Signal Randomiser

[Kromaatikse]

Living in a heavy commuter zone, I get to see all sorts of reasons for disruption.

Only this afternoon, I saw two trains outright cancelled due to lack of crew - apparently there's been an overtime backlog all year due to the heavy winter we've just had, and this has finally caught up with them as people take their major annual leave. Incidentally, the units scheduled for the missing trains were *not* appended to the following services, so those were full and standing, further extending stopping times.

Earlier this summer, part of the line I live on was singled for two weeks straight while they replaced the waterproofing of some underbridges. This necessitated halving the service frequency over that section. I observed the reversing arrangements at the near end of the section, and noted that the timing for the reversal was rather tight. requiring a driver at both ends - and that the trains not continuing to the far terminus sat idle for ten minutes, complete with both drivers. Not the most efficient operation in the world.

Occasionally lines get blocked unexpectedly, too, or there are signal and pointwork failures (especially in winter). This has a distinctly chaotic effect on train performance, even under emergency timetables designed to reduce the number of train paths required. Last week there was an OHLE fault in the main depot, resulting in some trains being unavailable for their traffic - and substituted by other stock which was approximately in position. A Pendolino was replaced by an "express train" - using the local equivalent of early Mk2 stock and thus unable to proceed at anywhere near the speed required by the Pendolino's timetable.

Disruption also occurs when long-distance trains are delayed by track maintenance in some far-off part of the network. Sometimes - by no means always - other trains are held to meet them, which of course results in those trains running out of course. Or the stopping time of a commuter train is extended at an intermediate station, because several mums are getting on and off with their prams, or someone has trouble with a bicycle, or it's simply extra-busy for some reason and people are having to squeeze into whatever space is available.

Realistically, randomisation can replicate that very last category of disruption nicely - the rest, I think, require specific scenarios written to include them.

[asfinmja]

Only this afternoon, I saw two trains outright cancelled due to lack of crew - apparently there's been an overtime backlog all year due to the heavy winter we've just had, and this has finally caught up with them as people take their major annual leave.

This time one big contributor for cancelations was problems with computer program used to create workshifts and allocating cabinet crew to trains.

Realistically, randomisation can replicate that very last category of disruption nicely - the rest, I think, require specific scenarios written to include them.

Randomization can quickly generate more disruption than non-human dispatcher can solve. One AI broken in wrong place can lead to situation where player train can not complete tasks assigned by scenario writer. More random events there are more likely it is scenario can not be completed at all. I this propability for total lockup will increase exponentially to number of random events.

And as earlier was said, testing of scenario of this kind would be nightmare of worst kind.

How about creating a scenario, then cloning it ten times, alter the AI trains' timings and via points (from main line to loops etc), alter the player train's timings etc, so that you end up with ten variations on the same service.

I think this would be manageable way to introduce "random" elements. You can make say morning commuter run. Then clone it for every weekday and introduce anomalies to normal running, different variations for every day.

[AndiS]

How about creating a scenario, then cloning it ten times, alter the AI trains' timings and via points (from main line to loops etc), alter the player train's timings etc, so that you end up with ten variations on the same service.
Finally, and the important bit, alter the scenario names to gibberish so that you or the player simply does not know which scenario s/he will be playing, could be any one of ten.
Good idea? No? oh.
regards Ted.

Actually, for the purpose of the game, I think it is a very good approach. You could name the scenarios by appending a running number to the original. Then you would still know what is going on basically, but not which variant you get today.

Only drawback is your scenario list will get pretty long, but with the above naming convention, you ought to still manage it.

I guess with about 5 variants of each scenario, it will appear random enough to you. All it takes to look new and/or random is going beyond your memory capacity.

[DescendingSadly]

Some really interesting discussion and constructive criticism has come from this, thanks guys I guess cloning the scenario again and again and again is the only way forward. My thanks to you all.

[asfinmja]

Some really interesting discussion and constructive criticism has come from this, thanks guys I guess cloning the scenario again and again and again is the only way forward. My thanks to you all.

RSPaul on Railworks America: "We can now do random events" http://www.railworksamerica.com/forum/v ... 159#p28702