News:

Simutrans Forum Archive
A complete record of the old Simutrans Forum.

RETB retirement date

Started by Mariculous, April 27, 2020, 10:51:28 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Mariculous

In the pakset, RETB is retired in 2010.
Ingame, there is no other signalling system capable to properly signal long single tracked (dead-end) section, so it should really not be retired!

Apart from gameplay reasons, there might also be realism reasons against the retirement of RETB.
First, it is right that it seems to get replaced on favor to ETCS.
QuoteRETB is being gradually replaced with the new European in-cab signalling system, ERTMS.[2] The Cambrian line was due to be changed over to the new system by spring 2010 but was delayed, being commissioned on 11 March 2011.

However, there seem to be exceptions from this rule, which caused further developement on RETB.
QuoteThe rugged terrain and light line traffic made it cost-prohibitive to install GSM-R for these areas and so a new radio system, with new base station and on-train equipment, was developed. This allows RETB to continue to operate on Band III Sub Band 1
The latter is called RETB NG.

So I guess it's fine to retain RETB ingame after 2010.
Alternatively, or additionally we could add an ETCS signal working as a token_block (purely pak side work), although the slowdown when approaching a clear signal would not be correct in ETCS.


edit: the little changes can be found in https://github.com/jamespetts/simutrans-pak128.britain/pull/85/commits/b64959fc034010c9f6684420daba8ea904ed6070

jamespetts

The RETB system was indeed retired in game to match reality, as it is no longer deployed and is being replaced with ETCS. May I ask what the source of the quotes is?

The intention is that it is simply replaced by track circuit block: the in-game advantages of RETB is that one can have a single, inexpensive signalling centre controlling a very large area of lightly trafficked lines. At the time (the 1980s), the only signalboxes capable of controlling track circuit block signalling have a much smaller range, and are thus not cost effective.

By the time that ETCS is introduced, however, signalboxes have a much larger range. The small ETCS signalling centre has, I believe, a range similar to that of the RETB cabin, so there is no longer a need for RETB signalling.

I note that you mention that no other system can signal single track lines as well as RETB (which is actually an implementation of token block). It is intended that track circuit block with bidirectional signalling is equally capable: may I ask why this does not work well?
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Mariculous

#2
The quote itself is from wikipedia:
https://en.wikipedia.org/wiki/Radio_Electronic_Token_Block#RETB_Next_Generation

Another one from commons design:
QuoteIn 2013 Comms Design were selected by Network Rail to design and deliver a new RETB system for the North and West of Scotland. Additionally, as experts in RF systems, we planned the full radio network to ensure high quality radio coverage throughout the route.
https://commsdesign.net/retb-for-network-rail-uk

And finally,
QuoteThe RETB Next Generation Project, described in issue 125 (March 2015), is more than a renewal of the current system, as it also mitigates the risk of radio interference when European frequencies for digital television change in December 2015. This means that the current ?Band III sub-band 2? radio equipment will be replaced with new radio equipment operating on ?Band III sub-band 1?.
https://www.railengineer.co.uk/2015/07/13/retb-next-generation-project/


Quote from: jamespetts on May 13, 2020, 11:55:02 AMmay I ask why this does not work well?
It's bugged and will deadlock in such situations. I have recently posted a bugreport for this.

Further, it is not capable of signalling single tracked dead-end sections. Token block and one train staff are the only signalling systems that can handle this properly, both being available in a quite ancient version in 2017, so RETB would be the solution here either.

Even further, in simutrans, RETB is more cost-effective than directional reservations in some cases, as it does not require a signal at each station.

And one of the huge advantages of directional reservations, allowing multiple trains to follow up in the same direction, can be a disadvantage in some cases.
A train entering such a single section is not guaranteed that it can leave the section!
If there is no free space in the passing loop, such a train will deadlock.
Thus, mostly single tracked lines with stops on the single tracked section and small passing loops are also best signalled with some kind of token block.
Using directional TCB on such lines can cause deadlocks as soon as a single train is delayed.



Conclusion:
Assuming the bugs are fixed, directional TCB and cab signalling can often replace token blocks.
In case of dead-end tracks with intermediate stations and in case of small passing loops with intermediate stops on single tracked sections, token blocks cannot be replaced with TCB nor cab signalling.
In case of larger passing loops and intermediate stops on single tracked sections, TCB might be replaced by directional TCB or cab signalling, but it will be less cost-effective.

If we wanted TCB/cab signalling to be a replacement for token block in all situations, these had to always reservate through to the next signal, even in case of intermediate stops.

jamespetts

Thank you for the reply: I will investigate the bug report when I get a chance. In reality, one train staff working is used in the cases of dead end tracks with intermediate stations. This has no retirement date in the pakset. May I ask why this together with track circuit block is not sufficient for the dead end cases?

For the passing loop cases, one should not be having to use token block so late in the game, as this is not what is generally done in reality, certainly not RETB, which is intended specifically for very sparsely populated areas with long single lines far from civilisation. Is the problem that the block reservation does not go all the way through to the next signal, or that the directional reservation does not go all the way through to the next signal? The latter would be much easier to solve than the former.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Mariculous

#4
Quote from: jamespetts on May 14, 2020, 10:23:54 AMMay I ask why this together with track circuit block is not sufficient for the dead end cases?

I guess that's simply a personal preference of "real" signals over playing torch relay.
It feeels quite wrong to me to equip a dead-end track, freshly built in 21st century with an ancient signalling system that requires stopping before enetring and after leaving a dead-end section. Such a setup is quite common to my tram-train and most DLR networks on stephenson-siemens.

Apart from one train staff "feeling wrong" in the 21st century, token block simply has the best behavior for such layouts, so I use it there.
That means, it doesn't require trains to slow down when leaving a single tracked section, nor forces trains into drive by sight after leaving, and it's fail-safe in case of depots on a single tracked section.

Anyway, from a German point of view, I'd expect TCB and ETCS to be capable of handling such situations. Most single-tracked tracks here use some kind of TCB on such lines, a few lines still use something quite simmilar to mechanical absolute block.
Very few tracks don't have any signalling at all, in which case the conductor may continue on order, which will either be handed out physically, using a printed form. Alternatively, the form might be dictated by the traffic controller via radio and the form will be written by the conductor.
Yes, that's quite old-fashioned and there are only very few such tracks.

Quote from: jamespetts on May 14, 2020, 10:23:54 AMIs the problem that the block reservation does not go all the way through to the next signal
That's exactly the issue.
Imagine a single tracked line, where there is a stop on the single tracked section, followed by a passing station that can hold exactly one train per-direction, followed by another stop on a single tracked section again.

We have installed a token block as an exit signal on the passing stations.
Trains will reservate the whole single tracked section up to the next platform. If the platform is not yet cleared, because that train got delayed, we will have to wait!
This setup will securely run deadlock-free.

Now, as time passes by and new fancy stuff like TCB gets available, we decide to change lines setup to TCB, using directional reservations.
That means, we have to install a signal at each platform in both directions, otherwise we would fall back to drive_by_sight at the first station.
Further, we need to ensure the first signal triggers a directional reservation and the directional reservation gets terminated by a one way sign before the next single tracked section.

Our trains will now create directional reservations to the next passing station.
We start on a double tracked section outside of the above described setup. The station in the single tracked section is clear, so we enter that section. After stopping there, we will attemt to continue to the passing station, but it's still occupied by some delayed train.
Unfortunately, there is also a delayed train waiting in the opposite direction of the passing loop and just behind, a train in the single tracked section waiting for clearance of the oppsoie direction passing station.
We just got a deadlock.

The same does btw. also apply generally to single tracked sections that are cut into smaller blocks to allow trains following up. When there are no intermediate stations, we can however freely decide to not cut them into sections i.e. not use directional reservations there at all.
In case of intermediate stations, we are forced to cut that section into blocks between stations.

Mariculous

#5
And here we are again.

B-B is in that era where the pakset has retired modern modes of token block aka RETB, so the only way to properly signal single-tracked sections with stops on them is to use token blocks from the 1800s

In the real-world, on many of such RETB signalled sections, ERTMS was installed.
This often is not an option in simutrans due to limitations in that working method.

Others kept operating on RETB, which was later on modernised as the original radio frequency was not usable by that system anymore. This is known as RETB next gen.

Is there any solution to this, as after the retirement of RETB ingame, we still have a situation that is bad in both terms, gameplay and realism.

The easy fix to this is still open: https://github.com/jamespetts/simutrans-pak128.britain/pull/85
The hard fix is a rather major coding project to get ERTMS capable of signalling such situations properly.

Edit: To understand why sometimes modern ERTMS with directional reservations is not an option, see the attached image.

There are 6 blocks on the city viaduct. If these are filled up due to a spike in traffic, the next vehicle will enter the single tracked section and cannot leave it.
The only signalling mehod available in simutrans extended that can reliably handle such setups is token block.

The setup in the image has locked twice, so I have increased the number of (virtual) blocks using moving block signalling, but that doesn't really solve the issue, it only increases the number of vehicles required to lock this.
Bear in mind that some passing loops are much smaller than the city viaduct of Darey.

Matthew

In Simutrans terms, I can't see any difference between traditional RETB and RETB-NG. They use different radio frequencies and equipment models, but we don't (and shouldn't!) simulate that level of detail. Since RETB is still in use in Scotland, I support removing the retirement date.
(Signature being tested) If you enjoy playing Simutrans, then you might also enjoy watching Japan Railway Journal
Available in English and simplified Chinese
如果您喜欢玩Simutrans的话,那么说不定就想看《日本铁路之旅》(英语也有简体中文字幕)。

jamespetts

Apologies that I have not had the time previously to consider this in detail. I do not think that I was aware when I implemented the RETB retirement date of the RETB-NG project. In light of that, I have added some new RETB-NG stop boards, which do not have a retirement date, and removed the retirement date for the RETB cabin. The new stop boards feature the blue light shown on modern RETB boards, but are more expensive as a result.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.