Signalling & CBTC

[BrizCommuter]

colinw: This topic has been split from the MBRL thread

Why is it that other countries can easily run 35tph on tracks and we cant?. Surely it must be time to upgrade the antiquated signalling system currently in use.

The highest frequency on any suburban rail system is 30tph (Paris RER A, Munich S-Bahn). That is with advanced signalling, and no major infrastructure constraints such as flat junctions. The CityTrain network has neither of these advantages, and putting them in place would cost $$$. I'm not aware of any other Australian suburban rail system currently running more than 21tph.  

There are a few metros that run up to 38tph (Paris, Moscow), but they are all self-contained, fully grade segregated, with end to end services, advanced signalling, with no infrastructure constraints.

[colinw]

Those trains will be on a moving block.

There's no moving block on London Underground yet. For the most part it is a conventional (and somewhat antiquated) signalling system with train stops and trip cocks to prevent SPADs.  The Jubilee line is in the process of being equipped with a CBTC system that uses moving block, based on active reporting of train position, and I believe the Victoria line is to receive this as well. If you want to see moving block running you need to visit the Docklands Light Railway.

To achieve metro level headways you don't necessarily need moving block.  CBTC systems typically use a concept called "virtual blocks" which partitions each physical block (as delineated by lineside signals & track circuits) into a sequence of shorter "virtual blocks".  CBTC based trains actively report their position to lineside equipment via a bidirectional comms link, and in return receive movement authorities which permit them to occupy virtual blocks. The system then presents these movement authorities to the driver via in-cab signalling which overrides the lineside signals (it is valid to pass a red if the movement authority permits). That way equipped trains can be run at headways shorter than would be permitted based on the physical blocks alone.

Most European (and many Chinese, etc.) metros are using CBTC (ATP + ATO) systems which still do not have moving blocks.  They mostly behave as described above. The limit with such systems is around 2 minute headway / 30 TPH per track.  To go to 1.5 minute headway (40 TPH) or better will require moving block.

cheers,
Colin

[#Metro]

What about the cost for both options and the east of transition to it?
40 tph would be a dream!

[colinw]

To do the whole CityTrain system I imagine we're looking at a project in the hundreds of millions to billion range.  You're talking about quite a bit of train carried, lineside & track based infrastructure: track mounted transponders (aka "balises"), various kinds of lineside equipment (block controllers, radios, interfaces to the interlocking), equipment cubicles in each EMU, in cab driver's interface.  Then there's the cost of systems engineering & configuration, probably some software tailoring, a whole heap of certification & testing leading to the safety certificate - the list goes on. A retrofit to an existing railway is more expensive than a greenfield installation on a new line or a metro.

Moving block is very much more difficult to implement & certify than CBTC with virtual block. So far the track record of projects to implement moving block is one of late delivery, cost overrun, and even outright failure.  The first attempt at moving block on the London Underground was a complete failure, and watered down to fixed block due to technical problems. Likewise, ERTMS level 3 with moving block remains out of reach due to cost & technical problems.

Another critical aspect with these systems is BRAKING PERFORMANCE of the trains.  The ATP component of a CBTC system enforces movement authorities by calculating a series of "braking curves" to predict where the train will stop if a service (or emergency) brake application occurs. When a train approaches a limit of authority, it crosses a series of ATP calculated braking curves which escalate the response of the ATP from warning through to service and even emergency brake application. If ATO equipped the ATO uses the same calculation to determine when to commence braking. With better braking performance, headways can be reduced as the movement authority can allow a train to follow more closely on the heels of the preceding service.

A further factor here is gradients.  Steep down grades extend the braking curve and hurt headway, i.e. you cannot permit as close a headway on a 1 in 30 down grade as on a 1 in 50, particularly if there is a stopping point (station) at the base of the grade.

In my opinion, a CBTC system with fixed blocks, virtual blocks for shorter headways, and retaining lineside signalling for non-equipped trains is the way to go. This is a very stable & mature technology in use on dozens of railways worldwide, and provides a path to robust 30 TPH operation if designed properly.

The fact that my employer is one vendor of such systems may influence my viewpoint somewhat.  

[#Metro]

And what about all these signal faults and delays? Can anything be done about that?

[colinw]

Can't say, I don't get into the signaling side of things much, the interlocking & signaling system is just another communications interface as far as the systems I work on are concerned.

[Fares_Fair]

And what about all these signal faults and delays? Can anything be done about that?

I am a firm believer in redundancy of important network systems.
I also understand that it is difficult due to the safety aspects involved with signals and their failures.

Are there any retired semaphore (flag waver variety) operators around ?
Give them direct contact with 'Control', and voila !



Regards,
Fares_Fair.

[colinw]

The headways that Sydney & Melbourne achieved with signal cabins and electromechanical interlocking was amazing.  And reliably as well.

I sometimes wonder if all this electronic gadgetry is actually a step forward in actual functionality at all (and I make my living peddling the stuff).

[Fares_Fair]

The headways that Sydney & Melbourne achieved with signal cabins and electromechanical interlocking was amazing.  And reliably as well.

I sometimes wonder if all this electronic gadgetry is actually a step forward in actual functionality at all (and I make my living peddling the stuff).

I reckon you could compare it to the pre-modern car.

In the good old days you had points and contacts (sound familiar) and spark plugs and carburettors.
If something went wrong you could always file the points, clean the contacts, clear out the carburettor jets and regap or replace the spark plugs. You could also push-start the car if it was a manual transmission.

In todays car.

You have a computer to run the whole show, coupled with (usually) a solid state ignition module.
These are extremely reliable for most of the time.
However, and it's a BIG however, when they do break down - you are stranded, with no redundancy available to allow for continuation of the journey. No other way around it, can't push-start, can't limp home on 3 out of 4 operational cylinders.

Sign of the times perhaps.

Regards,
Fares_Fair.

[BrizCommuter]

Those trains will be on a moving block.

There's no moving block on London Underground yet. For the most part it is a conventional (and somewhat antiquated) signalling system with train stops and trip cocks to prevent SPADs.  The Jubilee line is in the process of being equipped with a CBTC system that uses moving block, based on active reporting of train position, and I believe the Victoria line is to receive this as well. If you want to see moving block running you need to visit the Docklands Light Railway.

The Jubilee Line's Moving Block (Seltrac S40) is now in operation, but only on 2/3 of the line. The Victoria Line's signalling system is Distance to Go, instead of moving block.

[colinw]

They got it going?  Good!

Out of interest, what is the basic frequency on Jubilee Line these days?  24 TPH?

[BrizCommuter]

Many of the paris metro line run every 90 seconds in peak and the district/circle lines in london

None of Paris metro lines are timetabled to run every 90secs, just one line (ligne 13) with short 75m trains manages 95secs/38tph.

The District & Circle Lines in London currently run 28tph, and has trains timetabled at 120sec or 150sec headways.

[somebody]

In my opinion, a CBTC system with fixed blocks, virtual blocks for shorter headways, and retaining lineside signalling for non-equipped trains is the way to go. This is a very stable & mature technology in use on dozens of railways worldwide, and provides a path to robust 30 TPH operation if designed properly.

So, this is an in-cab signalling system, with fixed blocks and track circuiting?  Are the blocks shorter?  Does it give a speed sensitive "Distance to run?"

What effect would this have on freight, does it remove the limitation where speeds have to be restricted in suburban areas due to the signals being too close together for them?

[p858snake]

And as discussed many a time, assuming a realistic max capacity of 20tph on the suburbans, this cannot be done without either CRR, better core network signalling, or an approx. 71% decrease in services at Nundah and Toombul stations.

Could easily ditch the shornc trains and make them express for any stats that overlap the cab line...

Edit: now looking at the map, it would be only 3ish stations (not including eagle or the cbd) but that would still save about ~9ish (assuming 3min per station) minutes for the shornc trains

[somebody]

And as discussed many a time, assuming a realistic max capacity of 20tph on the suburbans, this cannot be done without either CRR, better core network signalling, or an approx. 71% decrease in services at Nundah and Toombul stations.

Could easily ditch the shornc trains and make them express for any stats that overlap the cab line...

That would be a logical interim measure.  But do you think it would be implemented?  *NO*

[colinw]

In my opinion, a CBTC system with fixed blocks, virtual blocks for shorter headways, and retaining lineside signalling for non-equipped trains is the way to go. This is a very stable & mature technology in use on dozens of railways worldwide, and provides a path to robust 30 TPH operation if designed properly.

So, this is an in-cab signalling system, with fixed blocks and track circuiting?  Are the blocks shorter?  Does it give a speed sensitive "Distance to run?"

What effect would this have on freight, does it remove the limitation where speeds have to be restricted in suburban areas due to the signals being too close together for them?

Yes to all the above.

The "virtual blocks" are shorter than the fixed blocks imposed by the track circuits & lineside signals, i.e. each physical block is broken into multiple virtual blocks.

Supervision is based on distance to go until a limit of authority.

Effect on freight is mimimal.  Freight trains are not equipped, and thus run to the lineside signals and physical blocks and cannot proceed past a signal at red to occupy a virtual block.  The same limitation applies to trains with failed CBTC equipment.  Rather than being stopped, they enter a "driver responsible" mode where they run to the lineside signals only.

If a non-equipped train is occupying a physical block, because there is no position report available to calculate where it really is within that block, then no movement authorities will be issued to allow equipped trains to follow closely.  I.e. each non-equipped train occupies entire physical blocks based on detection by track circuits or axle counters.

[colinw]

Excuse clumsy ASCII art.

Non equipped railway:

"-----" = unoccupied track
"xxxxx" = occuped track
"|" = block boundary


|-------------------|-------------------|xxxxxxxxxxxxxxxxxxx|-------------------|-------------------|xxxxxxxxxxxxxxxxxxx|


CBTC equipped railway:

"|" = physical block boundary
"v" = virtual block boundary


|----v----vxxxxv----|----v----v----vxxxx|----v----v----v----|xxxxv----v----v----|----vxxxxv----v----|----v----vxxxxv----|


CBTC equipped railway, physical block occupied by non-CBTC freight train:


|----v----vxxxxv----|----v----v----vxxxx|----v----v----v----|xxxxvxxxxvxxxxvxxxx|----v----v----v----|----v----vxxxxv----|


What I am trying to convey here is the tighter headways allowed by the virtual blocks, as the trains can now follow more closely due to the shorter length of the virtual blocks.

The final diagram shows the effect of a non-CBTC freight train.  It occupies an entire physical block, holding a following CBTC equipped train out because it may be ANYWHERE in the physical block.  It also cannot follow as closely behind the preceding equipped train as it has no way of getting the finer grained movement authorities, it thus runs to the lineside signals at the block boundaries.

Non CBTC equipped trains eat track capacity!

All of the above is a gross simplification.  In practice the blocks are probably not of uniform length, and virtual blocks are shorter and more tightly clustered in low speed areas and approaches to stations.  If retrofitting onto a legacy signaling system the major thing influencing the design is the existing positions of signals, track circuits &c.

[somebody]

Ah, so the "Track circuit" blocks are no shorter, it's just the "Virtual blocks".  How are these determined?  Speed at block entry and time?

Is it too expensive to equip freight?  There aren't that many of them are there, and higher speeds would surely be possible.

[Zoiks]

If we are going to try and move to virtual blocks you might as well go the whole hog and do it with moving blocks.
Ill post the link where I got the cost for Melbourne when I get home

[#Metro]

I guess some questions:

1. How costly are each option
2. How fast would it be possible to get up and running
3. What is the maximum tph possible on each system
4. how easy/hard would it be to install into the current system.

Remember we still have problems with single track etc too, so we have to weigh that up against duplications and also fixing up the seating arrangements.

With fixed up seating you can get an extra 3500 passengers/hour through the system -- that's equal to three and a half extra trains!!!

[colinw]

Ah, so the "Track circuit" blocks are no shorter, it's just the "Virtual blocks".  How are these determined?  Speed at block entry and time?

Is it too expensive to equip freight?  There aren't that many of them are there, and higher speeds would surely be possible.

That is correct.  The physical blocks remain unchanged, but are partitioned into smaller chunks for the purposes of granting movement authority.  The actual determination of the virtual blocks depends on a bunch of factors - line speed, desired headway, location of physical blocks, etc.

Equipping freight is feasible, but requires additional functionality over a a passenger-only solution. With passenger trains the characteristic of the train are class-specific & fixed (e.g. deceleration, length, top speed, etc.).  You only need to be able to say "this is a 3 car EMU", or "this is a 6 car SMU".  With freight you need to be able to tell the system the characteristics of the consist that has been marshalled - length of train, maximum speed, service brake delay, service brake deceleration rate, etc., all of which varies depending on the specific consist.  Metro CBTC systems typically do not have that kind of flexible parameter entry, they just know what type of train, and whether it has been coupled to one or more other sets to make a longer train.

If we are going to try and move to virtual blocks you might as well go the whole hog and do it with moving blocks.
Ill post the link where I got the cost for Melbourne when I get home

At this stage, moving block is significantly more expensive & risky than fixed block. It is a less mature technology.  The ultimate aim with moving block is to dispense with the track circuits & lineside signals entirely, in favour of in-cab signalling, determination of train position from transponders + tachos & doppler radar (and possibly GPS), and active reporting of train position "back to base". Early days yet with these kinds of system, and AFAIK has not been done at all for a general purpose mixed traffic railway.

1. How costly are each option
2. How fast would it be possible to get up and running
3. What is the maximum tph possible on each system
4. how easy/hard would it be to install into the current system.

My opinions only:

1. Hundreds of millions for fixed block, maybe as much as a billion. Moving block - no idea, but wouldn't be surprised if it was a LOT more by the time it was stable.
2. Metro systems are basically "turnkey". In China it is being done with 18 month projects start to stop. To do CityTrain, I'd estimate 3-5 years.
3. CBTC with fixed block - practical maximum would be about 30 TPH I think, and 24-28 definitely achievable (subject to other considerations like junction layouts, dwell times, etc.)  Moving block the aim is to go as high as 40 TPH or more, I have seen one claim of 48 TPH.
4. I expect it would be a moderately complex project, due to the "legacy" nature of the CityTrain area. There would be a lot of work involved in interfacing the existing interlocking to CBTC lineside equipment.  There would also be a lot of design work to overlay an optimal virtual block schema over the current system.  Hence my estimate above that it would be a 3-5 year project.

Note also that a system like this, even moving block, won't help you much if there are poorly designed junctions, single track sections, etc.  No matter how good the signaling and how close the headways are, something like Merivale bridge + the Roma St junctions is going to be a choke point anyway.  Ditto for single track sections like Cleveland line beyond Manly.

[#Metro]

This is a very interesting discussion. But can be break apart the last few posts into a new thread on signalling?

That would be great.

TT

[somebody]

Equipping freight is feasible, but requires additional functionality over a a passenger-only solution. With passenger trains the characteristic of the train are class-specific & fixed (e.g. deceleration, length, top speed, etc.).  You only need to be able to say "this is a 3 car EMU", or "this is a 6 car SMU".  With freight you need to be able to tell the system the characteristics of the consist that has been marshalled - length of train, maximum speed, service brake delay, service brake deceleration rate, etc., all of which varies depending on the specific consist.  Metro CBTC systems typically do not have that kind of flexible parameter entry, they just know what type of train, and whether it has been coupled to one or more other sets to make a longer train.

Perhaps, but the current system is that all freight trains have to work to the worst freight train's standard AIUI.  You could have standards for half a dozen levels of freight train easily.

[BrizCommuter]

They got it going?  Good!

Out of interest, what is the basic frequency on Jubilee Line these days?  24 TPH?

Still 24tph in the peak direction (Eastbound am, Westbound pm). Will apparently increase to 27tph when Seltrac is operational over the whole line, and then 31tph in the near future.

[Derwan]

I don't see the advantage of having trains following more closely without significant infrastructure upgrades.

Let's look at the mains through the city.  No matter how close a train is behind another train, it still has to wait for the one in front to leave a station.  Then by the time it's done its stop it has a clear track in front.  No close-following required.

The only time this would be an advantage is if every platform set-up at every station was like the suburbans at Central (and no doubt like a lot of the platforms at Roma St but I'm less familiar with that station).  This would allow two services to be stopped at the same time and then depart more quickly after each other - then do the same at the next station.

So to see any improvement with better signalling in Brisbane, I think you'd need 2 more platforms at Central and 4 more at Fortitude Valley and Bowen Hills.  CRR stations would need a similar set-up.

[#Metro]

So... where do we start?

[mufreight]

The short answer is we dont.
The current systems fail largely due to a lack of equipment redundency.
The more complicated the system the greater the probability that it will fail.  All of these hi tec systems fail safe with the system shutting down and no movement possible until the system is brought back on line.
There are alternatives that can be instituted to enable services to continue when there are system failures but it would seem with the passing of time there is a lessening of the motivation to utilise these systems which while slower than the hi tec counterparts do enable the continued safe operation of train services instead of the current throw up out hands approach of the present train control management.
It is far cheaper, quicker and more efficient to clip and lock affected facing points and transit trains using a form of train order/ proceed order working than to shut the system down and call in buses.
The question arises is why not, is it lack of motivation by management or a lack of motivated and skilled trained train controllers and safe working qualified staff within the system?

[#Metro]

Oh, mufreight, I meant with regards to what Derwan was suggesting. Derwan seems to think that there is some limiting factor-- and this might be station dwells. No good having excellent signalling if something wasn't done about that is what that seemed to say to me.

[BrizCommuter]

Let's look at the mains through the city.  No matter how close a train is behind another train, it still has to wait for the one in front to leave a station.  Then by the time it's done its stop it has a clear track in front.  No close-following required.

The only time this would be an advantage is if every platform set-up at every station was like the suburbans at Central (and no doubt like a lot of the platforms at Roma St but I'm less familiar with that station).  This would allow two services to be stopped at the same time and then depart more quickly after each other - then do the same at the next station.

So to see any improvement with better signalling in Brisbane, I think you'd need 2 more platforms at Central and 4 more at Fortitude Valley and Bowen Hills.  CRR stations would need a similar set-up.

Actually, that is the point of advanced signalling, decreasing the platform re-occupation time, also known as the run out run in time (RORI). This is the time taken from wheel start of the departing train, to wheel stop of the following train. Track capacity is limited to the worst case RORI+dwell time+operating margin*. For example Bowen Hills northbound suburban is a 90 secs RORI. Adding 60mins for dwell (for crew change), and 30 secs operating margin = 3 minutes/20tph. If a signalling system such as SACEM (as used on Paris RER A) was installed, the RORI can be reduced to around 50secs. 50+60+30=140secs/25.7tph.

* assuming no other limitations such as terminus operations, single tracks, at grade junctions.

Extra platforms should not be required unless you need to allow for delay recovery time as occurs at Central.

[#Metro]

Look, I have noticed compared to other systems, the train spends huge amounts of time hanging around the platform. What is with that.

Park Road especially! And the doors take ages to close. What's going on?

The other issue that was raised, and a good one, was the station dwells at central to allow for recovery time from delays etc. I think those trains easily sit on the platform for up to 5 minutes it feels. Do other places all operate like this?

And just out of curiosity. Where will the dwell for CRR be???

[BrizCommuter]

And just out of curiosity. Where will the dwell for CRR be???

CRR is currently dwelling thanks to the ALP, long enough for it to be cancelled by the LNP. 

[Fares_Fair]

And just out of curiosity. Where will the dwell for CRR be???

CRR is currently dwelling thanks to the ALP, long enough for it to be cancelled by the LNP.  

In light of all the talk about CRR, can anyone say exactly what it will entail, how many tracks, where do they link in ?

Regards,
Fares_Fair.

[Zoiks]

From the WBTNI:

An upgrade to Moving Block signalling would require a new signalling system to be installed across the CityTrain network. To avoid significant disruption, the new system would need to be overlaid with the existing system for a transition period. Following the full commissioning of the Moving Block system, comprehensive testing and the conversion of all rolling stock to be capable of operating with the new system, the existing signalling system could gradually be withdrawn.
It is very difficult to estimate the likely cost of installing a new Moving Block system including all the in-train equipment and the line side transponders. A strategic level estimate was undertaken which suggested it would cost in the region of $1.5 billion to convert the existing Brisbane rail network (note that this is a very high
level estimate and should not be used for any detailed evaluation). It should be remembered that despite an initial high installation cost the ongoing operating costs associated with a Moving Block system are significantly lower than for existing signalling systems.

[O_128]

And just out of curiosity. Where will the dwell for CRR be???

CRR is currently dwelling thanks to the ALP, long enough for it to be cancelled by the LNP. 

Ive spoken to a lot of independent people and translink and they are completely adamant that crr is a go.

[Zoiks]

In light of all the talk about CRR, can anyonesay exactlty what it will entail, how many tracks, where do they link in ?

Regards,
Fares_Fair.

Are you serious? 
Basically... 2 tracks from yerongpilly through to albert street and on towards spring hill.. original plan was it too hook up with the ekka loop, but i was always hoping for a track up through kelvin grove

[Fares_Fair]

In light of all the talk about CRR, can anyonesay exactlty what it will entail, how many tracks, where do they link in ?

Regards,
Fares_Fair.

Are you serious? 
Basically... 2 tracks from yerongpilly through to albert street and on towards spring hill.. original plan was it too hook up with the ekka loop, but i was always hoping for a track up through kelvin grove

Of course I am, and it's a perfectly legitimate question IMHO.
New stations involved ?
Thank you for the answer.

Regards,
Fares_Fair.

[Zoiks]

In light of all the talk about CRR, can anyonesay exactlty what it will entail, how many tracks, where do they link in ?

Regards,
Fares_Fair.

Are you serious?  
Basically... 2 tracks from yerongpilly through to albert street and on towards spring hill.. original plan was it too hook up with the ekka loop, but i was always hoping for a track up through kelvin grove

Of course I am, and it's a perfectly legitimate question IMHO.
New stations involved ?
Thank you for the answer.

Regards,
Fares_Fair.

Sorry I just figured most people on this forum would know the basics of the biggest project planned for our network.

New stations at albert street, gabba, etc. Its probably easier for you to read it yourself.

http://www.google.com/search?btnG=1&pws=0&q=cross+river+rail

[Golliwog]

In light of all the talk about CRR, can anyonesay exactlty what it will entail, how many tracks, where do they link in ?

Regards,
Fares_Fair.

Are you serious? 
Basically... 2 tracks from yerongpilly through to albert street and on towards spring hill.. original plan was it too hook up with the ekka loop, but i was always hoping for a track up through kelvin grove

Of course I am, and it's a perfectly legitimate question IMHO.
New stations involved ?
Thank you for the answer.

Regards,
Fares_Fair.

Or up the top of the RBOT page, to the right of the Logout button is one labelled "CRR". Thats the official Cross River Rail page. Has all the plans etc up there.

[BrizCommuter]

This is the document you need to read:
http://www.crossriverrail.qld.gov.au/images/stories/reference_design/pdf_cross_river_rail_reference_design_overview.pdf

[Fares_Fair]

Thank you BC and Zoiks, I will.

I am heavily involved in the Sunshine Coast line issues.
I do need to familiarise myself with the CRR project.
With it's unfortunate delay, I will get the time to do so.

Regards,
Fares_Fair.