Coupling an existing consist to another consist?

[Accurate]

This is using AI:
Couple a set of cars to the engine. Now I want to attach another set of cars to this consist. In effect assemble a long consist from various cars using AI.
I do not see a way to do this?

[mrennie]

This is using AI:
Couple a set of cars to the engine. Now I want to attach another set of cars to this consist. In effect assemble a long consist from various cars using AI.
I do not see a way to do this?

As far as I know, AI aren't capable of coupling up to consists by themselves - only the player engine can do that. I tried it some time ago but it always resulted in the AI crashing into the cut of cars it was trying to reach.

[Accurate]

That is indeed unfortunate. One of the major functions (as far as I know) is the assembly of a train from cars in the yard. That assembly is probably designed based upon a delivery sequence. It would be a very interesting operation to have AI assemble a train and then either another AI train, or a player train, take the assembled consist and deliver the cars to their proper destinations. For me this is a major issue since I never drive trains.

[wackyhuman]

I have created scenarios where AI couple up to other trains. But I only used an AI engine to do this, not an engine with other cars already attached. So I know it works for the AI engine only. One of my German shunting scenarios has a AI train come in, drop off 3 cars, head over to another siding and pick up some other cars and drive away again.

It took some trial and error, as sometimes it would work and I changed one little thing, like a different siding, and it suddenly did not work anymore and the engine would not drop off the cars for example. I have never tried it with some cars attached to the engine and trying to pick up other cars.

[mrennie]

I have created scenarios where AI couple up to other trains. But I only used an AI engine to do this, not an engine with other cars already attached. So I know it works for the AI engine only. One of my German shunting scenarios has a AI train come in, drop off 3 cars, head over to another siding and pick up some other cars and drive away again.

It took some trial and error, as sometimes it would work and I changed one little thing, like a different siding, and it suddenly did not work anymore and the engine would not drop off the cars for example. I have never tried it with some cars attached to the engine and trying to pick up other cars.

You did well to get that far! Did you use any special techniques to do it (such as setting stopping points or something like that)?

[CrimsonKing]

One thing I remember from experimenting with this myself is that the consist the AI is picking up has to be there at the start of the scenario. It cannot be dropped off by an (the same or another) AI or the player.

My experiment has the AI starting off as just the locomotive. It picks up one string of cars from a yard siding and then picks up 2 more consists from 2 other yard sidings before dropping them off at another place in the yard. I used no stopping points. Only pick up and drop off commands.

To sum things up: AIs can pick up consists but the consist can not have been moved before the AI couples to the consist.

[wackyhuman]

> Did you use any special techniques to do it
Nope, just creating the steps in the scenario editor.

> AIs can pick up consists but the consist can not have been moved before the AI couples to the consist.
This was the case in the scenario I did.

[Accurate]

I agree on the strange limitations as described. I now recall seeing the problem where locomotive a drops cars and another train refuses to add them . That is really too bad since it makes running a complex yard scenario, using AI, impossible. I will be gaining a nice chunk of disk space.

[Chacal]

If you think about it, it is not a strange limitation.
The TS2014 "dispatcher" checks instructions and calculates paths for every train at the start of the scenario.
If there's no car on a track segment, then a pick-up instruction is impossible.

Maybe it is possible to use stopping points for simulating a pick-up, but probably it will result in a collision, or by the dispatcher refusing to allow the train to proceed because the track sgment is not empty.

ALso, yards are not good for AI trains because they are usually not signalled.

[Accurate]

The ability of a locomotive to take its existing consist and add to it is done all the time in a competitive product. It is a matter of having the flag fields assigned in the consist descriptor.
Running AI in a yard is easy so long as you do not do the "double couple steps". Signals often are just eye candy. but almost every part of track in the yard needs them to create "blocks" for the residence of existing trains in the yard or those that show up later. No signals and every train is "seen" by every other train and conflicts occur.

Often people do not know how to place links and signals fail.

Finally, the poor program is unable to see a set of cars on a yard track so it directs a train on that path and runs over the cars.

Jackson has steadfastly refused to allocate funds to clean up this stuff so only partial reality can be achieved within the AI in yards. If fixed you could run multiple shunters breaking down trains and reallocating the cars to other consists. Then having the thru train go into the yard and collect the assembled train for attention by an "industry" elsewhere. Too bad, but those particular problems, if resolved, would make the "game" more of a "simulation".