Repository navigation
Building a routable network
Jip Claassens edited this page Aug 3, 2026
·
3 revisions
(outline — to be written out)
A geographic file of road geometry is not yet a graph. This page will describe the step from downloaded source data (Network data sources) to a connected, directed network that Dijkstra can traverse.
-
From arcs to links and nodes — splitting polylines at intersections; building the node set;
the
F1/F2from-node/to-node convention used throughout this wiki - Planarity and grade separation — crossings that must connect versus bridges and tunnels that must not; why naive intersection of geometry produces routes through viaducts
- Directionality — one-way streets, and representing them as directed links
- Turn restrictions and turn penalties — where they matter (urban car routing) and where they can be ignored
- Mode-specific subnetworks — filtering links by access rights per mode, and why a pedestrian network is not a subset of a car network
-
Connectivity checking — finding disconnected components with
connected_parts, and deciding which are genuine islands and which are data errors - Barriers and crossings — level crossings, ferries, gates
- Network optimisation algorithm — reducing the network once it is built
- Travel speeds per mode — assigning impedance to the links
- Connecting origins and destinations to the network
- GeoDMS network functions
Accessibility modelling documentation
© Object Vision BV. Documentation under CC BY-SA 4.0, source code under GNU GPL v3.

Concepts
Networks
- Network data sources
- OSM
- NWB
- GTFS
- Building a routable network
- Network optimisation algorithm
- Travel speeds per mode
Origins & destinations
Computation
- Shortest path and the Dijkstra algorithm
- Origin Destination matrix (trip table)
- Isochrones and catchment areas
- Public transport routing
Indicators
Applications
Network models
Reference