I need help or inspiration

I have been working on a large city based route for years. Columbus Ohio and four years. Very involved and detailed. I have since become a content creator though sometime I wonder why. Anyway...I have got to the point of testing my route using portals, schedule library and path rule. I have excel documents that list in detail what train is used, the time and which portal it comes from. I make notes snd critiques of problems, timing errors ect. Anyway, I let my session just run to see what happens and it seems after a certain period of time everything falls apart. Consists start "forgetting " commands and then all hell breaks loose as things back up..portals throw script errors...atls crossings get mad. I have almost lost my ****...sorry for the language but I have put ALOT of time into this. I guess what I want to know is if there are still folks using 2012 that can offer me suggestions or tips or anything. I am really quite close to giving up years and countless hours or work and I don't want to.

I think you will find the issues are in the fine detail. My fairly large session was moved to T:ANE when it was released not that I believe it matters in respect of TS2012 one way or another. I have 40 plus trains running in all directions, leave it to do its own thing and in time it will effectively lock up, watch it and it won't.

Because each schedule is an endless loop of varying lengths the interaction between the trains changes with time and sooner or later two or more trains either cross path where AI cannot determine which goes first or one train will wait at a junction for example for ages giving priority to another effectively miles away. In both instances everything then builds up behind those trains, the schedule is compromised and the log-jam has begun.

Watching it you tend to help it along then modify the instructions to provide that missing priority etc. but miss the event and you are unable to appreciate what started the crash and in turn try to prevent it happning in the future. Unfortunately of cause it could be months of continuous running before all the trains are back at their exact starting point at the same time. I'm sure I have yet to find all the clashes in my creation but fortunately with two computers I can let one just run, save and re-start where it left off looking for such conflicts while fiddling on the other. Peter

I will always lay track in TRS2006, and test those routes out as a WIP in TS12 and T:ANE ... but I will always be editing my route using TRS2006

Peter summed it up nicely. With a large number of AI drivers there's a 1 in N chance that one of them is going to screw up the works because they can. I agree it does get frustrating to have a nicely working session fall apart like someone unraveled a ball of string. I too have given up at times on some sessions then gone back only to find out it wasn't the AI after all and it was my fault to begin with. During my second looks, I have found misplaced track or direction markers, an incorrect signal, sometimes wrong commands, or even the right commands in the wrong order. This latter issue can really cause confused AI as they're following their cookie bits and tracing their routes.

Unfortunately this isn't just the problem and if this solved the problem once and for all we'd be dancing in the streets with joy. You see Trainz is actually quite complex. There's a lot of underlying systems which affect the operations as we see them. With all the scripts running, eventually things get bogged down. This issue is being addressed in the Trainz Dev, and these improvements are far off in the future. Unfortunately for me, I cannot explain them because they're being discussed in Trainz programmer geek talk.

With a real dispatching and a real scheduling system in place, this would change the balance in our favor since we would have the control over the drivers, more or less, and not allow them to just run and chase track marks. The interlocking towers have made it possible to control junctions, along with other rules, but having these other two subsystems in place would solve things on another level. The problem with the interlocking towers, as they are implemented, it is extremely complex to setup in my opinion. Why are they overly complex? If this is the case for other possible subsystems, then people would be more apt to not use them even though they maybe useful.

Hi Justin,

I always worry when I see Portals and ATLS in the same sentence.

This is obviously only part of your problem but just so you know, always make sure you finish an ATLS sequence before you enter a Portal, (i.e all Triggers have been cleared).

ATLS works by looking a train’s unique ID number which is given to it when you start Trainz. When a train enters a Portal to re-appear out of another one, that ID number changes. So as far as ATLS in concerned, the Alistair going in one Portal is a different Alastair to the one coming out of another, (or even the same Portal). That’s OK so long as the train has cleared all Triggers first. Don’t for example have it operate a crossing open trigger, then enter a Portal, then pop out of another/same Portal to operate a closing trigger. That won’t work.

Also make sure the train is fully out of the Portal before it hits a Trigger.

Other than that, ATLS shouldn’t get confused over time.

Boat

On another note, a vein of something that Boat mentioned, there's something else to watch out for not related to train ID. Watch how close your signals, track marks, and direction markers are from stations and other interactive assets including road crossings. These assets work with triggers and there's a specific radius around them.

When the AI triggers a road crossing and has encountered a track direction marker before that's really close, it can cause the AI to forget about the direction marker. The same with signals being placed too close to stations. I have run into this issue with a transit route. I had small dwarf signals placed just before and after the platforms. The setup works for one train, but a following consist actually ignores the signal at the platform entrance, or sometimes gets stuck at the signal. Other times the AI will ignore the signal and continue to go through the station and ignore the load command. I found moving the signals away from the platforms helped with this. This is the same situation with other interactive industries - remember stations are in reality interactive industries.

Road crossing will do the same. We like to have signals before a road crossing, but since they have a trigger for a script, the AI will get confused. The same situation exists as it does with the passenger stations. I found I had to move the signals back away from the road crossing - about 200 meters to work consistently. ATLS crossings, however, will vary due to the ability to place the trigger points closer or farther away. You will have to experiment with these to find that sweet spot for the AI.