Pre‐2025 public transport method
This page is an archive. It describes the method for public transport that the model used before 2025, and that PBL used in the reports Toegang voor iedereen? (2022) and Beter bereikbaar? (2024). The current method is described on Public transport and From chains to journeys, and the current component codes on Components and merging.
In 2025, the Planbureau voor de Leefomgeving (PBL) asked Object Vision to develop a method that not only calculates the fastest public transport routes, but also minimises travel costs. The method on this page calculates the costs only for the fastest route that it has found. It computed the fastest route with Dijkstra’s algorithm on a complete routable public transport network, including waiting at home, transfers and waiting at stops. Then it added the pre- and post-transport legs to the result. This worked well, because the travel time is known per link. The costs are not, because the Dutch fares of some operators have non-linear marginal cost components.
To cover different pre- and post-transport options, the method calculates several impedance matrices, each with other pre- and post-transport options attached, such as:
| Pre | PT | Post | Description |
|---|---|---|---|
| W | PT | W | Walking -> Public transport -> Walking |
| C | PT | W | Cycling -> Public transport -> Walking |
| C | PT | B | Cycling -> Public transport -> Cycling |
| C-Tic | PT | W | Cycle to InterCity rail station -> Public transport -> Walking |
| C-TMT | PT | W | Cycle to rail, metro, or tram stop -> Public transport -> Walking |
In the current model C-TMT is the pre-transport type Crmt, and B is the post-transport type C, cycling from the last stop; cycling to an intercity station (C-Tic) is not a type of the current model.
Different methods should also be combined, however. Cycling to intercity stations, for example, could be interesting. But from many areas no intercity station can be reached by bike within a time set by a parameter (for example 15 minutes). So the option to walk to a public transport stop should also be included. Different combinations could therefore be defined, such as C-Tic_PT_W with W_PT_W. This setup could, however, let a traveller walk from the origin to a stop and then walk on from that stop to the destination, when the origin and destination are close. In that case the traveller should walk directly to the destination. So the combination should always include a direct walking or a direct cycling trip.
An advantage of this setup is that the space-time-dependent public transport network has relatively few turns, so Dijkstra’s algorithm finds the fastest route relatively quickly. Also, when the origin set is split into geographic tiles, the tiles can be calculated in parallel, using the full processor and memory capacity of the computer.