AI physics - bad load, bad choice of loco, or bad luck?

C

Concrete_Bob

Member
So I'm running a session with multiple AI consists, using several different loco/commodity combinations. One consist is having persistent issues with collisions. I'm double heading with DD40X'S on a 2200tonne coal train. Ruling gradient for the layout is 2%, which is used in significant lengths.

So the question is.......Is this an excessive load for the conditions? Are the locos inadequately braked for the operation? Or is it just 'one of those AI things' where AI does whatever it likes?
 
The maxdecel setting is the key to AI braking capabilities. You can clone the config, tweak the values (try 10x higher to start with), and use Properties to select your new engine spec. You can get the loco to stop as quickly as you like.
 
I asked a question in English, the reply looks like English, and yet most of it seems utterly useless to me. I kinda thought that the point of AI was to run the trains autonomously, going full nerd and tinkering with files is not my idea of gameplay.

This seems to be one of the Trainz community's cultural weaknesses, trying to be more of a programmer's convention than an accurate simulation of rail operations.
 
Concrete_Bob said:
I asked a question in English, the reply looks like English, and yet most of it seems utterly useless to me. I kinda thought that the point of AI was to run the trains autonomously, going full nerd and tinkering with files is not my idea of gameplay.

This seems to be one of the Trainz community's cultural weaknesses, trying to be more of a programmer's convention than an accurate simulation of rail operations.
Click to expand...

How is this for simple. Trainz has NO artificial intelligence features at all. None, nada, not any. It is a marketing blurb. What trainz has is computer controlled trains that follow instructions either programmed into them by the game programmers or by the creator of the driver commands you apply to the train.

You're in essence, programming the train yourself. What Tony, who is the CEO of N3V by the way, offered is a way to tweak the internal behavior of the loco. If that is over your head then try the techniques suggested by MP242.

William
 
wreeder said:
-snip- You're in essence, programming the train yourself. What Tony, who is the CEO of N3V by the way, offered is a way to tweak the internal behavior of the loco. If that is over your head then try the techniques suggested by MP242.
William
Click to expand...

Or take the opportunity to learn something. So much of what I read on the forums is over my head. But I don't challenge the culture I'm a guest in. I either just ignore and move on or take the time to learn something about what is being said.
 
I currently have a rather modified setup of the Legends of the BNSF I and II running with another parallel mainline through the mountainous terrain and off to another merged route. There are some really steep grades that require a substantial amount of planning ahead due to a slow curve and tunnel located at the end of the long decent from the summit. Using slower speeds long ahead of the tunnel and curve means I can prepare for the turn and tunnel without speeding headlong off the tracks, in real life that is. Combine the slower speed ahead of the junction to single track, curve, and tunnel, along with the application of dynamic brakes, means I can successfully navigate the slope without an issue. With this setup, the AI manage to negotiate the line quite well since they know when to apply the throttle better than humans, and generally can brake faster and better than we can.

If your trains are derailing, you may want to look at the setup and see what's going on. the Centuries were among the most powerful locomotives in their day and a couple of them together is quite a powerful machine even in real life. The UPRR used them to climb the famous Sherman Hill and to negotiate other long open stretches due to their strong hauling power. The thing is, it's not just the locomotives doing the work here. It's all physics as in real life with the opposing forces and momentum coming into play here. As the locomotives pull, the consist behind the locomotives is doing two things. First it's pulling back behind as the locomotive climbs the hill and then second it's pushing the locomotives forward as the train travels downslope. This additional weight behind the locomotives can prevent the train from stopping on a steep grade, and a 2% grade is quite steep, as well as prevent a train from climbing over a hill. In addition it's these opposing forces that can break a train in two as they say due to the strain on the couplers.

With that said, you may also want to check the locomotives you are using for this line, and perhaps limit the length of your train. As in real life, the choice of motive power is something that needs consideration as well the train weight. I can't remember who made it, but someone at Jointed Rail put together an engine requirements calculator which can assist you in determining the motive power needed to bring your consists over the mountain.
 
wreeder said:
How is this for simple. Trainz has NO artificial intelligence features at all. None, nada, not any. It is a marketing blurb.

William
Click to expand...
Sounds about right. So much of Trainz is oversold. Glad To know that Tony is CEO, explains a lot as to what Trainz has evolved into. Pull your head in you cheeky git, there's a difference between being 'over someone's head' and 'being poorly explained'.

Tony is quite happy to sell this game to anyone as a train simulator. By that product labeling, a consumer should reasonably expect it to simulate trains. Moving the physics goalposts in order to make a consist work on your route is not a real simulation.

It's a great product for people who like to scratchbuild their own components and try to fix bugs and flaws in coding, and it's a great product for those who want to pretend to drive locomotives, but in terms of reality, as far as freight loading, unloading and AI running, Trainz has been stuck in a siding for the last 15 years.


MP242 appears to understand exactly where I am coming from, and has offered good reality based solutions. All you've done is act like a wanker.Good day to you.
 
MP242 said:
Most Trainz locos have a ridiculously overly powerful enginespec, and 2 Trainz DD40X would have no problem stopping a train on a -2% downhill grade. At a choke point, or nearing a Control Point interlocking, placing a 20 mph speed board, a mile or two distant, that would slow an AI train down, allowing it to stop in time, before sliding through the interlocking at track speed.

Some real RR's have a rule to stop a train, and test the brakes, before going down a grade, or speeding headlong into town at full speed.

I usually put 4 SD45's head end, and 2 more on the rear, and that usually allows enough traction pulling power for going up a grade, or enough stopping power to slow a loaded train down on a -2% grade

CSX has recently been running 2 to 3 locos head end, with a single mid train loco, and a single rear end loco, all synced together as DPU, as it provides even power and braking, with no slack run in/out, on an a relatively flat grade between Baltimore and Philadelphia, and beyond.
Click to expand...
Thanks for that, glad that you can offer good procedural advice rather than how bend rules to make a bad consist work.
 
As I have said before, Trainz is what it is. It is a sandbox game with a railroad theme. Make with it what you will. Does it make everyone happy, no.

You'll get out of it what you put in to it.

William
 
You must log in or register to reply here.

Similar threads

P
Session for IndustRail 1a - to explore and set up AI trainz
Replies
1
Views
3K
mtldrm38
M
Back
Top