[JohnnyS]

Hi,

Had noticed AI trains in timetable mode accelerating very slowly after a signal stop or station stop in recent versions of OR. At the start of tests, I had two x releases of OR installed x2386 and x2474. Subsequent testing with the available automatic builds shows the first occurrence of creeping at x2410, x2409 checks out OK.

Two videos to demonstrate.

x2386 https://www.youtube....h?v=7DhKnouDa2w
x2474 https://www.youtube....h?v=iFPOps8ZELo

A screenshots with the AI service concerned highlighted in the extended HUD using x2410.
https://farm6.staticflickr.com/5571/15048138438_a3946947e2_o.jpg


I don't have many traditional activities installed, so I don't know if the slow acceleration occurs in activity mode as well, I haven't seen it happen yet.

Is there a method for logging AI performance to provide more information? I don't think I have provided enough information for a proper bug report.

Cheers,
John.

[disc]

Same here.

[disc]

Do this route have hidden signals that are sometimes visible on track monitor?

[roeter]

@JohnnyS :

• You highlight train 145 (6:48 from Norwich), but the info shows that 141 (7:17 from Ipswich), which is just ahead of 145, is doing just 3.4 mph at 100% throttle. I can not see, however, if this train is just starting from a stop.
  What would be interesting is a number of screenshots with dispatcher info which are about 1 minute apart, showing the progress of both 141 and 145.
  
  
• You state that the problem probably started with version 2410. That version introduces the circuit breaker logic, which applies to electric engines only.
  Would it be possible to check if the problem only occurs for electric traction - for instance, could you run a specific test with both trains 141 and 145 hauled by diesel traction (e.g. class 47), and see what happens?

Thanks,
Rob Roeterdink

[JohnnyS]

Quote

Do this route have hidden signals that are sometimes visible on track monitor?

Yes.

Quote

You highlight train 145 (6:48 from Norwich), but the info shows that 141 (7:17 from Ipswich), which is just ahead of 145, is doing just 3.4 mph at 100% throttle. I can not see, however, if this train is just starting from a stop.

Yes the 0717 is accelerating from a station stop just in front of the Norwich train, I should have been clearer, the Norwich service stops only for the red signal it is not a timetabled station stop in the videos.

Quote

What would be interesting is a number of screenshots with dispatcher info which are about 1 minute apart, showing the progress of both 141 and 145.

OK, I'll do this some time later today.

Quote

You state that the problem probably started with version 2410. That version introduces the circuit breaker logic, which applies to electric engines only.
Would it be possible to check if the problem only occurs for electric traction - for instance, could you run a specific test with both trains 141 and 145 hauled by diesel traction (e.g. class 47), and see what happens?

I'll do this ASAP. I have another timetable for a different route that currently only has diesel traction, I have witnessed no crawling, though the timetable is considerably smaller (6 or 7 services at the moment as opposed to around 200).

Cheers,
John.

[disc]

I've tried to replace a creeping AI with diesel engine, and it seems that's the problem, as the diesel didn't do this. However the creeping of electrics always connected to signals, as they only do this if the next signal is not fully clear. If approach or something that is not CLR1-> creeping. If CLR1 then they accelerate normally.

[dennisat]

I have just noticed this same problem in a non-timetabled activity while waiting an age for an AI electric train to clear a station stop. It barely exceeded 3mph in the first few hundred metres then went off like a gazelle. It had a perfectly clear path, no other train had preceded it.

Dennis

[roeter]

Thanks for the info - I think I know what is happening.
IMHO, the new circuit-breaker logic does not work correctly for AI trains. It leaves the circuit-breaker open for AI trains, which means electric AI trains do not have any power.
Normally, this remains unnoticed as AI control has an acceleration 'override' : if the train is at 100% throttle and still is not moving as required, the 'override' kicks in and accelerates the train to near the allowed speed.
But, when approaching a restriction (in BRAKE control mode), there is no need to force the train to near the allowed speed as it is to stop soon anyway, so the override is not activated. But without any power, this leaves the train creeping along.

I do not know enough about the new circuit-breaker logic to investigate this any further, so I hope someone with enough knowledge will pick this up.

Thanks for your help,
Rob Roeterdink

[Serana]

Hello,

OK, so the circuit breaker is open in your case.
That means that either the TCS system doesn't authorize the power (but that's not the case on AI trains, or more precisely for trains with no players, because the power authorization is always given for these trains) or the pantograph isn't up.

In order to find which condition is the cause of this problem, we probably need a debug HUD that will tell us what is the state of the subsystems of the train.
I will start to write this HUD page.

[disc]

Pantograph was always up in my case. But as i wrote, the AI trains always accelerating in 100&000 mode, just very slowly.