Railworks America

Can somebody explain why are there train service classes (priorities) existing in RW and how do they work? As result of my tests permission to proceed has in most cases been given to player train no matter how train priorities have been set. This leads to conclusion that service classes do not work as expected. Dispatcher in RW don't care much of train priorities. Maybe some experienced scenario creators can comment this? I have constructed hundreds of advanced scenarios for Trainz. In RW it's almost impossible to get elementary things working. New timetable scenario tool helps to understand RW pathing logic but it didn't fix faulty dispatcher behavior (which trains can get priority). Everybodies opinions are welcome.
I was talking about scenario creation problems in UK Trainsim forums ... but seems they don't like such threads at all ... especially developers ... and I was banned for unknown reason ( http://forums.uktrainsim.com/viewtopic. ... 4&t=105462 ). Strange behavior.

I hope this is what you are looking for.
http://forums.uktrainsim.com/viewtopic. ... 6&t=105209
I'm pretty sure you will have read this.
I haven't written any scenarios and will not attempt to do so until I have a better understanding of the AI.
From what I have read, my understanding is this:
The signalling has compromises over which is layered, the dispatcher (a user coined term) that also has compromises thereby making AI train programming a very complex activity to attempt and get right.
There is no doubt that there are some authors out there that seem to have a better handle on this than others.
The good news is that the timetable editor goes a fair way towards unlocking the secrets.
Andy

RW dispatcher tests:

First is here:
http://www.mulgirailway.com/rw_traffic_test_1.html

As far as I can tell, the train priorities are simple numbering system except for Special. In most cases Special is like the lead actor in a play. He gets all the benefits. From the list on down from Special think one through the last one, Other Freight, with one being the higher priority and the others being the lesser priority. You can't make a freight train a passenger priority though. Freight has to stay freight and passenger has to stay passenger.

As far as the charts in that link: I didn't understand any of it nor did I find that thread helpful. And apparently the author didn't either because he said later on that when applied to signals, all the rules were out the window.

Player priority is only one small smidgeon of getting a scenario to work right. I suggest getting my Scenario expansion pack #41-56 for the Fort Kent route and take a look at how the scenarios work. In the manual I included a small briefing of each scenario to explain the highlights on how I got the scenario to work.

Most of the scenarios come with AI traffic, some of which are simple meets, and some of which are more complex with the player interacting with the AI train. There's always a method to my madness. First, see it work. Second, copy it. Then finally, adaptation.

Rich Garber

I downloaded the test.rwp after reviewing your post on the test and here are my best guesses and a solution as to why your test scenario didn't work.

If I remember correctly it was the UP train with the path failure, right? I already saved my version and quit Railworks so I can't go back and look. The failure in this scenario wasn't the priority but rather your trying to put two AI trains in the same yard. Let's get to some basics first. This is going to go way beyond class priority but it should help you appreciate why a simple looking scenario can become so complicated.

The yard is protected on both ends by signals and while it is a multi-track yard, remember, to the signals it's still a single piece of track. So any blockage or traffic in between the two signals would normally give you a red. And in this case, I believe that's what is the problem here. Granted the signals are the E type meaning they ignore that there's traffic beyond the signals - but only to a degree. Let's remember this as layer 1 as Andy referred to in his post where scenarios are complicated because layers of intelligence are at work here.

The second and probably more imporant layer here is the pathing. Both players are ending up on the same single strip of track we call the yard. So as far as the pathing is concerned you have two trains occupying the same block. In my own scenarios where I tried to include the player train and an AI train in the yard at the same time, wouldn't work. Whereas the E signals suggest this is not a problem (and maybe only with rollingstock!) the pathing overrules.

Pathing is key. And note it didn't matter when the pathing obstruction occurred, that it does, you're three strikes and out. This becomes problematic because you must learn to think through and anticipate the pathing problems. Pathing is more complicated in Railworks than MSTS because you don't have ultimate control over any of the train paths. You should. The AI is constantly trying to be clever and find a clear path much like you saw in the old Transport Tycoon games or some of the Railroad Tycoon games. This creates an unpredictability and it muddies up the solution because of it and only adds further to the confusion for how to get a scenario to work.

But ultimately the goal is to get the scenario to work, right? So I did this:



By changing the ultimate destination to Portal N for the UP train, the scenario should work. I didn't test it though I should've. Just cause everything looks free and clear now it doesn't mean it works, totally. I would also say that any priority used, like standard over other freight would have given you the same results. That's something you'll have to try though to be certain.

Hope this helps.

Rich

Thank You Rich for You explanations. Was very interesting to read. I'm coming from Trainz environment and in Trainz there are no such a doubtful limits like "no two trains per yard allowed". Mostly I used manual method of scenario creation. I hope You know that in Trainz You can control junction states with rules, triggers and variables. For example train B is activating trigger T and junctions J, I, K states will change and some other trains can move on. This is totally different world where You as scenario creator can control everything. No game engine will fight with You trying to be clever. I hope someday it'll be possible in RW too.
Anyway I already bought Your latest scenario pack and will check what's inside. Scenario creation in RW is a mystery for me. And not only for me.

Let's have a little fun here and expand your test scenario to something a little more interesting. In this version I did what you tried to do with your original test version and I'll show you even what you tried to do initially will work.

What I did was to simply extend the AI train's destination to the north portal and do a car drop-off in the yard. Your train follows behind and stops at yard track #4.

Here's the first pic:



You can see the AI train has dropped off 3 freight cars and is moving out to the north portal.

Here comes the player train:



To backtrack a moment, this situation works even if both trains are given a final destination of the Portal N. The difference is both trains are in their own signal blocks so you don't get that error you originally did with the two trains ending up in the yard at the same time.

Hope this helps.

Rich G

MrMurphy wrote:Thank You Rich for You explanations. Was very interesting to read. I'm coming from Trainz environment and in Trainz there are no such a doubtful limits like "no two trains per yard allowed". Mostly I used manual method of scenario creation. I hope You know that in Trainz You can control junction states with rules, triggers and variables. For example train B is activating trigger T and junctions J, I, K states will change and some other trains can move on. This is totally different world where You as scenario creator can control everything. No game engine will fight with You trying to be clever. I hope someday it'll be possible in RW too.
Anyway I already bought Your latest scenario pack and will check what's inside. Scenario creation in RW is a mystery for me. And not only for me.

I heard some talk somewhere that Railworks is going to some kind of scripting language where something like in Trainz will be available for Railworks. I think that's a good thing but really, it's tough already creating a scenario without all the other stuff to make rabbit trails out of.

Rich

Are you saying that IF each yard track were signaled that the original scenario would work?

jamesphh wrote:Are you saying that IF each yard track were signaled that the original scenario would work?

I think so.
I'm starting to understand the logic in RW scenario creation It's mostly in timing and not as much in train priorities.

jamesphh wrote:Are you saying that IF each yard track were signaled that the original scenario would work?

Good catch! I believe so. Because essentially the signal (or path AI) is seeing a train already in the yard and another is trying to enter the same block. But try it where you add the final destination to Portal N and you'll see no errors. That tells me the path isn't seeing a problem with two trains entering the yard or two trains running on the same yard track.

Rich

MrMurphy wrote:I think so.
I'm starting to understand the logic in RW scenario creation It's mostly in timing and not as much in train priorities.

That's what I'm starting to understand as well. I was able to successfully create a AI and Player train meet on the Fort Kent passing tracks, and I figured that even though both trains were "Standard Freight" if I wanted the player train to have priority, he'd have to be the first one into the block where the switches were.

I'm going to try to expand on this tonight with another scenario.

I agree on that one too. If you need to use the same classification for two different trains timing and what I call placement is imperative. With two trains coming at each other besides being routed to different tracks on a siding you want one to get there "first." There's no real decision making going on in Railworks so in a sense, you have to artificially make the decisions for the game. There's several ways to do it and I think I covered some of them earlier, I can't remember. Generally if you can assign different priorities, that's the best way to handle issues of meets. Railworks uses regular train labels for their priorities but that appears to me as misleading. One time I did ask them what the differences were in the priorities and I was told to think of it as a number priority. The lower the number, the higher the priority. So with Other Freight being the lowest position on the list, it has the least priority. Let me clarify this. In other words, don't think of the priority label that Railworks use as having to do with anything in its real world counterpart. It will trip you up if you do.

I made the same mistake. I thought giving two trains the same "real world" priority would actually mean something to the game. It doesn't. Basically, each train was treated the same - and nothing worked. When I started experimenting with the priorities I noticed things worked better leading me to the conclusion priority, or the way this game uses priority, is like saying "who goes first."

Rich

One of the early problems I had with the priority system was that it would lock out the entire route. Like if I put the player's train at Ft. Kent Yard and the AI train at Eagle's Lake, depending on who has the higher priority will get the entire route to themselves. However, I may have approached that wrong.

What I have done with the scenario that I took a screenshot of above is that both were "Standard Freight", the AI train started two signal blocks behind the block where the meet was suppose to take place (copied this from one of RIch's new scenario :) ), the AI train made it to the spot first, but I wanted the player's train to pass through the switch first. So I basically had the AI train wait until the player train passed through. I managed this by using the new scenario editor tool and basically using the tool to figure out how long it'd take for each train to reach the spot.

Makes sense how you got it to work. Each scenario seems to want to be treated differently. And it's compounded by our desire to choreograph the events and how and what order things happen. I keep an old naval war cannon parked just a few feet back from my computer just to let it know that if I don't get what I want, kaboom! Unfortunately Railworks has a mind of its own and doesn't care if I end up replacing the computer, desk and wall. In some ways it's the open ended way things are done in Railworks that makes scenario creation as difficult as it might be if it was a more controlled environment. What I mean is, take for instance if you would be happy still if the AI made up its own mind who went first - like a dispatcher was there. It might be less frustrating but then less flexible in the way things are choreographed. What I think would make everybody happy is if somebody from RS.com came and explained how the AI looks at things. It's software so it's gotta have some pecking order. Through observation we can figure some of the things that go on but things being as flexible as they are - it's taking a good while to see the patterns.

On a side note... some of the things being said about the scenario editor elsewhere are flat out wrong. And that these things are continually touted as things wrong with Railworks is making me mad to the point I want to make a video of some of things happening in my Fort Kent scenario pack that people are saying can't be done. The problem with the scenario editor isn't what you can't do, but what you can do. And because you can do it, it adds to the mystique, because it's not so easily accomplished, nothing works right. I'm working on this new route right now but I might take some time to make that video.

Rich