The International Simutrans Forum

 

Author Topic: Strange private car routing  (Read 78 times)

0 Members and 1 Guest are viewing this topic.

Offline Phystam

  • Devotee
  • *
  • Posts: 510
  • Pak256.Ex developer
    • Pak256 wiki page
  • Languages: JP, EN, EO
Strange private car routing
« on: June 06, 2021, 02:41:06 PM »
Sometimes, for whatever reason, the main route is not used at all.
Perhaps the effect of past congestion has led to this odd route being chosen. The main route used to be chronically congested, while the route to make a U-turn at the interchange was hardly congested at all.
The U-turn at the interchange route looks less costly in the route calculations, but if all the cars actually make a U-turn, the capacity of the intersection is so low that it causes congestion.
Measures such as "not allowing turns in the same direction at two consecutive intersections" need to be taken when calculating the route.

This is an example of a diamond-shaped interchange, where the mainline was similarly congested, so the route was chosen to enter immediately after the interchange.

Offline jamespetts

  • Simutrans-Extended project coordinator
  • Administrator
  • *
  • Posts: 20776
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Strange private car routing
« Reply #1 on: June 06, 2021, 02:46:59 PM »
Unfortunately, this sort of problem can be fantastically difficult to investigate effectively, especially when it exists on a large saved game, as it can be very difficult to extract the relevant data from such a huge and complex dataset.

The system for computing congestion has recently been updated to fix a number of previously reported anomalies, and certainly congestion has been the inferred cause of this sort of issue in the past. It may help to monitor that junction over the course of a number of refreshes of the private car routing algorithm to see whether this varies with congestion.