The New Settings for Friction, etc

[Junna]

In the past it was possible to severely limit the cornering calculations by changing the settings in simuconf, but in the new version, these settings are gone and replaced instead by that new corner force divider calculation. However, it is not possible to configure it in such a way as in the past (what I want is to disable the corner effect of 45 degree angle turns entirely within 2-3 tiles length). Changing the friction thing to 0 does not seem to change much.

/

And unrelated - but what is the ideal set-up with the new signals? The signal distance now has to be very great or the trains will invariably slow to 50km/h upon reaching a signal. This very much decreases the capacity of a line since very long distances are required between signals and having two stop signals after each other or closer than a certain number of tiles to a pre-signal (it is nice that they finally have some use, though) will make it brake.

[jamespetts]

The new system for cornering is designed to use realistic speed limits based on the actual corner radius, which means that it calculates the radius of a curve for 45 degree bends when it reaches the second in a sequence of them. The intention is to simulate gentle curves by multiples of 45 degree turns: the further that they are spaced, the gentler the curve. The curve friction factor is a different thing entirely, and does not affect the cornering speed limit at all. If you want to increase the speed limit for corners, reduce the corner force divider for the relevant waytype, but note that this will affect all types of cornering for that waytype, not just 45 degree turns, and also that it is currently calibrated to be realistic, so changing it is likely to produce unrealistic cornering speeds on all types of corner. The new system is not compatible with being able to have different types of rules for different types of corners: the old system worked that way only because it was somewhat half-baked.

As to the new signals, this system is not yet complete in that not all types of signalling systems are represented. The main type of signalling system that you will be using on railways at present is the absolute block system. In order to stop trains slowing before every signal, you need to use distant signals (formerly pre-signals), which need to be at least braking distance away from the stop signal that they are protecting. This is complicated furhter by the absolute block system's definition of a block section, which, unlike modern signalling, is not merely the space between two signals, but is the entire area on any given line controlled by a single signalbox. Each section would have only one distant signal but may have had multiple stop signals. Signalboxes have yet to be implemented (they are planned), so, in the interim, a distant signal will only show clear if all stop signals between the distant signal and the next distant signal (or an end of choose marker, whichever is closer) are clear. If the distant shows clear, then the train does not have to slow enough to stop in time for the stop signals: if the distant is at caution, then it will have to slow for each stop signal until it reaches the next distant signal.

[Vladki]

a distant signal will only show clear if all stop signals between the distant signal and the next distant signal (or an end of choose marker, whichever is closer) are clear. If the distant shows clear, then the train does not have to slow enough to stop in time for the stop signals: if the distant is at caution, then it will have to slow for each stop signal until it reaches the next distant signal.

That sounds weird... AFAIK real distant signals are affected only by the immediately next stop signal and nothing else. Moreover at least on czechoslovak railways there exists a "repeater" distant signal - usually on half-way between distant (pre-signal) and stop signal. Both show the same aspect, just the repeater has an indication that it is a repeater and you should break really hard ... So a more distant signals for one stop signal sounds ok, but one distant signal for more stop signals is IMHO plain wrong.

[jamespetts]

It may sound weird in modern terms, but I can assure you that that is how it used to work in days when signalmen knew where trains were only by looking at them out of the windows of their signalboxes and by telegraph signals sent from nearby signalboxes: see the Wikipedia article for a full explanation.

[Vladki]

Hmm, OK. That's new for me. Here in CS, the distant signal is showing only the aspect of home signal (we call it entry signal - as it is on the entry to station). And home signal acts as distant signal for start signal (departure). In earlier years, the home signal was not acting as distant signal for start signal, and the train was expected to slow down or stop at the station. But the distant signal was always showing only the aspect of the immediately next main signal.

We have also "block" signals - when the track between staions is split into several blocks and then there is only one main signal (home and start signal combined in one.)

[Junna]

Trains slow down at the stop signal (red) here for some reason. It is not entirely intuitive how to set these up adequately.

[Spenk009]

Replace the pre-signal with a normal stop signal. Then place a pre-signal after the stop signal. I guess the train wants a normal stop signal to tell it that the track ahead is free. If it only reads (and reserves) the pre-signal and gets an ok after the stop signal, it's still running on visual range (max 50km/h) up to the stop signal.

[jamespetts]

It should be easier to understand how to use the absolute block system when signalboxes are introduced. Apologies that I have not been working much on Simutrans-Experimental lately; it is difficult to concentrate when there is much outstanding on my house.

[Ves]

Dont stress yourself too much! We will still be around when the house is finish! :-)
I have a question about the new signal system:
Would you consider add animation frames to the signals so one could get blinking lamps?
I know that in Sweden, all presignal elements are shown by blinking lamps.
Eg, the presignal is blinking green to warn for danger and blinking white for free. Two aspect signals show solid green and blinking green to show free but expect danger, or solid green and blinking white to show drive and expect drive.
Think other countries uses blinking as well.

[jamespetts]

That is an interesting thought, but I am unfamiliar with the Simutrans graphics engine, and it would probably take me so long to work out how to do this that it would take too much time away from other, more pressing projects. If someone else were to do it, however, I should happily include it so long as it worked well.

[Junna]

That moving business is sure taking it's time. Hasn't it been like a year?

How is the signal boxes and the overall signal system intended to work when the work is done?

[jamespetts]

It is taking a very long time! I started looking for properties in June last year, found one in August, completed the sale in October, and have been in the process of having it renovated (and, in the meantime, staying with my parents) ever since. The renovation is making good progress now, though: the decorator has nearly finished. It is still consuming a great deal of my time as I spend my time researching things such as lightbulbs and antique furniture rather than coding for Simutrans, and find it difficult to concentrate on two large, long-term projects at once.

As to signals, the intention is for there to be a number of different signalling systems based on the types of signals used. This is already partially implemented, with drive by sight, absolute block and token block already in place.

Signal boxes will be required in order to allow signals to be used: the plan at this stage is for each signal to have to be connected to a particular signal box, and each type of signal box to be able to deal with different types of signals, with variable limits on the number of signals per box and the maximum signal box to signal distance.

Absolute block signalling would involve signal boxes with quite low limits (the smallest boxes having a limit of perhaps 6 signals, and all boxes having a maximum distance of only a few km, perhaps 2-3km), but the signal boxes would be much cheaper to build than more modern signal boxes (which would have much higher limits and work with more efficient, modern signalling). In absolute block, a distant signal connected to one signal box will clear only if all signals on the train's route controlled by that signal box are also clear, or else it will be at caution, and the train will have to slow enough to stop at each stop signal until the next distant signal is passed at clear. The idea would be for each signal box to control a section containing, on each linear route passing through that section, a single distant signal followed by between about 1 and 3 stop signals (the norm in real practice being 2), to reflect the practice of real railways using this system.

[Junna]

How will signal boxes be connected to stations? Would there be some manner of interface for connecting them?



2 is a stop signal for a single-track segment. Previously, the one-way signal was at the pink line after the pre-signal, but this makes the signals not work right (i.e. the train will slow for the stop signal even though the pre-signal displays green). I assume this is something that hasn't been fixed yet.

[Junna]

So...?

[jamespetts]

Apologies for not having answered: I moved into my house at the end of June, and was very busy dealing with connected matters.

I am not quite sure what you mean about signal boxes connecting to stations; there is no need in principle for them to connect in any specific way to stations in particular. The idea is that they will connect to signals: each signal box will be able to have a maximum number of connected signals within a maximum radius.

[Junna]

What I mean is, what happens if there are more signal boxes within the coverage area? How would they be distributed? Would they go simply to whichever is closest?

[jamespetts]

The tentative plan at present is for the build signal dialogue to be activated by clicking on the signal box, not through the main menu, although I am not sure how easy that that sort of change in GUI will be.

[Junna]

Hadn't these gaps been fixed?

Also, a while ago there was mention of some sort of fencing and stuff to be put around roadways in cities to protect them from being reclaimed by the city-- what happened to this feature?



How do I signal this to make the train not slow down? There's a stop signal and then a distant signal ahead here, and plenty of braking space.

Also, as I mentioned before, having one-way signals (and now that they are a must to make loops work right, this is quite a problem) now disrupts the communication between distant and stop signals, so that if there's any one-way between them they will not work as intended, and cause slow downs for no reason.

[jamespetts]

Can you upload a saved game for the second issue?

As to the fences, somebody else was working on them, and it seems that that work was never finished.

As to the gaps, despite many attempts, I have never managed to fix them.

[Junna]

http://www.mediafire.com/download/fk1y2wp5y3os2xf/testmap1.sve

[jamespetts]

Thank you for posting that. This is somewhat tricky to track down, and I think is related to the long sequence of signals without an end of zone (end of choose) marker. Because this issue will be treated differently when signalboxes are introduced, I think that it would be more efficient for me to seek to fix this only at that stage.

Incidentally, you are using a rather non-standard method of signalling here: you are signalling a single line with signals intended for double track working. You really need to use token block signalling for single lines, using the long block signals, rather than absolute block using stop and distant signals. Trains will then have to slow down at stations and other passing places, but this is what trains actually have to do in a token block system in any event.

Have you had any of these problems on double track lines?

[Junna]

Have you had any of these problems on double track lines?

Yeah, but invariably near a junction which would have a similar set-up to this (the picture exampled before-- in the older post #12 in this thread) or where there would be an approach to a station (which would have a similar set-up in practice).

[jamespetts]

I think that the operation of distant signals will need to be reviewed when I introduce signal boxes, as it is rather difficult to make it work without them, as the signal box was central to the notion of a section in absolute block signalling.