Class 345 progress

[kevin_roche]

Yes they've been much more reliable. In fact I'd hazard a guess that GWR's new 387 HeX units have probably had more (or at least equal) the number of ETCS failures, usually whilst starting at T5

Interesting that they happen at T5. As I understand it the Class 345 failures happened when switching from ETCS to TPWS/AWS at Airport Junction. Or were those just the more obvious ones?

I saw a class 345 coming out of the tunnel as I was going into Paddington. Today it was tfl rail branded so previously used on the main line.

Which probably outs paid to any rumours of different software for core testing and mainline usage

I think they swap them around occasionally and Mark Wild's report does mention testing the new software on the core first before using it on the TfL services. At that point, it is still on the Central Operating Section between the tunnel and Westbourne Park. I wonder if they are actually testing on the trains that will be going to Reading and Heathrow soon. I read somewhere that Trains were also being tested in and out of the tunnel at the Stratford end.

 

[Horizon22]

Interesting that they happen at T5. As I understand it the Class 345 failures happened when switching from ETCS to TPWS/AWS at Airport Junction. Or were those just the more obvious ones?

They certainly do happen at the transition too, but the continual T5 failures are something of a mystery. The location of the tunnels and electro-magnetic interference is genuinely being suggestion as a possible cause. This is because it is beyond the normal incidence rate as Paddington seems to have far fewer.

 

[rebmcr]

The location of the tunnels and electro-magnetic interference is genuinely being suggestion as a possible cause. This is because it is beyond the normal incidence rate as Paddington seems to have far fewer.

That would rule out my spidey-sense of it being a PICNIC issue.

 

[rd749249]

I've had it dump me in CTA. It happens everywhere in ETCS land.

 

[kevin_roche]

They certainly do happen at the transition too, but the continual T5 failures are something of a mystery. The location of the tunnels and electro-magnetic interference is genuinely being suggestion as a possible cause. This is because it is beyond the normal incidence rate as Paddington seems to have far fewer.

The signalling at T5 (ETCS) is quite different to that at Paddington (TPWS/AWS). For ETCS to work the train has to be able to receive a GSM-R data signal with the Movement Authority. No signal or poor signal means there is no Movement Authority. The issues with GSM-R at T4 were really bad and special measure had to be taken for it to work. GSM-R operating frequency is very close to the normal mobile phone frequencies and that can be a cause of interference I believe.

I understood that the issues with resets at Airport junction were with software, rather than lack of signal. In the reverse (toward heathrow) direction there is a short section of ETCS L1 where the MA is sent by the track Balise instead of GSM-R for the reason that the new session was expected to take a few seconds to establish. the word on the grapevine was that it wasn't actually needed in the end.

Edit: Fixed the spelling of ETCS above

 

[rebmcr]

The signalling at T5 (ECTS) is quite different to that at Paddington (TPWS/AWS). For ECTS to work the train has to be able to receive a GSM-R data signal with the Movement Authority. No signal or poor signal means there is no Movement Authority.

I believe GSM-R login is also required at Paddington, such that the train's reporting number is properly represented in the 'berth' — otherwise the signaller cannot set a route out of the platform.

 

[JN114]

I believe GSM-R login is also required at Paddington, such that the train's reporting number is properly represented in the 'berth' — otherwise the signaller cannot set a route out of the platform.

No, you have the logic backwards - the signaller being physically able to set route is in no way related to GSM-R setup at either end of the route. The driver has to input the correct/matching reporting number into GSM-R as is on the train describer, otherwise it will not register properly (it will come up “check headcode” or similar if wrong).

GSM-R login is required almost everywhere on the rail network, so that you can correctly receive and send calls to the signaller, including correct routing of emergency calls.

The signalling at T5 (ECTS) is quite different to that at Paddington (TPWS/AWS). For ECTS to work the train has to be able to receive a GSM-R data signal with the Movement Authority. No signal or poor signal means there is no Movement Authority. The issues with GSM-R at T4 were really bad and special measure had to be taken for it to work. GSM-R operating frequency is very close to the normal mobile phone frequencies and that can be a cause of interference I believe.

ETCS - European Train Control System

Loss of communication or poor signal will be indicated as such on the driver’s display, along with exceeded authority and full brake demand. This happens from time to time, but for the most part the issues nowadays are thought to be at a software level between the train and the RBC (Radio Block Centre).

HEx do anecdotally have most of their problems on setting up or starting off from T5; then software glitches while running under L2 (resulting in loss of Movement Authority etc); then fail to transition from L2 to NTC (National Train Control, AWS/TPWS). Fail to transition NTC to L2 seems very rare.

Whereas anecdotally for Crossrail I’d say most of their issues are the random glitches I referred to above for HEx, followed by fail to transition NTC to L2; followed by setup issues at T5; with fail to transition L2 to NTC is seemingly rare for them.

That both operators are having very similar sets of issues with very differing on train software architecture points to infrastructure not being quite up to scratch, if it were the trains I’d expect one or other architecture to work significantly better than the other. But they’re both failing about the same amount, in broadly the same ways - HEx have “caught up” Crossrail in only having a couple issues per day, compared to a dozen or more per day in the first weeks of operation.

There’s not been the extensive operating to T4 by either operator yet to judge if the known GSM-R signal issues have been resolved. That’ll come soon when HEx go back to 4 trains per hour.

 

[matt_world2004]

There’s not been the extensive operating to T4 by either operator yet to judge if the known GSM-R signal issues have been resolved. That’ll come soon when HEx go back to 4 trains per hour.

I am assuming the plan is to get 387s to terminal 4 at some point so they can see how they run when providing s Christmas Day service there.

 

[JN114]

I am assuming the plan is to get 387s to terminal 4 at some point so they can see how they run when providing s Christmas Day service there.

They’ve been many times already, during driver training most runs were to/from T4 - for example this trip from just this past Monday:- https://www.realtimetrains.co.uk/train/V40950/2021-02-15/detailed

 

[kevin_roche]

I believe GSM-R login is also required at Paddington, such that the train's reporting number is properly represented in the 'berth' — otherwise the signaller cannot set a route out of the platform.

I did not know that. Thanks for letting me know. Apologies for swapping the letters around in my earlier post. I fixed it now.

 

[Bigbru]

No, you have the logic backwards - the signaller being physically able to set route is in no way related to GSM-R setup at either end of the route. The driver has to input the correct/matching reporting number into GSM-R as is on the train describer, otherwise it will not register properly (it will come up “check headcode” or similar if wrong).

GSM-R login is required almost everywhere on the rail network, so that you can correctly receive and send calls to the signaller, including correct routing of emergency calls.



ETCS - European Train Control System

Loss of communication or poor signal will be indicated as such on the driver’s display, along with exceeded authority and full brake demand. This happens from time to time, but for the most part the issues nowadays are thought to be at a software level between the train and the RBC (Radio Block Centre).

HEx do anecdotally have most of their problems on setting up or starting off from T5; then software glitches while running under L2 (resulting in loss of Movement Authority etc); then fail to transition from L2 to NTC (National Train Control, AWS/TPWS). Fail to transition NTC to L2 seems very rare.

Whereas anecdotally for Crossrail I’d say most of their issues are the random glitches I referred to above for HEx, followed by fail to transition NTC to L2; followed by setup issues at T5; with fail to transition L2 to NTC is seemingly rare for them.

That both operators are having very similar sets of issues with very differing on train software architecture points to infrastructure not being quite up to scratch, if it were the trains I’d expect one or other architecture to work significantly better than the other. But they’re both failing about the same amount, in broadly the same ways - HEx have “caught up” Crossrail in only having a couple issues per day, compared to a dozen or more per day in the first weeks of operation.

There’s not been the extensive operating to T4 by either operator yet to judge if the known GSM-R signal issues have been resolved. That’ll come soon when HEx go back to 4 trains per hour.

Given Hex have purchased their ETCS kit from Bombardier though, would this mean it doesn't necessarily point to infrastructure not being upto scratch?

 

[JN114]

Given Hex have purchased their ETCS kit from Bombardier though, would this mean it doesn't necessarily point to infrastructure not being upto scratch?

The system on the 387s is wholly Bombardier.

The system on the 345s is Siemens CBTC emulating ETCS on the formers’ hardware.

They are wholly different architectures, but failing in very similar ways.

 

[Horizon22]

The system on the 387s is wholly Bombardier.

The system on the 345s is Siemens CBTC emulating ETCS on the formers’ hardware.

They are wholly different architectures, but failing in very similar ways.

Interesting that ETCS which is meant to be the future of railway signalling has had so many glitches on effectively a metro service and might pose further problems in the Central section. Not really a discussion for here, but the reliability will have to increase overall in future schemes.

 

[hwl]

The signalling at T5 (ETCS) is quite different to that at Paddington (TPWS/AWS). For ETCS to work the train has to be able to receive a GSM-R data signal with the Movement Authority. No signal or poor signal means there is no Movement Authority. The issues with GSM-R at T4 were really bad and special measure had to be taken for it to work. GSM-R operating frequency is very close to the normal mobile phone frequencies and that can be a cause of interference I believe.

I understood that the issues with resets at Airport junction were with software, rather than lack of signal. In the reverse (toward heathrow) direction there is a short section of ETCS L1 where the MA is sent by the track Balise instead of GSM-R for the reason that the new session was expected to take a few seconds to establish. the word on the grapevine was that it wasn't actually needed in the end.

Edit: Fixed the spelling of ETCS above

The ETCS positioning balises use 27Mhz which (wave length 11.03m) which is also used by GW-ATP. The tunnels having features in the quarter (2.76m) to full wavelength range is generally bad news as is the continued presence of GW-ATP.

Is it known whether the failure are loss of GSMR or balise related?

 

[kevin_roche]

The ETCS positioning balises use 27Mhz which (wave length 11.03m) which is also used by GW-ATP. The tunnels having features in the quarter (2.76m) to full wavelength range is generally bad news as is the continued presence of GW-ATP.

Is it known whether the failure are loss of GSMR or balise related?

I don't know if the current issue on the HEX trains is GSMR or Balise related but I understood that as of the change in HEX rolling stock that they were no longer using GW/ATP at Heathrow.

I may be wrong but I was told by one of the people working on ETCS infrastructure there that that was the plan. They other thing, that was also happening was the improvements to the software in the GSM-R receivers on the Class 345 which he indicated might solve the problems they had 6 months ago. Another of the Crossrail people told me that there was no problem with 27Mhz interference. That was more than a year ago. Unfortunately I have no more recent information other than what everyone has seen in the reports by Mark Wild.

 

[Horizon22]

The ETCS positioning balises use 27Mhz which (wave length 11.03m) which is also used by GW-ATP. The tunnels having features in the quarter (2.76m) to full wavelength range is generally bad news as is the continued presence of GW-ATP.

Is it known whether the failure are loss of GSMR or balise related?

The problem is generally at the "start of mission" phase from what I understand and some sort of loss of signal, but other than that it is beyond my technical knowledge!

 

[paulravuslante]

The problem is generally at the "start of mission" phase from what I understand and some sort of loss of signal, but other than that it is beyond my technical knowledge!

I previously had an issue at T5 where the train couldn’t connect to the RBC. On switching the train off and on again it then connected.

 

[kevin_roche]

The problem is generally at the "start of mission" phase from what I understand and some sort of loss of signal, but other than that it is beyond my technical knowledge!

I previously had an issue at T5 where the train couldn’t connect to the RBC. On switching the train off and on again it then connected.

Given these reports, this looks to me like a problem with GSM-R. It could be interference or it could equally be software since I believe the receivers in the GSM-R radios are digital devices with software defined radios (SDR) which could also be the issue.

SDR technology like any other software based thing can suffer from software bugs. I did wonder why there was a need for two different software updates on the train. Pure speculation but possibly one is an update to the radio reception.

 

[Horizon22]

Given these reports, this looks to me like a problem with GSM-R. It could be interference or it could equally be software since I believe the receivers in the GSM-R radios are digital devices with software defined radios (SDR) which could also be the issue.

SDR technology like any other software based thing can suffer from software bugs. I did wonder why there was a need for two different software updates on the train. Pure speculation but possibly one is an update to the radio reception.

GSM-R has been used in the tunnels for years though which means it is surely a software issue - and it is seemingly equally applying to both 387s and 345s.

 

[paulravuslante]

A H&S colleague of mine had previously mentioned the modems in the middle (coupled) cabs of the 387s had been switched off to improve reliability for all trains.

Also apparently the RBC was rebooted following a major issue at the end of January and this is also believed to have improved reliability.

The software upgrades on the 345s are for both ETCS reliability improvements as well as general software improvements and fixes.

 

[43096]

The system on the 387s is wholly Bombardier.

The system on the 345s is Siemens CBTC emulating ETCS on the formers’ hardware.

They are wholly different architectures, but failing in very similar ways.

You'd have to question the wisdom of installing Bombardier ETCS kit. Ask DB in Germany what they think of it!

 

[kevin_roche]

The latest project representative report for the period ending 12 Dec 2020 is now on the TfL website. https://content.tfl.gov.uk/project-representative-periodic-report-period-9-2020.pdf

It has the following:

Stage 2B Operations: Since the software release in early December 2020, FLU performance operating on the GWML has improved as a result of a significant reduction in ETCS faults. However, until a further software release is implemented in February 2021, we expect ATP failures to continue.

The response from Crossrail is:

Automatic Train Protection (ATP) failure is a subset of the European Train Control System (ETCS) faults that require the driver to do an ETCS reset to recover. This ATP failure subset has a fix included in the next software release for implementation in February 2021.

So that looks like the current problem is a software issue rather than interference.

 

[Gulf1159]

Liverpool Street platform extensions (16 and 17) should be commissioned over the easter weekend and 9 car operation is due to start on the east from Sunday 16 May 2021

 

[JonathanH]

Liverpool Street platform extensions (16 and 17) should be commissioned over the easter weekend and 9 car operation is due to start on the east from Sunday 16 May 2021

Presumably the 7 car units are swapped for 9 car units over a few weeks rather than a 'big bang' change given that the transfer of units has to be via the North London Line and there isn't loads of siding space available on the east side.

 

[Snow1964]

Presumably the 7 car units are swapped for 9 car units over a few weeks rather than a 'big bang' change given that the transfer of units has to be via the North London Line and there isn't loads of siding space available on the east side.

Is it possible to transfer the units via the tunnel (even if under a semi-manual signalling authority. Especially if slow lines through Stratford are closed

There is various updates in the Board committee papers for next week (See pages 40-42)

https://content.tfl.gov.uk/pic-20210303-agenda-and-documents-public.pdf.pdf

 

[JN114]

Is it possible to transfer the units via the tunnel (even if under a semi-manual signalling authority. Especially if slow lines through Stratford are closed

There is various updates in the Board committee papers for next week (See pages 40-42)

https://content.tfl.gov.uk/pic-20210303-agenda-and-documents-public.pdf.pdf

Theoretically yes, but it would require disruptive engineering possessions of the central core, as well as both the Great Eastern and Great Western to do it.

To expand a little -

If I recall correctly from various bits of weekend testing, it’s 3 of the 6 lines in the Paddington station throat; and the “electric” lines between Forest Gate and Bow that need to be taken as the signalling protection zone for transitions with the signalling not yet commissioned. On top of that you’d likely need the central core under possession as at present I don’t believe there is a software version that allows both running on central core section and under National Train Control (AWS/TPWS) or ETCS. So they’d have no signalling at all in the core section so would likely have to be low speed, 1 train at a time.

Whereas given a compatible Loco (of which ROG have many), it’s relatively painless to drag the sets between Western and Eastern “branches” via the NLL - it can be done at any time; and doesn’t impact other operators having to part-close the mainlines at the transition points to do it.

 

[matt_world2004]

If there is 9 cars on the Stratford end and the core is open. Wouldn't this allow the east end and the core end join up more quickly as part of the reason for a staggered opening was to allow 9 car conversions.

 

[rebmcr]

If there is 9 cars on the Stratford end and the core is open. Wouldn't this allow the east end and the core end join up more quickly as part of the reason for a staggered opening was to allow 9 car conversions.

I doubt there is any appetite to press forward with sooner use of signalling transitions than are currently planned.

 

[Snow1964]

If there is 9 cars on the Stratford end and the core is open. Wouldn't this allow the east end and the core end join up more quickly as part of the reason for a staggered opening was to allow 9 car conversions.

If I have read the new Investment Committee papers correctly, then the conversion from 7 to 9 car is stage 4A

If stage 1 was Liverpool St - Shenfield then sort of need to consider why the phase is 4A and not numbered say 1A or 1B

Theoretically yes, but it would require disruptive engineering possessions of the central core, as well as both the Great Eastern and Great Western to do it.

To expand a little -

If I recall correctly from various bits of weekend testing, it’s 3 of the 6 lines in the Paddington station throat; and the “electric” lines between Forest Gate and Bow that need to be taken as the signalling protection zone for transitions with the signalling not yet commissioned. On top of that you’d likely need the central core under possession as at present I don’t believe there is a software version that allows both running on central core section and under National Train Control (AWS/TPWS) or ETCS. So they’d have no signalling at all in the core section so would likely have to be low speed, 1 train at a time.

Whereas given a compatible Loco (of which ROG have many), it’s relatively painless to drag the sets between Western and Eastern “branches” via the NLL - it can be done at any time; and doesn’t impact other operators having to part-close the mainlines at the transition points to do it.

Ok, I only asked because I thought there was lack of siding space Eastern end, and it seemed easier to drive a load of trains to say Whitechapel, almost buffer to buffer in advance, then on changeover date move them forward to Stratford area

The alternative of about 20 trains (each way) needing to be coupled to barrier wagons, then shuttling back and forth on changeover date seems much harder. Even if it is easiest for irregular stock transfers.

 

[kevin_roche]

If I have read the new Investment Committee papers correctly, then the conversion from 7 to 9 car is stage 4A

If stage 1 was Liverpool St - Shenfield then sort of need to consider why the phase is 4A and not numbered say 1A or 1B

Stage 4 was to be the introduction of 9 car trains running from Shenfield to Paddington via the tunnel. It has been split into 4A and 4B. 4A is 9 Car trains to Liverpool Street using the lengthened platforms. 4B will be through the tunnel to Paddington.