Congestion and if it blocks a route

[jeffatsqi]

It is my understanding that "congestion" lead to no route for potential trailers. However, I may have this wrong because I am seeing the following:

  * A route with a single DC3 building to over 5,000 passengers waiting at the terminal

So, the question is what is the limit for congestion that blocks the route.

Also, on a similar note, I had a bus route from a airport to a cite build a waiting queue of over 3,000. I added a fast train from the airport station to just outside the city and a tram to the end point station. No one moved to the new train route. Not the 3,000 wait or the new passengers. They all chose to get in line behind the 3,000 already waiting for the 1 stop bus as opposed to the 2 stop train/tram. Is this expected behavior?

[VS]

(With "trailers" you perhaps meant "passengers"?)

A "route" can not get congested since it does not manifest itself in any way. These are only lists of places where vehicles should stop and that's it. No counters or capacities to fill.

Stations are the kind of entity that can be congested. This means that passengers coming from houses don't enter. However, these getting off vehicles to change lines still do get off and increase count.

The example with tram+train vs. bus is really classic. Route finding for cargo takes only one metric, and that is number of switches. The bus line has zero hops and tram+train one. 0<1 and all passengers go to bus.

It works that way since searching for all possible ways between two nodes (stations) works on what you called earlier "routes". And these hold almost no information about the connection. To decide correctly if train+tram is better than road, the game would have to take into account:

• speed of vehicles serving the connection (average? median? first found? from statistics?
• max. speed of the way (road, track)
• terrain profile and turns/curves

This would have to be done for all pathfinding, which is one of the most time consuming parts of game. Hence the metric is kept as simple as possible. Counting the nodes where cargo is switching lines is almost as simple as when you speak 1,2,3...

If you want to utilize the new connection, remove the station from bus route completely. Then the waiting passenger will be forced to take the "longer" route.

[jeffatsqi]

Thanks for the response. It is very helpful.

On the second  question of zero hops bus vs one hop tram+train I understand the algorithm. I had removed the bus and all the passengers did move to the train/tram.

On the first question – congestion blocking a route – I understand  your response to mean that only a the point of origination – first node in network – will passengers be blocked by congestion. However, once in the network they can unload at any station to transfer regardless of how congested. Is that right?

If so, then passengers in the simulation behave very differently than cargo. In the case of cargo there is "back pressure" to stop the loading of cargo for a factory that is already at max stocking levels. For people there is no "back pressure"  that blocks entry into a network that will provide poor serve because it is already overloaded (just like a factory is over supplied). The only screen is  at the initial stop. Correct?

Don't get me wrong. I am not complaining.  This is an excellent simulation that I am using it as the basis for an Economics Lab and thus I need to understand the behavior model for each component.

On follow-up question. If I have two alternative routes between A and B, both are direct, one is short and the other is long with many stops, the passengers waiting at station A will just take the first vehicle that show up. Correct?

Now what about a completive situation? If one of the AI players has a route A' to B' that also covers the destination that travelers want to go to, are the potential travelers split 50%/50% between station A and A'?

[prissi]

In reverse order: For goods the routes are split; passenger (and mail) will simple go to the nearest, non-overcrowded stop.

All goods just know their destination and their next transfer point. They will hop on any vehicle stopping there. However, they will decide for one of two complete routes; which one is not always clear beforehand, since it depends on building order of the stops.

And you are right, there is no "back-pressure" for passengers (and mail). The challenge with pasengers and mail is rather the need to upgrade your network to transport them all, after you reached break even. For goods it is rather to deliver nearly the correct amounts "just in time" (but the stockpiling can switched off, either by beginners mode or permanent in the simuconf.tab file).

[VS]

only a the point of origination – first node in network – will passengers be blocked by congestion. However, once in the network they can unload at any station to transfer regardless of how congested. Is that right?

Yes.

In the case of cargo there is "back pressure" to stop the loading of cargo for a factory that is already at max stocking levels.

Yes. It's a slightly different mechanism (not part of routing), but the end effect is what you describe.

For people there is no "back pressure"  that blocks entry into a network that will provide poor serve because it is already overloaded

Yes.

On follow-up question. If I have two alternative routes between A and B, both are direct, one is short and the other is long with many stops, the passengers waiting at station A will just take the first vehicle that show up. Correct?

If switching vehicles does not occur between A and B - yes.

In reverse order: For goods the routes are split; passenger (and mail) will simple go to the nearest, non-overcrowded stop.

A little clarification: this is before they (goods & psg.) enter the first stop on their way. And prissi speaks about situation when there are more valid "entrance points".