Setting Up Permissive Signals In Yard Properly

[AdvancedApproach]

I have two yards on this route which includes locomotive staging points. These are where I keep my local and switching motive power when not actively in use. Only problem is when returning my AI drivers get stuck and notify me that they're waiting for access. This is with type 5 dwarf signals with number boards, which makes them permissive. How can I fix this problem? Using invisible signals have been of no assistance for me and actually have caused even more headaches. Here's an example:

 

[mtldrm38]

You do not always need to add a entry signal to every sidings...
A L5 yellow invisible signal is ok after the entry junction.
04 outbound signals are ok or invisible to keep the parked engine from controlling the entry junction...
From what I see you are attempting?

 

[norfolksouthern37]

I have been using the following or similar setup for a while and do not experience problems with AI drivers. Filling up a route with invisible signals is NOT a recommendation I would make, but I see it done a lot. I tend to try not to use an invisible signal unless it is in this context of a yard, where the exit signal offsets are used (<kuid2:45324:24900:6> JR Signal Offset), or in a case where this YD permanent restricting signal is used (blue triangle - <kuid2:45324:24012:1> JR Invisible Signal (Yard)).

 

[AdvancedApproach]

You do not always need to add a entry signal to every sidings...
A L5 yellow invisible signal is ok after the entry junction.
04 outbound signals are ok or invisible to keep the parked engine from controlling the entry junction...
From what I see you are attempting?

So take out signals on the yard tracks? How would I account for AI trains? I'm worried that could cause some unwanted problems. Do you have an idea on what should be done afterwards?

 

[crazytrain]

I avoid signals in yards like the plague. The AI doesn't switch well anyways, without some really intense order filing or scripting, and then it's hit or miss if they do what you intended. OK, I'll just do it myself!

But, if you insist on it, try a #6 dwarf where the #5 is at the entrance. It should have 2 lights instead of 1 to cover both L/R diverging routes. Also, without seeing how the back side of the yard is configured/signaled, it could be any number of things that keeps Joe Engineer from parking where he's told.

 

[justinroth]

I have been using the following or similar setup for a while and do not experience problems with AI drivers. Filling up a route with invisible signals is NOT a recommendation I would make, but I see it done a lot. I tend to try not to use an invisible signal unless it is in this context of a yard, where the exit signal offsets are used (<kuid2:45324:24900:6> JR Signal Offset), or in a case where this YD permanent restricting signal is used (blue triangle - <kuid2:45324:24012:1> JR Invisible Signal (Yard)).

Oof I'm gonna need to rethink some things.

 

[AdvancedApproach]

I avoid signals in yards like the plague. The AI doesn't switch well anyways, without some really intense order filing or scripting, and then it's hit or miss if they do what you intended. OK, I'll just do it myself!

But, if you insist on it, try a #6 dwarf where the #5 is at the entrance. It should have 2 lights instead of 1 to cover both L/R diverging routes. Also, without seeing how the back side of the yard is configured/signaled, it could be any number of things that keeps Joe Engineer from parking where he's told.

Took out all of the yard signals. Only the entrance and exit signals remain. However, AI still gets stuck when putting trains on the locomotive staging point track mark. I had a locomotive occupying the track but not that particular point and it still was waiting for access. Too bad N3V won't do anything about that.

 

[TSwenson]

I'm thinking this is one of those funny rules things that signaling won't help and the program doesn't cover (correctly). In my experience with Trainz, there are some serious holes in terms of the AI operating rules, as there seems to be an inability of the AI to operate in anything other than signaled territory.

In my experience (NORAC), yards or "yard limits" are authority by verbal permission and are sort of a no-mans-land where you have to watch out for everyone else, while signaled territory gives you authority based on the signal indication and you "own" that track. There are exceptions that occur, read the rulebook, it's that thick for a reason. My guess, not knowing the specifics of AI operation coding, is that the AI driver gets confused somewhere it the transition from signals to unsignaled yard, probably getting stuck in a decision loop and defaulting to "stop and ask for help" mode.

Yards Limits are generally under a "restricted speed" operating condition where you have to be under a certain speed AND slow enough to be able to stop short of something in your way. Being able to stop in "one half of range of vision" is the NORAC standard with a maximum of 20MPH (or 15MPH, read the rulebook). Yes, that means you are probably creeping along with that 10,000 ton train, and flying along light engine.
The AI drivers don't seem to communicate with each other like real crews, and are decidedly shortsighted in managing their routing in unsignaled areas. I haven't found any kind of "yard limit" signal/sign/track mark that enforces that condition in Trainz yet (and PLEASE tell me if there is one).

 

[AdvancedApproach]

I'm thinking this is one of those funny rules things that signaling won't help and the program doesn't cover (correctly). In my experience with Trainz, there are some serious holes in terms of the AI operating rules, as there seems to be an inability of the AI to operate in anything other than signaled territory.

In my experience (NORAC), yards or "yard limits" are authority by verbal permission and are sort of a no-mans-land where you have to watch out for everyone else, while signaled territory gives you authority based on the signal indication and you "own" that track. There are exceptions that occur, read the rulebook, it's that thick for a reason. My guess, not knowing the specifics of AI operation coding, is that the AI driver gets confused somewhere it the transition from signals to unsignaled yard, probably getting stuck in a decision loop and defaulting to "stop and ask for help" mode.

Yards Limits are generally under a "restricted speed" operating condition where you have to be under a certain speed AND slow enough to be able to stop short of something in your way. Being able to stop in "one half of range of vision" is the NORAC standard with a maximum of 20MPH (or 15MPH, read the rulebook). Yes, that means you are probably creeping along with that 10,000 ton train, and flying along light engine.
The AI drivers don't seem to communicate with each other like real crews, and are decidedly shortsighted in managing their routing in unsignaled areas. I haven't found any kind of "yard limit" signal/sign/track mark that enforces that condition in Trainz yet (and PLEASE tell me if there is one).

Out here, GCOR has the same rules, at least, to my understanding.

 

[TSwenson]

Out here, GCOR has the same rules, at least, to my understanding.

Yes, that was my understanding, but I'm qualified on NORAC and know it well, so that's my default operating rules.
Anyways, across the versions ('04 to '19) I've found that the AI doesn't like to change between signaled/non-signaled operations, and it does rather poorly at the latter as well. I would LOVE for NV3 to actually release the Trainz AI Operating Rules as it would solve so many problems (and wouldn't the employee who did that be a hero to us).

As to the issue, I would make sure there isn't anything else besides your trackmark within 50m, I've found that anything (junctions, signals, speed signs) that is within approximately 25-50m of something else seems to affect the AI driver negatively.

 

[AdvancedApproach]

Yes, that was my understanding, but I'm qualified on NORAC and know it well, so that's my default operating rules.
Anyways, across the versions ('04 to '19) I've found that the AI doesn't like to change between signaled/non-signaled operations, and it does rather poorly at the latter as well. I would LOVE for NV3 to actually release the Trainz AI Operating Rules as it would solve so many problems (and wouldn't the employee who did that be a hero to us).

As to the issue, I would make sure there isn't anything else besides your trackmark within 50m, I've found that anything (junctions, signals, speed signs) that is within approximately 25-50m of something else seems to affect the AI driver negatively.

GCOR is easier to learn while NORAC is considered more complex. The trackmarks are at least 100m from the nearest junction. I eliminated the signals closest to those trackmarks. Now I do have invisible speed limit signs where my trackmarks are since 40 mph is the default speed for Trainz. Will removing them actually help? Lastly, the locomotives were more than 100m from the trackmark in question and that didn’t help.

 

[TSwenson]

GCOR is easier to learn while NORAC is considered more complex. The trackmarks are at least 100m from the nearest junction. I eliminated the signals closest to those trackmarks. Now I do have invisible speed limit signs where my trackmarks are since 40 mph is the default speed for Trainz. Will removing them actually help? Lastly, the locomotives were more than 100m from the trackmark in question and that didn’t help.

Yes, everyone says it's easier, but I use NORAC profesionaly so... "Ya go with what ya know..." I was once told that NORAC was designed for scheduled passenger trains and GCOR was made to move freight.

Well... so much for the easy bits. Try eliminating the speed limits... I always try to start with the simple and stupid stuff, I've found that about 80% of the time it's something simple/stupid that gets missed/overlooked. You probably don't want to hear some of the choice phases I use when I find how stupid a fix is.
After eliminating the simple/stupid stuff, I'm guessing there is a logic loop that the AI driver is getting stuck in, but without the rules or decision making tree for the AI it is going to be a painstaking process of trial and error to find the issue. I remember when the direction you put the track spline down in... affected things... despite Auran's claims to the contrary.

 

[JCitron]

Yes, everyone says it's easier, but I use NORAC profesionaly so... "Ya go with what ya know..." I was once told that NORAC was designed for scheduled passenger trains and GCOR was made to move freight.

Well... so much for the easy bits. Try eliminating the speed limits... I always try to start with the simple and stupid stuff, I've found that about 80% of the time it's something simple/stupid that gets missed/overlooked. You probably don't want to hear some of the choice phases I use when I find how stupid a fix is.
After eliminating the simple/stupid stuff, I'm guessing there is a logic loop that the AI driver is getting stuck in, but without the rules or decision making tree for the AI it is going to be a painstaking process of trial and error to find the issue. I remember when the direction you put the track spline down in... affected things... despite Auran's claims to the contrary.

KISS. It works every time.

I call your method top-down troubleshooting and I've used it for years not only for Trainz but also in the past as a hardware technician. Start with the easy things first then start digging deeper as if whatever is being worked on is made in layers.

I've noticed that while the AI is prone to many errors and there are many bugs that go hand-in-hand, many errors are also self-inflicted. You wouldn't believe how many times I've "fired" drivers for doing something insanely stupid only to find out later that I was sending them to the wrong track mark.

 

[AdvancedApproach]

@norfolksouthern37 what we be the most appropriate signal system for this setup? Is it the same as my other map you replied to?

 

[AdvancedApproach]

@norfolksouthern37 mind you this is for a locomotive refueling depot and two tracks where helper locomotives are staged.