Level crossing incident near Norwich new RAIB investigation

[hwl]

Err, presumably they'd have needed to tell NR they wanted to terminate at Ely instead of Cambridge?

But probably after the teams had been sent out...
and the teams not being arranged by the bit they are speaking to e.g. the signallers.

 

[O L Leigh]

Possibly the Orange ppl at Waterbeach are a red herring

Almost certainly, unless the other AHBs on the same stretch of line had also been taken under local control (Milton Fen, Bottisham Road, Dimmocks Cote, etc).

 

[w0033944]

Local press in this area reports seeing an email, sent yesterday by GA's deputy director and a union man that suggests that the problems with the 755s is that they have on-board flange lubricators, the grease from which is thought to be contaminating the tyre/rail interface such that the track circuits are failing to reliably detect the train. The solution, apears to be, in part at least, citrus oil:

https://www.eveningnews24.co.uk/new...glia-new-trains-problems-citrus-oil-1-6417275

Greater Anglia has revealed its secret ingredient to tackle the host of problems on its network this week - citrus oil.
More than 80 services are cancelled again today on its rural routes with the Norwich-Sheringham, Ipswich-Peterborough, Norwich-Cambridge, and Ipswich-Felixstowe lines particularly badly hit.

Greater Anglia has blamed the mass cancellations on "signalling problems".

But an email sent to staff yesterday shows that the rail operator is trying to fix issues with its new, £1.4 billion trains on the routes.

Greater Anglia brought in the new Stadler trains, called Class 755s, earlier this year, but they have been hit with a barrage of problems and are nicknamed "Basils" by staff because they have so many faults.

The dangers of the faults were exposed last week when it emerged the Rail Accident Investigation Branch is looking at how a new train was 0.25 seconds away from smashing into a car at Thorpe End level crossing on November 24.

It almost hit the car when the level crossing barriers went up too soon.

An email, sent on Monday by union organiser Nigel Gibson and deputy director at Greater Anglia Richard Packer, outlines the measures now being put in place on the new trains.

It includes treating the wheels with citrus oil to remove any "contamination" on them.

One of the problems exposed by the near miss at the level crossing is an issue between how the trains communicate with the rail system through sensors on the track.

The trains have also had their "flange lubricators" removed from their wheels.

They reduce friction between the train and track, but it is understood the lubrication system could have blocked the circuit at the level crossing which controls the automatic barriers.

The new trains are also being restricted on the rural routes and running at 20mph over some level crossings.

Network Rail, meanwhile, has changed the timings of the barriers at level crossings to stop a repeat and have put staff in place in case barriers fail.

It is not clear how long the disruption will last.

On Monday Greater Anglia's managing director Jamie Burles wrote to staff: "We had a tough week last week and we're not out of the woods yet."

 

[Edders23]

Are you seriously advocating a safety culture of trial and error? An error related to safety could be catastrophic and no rail operator would ever employ it because it is completely irresponsible.

That's the most shocking post I've ever seen on here.

No I wasn't advocating trial and error but it is the way a lot of things work in the engineering world the 737 max being a prime example. You cannot engineer perfection at the first attempt just try and get as close as possible

 

[Muzer]

So for this to happen would both the "entrance" treadle have to have failed AND the track circuit briefly failed wrong-side (detected no train when there was a train)? I can understand the crossing reopening in the event of a track circuit occupying and clearing, as that can happen at random. But I would be surprised if a crossing would reopen once the entrance treadle had been activated before the exit treadle does, as I wouldn't have thought you would get many false treadle activations? Could someone actually personally familiar with how this works please clarify this for me? Thanks!

 

[GB]

There are no treadles on this branch.

 

[Tio Terry]

There are no treadles on this branch.

Which rather points to a false clear for a track circuit. Back to the trains having problems with track circuit activation.

No doubt the RAIB will get to the bottom of it.

 

[ge-gn]

There are no treadles on this branch.

Not entirely correct. There are treadles, but not for track circuit actuation. Pedant out.

 

[w0033944]

Which rather points to a false clear for a track circuit. Back to the trains having problems with track circuit activation.

No doubt the RAIB will get to the bottom of it.

Are we jumping to conclusions, I wonder? Just because we've heard of the internal email regarding contamination doesn't mean that circuit activation is the problem. Maybe the predictors haven't been recalibrated for longer sets?

 

[GB]

Maybe the predictors haven't been recalibrated for longer sets?

There's a freight train that uses the branch 3 or 4 times a week so doubt that is the issue.

 

[LAX54]

Local press in this area reports seeing an email, sent yesterday by GA's deputy director and a union man that suggests that the problems with the 755s is that they have on-board flange lubricators, the grease from which is thought to be contaminating the tyre/rail interface such that the track circuits are failing to reliably detect the train. The solution, apears to be, in part at least, citrus oil:

https://www.eveningnews24.co.uk/new...glia-new-trains-problems-citrus-oil-1-6417275

Couple with the fact that the wheel contact with the rail, is miniscule, much less than normal UK stock.

 

[trebor79]

Couple with the fact that the wheel contact with the rail, is miniscule, much less than normal UK stock.

Is that tyre profile or wheel diameter, or both? The wheels do look like a smaller radius than most other stock, I guess that helps with a low floor design.

 

[w0033944]

Couple with the fact that the wheel contact with the rail, is miniscule, much less than normal UK stock.

Is that tyre profile or wheel diameter, or both? The wheels do look like a smaller radius than most other stock, I guess that helps with a low floor design.

Good point. Feeling somewhat guilty as a disabled person that the need for a low floor may have contributed to the problem.

 

[edwin_m]

Pretty sure container flats have smaller wheels, although they are limited to 75mph. The other difference is longer wheelbase bogies due to the articulation, which might make flange contact more likely on the tighter curves.

 

[trebor79]

Good point. Feeling somewhat guilty as a disabled person that the need for a low floor may have contributed to the problem.

Why should you feel guilty? No reason you shouldn't be able to use modern public transport.

 

[w0033944]

Why should you feel guilty? No reason you shouldn't be able to use modern public transport.

Many disabled people are used to feeling that way - anyway, that's OT. Presumably this problem can't easily be solved if loading gauges are to be maintained.

 

[eman_resu]

The predictors that we use down here have no settings related to train length, and are used by both multi locomotive heavy long freight trains, as well as short 2 car passenger sets.

Practice down here is to allow the crossing to time out* and recover if the no movement is detected after activation, or the train is detected at a stop. The crossing then activates immediately if movement is detected towards the crossing.

*However this is typically two minutes or more, therefore allowing a train that is detected and then subsequently 'lost' to traverse the crossing prior to the timeout and recovery of the crossing.

 

[LAX54]

As of this morning (13th) Normal working has been resumed with the 755's, although may take a while to get all up and running to the WTT again.

 

[MikeWM]

As of this morning (13th) Normal working has been resumed with the 755's, although may take a while to get all up and running to the WTT again.

Yes, I got the (late, but not due to losing time en-route) 0938 off Ely this morning.

All ready for Stansted services on Sunday then?

 

[w0033944]

As of this morning (13th) Normal working has been resumed with the 755's, although may take a while to get all up and running to the WTT again.

Great new providing the crossing problem has been sorted. GA cannot afford an accident.

 

[Belperpete]

The predictors that we use down here have no settings related to train length, and are used by both multi locomotive heavy long freight trains, as well as short 2 car passenger sets.

Practice down here is to allow the crossing to time out* and recover if the no movement is detected after activation, or the train is detected at a stop. The crossing then activates immediately if movement is detected towards the crossing.

*However this is typically two minutes or more, therefore allowing a train that is detected and then subsequently 'lost' to traverse the crossing prior to the timeout and recovery of the crossing.

Where is "down here"?

 

[Belperpete]

So for this to happen would both the "entrance" treadle have to have failed AND the track circuit briefly failed wrong-side (detected no train when there was a train)? I can understand the crossing reopening in the event of a track circuit occupying and clearing, as that can happen at random. But I would be surprised if a crossing would reopen once the entrance treadle had been activated before the exit treadle does, as I wouldn't have thought you would get many false treadle activations? Could someone actually personally familiar with how this works please clarify this for me? Thanks!

With conventional automatic level crossings, there are usually treadles at the strike-in points, in case the track-circuit fails to detect the train. However, with these crossing predictor track-circuits, there is no set strike-in point - the point at which the crossing is activated will depend on how fast the train is approaching. So, without a set strike-in point, presumably there is nowhere to put strike-in treadles.

It does look as though the train was initially detected, to start the crossing working, and then detection was lost causing the crossing to re-open. While it is undesirable for detection to be lost, the crossing should not re-open immediately, there should be a delay of several minutes to allow time for an approaching train to have passed clear. Either the train was going very slowly, or the timer was seriously short. I think it significant that NR have reportedly increased the timer settings.

An alternative explanation is that the train was never detected, that something else activated the crossing (e.g. a train on another track), and the crossing just happened to re-open as the undetected train arrived. However, there has been no mention of another train, and NR increasing the reset timers would not address this issue.

Personally, I am a bit uneasy about these predictors being used with AHBs. With AOCLs and ABCLs, at least the driver would get the flashing red light if the predictor closes the crossing late, or if detection is lost.

 

[LAX54]

With conventional automatic level crossings, there are usually treadles at the strike-in points, in case the track-circuit fails to detect the train. However, with these crossing predictor track-circuits, there is no set strike-in point - the point at which the crossing is activated will depend on how fast the train is approaching. So, without a set strike-in point, presumably there is nowhere to put strike-in treadles.

It does look as though the train was initially detected, to start the crossing working, and then detection was lost causing the crossing to re-open. While it is undesirable for detection to be lost, the crossing should not re-open immediately, there should be a delay of several minutes to allow time for an approaching train to have passed clear. Either the train was going very slowly, or the timer was seriously short. I think it significant that NR have reportedly increased the timer settings.

An alternative explanation is that the train was never detected, that something else activated the crossing (e.g. a train on another track), and the crossing just happened to re-open as the undetected train arrived. However, there has been no mention of another train, and NR increasing the reset timers would not address this issue.

Personally, I am a bit uneasy about these predictors being used with AHBs. With AOCLs and ABCLs, at least the driver would get the flashing red light if the predictor closes the crossing late, or if detection is lost.

From what I can gather this afternoon, is that 755's still have 'issues' on the Cromer, but if a 156 or 170 runs on the line, all is fine.

 

[trebor79]

With conventional automatic level crossings, there are usually treadles at the strike-in points, in case the track-circuit fails to detect the train. However, with these crossing predictor track-circuits, there is no set strike-in point - the point at which the crossing is activated will depend on how fast the train is approaching. So, without a set strike-in point, presumably there is nowhere to put strike-in treadles.

It does look as though the train was initially detected, to start the crossing working, and then detection was lost causing the crossing to re-open. While it is undesirable for detection to be lost, the crossing should not re-open immediately, there should be a delay of several minutes to allow time for an approaching train to have passed clear. Either the train was going very slowly, or the timer was seriously short. I think it significant that NR have reportedly increased the timer settings.

An alternative explanation is that the train was never detected, that something else activated the crossing (e.g. a train on another track), and the crossing just happened to re-open as the undetected train arrived. However, there has been no mention of another train, and NR increasing the reset timers would not address this issue.

Personally, I am a bit uneasy about these predictors being used with AHBs. With AOCLs and ABCLs, at least the driver would get the flashing red light if the predictor closes the crossing late, or if detection is lost.

Would be better not to have an automatic raise timer at all. That way if a train is detected and then subsequently disappears or isn't positively detected passing the crossing, the barriers stay down and the lights stay on. Ie it fails safe which is exactly what happens at treadle operated AHBs.
Raising after a set time to allow for a train having stopped might be appropriate for certain settings in America, where they seem to have marshalling yards running over streets. But I can't think of any UK setting making it a necessary or desirable feature.

 

[Belperpete]

Would be better not to have an automatic raise timer at all. That way if a train is detected and then subsequently disappears or isn't positively detected passing the crossing, the barriers stay down and the lights stay on. Ie it fails safe which is exactly what happens at treadle operated AHBs.
Raising after a set time to allow for a train having stopped might be appropriate for certain settings in America, where they seem to have marshalling yards running over streets. But I can't think of any UK setting making it a necessary or desirable feature.

Most AHBs have the same kind of reset timers, in case of false track-circuit operation.

 

[trebor79]

Most AHBs have the same kind of reset timers, in case of false track-circuit operation.

But presumably a much longer timer than was the case with this crossing?
As an aside, how often would a track circuit falsely indicate as occupied and then lose that indication? Just curious.

 

[alxndr]

But presumably a much longer timer than was the case with this crossing?
As an aside, how often would a track circuit falsely indicate as occupied and then lose that indication? Just curious.

If all is working within spec it should never happen, but failures do occasionally occur. The majority never fail, or if they do it's years apart, but failures do happen, and that needs to be catered for.

The problem if the barriers steady down is that the public are likely to get fed up, assume nothing is coming, and chance it.

 

[edwin_m]

But presumably a much longer timer than was the case with this crossing?
As an aside, how often would a track circuit falsely indicate as occupied and then lose that indication? Just curious.

A bad contact in the electrical circuit could cause a false occupation and could be intermittent. Potentially if the rail insulation isn't up to scratch it could fail when wet and recover when dried out. Both of these would be right side failures (showing occupied when not occupied). Too much electrical resistance between the train and the rails would be a wrong side failure.

I think the AHB timers raise the barriers after the time period even if the track circuit in question in still occupied, on the grounds that if it's been occupied for that long the train will be moving very slowly, if at all, so the driver will be able to stop if the barriers are seen to be up.

But predictors are more complicated than track circuits so their behavior in fault conditions is probably less straightforward too.

 

[Belperpete]

But presumably a much longer timer than was the case with this crossing?

I don't know what the timer was set to on this crossing. On conventional AHBs, I think it is set to three minutes. As the fastest train should arrive within about half a minute, three minutes should be more than adequate, even for a very slow train.

 

[LAX54]

I don't know what the timer was set to on this crossing. On conventional AHBs, I think it is set to three minutes. As the fastest train should arrive within about half a minute, three minutes should be more than adequate, even for a very slow train.

Conventional AHB's minimum time about 30 seconds, if they have not raised in 3 mins, there is a 'failed' alarm in the signalbox, the Signaller then carries out the R&R that goes with such an alarm, if it is a 'slow' train, then the barriers will of course raise after it has passed by, and then alarm will silence and the 'FAILED' indication go out. Cromer line, are however not conventional as far as I can see.