Bike Demand Model

The Utah Bike Demand Model estimates bicycle trips between microzones and assigns them to the bike network.

The model builds on the micromobility toolset from RSG, applied to Utah data. It estimates how many bicycle trips happen between small zones and which parts of the bike network they use.

How it works

flowchart LR
  A[Convert<br/>network] --> C[Trip<br/>generation]
  B[Create<br/>microzones] --> C
  C --> D[Trip<br/>distribution] --> E[Trip<br/>assignment] --> F[Post-process<br/>outputs]

1. Convert the network. Prepares the bike network for the model.
2. Create microzones. Builds the small zones the model works in (see below).
3. Run the bike demand model. Trip generation, trip distribution and trip assignment.
4. Post-process outputs. Produces monthly bike volume across links, microzone-to-microzone trip summaries and microzone centroid nodes, used for quality checks.

Microzones

Microzones, or micro analysis zones (MAZ), are smaller than the traffic analysis zones used by the Travel Demand Model. They are typically about the size of a census block and group areas with similar land use, such as residential, commercial, industrial, education, parks and open space. They also follow Wasatch Choice Centers, and buildings within a microzone tend to share similar points of access to the network. This finer scale suits bicycle travel, where short trips matter.

Running the model

The model needs at least 32 GB of RAM and Python 3.

pip install git+https://github.com/rsginc/micromobility_toolset
python Create_Microzones.py        # if microzone data is needed
python Convert_MM_Network.py       # if network data is needed
python utah_bike_demand_model.py   # run the full model

Add --name generate_demand or --name assign_demand to run one step, or --sample 100 to run only the first 100 microzones. Outputs are saved in the repository’s logs folder. See the full instructions.

Releases

The Bike Demand Model does not currently publish numbered releases. The repository’s latest version is the current model.

Documentation

• Source code and instructions
• Model design specification (PDF, dated January 28, 2021), in the repository’s Model_Design folder
• Design meeting notes
• Bike model coefficients tracker

Related bike work

• Bikeway by city compares bicycle infrastructure with the roadway network for each city.
• A feature-building script combines UGRC roads and trails data into existing and planned bike facility layers, choosing the best bike feature and its side of the road.
• A latent bike demand GIS tool is also kept in the catalog below.

Components

Entries marked Internal need WFRC GitHub access.