In douglascm/trafficr: Traffic Assignment Methods

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)
library(trafficr)

Traffic Assignment Package

trafficr is a package for static traffic assignment R code. Currently, the program can solve the static traffic assignment problem using user equilibrium (UE) for the city network in urban traffic assignment. The solution is achieved using the Frank-Wolfe algorithm. Package is based on dodgr package for all-or-nothing assignment, route and time data.

How to run traffic assignment

Following steps explain the setup and how to run the procedures and extract flows and maps.

1. Install the R package with:

devtools::install_github("douglascm/trafficr")

Package devtools is required for the installation, install with install.packages('devtools')

2. Data Preparation :

Import into the current environment all three required datasets:

• demand tabular data containing demand data for each origin and destination:

demand <- fread("path to file")

- Required columns:
    - orig: origin id (integer)
    - dest: destination id (integer)

head(sioux_demand)

• network sf object data containing road network:

network <- st_read("path to file")

- Required columns:
    - id_link: link id (integer)
    - ffs_fw: forward free flow speed in km/h (numeric)
    - ffs_bw: backward free flow speed in km/h (numeric)
    - cap_fw: forward link capacity in veh/h (numeric)
    - cap_bw: backward link capacity in veh/h (numeric)
- Optional columns:
    - addv_fw: forward additional cost (numeric)
    - addv_bw: backward additional cost (numeric)
    - wg_fw: forward cost weight (numeric)
    - wg_bw: backward cost weight (numeric)
    - alpha: alpha BPR function parameter (numeric)
    - beta: beta BPR function parameter (numeric)
    - cost: actual network cost, used instead of geometric distance and speed (numeric)

head(sioux_network)

• zones sf object data containing zones to be added to road network:

zones <- st_read("path to file")

- Required columns:
    - id_zone: link id (integer)
    - longitude: Longitude in WGS84 (numeric)
    - latitude: Latitude in WGS84 (numeric)

head(sioux_zones)

3. Running the code :

Use the following methods to perform operations:

configure_graph() Tranforms network data from source network and zones and outputs a list with 2 elemens, 1 being the graph and 2 being the transformed zones:

configure_graph(network,zones,turn_penalty,left_side,addvalue_cols,weighted,performance_measures,use_cost_col)

• network: sf object containing network data
• zones: sf object containing zone data
• turn_penalty: Adds turn penalty (see dodgr documentation on turn penalty from function 'R/weight-streetnet.R')
• left_side: Does traffic travel on the left side of the road (TRUE) or the right side (FALSE)? - only has effect on turn angle calculations for edge times. (see dodgr documentation on turn penalty from function 'R/weight-streetnet.R')
• addvalue_cols: Does network shapefile has columns 'addv_fw' and 'addv_bw' for added cost in links? (TRUE) or (FALSE)
• weighted: Does network shapefile has columns 'wg_fw' and 'wg_bw' for a cost multiplication factor in links? (TRUE) or (FALSE)
• performance_measures: Does network shapefile has columns 'alpha' and 'beta' for usage on BPR funcion? (TRUE) or (FALSE)
• use_cost_col: Use cost column as cost (opposite to speed and distance)? (TRUE) or (FALSE)

configure_demand() Tranforms demand data from source origin and destination to match origin and destination from graph/zone objects:

configure_demand(demand,zones)

• demand: demand data
• zones: zones object created with configure_graph()

solve_ue() Solve the traffic flow assignment model (user equilibrium) by Frank-Wolfe algorithm (all the necessary data must be properly input into the model in advance).

solve_ue(graph,demand,col,max_iterations)

• graph: Graph object created with configure_graph() function
• demand: Demand object created with configure_demand() function
• col: Column from demand file selected for assignment
• max_iterations: Maximum number of iterations for Frank-Wolfe algorithm

Next is an example of the structure to run the entire code, with additional functions that are documented inside R:

flist <- configure_graph(sioux_network,sioux_zones,use_cost_col=TRUE)

graph <- flist[[1]]
zones <- flist[[2]]
demand <- configure_demand(sioux_demand,zones)

graph <- solve_ue(graph,demand,col = 'demand')
times <- get_times(graph,demand)
paths <- get_paths(graph,demand)

network <- graph_to_sf(graph,sioux_network)
plot_flowmap(network,show_labels = T)

Questions

Feel free to send an email to [e-mail] if you have questions or concerns.

Future releases

Future releases will have an implementation of other traffic assignment algorithms such as Gradient Projection, Origin-based assignment, and Algorithm B.

douglascm/trafficr documentation built on March 13, 2020, 11:44 a.m.