News:

Simutrans Tools
Know our tools that can help you to create add-ons, install and customize Simutrans.

Waiting areas with choose signals

Started by okurz, August 13, 2015, 11:32:57 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

okurz

Hi, I have been researching all topics and web hits I could find about arranging waiting areas using choose signals. The most promising layout suggestion was this: http://archive.forum.simutrans.com/topic/02093.0/index.html

I want to setup waiting areas similar to the one shown in the referenced topic. So far I did manage to setup waiting areas but only by building platforms with the length of the train in the waiting areas for each waiting rail, see




This seems weird to me as so far I did see the schedule and line reservation taking platform length into account. E.g. I could wrongly order a train of length 4 to go to a platform of length 3, order to load up 100% and make it wait indefinitely as the last 2 waggons are not adjacent to a platform. But for the choose signal it seems to make a difference. If I build waiting areas with just single-tile platforms as a wait stop the trains will not choose the platforms as destination after the choose signal except for the one tile that has been selected as the destination in the schedule, see



BTW, I know that choose signals work with the next stop in the schedule, be it a station or a waypoint, so I guess this is not the problem.

My questions:

  • Is it correct that waiting areas as described need full train length platforms? This would deny the purpose of saving on maintenance but still offer a benefit when the target platform can not accomodate more space
  • Are there other, better approaches for setting up waiting areas? I know only one alternative which is a long winding single trail with signals through which each train has to wind through in all cases.

I am using simutrans 120.0.1 with pak64

Other references to the topic:
- http://forum.simutrans.com/index.php?topic=1091.0
- http://forum.simutrans.com/index.php?topic=12966

Ters

All waiting areas in Simutrans are hacks. There is no such concept coded into the game. For the most part, I have built my waiting areas lengthwise. If trains wait very long, that will cause a great deal of false alarms that the trains are stuck. In that case, you have however probably set up the line wrong, with too many and/or too long trains. On long lines, this can be near impossible to get right, though, at least with the default JIT algorithm.

okurz

Of course there is no concept necessarily coded into the game but the choose signal is available and it helps me anyway :-) I am just interested in your experiences, if they support my observations or if you have other cool ideas.

Quote from: Ters on August 13, 2015, 12:44:33 PM
On long lines, this can be near impossible to get right, though, at least with the default JIT algorithm.
Interesting note. Seems like I have this issue on some lines. For one, I guess the train with length of 6 is just too long. So do I understand you right that I am probably not safe of a sporadic "train stuck" message?

Ters

Quote from: okurz on August 13, 2015, 01:51:05 PM
Interesting note. Seems like I have this issue on some lines. For one, I guess the train with length of 6 is just too long.

Ideally, trains should arrive with shipments at the same rate they are consumed at the destination, but the supplier doesn't provide you with goods at that rate. With some consumers, there is too little input storage for a short enough train to make a profit.

Quote from: okurz on August 13, 2015, 01:51:05 PM
So do I understand you right that I am probably not safe of a sporadic "train stuck" message?

I actually don't remember if I ever got this on the only station based waiting area I have ever built. As long as the train is waiting for a free path and not a full load, I would think it will consider itself stuck eventually. I might just never have had trains waiting long enough. Generally, although my waiting areas are big enough to hold all trains on a line (minus one waiting at the actual station), that's just a safety precaution to avoid holding up the main lines should something go wrong. I have never bothered trying to fix those few lines giving me false stuck messages from time to time, so I don't know if it's actually possible or not.

DrSuperGood

The solution is JIT2 (nightly build only ATM) where industries will constantly receive orders at the rate they order products. This eliminates the need for waiting areas and likely also the need for spare platforms.

With JIT2 if a line has 10 coal trains on it then chances are only 0-1 will be waiting for load at any given time. If more than 1 is waiting then either the line has high traffic (causing timing between the trains to vary) or you have 1 train too many (you could use 9 trains for same amount of shipment). This also means that at any given time 9-10 trains will be moving on the track. The overall result is that the trains do not need waiting areas and instead are occupying the buffering capacity of the tracks.

Ters

But if I understand JIT2 correctly, the capacity of each individual train is still as important with regards to the recipient's input storage capacity. Or more so, because the surplus is just discarded without payment?

DrSuperGood

QuoteBut if I understand JIT2 correctly, the capacity of each individual train is still as important with regards to the recipient's input storage capacity.
Correct. So you should be using 3-6 tile long coal trains (both of which are economical), and not 9-12 tile coal trains. The input storage for coal is not really a problem in pak64 as long as you stay away from maximum length trains.

QuoteOr more so, because the surplus is just discarded without payment?
You do get paid, just the consumer suffers an order penalty (so you lose out on shipping some amount of product).

prissi

JUst to confirm: Choosesignal won't choose a too short stop to avoid trains hanging for loads which could be never achieved, just because they ended up at the wrong platform.

okurz

prissi, thanks very much. This fully answers my question number 1, too :-)