Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Locked thread
Pinball Jizzard
Jun 23, 2010

spamman posted:

Interesting post. Didn't really know much about how the UK did their level crossings.

I'm a train control systems engineer (that does some pure signalling stuff occasionally too) down here in Victoria and we don't use those LIDAR or RADAR crossings. A lot of predictors generally (in areas that aren't surrounded by other signalling infrastructure), but they're starting to get ripped out now too because of track contamination issues. A lot of axle counter crossings going in in regional areas (which lead to all kinds of axle counter reset issues). The big focus in the Metro areas is on grade separations though.

This is the bit that interests me here. I live and work in Aus in the signalling trade. What kind of contamination are you seeing with your predictors causing them to fail? Generally there are systems you can use to ensure good readings even on the moat disused lines (in WA there is a GCP4000 that operates well on a line that sees one train per year).

Also what issues do you face with regards to resetting on remote crossings? Locals traversing the axle counters would be about the only situation that would be of any issue where you'd leave a track unable to non co-op reset.

Adbot
ADBOT LOVES YOU

Pinball Jizzard
Jun 23, 2010

spamman posted:

In regards to track contamination, particular grasses grow over the tracks to the point where they won't drop, and there are also areas where there are millipede breeding seasons (I kid you not) where millions of them go over the tracks, get smooshed into a paste which then stops tracks from dropping.

In terms of axle counters, bored kids in the country have figured out how to make heads think there has been a count and drop level crossings.

Have you done any testing with DC shunt enhancement panels? By feeding a 5v constant DC feed you can generally break through small portions of rail covering to provide a better shunt. The other option to consider on those lines might be using Jeumont Schneiders (or TI.21 depending on your upbringing). It's hard to not significantly shunt a 500V pulse but you'd end up having to sacrifice distance and put in multiple tracks on each approach.

With axle counters and kids you're pretty much hosed unless you can build a plastic housing to stop it. Should be possible with most Frauschers and Thales but not so with Siemens given their size. Getting approval for it on the other hand...

Pinball Jizzard
Jun 23, 2010

Bozza posted:

The Fraucher ones? I like those they're pretty neat but from a maintenance technician perspective I'd prefer the Azlm types.

Axle counters are the future though, especially now we've got Engineers Possession Resets sorted.

The Thales Azlms are good don't get me wrong but the Frauschers are a maintainers dream. They are effectively now just plug and play once setup. If you drop them off the rail for track works you can throw them on, ensure the height is right (which 99 times in 100 it will be because of how the claw is fitted) and then have it automatically recalibrate.

Add into that the Frauscher development into high rail detection being at the best level it ever has been and I prefer them over all other types.

The big downfall and the reason we won't see full use of axle counters will always be the lack of broken rail detection, we as an industry haven't yet tried the workarounds to see how well they work. For instance a DC track (or high gain coded tracks) can work under an axle counter as a non-vital broken rail checking system.

spamman posted:

The option they've used so far is HVI track circuits in small doses. Not sure actually on the axle counters (like I said, I'm normally train control so I wasn't too involved with it), from memory though the Siemens axle counters are the only ones type approved currently by this ARO.

Which ARO do you mind if I ask?

Pinball Jizzard
Jun 23, 2010

If the distant lines are the Maryborough line then they are definitely Siemens.

I hasten to add that I'm incredibly sorry to hear that you work for V/Line.

Pinball Jizzard fucked around with this message at 02:10 on Aug 26, 2015

Pinball Jizzard
Jun 23, 2010

spamman posted:

I don't actually work for them but rather work for one of those millions of firms that does stuff for V/Line.

Although yeah, from what I've heard the Maryborough project basically killed Ansaldo's attempt to get into Victoria.

The entire RRL line is also Siemens too. ACM100s or something. Not sure what they used on the Bairnsdale project though .

Lets not discuss Maryborough!

I thought Siemens (Invensys) at the time were going to have a Frauscher trial site on RRL but I could have been wrong.

Pinball Jizzard
Jun 23, 2010

Bozza posted:

The new K types are like that for the AZLM, two bolts rather than the 15,000,000 parts of the old H type to get them calibrated properly.

We had the same hand wringing in the UK over tracks vs axle counters, track engineer ended up having to do a long rear end report called a Rail Integrity Management Plan which basically listed every bit of track pre 1979 (I think, I'm a signalling designer...) and how it was gonna be dealt with. Ended up our asset management guys thought this was just pissing away cash as we have fairly good records for dealing with rail breaks and most tracks here actually don't provide that robust broken rail detection, so it's a bit of a red herring.

Professionally (and personally) I don't think it's the job of the S&T equipment to monitor p-ways asset to the detriment of their own. Not sure how well that'd go down if I dropped that bombshell at work...

I've only had experience with the SK30H's combine with the ZP30H trackside units. Whilst calibrating wasn't hard; it also wasn't just manipulating two switches from the location as it is with the Frauschers. I personally just see the Frauscher as the current industry leader with regards to what we actually want from axle counters.

I do believe that broken rail detection is partly our responsibility given have the ability in some cases (not all given the way in which track circuits work) to give advanced warning of a major rail break. It should never come at the cost of proper P-Way checks but we can't just throw up our hands and say it's not our problem at all. If we can put a track circuit in under an axle counter section as a non-vital check then we should be doing it for that one in a million chance we stop a passenger train derailing.

Originally I'm from the UK and I fully understand how important protecting the passengers is to the industry there. It's just a minor addition with minimal development of required standards.

Pinball Jizzard
Jun 23, 2010

Bozza posted:

I must disagree, if you're providing tracks you're providing block joints which are much more likely points of failure than plain line rail breaks.

You're removing loads of significant points of failure and it makes your traction return bonding much more simple, which means that your overall systemic failure is reduced.

One of the big talking points going on at the moment is how to deal with failures (and SPADs) without it becoming a total job stopper as we switch from a fully interlocked signalling system which fails to a uselessly safe state to blokes on phones working via instruction to try and get stuff moving again - what is more risky?

I'm sure we could have a multi page debate about this though as its a bit of a hot topic at the moment. Management of overruns has just been totally overhauled in the data prep world following similar logic as we used to remove the 'inherent' rail break detection of track circuits.

When did you move over to Aus if you don't mind me asking?

Not every track requires a block joint and not every axle counter section would require it's own joints. CSEE/UM71 tracks with tuned loops could be used for the task; or you could cover a kilometer with only two joints and provide what you required. This way you reduce overall systemic failure points but also maintain current levels of broken rail detection. Stepping up to coded tracks you can run into the tens of kilometers.

As for moving block systems and dealing with "SPADs" (exceeding limits of authority); I can tell you the situation isn't much better. Instead of having an interlocking in an area, our current method of making things work is to make each individual train an interlocking of its own (in effect). Due to that the train can't just move under the authority of a bloke on the phone when something goes wrong. Sadly I can't go into much detail on the hows and what nots but I can say I had a very interesting experience of a failure situation on a train fitted with in cab signalling.

That's just the mechanical side of it. Inherently we have to accept there's going to be more risk involved when a driver has to take responsibility for failures which traditionally were locked by the signalling system. Without fixed points dictating safety we're eventually going to have an accident when someone moves too far forward (or backwards) after a failure, something that to a reasonable extent was protected by signalling principles and equipment. To put a train in imminent danger he would need to exceed his overlap or when talked forward pass the next signal also at red. With moving block or any in-cab system; should there be a failure the driver mught have innacurate or no indication on where exactly he is in relation to where he should be if the system on board is gone. Something we'll design to protect against but not removing the issue completely.

I first came to Aus in 2009, moved full time in late 2010. I've worked in Aus/NZ since.

Pinball Jizzard fucked around with this message at 13:29 on Aug 27, 2015

Pinball Jizzard
Jun 23, 2010
S&Cs aee a different matter completely. Given they make up such a small proportion of your overall line. I agree that the idea becomes kind of broken at that point for having tracks under the axle counters.

I worked on the Cambrian lines that you mentioned earlier in the thread as well as another level 2 system here (using L15K ATP) which was being upgraded. For me the problem is complexity. The UK is a far too complicated in many places to allow level 2 to work easily. Level 2s by definition need to be run on simple railways, it's difficult to convey simply to a driver his exact route when there's a lot of crossings and lines and all he has is a screen with limited output. Think of how many transponders you'd need at somewhere like Bristol Temple Meads never mind New Street, Paddington or Euston. How do you convey to a driver which line he's going to be on when there's 14 lines he could potentially end up on? Never mind in a failure situation where a level 3 alarm causes the emergency brake application and he needs a transponder (or two depending on standard) to allow full functionality to return. There's a good reason why in many metro projects worldwide ETCS has been removed from the scopes until a later date. I am glad that Network Rail is pushing development though; one of the reasons I moved here was because the British industry had become quite stagnant.

Pinball Jizzard
Jun 23, 2010

spamman posted:

Wow, that's monstrous. Is the hard coding of SPADs into the interlocking data a feature of the interlocking that's used, or is it for compliance with local industry standards? I'm not sure I've ever seen that before, SPADs here are detected at a non vital level.

SPAD detection is a non-vital alarm yes. But SPAD protection is in the interlocking circuitry and is definitely vital. The most basic example is the overlap of a route. This is a vital level protection in case a train SPADs the signal. The overlap distance is calculated based on line speed and train braking distances and stops a SPADing train from ever coming into contact with another train route or SPADing train. We have a lot more protection levels which aren't as obvious and a bit more complex at the interlocking level (flank protection, swinging overlaps, approach locking etc etc)

Pinball Jizzard
Jun 23, 2010

Bozza posted:

The current sway of the debate is that it's not relevant that the driver knows where they're going. It's a simple statement that has absolutely huge implications for how we operate as a network as it changes some real fundamentals.

That's a massive statement to make and a dangerous one at that.

Pinball Jizzard
Jun 23, 2010

Bozza posted:

It's one that's fundamental to the system - ETCS inherently does not provide routing information (though we are currently looking at proving text messages before major diverging routes eg Airport Jn at Stockley which is route barred to non electrics towards Heathrow) so you need to change your entire operating model to one where the driver has gone from having route knowledge to one where they don't.

That's the big issue for me with complexity though. On a simple system a driver knows what is happening and where he's going based on the limited information they have, on a complex railway that information is important but difficult to learn instinctively. For instance (and perhaps a very fringe situation). In the north west of Australia there's a hill which the railway is built upon which requires the drivers to be prepared for roughly 10km before going up and down. If you come from the top too quickly the brakes can't stop you and if you approach it too slowly from the bottom you're going to end up going back down. The level 2 system doesn't cater to that it just shows an exit speed and LoA; the only information the drivers have is from a Driver Assist screen (which shows rail profile). What if you remove route knowledge from those drivers?

As an aside, do you know if NR is planning on developing a driver assist system or driver strategy engine for use in conjunction? Not particularly signalling but it does feed off and into development.

*edit* There's a hill in the western region I believe near Gloucester with a similar profile. I know that they have to have to use a banksman if trains are over a certain length to get up it.

Pinball Jizzard fucked around with this message at 14:23 on Aug 28, 2015

Pinball Jizzard
Jun 23, 2010

Bozza posted:

Defined within the speed profile (supposedly) but I take your point. This happened when the ETCS went in on Cambrian, the journey times got longer as the system didn't allow the train to ever go above 40mph for long stretches, as defined in the sectional appendix, but drivers had been speeding to make it

You see that's the thing though. Unless you specifically set the speed limit low to go down a hill (increasing headway) you don't have it set in the system and in many cases that's just route knowledge, we both know how important headway is on a mainline and there's not many good workarounds to it. Going beyond the speed limit to go up a hill is a problem when you use a protected system similar to ATP where excessive speed leads to an alarm and ultimately an emergency brake application.

These are complexities you find (like the Cambrian line) during implementation and trust when I say the impact can be incredibly severe; which I'd like to discuss but I'm unsure if I can. What I can say is that because of the type of traffic and frequency of traffic on the Cambrians, there was never going to be an incident of major note.

I'm sorry for being so cagey about real world experiences with these kind of things, I'd love to discuss but it would most likely be breaking a few contract clauses and that would be a bad idea.

Pinball Jizzard
Jun 23, 2010

Bozza posted:

Trying not to hang myself working it out at the moment...



Effectively where there is a primary form of protection of a route (catch points in the overlap, facing points locked normal) against other routes there is no need to replace the other routes. In effect a SPAD will be less restrictive against indirectly opposing routes.

Pinball Jizzard
Jun 23, 2010

Bozza posted:

Pretty much, it's designed to allow the signaller a bit of breathing room if a SPAD occurs as it auto protects the potential conflicts at the moment of detection whilst leaving all other signals alone. It's less crude than a Signal Group Replacement Control which puts all the signals in an area back which is what the signaller should hit of a SPAD alarm sounds.

However, once you get into the detail of the layout it becomes a bit more tricky.

It's been giving me grief for weeks because both myself and the data/systems designs guys at Siemens are basically making it up as we go along as the standards are still maturing ie are poo poo.

From the quick read I've done if you lock the flank protection you're going to put every route outside the POPG at the cost of less availability to the signallers; are you allowed in the UK to lock flank protection once a train has passed "x" point where "x" is within braking distance of the signal (thereby only activating the lock when it becomes critical in the event of a SPAD)?

What system are you designing this for?

Pinball Jizzard fucked around with this message at 07:28 on Nov 23, 2015

Pinball Jizzard
Jun 23, 2010

Bozza posted:

Time of operation locking? Yeah it can be provided if required but not often.

Worth noting that modern UK standards say you don't detect facing points in the overlap, finally catching up with what the Western Region has done since forever.

It's going in a Siemens Westlock.

If you did detect and not Time of Operation locking, you'd be looking at effectively tie overlap points locking in with the approach locking. It would provide stronger flank protection, meet the primary overlap protection group standard and only slightly inhibit signalling availability as a cost. The idea being that as soon as a train is within the lookback of the end of route and points in the overlap would be locked thereby only having SPAD protection for that route on directly opposing routes.

It does mean going away from the standards of the Great Western (and now the national standard) but would make the design and principles testing a lot less involved. This way you're not going to have lots of conditional requirements depending on points being moved in the overlap changing which routes are affected in the POPG.

I do get the feeling I'm reading this wrong though because that seems too simple.

For those reading that might be interested. Time of operation locking applies to points within the overlap which can "swing" after the route is set. These points are free to move all the way up to the point where moving may potentially allow a train to cross them whilst moving. It's simply worked out by using the linespeed to work out the distance a train travels per second and then taking into account the worst case operation time of the points (per standard). For instance; at a linespeed of 36km/h a train travels 600m per minute or 10m per second; if the operation time of the points is within 15 seconds then the points need to be locked when a train is at closest 150m away.

*edit* I could potentially have thought about it and work out the circuitry required. Iirc Westlock is written in ladder logic? I remember Westrace I and II are. But i could certainly produce something in boolean for it.

Pinball Jizzard fucked around with this message at 13:49 on Nov 23, 2015

Pinball Jizzard
Jun 23, 2010

Bozza posted:

Westlock is ladder logic but the data prep is done in yer bog standard SSI data constructs with a few specials.

I'll have a think about the rest when I get some free time however this bit here is confusing me. Are you designing the ladder logic by writing it as SSI data constructs? Or are you writing this stuff for a SSI and a Westlock is going to sit above it?

Pinball Jizzard
Jun 23, 2010

Bozza posted:

I'll be doing no such thing, don't dirty my hands with data prep as I'm a schemes designer (GRIP0-4 if you know that process)

I think I may have confused Smartlock (the SSL/Alstom product) which talks pure SSI with the Westlock which doesn't. No idea about how the data constructs work as that's contracted out but think it's ladder logic. I'll ask my mate if I see him at 5 a side this week :)

That would make more sense. I had my head in my hands at the thought of ladder logic being designed first in SSI data!

And such a narrow field. The UK is far too easy nowadays giving the luxuries of a single role! I am unaware of the GRIP0-4 process having not been in a UK design house since 2001 or so.

Pinball Jizzard
Jun 23, 2010
Which is the thing that gets me. Why would you want to convert SSI data to ladder logic when in basic terms SSI data is a derivative language of ladder logic?

It's like translating English to French and then back to English.

Pinball Jizzard fucked around with this message at 04:39 on Nov 24, 2015

Pinball Jizzard
Jun 23, 2010
I understand why the tools were developed and in many ways it'd save a whole lot of design time. Transferring ASCON (for instance) data into the individual circuits would be easy enough but it's the fringes that would be dangerous. SSI data has many struggles with regards to timing issues and swinging overlaps (to name two of the glaring ones I've experienced), how you'd translate that well without enormous amounts of checking or redesign after principles tests I'm not comfortable with.

Pinball Jizzard
Jun 23, 2010
I am, but I'm a tester and not a designer. 2009 I left the UK to come over and see something different. I do a bit of design (and everything else) whenever I'm needed to.

There's not the jobs here at the moment which has seen the reverse happening and a lot of people are now heading to the UK. The mining industry downturn has seriously harmed the industry to the point where we're losing people all over.

Pinball Jizzard
Jun 23, 2010

Brovine posted:

The theory is that instead of allowing a train to go through a red signal and into the path of another train, you stop it safely by dumping it off the track before it conflicts with anything else. Particularly with runaways; when they were first installed most goods wagons had only hand brakes and might occasionally go for a wander of their own accord.

It's neater if you've got the space to build effectively an extra siding to send it down instead of just dumping it in the ballast, but both work.

Very rarely do you have a set of mainline catch points. The cases that I'm aware that they are installed are for run backs on steep hills (Outside Gloucester for instance). More often than not, sidings will have catch points just beyond either exit signal to prevent run-away wagons or trains.

Interlocking and line separation are the primary forms of mainline defence; with the addition of AWS and TWS to onboard systems to further prevent it happening.

Pinball Jizzard
Jun 23, 2010

Endjinneer posted:

Bozza's a signalling engineer. Politely, they're like the witch doctors in the railway engineering tribe. They traffic in a world that touches only lightly on our own, concerning themselves with the rites of flank protection and the proper ways to appease the sighting committee. Something as mundane as a collapsing wall wouldn't even register.

I sadly don't read this thread enough. This is perfection.

How is the UK faring right now with larger scale signalling projects? I know Thameslink is still underway, has Reading finished now? Is there any other large scale resignalling of note?

Pinball Jizzard
Jun 23, 2010
Gentlemen, gentlemen. When discussing the western region of the U.K railway, you should always remember to prefix it with "Great"

And on another note. How extensive is the OLE rollout in the Great a Western? Is it Paddington to Bristol mains and reliefs only or are they including major branches? Any plans to electrify Bristol to Plymouth as well for instance?

Pinball Jizzard fucked around with this message at 22:55 on May 9, 2017

Adbot
ADBOT LOVES YOU

Pinball Jizzard
Jun 23, 2010

Bozza posted:

When I worked for Western route in NR we were explicitly told not to refer to it as Great Western. So we used to allude to it instead 😂

gently caress ya rules.

Branches were planned originally when I was working on Crossrail, so Windsor and Bourne End/Marlow but not sure if it was just pie in the sky. Imagine that's been shelved.

Bristol South resignalling is electrification ready I believe but until Exeter and Plymouth PSBs are resignalled with something other than relays it's a non-starter. Think the plan is currently electric to Temple Meads, Swansea and the south Wales valleys. Everyone else gets bimode.

Not even sure how much of the Westbury diversion is going to be wired...

I sadly don't think there's enough benefit to resignalling Exeter beyond life expiry issues, Plymouth on the other hand needed it 15 years ago, I can't remember specifics but I remember the approaches were pretty testy, and the TDM system failed on a weekly basis (from my very old memories of the place).

Cardiff and Swansea were electrification ready when those jobs were done weren't they? I know Cardiff would have been about 10 years ago when it was resignalled.

  • Locked thread