTP and thus has not been adopted by the TPB. Instead, in spring of 2026, it has been deemed ready for production use based on staff evaluation. More information about the model can be found on the Data Requests webpage.
Following a multi-year strategic plan that guides the development of the TPB’s travel forecasting methods, COG/TPB staff set out in late 2019 to develop a next-generation, activity-based travel demand model. In the next six years, the Gen3 Model was developed in three phases. Phase 1 created a prototype model for testing by February 2022. Phase 2, which finished in March 2024, developed a model for production use (Version 1.0.0). Phase 3 evaluated and further improved the model’s readiness for production work. Consequently, the Gen3 Model was released for beta testing (Version 1.0.5) in November 2025, and an official model was released in the spring of 2026 and was recommended by COG/TPB staff for production use (Version 1.1.0).
The Gen3/Ver. 1.1.0 Model is implemented in the ActivitySim software platform. ActivitySim is an open platform for activity-based modeling. It is the product of a collaborative activity-based travel behavior modeling software development project that is supported both financially and managerially by a consortium of public-sector agencies. As of 2025, the administrative agent for the consortium is the Zephyr Foundation, and there are 14 member agencies in the consortium, including COG/TPB.
The Gen3 Model was implemented based on a fully functional activity-based model that was originally designed for the San Francisco Bay area’s Metropolitan Transportation Commission (MTC) and later deployed for Southeast Michigan Council of Governments (SEMCOG, the MPO for Detroit). The model system relies on logit choice models to represent travel decisions (how frequently to travel, where to travel, by what mode, etc.) and was designed to achieve behavioral realism within a practical system of components.
Compared to the Gen2 Travel Model, the TPB's Gen3/Ver. 1.1.0 Model has the following distinguishing features:
- A disaggregate, activity-based, tour-based travel model, estimated, calibrated and validated to observed conditions in 2018.
- Treatment of time:
- Temporal resolution of daily activity scheduling in ActivitySim: Half-hour
- Temporal resolution of highway/transit network simulation (path skimming and assignment): Four aggregate time-of-day periods - AM Peak, Midday, PM Peak, and Night-Time (Evening/Early AM)
- Representative population: The Gen3 Model features a representative population (or synthetic population) that is generated in PopulationSim. The representative population is used in the model to simulate mobility and travel choices for each decision maker in the population according to contextual probability distributions.
- Demand micro-simulation: The Gen3 Model simulates the travel-related choices made by residents in a series of Monte Carlo simulations in ActivitySim, which replaces the trip-based demand model components (trip generation, trip distribution, mode choice and time-of-day choice) in the Gen2 Model and enables new model sensitivities in terms of changes in household and person variables, network levels of service by time-of-day, telecommuting, transit subsidies, ride-hail modes and autonomous vehicles, etc.
- Toll setting:
- A heuristic toll-setting procedure is included as part of the model. The model user has the option to turn it on or off.
- The procedure is the same as that used in the Gen2 modeling processes, but the resulting toll rates are different due to different travel demand patterns in the two models.
- This toll setting process is primarily designed to provide “planning-grade” estimates of variably priced tolls where tolls could serve as an indicator of bottlenecks or lack of SOV demand in the freeway system. The toll setting output has never been calibrated or validated to observed data, and thus, it should not be viewed as a realistic representation of the tolling system on the ground. TPB staff strongly recommend against the practice of directly using the outputs from the COG/TPB toll setting process in any type of operation-level studies.
- Representation of external travel
- External-internal transit travel: Used transit on-board survey data to generate transit trip tables representing non-resident travel into and out of modeled area.
- External-internal auto travel: Used a new methodology to estimate internal-external and external-internal auto person trips.
- Transit modeling procedures
- Transit skimming: Replaced Cube TRNBUILD with Cube Public Transport (PT) module to generate multi-path transit skims with post-processing to ensure transit paths are more consistent with mode choice model alternatives.
- Transit assignment: Replaced Cube TRNBUILD with Cube Public Transport (PT) module to assign transit O-D trips in Origin-Destination format in four time-of-day periods.
Click here for information on how to request models and data.
The model user’s guide and the transmittal package memo for the Gen3/Ver. 1.1.0 Travel Model can be found below:
The highway and transit networks report can be found in the following link:
The Gen3 Model was calibrated and validated to year-2018 conditions. The following report documents the model calibration and validation results for the Gen3/Ver. 1.0.4 Model.
The report was amended by two memos, which document the additional calibration results and updated validation results, respectively, for the Gen3/Ver. 1.1.0 Model.
Gen3, Phase 1, Model Documentation
RSG and Baseline Mobility Group, Inc. Gen3 Model Design Plan. Metropolitan Washington Council of Governments, National Capital Region Transportation Planning Board, July 2, 2020
RSG. MWCOG Population Synthesizer. Final Report. Metropolitan Washington Council of Governments, National Capital Region Transportation Planning Board, August 4, 2021
RSG. Gen3 Data Development. Washington, D.C.: Metropolitan Washington Council of Governments, National Capital Region Transportation Planning Board, December 29, 2021
Tour Mode and Destination Choice Model Estimation. Washington, D.C.: Metropolitan Washington Council of Governments, National Capital Region Transportation Planning Board, January 19, 2022
RSG and Baseline Mobility Group. Gen3 Model Phase 1 Calibration and Validation. Washington, D.C.: Metropolitan Washington Council of Governments, National Capital Region Transportation Planning Board, February 1, 2022
Gen3 Model Phase 1 Sensitivity Testing Results. Washington, D.C.: Metropolitan Washington Council of Governments, National Capital Region Transportation Planning Board, March 29, 2022
Gen3, Phase 2, Model Documentation
RSG, Baseline Mobility Group, and Metropolitan Washington Council of Governments. “Gen3 Model User Guide.” Washington, D.C.: Metropolitan Washington Council of Governments, National Capital Region Transportation Planning Board, January 31, 2024.
RSG and Baseline Mobility Group. “Gen3 Model Calibration and Validation Report.” Final Report. Washington, D.C.: Metropolitan Washington Council of Governments, National Capital Region Transportation Planning Board, February 7, 2024.
RSG, Baseline Mobility Group, and Metropolitan Washington Council of Governments. “Gen3 Model Phase 2 Sensitivity Testing Results.” Final Report. Washington, D.C.: Metropolitan Washington Council of Governments, National Capital Region Transportation Planning Board, May 22, 2024.
Page updated 4/16/2026.
↑ Back to Top