/Traffic-Assignment-Frank-Wolfe-2021

This simple script computes the traffic assignment using the Frank-Wolfe algorithm (FW) or the Method of successive averages (MSA). It can compute the User Equilibrium (UE) assignment or the System Optimal (SO) assignment. The travel time cost function that models the effect of congestion on travel time is pluggable and definable by the users.

Primary LanguagePython

Traffic-Assignment-Frank-Wolfe-2021

This simple script computes the traffic assignment using the Frank-Wolfe algorithm (FW) or the Method of successive averages (MSA).

It can compute the User Equilibrium (UE) assignment or the System Optimal (SO) assignment.

The travel time cost function that models the effect of congestion on travel time is pluggable and definable by the users.

Currently, three cost function implementations are available:

  • BPR cost function (see more)
  • Greenshields cost function (see Greenshields, B. D., et al. "A study of traffic capacity." Highway research board proceedings. Vol. 1935. National Research Council (USA), Highway Research Board, 1935.)
  • Constant cost function (no congestion effects)

Our implementation has been tested against all the networks for which a solution is available on TransportationNetworks and has always obtained the correct solution.

How to use

First, clone the repository to a local directory.

To use the algorithm, simply call the computeAssingment() method in the assignment.py script.

The documentation of the method provides a through description of all the available parameters and their meaning.

Importing networks

Networks and demand files must be specified in the TNTP data format.

A through description of the TNTP format and a wide range of real transportation networks to test the algorithm on is avaialble at TransportationNetworks.

Acknowledgments

  • This work is based on Traffic-Assignment. I focused on fixing this implementation and extending it to pluggable cost functions and user optimal flows.
  • All the networks I used for testing the correctness of the algorithm are available at TransportationNetworks.